From 32b3ed0a1362a4b0798ad71fac3450fb77cb7e41 Mon Sep 17 00:00:00 2001 From: Thomas Groman Date: Thu, 19 Sep 2019 00:41:48 -0700 Subject: merged from 0.6.7 codebase --- libraries/pack200/src/bands.cpp | 642 +-- libraries/pack200/src/bands.h | 604 +-- libraries/pack200/src/bytes.cpp | 246 +- libraries/pack200/src/bytes.h | 472 +- libraries/pack200/src/coding.cpp | 1762 ++++---- libraries/pack200/src/coding.h | 348 +- libraries/pack200/src/constants.h | 712 +-- libraries/pack200/src/unpack.cpp | 8363 +++++++++++++++++------------------ libraries/pack200/src/unpack.h | 992 ++--- libraries/pack200/src/unpack200.cpp | 188 +- libraries/pack200/src/utils.cpp | 28 +- libraries/pack200/src/utils.h | 6 +- libraries/pack200/src/zip.cpp | 838 ++-- libraries/pack200/src/zip.h | 136 +- 14 files changed, 7669 insertions(+), 7668 deletions(-) (limited to 'libraries/pack200/src') diff --git a/libraries/pack200/src/bands.cpp b/libraries/pack200/src/bands.cpp index 1608d838..e82613b5 100644 --- a/libraries/pack200/src/bands.cpp +++ b/libraries/pack200/src/bands.cpp @@ -47,188 +47,188 @@ void band::readData(int expectedLength) { - assert(expectedLength >= 0); - assert(vs[0].cmk == cmk_ERROR); - if (expectedLength != 0) - { - assert(length == 0); - length = expectedLength; - } - if (length == 0) - { - assert((rplimit = cm.vs0.rp = u->rp) != nullptr); - return; - } - assert(length > 0); + assert(expectedLength >= 0); + assert(vs[0].cmk == cmk_ERROR); + if (expectedLength != 0) + { + assert(length == 0); + length = expectedLength; + } + if (length == 0) + { + assert((rplimit = cm.vs0.rp = u->rp) != nullptr); + return; + } + assert(length > 0); - bool is_BYTE1 = (defc->spec == BYTE1_spec); + bool is_BYTE1 = (defc->spec == BYTE1_spec); - if (is_BYTE1) - { - // No possibility of coding change. Sizing is exact. - u->ensure_input(length); - } - else - { - // Make a conservatively generous estimate of band size in bytes. - // Assume B == 5 everywhere. - // Assume awkward pop with all {U} values (2*5 per value) - int64_t generous = (int64_t)length * (B_MAX * 3 + 1) + C_SLOP; - u->ensure_input(generous); - } + if (is_BYTE1) + { + // No possibility of coding change. Sizing is exact. + u->ensure_input(length); + } + else + { + // Make a conservatively generous estimate of band size in bytes. + // Assume B == 5 everywhere. + // Assume awkward pop with all {U} values (2*5 per value) + int64_t generous = (int64_t)length * (B_MAX * 3 + 1) + C_SLOP; + u->ensure_input(generous); + } - // Read one value to see what it might be. - int XB = _meta_default; - if (!is_BYTE1) - { - // must be a variable-length coding - assert(defc->B() > 1 && defc->L() > 0); + // Read one value to see what it might be. + int XB = _meta_default; + if (!is_BYTE1) + { + // must be a variable-length coding + assert(defc->B() > 1 && defc->L() > 0); - value_stream xvs; - coding *valc = defc; - if (valc->D() != 0) - { - valc = coding::findBySpec(defc->B(), defc->H(), defc->S()); - assert(!valc->isMalloc); - } - xvs.init(u->rp, u->rplimit, valc); - int X = xvs.getInt(); - if (valc->S() != 0) - { - assert(valc->min <= -256); - XB = -1 - X; - } - else - { - int L = valc->L(); - assert(valc->max >= L + 255); - XB = X - L; - } - if (0 <= XB && XB < 256) - { - // Skip over the escape value. - u->rp = xvs.rp; - } - else - { - // No, it's still default. - XB = _meta_default; - } - } + value_stream xvs; + coding *valc = defc; + if (valc->D() != 0) + { + valc = coding::findBySpec(defc->B(), defc->H(), defc->S()); + assert(!valc->isMalloc); + } + xvs.init(u->rp, u->rplimit, valc); + int X = xvs.getInt(); + if (valc->S() != 0) + { + assert(valc->min <= -256); + XB = -1 - X; + } + else + { + int L = valc->L(); + assert(valc->max >= L + 255); + XB = X - L; + } + if (0 <= XB && XB < 256) + { + // Skip over the escape value. + u->rp = xvs.rp; + } + else + { + // No, it's still default. + XB = _meta_default; + } + } - if (XB <= _meta_canon_max) - { - byte XB_byte = (byte)XB; - byte *XB_ptr = &XB_byte; - cm.init(u->rp, u->rplimit, XB_ptr, 0, defc, length, nullptr); - } - else - { - assert(u->meta_rp != nullptr); - // Scribble the initial byte onto the band. - byte *save_meta_rp = --u->meta_rp; - byte save_meta_xb = (*save_meta_rp); - (*save_meta_rp) = (byte)XB; - cm.init(u->rp, u->rplimit, u->meta_rp, 0, defc, length, nullptr); - (*save_meta_rp) = save_meta_xb; // put it back, just to be tidy - } - rplimit = u->rp; + if (XB <= _meta_canon_max) + { + byte XB_byte = (byte)XB; + byte *XB_ptr = &XB_byte; + cm.init(u->rp, u->rplimit, XB_ptr, 0, defc, length, nullptr); + } + else + { + assert(u->meta_rp != nullptr); + // Scribble the initial byte onto the band. + byte *save_meta_rp = --u->meta_rp; + byte save_meta_xb = (*save_meta_rp); + (*save_meta_rp) = (byte)XB; + cm.init(u->rp, u->rplimit, u->meta_rp, 0, defc, length, nullptr); + (*save_meta_rp) = save_meta_xb; // put it back, just to be tidy + } + rplimit = u->rp; - rewind(); + rewind(); } void band::setIndex(cpindex *ix_) { - assert(ix_ == nullptr || ixTag == ix_->ixTag); - ix = ix_; + assert(ix_ == nullptr || ixTag == ix_->ixTag); + ix = ix_; } void band::setIndexByTag(byte tag) { - setIndex(u->cp.getIndex(tag)); + setIndex(u->cp.getIndex(tag)); } entry *band::getRefCommon(cpindex *ix_, bool nullOKwithCaller) { - assert(ix_->ixTag == ixTag || - (ixTag == CONSTANT_Literal && ix_->ixTag >= CONSTANT_Integer && - ix_->ixTag <= CONSTANT_String)); - int n = vs[0].getInt() - nullOK; - // Note: band-local nullOK means nullptr encodes as 0. - // But nullOKwithCaller means caller is willing to tolerate a nullptr. - entry *ref = ix_->get(n); - if (ref == nullptr && !(nullOKwithCaller && n == -1)) - unpack_abort(n == -1 ? "nullptr ref" : "bad ref"); - return ref; + assert(ix_->ixTag == ixTag || + (ixTag == CONSTANT_Literal && ix_->ixTag >= CONSTANT_Integer && + ix_->ixTag <= CONSTANT_String)); + int n = vs[0].getInt() - nullOK; + // Note: band-local nullOK means nullptr encodes as 0. + // But nullOKwithCaller means caller is willing to tolerate a nullptr. + entry *ref = ix_->get(n); + if (ref == nullptr && !(nullOKwithCaller && n == -1)) + unpack_abort(n == -1 ? "nullptr ref" : "bad ref"); + return ref; } int64_t band::getLong(band &lo_band, bool have_hi) { - band &hi_band = (*this); - assert(lo_band.bn == hi_band.bn + 1); - uint32_t lo = lo_band.getInt(); - if (!have_hi) - { - assert(hi_band.length == 0); - return makeLong(0, lo); - } - uint32_t hi = hi_band.getInt(); - return makeLong(hi, lo); + band &hi_band = (*this); + assert(lo_band.bn == hi_band.bn + 1); + uint32_t lo = lo_band.getInt(); + if (!have_hi) + { + assert(hi_band.length == 0); + return makeLong(0, lo); + } + uint32_t hi = hi_band.getInt(); + return makeLong(hi, lo); } int band::getIntTotal() { - if (length == 0) - return 0; - if (total_memo > 0) - return total_memo - 1; - int total = getInt(); - // overflow checks require that none of the addends are <0, - // and that the partial sums never overflow (wrap negative) - if (total < 0) - { - unpack_abort("overflow detected"); - } - for (int k = length - 1; k > 0; k--) - { - int prev_total = total; - total += vs[0].getInt(); - if (total < prev_total) - { - unpack_abort("overflow detected"); - } - } - rewind(); - total_memo = total + 1; - return total; + if (length == 0) + return 0; + if (total_memo > 0) + return total_memo - 1; + int total = getInt(); + // overflow checks require that none of the addends are <0, + // and that the partial sums never overflow (wrap negative) + if (total < 0) + { + unpack_abort("overflow detected"); + } + for (int k = length - 1; k > 0; k--) + { + int prev_total = total; + total += vs[0].getInt(); + if (total < prev_total) + { + unpack_abort("overflow detected"); + } + } + rewind(); + total_memo = total + 1; + return total; } int band::getIntCount(int tag) { - if (length == 0) - return 0; - if (tag >= HIST0_MIN && tag <= HIST0_MAX) - { - if (hist0 == nullptr) - { - // Lazily calculate an approximate histogram. - hist0 = U_NEW(int, (HIST0_MAX - HIST0_MIN) + 1); - for (int k = length; k > 0; k--) - { - int x = vs[0].getInt(); - if (x >= HIST0_MIN && x <= HIST0_MAX) - hist0[x - HIST0_MIN] += 1; - } - rewind(); - } - return hist0[tag - HIST0_MIN]; - } - int total = 0; - for (int k = length; k > 0; k--) - { - total += (vs[0].getInt() == tag) ? 1 : 0; - } - rewind(); - return total; + if (length == 0) + return 0; + if (tag >= HIST0_MIN && tag <= HIST0_MAX) + { + if (hist0 == nullptr) + { + // Lazily calculate an approximate histogram. + hist0 = U_NEW(int, (HIST0_MAX - HIST0_MIN) + 1); + for (int k = length; k > 0; k--) + { + int x = vs[0].getInt(); + if (x >= HIST0_MIN && x <= HIST0_MAX) + hist0[x - HIST0_MIN] += 1; + } + rewind(); + } + return hist0[tag - HIST0_MIN]; + } + int total = 0; + for (int k = length; k > 0; k--) + { + total += (vs[0].getInt() == tag) ? 1 : 0; + } + rewind(); + return total; } #define INDEX_INIT(tag, nullOK, subindex) ((tag) + (subindex) * SUBINDEX_BIT + (nullOK) * 256) @@ -240,184 +240,184 @@ int band::getIntCount(int tag) struct band_init { - int defc; - int index; + int defc; + int index; }; #define BAND_INIT(name, cspec, ix) \ - { \ - cspec, ix \ - } + { \ + cspec, ix \ + } const band_init all_band_inits[] = - { - // BAND_INIT(archive_magic, BYTE1_spec, 0), - // BAND_INIT(archive_header, UNSIGNED5_spec, 0), - // BAND_INIT(band_headers, BYTE1_spec, 0), - BAND_INIT(cp_Utf8_prefix, DELTA5_spec, 0), BAND_INIT(cp_Utf8_suffix, UNSIGNED5_spec, 0), - BAND_INIT(cp_Utf8_chars, CHAR3_spec, 0), BAND_INIT(cp_Utf8_big_suffix, DELTA5_spec, 0), - BAND_INIT(cp_Utf8_big_chars, DELTA5_spec, 0), BAND_INIT(cp_Int, UDELTA5_spec, 0), - BAND_INIT(cp_Float, UDELTA5_spec, 0), BAND_INIT(cp_Long_hi, UDELTA5_spec, 0), - BAND_INIT(cp_Long_lo, DELTA5_spec, 0), BAND_INIT(cp_Double_hi, UDELTA5_spec, 0), - BAND_INIT(cp_Double_lo, DELTA5_spec, 0), - BAND_INIT(cp_String, UDELTA5_spec, INDEX(CONSTANT_Utf8)), - BAND_INIT(cp_Class, UDELTA5_spec, INDEX(CONSTANT_Utf8)), - BAND_INIT(cp_Signature_form, DELTA5_spec, INDEX(CONSTANT_Utf8)), - BAND_INIT(cp_Signature_classes, UDELTA5_spec, INDEX(CONSTANT_Class)), - BAND_INIT(cp_Descr_name, DELTA5_spec, INDEX(CONSTANT_Utf8)), - BAND_INIT(cp_Descr_type, UDELTA5_spec, INDEX(CONSTANT_Signature)), - BAND_INIT(cp_Field_class, DELTA5_spec, INDEX(CONSTANT_Class)), - BAND_INIT(cp_Field_desc, UDELTA5_spec, INDEX(CONSTANT_NameandType)), - BAND_INIT(cp_Method_class, DELTA5_spec, INDEX(CONSTANT_Class)), - BAND_INIT(cp_Method_desc, UDELTA5_spec, INDEX(CONSTANT_NameandType)), - BAND_INIT(cp_Imethod_class, DELTA5_spec, INDEX(CONSTANT_Class)), - BAND_INIT(cp_Imethod_desc, UDELTA5_spec, INDEX(CONSTANT_NameandType)), - BAND_INIT(attr_definition_headers, BYTE1_spec, 0), - BAND_INIT(attr_definition_name, UNSIGNED5_spec, INDEX(CONSTANT_Utf8)), - BAND_INIT(attr_definition_layout, UNSIGNED5_spec, INDEX(CONSTANT_Utf8)), - BAND_INIT(ic_this_class, UDELTA5_spec, INDEX(CONSTANT_Class)), - BAND_INIT(ic_flags, UNSIGNED5_spec, 0), - BAND_INIT(ic_outer_class, DELTA5_spec, NULL_OR_INDEX(CONSTANT_Class)), - BAND_INIT(ic_name, DELTA5_spec, NULL_OR_INDEX(CONSTANT_Utf8)), - BAND_INIT(class_this, DELTA5_spec, INDEX(CONSTANT_Class)), - BAND_INIT(class_super, DELTA5_spec, INDEX(CONSTANT_Class)), - BAND_INIT(class_interface_count, DELTA5_spec, 0), - BAND_INIT(class_interface, DELTA5_spec, INDEX(CONSTANT_Class)), - BAND_INIT(class_field_count, DELTA5_spec, 0), - BAND_INIT(class_method_count, DELTA5_spec, 0), - BAND_INIT(field_descr, DELTA5_spec, INDEX(CONSTANT_NameandType)), - BAND_INIT(field_flags_hi, UNSIGNED5_spec, 0), - BAND_INIT(field_flags_lo, UNSIGNED5_spec, 0), - BAND_INIT(field_attr_count, UNSIGNED5_spec, 0), - BAND_INIT(field_attr_indexes, UNSIGNED5_spec, 0), - BAND_INIT(field_attr_calls, UNSIGNED5_spec, 0), - BAND_INIT(field_ConstantValue_KQ, UNSIGNED5_spec, INDEX(CONSTANT_Literal)), - BAND_INIT(field_Signature_RS, UNSIGNED5_spec, INDEX(CONSTANT_Signature)), - BAND_INIT(field_metadata_bands, -1, -1), BAND_INIT(field_attr_bands, -1, -1), - BAND_INIT(method_descr, MDELTA5_spec, INDEX(CONSTANT_NameandType)), - BAND_INIT(method_flags_hi, UNSIGNED5_spec, 0), - BAND_INIT(method_flags_lo, UNSIGNED5_spec, 0), - BAND_INIT(method_attr_count, UNSIGNED5_spec, 0), - BAND_INIT(method_attr_indexes, UNSIGNED5_spec, 0), - BAND_INIT(method_attr_calls, UNSIGNED5_spec, 0), - BAND_INIT(method_Exceptions_N, UNSIGNED5_spec, 0), - BAND_INIT(method_Exceptions_RC, UNSIGNED5_spec, INDEX(CONSTANT_Class)), - BAND_INIT(method_Signature_RS, UNSIGNED5_spec, INDEX(CONSTANT_Signature)), - BAND_INIT(method_metadata_bands, -1, -1), BAND_INIT(method_attr_bands, -1, -1), - BAND_INIT(class_flags_hi, UNSIGNED5_spec, 0), - BAND_INIT(class_flags_lo, UNSIGNED5_spec, 0), - BAND_INIT(class_attr_count, UNSIGNED5_spec, 0), - BAND_INIT(class_attr_indexes, UNSIGNED5_spec, 0), - BAND_INIT(class_attr_calls, UNSIGNED5_spec, 0), - BAND_INIT(class_SourceFile_RUN, UNSIGNED5_spec, NULL_OR_INDEX(CONSTANT_Utf8)), - BAND_INIT(class_EnclosingMethod_RC, UNSIGNED5_spec, INDEX(CONSTANT_Class)), - BAND_INIT(class_EnclosingMethod_RDN, UNSIGNED5_spec, - NULL_OR_INDEX(CONSTANT_NameandType)), - BAND_INIT(class_Signature_RS, UNSIGNED5_spec, INDEX(CONSTANT_Signature)), - BAND_INIT(class_metadata_bands, -1, -1), - BAND_INIT(class_InnerClasses_N, UNSIGNED5_spec, 0), - BAND_INIT(class_InnerClasses_RC, UNSIGNED5_spec, INDEX(CONSTANT_Class)), - BAND_INIT(class_InnerClasses_F, UNSIGNED5_spec, 0), - BAND_INIT(class_InnerClasses_outer_RCN, UNSIGNED5_spec, NULL_OR_INDEX(CONSTANT_Class)), - BAND_INIT(class_InnerClasses_name_RUN, UNSIGNED5_spec, NULL_OR_INDEX(CONSTANT_Utf8)), - BAND_INIT(class_ClassFile_version_minor_H, UNSIGNED5_spec, 0), - BAND_INIT(class_ClassFile_version_major_H, UNSIGNED5_spec, 0), - BAND_INIT(class_attr_bands, -1, -1), BAND_INIT(code_headers, BYTE1_spec, 0), - BAND_INIT(code_max_stack, UNSIGNED5_spec, 0), - BAND_INIT(code_max_na_locals, UNSIGNED5_spec, 0), - BAND_INIT(code_handler_count, UNSIGNED5_spec, 0), - BAND_INIT(code_handler_start_P, BCI5_spec, 0), - BAND_INIT(code_handler_end_PO, BRANCH5_spec, 0), - BAND_INIT(code_handler_catch_PO, BRANCH5_spec, 0), - BAND_INIT(code_handler_class_RCN, UNSIGNED5_spec, NULL_OR_INDEX(CONSTANT_Class)), - BAND_INIT(code_flags_hi, UNSIGNED5_spec, 0), - BAND_INIT(code_flags_lo, UNSIGNED5_spec, 0), - BAND_INIT(code_attr_count, UNSIGNED5_spec, 0), - BAND_INIT(code_attr_indexes, UNSIGNED5_spec, 0), - BAND_INIT(code_attr_calls, UNSIGNED5_spec, 0), - BAND_INIT(code_StackMapTable_N, UNSIGNED5_spec, 0), - BAND_INIT(code_StackMapTable_frame_T, BYTE1_spec, 0), - BAND_INIT(code_StackMapTable_local_N, UNSIGNED5_spec, 0), - BAND_INIT(code_StackMapTable_stack_N, UNSIGNED5_spec, 0), - BAND_INIT(code_StackMapTable_offset, UNSIGNED5_spec, 0), - BAND_INIT(code_StackMapTable_T, BYTE1_spec, 0), - BAND_INIT(code_StackMapTable_RC, UNSIGNED5_spec, INDEX(CONSTANT_Class)), - BAND_INIT(code_StackMapTable_P, BCI5_spec, 0), - BAND_INIT(code_LineNumberTable_N, UNSIGNED5_spec, 0), - BAND_INIT(code_LineNumberTable_bci_P, BCI5_spec, 0), - BAND_INIT(code_LineNumberTable_line, UNSIGNED5_spec, 0), - BAND_INIT(code_LocalVariableTable_N, UNSIGNED5_spec, 0), - BAND_INIT(code_LocalVariableTable_bci_P, BCI5_spec, 0), - BAND_INIT(code_LocalVariableTable_span_O, BRANCH5_spec, 0), - BAND_INIT(code_LocalVariableTable_name_RU, UNSIGNED5_spec, INDEX(CONSTANT_Utf8)), - BAND_INIT(code_LocalVariableTable_type_RS, UNSIGNED5_spec, INDEX(CONSTANT_Signature)), - BAND_INIT(code_LocalVariableTable_slot, UNSIGNED5_spec, 0), - BAND_INIT(code_LocalVariableTypeTable_N, UNSIGNED5_spec, 0), - BAND_INIT(code_LocalVariableTypeTable_bci_P, BCI5_spec, 0), - BAND_INIT(code_LocalVariableTypeTable_span_O, BRANCH5_spec, 0), - BAND_INIT(code_LocalVariableTypeTable_name_RU, UNSIGNED5_spec, INDEX(CONSTANT_Utf8)), - BAND_INIT(code_LocalVariableTypeTable_type_RS, UNSIGNED5_spec, - INDEX(CONSTANT_Signature)), - BAND_INIT(code_LocalVariableTypeTable_slot, UNSIGNED5_spec, 0), - BAND_INIT(code_attr_bands, -1, -1), BAND_INIT(bc_codes, BYTE1_spec, 0), - BAND_INIT(bc_case_count, UNSIGNED5_spec, 0), BAND_INIT(bc_case_value, DELTA5_spec, 0), - BAND_INIT(bc_byte, BYTE1_spec, 0), BAND_INIT(bc_short, DELTA5_spec, 0), - BAND_INIT(bc_local, UNSIGNED5_spec, 0), BAND_INIT(bc_label, BRANCH5_spec, 0), - BAND_INIT(bc_intref, DELTA5_spec, INDEX(CONSTANT_Integer)), - BAND_INIT(bc_floatref, DELTA5_spec, INDEX(CONSTANT_Float)), - BAND_INIT(bc_longref, DELTA5_spec, INDEX(CONSTANT_Long)), - BAND_INIT(bc_doubleref, DELTA5_spec, INDEX(CONSTANT_Double)), - BAND_INIT(bc_stringref, DELTA5_spec, INDEX(CONSTANT_String)), - BAND_INIT(bc_classref, UNSIGNED5_spec, NULL_OR_INDEX(CONSTANT_Class)), - BAND_INIT(bc_fieldref, DELTA5_spec, INDEX(CONSTANT_Fieldref)), - BAND_INIT(bc_methodref, UNSIGNED5_spec, INDEX(CONSTANT_Methodref)), - BAND_INIT(bc_imethodref, DELTA5_spec, INDEX(CONSTANT_InterfaceMethodref)), - BAND_INIT(bc_thisfield, UNSIGNED5_spec, SUB_INDEX(CONSTANT_Fieldref)), - BAND_INIT(bc_superfield, UNSIGNED5_spec, SUB_INDEX(CONSTANT_Fieldref)), - BAND_INIT(bc_thismethod, UNSIGNED5_spec, SUB_INDEX(CONSTANT_Methodref)), - BAND_INIT(bc_supermethod, UNSIGNED5_spec, SUB_INDEX(CONSTANT_Methodref)), - BAND_INIT(bc_initref, UNSIGNED5_spec, SUB_INDEX(CONSTANT_Methodref)), - BAND_INIT(bc_escref, UNSIGNED5_spec, INDEX(CONSTANT_All)), - BAND_INIT(bc_escrefsize, UNSIGNED5_spec, 0), BAND_INIT(bc_escsize, UNSIGNED5_spec, 0), - BAND_INIT(bc_escbyte, BYTE1_spec, 0), - BAND_INIT(file_name, UNSIGNED5_spec, INDEX(CONSTANT_Utf8)), - BAND_INIT(file_size_hi, UNSIGNED5_spec, 0), BAND_INIT(file_size_lo, UNSIGNED5_spec, 0), - BAND_INIT(file_modtime, DELTA5_spec, 0), BAND_INIT(file_options, UNSIGNED5_spec, 0), - // BAND_INIT(file_bits, BYTE1_spec, 0), - {0, 0}}; + { + // BAND_INIT(archive_magic, BYTE1_spec, 0), + // BAND_INIT(archive_header, UNSIGNED5_spec, 0), + // BAND_INIT(band_headers, BYTE1_spec, 0), + BAND_INIT(cp_Utf8_prefix, DELTA5_spec, 0), BAND_INIT(cp_Utf8_suffix, UNSIGNED5_spec, 0), + BAND_INIT(cp_Utf8_chars, CHAR3_spec, 0), BAND_INIT(cp_Utf8_big_suffix, DELTA5_spec, 0), + BAND_INIT(cp_Utf8_big_chars, DELTA5_spec, 0), BAND_INIT(cp_Int, UDELTA5_spec, 0), + BAND_INIT(cp_Float, UDELTA5_spec, 0), BAND_INIT(cp_Long_hi, UDELTA5_spec, 0), + BAND_INIT(cp_Long_lo, DELTA5_spec, 0), BAND_INIT(cp_Double_hi, UDELTA5_spec, 0), + BAND_INIT(cp_Double_lo, DELTA5_spec, 0), + BAND_INIT(cp_String, UDELTA5_spec, INDEX(CONSTANT_Utf8)), + BAND_INIT(cp_Class, UDELTA5_spec, INDEX(CONSTANT_Utf8)), + BAND_INIT(cp_Signature_form, DELTA5_spec, INDEX(CONSTANT_Utf8)), + BAND_INIT(cp_Signature_classes, UDELTA5_spec, INDEX(CONSTANT_Class)), + BAND_INIT(cp_Descr_name, DELTA5_spec, INDEX(CONSTANT_Utf8)), + BAND_INIT(cp_Descr_type, UDELTA5_spec, INDEX(CONSTANT_Signature)), + BAND_INIT(cp_Field_class, DELTA5_spec, INDEX(CONSTANT_Class)), + BAND_INIT(cp_Field_desc, UDELTA5_spec, INDEX(CONSTANT_NameandType)), + BAND_INIT(cp_Method_class, DELTA5_spec, INDEX(CONSTANT_Class)), + BAND_INIT(cp_Method_desc, UDELTA5_spec, INDEX(CONSTANT_NameandType)), + BAND_INIT(cp_Imethod_class, DELTA5_spec, INDEX(CONSTANT_Class)), + BAND_INIT(cp_Imethod_desc, UDELTA5_spec, INDEX(CONSTANT_NameandType)), + BAND_INIT(attr_definition_headers, BYTE1_spec, 0), + BAND_INIT(attr_definition_name, UNSIGNED5_spec, INDEX(CONSTANT_Utf8)), + BAND_INIT(attr_definition_layout, UNSIGNED5_spec, INDEX(CONSTANT_Utf8)), + BAND_INIT(ic_this_class, UDELTA5_spec, INDEX(CONSTANT_Class)), + BAND_INIT(ic_flags, UNSIGNED5_spec, 0), + BAND_INIT(ic_outer_class, DELTA5_spec, NULL_OR_INDEX(CONSTANT_Class)), + BAND_INIT(ic_name, DELTA5_spec, NULL_OR_INDEX(CONSTANT_Utf8)), + BAND_INIT(class_this, DELTA5_spec, INDEX(CONSTANT_Class)), + BAND_INIT(class_super, DELTA5_spec, INDEX(CONSTANT_Class)), + BAND_INIT(class_interface_count, DELTA5_spec, 0), + BAND_INIT(class_interface, DELTA5_spec, INDEX(CONSTANT_Class)), + BAND_INIT(class_field_count, DELTA5_spec, 0), + BAND_INIT(class_method_count, DELTA5_spec, 0), + BAND_INIT(field_descr, DELTA5_spec, INDEX(CONSTANT_NameandType)), + BAND_INIT(field_flags_hi, UNSIGNED5_spec, 0), + BAND_INIT(field_flags_lo, UNSIGNED5_spec, 0), + BAND_INIT(field_attr_count, UNSIGNED5_spec, 0), + BAND_INIT(field_attr_indexes, UNSIGNED5_spec, 0), + BAND_INIT(field_attr_calls, UNSIGNED5_spec, 0), + BAND_INIT(field_ConstantValue_KQ, UNSIGNED5_spec, INDEX(CONSTANT_Literal)), + BAND_INIT(field_Signature_RS, UNSIGNED5_spec, INDEX(CONSTANT_Signature)), + BAND_INIT(field_metadata_bands, -1, -1), BAND_INIT(field_attr_bands, -1, -1), + BAND_INIT(method_descr, MDELTA5_spec, INDEX(CONSTANT_NameandType)), + BAND_INIT(method_flags_hi, UNSIGNED5_spec, 0), + BAND_INIT(method_flags_lo, UNSIGNED5_spec, 0), + BAND_INIT(method_attr_count, UNSIGNED5_spec, 0), + BAND_INIT(method_attr_indexes, UNSIGNED5_spec, 0), + BAND_INIT(method_attr_calls, UNSIGNED5_spec, 0), + BAND_INIT(method_Exceptions_N, UNSIGNED5_spec, 0), + BAND_INIT(method_Exceptions_RC, UNSIGNED5_spec, INDEX(CONSTANT_Class)), + BAND_INIT(method_Signature_RS, UNSIGNED5_spec, INDEX(CONSTANT_Signature)), + BAND_INIT(method_metadata_bands, -1, -1), BAND_INIT(method_attr_bands, -1, -1), + BAND_INIT(class_flags_hi, UNSIGNED5_spec, 0), + BAND_INIT(class_flags_lo, UNSIGNED5_spec, 0), + BAND_INIT(class_attr_count, UNSIGNED5_spec, 0), + BAND_INIT(class_attr_indexes, UNSIGNED5_spec, 0), + BAND_INIT(class_attr_calls, UNSIGNED5_spec, 0), + BAND_INIT(class_SourceFile_RUN, UNSIGNED5_spec, NULL_OR_INDEX(CONSTANT_Utf8)), + BAND_INIT(class_EnclosingMethod_RC, UNSIGNED5_spec, INDEX(CONSTANT_Class)), + BAND_INIT(class_EnclosingMethod_RDN, UNSIGNED5_spec, + NULL_OR_INDEX(CONSTANT_NameandType)), + BAND_INIT(class_Signature_RS, UNSIGNED5_spec, INDEX(CONSTANT_Signature)), + BAND_INIT(class_metadata_bands, -1, -1), + BAND_INIT(class_InnerClasses_N, UNSIGNED5_spec, 0), + BAND_INIT(class_InnerClasses_RC, UNSIGNED5_spec, INDEX(CONSTANT_Class)), + BAND_INIT(class_InnerClasses_F, UNSIGNED5_spec, 0), + BAND_INIT(class_InnerClasses_outer_RCN, UNSIGNED5_spec, NULL_OR_INDEX(CONSTANT_Class)), + BAND_INIT(class_InnerClasses_name_RUN, UNSIGNED5_spec, NULL_OR_INDEX(CONSTANT_Utf8)), + BAND_INIT(class_ClassFile_version_minor_H, UNSIGNED5_spec, 0), + BAND_INIT(class_ClassFile_version_major_H, UNSIGNED5_spec, 0), + BAND_INIT(class_attr_bands, -1, -1), BAND_INIT(code_headers, BYTE1_spec, 0), + BAND_INIT(code_max_stack, UNSIGNED5_spec, 0), + BAND_INIT(code_max_na_locals, UNSIGNED5_spec, 0), + BAND_INIT(code_handler_count, UNSIGNED5_spec, 0), + BAND_INIT(code_handler_start_P, BCI5_spec, 0), + BAND_INIT(code_handler_end_PO, BRANCH5_spec, 0), + BAND_INIT(code_handler_catch_PO, BRANCH5_spec, 0), + BAND_INIT(code_handler_class_RCN, UNSIGNED5_spec, NULL_OR_INDEX(CONSTANT_Class)), + BAND_INIT(code_flags_hi, UNSIGNED5_spec, 0), + BAND_INIT(code_flags_lo, UNSIGNED5_spec, 0), + BAND_INIT(code_attr_count, UNSIGNED5_spec, 0), + BAND_INIT(code_attr_indexes, UNSIGNED5_spec, 0), + BAND_INIT(code_attr_calls, UNSIGNED5_spec, 0), + BAND_INIT(code_StackMapTable_N, UNSIGNED5_spec, 0), + BAND_INIT(code_StackMapTable_frame_T, BYTE1_spec, 0), + BAND_INIT(code_StackMapTable_local_N, UNSIGNED5_spec, 0), + BAND_INIT(code_StackMapTable_stack_N, UNSIGNED5_spec, 0), + BAND_INIT(code_StackMapTable_offset, UNSIGNED5_spec, 0), + BAND_INIT(code_StackMapTable_T, BYTE1_spec, 0), + BAND_INIT(code_StackMapTable_RC, UNSIGNED5_spec, INDEX(CONSTANT_Class)), + BAND_INIT(code_StackMapTable_P, BCI5_spec, 0), + BAND_INIT(code_LineNumberTable_N, UNSIGNED5_spec, 0), + BAND_INIT(code_LineNumberTable_bci_P, BCI5_spec, 0), + BAND_INIT(code_LineNumberTable_line, UNSIGNED5_spec, 0), + BAND_INIT(code_LocalVariableTable_N, UNSIGNED5_spec, 0), + BAND_INIT(code_LocalVariableTable_bci_P, BCI5_spec, 0), + BAND_INIT(code_LocalVariableTable_span_O, BRANCH5_spec, 0), + BAND_INIT(code_LocalVariableTable_name_RU, UNSIGNED5_spec, INDEX(CONSTANT_Utf8)), + BAND_INIT(code_LocalVariableTable_type_RS, UNSIGNED5_spec, INDEX(CONSTANT_Signature)), + BAND_INIT(code_LocalVariableTable_slot, UNSIGNED5_spec, 0), + BAND_INIT(code_LocalVariableTypeTable_N, UNSIGNED5_spec, 0), + BAND_INIT(code_LocalVariableTypeTable_bci_P, BCI5_spec, 0), + BAND_INIT(code_LocalVariableTypeTable_span_O, BRANCH5_spec, 0), + BAND_INIT(code_LocalVariableTypeTable_name_RU, UNSIGNED5_spec, INDEX(CONSTANT_Utf8)), + BAND_INIT(code_LocalVariableTypeTable_type_RS, UNSIGNED5_spec, + INDEX(CONSTANT_Signature)), + BAND_INIT(code_LocalVariableTypeTable_slot, UNSIGNED5_spec, 0), + BAND_INIT(code_attr_bands, -1, -1), BAND_INIT(bc_codes, BYTE1_spec, 0), + BAND_INIT(bc_case_count, UNSIGNED5_spec, 0), BAND_INIT(bc_case_value, DELTA5_spec, 0), + BAND_INIT(bc_byte, BYTE1_spec, 0), BAND_INIT(bc_short, DELTA5_spec, 0), + BAND_INIT(bc_local, UNSIGNED5_spec, 0), BAND_INIT(bc_label, BRANCH5_spec, 0), + BAND_INIT(bc_intref, DELTA5_spec, INDEX(CONSTANT_Integer)), + BAND_INIT(bc_floatref, DELTA5_spec, INDEX(CONSTANT_Float)), + BAND_INIT(bc_longref, DELTA5_spec, INDEX(CONSTANT_Long)), + BAND_INIT(bc_doubleref, DELTA5_spec, INDEX(CONSTANT_Double)), + BAND_INIT(bc_stringref, DELTA5_spec, INDEX(CONSTANT_String)), + BAND_INIT(bc_classref, UNSIGNED5_spec, NULL_OR_INDEX(CONSTANT_Class)), + BAND_INIT(bc_fieldref, DELTA5_spec, INDEX(CONSTANT_Fieldref)), + BAND_INIT(bc_methodref, UNSIGNED5_spec, INDEX(CONSTANT_Methodref)), + BAND_INIT(bc_imethodref, DELTA5_spec, INDEX(CONSTANT_InterfaceMethodref)), + BAND_INIT(bc_thisfield, UNSIGNED5_spec, SUB_INDEX(CONSTANT_Fieldref)), + BAND_INIT(bc_superfield, UNSIGNED5_spec, SUB_INDEX(CONSTANT_Fieldref)), + BAND_INIT(bc_thismethod, UNSIGNED5_spec, SUB_INDEX(CONSTANT_Methodref)), + BAND_INIT(bc_supermethod, UNSIGNED5_spec, SUB_INDEX(CONSTANT_Methodref)), + BAND_INIT(bc_initref, UNSIGNED5_spec, SUB_INDEX(CONSTANT_Methodref)), + BAND_INIT(bc_escref, UNSIGNED5_spec, INDEX(CONSTANT_All)), + BAND_INIT(bc_escrefsize, UNSIGNED5_spec, 0), BAND_INIT(bc_escsize, UNSIGNED5_spec, 0), + BAND_INIT(bc_escbyte, BYTE1_spec, 0), + BAND_INIT(file_name, UNSIGNED5_spec, INDEX(CONSTANT_Utf8)), + BAND_INIT(file_size_hi, UNSIGNED5_spec, 0), BAND_INIT(file_size_lo, UNSIGNED5_spec, 0), + BAND_INIT(file_modtime, DELTA5_spec, 0), BAND_INIT(file_options, UNSIGNED5_spec, 0), + // BAND_INIT(file_bits, BYTE1_spec, 0), + {0, 0}}; band *band::makeBands(unpacker *u) { - band *tmp_all_bands = U_NEW(band, BAND_LIMIT); - for (int i = 0; i < BAND_LIMIT; i++) - { - assert((byte *)&all_band_inits[i + 1] < - (byte *)all_band_inits + sizeof(all_band_inits)); - const band_init &bi = all_band_inits[i]; - band &b = tmp_all_bands[i]; - coding *defc = coding::findBySpec(bi.defc); - assert((defc == nullptr) == (bi.defc == -1)); // no garbage, please - assert(defc == nullptr || !defc->isMalloc); - b.init(u, i, defc); - if (bi.index > 0) - { - b.nullOK = ((bi.index >> 8) & 1); - b.ixTag = (bi.index & 0xFF); - } - } - return tmp_all_bands; + band *tmp_all_bands = U_NEW(band, BAND_LIMIT); + for (int i = 0; i < BAND_LIMIT; i++) + { + assert((byte *)&all_band_inits[i + 1] < + (byte *)all_band_inits + sizeof(all_band_inits)); + const band_init &bi = all_band_inits[i]; + band &b = tmp_all_bands[i]; + coding *defc = coding::findBySpec(bi.defc); + assert((defc == nullptr) == (bi.defc == -1)); // no garbage, please + assert(defc == nullptr || !defc->isMalloc); + b.init(u, i, defc); + if (bi.index > 0) + { + b.nullOK = ((bi.index >> 8) & 1); + b.ixTag = (bi.index & 0xFF); + } + } + return tmp_all_bands; } void band::initIndexes(unpacker *u) { - band *tmp_all_bands = u->all_bands; - for (int i = 0; i < BAND_LIMIT; i++) - { - band *scan = &tmp_all_bands[i]; - uint32_t tag = scan->ixTag; // Cf. #define INDEX(tag) above - if (tag != 0 && tag != CONSTANT_Literal && (tag & SUBINDEX_BIT) == 0) - { - scan->setIndex(u->cp.getIndex(tag)); - } - } + band *tmp_all_bands = u->all_bands; + for (int i = 0; i < BAND_LIMIT; i++) + { + band *scan = &tmp_all_bands[i]; + uint32_t tag = scan->ixTag; // Cf. #define INDEX(tag) above + if (tag != 0 && tag != CONSTANT_Literal && (tag & SUBINDEX_BIT) == 0) + { + scan->setIndex(u->cp.getIndex(tag)); + } + } } diff --git a/libraries/pack200/src/bands.h b/libraries/pack200/src/bands.h index a56cd7d5..66c5aec4 100644 --- a/libraries/pack200/src/bands.h +++ b/libraries/pack200/src/bands.h @@ -30,138 +30,138 @@ struct unpacker; struct band { - int bn; // band_number of this band - coding *defc; // default coding method - cpindex *ix; // CP entry mapping, if CPRefBand - byte ixTag; // 0 or 1; nullptr is coded as (nullOK?0:-1) - byte nullOK; // 0 or 1; nullptr is coded as (nullOK?0:-1) - int length; // expected # values - unpacker *u; // back pointer - - value_stream vs[2]; // source of values - coding_method cm; // method used for initial state of vs[0] - byte *rplimit; // end of band (encoded, transmitted) - - int total_memo; // cached value of getIntTotal, or -1 - int *hist0; // approximate. histogram - enum - { - HIST0_MIN = 0, - HIST0_MAX = 255 - }; // catches the usual cases - - // properties for attribute layout elements: - byte le_kind; // EK_XXX - byte le_bci; // 0,EK_BCI,EK_BCD,EK_BCO - byte le_back; // ==EF_BACK - byte le_len; // 0,1,2,4 (size in classfile), or call addr - band **le_body; // body of repl, union, call (nullptr-terminated) + int bn; // band_number of this band + coding *defc; // default coding method + cpindex *ix; // CP entry mapping, if CPRefBand + byte ixTag; // 0 or 1; nullptr is coded as (nullOK?0:-1) + byte nullOK; // 0 or 1; nullptr is coded as (nullOK?0:-1) + int length; // expected # values + unpacker *u; // back pointer + + value_stream vs[2]; // source of values + coding_method cm; // method used for initial state of vs[0] + byte *rplimit; // end of band (encoded, transmitted) + + int total_memo; // cached value of getIntTotal, or -1 + int *hist0; // approximate. histogram + enum + { + HIST0_MIN = 0, + HIST0_MAX = 255 + }; // catches the usual cases + + // properties for attribute layout elements: + byte le_kind; // EK_XXX + byte le_bci; // 0,EK_BCI,EK_BCD,EK_BCO + byte le_back; // ==EF_BACK + byte le_len; // 0,1,2,4 (size in classfile), or call addr + band **le_body; // body of repl, union, call (nullptr-terminated) // Note: EK_CASE elements use hist0 to record union tags. #define le_casetags hist0 - band &nextBand() - { - return this[1]; - } - band &prevBand() - { - return this[-1]; - } - - void init(unpacker *u_, int bn_, coding *defc_) - { - u = u_; - cm.u = u_; - bn = bn_; - defc = defc_; - } - void init(unpacker *u_, int bn_, int defcSpec) - { - init(u_, bn_, coding::findBySpec(defcSpec)); - } - void initRef(int ixTag_ = 0, bool nullOK_ = false) - { - ixTag = ixTag_; - nullOK = nullOK_; - setIndexByTag(ixTag); - } - - void expectMoreLength(int l) - { - assert(length >= 0); // able to accept a length - assert((int)l >= 0); // no overflow - assert(rplimit == nullptr); // readData not yet called - length += l; - assert(length >= l); // no overflow - } - - void setIndex(cpindex *ix_); - void setIndexByTag(byte tag); - - // Parse the band and its meta-coding header. - void readData(int expectedLength = 0); - - // Reset the band for another pass (Cf. Java Band.resetForSecondPass.) - void rewind() - { - cm.reset(&vs[0]); - } - - byte *&curRP() - { - return vs[0].rp; - } - byte *minRP() - { - return cm.vs0.rp; - } - byte *maxRP() - { - return rplimit; - } - size_t size() - { - return maxRP() - minRP(); - } - - int getByte() - { - assert(ix == nullptr); - return vs[0].getByte(); - } - int getInt() - { - assert(ix == nullptr); - return vs[0].getInt(); - } - entry *getRefN() - { - assert(ix != nullptr); - return getRefCommon(ix, true); - } - entry *getRef() - { - assert(ix != nullptr); - return getRefCommon(ix, false); - } - entry *getRefUsing(cpindex *ix2) - { - assert(ix == nullptr); - return getRefCommon(ix2, true); - } - entry *getRefCommon(cpindex *ix, bool nullOK); - int64_t getLong(band &lo_band, bool have_hi); - - static int64_t makeLong(uint32_t hi, uint32_t lo) - { - return ((uint64_t)hi << 32) + (((uint64_t)lo << 32) >> 32); - } - - int getIntTotal(); - int getIntCount(int tag); - - static band *makeBands(unpacker *u); - static void initIndexes(unpacker *u); + band &nextBand() + { + return this[1]; + } + band &prevBand() + { + return this[-1]; + } + + void init(unpacker *u_, int bn_, coding *defc_) + { + u = u_; + cm.u = u_; + bn = bn_; + defc = defc_; + } + void init(unpacker *u_, int bn_, int defcSpec) + { + init(u_, bn_, coding::findBySpec(defcSpec)); + } + void initRef(int ixTag_ = 0, bool nullOK_ = false) + { + ixTag = ixTag_; + nullOK = nullOK_; + setIndexByTag(ixTag); + } + + void expectMoreLength(int l) + { + assert(length >= 0); // able to accept a length + assert((int)l >= 0); // no overflow + assert(rplimit == nullptr); // readData not yet called + length += l; + assert(length >= l); // no overflow + } + + void setIndex(cpindex *ix_); + void setIndexByTag(byte tag); + + // Parse the band and its meta-coding header. + void readData(int expectedLength = 0); + + // Reset the band for another pass (Cf. Java Band.resetForSecondPass.) + void rewind() + { + cm.reset(&vs[0]); + } + + byte *&curRP() + { + return vs[0].rp; + } + byte *minRP() + { + return cm.vs0.rp; + } + byte *maxRP() + { + return rplimit; + } + size_t size() + { + return maxRP() - minRP(); + } + + int getByte() + { + assert(ix == nullptr); + return vs[0].getByte(); + } + int getInt() + { + assert(ix == nullptr); + return vs[0].getInt(); + } + entry *getRefN() + { + assert(ix != nullptr); + return getRefCommon(ix, true); + } + entry *getRef() + { + assert(ix != nullptr); + return getRefCommon(ix, false); + } + entry *getRefUsing(cpindex *ix2) + { + assert(ix == nullptr); + return getRefCommon(ix2, true); + } + entry *getRefCommon(cpindex *ix, bool nullOK); + int64_t getLong(band &lo_band, bool have_hi); + + static int64_t makeLong(uint32_t hi, uint32_t lo) + { + return ((uint64_t)hi << 32) + (((uint64_t)lo << 32) >> 32); + } + + int getIntTotal(); + int getIntCount(int tag); + + static band *makeBands(unpacker *u); + static void initIndexes(unpacker *u); }; extern band all_bands[]; @@ -173,179 +173,179 @@ extern band all_bands[]; // Band schema: enum band_number { - // e_archive_magic, - // e_archive_header, - // e_band_headers, - - // constant pool contents - e_cp_Utf8_prefix, - e_cp_Utf8_suffix, - e_cp_Utf8_chars, - e_cp_Utf8_big_suffix, - e_cp_Utf8_big_chars, - e_cp_Int, - e_cp_Float, - e_cp_Long_hi, - e_cp_Long_lo, - e_cp_Double_hi, - e_cp_Double_lo, - e_cp_String, - e_cp_Class, - e_cp_Signature_form, - e_cp_Signature_classes, - e_cp_Descr_name, - e_cp_Descr_type, - e_cp_Field_class, - e_cp_Field_desc, - e_cp_Method_class, - e_cp_Method_desc, - e_cp_Imethod_class, - e_cp_Imethod_desc, - - // bands which define transmission of attributes - e_attr_definition_headers, - e_attr_definition_name, - e_attr_definition_layout, - - // band for hardwired InnerClasses attribute (shared across the package) - e_ic_this_class, - e_ic_flags, - // These bands contain data only where flags sets ACC_IC_LONG_FORM: - e_ic_outer_class, - e_ic_name, - - // bands for carrying class schema information: - e_class_this, - e_class_super, - e_class_interface_count, - e_class_interface, - - // bands for class members - e_class_field_count, - e_class_method_count, - e_field_descr, - e_field_flags_hi, - e_field_flags_lo, - e_field_attr_count, - e_field_attr_indexes, - e_field_attr_calls, - e_field_ConstantValue_KQ, - e_field_Signature_RS, - e_field_metadata_bands, - e_field_attr_bands, - e_method_descr, - e_method_flags_hi, - e_method_flags_lo, - e_method_attr_count, - e_method_attr_indexes, - e_method_attr_calls, - e_method_Exceptions_N, - e_method_Exceptions_RC, - e_method_Signature_RS, - e_method_metadata_bands, - e_method_attr_bands, - e_class_flags_hi, - e_class_flags_lo, - e_class_attr_count, - e_class_attr_indexes, - e_class_attr_calls, - e_class_SourceFile_RUN, - e_class_EnclosingMethod_RC, - e_class_EnclosingMethod_RDN, - e_class_Signature_RS, - e_class_metadata_bands, - e_class_InnerClasses_N, - e_class_InnerClasses_RC, - e_class_InnerClasses_F, - e_class_InnerClasses_outer_RCN, - e_class_InnerClasses_name_RUN, - e_class_ClassFile_version_minor_H, - e_class_ClassFile_version_major_H, - e_class_attr_bands, - e_code_headers, - e_code_max_stack, - e_code_max_na_locals, - e_code_handler_count, - e_code_handler_start_P, - e_code_handler_end_PO, - e_code_handler_catch_PO, - e_code_handler_class_RCN, - - // code attributes - e_code_flags_hi, - e_code_flags_lo, - e_code_attr_count, - e_code_attr_indexes, - e_code_attr_calls, - e_code_StackMapTable_N, - e_code_StackMapTable_frame_T, - e_code_StackMapTable_local_N, - e_code_StackMapTable_stack_N, - e_code_StackMapTable_offset, - e_code_StackMapTable_T, - e_code_StackMapTable_RC, - e_code_StackMapTable_P, - e_code_LineNumberTable_N, - e_code_LineNumberTable_bci_P, - e_code_LineNumberTable_line, - e_code_LocalVariableTable_N, - e_code_LocalVariableTable_bci_P, - e_code_LocalVariableTable_span_O, - e_code_LocalVariableTable_name_RU, - e_code_LocalVariableTable_type_RS, - e_code_LocalVariableTable_slot, - e_code_LocalVariableTypeTable_N, - e_code_LocalVariableTypeTable_bci_P, - e_code_LocalVariableTypeTable_span_O, - e_code_LocalVariableTypeTable_name_RU, - e_code_LocalVariableTypeTable_type_RS, - e_code_LocalVariableTypeTable_slot, - e_code_attr_bands, - - // bands for bytecodes - e_bc_codes, - // remaining bands provide typed opcode fields required by the bc_codes - e_bc_case_count, - e_bc_case_value, - e_bc_byte, - e_bc_short, - e_bc_local, - e_bc_label, - - // ldc* operands: - e_bc_intref, - e_bc_floatref, - e_bc_longref, - e_bc_doubleref, - e_bc_stringref, - e_bc_classref, - e_bc_fieldref, - e_bc_methodref, - e_bc_imethodref, - - // _self_linker_op family - e_bc_thisfield, - e_bc_superfield, - e_bc_thismethod, - e_bc_supermethod, - - // bc_invokeinit family: - e_bc_initref, - - // bytecode escape sequences - e_bc_escref, - e_bc_escrefsize, - e_bc_escsize, - e_bc_escbyte, - - // file attributes and contents - e_file_name, - e_file_size_hi, - e_file_size_lo, - e_file_modtime, - e_file_options, - // e_file_bits, // handled specially as an appendix - BAND_LIMIT + // e_archive_magic, + // e_archive_header, + // e_band_headers, + + // constant pool contents + e_cp_Utf8_prefix, + e_cp_Utf8_suffix, + e_cp_Utf8_chars, + e_cp_Utf8_big_suffix, + e_cp_Utf8_big_chars, + e_cp_Int, + e_cp_Float, + e_cp_Long_hi, + e_cp_Long_lo, + e_cp_Double_hi, + e_cp_Double_lo, + e_cp_String, + e_cp_Class, + e_cp_Signature_form, + e_cp_Signature_classes, + e_cp_Descr_name, + e_cp_Descr_type, + e_cp_Field_class, + e_cp_Field_desc, + e_cp_Method_class, + e_cp_Method_desc, + e_cp_Imethod_class, + e_cp_Imethod_desc, + + // bands which define transmission of attributes + e_attr_definition_headers, + e_attr_definition_name, + e_attr_definition_layout, + + // band for hardwired InnerClasses attribute (shared across the package) + e_ic_this_class, + e_ic_flags, + // These bands contain data only where flags sets ACC_IC_LONG_FORM: + e_ic_outer_class, + e_ic_name, + + // bands for carrying class schema information: + e_class_this, + e_class_super, + e_class_interface_count, + e_class_interface, + + // bands for class members + e_class_field_count, + e_class_method_count, + e_field_descr, + e_field_flags_hi, + e_field_flags_lo, + e_field_attr_count, + e_field_attr_indexes, + e_field_attr_calls, + e_field_ConstantValue_KQ, + e_field_Signature_RS, + e_field_metadata_bands, + e_field_attr_bands, + e_method_descr, + e_method_flags_hi, + e_method_flags_lo, + e_method_attr_count, + e_method_attr_indexes, + e_method_attr_calls, + e_method_Exceptions_N, + e_method_Exceptions_RC, + e_method_Signature_RS, + e_method_metadata_bands, + e_method_attr_bands, + e_class_flags_hi, + e_class_flags_lo, + e_class_attr_count, + e_class_attr_indexes, + e_class_attr_calls, + e_class_SourceFile_RUN, + e_class_EnclosingMethod_RC, + e_class_EnclosingMethod_RDN, + e_class_Signature_RS, + e_class_metadata_bands, + e_class_InnerClasses_N, + e_class_InnerClasses_RC, + e_class_InnerClasses_F, + e_class_InnerClasses_outer_RCN, + e_class_InnerClasses_name_RUN, + e_class_ClassFile_version_minor_H, + e_class_ClassFile_version_major_H, + e_class_attr_bands, + e_code_headers, + e_code_max_stack, + e_code_max_na_locals, + e_code_handler_count, + e_code_handler_start_P, + e_code_handler_end_PO, + e_code_handler_catch_PO, + e_code_handler_class_RCN, + + // code attributes + e_code_flags_hi, + e_code_flags_lo, + e_code_attr_count, + e_code_attr_indexes, + e_code_attr_calls, + e_code_StackMapTable_N, + e_code_StackMapTable_frame_T, + e_code_StackMapTable_local_N, + e_code_StackMapTable_stack_N, + e_code_StackMapTable_offset, + e_code_StackMapTable_T, + e_code_StackMapTable_RC, + e_code_StackMapTable_P, + e_code_LineNumberTable_N, + e_code_LineNumberTable_bci_P, + e_code_LineNumberTable_line, + e_code_LocalVariableTable_N, + e_code_LocalVariableTable_bci_P, + e_code_LocalVariableTable_span_O, + e_code_LocalVariableTable_name_RU, + e_code_LocalVariableTable_type_RS, + e_code_LocalVariableTable_slot, + e_code_LocalVariableTypeTable_N, + e_code_LocalVariableTypeTable_bci_P, + e_code_LocalVariableTypeTable_span_O, + e_code_LocalVariableTypeTable_name_RU, + e_code_LocalVariableTypeTable_type_RS, + e_code_LocalVariableTypeTable_slot, + e_code_attr_bands, + + // bands for bytecodes + e_bc_codes, + // remaining bands provide typed opcode fields required by the bc_codes + e_bc_case_count, + e_bc_case_value, + e_bc_byte, + e_bc_short, + e_bc_local, + e_bc_label, + + // ldc* operands: + e_bc_intref, + e_bc_floatref, + e_bc_longref, + e_bc_doubleref, + e_bc_stringref, + e_bc_classref, + e_bc_fieldref, + e_bc_methodref, + e_bc_imethodref, + + // _self_linker_op family + e_bc_thisfield, + e_bc_superfield, + e_bc_thismethod, + e_bc_supermethod, + + // bc_invokeinit family: + e_bc_initref, + + // bytecode escape sequences + e_bc_escref, + e_bc_escrefsize, + e_bc_escsize, + e_bc_escbyte, + + // file attributes and contents + e_file_name, + e_file_size_hi, + e_file_size_lo, + e_file_modtime, + e_file_options, + // e_file_bits, // handled specially as an appendix + BAND_LIMIT }; // Symbolic names for bands, as if in a giant global struct: diff --git a/libraries/pack200/src/bytes.cpp b/libraries/pack200/src/bytes.cpp index d3808afa..767fe0a5 100644 --- a/libraries/pack200/src/bytes.cpp +++ b/libraries/pack200/src/bytes.cpp @@ -36,182 +36,182 @@ static byte dummy[1 << 10]; bool bytes::inBounds(const void *p) { - return p >= ptr && p < limit(); + return p >= ptr && p < limit(); } void bytes::malloc(size_t len_) { - len = len_; - ptr = NEW(byte, add_size(len_, 1)); // add trailing zero byte always - if (ptr == nullptr) - { - // set ptr to some victim memory, to ease escape - set(dummy, sizeof(dummy) - 1); - unpack_abort(ERROR_ENOMEM); - } + len = len_; + ptr = NEW(byte, add_size(len_, 1)); // add trailing zero byte always + if (ptr == nullptr) + { + // set ptr to some victim memory, to ease escape + set(dummy, sizeof(dummy) - 1); + unpack_abort(ERROR_ENOMEM); + } } void bytes::realloc(size_t len_) { - if (len == len_) - return; // nothing to do - if (ptr == dummy) - return; // escaping from an error - if (ptr == nullptr) - { - malloc(len_); - return; - } - byte *oldptr = ptr; - ptr = (len_ >= PSIZE_MAX) ? nullptr : (byte *)::realloc(ptr, add_size(len_, 1)); - if (ptr != nullptr) - { - if (len < len_) - memset(ptr + len, 0, len_ - len); - ptr[len_] = 0; - len = len_; - } - else - { - ptr = oldptr; // ease our escape - unpack_abort(ERROR_ENOMEM); - } + if (len == len_) + return; // nothing to do + if (ptr == dummy) + return; // escaping from an error + if (ptr == nullptr) + { + malloc(len_); + return; + } + byte *oldptr = ptr; + ptr = (len_ >= PSIZE_MAX) ? nullptr : (byte *)::realloc(ptr, add_size(len_, 1)); + if (ptr != nullptr) + { + if (len < len_) + memset(ptr + len, 0, len_ - len); + ptr[len_] = 0; + len = len_; + } + else + { + ptr = oldptr; // ease our escape + unpack_abort(ERROR_ENOMEM); + } } void bytes::free() { - if (ptr == dummy) - return; // escaping from an error - if (ptr != nullptr) - { - ::free(ptr); - } - len = 0; - ptr = 0; + if (ptr == dummy) + return; // escaping from an error + if (ptr != nullptr) + { + ::free(ptr); + } + len = 0; + ptr = 0; } int bytes::indexOf(byte c) { - byte *p = (byte *)memchr(ptr, c, len); - return (p == 0) ? -1 : (int)(p - ptr); + byte *p = (byte *)memchr(ptr, c, len); + return (p == 0) ? -1 : (int)(p - ptr); } byte *bytes::writeTo(byte *bp) { - memcpy(bp, ptr, len); - return bp + len; + memcpy(bp, ptr, len); + return bp + len; } int bytes::compareTo(bytes &other) { - size_t l1 = len; - size_t l2 = other.len; - int cmp = memcmp(ptr, other.ptr, (l1 < l2) ? l1 : l2); - if (cmp != 0) - return cmp; - return (l1 < l2) ? -1 : (l1 > l2) ? 1 : 0; + size_t l1 = len; + size_t l2 = other.len; + int cmp = memcmp(ptr, other.ptr, (l1 < l2) ? l1 : l2); + if (cmp != 0) + return cmp; + return (l1 < l2) ? -1 : (l1 > l2) ? 1 : 0; } void bytes::saveFrom(const void *ptr_, size_t len_) { - malloc(len_); - // Save as much as possible. - if (len_ > len) - { - assert(ptr == dummy); // error recovery - len_ = len; - } - copyFrom(ptr_, len_); + malloc(len_); + // Save as much as possible. + if (len_ > len) + { + assert(ptr == dummy); // error recovery + len_ = len; + } + copyFrom(ptr_, len_); } //#TODO: Need to fix for exception handling void bytes::copyFrom(const void *ptr_, size_t len_, size_t offset) { - assert(len_ == 0 || inBounds(ptr + offset)); - assert(len_ == 0 || inBounds(ptr + offset + len_ - 1)); - memcpy(ptr + offset, ptr_, len_); + assert(len_ == 0 || inBounds(ptr + offset)); + assert(len_ == 0 || inBounds(ptr + offset + len_ - 1)); + memcpy(ptr + offset, ptr_, len_); } // Make sure there are 'o' bytes beyond the fill pointer, // advance the fill pointer, and return the old fill pointer. byte *fillbytes::grow(size_t s) { - size_t nlen = add_size(b.len, s); - if (nlen <= allocated) - { - b.len = nlen; - return limit() - s; - } - size_t maxlen = nlen; - if (maxlen < 128) - maxlen = 128; - if (maxlen < allocated * 2) - maxlen = allocated * 2; - if (allocated == 0) - { - // Initial buffer was not malloced. Do not reallocate it. - bytes old = b; - b.malloc(maxlen); - if (b.len == maxlen) - old.writeTo(b.ptr); - } - else - { - b.realloc(maxlen); - } - allocated = b.len; - if (allocated != maxlen) - { - b.len = nlen - s; // back up - return dummy; // scribble during error recov. - } - // after realloc, recompute pointers - b.len = nlen; - assert(b.len <= allocated); - return limit() - s; + size_t nlen = add_size(b.len, s); + if (nlen <= allocated) + { + b.len = nlen; + return limit() - s; + } + size_t maxlen = nlen; + if (maxlen < 128) + maxlen = 128; + if (maxlen < allocated * 2) + maxlen = allocated * 2; + if (allocated == 0) + { + // Initial buffer was not malloced. Do not reallocate it. + bytes old = b; + b.malloc(maxlen); + if (b.len == maxlen) + old.writeTo(b.ptr); + } + else + { + b.realloc(maxlen); + } + allocated = b.len; + if (allocated != maxlen) + { + b.len = nlen - s; // back up + return dummy; // scribble during error recov. + } + // after realloc, recompute pointers + b.len = nlen; + assert(b.len <= allocated); + return limit() - s; } void fillbytes::ensureSize(size_t s) { - if (allocated >= s) - return; - size_t len0 = b.len; - grow(s - size()); - b.len = len0; // put it back + if (allocated >= s) + return; + size_t len0 = b.len; + grow(s - size()); + b.len = len0; // put it back } int ptrlist::indexOf(const void *x) { - int len = length(); - for (int i = 0; i < len; i++) - { - if (get(i) == x) - return i; - } - return -1; + int len = length(); + for (int i = 0; i < len; i++) + { + if (get(i) == x) + return i; + } + return -1; } void ptrlist::freeAll() { - int len = length(); - for (int i = 0; i < len; i++) - { - void *p = (void *)get(i); - if (p != nullptr) - { - ::free(p); - } - } - free(); + int len = length(); + for (int i = 0; i < len; i++) + { + void *p = (void *)get(i); + if (p != nullptr) + { + ::free(p); + } + } + free(); } int intlist::indexOf(int x) { - int len = length(); - for (int i = 0; i < len; i++) - { - if (get(i) == x) - return i; - } - return -1; + int len = length(); + for (int i = 0; i < len; i++) + { + if (get(i) == x) + return i; + } + return -1; } diff --git a/libraries/pack200/src/bytes.h b/libraries/pack200/src/bytes.h index b116efda..2ce1f7f4 100644 --- a/libraries/pack200/src/bytes.h +++ b/libraries/pack200/src/bytes.h @@ -27,225 +27,225 @@ struct bytes { - int8_t *ptr; - size_t len; - int8_t *limit() - { - return ptr + len; - } + int8_t *ptr; + size_t len; + int8_t *limit() + { + return ptr + len; + } - void set(int8_t *ptr_, size_t len_) - { - ptr = ptr_; - len = len_; - } - void set(const char *str) - { - ptr = (int8_t *)str; - len = strlen(str); - } - bool inBounds(const void *p); // p in [ptr, limit) - void malloc(size_t len_); - void realloc(size_t len_); - void free(); - void copyFrom(const void *ptr_, size_t len_, size_t offset = 0); - void saveFrom(const void *ptr_, size_t len_); - void saveFrom(const char *str) - { - saveFrom(str, strlen(str)); - } - void copyFrom(bytes &other, size_t offset = 0) - { - copyFrom(other.ptr, other.len, offset); - } - void saveFrom(bytes &other) - { - saveFrom(other.ptr, other.len); - } - void clear(int fill_byte = 0) - { - memset(ptr, fill_byte, len); - } - int8_t *writeTo(int8_t *bp); - bool equals(bytes &other) - { - return 0 == compareTo(other); - } - int compareTo(bytes &other); - bool contains(int8_t c) - { - return indexOf(c) >= 0; - } - int indexOf(int8_t c); - // substrings: - static bytes of(int8_t *ptr, size_t len) - { - bytes res; - res.set(ptr, len); - return res; - } - bytes slice(size_t beg, size_t end) - { - bytes res; - res.ptr = ptr + beg; - res.len = end - beg; - assert(res.len == 0 ||(inBounds(res.ptr) && inBounds(res.limit() - 1))); - return res; - } - // building C strings inside byte buffers: - bytes &strcat(const char *str) - { - ::strcat((char *)ptr, str); - return *this; - } - bytes &strcat(bytes &other) - { - ::strncat((char *)ptr, (char *)other.ptr, other.len); - return *this; - } - char *strval() - { - assert(strlen((char *)ptr) == len); - return (char *)ptr; - } + void set(int8_t *ptr_, size_t len_) + { + ptr = ptr_; + len = len_; + } + void set(const char *str) + { + ptr = (int8_t *)str; + len = strlen(str); + } + bool inBounds(const void *p); // p in [ptr, limit) + void malloc(size_t len_); + void realloc(size_t len_); + void free(); + void copyFrom(const void *ptr_, size_t len_, size_t offset = 0); + void saveFrom(const void *ptr_, size_t len_); + void saveFrom(const char *str) + { + saveFrom(str, strlen(str)); + } + void copyFrom(bytes &other, size_t offset = 0) + { + copyFrom(other.ptr, other.len, offset); + } + void saveFrom(bytes &other) + { + saveFrom(other.ptr, other.len); + } + void clear(int fill_byte = 0) + { + memset(ptr, fill_byte, len); + } + int8_t *writeTo(int8_t *bp); + bool equals(bytes &other) + { + return 0 == compareTo(other); + } + int compareTo(bytes &other); + bool contains(int8_t c) + { + return indexOf(c) >= 0; + } + int indexOf(int8_t c); + // substrings: + static bytes of(int8_t *ptr, size_t len) + { + bytes res; + res.set(ptr, len); + return res; + } + bytes slice(size_t beg, size_t end) + { + bytes res; + res.ptr = ptr + beg; + res.len = end - beg; + assert(res.len == 0 ||(inBounds(res.ptr) && inBounds(res.limit() - 1))); + return res; + } + // building C strings inside byte buffers: + bytes &strcat(const char *str) + { + ::strcat((char *)ptr, str); + return *this; + } + bytes &strcat(bytes &other) + { + ::strncat((char *)ptr, (char *)other.ptr, other.len); + return *this; + } + char *strval() + { + assert(strlen((char *)ptr) == len); + return (char *)ptr; + } }; #define BYTES_OF(var) (bytes::of((int8_t *)&(var), sizeof(var))) struct fillbytes { - bytes b; - size_t allocated; + bytes b; + size_t allocated; - int8_t *base() - { - return b.ptr; - } - size_t size() - { - return b.len; - } - int8_t *limit() - { - return b.limit(); - } // logical limit - void setLimit(int8_t *lp) - { - assert(isAllocated(lp)); - b.len = lp - b.ptr; - } - int8_t *end() - { - return b.ptr + allocated; - } // physical limit - int8_t *loc(size_t o) - { - assert(o < b.len); - return b.ptr + o; - } - void init() - { - allocated = 0; - b.set(nullptr, 0); - } - void init(size_t s) - { - init(); - ensureSize(s); - } - void free() - { - if (allocated != 0) - b.free(); - allocated = 0; - } - void empty() - { - b.len = 0; - } - int8_t *grow(size_t s); // grow so that limit() += s - int getByte(uint32_t i) - { - return *loc(i) & 0xFF; - } - void addByte(int8_t x) - { - *grow(1) = x; - } - void ensureSize(size_t s); // make sure allocated >= s - void trimToSize() - { - if (allocated > size()) - b.realloc(allocated = size()); - } - bool canAppend(size_t s) - { - return allocated > b.len + s; - } - bool isAllocated(int8_t *p) - { - return p >= base() && p <= end(); - } // asserts - void set(bytes &src) - { - set(src.ptr, src.len); - } + int8_t *base() + { + return b.ptr; + } + size_t size() + { + return b.len; + } + int8_t *limit() + { + return b.limit(); + } // logical limit + void setLimit(int8_t *lp) + { + assert(isAllocated(lp)); + b.len = lp - b.ptr; + } + int8_t *end() + { + return b.ptr + allocated; + } // physical limit + int8_t *loc(size_t o) + { + assert(o < b.len); + return b.ptr + o; + } + void init() + { + allocated = 0; + b.set(nullptr, 0); + } + void init(size_t s) + { + init(); + ensureSize(s); + } + void free() + { + if (allocated != 0) + b.free(); + allocated = 0; + } + void empty() + { + b.len = 0; + } + int8_t *grow(size_t s); // grow so that limit() += s + int getByte(uint32_t i) + { + return *loc(i) & 0xFF; + } + void addByte(int8_t x) + { + *grow(1) = x; + } + void ensureSize(size_t s); // make sure allocated >= s + void trimToSize() + { + if (allocated > size()) + b.realloc(allocated = size()); + } + bool canAppend(size_t s) + { + return allocated > b.len + s; + } + bool isAllocated(int8_t *p) + { + return p >= base() && p <= end(); + } // asserts + void set(bytes &src) + { + set(src.ptr, src.len); + } - void set(int8_t *ptr, size_t len) - { - b.set(ptr, len); - allocated = 0; // mark as not reallocatable - } + void set(int8_t *ptr, size_t len) + { + b.set(ptr, len); + allocated = 0; // mark as not reallocatable + } - // block operations on resizing byte buffer: - fillbytes &append(const void *ptr_, size_t len_) - { - memcpy(grow(len_), ptr_, len_); - return (*this); - } - fillbytes &append(bytes &other) - { - return append(other.ptr, other.len); - } - fillbytes &append(const char *str) - { - return append(str, strlen(str)); - } + // block operations on resizing byte buffer: + fillbytes &append(const void *ptr_, size_t len_) + { + memcpy(grow(len_), ptr_, len_); + return (*this); + } + fillbytes &append(bytes &other) + { + return append(other.ptr, other.len); + } + fillbytes &append(const char *str) + { + return append(str, strlen(str)); + } }; struct ptrlist : fillbytes { - typedef const void *cvptr; - int length() - { - return (int)(size() / sizeof(cvptr)); - } - cvptr *base() - { - return (cvptr *)fillbytes::base(); - } - cvptr &get(int i) - { - return *(cvptr *)loc(i * sizeof(cvptr)); - } - cvptr *limit() - { - return (cvptr *)fillbytes::limit(); - } - void add(cvptr x) - { - *(cvptr *)grow(sizeof(x)) = x; - } - void popTo(int l) - { - assert(l <= length()); - b.len = l * sizeof(cvptr); - } - int indexOf(cvptr x); - bool contains(cvptr x) - { - return indexOf(x) >= 0; - } - void freeAll(); // frees every ptr on the list, plus the list itself + typedef const void *cvptr; + int length() + { + return (int)(size() / sizeof(cvptr)); + } + cvptr *base() + { + return (cvptr *)fillbytes::base(); + } + cvptr &get(int i) + { + return *(cvptr *)loc(i * sizeof(cvptr)); + } + cvptr *limit() + { + return (cvptr *)fillbytes::limit(); + } + void add(cvptr x) + { + *(cvptr *)grow(sizeof(x)) = x; + } + void popTo(int l) + { + assert(l <= length()); + b.len = l * sizeof(cvptr); + } + int indexOf(cvptr x); + bool contains(cvptr x) + { + return indexOf(x) >= 0; + } + void freeAll(); // frees every ptr on the list, plus the list itself }; // Use a macro rather than mess with subtle mismatches // between member and non-member function pointers. @@ -253,34 +253,34 @@ struct ptrlist : fillbytes struct intlist : fillbytes { - int length() - { - return (int)(size() / sizeof(int)); - } - int *base() - { - return (int *)fillbytes::base(); - } - int &get(int i) - { - return *(int *)loc(i * sizeof(int)); - } - int *limit() - { - return (int *)fillbytes::limit(); - } - void add(int x) - { - *(int *)grow(sizeof(x)) = x; - } - void popTo(int l) - { - assert(l <= length()); - b.len = l * sizeof(int); - } - int indexOf(int x); - bool contains(int x) - { - return indexOf(x) >= 0; - } + int length() + { + return (int)(size() / sizeof(int)); + } + int *base() + { + return (int *)fillbytes::base(); + } + int &get(int i) + { + return *(int *)loc(i * sizeof(int)); + } + int *limit() + { + return (int *)fillbytes::limit(); + } + void add(int x) + { + *(int *)grow(sizeof(x)) = x; + } + void popTo(int l) + { + assert(l <= length()); + b.len = l * sizeof(int); + } + int indexOf(int x); + bool contains(int x) + { + return indexOf(x) >= 0; + } }; diff --git a/libraries/pack200/src/coding.cpp b/libraries/pack200/src/coding.cpp index 6bd17a3c..8e872013 100644 --- a/libraries/pack200/src/coding.cpp +++ b/libraries/pack200/src/coding.cpp @@ -48,12 +48,12 @@ extern coding basic_codings[]; #pragma GCC diagnostic ignored "-Wunused-variable" #define CODING_PRIVATE(spec) \ - int spec_ = spec; \ - int B = CODING_B(spec_); \ - int H = CODING_H(spec_); \ - int L = 256 - H; \ - int S = CODING_S(spec_); \ - int D = CODING_D(spec_) + int spec_ = spec; \ + int B = CODING_B(spec_); \ + int H = CODING_H(spec_); \ + int L = 256 - H; \ + int S = CODING_S(spec_); \ + int D = CODING_D(spec_) #define IS_NEG_CODE(S, codeVal) ((((int)(codeVal) + 1) & ((1 << S) - 1)) == 0) @@ -61,568 +61,568 @@ extern coding basic_codings[]; static int decode_sign(int S, uint32_t ux) { // == Coding.decodeSign32 - assert(S > 0); - uint32_t sigbits = (ux >> S); - if (IS_NEG_CODE(S, ux)) - return (int)(~sigbits); - else - return (int)(ux - sigbits); - // Note that (int)(ux-sigbits) can be negative, if ux is large enough. + assert(S > 0); + uint32_t sigbits = (ux >> S); + if (IS_NEG_CODE(S, ux)) + return (int)(~sigbits); + else + return (int)(ux - sigbits); + // Note that (int)(ux-sigbits) can be negative, if ux is large enough. } coding *coding::init() { - if (umax > 0) - return this; // already done - assert(spec != 0); // sanity - - // fill in derived fields - CODING_PRIVATE(spec); - - // Return nullptr if 'arb(BHSD)' parameter constraints are not met: - if (B < 1 || B > B_MAX) - return nullptr; - if (H < 1 || H > 256) - return nullptr; - if (S < 0 || S > 2) - return nullptr; - if (D < 0 || D > 1) - return nullptr; - if (B == 1 && H != 256) - return nullptr; // 1-byte coding must be fixed-size - if (B >= 5 && H == 256) - return nullptr; // no 5-byte fixed-size coding - - // first compute the range of the coding, in 64 bits - int64_t range = 0; - { - int64_t H_i = 1; - for (int i = 0; i < B; i++) - { - range += H_i; - H_i *= H; - } - range *= L; - range += H_i; - } - assert(range > 0); // no useless codings, please - - int this_umax; - - // now, compute min and max - if (range >= ((int64_t)1 << 32)) - { - this_umax = INT_MAX_VALUE; - this->umin = INT_MIN_VALUE; - this->max = INT_MAX_VALUE; - this->min = INT_MIN_VALUE; - } - else - { - this_umax = (range > INT_MAX_VALUE) ? INT_MAX_VALUE : (int)range - 1; - this->max = this_umax; - this->min = this->umin = 0; - if (S != 0 && range != 0) - { - int64_t maxPosCode = range - 1; - int64_t maxNegCode = range - 1; - while (IS_NEG_CODE(S, maxPosCode)) - --maxPosCode; - while (!IS_NEG_CODE(S, maxNegCode)) - --maxNegCode; - int maxPos = decode_sign(S, (uint32_t)maxPosCode); - if (maxPos < 0) - this->max = INT_MAX_VALUE; // 32-bit wraparound - else - this->max = maxPos; - if (maxNegCode < 0) - this->min = 0; // No negative codings at all. - else - this->min = decode_sign(S, (uint32_t)maxNegCode); - } - } - - assert(!(isFullRange | isSigned | isSubrange)); // init - if (min < 0) - this->isSigned = true; - if (max < INT_MAX_VALUE && range <= INT_MAX_VALUE) - this->isSubrange = true; - if (max == INT_MAX_VALUE && min == INT_MIN_VALUE) - this->isFullRange = true; - - // do this last, to reduce MT exposure (should have a membar too) - this->umax = this_umax; - - return this; + if (umax > 0) + return this; // already done + assert(spec != 0); // sanity + + // fill in derived fields + CODING_PRIVATE(spec); + + // Return nullptr if 'arb(BHSD)' parameter constraints are not met: + if (B < 1 || B > B_MAX) + return nullptr; + if (H < 1 || H > 256) + return nullptr; + if (S < 0 || S > 2) + return nullptr; + if (D < 0 || D > 1) + return nullptr; + if (B == 1 && H != 256) + return nullptr; // 1-byte coding must be fixed-size + if (B >= 5 && H == 256) + return nullptr; // no 5-byte fixed-size coding + + // first compute the range of the coding, in 64 bits + int64_t range = 0; + { + int64_t H_i = 1; + for (int i = 0; i < B; i++) + { + range += H_i; + H_i *= H; + } + range *= L; + range += H_i; + } + assert(range > 0); // no useless codings, please + + int this_umax; + + // now, compute min and max + if (range >= ((int64_t)1 << 32)) + { + this_umax = INT_MAX_VALUE; + this->umin = INT_MIN_VALUE; + this->max = INT_MAX_VALUE; + this->min = INT_MIN_VALUE; + } + else + { + this_umax = (range > INT_MAX_VALUE) ? INT_MAX_VALUE : (int)range - 1; + this->max = this_umax; + this->min = this->umin = 0; + if (S != 0 && range != 0) + { + int64_t maxPosCode = range - 1; + int64_t maxNegCode = range - 1; + while (IS_NEG_CODE(S, maxPosCode)) + --maxPosCode; + while (!IS_NEG_CODE(S, maxNegCode)) + --maxNegCode; + int maxPos = decode_sign(S, (uint32_t)maxPosCode); + if (maxPos < 0) + this->max = INT_MAX_VALUE; // 32-bit wraparound + else + this->max = maxPos; + if (maxNegCode < 0) + this->min = 0; // No negative codings at all. + else + this->min = decode_sign(S, (uint32_t)maxNegCode); + } + } + + assert(!(isFullRange | isSigned | isSubrange)); // init + if (min < 0) + this->isSigned = true; + if (max < INT_MAX_VALUE && range <= INT_MAX_VALUE) + this->isSubrange = true; + if (max == INT_MAX_VALUE && min == INT_MIN_VALUE) + this->isFullRange = true; + + // do this last, to reduce MT exposure (should have a membar too) + this->umax = this_umax; + + return this; } coding *coding::findBySpec(int spec) { - for (coding *scan = &basic_codings[0];; scan++) - { - if (scan->spec == spec) - return scan->init(); - if (scan->spec == 0) - break; - } - coding *ptr = NEW(coding, 1); - if (!ptr) - return nullptr; - coding *c = ptr->initFrom(spec); - if (c == nullptr) - { - ::free(ptr); - } - else - // else caller should free it... - c->isMalloc = true; - return c; + for (coding *scan = &basic_codings[0];; scan++) + { + if (scan->spec == spec) + return scan->init(); + if (scan->spec == 0) + break; + } + coding *ptr = NEW(coding, 1); + if (!ptr) + return nullptr; + coding *c = ptr->initFrom(spec); + if (c == nullptr) + { + ::free(ptr); + } + else + // else caller should free it... + c->isMalloc = true; + return c; } coding *coding::findBySpec(int B, int H, int S, int D) { - if (B < 1 || B > B_MAX) - return nullptr; - if (H < 1 || H > 256) - return nullptr; - if (S < 0 || S > 2) - return nullptr; - if (D < 0 || D > 1) - return nullptr; - return findBySpec(CODING_SPEC(B, H, S, D)); + if (B < 1 || B > B_MAX) + return nullptr; + if (H < 1 || H > 256) + return nullptr; + if (S < 0 || S > 2) + return nullptr; + if (D < 0 || D > 1) + return nullptr; + return findBySpec(CODING_SPEC(B, H, S, D)); } void coding::free() { - if (isMalloc) - { - ::free(this); - } + if (isMalloc) + { + ::free(this); + } } void coding_method::reset(value_stream *state) { - assert(state->rp == state->rplimit); // not in mid-stream, please - // assert(this == vs0.cm); - state[0] = vs0; - if (uValues != nullptr) - { - uValues->reset(state->helper()); - } + assert(state->rp == state->rplimit); // not in mid-stream, please + // assert(this == vs0.cm); + state[0] = vs0; + if (uValues != nullptr) + { + uValues->reset(state->helper()); + } } uint32_t coding::parse(byte *&rp, int B, int H) { - int L = 256 - H; - byte *ptr = rp; - // hand peel the i==0 part of the loop: - uint32_t b_i = *ptr++ & 0xFF; - if (B == 1 || b_i < (uint32_t)L) - { - rp = ptr; - return b_i; - } - uint32_t sum = b_i; - uint32_t H_i = H; - assert(B <= B_MAX); - for (int i = 2; i <= B_MAX; i++) - { // easy for compilers to unroll if desired - b_i = *ptr++ & 0xFF; - sum += b_i * H_i; - if (i == B || b_i < (uint32_t)L) - { - rp = ptr; - return sum; - } - H_i *= H; - } - assert(false); - return 0; + int L = 256 - H; + byte *ptr = rp; + // hand peel the i==0 part of the loop: + uint32_t b_i = *ptr++ & 0xFF; + if (B == 1 || b_i < (uint32_t)L) + { + rp = ptr; + return b_i; + } + uint32_t sum = b_i; + uint32_t H_i = H; + assert(B <= B_MAX); + for (int i = 2; i <= B_MAX; i++) + { // easy for compilers to unroll if desired + b_i = *ptr++ & 0xFF; + sum += b_i * H_i; + if (i == B || b_i < (uint32_t)L) + { + rp = ptr; + return sum; + } + H_i *= H; + } + assert(false); + return 0; } uint32_t coding::parse_lgH(byte *&rp, int B, int H, int lgH) { - assert(H == (1 << lgH)); - int L = 256 - (1 << lgH); - byte *ptr = rp; - // hand peel the i==0 part of the loop: - uint32_t b_i = *ptr++ & 0xFF; - if (B == 1 || b_i < (uint32_t)L) - { - rp = ptr; - return b_i; - } - uint32_t sum = b_i; - uint32_t lg_H_i = lgH; - assert(B <= B_MAX); - for (int i = 2; i <= B_MAX; i++) - { // easy for compilers to unroll if desired - b_i = *ptr++ & 0xFF; - sum += b_i << lg_H_i; - if (i == B || b_i < (uint32_t)L) - { - rp = ptr; - return sum; - } - lg_H_i += lgH; - } - assert(false); - return 0; + assert(H == (1 << lgH)); + int L = 256 - (1 << lgH); + byte *ptr = rp; + // hand peel the i==0 part of the loop: + uint32_t b_i = *ptr++ & 0xFF; + if (B == 1 || b_i < (uint32_t)L) + { + rp = ptr; + return b_i; + } + uint32_t sum = b_i; + uint32_t lg_H_i = lgH; + assert(B <= B_MAX); + for (int i = 2; i <= B_MAX; i++) + { // easy for compilers to unroll if desired + b_i = *ptr++ & 0xFF; + sum += b_i << lg_H_i; + if (i == B || b_i < (uint32_t)L) + { + rp = ptr; + return sum; + } + lg_H_i += lgH; + } + assert(false); + return 0; } static const char ERB[] = "EOF reading band"; void coding::parseMultiple(byte *&rp, int N, byte *limit, int B, int H) { - if (N < 0) - { - unpack_abort("bad value count"); - return; - } - byte *ptr = rp; - if (B == 1 || H == 256) - { - size_t len = (size_t)N * B; - if (len / B != (size_t)N || ptr + len > limit) - { - unpack_abort(ERB); - return; - } - rp = ptr + len; - return; - } - // Note: We assume rp has enough zero-padding. - int L = 256 - H; - int n = B; - while (N > 0) - { - ptr += 1; - if (--n == 0) - { - // end of encoding at B bytes, regardless of byte value - } - else - { - int b = (ptr[-1] & 0xFF); - if (b >= L) - { - // keep going, unless we find a byte < L - continue; - } - } - // found the last byte - N -= 1; - n = B; // reset length counter - // do an error check here - if (ptr > limit) - { - unpack_abort(ERB); - return; - } - } - rp = ptr; - return; + if (N < 0) + { + unpack_abort("bad value count"); + return; + } + byte *ptr = rp; + if (B == 1 || H == 256) + { + size_t len = (size_t)N * B; + if (len / B != (size_t)N || ptr + len > limit) + { + unpack_abort(ERB); + return; + } + rp = ptr + len; + return; + } + // Note: We assume rp has enough zero-padding. + int L = 256 - H; + int n = B; + while (N > 0) + { + ptr += 1; + if (--n == 0) + { + // end of encoding at B bytes, regardless of byte value + } + else + { + int b = (ptr[-1] & 0xFF); + if (b >= L) + { + // keep going, unless we find a byte < L + continue; + } + } + // found the last byte + N -= 1; + n = B; // reset length counter + // do an error check here + if (ptr > limit) + { + unpack_abort(ERB); + return; + } + } + rp = ptr; + return; } bool value_stream::hasHelper() { - // If my coding method is a pop-style method, - // then I need a second value stream to transmit - // unfavored values. - // This can be determined by examining fValues. - return cm->fValues != nullptr; + // If my coding method is a pop-style method, + // then I need a second value stream to transmit + // unfavored values. + // This can be determined by examining fValues. + return cm->fValues != nullptr; } void value_stream::init(byte *rp_, byte *rplimit_, coding *defc) { - rp = rp_; - rplimit = rplimit_; - sum = 0; - cm = nullptr; // no need in the simple case - setCoding(defc); + rp = rp_; + rplimit = rplimit_; + sum = 0; + cm = nullptr; // no need in the simple case + setCoding(defc); } void value_stream::setCoding(coding *defc) { - if (defc == nullptr) - { - unpack_abort("bad coding"); - defc = coding::findByIndex(_meta_canon_min); // random pick for recovery - } - - c = (*defc); - - // choose cmk - cmk = cmk_ERROR; - switch (c.spec) - { - case BYTE1_spec: - cmk = cmk_BYTE1; - break; - case CHAR3_spec: - cmk = cmk_CHAR3; - break; - case UNSIGNED5_spec: - cmk = cmk_UNSIGNED5; - break; - case DELTA5_spec: - cmk = cmk_DELTA5; - break; - case BCI5_spec: - cmk = cmk_BCI5; - break; - case BRANCH5_spec: - cmk = cmk_BRANCH5; - break; - default: - if (c.D() == 0) - { - switch (c.S()) - { - case 0: - cmk = cmk_BHS0; - break; - case 1: - cmk = cmk_BHS1; - break; - default: - cmk = cmk_BHS; - break; - } - } - else - { - if (c.S() == 1) - { - if (c.isFullRange) - cmk = cmk_BHS1D1full; - if (c.isSubrange) - cmk = cmk_BHS1D1sub; - } - if (cmk == cmk_ERROR) - cmk = cmk_BHSD1; - } - } + if (defc == nullptr) + { + unpack_abort("bad coding"); + defc = coding::findByIndex(_meta_canon_min); // random pick for recovery + } + + c = (*defc); + + // choose cmk + cmk = cmk_ERROR; + switch (c.spec) + { + case BYTE1_spec: + cmk = cmk_BYTE1; + break; + case CHAR3_spec: + cmk = cmk_CHAR3; + break; + case UNSIGNED5_spec: + cmk = cmk_UNSIGNED5; + break; + case DELTA5_spec: + cmk = cmk_DELTA5; + break; + case BCI5_spec: + cmk = cmk_BCI5; + break; + case BRANCH5_spec: + cmk = cmk_BRANCH5; + break; + default: + if (c.D() == 0) + { + switch (c.S()) + { + case 0: + cmk = cmk_BHS0; + break; + case 1: + cmk = cmk_BHS1; + break; + default: + cmk = cmk_BHS; + break; + } + } + else + { + if (c.S() == 1) + { + if (c.isFullRange) + cmk = cmk_BHS1D1full; + if (c.isSubrange) + cmk = cmk_BHS1D1sub; + } + if (cmk == cmk_ERROR) + cmk = cmk_BHSD1; + } + } } static int getPopValue(value_stream *self, uint32_t uval) { - if (uval > 0) - { - // note that the initial parse performed a range check - assert(uval <= (uint32_t)self->cm->fVlength); - return self->cm->fValues[uval - 1]; - } - else - { - // take an unfavored value - return self->helper()->getInt(); - } + if (uval > 0) + { + // note that the initial parse performed a range check + assert(uval <= (uint32_t)self->cm->fVlength); + return self->cm->fValues[uval - 1]; + } + else + { + // take an unfavored value + return self->helper()->getInt(); + } } int coding::sumInUnsignedRange(int x, int y) { - assert(isSubrange); - int range = (int)(umax + 1); - assert(range > 0); - x += y; - if (x != (int)((int64_t)(x - y) + (int64_t)y)) - { - // 32-bit overflow interferes with range reduction. - // Back off from the overflow by adding a multiple of range: - if (x < 0) - { - x -= range; - assert(x >= 0); - } - else - { - x += range; - assert(x < 0); - } - } - if (x < 0) - { - x += range; - if (x >= 0) - return x; - } - else if (x >= range) - { - x -= range; - if (x < range) - return x; - } - else - { - // in range - return x; - } - // do it the hard way - x %= range; - if (x < 0) - x += range; - return x; + assert(isSubrange); + int range = (int)(umax + 1); + assert(range > 0); + x += y; + if (x != (int)((int64_t)(x - y) + (int64_t)y)) + { + // 32-bit overflow interferes with range reduction. + // Back off from the overflow by adding a multiple of range: + if (x < 0) + { + x -= range; + assert(x >= 0); + } + else + { + x += range; + assert(x < 0); + } + } + if (x < 0) + { + x += range; + if (x >= 0) + return x; + } + else if (x >= range) + { + x -= range; + if (x < range) + return x; + } + else + { + // in range + return x; + } + // do it the hard way + x %= range; + if (x < 0) + x += range; + return x; } static int getDeltaValue(value_stream *self, uint32_t uval, bool isSubrange) { - assert((uint32_t)(self->c.isSubrange) == (uint32_t)isSubrange); - assert(self->c.isSubrange | self->c.isFullRange); - if (isSubrange) - return self->sum = self->c.sumInUnsignedRange(self->sum, (int)uval); - else - return self->sum += (int)uval; + assert((uint32_t)(self->c.isSubrange) == (uint32_t)isSubrange); + assert(self->c.isSubrange | self->c.isFullRange); + if (isSubrange) + return self->sum = self->c.sumInUnsignedRange(self->sum, (int)uval); + else + return self->sum += (int)uval; } bool value_stream::hasValue() { - if (rp < rplimit) - return true; - if (cm == nullptr) - return false; - if (cm->next == nullptr) - return false; - cm->next->reset(this); - return hasValue(); + if (rp < rplimit) + return true; + if (cm == nullptr) + return false; + if (cm->next == nullptr) + return false; + cm->next->reset(this); + return hasValue(); } int value_stream::getInt() { - if (rp >= rplimit) - { - // Advance to next coding segment. - if (rp > rplimit || cm == nullptr || cm->next == nullptr) - { - // Must perform this check and throw an exception on bad input. - unpack_abort(ERB); - return 0; - } - cm->next->reset(this); - return getInt(); - } - - CODING_PRIVATE(c.spec); - uint32_t uval; - enum - { - B5 = 5, - B3 = 3, - H128 = 128, - H64 = 64, - H4 = 4 - }; - switch (cmk) - { - case cmk_BHS: - assert(D == 0); - uval = coding::parse(rp, B, H); - if (S == 0) - return (int)uval; - return decode_sign(S, uval); - - case cmk_BHS0: - assert(S == 0 && D == 0); - uval = coding::parse(rp, B, H); - return (int)uval; - - case cmk_BHS1: - assert(S == 1 && D == 0); - uval = coding::parse(rp, B, H); - return DECODE_SIGN_S1(uval); - - case cmk_BYTE1: - assert(c.spec == BYTE1_spec); - assert(B == 1 && H == 256 && S == 0 && D == 0); - return *rp++ & 0xFF; - - case cmk_CHAR3: - assert(c.spec == CHAR3_spec); - assert(B == B3 && H == H128 && S == 0 && D == 0); - return coding::parse_lgH(rp, B3, H128, 7); - - case cmk_UNSIGNED5: - assert(c.spec == UNSIGNED5_spec); - assert(B == B5 && H == H64 && S == 0 && D == 0); - return coding::parse_lgH(rp, B5, H64, 6); - - case cmk_BHSD1: - assert(D == 1); - uval = coding::parse(rp, B, H); - if (S != 0) - uval = (uint32_t)decode_sign(S, uval); - return getDeltaValue(this, uval, (bool)c.isSubrange); - - case cmk_BHS1D1full: - assert(S == 1 && D == 1 && c.isFullRange); - uval = coding::parse(rp, B, H); - uval = (uint32_t)DECODE_SIGN_S1(uval); - return getDeltaValue(this, uval, false); - - case cmk_BHS1D1sub: - assert(S == 1 && D == 1 && c.isSubrange); - uval = coding::parse(rp, B, H); - uval = (uint32_t)DECODE_SIGN_S1(uval); - return getDeltaValue(this, uval, true); - - case cmk_DELTA5: - assert(c.spec == DELTA5_spec); - assert(B == B5 && H == H64 && S == 1 && D == 1 && c.isFullRange); - uval = coding::parse_lgH(rp, B5, H64, 6); - sum += DECODE_SIGN_S1(uval); - return sum; - - case cmk_BCI5: - assert(c.spec == BCI5_spec); - assert(B == B5 && H == H4 && S == 0 && D == 0); - return coding::parse_lgH(rp, B5, H4, 2); - - case cmk_BRANCH5: - assert(c.spec == BRANCH5_spec); - assert(B == B5 && H == H4 && S == 2 && D == 0); - uval = coding::parse_lgH(rp, B5, H4, 2); - return decode_sign(S, uval); - - case cmk_pop: - uval = coding::parse(rp, B, H); - if (S != 0) - { - uval = (uint32_t)decode_sign(S, uval); - } - if (D != 0) - { - assert(c.isSubrange | c.isFullRange); - if (c.isSubrange) - sum = c.sumInUnsignedRange(sum, (int)uval); - else - sum += (int)uval; - uval = (uint32_t)sum; - } - return getPopValue(this, uval); - - case cmk_pop_BHS0: - assert(S == 0 && D == 0); - uval = coding::parse(rp, B, H); - return getPopValue(this, uval); - - case cmk_pop_BYTE1: - assert(c.spec == BYTE1_spec); - assert(B == 1 && H == 256 && S == 0 && D == 0); - return getPopValue(this, *rp++ & 0xFF); - - default: - break; - } - assert(false); - return 0; + if (rp >= rplimit) + { + // Advance to next coding segment. + if (rp > rplimit || cm == nullptr || cm->next == nullptr) + { + // Must perform this check and throw an exception on bad input. + unpack_abort(ERB); + return 0; + } + cm->next->reset(this); + return getInt(); + } + + CODING_PRIVATE(c.spec); + uint32_t uval; + enum + { + B5 = 5, + B3 = 3, + H128 = 128, + H64 = 64, + H4 = 4 + }; + switch (cmk) + { + case cmk_BHS: + assert(D == 0); + uval = coding::parse(rp, B, H); + if (S == 0) + return (int)uval; + return decode_sign(S, uval); + + case cmk_BHS0: + assert(S == 0 && D == 0); + uval = coding::parse(rp, B, H); + return (int)uval; + + case cmk_BHS1: + assert(S == 1 && D == 0); + uval = coding::parse(rp, B, H); + return DECODE_SIGN_S1(uval); + + case cmk_BYTE1: + assert(c.spec == BYTE1_spec); + assert(B == 1 && H == 256 && S == 0 && D == 0); + return *rp++ & 0xFF; + + case cmk_CHAR3: + assert(c.spec == CHAR3_spec); + assert(B == B3 && H == H128 && S == 0 && D == 0); + return coding::parse_lgH(rp, B3, H128, 7); + + case cmk_UNSIGNED5: + assert(c.spec == UNSIGNED5_spec); + assert(B == B5 && H == H64 && S == 0 && D == 0); + return coding::parse_lgH(rp, B5, H64, 6); + + case cmk_BHSD1: + assert(D == 1); + uval = coding::parse(rp, B, H); + if (S != 0) + uval = (uint32_t)decode_sign(S, uval); + return getDeltaValue(this, uval, (bool)c.isSubrange); + + case cmk_BHS1D1full: + assert(S == 1 && D == 1 && c.isFullRange); + uval = coding::parse(rp, B, H); + uval = (uint32_t)DECODE_SIGN_S1(uval); + return getDeltaValue(this, uval, false); + + case cmk_BHS1D1sub: + assert(S == 1 && D == 1 && c.isSubrange); + uval = coding::parse(rp, B, H); + uval = (uint32_t)DECODE_SIGN_S1(uval); + return getDeltaValue(this, uval, true); + + case cmk_DELTA5: + assert(c.spec == DELTA5_spec); + assert(B == B5 && H == H64 && S == 1 && D == 1 && c.isFullRange); + uval = coding::parse_lgH(rp, B5, H64, 6); + sum += DECODE_SIGN_S1(uval); + return sum; + + case cmk_BCI5: + assert(c.spec == BCI5_spec); + assert(B == B5 && H == H4 && S == 0 && D == 0); + return coding::parse_lgH(rp, B5, H4, 2); + + case cmk_BRANCH5: + assert(c.spec == BRANCH5_spec); + assert(B == B5 && H == H4 && S == 2 && D == 0); + uval = coding::parse_lgH(rp, B5, H4, 2); + return decode_sign(S, uval); + + case cmk_pop: + uval = coding::parse(rp, B, H); + if (S != 0) + { + uval = (uint32_t)decode_sign(S, uval); + } + if (D != 0) + { + assert(c.isSubrange | c.isFullRange); + if (c.isSubrange) + sum = c.sumInUnsignedRange(sum, (int)uval); + else + sum += (int)uval; + uval = (uint32_t)sum; + } + return getPopValue(this, uval); + + case cmk_pop_BHS0: + assert(S == 0 && D == 0); + uval = coding::parse(rp, B, H); + return getPopValue(this, uval); + + case cmk_pop_BYTE1: + assert(c.spec == BYTE1_spec); + assert(B == 1 && H == 256 && S == 0 && D == 0); + return getPopValue(this, *rp++ & 0xFF); + + default: + break; + } + assert(false); + return 0; } static int moreCentral(int x, int y) { // used to find end of Pop.{F} - // Suggested implementation from the Pack200 specification: - uint32_t kx = (x >> 31) ^ (x << 1); - uint32_t ky = (y >> 31) ^ (y << 1); - return (kx < ky ? x : y); + // Suggested implementation from the Pack200 specification: + uint32_t kx = (x >> 31) ^ (x << 1); + uint32_t ky = (y >> 31) ^ (y << 1); + return (kx < ky ? x : y); } // static maybe_inline // int moreCentral2(int x, int y, int min) { @@ -641,7 +641,7 @@ static const byte *no_meta[] = {nullptr}; #define NO_META (*(byte **)no_meta) enum { - POP_FAVORED_N = -2 + POP_FAVORED_N = -2 }; // mode bits @@ -650,395 +650,395 @@ enum // This function knows all about meta-coding. void coding_method::init(byte *&band_rp, byte *band_limit, byte *&meta_rp, int mode, - coding *defc, int N, intlist *valueSink) + coding *defc, int N, intlist *valueSink) { - assert(N != 0); - - assert(u != nullptr); // must be pre-initialized - // if (u == nullptr) u = unpacker::current(); // expensive - - int op = (meta_rp == nullptr) ? _meta_default : (*meta_rp++ & 0xFF); - coding *foundc = nullptr; - coding *to_free = nullptr; - - if (op == _meta_default) - { - foundc = defc; - // and fall through - } - else if (op >= _meta_canon_min && op <= _meta_canon_max) - { - foundc = coding::findByIndex(op); - // and fall through - } - else if (op == _meta_arb) - { - int args = (*meta_rp++ & 0xFF); - // args = (D:[0..1] + 2*S[0..2] + 8*(B:[1..5]-1)) - int D = ((args >> 0) & 1); - int S = ((args >> 1) & 3); - int B = ((args >> 3) & -1) + 1; - // & (H[1..256]-1) - int H = (*meta_rp++ & 0xFF) + 1; - foundc = coding::findBySpec(B, H, S, D); - to_free = foundc; // findBySpec may dynamically allocate - if (foundc == nullptr) - { - unpack_abort("illegal arbitrary coding"); - return; - } - // and fall through - } - else if (op >= _meta_run && op < _meta_pop) - { - int args = (op - _meta_run); - // args: KX:[0..3] + 4*(KBFlag:[0..1]) + 8*(ABDef:[0..2]) - int KX = ((args >> 0) & 3); - int KBFlag = ((args >> 2) & 1); - int ABDef = ((args >> 3) & -1); - assert(ABDef <= 2); - // & KB: one of [0..255] if KBFlag=1 - int KB = (!KBFlag ? 3 : (*meta_rp++ & 0xFF)); - int K = (KB + 1) << (KX * 4); - int N2 = (N >= 0) ? N - K : N; - if (N == 0 || (N2 <= 0 && N2 != N)) - { - unpack_abort("illegal run encoding"); - } - if ((mode & DISABLE_RUN) != 0) - { - unpack_abort("illegal nested run encoding"); - } - - // & Enc{ ACode } if ADef=0 (ABDef != 1) - // No direct nesting of 'run' in ACode, but in BCode it's OK. - int disRun = mode | DISABLE_RUN; - if (ABDef == 1) - { - this->init(band_rp, band_limit, NO_META, disRun, defc, K, valueSink); - } - else - { - this->init(band_rp, band_limit, meta_rp, disRun, defc, K, valueSink); - } - - // & Enc{ BCode } if BDef=0 (ABDef != 2) - coding_method *tail = U_NEW(coding_method, 1); - if (!tail) - return; - tail->u = u; - - // The 'run' codings may be nested indirectly via 'pop' codings. - // This means that this->next may already be filled in, if - // ACode was of type 'pop' with a 'run' token coding. - // No problem: Just chain the upcoming BCode onto the end. - for (coding_method *self = this;; self = self->next) - { - if (self->next == nullptr) - { - self->next = tail; - break; - } - } - - if (ABDef == 2) - { - tail->init(band_rp, band_limit, NO_META, mode, defc, N2, valueSink); - } - else - { - tail->init(band_rp, band_limit, meta_rp, mode, defc, N2, valueSink); - } - // Note: The preceding calls to init should be tail-recursive. - - return; // done; no falling through - } - else if (op >= _meta_pop && op < _meta_limit) - { - int args = (op - _meta_pop); - // args: (FDef:[0..1]) + 2*UDef:[0..1] + 4*(TDefL:[0..11]) - int FDef = ((args >> 0) & 1); - int UDef = ((args >> 1) & 1); - int TDefL = ((args >> 2) & -1); - assert(TDefL <= 11); - int TDef = (TDefL > 0); - int TL = (TDefL <= 6) ? (2 << TDefL) : (256 - (4 << (11 - TDefL))); - int TH = (256 - TL); - if (N <= 0) - { - unpack_abort("illegal pop encoding"); - } - if ((mode & DISABLE_POP) != 0) - { - unpack_abort("illegal nested pop encoding"); - } - - // No indirect nesting of 'pop', but 'run' is OK. - int disPop = DISABLE_POP; - - // & Enc{ FCode } if FDef=0 - int FN = POP_FAVORED_N; - assert(valueSink == nullptr); - intlist fValueSink; - fValueSink.init(); - coding_method fval; - BYTES_OF(fval).clear(); - fval.u = u; - if (FDef != 0) - { - fval.init(band_rp, band_limit, NO_META, disPop, defc, FN, &fValueSink); - } - else - { - fval.init(band_rp, band_limit, meta_rp, disPop, defc, FN, &fValueSink); - } - bytes fvbuf; - fValues = (u->saveTo(fvbuf, fValueSink.b), (int *)fvbuf.ptr); - fVlength = fValueSink.length(); // i.e., the parameter K - fValueSink.free(); - - // Skip the first {F} run in all subsequent passes. - // The next call to this->init(...) will set vs0.rp to point after the {F}. - - // & Enc{ TCode } if TDef=0 (TDefL==0) - if (TDef != 0) - { - coding *tcode = coding::findBySpec(1, 256); // BYTE1 - // find the most narrowly sufficient code: - for (int B = 2; B <= B_MAX; B++) - { - if (fVlength <= tcode->umax) - break; // found it - tcode->free(); - tcode = coding::findBySpec(B, TH); - if (!tcode) - return; - } - if (!(fVlength <= tcode->umax)) - { - unpack_abort("pop.L value too small"); - } - this->init(band_rp, band_limit, NO_META, disPop, tcode, N, nullptr); - tcode->free(); - } - else - { - this->init(band_rp, band_limit, meta_rp, disPop, defc, N, nullptr); - } - - // Count the number of zero tokens right now. - // Also verify that they are in bounds. - int UN = 0; // one {U} for each zero in {T} - value_stream vs = vs0; - for (int i = 0; i < N; i++) - { - uint32_t val = vs.getInt(); - if (val == 0) - UN += 1; - if (!(val <= (uint32_t)fVlength)) - { - unpack_abort("pop token out of range"); - } - } - vs.done(); - - // & Enc{ UCode } if UDef=0 - if (UN != 0) - { - uValues = U_NEW(coding_method, 1); - if (uValues == nullptr) - return; - uValues->u = u; - if (UDef != 0) - { - uValues->init(band_rp, band_limit, NO_META, disPop, defc, UN, nullptr); - } - else - { - uValues->init(band_rp, band_limit, meta_rp, disPop, defc, UN, nullptr); - } - } - else - { - if (UDef == 0) - { - int uop = (*meta_rp++ & 0xFF); - if (uop > _meta_canon_max) - // %%% Spec. requires the more strict (uop != _meta_default). - unpack_abort("bad meta-coding for empty pop/U"); - } - } - - // Bug fix for 6259542 - // Last of all, adjust vs0.cmk to the 'pop' flavor - for (coding_method *self = this; self != nullptr; self = self->next) - { - coding_method_kind cmk2 = cmk_pop; - switch (self->vs0.cmk) - { - case cmk_BHS0: - cmk2 = cmk_pop_BHS0; - break; - case cmk_BYTE1: - cmk2 = cmk_pop_BYTE1; - break; - default: - break; - } - self->vs0.cmk = cmk2; - if (self != this) - { - assert(self->fValues == nullptr); // no double init - self->fValues = this->fValues; - self->fVlength = this->fVlength; - assert(self->uValues == nullptr); // must stay nullptr - } - } - - return; // done; no falling through - } - else - { - unpack_abort("bad meta-coding"); - } - - // Common code here skips a series of values with one coding. - assert(foundc != nullptr); - - assert(vs0.cmk == cmk_ERROR); // no garbage, please - assert(vs0.rp == nullptr); // no garbage, please - assert(vs0.rplimit == nullptr); // no garbage, please - assert(vs0.sum == 0); // no garbage, please - - vs0.init(band_rp, band_limit, foundc); - - // Done with foundc. Free if necessary. - if (to_free != nullptr) - { - to_free->free(); - to_free = nullptr; - } - foundc = nullptr; - - coding &c = vs0.c; - CODING_PRIVATE(c.spec); - // assert sane N - assert((uint32_t)N < INT_MAX_VALUE || N == POP_FAVORED_N); - - // Look at the values, or at least skip over them quickly. - if (valueSink == nullptr) - { - // Skip and ignore values in the first pass. - c.parseMultiple(band_rp, N, band_limit, B, H); - } - else if (N >= 0) - { - // Pop coding, {F} sequence, initial run of values... - assert((mode & DISABLE_POP) != 0); - value_stream vs = vs0; - for (int n = 0; n < N; n++) - { - int val = vs.getInt(); - valueSink->add(val); - } - band_rp = vs.rp; - } - else - { - // Pop coding, {F} sequence, final run of values... - assert((mode & DISABLE_POP) != 0); - assert(N == POP_FAVORED_N); - int min = INT_MIN_VALUE; // farthest from the center - // min2 is based on the buggy specification of centrality in version 150.7 - // no known implementations transmit this value, but just in case... - // int min2 = INT_MIN_VALUE; - int last = 0; - // if there were initial runs, find the potential sentinels in them: - for (int i = 0; i < valueSink->length(); i++) - { - last = valueSink->get(i); - min = moreCentral(min, last); - // min2 = moreCentral2(min2, last, min); - } - value_stream vs = vs0; - for (;;) - { - int val = vs.getInt(); - if (valueSink->length() > 0 && (val == last || val == min)) //|| val == min2 - break; - valueSink->add(val); - last = val; - min = moreCentral(min, last); - // min2 = moreCentral2(min2, last, min); - } - band_rp = vs.rp; - } - - // Get an accurate upper limit now. - vs0.rplimit = band_rp; - vs0.cm = this; - - return; // success + assert(N != 0); + + assert(u != nullptr); // must be pre-initialized + // if (u == nullptr) u = unpacker::current(); // expensive + + int op = (meta_rp == nullptr) ? _meta_default : (*meta_rp++ & 0xFF); + coding *foundc = nullptr; + coding *to_free = nullptr; + + if (op == _meta_default) + { + foundc = defc; + // and fall through + } + else if (op >= _meta_canon_min && op <= _meta_canon_max) + { + foundc = coding::findByIndex(op); + // and fall through + } + else if (op == _meta_arb) + { + int args = (*meta_rp++ & 0xFF); + // args = (D:[0..1] + 2*S[0..2] + 8*(B:[1..5]-1)) + int D = ((args >> 0) & 1); + int S = ((args >> 1) & 3); + int B = ((args >> 3) & -1) + 1; + // & (H[1..256]-1) + int H = (*meta_rp++ & 0xFF) + 1; + foundc = coding::findBySpec(B, H, S, D); + to_free = foundc; // findBySpec may dynamically allocate + if (foundc == nullptr) + { + unpack_abort("illegal arbitrary coding"); + return; + } + // and fall through + } + else if (op >= _meta_run && op < _meta_pop) + { + int args = (op - _meta_run); + // args: KX:[0..3] + 4*(KBFlag:[0..1]) + 8*(ABDef:[0..2]) + int KX = ((args >> 0) & 3); + int KBFlag = ((args >> 2) & 1); + int ABDef = ((args >> 3) & -1); + assert(ABDef <= 2); + // & KB: one of [0..255] if KBFlag=1 + int KB = (!KBFlag ? 3 : (*meta_rp++ & 0xFF)); + int K = (KB + 1) << (KX * 4); + int N2 = (N >= 0) ? N - K : N; + if (N == 0 || (N2 <= 0 && N2 != N)) + { + unpack_abort("illegal run encoding"); + } + if ((mode & DISABLE_RUN) != 0) + { + unpack_abort("illegal nested run encoding"); + } + + // & Enc{ ACode } if ADef=0 (ABDef != 1) + // No direct nesting of 'run' in ACode, but in BCode it's OK. + int disRun = mode | DISABLE_RUN; + if (ABDef == 1) + { + this->init(band_rp, band_limit, NO_META, disRun, defc, K, valueSink); + } + else + { + this->init(band_rp, band_limit, meta_rp, disRun, defc, K, valueSink); + } + + // & Enc{ BCode } if BDef=0 (ABDef != 2) + coding_method *tail = U_NEW(coding_method, 1); + if (!tail) + return; + tail->u = u; + + // The 'run' codings may be nested indirectly via 'pop' codings. + // This means that this->next may already be filled in, if + // ACode was of type 'pop' with a 'run' token coding. + // No problem: Just chain the upcoming BCode onto the end. + for (coding_method *self = this;; self = self->next) + { + if (self->next == nullptr) + { + self->next = tail; + break; + } + } + + if (ABDef == 2) + { + tail->init(band_rp, band_limit, NO_META, mode, defc, N2, valueSink); + } + else + { + tail->init(band_rp, band_limit, meta_rp, mode, defc, N2, valueSink); + } + // Note: The preceding calls to init should be tail-recursive. + + return; // done; no falling through + } + else if (op >= _meta_pop && op < _meta_limit) + { + int args = (op - _meta_pop); + // args: (FDef:[0..1]) + 2*UDef:[0..1] + 4*(TDefL:[0..11]) + int FDef = ((args >> 0) & 1); + int UDef = ((args >> 1) & 1); + int TDefL = ((args >> 2) & -1); + assert(TDefL <= 11); + int TDef = (TDefL > 0); + int TL = (TDefL <= 6) ? (2 << TDefL) : (256 - (4 << (11 - TDefL))); + int TH = (256 - TL); + if (N <= 0) + { + unpack_abort("illegal pop encoding"); + } + if ((mode & DISABLE_POP) != 0) + { + unpack_abort("illegal nested pop encoding"); + } + + // No indirect nesting of 'pop', but 'run' is OK. + int disPop = DISABLE_POP; + + // & Enc{ FCode } if FDef=0 + int FN = POP_FAVORED_N; + assert(valueSink == nullptr); + intlist fValueSink; + fValueSink.init(); + coding_method fval; + BYTES_OF(fval).clear(); + fval.u = u; + if (FDef != 0) + { + fval.init(band_rp, band_limit, NO_META, disPop, defc, FN, &fValueSink); + } + else + { + fval.init(band_rp, band_limit, meta_rp, disPop, defc, FN, &fValueSink); + } + bytes fvbuf; + fValues = (u->saveTo(fvbuf, fValueSink.b), (int *)fvbuf.ptr); + fVlength = fValueSink.length(); // i.e., the parameter K + fValueSink.free(); + + // Skip the first {F} run in all subsequent passes. + // The next call to this->init(...) will set vs0.rp to point after the {F}. + + // & Enc{ TCode } if TDef=0 (TDefL==0) + if (TDef != 0) + { + coding *tcode = coding::findBySpec(1, 256); // BYTE1 + // find the most narrowly sufficient code: + for (int B = 2; B <= B_MAX; B++) + { + if (fVlength <= tcode->umax) + break; // found it + tcode->free(); + tcode = coding::findBySpec(B, TH); + if (!tcode) + return; + } + if (!(fVlength <= tcode->umax)) + { + unpack_abort("pop.L value too small"); + } + this->init(band_rp, band_limit, NO_META, disPop, tcode, N, nullptr); + tcode->free(); + } + else + { + this->init(band_rp, band_limit, meta_rp, disPop, defc, N, nullptr); + } + + // Count the number of zero tokens right now. + // Also verify that they are in bounds. + int UN = 0; // one {U} for each zero in {T} + value_stream vs = vs0; + for (int i = 0; i < N; i++) + { + uint32_t val = vs.getInt(); + if (val == 0) + UN += 1; + if (!(val <= (uint32_t)fVlength)) + { + unpack_abort("pop token out of range"); + } + } + vs.done(); + + // & Enc{ UCode } if UDef=0 + if (UN != 0) + { + uValues = U_NEW(coding_method, 1); + if (uValues == nullptr) + return; + uValues->u = u; + if (UDef != 0) + { + uValues->init(band_rp, band_limit, NO_META, disPop, defc, UN, nullptr); + } + else + { + uValues->init(band_rp, band_limit, meta_rp, disPop, defc, UN, nullptr); + } + } + else + { + if (UDef == 0) + { + int uop = (*meta_rp++ & 0xFF); + if (uop > _meta_canon_max) + // %%% Spec. requires the more strict (uop != _meta_default). + unpack_abort("bad meta-coding for empty pop/U"); + } + } + + // Bug fix for 6259542 + // Last of all, adjust vs0.cmk to the 'pop' flavor + for (coding_method *self = this; self != nullptr; self = self->next) + { + coding_method_kind cmk2 = cmk_pop; + switch (self->vs0.cmk) + { + case cmk_BHS0: + cmk2 = cmk_pop_BHS0; + break; + case cmk_BYTE1: + cmk2 = cmk_pop_BYTE1; + break; + default: + break; + } + self->vs0.cmk = cmk2; + if (self != this) + { + assert(self->fValues == nullptr); // no double init + self->fValues = this->fValues; + self->fVlength = this->fVlength; + assert(self->uValues == nullptr); // must stay nullptr + } + } + + return; // done; no falling through + } + else + { + unpack_abort("bad meta-coding"); + } + + // Common code here skips a series of values with one coding. + assert(foundc != nullptr); + + assert(vs0.cmk == cmk_ERROR); // no garbage, please + assert(vs0.rp == nullptr); // no garbage, please + assert(vs0.rplimit == nullptr); // no garbage, please + assert(vs0.sum == 0); // no garbage, please + + vs0.init(band_rp, band_limit, foundc); + + // Done with foundc. Free if necessary. + if (to_free != nullptr) + { + to_free->free(); + to_free = nullptr; + } + foundc = nullptr; + + coding &c = vs0.c; + CODING_PRIVATE(c.spec); + // assert sane N + assert((uint32_t)N < INT_MAX_VALUE || N == POP_FAVORED_N); + + // Look at the values, or at least skip over them quickly. + if (valueSink == nullptr) + { + // Skip and ignore values in the first pass. + c.parseMultiple(band_rp, N, band_limit, B, H); + } + else if (N >= 0) + { + // Pop coding, {F} sequence, initial run of values... + assert((mode & DISABLE_POP) != 0); + value_stream vs = vs0; + for (int n = 0; n < N; n++) + { + int val = vs.getInt(); + valueSink->add(val); + } + band_rp = vs.rp; + } + else + { + // Pop coding, {F} sequence, final run of values... + assert((mode & DISABLE_POP) != 0); + assert(N == POP_FAVORED_N); + int min = INT_MIN_VALUE; // farthest from the center + // min2 is based on the buggy specification of centrality in version 150.7 + // no known implementations transmit this value, but just in case... + // int min2 = INT_MIN_VALUE; + int last = 0; + // if there were initial runs, find the potential sentinels in them: + for (int i = 0; i < valueSink->length(); i++) + { + last = valueSink->get(i); + min = moreCentral(min, last); + // min2 = moreCentral2(min2, last, min); + } + value_stream vs = vs0; + for (;;) + { + int val = vs.getInt(); + if (valueSink->length() > 0 && (val == last || val == min)) //|| val == min2 + break; + valueSink->add(val); + last = val; + min = moreCentral(min, last); + // min2 = moreCentral2(min2, last, min); + } + band_rp = vs.rp; + } + + // Get an accurate upper limit now. + vs0.rplimit = band_rp; + vs0.cm = this; + + return; // success } coding basic_codings[] = { - // This one is not a usable irregular coding, but is used by cp_Utf8_chars. - CODING_INIT(3, 128, 0, 0), - - // Fixed-length codings: - CODING_INIT(1, 256, 0, 0), CODING_INIT(1, 256, 1, 0), CODING_INIT(1, 256, 0, 1), - CODING_INIT(1, 256, 1, 1), CODING_INIT(2, 256, 0, 0), CODING_INIT(2, 256, 1, 0), - CODING_INIT(2, 256, 0, 1), CODING_INIT(2, 256, 1, 1), CODING_INIT(3, 256, 0, 0), - CODING_INIT(3, 256, 1, 0), CODING_INIT(3, 256, 0, 1), CODING_INIT(3, 256, 1, 1), - CODING_INIT(4, 256, 0, 0), CODING_INIT(4, 256, 1, 0), CODING_INIT(4, 256, 0, 1), - CODING_INIT(4, 256, 1, 1), - - // Full-range variable-length codings: - CODING_INIT(5, 4, 0, 0), CODING_INIT(5, 4, 1, 0), CODING_INIT(5, 4, 2, 0), - CODING_INIT(5, 16, 0, 0), CODING_INIT(5, 16, 1, 0), CODING_INIT(5, 16, 2, 0), - CODING_INIT(5, 32, 0, 0), CODING_INIT(5, 32, 1, 0), CODING_INIT(5, 32, 2, 0), - CODING_INIT(5, 64, 0, 0), CODING_INIT(5, 64, 1, 0), CODING_INIT(5, 64, 2, 0), - CODING_INIT(5, 128, 0, 0), CODING_INIT(5, 128, 1, 0), CODING_INIT(5, 128, 2, 0), - CODING_INIT(5, 4, 0, 1), CODING_INIT(5, 4, 1, 1), CODING_INIT(5, 4, 2, 1), - CODING_INIT(5, 16, 0, 1), CODING_INIT(5, 16, 1, 1), CODING_INIT(5, 16, 2, 1), - CODING_INIT(5, 32, 0, 1), CODING_INIT(5, 32, 1, 1), CODING_INIT(5, 32, 2, 1), - CODING_INIT(5, 64, 0, 1), CODING_INIT(5, 64, 1, 1), CODING_INIT(5, 64, 2, 1), - CODING_INIT(5, 128, 0, 1), CODING_INIT(5, 128, 1, 1), CODING_INIT(5, 128, 2, 1), - - // Variable length subrange codings: - CODING_INIT(2, 192, 0, 0), CODING_INIT(2, 224, 0, 0), CODING_INIT(2, 240, 0, 0), - CODING_INIT(2, 248, 0, 0), CODING_INIT(2, 252, 0, 0), CODING_INIT(2, 8, 0, 1), - CODING_INIT(2, 8, 1, 1), CODING_INIT(2, 16, 0, 1), CODING_INIT(2, 16, 1, 1), - CODING_INIT(2, 32, 0, 1), CODING_INIT(2, 32, 1, 1), CODING_INIT(2, 64, 0, 1), - CODING_INIT(2, 64, 1, 1), CODING_INIT(2, 128, 0, 1), CODING_INIT(2, 128, 1, 1), - CODING_INIT(2, 192, 0, 1), CODING_INIT(2, 192, 1, 1), CODING_INIT(2, 224, 0, 1), - CODING_INIT(2, 224, 1, 1), CODING_INIT(2, 240, 0, 1), CODING_INIT(2, 240, 1, 1), - CODING_INIT(2, 248, 0, 1), CODING_INIT(2, 248, 1, 1), CODING_INIT(3, 192, 0, 0), - CODING_INIT(3, 224, 0, 0), CODING_INIT(3, 240, 0, 0), CODING_INIT(3, 248, 0, 0), - CODING_INIT(3, 252, 0, 0), CODING_INIT(3, 8, 0, 1), CODING_INIT(3, 8, 1, 1), - CODING_INIT(3, 16, 0, 1), CODING_INIT(3, 16, 1, 1), CODING_INIT(3, 32, 0, 1), - CODING_INIT(3, 32, 1, 1), CODING_INIT(3, 64, 0, 1), CODING_INIT(3, 64, 1, 1), - CODING_INIT(3, 128, 0, 1), CODING_INIT(3, 128, 1, 1), CODING_INIT(3, 192, 0, 1), - CODING_INIT(3, 192, 1, 1), CODING_INIT(3, 224, 0, 1), CODING_INIT(3, 224, 1, 1), - CODING_INIT(3, 240, 0, 1), CODING_INIT(3, 240, 1, 1), CODING_INIT(3, 248, 0, 1), - CODING_INIT(3, 248, 1, 1), CODING_INIT(4, 192, 0, 0), CODING_INIT(4, 224, 0, 0), - CODING_INIT(4, 240, 0, 0), CODING_INIT(4, 248, 0, 0), CODING_INIT(4, 252, 0, 0), - CODING_INIT(4, 8, 0, 1), CODING_INIT(4, 8, 1, 1), CODING_INIT(4, 16, 0, 1), - CODING_INIT(4, 16, 1, 1), CODING_INIT(4, 32, 0, 1), CODING_INIT(4, 32, 1, 1), - CODING_INIT(4, 64, 0, 1), CODING_INIT(4, 64, 1, 1), CODING_INIT(4, 128, 0, 1), - CODING_INIT(4, 128, 1, 1), CODING_INIT(4, 192, 0, 1), CODING_INIT(4, 192, 1, 1), - CODING_INIT(4, 224, 0, 1), CODING_INIT(4, 224, 1, 1), CODING_INIT(4, 240, 0, 1), - CODING_INIT(4, 240, 1, 1), CODING_INIT(4, 248, 0, 1), CODING_INIT(4, 248, 1, 1), - CODING_INIT(0, 0, 0, 0)}; + // This one is not a usable irregular coding, but is used by cp_Utf8_chars. + CODING_INIT(3, 128, 0, 0), + + // Fixed-length codings: + CODING_INIT(1, 256, 0, 0), CODING_INIT(1, 256, 1, 0), CODING_INIT(1, 256, 0, 1), + CODING_INIT(1, 256, 1, 1), CODING_INIT(2, 256, 0, 0), CODING_INIT(2, 256, 1, 0), + CODING_INIT(2, 256, 0, 1), CODING_INIT(2, 256, 1, 1), CODING_INIT(3, 256, 0, 0), + CODING_INIT(3, 256, 1, 0), CODING_INIT(3, 256, 0, 1), CODING_INIT(3, 256, 1, 1), + CODING_INIT(4, 256, 0, 0), CODING_INIT(4, 256, 1, 0), CODING_INIT(4, 256, 0, 1), + CODING_INIT(4, 256, 1, 1), + + // Full-range variable-length codings: + CODING_INIT(5, 4, 0, 0), CODING_INIT(5, 4, 1, 0), CODING_INIT(5, 4, 2, 0), + CODING_INIT(5, 16, 0, 0), CODING_INIT(5, 16, 1, 0), CODING_INIT(5, 16, 2, 0), + CODING_INIT(5, 32, 0, 0), CODING_INIT(5, 32, 1, 0), CODING_INIT(5, 32, 2, 0), + CODING_INIT(5, 64, 0, 0), CODING_INIT(5, 64, 1, 0), CODING_INIT(5, 64, 2, 0), + CODING_INIT(5, 128, 0, 0), CODING_INIT(5, 128, 1, 0), CODING_INIT(5, 128, 2, 0), + CODING_INIT(5, 4, 0, 1), CODING_INIT(5, 4, 1, 1), CODING_INIT(5, 4, 2, 1), + CODING_INIT(5, 16, 0, 1), CODING_INIT(5, 16, 1, 1), CODING_INIT(5, 16, 2, 1), + CODING_INIT(5, 32, 0, 1), CODING_INIT(5, 32, 1, 1), CODING_INIT(5, 32, 2, 1), + CODING_INIT(5, 64, 0, 1), CODING_INIT(5, 64, 1, 1), CODING_INIT(5, 64, 2, 1), + CODING_INIT(5, 128, 0, 1), CODING_INIT(5, 128, 1, 1), CODING_INIT(5, 128, 2, 1), + + // Variable length subrange codings: + CODING_INIT(2, 192, 0, 0), CODING_INIT(2, 224, 0, 0), CODING_INIT(2, 240, 0, 0), + CODING_INIT(2, 248, 0, 0), CODING_INIT(2, 252, 0, 0), CODING_INIT(2, 8, 0, 1), + CODING_INIT(2, 8, 1, 1), CODING_INIT(2, 16, 0, 1), CODING_INIT(2, 16, 1, 1), + CODING_INIT(2, 32, 0, 1), CODING_INIT(2, 32, 1, 1), CODING_INIT(2, 64, 0, 1), + CODING_INIT(2, 64, 1, 1), CODING_INIT(2, 128, 0, 1), CODING_INIT(2, 128, 1, 1), + CODING_INIT(2, 192, 0, 1), CODING_INIT(2, 192, 1, 1), CODING_INIT(2, 224, 0, 1), + CODING_INIT(2, 224, 1, 1), CODING_INIT(2, 240, 0, 1), CODING_INIT(2, 240, 1, 1), + CODING_INIT(2, 248, 0, 1), CODING_INIT(2, 248, 1, 1), CODING_INIT(3, 192, 0, 0), + CODING_INIT(3, 224, 0, 0), CODING_INIT(3, 240, 0, 0), CODING_INIT(3, 248, 0, 0), + CODING_INIT(3, 252, 0, 0), CODING_INIT(3, 8, 0, 1), CODING_INIT(3, 8, 1, 1), + CODING_INIT(3, 16, 0, 1), CODING_INIT(3, 16, 1, 1), CODING_INIT(3, 32, 0, 1), + CODING_INIT(3, 32, 1, 1), CODING_INIT(3, 64, 0, 1), CODING_INIT(3, 64, 1, 1), + CODING_INIT(3, 128, 0, 1), CODING_INIT(3, 128, 1, 1), CODING_INIT(3, 192, 0, 1), + CODING_INIT(3, 192, 1, 1), CODING_INIT(3, 224, 0, 1), CODING_INIT(3, 224, 1, 1), + CODING_INIT(3, 240, 0, 1), CODING_INIT(3, 240, 1, 1), CODING_INIT(3, 248, 0, 1), + CODING_INIT(3, 248, 1, 1), CODING_INIT(4, 192, 0, 0), CODING_INIT(4, 224, 0, 0), + CODING_INIT(4, 240, 0, 0), CODING_INIT(4, 248, 0, 0), CODING_INIT(4, 252, 0, 0), + CODING_INIT(4, 8, 0, 1), CODING_INIT(4, 8, 1, 1), CODING_INIT(4, 16, 0, 1), + CODING_INIT(4, 16, 1, 1), CODING_INIT(4, 32, 0, 1), CODING_INIT(4, 32, 1, 1), + CODING_INIT(4, 64, 0, 1), CODING_INIT(4, 64, 1, 1), CODING_INIT(4, 128, 0, 1), + CODING_INIT(4, 128, 1, 1), CODING_INIT(4, 192, 0, 1), CODING_INIT(4, 192, 1, 1), + CODING_INIT(4, 224, 0, 1), CODING_INIT(4, 224, 1, 1), CODING_INIT(4, 240, 0, 1), + CODING_INIT(4, 240, 1, 1), CODING_INIT(4, 248, 0, 1), CODING_INIT(4, 248, 1, 1), + CODING_INIT(0, 0, 0, 0)}; #define BASIC_INDEX_LIMIT (int)(sizeof(basic_codings) / sizeof(basic_codings[0]) - 1) coding *coding::findByIndex(int idx) { - int index_limit = BASIC_INDEX_LIMIT; - assert(_meta_canon_min == 1 && _meta_canon_max + 1 == index_limit); + int index_limit = BASIC_INDEX_LIMIT; + assert(_meta_canon_min == 1 && _meta_canon_max + 1 == index_limit); - if (idx >= _meta_canon_min && idx <= _meta_canon_max) - return basic_codings[idx].init(); - else - return nullptr; + if (idx >= _meta_canon_min && idx <= _meta_canon_max) + return basic_codings[idx].init(); + else + return nullptr; } diff --git a/libraries/pack200/src/coding.h b/libraries/pack200/src/coding.h index f9bd6ca2..bfdd252e 100644 --- a/libraries/pack200/src/coding.h +++ b/libraries/pack200/src/coding.h @@ -35,9 +35,9 @@ struct unpacker; #define CODING_D(x) ((x) >> 0 & 0xF) #define CODING_INIT(B, H, S, D) \ - { \ - CODING_SPEC(B, H, S, D), 0, 0, 0, 0, 0, 0, 0, 0 \ - } + { \ + CODING_SPEC(B, H, S, D), 0, 0, 0, 0, 0, 0, 0, 0 \ + } // For debugging purposes, some compilers do not like this and will complain. // #define long do_not_use_C_long_types_use_jlong_or_int @@ -45,126 +45,126 @@ struct unpacker; struct coding { - int spec; // B,H,S,D - - // Handy values derived from the spec: - int B() - { - return CODING_B(spec); - } - int H() - { - return CODING_H(spec); - } - int S() - { - return CODING_S(spec); - } - int D() - { - return CODING_D(spec); - } - int L() - { - return 256 - CODING_H(spec); - } - int min, max; - int umin, umax; - char isSigned, isSubrange, isFullRange, isMalloc; - - coding *init(); // returns self or nullptr if error - coding *initFrom(int spec_) - { - assert(this->spec == 0); - this->spec = spec_; - return init(); - } - - static coding *findBySpec(int spec); - static coding *findBySpec(int B, int H, int S = 0, int D = 0); - static coding *findByIndex(int irregularCodingIndex); - - static uint32_t parse(byte *&rp, int B, int H); - static uint32_t parse_lgH(byte *&rp, int B, int H, int lgH); - static void parseMultiple(byte *&rp, int N, byte *limit, int B, int H); - - uint32_t parse(byte *&rp) - { - return parse(rp, CODING_B(spec), CODING_H(spec)); - } - void parseMultiple(byte *&rp, int N, byte *limit) - { - parseMultiple(rp, N, limit, CODING_B(spec), CODING_H(spec)); - } - - bool canRepresent(int x) - { - return (x >= min && x <= max); - } - bool canRepresentUnsigned(int x) - { - return (x >= umin && x <= umax); - } - - int sumInUnsignedRange(int x, int y); - - int readFrom(byte *&rpVar, int *dbase); - void readArrayFrom(byte *&rpVar, int *dbase, int length, int *values); - void skipArrayFrom(byte *&rpVar, int length) - { - readArrayFrom(rpVar, (int *)NULL, length, (int *)NULL); - } - - void free(); // free self if isMalloc + int spec; // B,H,S,D + + // Handy values derived from the spec: + int B() + { + return CODING_B(spec); + } + int H() + { + return CODING_H(spec); + } + int S() + { + return CODING_S(spec); + } + int D() + { + return CODING_D(spec); + } + int L() + { + return 256 - CODING_H(spec); + } + int min, max; + int umin, umax; + char isSigned, isSubrange, isFullRange, isMalloc; + + coding *init(); // returns self or nullptr if error + coding *initFrom(int spec_) + { + assert(this->spec == 0); + this->spec = spec_; + return init(); + } + + static coding *findBySpec(int spec); + static coding *findBySpec(int B, int H, int S = 0, int D = 0); + static coding *findByIndex(int irregularCodingIndex); + + static uint32_t parse(byte *&rp, int B, int H); + static uint32_t parse_lgH(byte *&rp, int B, int H, int lgH); + static void parseMultiple(byte *&rp, int N, byte *limit, int B, int H); + + uint32_t parse(byte *&rp) + { + return parse(rp, CODING_B(spec), CODING_H(spec)); + } + void parseMultiple(byte *&rp, int N, byte *limit) + { + parseMultiple(rp, N, limit, CODING_B(spec), CODING_H(spec)); + } + + bool canRepresent(int x) + { + return (x >= min && x <= max); + } + bool canRepresentUnsigned(int x) + { + return (x >= umin && x <= umax); + } + + int sumInUnsignedRange(int x, int y); + + int readFrom(byte *&rpVar, int *dbase); + void readArrayFrom(byte *&rpVar, int *dbase, int length, int *values); + void skipArrayFrom(byte *&rpVar, int length) + { + readArrayFrom(rpVar, (int *)NULL, length, (int *)NULL); + } + + void free(); // free self if isMalloc }; enum coding_method_kind { - cmk_ERROR, - cmk_BHS, - cmk_BHS0, - cmk_BHS1, - cmk_BHSD1, - cmk_BHS1D1full, // isFullRange - cmk_BHS1D1sub, // isSubRange - - // special cases hand-optimized (~50% of all decoded values) - cmk_BYTE1, //(1,256) 6% - cmk_CHAR3, //(3,128) 7% - cmk_UNSIGNED5, //(5,64) 13% - cmk_DELTA5, //(5,64,1,1) 5% - cmk_BCI5, //(5,4) 18% - cmk_BRANCH5, //(5,4,2) 4% - // cmk_UNSIGNED5H16, //(5,16) 5% - // cmk_UNSIGNED2H4, //(2,4) 6% - // cmk_DELTA4H8, //(4,8,1,1) 10% - // cmk_DELTA3H16, //(3,16,1,1) 9% - cmk_BHS_LIMIT, - cmk_pop, - cmk_pop_BHS0, - cmk_pop_BYTE1, - cmk_pop_LIMIT, - cmk_LIMIT + cmk_ERROR, + cmk_BHS, + cmk_BHS0, + cmk_BHS1, + cmk_BHSD1, + cmk_BHS1D1full, // isFullRange + cmk_BHS1D1sub, // isSubRange + + // special cases hand-optimized (~50% of all decoded values) + cmk_BYTE1, //(1,256) 6% + cmk_CHAR3, //(3,128) 7% + cmk_UNSIGNED5, //(5,64) 13% + cmk_DELTA5, //(5,64,1,1) 5% + cmk_BCI5, //(5,4) 18% + cmk_BRANCH5, //(5,4,2) 4% + // cmk_UNSIGNED5H16, //(5,16) 5% + // cmk_UNSIGNED2H4, //(2,4) 6% + // cmk_DELTA4H8, //(4,8,1,1) 10% + // cmk_DELTA3H16, //(3,16,1,1) 9% + cmk_BHS_LIMIT, + cmk_pop, + cmk_pop_BHS0, + cmk_pop_BYTE1, + cmk_pop_LIMIT, + cmk_LIMIT }; enum { - BYTE1_spec = CODING_SPEC(1, 256, 0, 0), - CHAR3_spec = CODING_SPEC(3, 128, 0, 0), - UNSIGNED4_spec = CODING_SPEC(4, 256, 0, 0), - UNSIGNED5_spec = CODING_SPEC(5, 64, 0, 0), - SIGNED5_spec = CODING_SPEC(5, 64, 1, 0), - DELTA5_spec = CODING_SPEC(5, 64, 1, 1), - UDELTA5_spec = CODING_SPEC(5, 64, 0, 1), - MDELTA5_spec = CODING_SPEC(5, 64, 2, 1), - BCI5_spec = CODING_SPEC(5, 4, 0, 0), - BRANCH5_spec = CODING_SPEC(5, 4, 2, 0) + BYTE1_spec = CODING_SPEC(1, 256, 0, 0), + CHAR3_spec = CODING_SPEC(3, 128, 0, 0), + UNSIGNED4_spec = CODING_SPEC(4, 256, 0, 0), + UNSIGNED5_spec = CODING_SPEC(5, 64, 0, 0), + SIGNED5_spec = CODING_SPEC(5, 64, 1, 0), + DELTA5_spec = CODING_SPEC(5, 64, 1, 1), + UDELTA5_spec = CODING_SPEC(5, 64, 0, 1), + MDELTA5_spec = CODING_SPEC(5, 64, 2, 1), + BCI5_spec = CODING_SPEC(5, 4, 0, 0), + BRANCH5_spec = CODING_SPEC(5, 4, 2, 0) }; enum { - B_MAX = 5, - C_SLOP = B_MAX * 10 + B_MAX = 5, + C_SLOP = B_MAX * 10 }; struct coding_method; @@ -172,76 +172,76 @@ struct coding_method; // iterator under the control of a meta-coding struct value_stream { - // current coding of values or values - coding c; // B,H,S,D,etc. - coding_method_kind cmk; // type of decoding needed - byte *rp; // read pointer - byte *rplimit; // final value of read pointer - int sum; // partial sum of all values so far (D=1 only) - coding_method *cm; // coding method that defines this stream - - void init(byte *band_rp, byte *band_limit, coding *defc); - void init(byte *band_rp, byte *band_limit, int spec) - { - init(band_rp, band_limit, coding::findBySpec(spec)); - } - - void setCoding(coding *c); - void setCoding(int spec) - { - setCoding(coding::findBySpec(spec)); - } - - // Parse and decode a single value. - int getInt(); - - // Parse and decode a single byte, with no error checks. - int getByte() - { - assert(cmk == cmk_BYTE1); - assert(rp < rplimit); - return *rp++ & 0xFF; - } - - // Used only for asserts. - bool hasValue(); - - void done() - { - assert(!hasValue()); - } - - // Sometimes a value stream has an auxiliary (but there are never two). - value_stream *helper() - { - assert(hasHelper()); - return this + 1; - } - bool hasHelper(); + // current coding of values or values + coding c; // B,H,S,D,etc. + coding_method_kind cmk; // type of decoding needed + byte *rp; // read pointer + byte *rplimit; // final value of read pointer + int sum; // partial sum of all values so far (D=1 only) + coding_method *cm; // coding method that defines this stream + + void init(byte *band_rp, byte *band_limit, coding *defc); + void init(byte *band_rp, byte *band_limit, int spec) + { + init(band_rp, band_limit, coding::findBySpec(spec)); + } + + void setCoding(coding *c); + void setCoding(int spec) + { + setCoding(coding::findBySpec(spec)); + } + + // Parse and decode a single value. + int getInt(); + + // Parse and decode a single byte, with no error checks. + int getByte() + { + assert(cmk == cmk_BYTE1); + assert(rp < rplimit); + return *rp++ & 0xFF; + } + + // Used only for asserts. + bool hasValue(); + + void done() + { + assert(!hasValue()); + } + + // Sometimes a value stream has an auxiliary (but there are never two). + value_stream *helper() + { + assert(hasHelper()); + return this + 1; + } + bool hasHelper(); }; struct coding_method { - value_stream vs0; // initial state snapshot (vs.meta==this) + value_stream vs0; // initial state snapshot (vs.meta==this) - coding_method *next; // what to do when we run out of bytes + coding_method *next; // what to do when we run out of bytes - // these fields are used for pop codes only: - int *fValues; // favored value array - int fVlength; // maximum favored value token - coding_method *uValues; // unfavored value stream + // these fields are used for pop codes only: + int *fValues; // favored value array + int fVlength; // maximum favored value token + coding_method *uValues; // unfavored value stream - // pointer to outer unpacker, for error checks etc. - unpacker *u; + // pointer to outer unpacker, for error checks etc. + unpacker *u; - // Initialize a value stream. - void reset(value_stream *state); + // Initialize a value stream. + void reset(value_stream *state); - // Parse a band header, size a band, and initialize for further action. - // band_rp advances (but not past band_limit), and meta_rp advances. - // The mode gives context, such as "inside a pop". - // The defc and N are the incoming parameters to a meta-coding. - // The value sink is used to collect output values, when desired. - void init(byte *&band_rp, byte *band_limit, byte *&meta_rp, int mode, coding *defc, int N, - intlist *valueSink); + // Parse a band header, size a band, and initialize for further action. + // band_rp advances (but not past band_limit), and meta_rp advances. + // The mode gives context, such as "inside a pop". + // The defc and N are the incoming parameters to a meta-coding. + // The value sink is used to collect output values, when desired. + void init(byte *&band_rp, byte *band_limit, byte *&meta_rp, int mode, coding *defc, int N, + intlist *valueSink); }; diff --git a/libraries/pack200/src/constants.h b/libraries/pack200/src/constants.h index 2cc14b7d..f1baf42a 100644 --- a/libraries/pack200/src/constants.h +++ b/libraries/pack200/src/constants.h @@ -55,388 +55,388 @@ enum { - CONSTANT_None, - CONSTANT_Utf8, - CONSTANT_unused2, /* unused, was Unicode */ - CONSTANT_Integer, - CONSTANT_Float, - CONSTANT_Long, - CONSTANT_Double, - CONSTANT_Class, - CONSTANT_String, - CONSTANT_Fieldref, - CONSTANT_Methodref, - CONSTANT_InterfaceMethodref, - CONSTANT_NameandType, - CONSTANT_Signature = 13, - CONSTANT_All = 14, - CONSTANT_Limit = 15, - CONSTANT_NONE = 0, - CONSTANT_Literal = 20, // pseudo-tag for debugging - CONSTANT_Member = 21, // pseudo-tag for debugging - SUBINDEX_BIT = 64, // combined with CONSTANT_xxx for ixTag - ACC_STATIC = 0x0008, - ACC_IC_LONG_FORM = (1 << 16), // for ic_flags - CLASS_ATTR_SourceFile = 17, - CLASS_ATTR_EnclosingMethod = 18, - CLASS_ATTR_InnerClasses = 23, - CLASS_ATTR_ClassFile_version = 24, - FIELD_ATTR_ConstantValue = 17, - METHOD_ATTR_Code = 17, - METHOD_ATTR_Exceptions = 18, - METHOD_ATTR_RuntimeVisibleParameterAnnotations = 23, - METHOD_ATTR_RuntimeInvisibleParameterAnnotations = 24, - METHOD_ATTR_AnnotationDefault = 25, - CODE_ATTR_StackMapTable = 0, - CODE_ATTR_LineNumberTable = 1, - CODE_ATTR_LocalVariableTable = 2, - CODE_ATTR_LocalVariableTypeTable = 3, - // X_ATTR_Synthetic = 12, // ACC_SYNTHETIC; not predefined - X_ATTR_Signature = 19, - X_ATTR_Deprecated = 20, - X_ATTR_RuntimeVisibleAnnotations = 21, - X_ATTR_RuntimeInvisibleAnnotations = 22, - X_ATTR_OVERFLOW = 16, - X_ATTR_LIMIT_NO_FLAGS_HI = 32, - X_ATTR_LIMIT_FLAGS_HI = 63, + CONSTANT_None, + CONSTANT_Utf8, + CONSTANT_unused2, /* unused, was Unicode */ + CONSTANT_Integer, + CONSTANT_Float, + CONSTANT_Long, + CONSTANT_Double, + CONSTANT_Class, + CONSTANT_String, + CONSTANT_Fieldref, + CONSTANT_Methodref, + CONSTANT_InterfaceMethodref, + CONSTANT_NameandType, + CONSTANT_Signature = 13, + CONSTANT_All = 14, + CONSTANT_Limit = 15, + CONSTANT_NONE = 0, + CONSTANT_Literal = 20, // pseudo-tag for debugging + CONSTANT_Member = 21, // pseudo-tag for debugging + SUBINDEX_BIT = 64, // combined with CONSTANT_xxx for ixTag + ACC_STATIC = 0x0008, + ACC_IC_LONG_FORM = (1 << 16), // for ic_flags + CLASS_ATTR_SourceFile = 17, + CLASS_ATTR_EnclosingMethod = 18, + CLASS_ATTR_InnerClasses = 23, + CLASS_ATTR_ClassFile_version = 24, + FIELD_ATTR_ConstantValue = 17, + METHOD_ATTR_Code = 17, + METHOD_ATTR_Exceptions = 18, + METHOD_ATTR_RuntimeVisibleParameterAnnotations = 23, + METHOD_ATTR_RuntimeInvisibleParameterAnnotations = 24, + METHOD_ATTR_AnnotationDefault = 25, + CODE_ATTR_StackMapTable = 0, + CODE_ATTR_LineNumberTable = 1, + CODE_ATTR_LocalVariableTable = 2, + CODE_ATTR_LocalVariableTypeTable = 3, + // X_ATTR_Synthetic = 12, // ACC_SYNTHETIC; not predefined + X_ATTR_Signature = 19, + X_ATTR_Deprecated = 20, + X_ATTR_RuntimeVisibleAnnotations = 21, + X_ATTR_RuntimeInvisibleAnnotations = 22, + X_ATTR_OVERFLOW = 16, + X_ATTR_LIMIT_NO_FLAGS_HI = 32, + X_ATTR_LIMIT_FLAGS_HI = 63, #define O_ATTR_DO(F) \ - F(X_ATTR_OVERFLOW, 01) \ - /*(end)*/ + F(X_ATTR_OVERFLOW, 01) \ + /*(end)*/ #define X_ATTR_DO(F) \ - O_ATTR_DO(F) F(X_ATTR_Signature, Signature) F(X_ATTR_Deprecated, Deprecated) \ - F(X_ATTR_RuntimeVisibleAnnotations, RuntimeVisibleAnnotations) \ - F(X_ATTR_RuntimeInvisibleAnnotations, RuntimeInvisibleAnnotations) \ - /*F(X_ATTR_Synthetic,Synthetic)*/ \ - /*(end)*/ + O_ATTR_DO(F) F(X_ATTR_Signature, Signature) F(X_ATTR_Deprecated, Deprecated) \ + F(X_ATTR_RuntimeVisibleAnnotations, RuntimeVisibleAnnotations) \ + F(X_ATTR_RuntimeInvisibleAnnotations, RuntimeInvisibleAnnotations) \ + /*F(X_ATTR_Synthetic,Synthetic)*/ \ + /*(end)*/ #define CLASS_ATTR_DO(F) \ - F(CLASS_ATTR_SourceFile, SourceFile) F(CLASS_ATTR_InnerClasses, InnerClasses) \ - F(CLASS_ATTR_EnclosingMethod, EnclosingMethod) F(CLASS_ATTR_ClassFile_version, 02) \ - /*(end)*/ + F(CLASS_ATTR_SourceFile, SourceFile) F(CLASS_ATTR_InnerClasses, InnerClasses) \ + F(CLASS_ATTR_EnclosingMethod, EnclosingMethod) F(CLASS_ATTR_ClassFile_version, 02) \ + /*(end)*/ #define FIELD_ATTR_DO(F) \ - F(FIELD_ATTR_ConstantValue, ConstantValue) \ - /*(end)*/ + F(FIELD_ATTR_ConstantValue, ConstantValue) \ + /*(end)*/ #define METHOD_ATTR_DO(F) \ - F(METHOD_ATTR_Code, Code) F(METHOD_ATTR_Exceptions, Exceptions) \ - F(METHOD_ATTR_RuntimeVisibleParameterAnnotations, RuntimeVisibleParameterAnnotations) \ - F(METHOD_ATTR_RuntimeInvisibleParameterAnnotations, \ - RuntimeInvisibleParameterAnnotations) \ - F(METHOD_ATTR_AnnotationDefault, AnnotationDefault) \ - /*(end)*/ + F(METHOD_ATTR_Code, Code) F(METHOD_ATTR_Exceptions, Exceptions) \ + F(METHOD_ATTR_RuntimeVisibleParameterAnnotations, RuntimeVisibleParameterAnnotations) \ + F(METHOD_ATTR_RuntimeInvisibleParameterAnnotations, \ + RuntimeInvisibleParameterAnnotations) \ + F(METHOD_ATTR_AnnotationDefault, AnnotationDefault) \ + /*(end)*/ #define CODE_ATTR_DO(F) \ - F(CODE_ATTR_StackMapTable, StackMapTable) F(CODE_ATTR_LineNumberTable, LineNumberTable) \ - F(CODE_ATTR_LocalVariableTable, LocalVariableTable) \ - F(CODE_ATTR_LocalVariableTypeTable, LocalVariableTypeTable) \ - /*(end)*/ + F(CODE_ATTR_StackMapTable, StackMapTable) F(CODE_ATTR_LineNumberTable, LineNumberTable) \ + F(CODE_ATTR_LocalVariableTable, LocalVariableTable) \ + F(CODE_ATTR_LocalVariableTypeTable, LocalVariableTypeTable) \ + /*(end)*/ #define ALL_ATTR_DO(F) \ - X_ATTR_DO(F) CLASS_ATTR_DO(F) FIELD_ATTR_DO(F) METHOD_ATTR_DO(F) CODE_ATTR_DO(F) \ - /*(end)*/ + X_ATTR_DO(F) CLASS_ATTR_DO(F) FIELD_ATTR_DO(F) METHOD_ATTR_DO(F) CODE_ATTR_DO(F) \ + /*(end)*/ - // attribute "context types" - ATTR_CONTEXT_CLASS = 0, - ATTR_CONTEXT_FIELD = 1, - ATTR_CONTEXT_METHOD = 2, - ATTR_CONTEXT_CODE = 3, - ATTR_CONTEXT_LIMIT = 4, + // attribute "context types" + ATTR_CONTEXT_CLASS = 0, + ATTR_CONTEXT_FIELD = 1, + ATTR_CONTEXT_METHOD = 2, + ATTR_CONTEXT_CODE = 3, + ATTR_CONTEXT_LIMIT = 4, - // constants for parsed layouts (stored in band::le_kind) - EK_NONE = 0, // not a layout element - EK_INT = 'I', // B H I SH etc., also FH etc. - EK_BCI = 'P', // PH etc. - EK_BCID = 'Q', // POH etc. - EK_BCO = 'O', // OH etc. - EK_REPL = 'N', // NH[...] etc. - EK_REF = 'R', // RUH, RUNH, KQH, etc. - EK_UN = 'T', // TB(...)[...] etc. - EK_CASE = 'K', // (...)[...] etc. - EK_CALL = '(', // (0), (1), etc. - EK_CBLE = '[', // [...][...] etc. - NO_BAND_INDEX = -1, + // constants for parsed layouts (stored in band::le_kind) + EK_NONE = 0, // not a layout element + EK_INT = 'I', // B H I SH etc., also FH etc. + EK_BCI = 'P', // PH etc. + EK_BCID = 'Q', // POH etc. + EK_BCO = 'O', // OH etc. + EK_REPL = 'N', // NH[...] etc. + EK_REF = 'R', // RUH, RUNH, KQH, etc. + EK_UN = 'T', // TB(...)[...] etc. + EK_CASE = 'K', // (...)[...] etc. + EK_CALL = '(', // (0), (1), etc. + EK_CBLE = '[', // [...][...] etc. + NO_BAND_INDEX = -1, - // File option bits, from LSB in ascending bit position. - FO_DEFLATE_HINT = 1 << 0, - FO_IS_CLASS_STUB = 1 << 1, + // File option bits, from LSB in ascending bit position. + FO_DEFLATE_HINT = 1 << 0, + FO_IS_CLASS_STUB = 1 << 1, - // Archive option bits, from LSB in ascending bit position: - AO_HAVE_SPECIAL_FORMATS = 1 << 0, - AO_HAVE_CP_NUMBERS = 1 << 1, - AO_HAVE_ALL_CODE_FLAGS = 1 << 2, - AO_3_UNUSED_MBZ = 1 << 3, - AO_HAVE_FILE_HEADERS = 1 << 4, - AO_DEFLATE_HINT = 1 << 5, - AO_HAVE_FILE_MODTIME = 1 << 6, - AO_HAVE_FILE_OPTIONS = 1 << 7, - AO_HAVE_FILE_SIZE_HI = 1 << 8, - AO_HAVE_CLASS_FLAGS_HI = 1 << 9, - AO_HAVE_FIELD_FLAGS_HI = 1 << 10, - AO_HAVE_METHOD_FLAGS_HI = 1 << 11, - AO_HAVE_CODE_FLAGS_HI = 1 << 12, + // Archive option bits, from LSB in ascending bit position: + AO_HAVE_SPECIAL_FORMATS = 1 << 0, + AO_HAVE_CP_NUMBERS = 1 << 1, + AO_HAVE_ALL_CODE_FLAGS = 1 << 2, + AO_3_UNUSED_MBZ = 1 << 3, + AO_HAVE_FILE_HEADERS = 1 << 4, + AO_DEFLATE_HINT = 1 << 5, + AO_HAVE_FILE_MODTIME = 1 << 6, + AO_HAVE_FILE_OPTIONS = 1 << 7, + AO_HAVE_FILE_SIZE_HI = 1 << 8, + AO_HAVE_CLASS_FLAGS_HI = 1 << 9, + AO_HAVE_FIELD_FLAGS_HI = 1 << 10, + AO_HAVE_METHOD_FLAGS_HI = 1 << 11, + AO_HAVE_CODE_FLAGS_HI = 1 << 12, #define ARCHIVE_BIT_DO(F) \ - F(AO_HAVE_SPECIAL_FORMATS) F(AO_HAVE_CP_NUMBERS) F(AO_HAVE_ALL_CODE_FLAGS) \ - /*F(AO_3_UNUSED_MBZ)*/ \ - F(AO_HAVE_FILE_HEADERS) F(AO_DEFLATE_HINT) F(AO_HAVE_FILE_MODTIME) \ - F(AO_HAVE_FILE_OPTIONS) F(AO_HAVE_FILE_SIZE_HI) F(AO_HAVE_CLASS_FLAGS_HI) \ - F(AO_HAVE_FIELD_FLAGS_HI) F(AO_HAVE_METHOD_FLAGS_HI) F(AO_HAVE_CODE_FLAGS_HI) \ - /*(end)*/ + F(AO_HAVE_SPECIAL_FORMATS) F(AO_HAVE_CP_NUMBERS) F(AO_HAVE_ALL_CODE_FLAGS) \ + /*F(AO_3_UNUSED_MBZ)*/ \ + F(AO_HAVE_FILE_HEADERS) F(AO_DEFLATE_HINT) F(AO_HAVE_FILE_MODTIME) \ + F(AO_HAVE_FILE_OPTIONS) F(AO_HAVE_FILE_SIZE_HI) F(AO_HAVE_CLASS_FLAGS_HI) \ + F(AO_HAVE_FIELD_FLAGS_HI) F(AO_HAVE_METHOD_FLAGS_HI) F(AO_HAVE_CODE_FLAGS_HI) \ + /*(end)*/ - // Constants for decoding attribute definition header bytes. - ADH_CONTEXT_MASK = 0x3, // (hdr & ADH_CONTEXT_MASK) - ADH_BIT_SHIFT = 0x2, // (hdr >> ADH_BIT_SHIFT) - ADH_BIT_IS_LSB = 1, // (hdr >> ADH_BIT_SHIFT) - ADH_BIT_IS_LSB + // Constants for decoding attribute definition header bytes. + ADH_CONTEXT_MASK = 0x3, // (hdr & ADH_CONTEXT_MASK) + ADH_BIT_SHIFT = 0x2, // (hdr >> ADH_BIT_SHIFT) + ADH_BIT_IS_LSB = 1, // (hdr >> ADH_BIT_SHIFT) - ADH_BIT_IS_LSB #define ADH_BYTE(context, index) ((((index) + ADH_BIT_IS_LSB) << ADH_BIT_SHIFT) + (context)) #define ADH_BYTE_CONTEXT(adhb) ((adhb) & ADH_CONTEXT_MASK) #define ADH_BYTE_INDEX(adhb) (((adhb) >> ADH_BIT_SHIFT) - ADH_BIT_IS_LSB) - NO_MODTIME = 0, // nullptr modtime value + NO_MODTIME = 0, // nullptr modtime value - // meta-coding - _meta_default = 0, - _meta_canon_min = 1, - _meta_canon_max = 115, - _meta_arb = 116, - _meta_run = 117, - _meta_pop = 141, - _meta_limit = 189, - _meta_error = 255, - _xxx_1_end + // meta-coding + _meta_default = 0, + _meta_canon_min = 1, + _meta_canon_max = 115, + _meta_arb = 116, + _meta_run = 117, + _meta_pop = 141, + _meta_limit = 189, + _meta_error = 255, + _xxx_1_end }; // Bytecodes. enum { - bc_nop = 0, // 0x00 - bc_aconst_null = 1, // 0x01 - bc_iconst_m1 = 2, // 0x02 - bc_iconst_0 = 3, // 0x03 - bc_iconst_1 = 4, // 0x04 - bc_iconst_2 = 5, // 0x05 - bc_iconst_3 = 6, // 0x06 - bc_iconst_4 = 7, // 0x07 - bc_iconst_5 = 8, // 0x08 - bc_lconst_0 = 9, // 0x09 - bc_lconst_1 = 10, // 0x0a - bc_fconst_0 = 11, // 0x0b - bc_fconst_1 = 12, // 0x0c - bc_fconst_2 = 13, // 0x0d - bc_dconst_0 = 14, // 0x0e - bc_dconst_1 = 15, // 0x0f - bc_bipush = 16, // 0x10 - bc_sipush = 17, // 0x11 - bc_ldc = 18, // 0x12 - bc_ldc_w = 19, // 0x13 - bc_ldc2_w = 20, // 0x14 - bc_iload = 21, // 0x15 - bc_lload = 22, // 0x16 - bc_fload = 23, // 0x17 - bc_dload = 24, // 0x18 - bc_aload = 25, // 0x19 - bc_iload_0 = 26, // 0x1a - bc_iload_1 = 27, // 0x1b - bc_iload_2 = 28, // 0x1c - bc_iload_3 = 29, // 0x1d - bc_lload_0 = 30, // 0x1e - bc_lload_1 = 31, // 0x1f - bc_lload_2 = 32, // 0x20 - bc_lload_3 = 33, // 0x21 - bc_fload_0 = 34, // 0x22 - bc_fload_1 = 35, // 0x23 - bc_fload_2 = 36, // 0x24 - bc_fload_3 = 37, // 0x25 - bc_dload_0 = 38, // 0x26 - bc_dload_1 = 39, // 0x27 - bc_dload_2 = 40, // 0x28 - bc_dload_3 = 41, // 0x29 - bc_aload_0 = 42, // 0x2a - bc_aload_1 = 43, // 0x2b - bc_aload_2 = 44, // 0x2c - bc_aload_3 = 45, // 0x2d - bc_iaload = 46, // 0x2e - bc_laload = 47, // 0x2f - bc_faload = 48, // 0x30 - bc_daload = 49, // 0x31 - bc_aaload = 50, // 0x32 - bc_baload = 51, // 0x33 - bc_caload = 52, // 0x34 - bc_saload = 53, // 0x35 - bc_istore = 54, // 0x36 - bc_lstore = 55, // 0x37 - bc_fstore = 56, // 0x38 - bc_dstore = 57, // 0x39 - bc_astore = 58, // 0x3a - bc_istore_0 = 59, // 0x3b - bc_istore_1 = 60, // 0x3c - bc_istore_2 = 61, // 0x3d - bc_istore_3 = 62, // 0x3e - bc_lstore_0 = 63, // 0x3f - bc_lstore_1 = 64, // 0x40 - bc_lstore_2 = 65, // 0x41 - bc_lstore_3 = 66, // 0x42 - bc_fstore_0 = 67, // 0x43 - bc_fstore_1 = 68, // 0x44 - bc_fstore_2 = 69, // 0x45 - bc_fstore_3 = 70, // 0x46 - bc_dstore_0 = 71, // 0x47 - bc_dstore_1 = 72, // 0x48 - bc_dstore_2 = 73, // 0x49 - bc_dstore_3 = 74, // 0x4a - bc_astore_0 = 75, // 0x4b - bc_astore_1 = 76, // 0x4c - bc_astore_2 = 77, // 0x4d - bc_astore_3 = 78, // 0x4e - bc_iastore = 79, // 0x4f - bc_lastore = 80, // 0x50 - bc_fastore = 81, // 0x51 - bc_dastore = 82, // 0x52 - bc_aastore = 83, // 0x53 - bc_bastore = 84, // 0x54 - bc_castore = 85, // 0x55 - bc_sastore = 86, // 0x56 - bc_pop = 87, // 0x57 - bc_pop2 = 88, // 0x58 - bc_dup = 89, // 0x59 - bc_dup_x1 = 90, // 0x5a - bc_dup_x2 = 91, // 0x5b - bc_dup2 = 92, // 0x5c - bc_dup2_x1 = 93, // 0x5d - bc_dup2_x2 = 94, // 0x5e - bc_swap = 95, // 0x5f - bc_iadd = 96, // 0x60 - bc_ladd = 97, // 0x61 - bc_fadd = 98, // 0x62 - bc_dadd = 99, // 0x63 - bc_isub = 100, // 0x64 - bc_lsub = 101, // 0x65 - bc_fsub = 102, // 0x66 - bc_dsub = 103, // 0x67 - bc_imul = 104, // 0x68 - bc_lmul = 105, // 0x69 - bc_fmul = 106, // 0x6a - bc_dmul = 107, // 0x6b - bc_idiv = 108, // 0x6c - bc_ldiv = 109, // 0x6d - bc_fdiv = 110, // 0x6e - bc_ddiv = 111, // 0x6f - bc_irem = 112, // 0x70 - bc_lrem = 113, // 0x71 - bc_frem = 114, // 0x72 - bc_drem = 115, // 0x73 - bc_ineg = 116, // 0x74 - bc_lneg = 117, // 0x75 - bc_fneg = 118, // 0x76 - bc_dneg = 119, // 0x77 - bc_ishl = 120, // 0x78 - bc_lshl = 121, // 0x79 - bc_ishr = 122, // 0x7a - bc_lshr = 123, // 0x7b - bc_iushr = 124, // 0x7c - bc_lushr = 125, // 0x7d - bc_iand = 126, // 0x7e - bc_land = 127, // 0x7f - bc_ior = 128, // 0x80 - bc_lor = 129, // 0x81 - bc_ixor = 130, // 0x82 - bc_lxor = 131, // 0x83 - bc_iinc = 132, // 0x84 - bc_i2l = 133, // 0x85 - bc_i2f = 134, // 0x86 - bc_i2d = 135, // 0x87 - bc_l2i = 136, // 0x88 - bc_l2f = 137, // 0x89 - bc_l2d = 138, // 0x8a - bc_f2i = 139, // 0x8b - bc_f2l = 140, // 0x8c - bc_f2d = 141, // 0x8d - bc_d2i = 142, // 0x8e - bc_d2l = 143, // 0x8f - bc_d2f = 144, // 0x90 - bc_i2b = 145, // 0x91 - bc_i2c = 146, // 0x92 - bc_i2s = 147, // 0x93 - bc_lcmp = 148, // 0x94 - bc_fcmpl = 149, // 0x95 - bc_fcmpg = 150, // 0x96 - bc_dcmpl = 151, // 0x97 - bc_dcmpg = 152, // 0x98 - bc_ifeq = 153, // 0x99 - bc_ifne = 154, // 0x9a - bc_iflt = 155, // 0x9b - bc_ifge = 156, // 0x9c - bc_ifgt = 157, // 0x9d - bc_ifle = 158, // 0x9e - bc_if_icmpeq = 159, // 0x9f - bc_if_icmpne = 160, // 0xa0 - bc_if_icmplt = 161, // 0xa1 - bc_if_icmpge = 162, // 0xa2 - bc_if_icmpgt = 163, // 0xa3 - bc_if_icmple = 164, // 0xa4 - bc_if_acmpeq = 165, // 0xa5 - bc_if_acmpne = 166, // 0xa6 - bc_goto = 167, // 0xa7 - bc_jsr = 168, // 0xa8 - bc_ret = 169, // 0xa9 - bc_tableswitch = 170, // 0xaa - bc_lookupswitch = 171, // 0xab - bc_ireturn = 172, // 0xac - bc_lreturn = 173, // 0xad - bc_freturn = 174, // 0xae - bc_dreturn = 175, // 0xaf - bc_areturn = 176, // 0xb0 - bc_return = 177, // 0xb1 - bc_getstatic = 178, // 0xb2 - bc_putstatic = 179, // 0xb3 - bc_getfield = 180, // 0xb4 - bc_putfield = 181, // 0xb5 - bc_invokevirtual = 182, // 0xb6 - bc_invokespecial = 183, // 0xb7 - bc_invokestatic = 184, // 0xb8 - bc_invokeinterface = 185, // 0xb9 - bc_xxxunusedxxx = 186, // 0xba - bc_new = 187, // 0xbb - bc_newarray = 188, // 0xbc - bc_anewarray = 189, // 0xbd - bc_arraylength = 190, // 0xbe - bc_athrow = 191, // 0xbf - bc_checkcast = 192, // 0xc0 - bc_instanceof = 193, // 0xc1 - bc_monitorenter = 194, // 0xc2 - bc_monitorexit = 195, // 0xc3 - bc_wide = 196, // 0xc4 - bc_multianewarray = 197, // 0xc5 - bc_ifnull = 198, // 0xc6 - bc_ifnonnull = 199, // 0xc7 - bc_goto_w = 200, // 0xc8 - bc_jsr_w = 201, // 0xc9 - bc_bytecode_limit = 202 // 0xca + bc_nop = 0, // 0x00 + bc_aconst_null = 1, // 0x01 + bc_iconst_m1 = 2, // 0x02 + bc_iconst_0 = 3, // 0x03 + bc_iconst_1 = 4, // 0x04 + bc_iconst_2 = 5, // 0x05 + bc_iconst_3 = 6, // 0x06 + bc_iconst_4 = 7, // 0x07 + bc_iconst_5 = 8, // 0x08 + bc_lconst_0 = 9, // 0x09 + bc_lconst_1 = 10, // 0x0a + bc_fconst_0 = 11, // 0x0b + bc_fconst_1 = 12, // 0x0c + bc_fconst_2 = 13, // 0x0d + bc_dconst_0 = 14, // 0x0e + bc_dconst_1 = 15, // 0x0f + bc_bipush = 16, // 0x10 + bc_sipush = 17, // 0x11 + bc_ldc = 18, // 0x12 + bc_ldc_w = 19, // 0x13 + bc_ldc2_w = 20, // 0x14 + bc_iload = 21, // 0x15 + bc_lload = 22, // 0x16 + bc_fload = 23, // 0x17 + bc_dload = 24, // 0x18 + bc_aload = 25, // 0x19 + bc_iload_0 = 26, // 0x1a + bc_iload_1 = 27, // 0x1b + bc_iload_2 = 28, // 0x1c + bc_iload_3 = 29, // 0x1d + bc_lload_0 = 30, // 0x1e + bc_lload_1 = 31, // 0x1f + bc_lload_2 = 32, // 0x20 + bc_lload_3 = 33, // 0x21 + bc_fload_0 = 34, // 0x22 + bc_fload_1 = 35, // 0x23 + bc_fload_2 = 36, // 0x24 + bc_fload_3 = 37, // 0x25 + bc_dload_0 = 38, // 0x26 + bc_dload_1 = 39, // 0x27 + bc_dload_2 = 40, // 0x28 + bc_dload_3 = 41, // 0x29 + bc_aload_0 = 42, // 0x2a + bc_aload_1 = 43, // 0x2b + bc_aload_2 = 44, // 0x2c + bc_aload_3 = 45, // 0x2d + bc_iaload = 46, // 0x2e + bc_laload = 47, // 0x2f + bc_faload = 48, // 0x30 + bc_daload = 49, // 0x31 + bc_aaload = 50, // 0x32 + bc_baload = 51, // 0x33 + bc_caload = 52, // 0x34 + bc_saload = 53, // 0x35 + bc_istore = 54, // 0x36 + bc_lstore = 55, // 0x37 + bc_fstore = 56, // 0x38 + bc_dstore = 57, // 0x39 + bc_astore = 58, // 0x3a + bc_istore_0 = 59, // 0x3b + bc_istore_1 = 60, // 0x3c + bc_istore_2 = 61, // 0x3d + bc_istore_3 = 62, // 0x3e + bc_lstore_0 = 63, // 0x3f + bc_lstore_1 = 64, // 0x40 + bc_lstore_2 = 65, // 0x41 + bc_lstore_3 = 66, // 0x42 + bc_fstore_0 = 67, // 0x43 + bc_fstore_1 = 68, // 0x44 + bc_fstore_2 = 69, // 0x45 + bc_fstore_3 = 70, // 0x46 + bc_dstore_0 = 71, // 0x47 + bc_dstore_1 = 72, // 0x48 + bc_dstore_2 = 73, // 0x49 + bc_dstore_3 = 74, // 0x4a + bc_astore_0 = 75, // 0x4b + bc_astore_1 = 76, // 0x4c + bc_astore_2 = 77, // 0x4d + bc_astore_3 = 78, // 0x4e + bc_iastore = 79, // 0x4f + bc_lastore = 80, // 0x50 + bc_fastore = 81, // 0x51 + bc_dastore = 82, // 0x52 + bc_aastore = 83, // 0x53 + bc_bastore = 84, // 0x54 + bc_castore = 85, // 0x55 + bc_sastore = 86, // 0x56 + bc_pop = 87, // 0x57 + bc_pop2 = 88, // 0x58 + bc_dup = 89, // 0x59 + bc_dup_x1 = 90, // 0x5a + bc_dup_x2 = 91, // 0x5b + bc_dup2 = 92, // 0x5c + bc_dup2_x1 = 93, // 0x5d + bc_dup2_x2 = 94, // 0x5e + bc_swap = 95, // 0x5f + bc_iadd = 96, // 0x60 + bc_ladd = 97, // 0x61 + bc_fadd = 98, // 0x62 + bc_dadd = 99, // 0x63 + bc_isub = 100, // 0x64 + bc_lsub = 101, // 0x65 + bc_fsub = 102, // 0x66 + bc_dsub = 103, // 0x67 + bc_imul = 104, // 0x68 + bc_lmul = 105, // 0x69 + bc_fmul = 106, // 0x6a + bc_dmul = 107, // 0x6b + bc_idiv = 108, // 0x6c + bc_ldiv = 109, // 0x6d + bc_fdiv = 110, // 0x6e + bc_ddiv = 111, // 0x6f + bc_irem = 112, // 0x70 + bc_lrem = 113, // 0x71 + bc_frem = 114, // 0x72 + bc_drem = 115, // 0x73 + bc_ineg = 116, // 0x74 + bc_lneg = 117, // 0x75 + bc_fneg = 118, // 0x76 + bc_dneg = 119, // 0x77 + bc_ishl = 120, // 0x78 + bc_lshl = 121, // 0x79 + bc_ishr = 122, // 0x7a + bc_lshr = 123, // 0x7b + bc_iushr = 124, // 0x7c + bc_lushr = 125, // 0x7d + bc_iand = 126, // 0x7e + bc_land = 127, // 0x7f + bc_ior = 128, // 0x80 + bc_lor = 129, // 0x81 + bc_ixor = 130, // 0x82 + bc_lxor = 131, // 0x83 + bc_iinc = 132, // 0x84 + bc_i2l = 133, // 0x85 + bc_i2f = 134, // 0x86 + bc_i2d = 135, // 0x87 + bc_l2i = 136, // 0x88 + bc_l2f = 137, // 0x89 + bc_l2d = 138, // 0x8a + bc_f2i = 139, // 0x8b + bc_f2l = 140, // 0x8c + bc_f2d = 141, // 0x8d + bc_d2i = 142, // 0x8e + bc_d2l = 143, // 0x8f + bc_d2f = 144, // 0x90 + bc_i2b = 145, // 0x91 + bc_i2c = 146, // 0x92 + bc_i2s = 147, // 0x93 + bc_lcmp = 148, // 0x94 + bc_fcmpl = 149, // 0x95 + bc_fcmpg = 150, // 0x96 + bc_dcmpl = 151, // 0x97 + bc_dcmpg = 152, // 0x98 + bc_ifeq = 153, // 0x99 + bc_ifne = 154, // 0x9a + bc_iflt = 155, // 0x9b + bc_ifge = 156, // 0x9c + bc_ifgt = 157, // 0x9d + bc_ifle = 158, // 0x9e + bc_if_icmpeq = 159, // 0x9f + bc_if_icmpne = 160, // 0xa0 + bc_if_icmplt = 161, // 0xa1 + bc_if_icmpge = 162, // 0xa2 + bc_if_icmpgt = 163, // 0xa3 + bc_if_icmple = 164, // 0xa4 + bc_if_acmpeq = 165, // 0xa5 + bc_if_acmpne = 166, // 0xa6 + bc_goto = 167, // 0xa7 + bc_jsr = 168, // 0xa8 + bc_ret = 169, // 0xa9 + bc_tableswitch = 170, // 0xaa + bc_lookupswitch = 171, // 0xab + bc_ireturn = 172, // 0xac + bc_lreturn = 173, // 0xad + bc_freturn = 174, // 0xae + bc_dreturn = 175, // 0xaf + bc_areturn = 176, // 0xb0 + bc_return = 177, // 0xb1 + bc_getstatic = 178, // 0xb2 + bc_putstatic = 179, // 0xb3 + bc_getfield = 180, // 0xb4 + bc_putfield = 181, // 0xb5 + bc_invokevirtual = 182, // 0xb6 + bc_invokespecial = 183, // 0xb7 + bc_invokestatic = 184, // 0xb8 + bc_invokeinterface = 185, // 0xb9 + bc_xxxunusedxxx = 186, // 0xba + bc_new = 187, // 0xbb + bc_newarray = 188, // 0xbc + bc_anewarray = 189, // 0xbd + bc_arraylength = 190, // 0xbe + bc_athrow = 191, // 0xbf + bc_checkcast = 192, // 0xc0 + bc_instanceof = 193, // 0xc1 + bc_monitorenter = 194, // 0xc2 + bc_monitorexit = 195, // 0xc3 + bc_wide = 196, // 0xc4 + bc_multianewarray = 197, // 0xc5 + bc_ifnull = 198, // 0xc6 + bc_ifnonnull = 199, // 0xc7 + bc_goto_w = 200, // 0xc8 + bc_jsr_w = 201, // 0xc9 + bc_bytecode_limit = 202 // 0xca }; enum { - bc_end_marker = 255, - bc_byte_escape = 254, - bc_ref_escape = 253, - _first_linker_op = bc_getstatic, - _last_linker_op = bc_invokestatic, - _num_linker_ops = (_last_linker_op - _first_linker_op) + 1, - _self_linker_op = bc_bytecode_limit, - _self_linker_aload_flag = 1 * _num_linker_ops, - _self_linker_super_flag = 2 * _num_linker_ops, - _self_linker_limit = _self_linker_op + 4 * _num_linker_ops, - _invokeinit_op = _self_linker_limit, - _invokeinit_self_option = 0, - _invokeinit_super_option = 1, - _invokeinit_new_option = 2, - _invokeinit_limit = _invokeinit_op + 3, - _xldc_op = _invokeinit_limit, - bc_aldc = bc_ldc, - bc_cldc = _xldc_op + 0, - bc_ildc = _xldc_op + 1, - bc_fldc = _xldc_op + 2, - bc_aldc_w = bc_ldc_w, - bc_cldc_w = _xldc_op + 3, - bc_ildc_w = _xldc_op + 4, - bc_fldc_w = _xldc_op + 5, - bc_lldc2_w = bc_ldc2_w, - bc_dldc2_w = _xldc_op + 6, - _xldc_limit = _xldc_op + 7, - _xxx_3_end + bc_end_marker = 255, + bc_byte_escape = 254, + bc_ref_escape = 253, + _first_linker_op = bc_getstatic, + _last_linker_op = bc_invokestatic, + _num_linker_ops = (_last_linker_op - _first_linker_op) + 1, + _self_linker_op = bc_bytecode_limit, + _self_linker_aload_flag = 1 * _num_linker_ops, + _self_linker_super_flag = 2 * _num_linker_ops, + _self_linker_limit = _self_linker_op + 4 * _num_linker_ops, + _invokeinit_op = _self_linker_limit, + _invokeinit_self_option = 0, + _invokeinit_super_option = 1, + _invokeinit_new_option = 2, + _invokeinit_limit = _invokeinit_op + 3, + _xldc_op = _invokeinit_limit, + bc_aldc = bc_ldc, + bc_cldc = _xldc_op + 0, + bc_ildc = _xldc_op + 1, + bc_fldc = _xldc_op + 2, + bc_aldc_w = bc_ldc_w, + bc_cldc_w = _xldc_op + 3, + bc_ildc_w = _xldc_op + 4, + bc_fldc_w = _xldc_op + 5, + bc_lldc2_w = bc_ldc2_w, + bc_dldc2_w = _xldc_op + 6, + _xldc_limit = _xldc_op + 7, + _xxx_3_end }; diff --git a/libraries/pack200/src/unpack.cpp b/libraries/pack200/src/unpack.cpp index 55d253b2..9c4c633c 100644 --- a/libraries/pack200/src/unpack.cpp +++ b/libraries/pack200/src/unpack.cpp @@ -34,13 +34,7 @@ * _LP64 can be explicitly set (used on Linux). * Solaris compilers will define __sparcv9 or __x86_64 on 64bit compilations. */ -#if defined(_LP64) || defined(__sparcv9) || defined(__x86_64) -#define LONG_LONG_FORMAT "%ld" -#define LONG_LONG_HEX_FORMAT "%lx" -#else -#define LONG_LONG_FORMAT "%lld" -#define LONG_LONG_HEX_FORMAT "%016llx" -#endif +#include #include @@ -67,248 +61,248 @@ // tags, in canonical order: static const byte TAGS_IN_ORDER[] = { - CONSTANT_Utf8, CONSTANT_Integer, CONSTANT_Float, CONSTANT_Long, - CONSTANT_Double, CONSTANT_String, CONSTANT_Class, CONSTANT_Signature, - CONSTANT_NameandType, CONSTANT_Fieldref, CONSTANT_Methodref, CONSTANT_InterfaceMethodref}; + CONSTANT_Utf8, CONSTANT_Integer, CONSTANT_Float, CONSTANT_Long, + CONSTANT_Double, CONSTANT_String, CONSTANT_Class, CONSTANT_Signature, + CONSTANT_NameandType, CONSTANT_Fieldref, CONSTANT_Methodref, CONSTANT_InterfaceMethodref}; #define N_TAGS_IN_ORDER (sizeof TAGS_IN_ORDER) // REQUESTED must be -2 for u2 and REQUESTED_LDC must be -1 for u1 enum { - NOT_REQUESTED = 0, - REQUESTED = -2, - REQUESTED_LDC = -1 + NOT_REQUESTED = 0, + REQUESTED = -2, + REQUESTED_LDC = -1 }; #define NO_INORD ((uint32_t) - 1) struct entry { - byte tag; - unsigned short nrefs; // pack w/ tag - - int outputIndex; - uint32_t inord; // &cp.entries[cp.tag_base[this->tag]+this->inord] == this - - entry **refs; - - // put last to pack best - union - { - bytes b; - int i; - int64_t l; - } value; - - void requestOutputIndex(constant_pool &cp, int req = REQUESTED); - int getOutputIndex() - { - assert(outputIndex > NOT_REQUESTED); - return outputIndex; - } - - entry *ref(int refnum) - { - assert((uint32_t)refnum < nrefs); - return refs[refnum]; - } - - const char *utf8String() - { - assert(tagMatches(CONSTANT_Utf8)); - assert(value.b.len == strlen((const char *)value.b.ptr)); - return (const char *)value.b.ptr; - } - - entry *className() - { - assert(tagMatches(CONSTANT_Class)); - return ref(0); - } - - entry *memberClass() - { - assert(tagMatches(CONSTANT_Member)); - return ref(0); - } - - entry *memberDescr() - { - assert(tagMatches(CONSTANT_Member)); - return ref(1); - } - - entry *descrName() - { - assert(tagMatches(CONSTANT_NameandType)); - return ref(0); - } - - entry *descrType() - { - assert(tagMatches(CONSTANT_NameandType)); - return ref(1); - } - - int typeSize(); - - bytes &asUtf8(); - int asInteger() - { - assert(tag == CONSTANT_Integer); - return value.i; - } - - bool isUtf8(bytes &b) - { - return tagMatches(CONSTANT_Utf8) && value.b.equals(b); - } - - bool isDoubleWord() - { - return tag == CONSTANT_Double || tag == CONSTANT_Long; - } - - bool tagMatches(byte tag2) - { - return (tag2 == tag) || (tag2 == CONSTANT_Utf8 && tag == CONSTANT_Signature) || - (tag2 == CONSTANT_Literal && tag >= CONSTANT_Integer && tag <= CONSTANT_String && - tag != CONSTANT_Class) || - (tag2 == CONSTANT_Member && tag >= CONSTANT_Fieldref && - tag <= CONSTANT_InterfaceMethodref); - } + byte tag; + unsigned short nrefs; // pack w/ tag + + int outputIndex; + uint32_t inord; // &cp.entries[cp.tag_base[this->tag]+this->inord] == this + + entry **refs; + + // put last to pack best + union + { + bytes b; + int i; + int64_t l; + } value; + + void requestOutputIndex(constant_pool &cp, int req = REQUESTED); + int getOutputIndex() + { + assert(outputIndex > NOT_REQUESTED); + return outputIndex; + } + + entry *ref(int refnum) + { + assert((uint32_t)refnum < nrefs); + return refs[refnum]; + } + + const char *utf8String() + { + assert(tagMatches(CONSTANT_Utf8)); + assert(value.b.len == strlen((const char *)value.b.ptr)); + return (const char *)value.b.ptr; + } + + entry *className() + { + assert(tagMatches(CONSTANT_Class)); + return ref(0); + } + + entry *memberClass() + { + assert(tagMatches(CONSTANT_Member)); + return ref(0); + } + + entry *memberDescr() + { + assert(tagMatches(CONSTANT_Member)); + return ref(1); + } + + entry *descrName() + { + assert(tagMatches(CONSTANT_NameandType)); + return ref(0); + } + + entry *descrType() + { + assert(tagMatches(CONSTANT_NameandType)); + return ref(1); + } + + int typeSize(); + + bytes &asUtf8(); + int asInteger() + { + assert(tag == CONSTANT_Integer); + return value.i; + } + + bool isUtf8(bytes &b) + { + return tagMatches(CONSTANT_Utf8) && value.b.equals(b); + } + + bool isDoubleWord() + { + return tag == CONSTANT_Double || tag == CONSTANT_Long; + } + + bool tagMatches(byte tag2) + { + return (tag2 == tag) || (tag2 == CONSTANT_Utf8 && tag == CONSTANT_Signature) || + (tag2 == CONSTANT_Literal && tag >= CONSTANT_Integer && tag <= CONSTANT_String && + tag != CONSTANT_Class) || + (tag2 == CONSTANT_Member && tag >= CONSTANT_Fieldref && + tag <= CONSTANT_InterfaceMethodref); + } }; entry *cpindex::get(uint32_t i) { - if (i >= len) - return nullptr; - else if (base1 != nullptr) - // primary index - return &base1[i]; - else - // secondary index - return base2[i]; + if (i >= len) + return nullptr; + else if (base1 != nullptr) + // primary index + return &base1[i]; + else + // secondary index + return base2[i]; } inline bytes &entry::asUtf8() { - assert(tagMatches(CONSTANT_Utf8)); - return value.b; + assert(tagMatches(CONSTANT_Utf8)); + return value.b; } int entry::typeSize() { - assert(tagMatches(CONSTANT_Utf8)); - const char *sigp = (char *)value.b.ptr; - switch (*sigp) - { - case '(': - sigp++; - break; // skip opening '(' - case 'D': - case 'J': - return 2; // double field - default: - return 1; // field - } - int siglen = 0; - for (;;) - { - int ch = *sigp++; - switch (ch) - { - case 'D': - case 'J': - siglen += 1; - break; - case '[': - // Skip rest of array info. - while (ch == '[') - { - ch = *sigp++; - } - if (ch != 'L') - break; - // else fall through - case 'L': - sigp = strchr(sigp, ';'); - if (sigp == nullptr) - { - unpack_abort("bad data"); - return 0; - } - sigp += 1; - break; - case ')': // closing ')' - return siglen; - } - siglen += 1; - } + assert(tagMatches(CONSTANT_Utf8)); + const char *sigp = (char *)value.b.ptr; + switch (*sigp) + { + case '(': + sigp++; + break; // skip opening '(' + case 'D': + case 'J': + return 2; // double field + default: + return 1; // field + } + int siglen = 0; + for (;;) + { + int ch = *sigp++; + switch (ch) + { + case 'D': + case 'J': + siglen += 1; + break; + case '[': + // Skip rest of array info. + while (ch == '[') + { + ch = *sigp++; + } + if (ch != 'L') + break; + // else fall through + case 'L': + sigp = strchr(sigp, ';'); + if (sigp == nullptr) + { + unpack_abort("bad data"); + return 0; + } + sigp += 1; + break; + case ')': // closing ')' + return siglen; + } + siglen += 1; + } } inline cpindex *constant_pool::getFieldIndex(entry *classRef) { - assert(classRef->tagMatches(CONSTANT_Class)); - assert((uint32_t)classRef->inord < (uint32_t)tag_count[CONSTANT_Class]); - return &member_indexes[classRef->inord * 2 + 0]; + assert(classRef->tagMatches(CONSTANT_Class)); + assert((uint32_t)classRef->inord < (uint32_t)tag_count[CONSTANT_Class]); + return &member_indexes[classRef->inord * 2 + 0]; } inline cpindex *constant_pool::getMethodIndex(entry *classRef) { - assert(classRef->tagMatches(CONSTANT_Class)); - assert((uint32_t)classRef->inord < (uint32_t)tag_count[CONSTANT_Class]); - return &member_indexes[classRef->inord * 2 + 1]; + assert(classRef->tagMatches(CONSTANT_Class)); + assert((uint32_t)classRef->inord < (uint32_t)tag_count[CONSTANT_Class]); + return &member_indexes[classRef->inord * 2 + 1]; } struct inner_class { - entry *inner; - entry *outer; - entry *name; - int flags; - inner_class *next_sibling; - bool requested; + entry *inner; + entry *outer; + entry *name; + int flags; + inner_class *next_sibling; + bool requested; }; // Here is where everything gets deallocated: void unpacker::free() { - int i; - if (jarout != nullptr) - jarout->reset(); - if (gzin != nullptr) - { - gzin->free(); - gzin = nullptr; - } - if (free_input) - input.free(); - /* - * free everybody ever allocated with U_NEW or (recently) with T_NEW - */ - assert(smallbuf.base() == nullptr || mallocs.contains(smallbuf.base())); - assert(tsmallbuf.base() == nullptr || tmallocs.contains(tsmallbuf.base())); - mallocs.freeAll(); - tmallocs.freeAll(); - smallbuf.init(); - tsmallbuf.init(); - bcimap.free(); - class_fixup_type.free(); - class_fixup_offset.free(); - class_fixup_ref.free(); - code_fixup_type.free(); - code_fixup_offset.free(); - code_fixup_source.free(); - requested_ics.free(); - cur_classfile_head.free(); - cur_classfile_tail.free(); - for (i = 0; i < ATTR_CONTEXT_LIMIT; i++) - attr_defs[i].free(); - - // free CP state - cp.outputEntries.free(); - for (i = 0; i < CONSTANT_Limit; i++) - cp.tag_extras[i].free(); + int i; + if (jarout != nullptr) + jarout->reset(); + if (gzin != nullptr) + { + gzin->free(); + gzin = nullptr; + } + if (free_input) + input.free(); + /* + * free everybody ever allocated with U_NEW or (recently) with T_NEW + */ + assert(smallbuf.base() == nullptr || mallocs.contains(smallbuf.base())); + assert(tsmallbuf.base() == nullptr || tmallocs.contains(tsmallbuf.base())); + mallocs.freeAll(); + tmallocs.freeAll(); + smallbuf.init(); + tsmallbuf.init(); + bcimap.free(); + class_fixup_type.free(); + class_fixup_offset.free(); + class_fixup_ref.free(); + code_fixup_type.free(); + code_fixup_offset.free(); + code_fixup_source.free(); + requested_ics.free(); + cur_classfile_head.free(); + cur_classfile_tail.free(); + for (i = 0; i < ATTR_CONTEXT_LIMIT; i++) + attr_defs[i].free(); + + // free CP state + cp.outputEntries.free(); + for (i = 0; i < CONSTANT_Limit; i++) + cp.tag_extras[i].free(); } // input handling @@ -318,1933 +312,1933 @@ void unpacker::free() // unless rplimit hits input.limit(). bool unpacker::ensure_input(int64_t more) { - uint64_t want = more - input_remaining(); - if ((int64_t)want <= 0) - return true; // it's already in the buffer - if (rplimit == input.limit()) - return true; // not expecting any more - - if (read_input_fn == nullptr) - { - // assume it is already all there - bytes_read += input.limit() - rplimit; - rplimit = input.limit(); - return true; - } - - uint64_t remaining = (input.limit() - rplimit); // how much left to read? - byte *rpgoal = (want >= remaining) ? input.limit() : rplimit + (size_t)want; - enum - { - CHUNK_SIZE = (1 << 14) - }; - uint64_t fetch = want; - if (fetch < CHUNK_SIZE) - fetch = CHUNK_SIZE; - if (fetch > remaining * 3 / 4) - fetch = remaining; - // Try to fetch at least "more" bytes. - while ((int64_t)fetch > 0) - { - int64_t nr = (*read_input_fn)(this, rplimit, fetch, remaining); - if (nr <= 0) - { - return (rplimit >= rpgoal); - } - remaining -= nr; - rplimit += nr; - fetch -= nr; - bytes_read += nr; - assert(remaining == (uint64_t)(input.limit() - rplimit)); - } - return true; + uint64_t want = more - input_remaining(); + if ((int64_t)want <= 0) + return true; // it's already in the buffer + if (rplimit == input.limit()) + return true; // not expecting any more + + if (read_input_fn == nullptr) + { + // assume it is already all there + bytes_read += input.limit() - rplimit; + rplimit = input.limit(); + return true; + } + + uint64_t remaining = (input.limit() - rplimit); // how much left to read? + byte *rpgoal = (want >= remaining) ? input.limit() : rplimit + (size_t)want; + enum + { + CHUNK_SIZE = (1 << 14) + }; + uint64_t fetch = want; + if (fetch < CHUNK_SIZE) + fetch = CHUNK_SIZE; + if (fetch > remaining * 3 / 4) + fetch = remaining; + // Try to fetch at least "more" bytes. + while ((int64_t)fetch > 0) + { + int64_t nr = (*read_input_fn)(this, rplimit, fetch, remaining); + if (nr <= 0) + { + return (rplimit >= rpgoal); + } + remaining -= nr; + rplimit += nr; + fetch -= nr; + bytes_read += nr; + assert(remaining == (uint64_t)(input.limit() - rplimit)); + } + return true; } // output handling fillbytes *unpacker::close_output(fillbytes *which) { - assert(wp != nullptr); - if (which == nullptr) - { - if (wpbase == cur_classfile_head.base()) - { - which = &cur_classfile_head; - } - else - { - which = &cur_classfile_tail; - } - } - assert(wpbase == which->base()); - assert(wplimit == which->end()); - which->setLimit(wp); - wp = nullptr; - wplimit = nullptr; - // wpbase = nullptr; - return which; + assert(wp != nullptr); + if (which == nullptr) + { + if (wpbase == cur_classfile_head.base()) + { + which = &cur_classfile_head; + } + else + { + which = &cur_classfile_tail; + } + } + assert(wpbase == which->base()); + assert(wplimit == which->end()); + which->setLimit(wp); + wp = nullptr; + wplimit = nullptr; + // wpbase = nullptr; + return which; } // maybe_inline void unpacker::ensure_put_space(size_t size) { - if (wp + size <= wplimit) - return; - // Determine which segment needs expanding. - fillbytes *which = close_output(); - byte *wp0 = which->grow(size); - wpbase = which->base(); - wplimit = which->end(); - wp = wp0; + if (wp + size <= wplimit) + return; + // Determine which segment needs expanding. + fillbytes *which = close_output(); + byte *wp0 = which->grow(size); + wpbase = which->base(); + wplimit = which->end(); + wp = wp0; } byte *unpacker::put_space(size_t size) { - byte *wp0 = wp; - byte *wp1 = wp0 + size; - if (wp1 > wplimit) - { - ensure_put_space(size); - wp0 = wp; - wp1 = wp0 + size; - } - wp = wp1; - return wp0; + byte *wp0 = wp; + byte *wp1 = wp0 + size; + if (wp1 > wplimit) + { + ensure_put_space(size); + wp0 = wp; + wp1 = wp0 + size; + } + wp = wp1; + return wp0; } void unpacker::putu2_at(byte *wp, int n) { - if (n != (unsigned short)n) - { - unpack_abort(ERROR_OVERFLOW); - return; - } - wp[0] = (n) >> 8; - wp[1] = (n) >> 0; + if (n != (unsigned short)n) + { + unpack_abort(ERROR_OVERFLOW); + return; + } + wp[0] = (n) >> 8; + wp[1] = (n) >> 0; } void unpacker::putu4_at(byte *wp, int n) { - wp[0] = (n) >> 24; - wp[1] = (n) >> 16; - wp[2] = (n) >> 8; - wp[3] = (n) >> 0; + wp[0] = (n) >> 24; + wp[1] = (n) >> 16; + wp[2] = (n) >> 8; + wp[3] = (n) >> 0; } void unpacker::putu8_at(byte *wp, int64_t n) { - putu4_at(wp + 0, (int)((uint64_t)n >> 32)); - putu4_at(wp + 4, (int)((uint64_t)n >> 0)); + putu4_at(wp + 0, (int)((uint64_t)n >> 32)); + putu4_at(wp + 4, (int)((uint64_t)n >> 0)); } void unpacker::putu2(int n) { - putu2_at(put_space(2), n); + putu2_at(put_space(2), n); } void unpacker::putu4(int n) { - putu4_at(put_space(4), n); + putu4_at(put_space(4), n); } void unpacker::putu8(int64_t n) { - putu8_at(put_space(8), n); + putu8_at(put_space(8), n); } int unpacker::putref_index(entry *e, int size) { - if (e == nullptr) - return 0; - else if (e->outputIndex > NOT_REQUESTED) - return e->outputIndex; - else if (e->tag == CONSTANT_Signature) - return putref_index(e->ref(0), size); - else - { - e->requestOutputIndex(cp, -size); - // Later on we'll fix the bits. - class_fixup_type.addByte(size); - class_fixup_offset.add((int)wpoffset()); - class_fixup_ref.add(e); - return 0; - } + if (e == nullptr) + return 0; + else if (e->outputIndex > NOT_REQUESTED) + return e->outputIndex; + else if (e->tag == CONSTANT_Signature) + return putref_index(e->ref(0), size); + else + { + e->requestOutputIndex(cp, -size); + // Later on we'll fix the bits. + class_fixup_type.addByte(size); + class_fixup_offset.add((int)wpoffset()); + class_fixup_ref.add(e); + return 0; + } } void unpacker::putref(entry *e) { - int oidx = putref_index(e, 2); - putu2_at(put_space(2), oidx); + int oidx = putref_index(e, 2); + putu2_at(put_space(2), oidx); } void unpacker::putu1ref(entry *e) { - int oidx = putref_index(e, 1); - putu1_at(put_space(1), oidx); + int oidx = putref_index(e, 1); + putu1_at(put_space(1), oidx); } // Allocation of small and large blocks. enum { - CHUNK = (1 << 14), - SMALL = (1 << 9) + CHUNK = (1 << 14), + SMALL = (1 << 9) }; // Call malloc. Try to combine small blocks and free much later. void *unpacker::alloc_heap(size_t size, bool smallOK, bool temp) { - if (!smallOK || size > SMALL) - { - void *res = must_malloc((int)size); - (temp ? &tmallocs : &mallocs)->add(res); - return res; - } - fillbytes &xsmallbuf = *(temp ? &tsmallbuf : &smallbuf); - if (!xsmallbuf.canAppend(size + 1)) - { - xsmallbuf.init(CHUNK); - (temp ? &tmallocs : &mallocs)->add(xsmallbuf.base()); - } - int growBy = (int)size; - growBy += -growBy & 7; // round up mod 8 - return xsmallbuf.grow(growBy); + if (!smallOK || size > SMALL) + { + void *res = must_malloc((int)size); + (temp ? &tmallocs : &mallocs)->add(res); + return res; + } + fillbytes &xsmallbuf = *(temp ? &tsmallbuf : &smallbuf); + if (!xsmallbuf.canAppend(size + 1)) + { + xsmallbuf.init(CHUNK); + (temp ? &tmallocs : &mallocs)->add(xsmallbuf.base()); + } + int growBy = (int)size; + growBy += -growBy & 7; // round up mod 8 + return xsmallbuf.grow(growBy); } void unpacker::saveTo(bytes &b, byte *ptr, size_t len) { - b.ptr = U_NEW(byte, add_size(len, 1)); - b.len = len; - b.copyFrom(ptr, len); + b.ptr = U_NEW(byte, add_size(len, 1)); + b.len = len; + b.copyFrom(ptr, len); } // Read up through band_headers. // Do the archive_size dance to set the size of the input mega-buffer. void unpacker::read_file_header() { - // Read file header to determine file type and total size. - enum - { - MAGIC_BYTES = 4, - AH_LENGTH_0 = 3, // minver, majver, options are outside of archive_size - AH_LENGTH_0_MAX = AH_LENGTH_0 + 1, // options might have 2 bytes - AH_LENGTH = 26, // maximum archive header length (w/ all fields) - // Length contributions from optional header fields: - AH_FILE_HEADER_LEN = 5, // sizehi/lo/next/modtime/files - AH_ARCHIVE_SIZE_LEN = 2, // sizehi/lo only; part of AH_FILE_HEADER_LEN - AH_CP_NUMBER_LEN = 4, // int/float/long/double - AH_SPECIAL_FORMAT_LEN = 2, // layouts/band-headers - AH_LENGTH_MIN = - AH_LENGTH - (AH_FILE_HEADER_LEN + AH_SPECIAL_FORMAT_LEN + AH_CP_NUMBER_LEN), - ARCHIVE_SIZE_MIN = AH_LENGTH_MIN - (AH_LENGTH_0 + AH_ARCHIVE_SIZE_LEN), - FIRST_READ = MAGIC_BYTES + AH_LENGTH_MIN - }; - - assert(AH_LENGTH_MIN == 15); // # of UNSIGNED5 fields required after archive_magic - assert(ARCHIVE_SIZE_MIN == 10); // # of UNSIGNED5 fields required after archive_size - // An absolute minimum nullptr archive is magic[4], {minver,majver,options}[3], - // archive_size[0], cp_counts[8], class_counts[4], for a total of 19 bytes. - // (Note that archive_size is optional; it may be 0..10 bytes in length.) - // The first read must capture everything up through the options field. - // This happens to work even if {minver,majver,options} is a pathological - // 15 bytes long. Legal pack files limit those three fields to 1+1+2 bytes. - assert(FIRST_READ >= MAGIC_BYTES + AH_LENGTH_0 * B_MAX); - - // Up through archive_size, the largest possible archive header is - // magic[4], {minver,majver,options}[4], archive_size[10]. - // (Note only the low 12 bits of options are allowed to be non-zero.) - // In order to parse archive_size, we need at least this many bytes - // in the first read. Of course, if archive_size_hi is more than - // a byte, we probably will fail to allocate the buffer, since it - // will be many gigabytes long. This is a practical, not an - // architectural limit to Pack200 archive sizes. - assert(FIRST_READ >= MAGIC_BYTES + AH_LENGTH_0_MAX + 2 * B_MAX); - - bool foreign_buf = (read_input_fn == nullptr); - byte initbuf[(int)FIRST_READ + (int)C_SLOP + 200]; // 200 is for JAR I/O - if (foreign_buf) - { - // inbytes is all there is - input.set(inbytes); - rp = input.base(); - rplimit = input.limit(); - } - else - { - // inbytes, if not empty, contains some read-ahead we must use first - // ensure_input will take care of copying it into initbuf, - // then querying read_input_fn for any additional data needed. - // However, the caller must assume that we use up all of inbytes. - // There is no way to tell the caller that we used only part of them. - // Therefore, the caller must use only a bare minimum of read-ahead. - if (inbytes.len > FIRST_READ) - { - unpack_abort("too much read-ahead"); - } - input.set(initbuf, sizeof(initbuf)); - input.b.clear(); - input.b.copyFrom(inbytes); - rplimit = rp = input.base(); - rplimit += inbytes.len; - bytes_read += inbytes.len; - } - // Read only 19 bytes, which is certain to contain #archive_options fields, - // but is certain not to overflow past the archive_header. - input.b.len = FIRST_READ; - if (!ensure_input(FIRST_READ)) - unpack_abort("EOF reading archive magic number"); - - if (rp[0] == 'P' && rp[1] == 'K') - { - // In the Unix-style program, we simply simulate a copy command. - // Copy until EOF; assume the JAR file is the last segment. - fprintf(stderr, "Copy-mode.\n"); - for (;;) - { - jarout->write_data(rp, (int)input_remaining()); - if (foreign_buf) - break; // one-time use of a passed in buffer - if (input.size() < CHUNK) - { - // Get some breathing room. - input.set(U_NEW(byte, (size_t)CHUNK + C_SLOP), (size_t)CHUNK); - } - rp = rplimit = input.base(); - if (!ensure_input(1)) - break; - } - jarout->closeJarFile(false); - return; - } - - // Read the magic number. - magic = 0; - for (int i1 = 0; i1 < (int)sizeof(magic); i1++) - { - magic <<= 8; - magic += (*rp++ & 0xFF); - } - - // Read the first 3 values from the header. - value_stream hdr; - int hdrVals = 0; - int hdrValsSkipped = 0; // debug only - hdr.init(rp, rplimit, UNSIGNED5_spec); - minver = hdr.getInt(); - majver = hdr.getInt(); - hdrVals += 2; - - if (magic != (int)JAVA_PACKAGE_MAGIC || - (majver != JAVA5_PACKAGE_MAJOR_VERSION && majver != JAVA6_PACKAGE_MAJOR_VERSION) || - (minver != JAVA5_PACKAGE_MINOR_VERSION && minver != JAVA6_PACKAGE_MINOR_VERSION)) - { - char message[200]; - sprintf(message, "@" ERROR_FORMAT ": magic/ver = " - "%08X/%d.%d should be %08X/%d.%d OR %08X/%d.%d\n", - magic, majver, minver, JAVA_PACKAGE_MAGIC, JAVA5_PACKAGE_MAJOR_VERSION, - JAVA5_PACKAGE_MINOR_VERSION, JAVA_PACKAGE_MAGIC, JAVA6_PACKAGE_MAJOR_VERSION, - JAVA6_PACKAGE_MINOR_VERSION); - unpack_abort(message); - } - - archive_options = hdr.getInt(); - hdrVals += 1; - assert(hdrVals == AH_LENGTH_0); // first three fields only + // Read file header to determine file type and total size. + enum + { + MAGIC_BYTES = 4, + AH_LENGTH_0 = 3, // minver, majver, options are outside of archive_size + AH_LENGTH_0_MAX = AH_LENGTH_0 + 1, // options might have 2 bytes + AH_LENGTH = 26, // maximum archive header length (w/ all fields) + // Length contributions from optional header fields: + AH_FILE_HEADER_LEN = 5, // sizehi/lo/next/modtime/files + AH_ARCHIVE_SIZE_LEN = 2, // sizehi/lo only; part of AH_FILE_HEADER_LEN + AH_CP_NUMBER_LEN = 4, // int/float/long/double + AH_SPECIAL_FORMAT_LEN = 2, // layouts/band-headers + AH_LENGTH_MIN = + AH_LENGTH - (AH_FILE_HEADER_LEN + AH_SPECIAL_FORMAT_LEN + AH_CP_NUMBER_LEN), + ARCHIVE_SIZE_MIN = AH_LENGTH_MIN - (AH_LENGTH_0 + AH_ARCHIVE_SIZE_LEN), + FIRST_READ = MAGIC_BYTES + AH_LENGTH_MIN + }; + + assert(AH_LENGTH_MIN == 15); // # of UNSIGNED5 fields required after archive_magic + assert(ARCHIVE_SIZE_MIN == 10); // # of UNSIGNED5 fields required after archive_size + // An absolute minimum nullptr archive is magic[4], {minver,majver,options}[3], + // archive_size[0], cp_counts[8], class_counts[4], for a total of 19 bytes. + // (Note that archive_size is optional; it may be 0..10 bytes in length.) + // The first read must capture everything up through the options field. + // This happens to work even if {minver,majver,options} is a pathological + // 15 bytes long. Legal pack files limit those three fields to 1+1+2 bytes. + assert(FIRST_READ >= MAGIC_BYTES + AH_LENGTH_0 * B_MAX); + + // Up through archive_size, the largest possible archive header is + // magic[4], {minver,majver,options}[4], archive_size[10]. + // (Note only the low 12 bits of options are allowed to be non-zero.) + // In order to parse archive_size, we need at least this many bytes + // in the first read. Of course, if archive_size_hi is more than + // a byte, we probably will fail to allocate the buffer, since it + // will be many gigabytes long. This is a practical, not an + // architectural limit to Pack200 archive sizes. + assert(FIRST_READ >= MAGIC_BYTES + AH_LENGTH_0_MAX + 2 * B_MAX); + + bool foreign_buf = (read_input_fn == nullptr); + byte initbuf[(int)FIRST_READ + (int)C_SLOP + 200]; // 200 is for JAR I/O + if (foreign_buf) + { + // inbytes is all there is + input.set(inbytes); + rp = input.base(); + rplimit = input.limit(); + } + else + { + // inbytes, if not empty, contains some read-ahead we must use first + // ensure_input will take care of copying it into initbuf, + // then querying read_input_fn for any additional data needed. + // However, the caller must assume that we use up all of inbytes. + // There is no way to tell the caller that we used only part of them. + // Therefore, the caller must use only a bare minimum of read-ahead. + if (inbytes.len > FIRST_READ) + { + unpack_abort("too much read-ahead"); + } + input.set(initbuf, sizeof(initbuf)); + input.b.clear(); + input.b.copyFrom(inbytes); + rplimit = rp = input.base(); + rplimit += inbytes.len; + bytes_read += inbytes.len; + } + // Read only 19 bytes, which is certain to contain #archive_options fields, + // but is certain not to overflow past the archive_header. + input.b.len = FIRST_READ; + if (!ensure_input(FIRST_READ)) + unpack_abort("EOF reading archive magic number"); + + if (rp[0] == 'P' && rp[1] == 'K') + { + // In the Unix-style program, we simply simulate a copy command. + // Copy until EOF; assume the JAR file is the last segment. + fprintf(stderr, "Copy-mode.\n"); + for (;;) + { + jarout->write_data(rp, (int)input_remaining()); + if (foreign_buf) + break; // one-time use of a passed in buffer + if (input.size() < CHUNK) + { + // Get some breathing room. + input.set(U_NEW(byte, (size_t)CHUNK + C_SLOP), (size_t)CHUNK); + } + rp = rplimit = input.base(); + if (!ensure_input(1)) + break; + } + jarout->closeJarFile(false); + return; + } + + // Read the magic number. + magic = 0; + for (int i1 = 0; i1 < (int)sizeof(magic); i1++) + { + magic <<= 8; + magic += (*rp++ & 0xFF); + } + + // Read the first 3 values from the header. + value_stream hdr; + int hdrVals = 0; + int hdrValsSkipped = 0; // debug only + hdr.init(rp, rplimit, UNSIGNED5_spec); + minver = hdr.getInt(); + majver = hdr.getInt(); + hdrVals += 2; + + if (magic != (int)JAVA_PACKAGE_MAGIC || + (majver != JAVA5_PACKAGE_MAJOR_VERSION && majver != JAVA6_PACKAGE_MAJOR_VERSION) || + (minver != JAVA5_PACKAGE_MINOR_VERSION && minver != JAVA6_PACKAGE_MINOR_VERSION)) + { + char message[200]; + sprintf(message, "@" ERROR_FORMAT ": magic/ver = " + "%08X/%d.%d should be %08X/%d.%d OR %08X/%d.%d\n", + magic, majver, minver, JAVA_PACKAGE_MAGIC, JAVA5_PACKAGE_MAJOR_VERSION, + JAVA5_PACKAGE_MINOR_VERSION, JAVA_PACKAGE_MAGIC, JAVA6_PACKAGE_MAJOR_VERSION, + JAVA6_PACKAGE_MINOR_VERSION); + unpack_abort(message); + } + + archive_options = hdr.getInt(); + hdrVals += 1; + assert(hdrVals == AH_LENGTH_0); // first three fields only #define ORBIT(bit) | (bit) - int OPTION_LIMIT = (0 ARCHIVE_BIT_DO(ORBIT)); + int OPTION_LIMIT = (0 ARCHIVE_BIT_DO(ORBIT)); #undef ORBIT - if ((archive_options & ~OPTION_LIMIT) != 0) - { - fprintf(stderr, "Warning: Illegal archive options 0x%x\n", archive_options); - unpack_abort("illegal archive options"); - return; - } - - if ((archive_options & AO_HAVE_FILE_HEADERS) != 0) - { - uint32_t hi = hdr.getInt(); - uint32_t lo = hdr.getInt(); - uint64_t x = band::makeLong(hi, lo); - archive_size = (size_t)x; - if (archive_size != x) - { - // Silly size specified; force overflow. - archive_size = PSIZE_MAX + 1; - } - hdrVals += 2; - } - else - { - hdrValsSkipped += 2; - } - - // Now we can size the whole archive. - // Read everything else into a mega-buffer. - rp = hdr.rp; - int header_size_0 = (int)(rp - input.base()); // used-up header (4byte + 3int) - int header_size_1 = (int)(rplimit - rp); // buffered unused initial fragment - int header_size = header_size_0 + header_size_1; - unsized_bytes_read = header_size_0; - if (foreign_buf) - { - if (archive_size > (size_t)header_size_1) - { - unpack_abort("EOF reading fixed input buffer"); - return; - } - } - else if (archive_size != 0) - { - if (archive_size < ARCHIVE_SIZE_MIN) - { - unpack_abort("impossible archive size"); // bad input data - return; - } - if (archive_size < (size_t)header_size_1) - { - unpack_abort("too much read-ahead"); // somehow we pre-fetched too much? - return; - } - input.set(U_NEW(byte, add_size(header_size_0, archive_size, C_SLOP)), - (size_t)header_size_0 + archive_size); - assert(input.limit()[0] == 0); - // Move all the bytes we read initially into the real buffer. - input.b.copyFrom(initbuf, header_size); - rp = input.b.ptr + header_size_0; - rplimit = input.b.ptr + header_size; - } - else - { - // It's more complicated and painful. - // A zero archive_size means that we must read until EOF. - input.init(CHUNK * 2); - input.b.len = input.allocated; - rp = rplimit = input.base(); - // Set up input buffer as if we already read the header: - input.b.copyFrom(initbuf, header_size); - rplimit += header_size; - while (ensure_input(input.limit() - rp)) - { - size_t dataSoFar = input_remaining(); - size_t nextSize = add_size(dataSoFar, CHUNK); - input.ensureSize(nextSize); - input.b.len = input.allocated; - rp = rplimit = input.base(); - rplimit += dataSoFar; - } - size_t dataSize = (rplimit - input.base()); - input.b.len = dataSize; - input.grow(C_SLOP); - free_input = true; // free it later - input.b.len = dataSize; - assert(input.limit()[0] == 0); - rp = rplimit = input.base(); - rplimit += dataSize; - rp += header_size_0; // already scanned these bytes... - } - live_input = true; // mark as "do not reuse" - - // read the rest of the header fields - ensure_input((AH_LENGTH - AH_LENGTH_0) * B_MAX); - hdr.rp = rp; - hdr.rplimit = rplimit; - - if ((archive_options & AO_HAVE_FILE_HEADERS) != 0) - { - archive_next_count = hdr.getInt(); - if (archive_next_count < 0) - unpack_abort("bad archive_next_count"); - archive_modtime = hdr.getInt(); - file_count = hdr.getInt(); - if (file_count < 0) - unpack_abort("bad file_count"); - hdrVals += 3; - } - else - { - hdrValsSkipped += 3; - } - - if ((archive_options & AO_HAVE_SPECIAL_FORMATS) != 0) - { - band_headers_size = hdr.getInt(); - if (band_headers_size < 0) - unpack_abort("bad band_headers_size"); - attr_definition_count = hdr.getInt(); - if (attr_definition_count < 0) - unpack_abort("bad attr_definition_count"); - hdrVals += 2; - } - else - { - hdrValsSkipped += 2; - } - - int cp_counts[N_TAGS_IN_ORDER]; - for (int k = 0; k < (int)N_TAGS_IN_ORDER; k++) - { - if (!(archive_options & AO_HAVE_CP_NUMBERS)) - { - switch (TAGS_IN_ORDER[k]) - { - case CONSTANT_Integer: - case CONSTANT_Float: - case CONSTANT_Long: - case CONSTANT_Double: - cp_counts[k] = 0; - hdrValsSkipped += 1; - continue; - } - } - cp_counts[k] = hdr.getInt(); - if (cp_counts[k] < 0) - unpack_abort("bad cp_counts"); - hdrVals += 1; - } - - ic_count = hdr.getInt(); - if (ic_count < 0) - unpack_abort("bad ic_count"); - - default_class_minver = hdr.getInt(); - default_class_majver = hdr.getInt(); - - class_count = hdr.getInt(); - if (class_count < 0) - unpack_abort("bad class_count"); - - hdrVals += 4; - - // done with archive_header - hdrVals += hdrValsSkipped; - assert(hdrVals == AH_LENGTH); - - rp = hdr.rp; - if (rp > rplimit) - unpack_abort("EOF reading archive header"); - - // Now size the CP. - cp.init(this, cp_counts); - - default_file_modtime = archive_modtime; - if (default_file_modtime == 0 && !(archive_options & AO_HAVE_FILE_MODTIME)) - default_file_modtime = DEFAULT_ARCHIVE_MODTIME; // taken from driver - if ((archive_options & AO_DEFLATE_HINT) != 0) - default_file_options |= FO_DEFLATE_HINT; - - // meta-bytes, if any, immediately follow archive header - // band_headers.readData(band_headers_size); - ensure_input(band_headers_size); - if (input_remaining() < (size_t)band_headers_size) - { - unpack_abort("EOF reading band headers"); - return; - } - bytes band_headers; - // The "1+" allows an initial byte to be pushed on the front. - band_headers.set(1 + U_NEW(byte, 1 + band_headers_size + C_SLOP), band_headers_size); - - // Start scanning band headers here: - band_headers.copyFrom(rp, band_headers.len); - rp += band_headers.len; - assert(rp <= rplimit); - meta_rp = band_headers.ptr; - // Put evil meta-codes at the end of the band headers, - // so we are sure to throw an error if we run off the end. - bytes::of(band_headers.limit(), C_SLOP).clear(_meta_error); + if ((archive_options & ~OPTION_LIMIT) != 0) + { + fprintf(stderr, "Warning: Illegal archive options 0x%x\n", archive_options); + unpack_abort("illegal archive options"); + return; + } + + if ((archive_options & AO_HAVE_FILE_HEADERS) != 0) + { + uint32_t hi = hdr.getInt(); + uint32_t lo = hdr.getInt(); + uint64_t x = band::makeLong(hi, lo); + archive_size = (size_t)x; + if (archive_size != x) + { + // Silly size specified; force overflow. + archive_size = PSIZE_MAX + 1; + } + hdrVals += 2; + } + else + { + hdrValsSkipped += 2; + } + + // Now we can size the whole archive. + // Read everything else into a mega-buffer. + rp = hdr.rp; + int header_size_0 = (int)(rp - input.base()); // used-up header (4byte + 3int) + int header_size_1 = (int)(rplimit - rp); // buffered unused initial fragment + int header_size = header_size_0 + header_size_1; + unsized_bytes_read = header_size_0; + if (foreign_buf) + { + if (archive_size > (size_t)header_size_1) + { + unpack_abort("EOF reading fixed input buffer"); + return; + } + } + else if (archive_size != 0) + { + if (archive_size < ARCHIVE_SIZE_MIN) + { + unpack_abort("impossible archive size"); // bad input data + return; + } + if (archive_size < (size_t)header_size_1) + { + unpack_abort("too much read-ahead"); // somehow we pre-fetched too much? + return; + } + input.set(U_NEW(byte, add_size(header_size_0, archive_size, C_SLOP)), + (size_t)header_size_0 + archive_size); + assert(input.limit()[0] == 0); + // Move all the bytes we read initially into the real buffer. + input.b.copyFrom(initbuf, header_size); + rp = input.b.ptr + header_size_0; + rplimit = input.b.ptr + header_size; + } + else + { + // It's more complicated and painful. + // A zero archive_size means that we must read until EOF. + input.init(CHUNK * 2); + input.b.len = input.allocated; + rp = rplimit = input.base(); + // Set up input buffer as if we already read the header: + input.b.copyFrom(initbuf, header_size); + rplimit += header_size; + while (ensure_input(input.limit() - rp)) + { + size_t dataSoFar = input_remaining(); + size_t nextSize = add_size(dataSoFar, CHUNK); + input.ensureSize(nextSize); + input.b.len = input.allocated; + rp = rplimit = input.base(); + rplimit += dataSoFar; + } + size_t dataSize = (rplimit - input.base()); + input.b.len = dataSize; + input.grow(C_SLOP); + free_input = true; // free it later + input.b.len = dataSize; + assert(input.limit()[0] == 0); + rp = rplimit = input.base(); + rplimit += dataSize; + rp += header_size_0; // already scanned these bytes... + } + live_input = true; // mark as "do not reuse" + + // read the rest of the header fields + ensure_input((AH_LENGTH - AH_LENGTH_0) * B_MAX); + hdr.rp = rp; + hdr.rplimit = rplimit; + + if ((archive_options & AO_HAVE_FILE_HEADERS) != 0) + { + archive_next_count = hdr.getInt(); + if (archive_next_count < 0) + unpack_abort("bad archive_next_count"); + archive_modtime = hdr.getInt(); + file_count = hdr.getInt(); + if (file_count < 0) + unpack_abort("bad file_count"); + hdrVals += 3; + } + else + { + hdrValsSkipped += 3; + } + + if ((archive_options & AO_HAVE_SPECIAL_FORMATS) != 0) + { + band_headers_size = hdr.getInt(); + if (band_headers_size < 0) + unpack_abort("bad band_headers_size"); + attr_definition_count = hdr.getInt(); + if (attr_definition_count < 0) + unpack_abort("bad attr_definition_count"); + hdrVals += 2; + } + else + { + hdrValsSkipped += 2; + } + + int cp_counts[N_TAGS_IN_ORDER]; + for (int k = 0; k < (int)N_TAGS_IN_ORDER; k++) + { + if (!(archive_options & AO_HAVE_CP_NUMBERS)) + { + switch (TAGS_IN_ORDER[k]) + { + case CONSTANT_Integer: + case CONSTANT_Float: + case CONSTANT_Long: + case CONSTANT_Double: + cp_counts[k] = 0; + hdrValsSkipped += 1; + continue; + } + } + cp_counts[k] = hdr.getInt(); + if (cp_counts[k] < 0) + unpack_abort("bad cp_counts"); + hdrVals += 1; + } + + ic_count = hdr.getInt(); + if (ic_count < 0) + unpack_abort("bad ic_count"); + + default_class_minver = hdr.getInt(); + default_class_majver = hdr.getInt(); + + class_count = hdr.getInt(); + if (class_count < 0) + unpack_abort("bad class_count"); + + hdrVals += 4; + + // done with archive_header + hdrVals += hdrValsSkipped; + assert(hdrVals == AH_LENGTH); + + rp = hdr.rp; + if (rp > rplimit) + unpack_abort("EOF reading archive header"); + + // Now size the CP. + cp.init(this, cp_counts); + + default_file_modtime = archive_modtime; + if (default_file_modtime == 0 && !(archive_options & AO_HAVE_FILE_MODTIME)) + default_file_modtime = DEFAULT_ARCHIVE_MODTIME; // taken from driver + if ((archive_options & AO_DEFLATE_HINT) != 0) + default_file_options |= FO_DEFLATE_HINT; + + // meta-bytes, if any, immediately follow archive header + // band_headers.readData(band_headers_size); + ensure_input(band_headers_size); + if (input_remaining() < (size_t)band_headers_size) + { + unpack_abort("EOF reading band headers"); + return; + } + bytes band_headers; + // The "1+" allows an initial byte to be pushed on the front. + band_headers.set(1 + U_NEW(byte, 1 + band_headers_size + C_SLOP), band_headers_size); + + // Start scanning band headers here: + band_headers.copyFrom(rp, band_headers.len); + rp += band_headers.len; + assert(rp <= rplimit); + meta_rp = band_headers.ptr; + // Put evil meta-codes at the end of the band headers, + // so we are sure to throw an error if we run off the end. + bytes::of(band_headers.limit(), C_SLOP).clear(_meta_error); } void unpacker::finish() { - if (verbose >= 1) - { - fprintf(stderr, "A total of " LONG_LONG_FORMAT " bytes were read in %d segment(s).\n", - (bytes_read_before_reset + bytes_read), segments_read_before_reset + 1); - fprintf(stderr, "A total of " LONG_LONG_FORMAT " file content bytes were written.\n", - (bytes_written_before_reset + bytes_written)); - fprintf(stderr, - "A total of %d files (of which %d are classes) were written to output.\n", - files_written_before_reset + files_written, - classes_written_before_reset + classes_written); - } - if (jarout != nullptr) - jarout->closeJarFile(true); + if (verbose >= 1) + { + fprintf(stderr, "A total of %" PRIu64 " bytes were read in %d segment(s).\n", + (bytes_read_before_reset + bytes_read), segments_read_before_reset + 1); + fprintf(stderr, "A total of %" PRIu64 " file content bytes were written.\n", + (bytes_written_before_reset + bytes_written)); + fprintf(stderr, + "A total of %d files (of which %d are classes) were written to output.\n", + files_written_before_reset + files_written, + classes_written_before_reset + classes_written); + } + if (jarout != nullptr) + jarout->closeJarFile(true); } // Cf. PackageReader.readConstantPoolCounts void constant_pool::init(unpacker *u_, int counts[NUM_COUNTS]) { - this->u = u_; - - // Fill-pointer for CP. - int next_entry = 0; - - // Size the constant pool: - for (int k = 0; k < (int)N_TAGS_IN_ORDER; k++) - { - byte tag = TAGS_IN_ORDER[k]; - int len = counts[k]; - tag_count[tag] = len; - tag_base[tag] = next_entry; - next_entry += len; - // Detect and defend against constant pool size overflow. - // (Pack200 forbids the sum of CP counts to exceed 2^29-1.) - enum - { - CP_SIZE_LIMIT = (1 << 29), - IMPLICIT_ENTRY_COUNT = 1 // empty Utf8 string - }; - if (len >= (1 << 29) || len < 0 || next_entry >= CP_SIZE_LIMIT + IMPLICIT_ENTRY_COUNT) - { - unpack_abort("archive too large: constant pool limit exceeded"); - } - } - - // Close off the end of the CP: - nentries = next_entry; - - // place a limit on future CP growth: - int generous = 0; - generous = add_size(generous, u->ic_count); // implicit name - generous = add_size(generous, u->ic_count); // outer - generous = add_size(generous, u->ic_count); // outer.utf8 - generous = add_size(generous, 40); // WKUs, misc - generous = add_size(generous, u->class_count); // implicit SourceFile strings - maxentries = add_size(nentries, generous); - - // Note that this CP does not include "empty" entries - // for longs and doubles. Those are introduced when - // the entries are renumbered for classfile output. - - entries = U_NEW(entry, maxentries); - - first_extra_entry = &entries[nentries]; - - // Initialize the standard indexes. - tag_count[CONSTANT_All] = nentries; - tag_base[CONSTANT_All] = 0; - for (int tag = 0; tag < CONSTANT_Limit; tag++) - { - entry *cpMap = &entries[tag_base[tag]]; - tag_index[tag].init(tag_count[tag], cpMap, tag); - } - - // Initialize hashTab to a generous power-of-two size. - uint32_t pow2 = 1; - uint32_t target = maxentries + maxentries / 2; // 60% full - while (pow2 < target) - pow2 <<= 1; - hashTab = U_NEW(entry *, hashTabLength = pow2); + this->u = u_; + + // Fill-pointer for CP. + int next_entry = 0; + + // Size the constant pool: + for (int k = 0; k < (int)N_TAGS_IN_ORDER; k++) + { + byte tag = TAGS_IN_ORDER[k]; + int len = counts[k]; + tag_count[tag] = len; + tag_base[tag] = next_entry; + next_entry += len; + // Detect and defend against constant pool size overflow. + // (Pack200 forbids the sum of CP counts to exceed 2^29-1.) + enum + { + CP_SIZE_LIMIT = (1 << 29), + IMPLICIT_ENTRY_COUNT = 1 // empty Utf8 string + }; + if (len >= (1 << 29) || len < 0 || next_entry >= CP_SIZE_LIMIT + IMPLICIT_ENTRY_COUNT) + { + unpack_abort("archive too large: constant pool limit exceeded"); + } + } + + // Close off the end of the CP: + nentries = next_entry; + + // place a limit on future CP growth: + int generous = 0; + generous = add_size(generous, u->ic_count); // implicit name + generous = add_size(generous, u->ic_count); // outer + generous = add_size(generous, u->ic_count); // outer.utf8 + generous = add_size(generous, 40); // WKUs, misc + generous = add_size(generous, u->class_count); // implicit SourceFile strings + maxentries = add_size(nentries, generous); + + // Note that this CP does not include "empty" entries + // for longs and doubles. Those are introduced when + // the entries are renumbered for classfile output. + + entries = U_NEW(entry, maxentries); + + first_extra_entry = &entries[nentries]; + + // Initialize the standard indexes. + tag_count[CONSTANT_All] = nentries; + tag_base[CONSTANT_All] = 0; + for (int tag = 0; tag < CONSTANT_Limit; tag++) + { + entry *cpMap = &entries[tag_base[tag]]; + tag_index[tag].init(tag_count[tag], cpMap, tag); + } + + // Initialize hashTab to a generous power-of-two size. + uint32_t pow2 = 1; + uint32_t target = maxentries + maxentries / 2; // 60% full + while (pow2 < target) + pow2 <<= 1; + hashTab = U_NEW(entry *, hashTabLength = pow2); } static byte *store_Utf8_char(byte *cp, unsigned short ch) { - if (ch >= 0x001 && ch <= 0x007F) - { - *cp++ = (byte)ch; - } - else if (ch <= 0x07FF) - { - *cp++ = (byte)(0xC0 | ((ch >> 6) & 0x1F)); - *cp++ = (byte)(0x80 | ((ch >> 0) & 0x3F)); - } - else - { - *cp++ = (byte)(0xE0 | ((ch >> 12) & 0x0F)); - *cp++ = (byte)(0x80 | ((ch >> 6) & 0x3F)); - *cp++ = (byte)(0x80 | ((ch >> 0) & 0x3F)); - } - return cp; + if (ch >= 0x001 && ch <= 0x007F) + { + *cp++ = (byte)ch; + } + else if (ch <= 0x07FF) + { + *cp++ = (byte)(0xC0 | ((ch >> 6) & 0x1F)); + *cp++ = (byte)(0x80 | ((ch >> 0) & 0x3F)); + } + else + { + *cp++ = (byte)(0xE0 | ((ch >> 12) & 0x0F)); + *cp++ = (byte)(0x80 | ((ch >> 6) & 0x3F)); + *cp++ = (byte)(0x80 | ((ch >> 0) & 0x3F)); + } + return cp; } static byte *skip_Utf8_chars(byte *cp, int len) { - for (;; cp++) - { - int ch = *cp & 0xFF; - if ((ch & 0xC0) != 0x80) - { - if (len-- == 0) - return cp; - if (ch < 0x80 && len == 0) - return cp + 1; - } - } + for (;; cp++) + { + int ch = *cp & 0xFF; + if ((ch & 0xC0) != 0x80) + { + if (len-- == 0) + return cp; + if (ch < 0x80 && len == 0) + return cp + 1; + } + } } static int compare_Utf8_chars(bytes &b1, bytes &b2) { - int l1 = (int)b1.len; - int l2 = (int)b2.len; - int l0 = (l1 < l2) ? l1 : l2; - byte *p1 = b1.ptr; - byte *p2 = b2.ptr; - int c0 = 0; - for (int i = 0; i < l0; i++) - { - int c1 = p1[i] & 0xFF; - int c2 = p2[i] & 0xFF; - if (c1 != c2) - { - // Before returning the obvious answer, - // check to see if c1 or c2 is part of a 0x0000, - // which encodes as {0xC0,0x80}. The 0x0000 is the - // lowest-sorting Java char value, and yet it encodes - // as if it were the first char after 0x7F, which causes - // strings containing nulls to sort too high. All other - // comparisons are consistent between Utf8 and Java chars. - if (c1 == 0xC0 && (p1[i + 1] & 0xFF) == 0x80) - c1 = 0; - if (c2 == 0xC0 && (p2[i + 1] & 0xFF) == 0x80) - c2 = 0; - if (c0 == 0xC0) - { - assert(((c1 | c2) & 0xC0) == 0x80); // c1 & c2 are extension chars - if (c1 == 0x80) - c1 = 0; // will sort below c2 - if (c2 == 0x80) - c2 = 0; // will sort below c1 - } - return c1 - c2; - } - c0 = c1; // save away previous char - } - // common prefix is identical; return length difference if any - return l1 - l2; + int l1 = (int)b1.len; + int l2 = (int)b2.len; + int l0 = (l1 < l2) ? l1 : l2; + byte *p1 = b1.ptr; + byte *p2 = b2.ptr; + int c0 = 0; + for (int i = 0; i < l0; i++) + { + int c1 = p1[i] & 0xFF; + int c2 = p2[i] & 0xFF; + if (c1 != c2) + { + // Before returning the obvious answer, + // check to see if c1 or c2 is part of a 0x0000, + // which encodes as {0xC0,0x80}. The 0x0000 is the + // lowest-sorting Java char value, and yet it encodes + // as if it were the first char after 0x7F, which causes + // strings containing nulls to sort too high. All other + // comparisons are consistent between Utf8 and Java chars. + if (c1 == 0xC0 && (p1[i + 1] & 0xFF) == 0x80) + c1 = 0; + if (c2 == 0xC0 && (p2[i + 1] & 0xFF) == 0x80) + c2 = 0; + if (c0 == 0xC0) + { + assert(((c1 | c2) & 0xC0) == 0x80); // c1 & c2 are extension chars + if (c1 == 0x80) + c1 = 0; // will sort below c2 + if (c2 == 0x80) + c2 = 0; // will sort below c1 + } + return c1 - c2; + } + c0 = c1; // save away previous char + } + // common prefix is identical; return length difference if any + return l1 - l2; } // Cf. PackageReader.readUtf8Bands void unpacker::read_Utf8_values(entry *cpMap, int len) { - // Implicit first Utf8 string is the empty string. - enum - { - // certain bands begin with implicit zeroes - PREFIX_SKIP_2 = 2, - SUFFIX_SKIP_1 = 1 - }; - - int i; - - // First band: Read lengths of shared prefixes. - if (len > PREFIX_SKIP_2) - cp_Utf8_prefix.readData(len - PREFIX_SKIP_2); - - // Second band: Read lengths of unshared suffixes: - if (len > SUFFIX_SKIP_1) - cp_Utf8_suffix.readData(len - SUFFIX_SKIP_1); - - bytes *allsuffixes = T_NEW(bytes, len); - - int nbigsuf = 0; - fillbytes charbuf; // buffer to allocate small strings - charbuf.init(); - - // Third band: Read the char values in the unshared suffixes: - cp_Utf8_chars.readData(cp_Utf8_suffix.getIntTotal()); - for (i = 0; i < len; i++) - { - int suffix = (i < SUFFIX_SKIP_1) ? 0 : cp_Utf8_suffix.getInt(); - if (suffix < 0) - { - unpack_abort("bad utf8 suffix"); - } - if (suffix == 0 && i >= SUFFIX_SKIP_1) - { - // chars are packed in cp_Utf8_big_chars - nbigsuf += 1; - continue; - } - bytes &chars = allsuffixes[i]; - uint32_t size3 = suffix * 3; // max Utf8 length - bool isMalloc = (suffix > SMALL); - if (isMalloc) - { - chars.malloc(size3); - } - else - { - if (!charbuf.canAppend(size3 + 1)) - { - assert(charbuf.allocated == 0 || tmallocs.contains(charbuf.base())); - charbuf.init(CHUNK); // Reset to new buffer. - tmallocs.add(charbuf.base()); - } - chars.set(charbuf.grow(size3 + 1), size3); - } - - byte *chp = chars.ptr; - for (int j = 0; j < suffix; j++) - { - unsigned short ch = cp_Utf8_chars.getInt(); - chp = store_Utf8_char(chp, ch); - } - // shrink to fit: - if (isMalloc) - { - chars.realloc(chp - chars.ptr); - tmallocs.add(chars.ptr); // free it later - } - else - { - int shrink = (int)(chars.limit() - chp); - chars.len -= shrink; - charbuf.b.len -= shrink; // ungrow to reclaim buffer space - // Note that we did not reclaim the final '\0'. - assert(chars.limit() == charbuf.limit() - 1); - assert(strlen((char *)chars.ptr) == chars.len); - } - } - // cp_Utf8_chars.done(); - - // Fourth band: Go back and size the specially packed strings. - int maxlen = 0; - cp_Utf8_big_suffix.readData(nbigsuf); - cp_Utf8_suffix.rewind(); - for (i = 0; i < len; i++) - { - int suffix = (i < SUFFIX_SKIP_1) ? 0 : cp_Utf8_suffix.getInt(); - int prefix = (i < PREFIX_SKIP_2) ? 0 : cp_Utf8_prefix.getInt(); - if (prefix < 0 || prefix + suffix < 0) - { - unpack_abort("bad utf8 prefix"); - } - bytes &chars = allsuffixes[i]; - if (suffix == 0 && i >= SUFFIX_SKIP_1) - { - suffix = cp_Utf8_big_suffix.getInt(); - assert(chars.ptr == nullptr); - chars.len = suffix; // just a momentary hack - } - else - { - assert(chars.ptr != nullptr); - } - if (maxlen < prefix + suffix) - { - maxlen = prefix + suffix; - } - } - // cp_Utf8_suffix.done(); // will use allsuffixes[i].len (ptr!=nullptr) - // cp_Utf8_big_suffix.done(); // will use allsuffixes[i].len - - // Fifth band(s): Get the specially packed characters. - cp_Utf8_big_suffix.rewind(); - for (i = 0; i < len; i++) - { - bytes &chars = allsuffixes[i]; - if (chars.ptr != nullptr) - continue; // already input - int suffix = (int)chars.len; // pick up the hack - uint32_t size3 = suffix * 3; - if (suffix == 0) - continue; // done with empty string - chars.malloc(size3); - byte *chp = chars.ptr; - band saved_band = cp_Utf8_big_chars; - cp_Utf8_big_chars.readData(suffix); - for (int j = 0; j < suffix; j++) - { - unsigned short ch = cp_Utf8_big_chars.getInt(); - chp = store_Utf8_char(chp, ch); - } - chars.realloc(chp - chars.ptr); - tmallocs.add(chars.ptr); // free it later - // cp_Utf8_big_chars.done(); - cp_Utf8_big_chars = saved_band; // reset the band for the next string - } - cp_Utf8_big_chars.readData(0); // zero chars - // cp_Utf8_big_chars.done(); - - // Finally, sew together all the prefixes and suffixes. - bytes bigbuf; - bigbuf.malloc(maxlen * 3 + 1); // max Utf8 length, plus slop for nullptr - int prevlen = 0; // previous string length (in chars) - tmallocs.add(bigbuf.ptr); // free after this block - cp_Utf8_prefix.rewind(); - for (i = 0; i < len; i++) - { - bytes &chars = allsuffixes[i]; - int prefix = (i < PREFIX_SKIP_2) ? 0 : cp_Utf8_prefix.getInt(); - int suffix = (int)chars.len; - byte *fillp; - // by induction, the buffer is already filled with the prefix - // make sure the prefix value is not corrupted, though: - if (prefix > prevlen) - { - unpack_abort("utf8 prefix overflow"); - return; - } - fillp = skip_Utf8_chars(bigbuf.ptr, prefix); - // copy the suffix into the same buffer: - fillp = chars.writeTo(fillp); - assert(bigbuf.inBounds(fillp)); - *fillp = 0; // bigbuf must contain a well-formed Utf8 string - int length = (int)(fillp - bigbuf.ptr); - bytes &value = cpMap[i].value.b; - value.set(U_NEW(byte, add_size(length, 1)), length); - value.copyFrom(bigbuf.ptr, length); - // Index all Utf8 strings - entry *&htref = cp.hashTabRef(CONSTANT_Utf8, value); - if (htref == nullptr) - { - // Note that if two identical strings are transmitted, - // the first is taken to be the canonical one. - htref = &cpMap[i]; - } - prevlen = prefix + suffix; - } - // cp_Utf8_prefix.done(); - - // Free intermediate buffers. - free_temps(); + // Implicit first Utf8 string is the empty string. + enum + { + // certain bands begin with implicit zeroes + PREFIX_SKIP_2 = 2, + SUFFIX_SKIP_1 = 1 + }; + + int i; + + // First band: Read lengths of shared prefixes. + if (len > PREFIX_SKIP_2) + cp_Utf8_prefix.readData(len - PREFIX_SKIP_2); + + // Second band: Read lengths of unshared suffixes: + if (len > SUFFIX_SKIP_1) + cp_Utf8_suffix.readData(len - SUFFIX_SKIP_1); + + bytes *allsuffixes = T_NEW(bytes, len); + + int nbigsuf = 0; + fillbytes charbuf; // buffer to allocate small strings + charbuf.init(); + + // Third band: Read the char values in the unshared suffixes: + cp_Utf8_chars.readData(cp_Utf8_suffix.getIntTotal()); + for (i = 0; i < len; i++) + { + int suffix = (i < SUFFIX_SKIP_1) ? 0 : cp_Utf8_suffix.getInt(); + if (suffix < 0) + { + unpack_abort("bad utf8 suffix"); + } + if (suffix == 0 && i >= SUFFIX_SKIP_1) + { + // chars are packed in cp_Utf8_big_chars + nbigsuf += 1; + continue; + } + bytes &chars = allsuffixes[i]; + uint32_t size3 = suffix * 3; // max Utf8 length + bool isMalloc = (suffix > SMALL); + if (isMalloc) + { + chars.malloc(size3); + } + else + { + if (!charbuf.canAppend(size3 + 1)) + { + assert(charbuf.allocated == 0 || tmallocs.contains(charbuf.base())); + charbuf.init(CHUNK); // Reset to new buffer. + tmallocs.add(charbuf.base()); + } + chars.set(charbuf.grow(size3 + 1), size3); + } + + byte *chp = chars.ptr; + for (int j = 0; j < suffix; j++) + { + unsigned short ch = cp_Utf8_chars.getInt(); + chp = store_Utf8_char(chp, ch); + } + // shrink to fit: + if (isMalloc) + { + chars.realloc(chp - chars.ptr); + tmallocs.add(chars.ptr); // free it later + } + else + { + int shrink = (int)(chars.limit() - chp); + chars.len -= shrink; + charbuf.b.len -= shrink; // ungrow to reclaim buffer space + // Note that we did not reclaim the final '\0'. + assert(chars.limit() == charbuf.limit() - 1); + assert(strlen((char *)chars.ptr) == chars.len); + } + } + // cp_Utf8_chars.done(); + + // Fourth band: Go back and size the specially packed strings. + int maxlen = 0; + cp_Utf8_big_suffix.readData(nbigsuf); + cp_Utf8_suffix.rewind(); + for (i = 0; i < len; i++) + { + int suffix = (i < SUFFIX_SKIP_1) ? 0 : cp_Utf8_suffix.getInt(); + int prefix = (i < PREFIX_SKIP_2) ? 0 : cp_Utf8_prefix.getInt(); + if (prefix < 0 || prefix + suffix < 0) + { + unpack_abort("bad utf8 prefix"); + } + bytes &chars = allsuffixes[i]; + if (suffix == 0 && i >= SUFFIX_SKIP_1) + { + suffix = cp_Utf8_big_suffix.getInt(); + assert(chars.ptr == nullptr); + chars.len = suffix; // just a momentary hack + } + else + { + assert(chars.ptr != nullptr); + } + if (maxlen < prefix + suffix) + { + maxlen = prefix + suffix; + } + } + // cp_Utf8_suffix.done(); // will use allsuffixes[i].len (ptr!=nullptr) + // cp_Utf8_big_suffix.done(); // will use allsuffixes[i].len + + // Fifth band(s): Get the specially packed characters. + cp_Utf8_big_suffix.rewind(); + for (i = 0; i < len; i++) + { + bytes &chars = allsuffixes[i]; + if (chars.ptr != nullptr) + continue; // already input + int suffix = (int)chars.len; // pick up the hack + uint32_t size3 = suffix * 3; + if (suffix == 0) + continue; // done with empty string + chars.malloc(size3); + byte *chp = chars.ptr; + band saved_band = cp_Utf8_big_chars; + cp_Utf8_big_chars.readData(suffix); + for (int j = 0; j < suffix; j++) + { + unsigned short ch = cp_Utf8_big_chars.getInt(); + chp = store_Utf8_char(chp, ch); + } + chars.realloc(chp - chars.ptr); + tmallocs.add(chars.ptr); // free it later + // cp_Utf8_big_chars.done(); + cp_Utf8_big_chars = saved_band; // reset the band for the next string + } + cp_Utf8_big_chars.readData(0); // zero chars + // cp_Utf8_big_chars.done(); + + // Finally, sew together all the prefixes and suffixes. + bytes bigbuf; + bigbuf.malloc(maxlen * 3 + 1); // max Utf8 length, plus slop for nullptr + int prevlen = 0; // previous string length (in chars) + tmallocs.add(bigbuf.ptr); // free after this block + cp_Utf8_prefix.rewind(); + for (i = 0; i < len; i++) + { + bytes &chars = allsuffixes[i]; + int prefix = (i < PREFIX_SKIP_2) ? 0 : cp_Utf8_prefix.getInt(); + int suffix = (int)chars.len; + byte *fillp; + // by induction, the buffer is already filled with the prefix + // make sure the prefix value is not corrupted, though: + if (prefix > prevlen) + { + unpack_abort("utf8 prefix overflow"); + return; + } + fillp = skip_Utf8_chars(bigbuf.ptr, prefix); + // copy the suffix into the same buffer: + fillp = chars.writeTo(fillp); + assert(bigbuf.inBounds(fillp)); + *fillp = 0; // bigbuf must contain a well-formed Utf8 string + int length = (int)(fillp - bigbuf.ptr); + bytes &value = cpMap[i].value.b; + value.set(U_NEW(byte, add_size(length, 1)), length); + value.copyFrom(bigbuf.ptr, length); + // Index all Utf8 strings + entry *&htref = cp.hashTabRef(CONSTANT_Utf8, value); + if (htref == nullptr) + { + // Note that if two identical strings are transmitted, + // the first is taken to be the canonical one. + htref = &cpMap[i]; + } + prevlen = prefix + suffix; + } + // cp_Utf8_prefix.done(); + + // Free intermediate buffers. + free_temps(); } void unpacker::read_single_words(band &cp_band, entry *cpMap, int len) { - cp_band.readData(len); - for (int i = 0; i < len; i++) - { - cpMap[i].value.i = cp_band.getInt(); // coding handles signs OK - } + cp_band.readData(len); + for (int i = 0; i < len; i++) + { + cpMap[i].value.i = cp_band.getInt(); // coding handles signs OK + } } void unpacker::read_double_words(band &cp_bands, entry *cpMap, int len) { - band &cp_band_hi = cp_bands; - band &cp_band_lo = cp_bands.nextBand(); - cp_band_hi.readData(len); - cp_band_lo.readData(len); - for (int i = 0; i < len; i++) - { - cpMap[i].value.l = cp_band_hi.getLong(cp_band_lo, true); - } - // cp_band_hi.done(); - // cp_band_lo.done(); + band &cp_band_hi = cp_bands; + band &cp_band_lo = cp_bands.nextBand(); + cp_band_hi.readData(len); + cp_band_lo.readData(len); + for (int i = 0; i < len; i++) + { + cpMap[i].value.l = cp_band_hi.getLong(cp_band_lo, true); + } + // cp_band_hi.done(); + // cp_band_lo.done(); } void unpacker::read_single_refs(band &cp_band, byte refTag, entry *cpMap, int len) { - assert(refTag == CONSTANT_Utf8); - cp_band.setIndexByTag(refTag); - cp_band.readData(len); - int indexTag = (cp_band.bn == e_cp_Class) ? CONSTANT_Class : 0; - for (int i = 0; i < len; i++) - { - entry &e = cpMap[i]; - e.refs = U_NEW(entry *, e.nrefs = 1); - entry *utf = cp_band.getRef(); - e.refs[0] = utf; - e.value.b = utf->value.b; // copy value of Utf8 string to self - if (indexTag != 0) - { - // Maintain cross-reference: - entry *&htref = cp.hashTabRef(indexTag, e.value.b); - if (htref == nullptr) - { - // Note that if two identical classes are transmitted, - // the first is taken to be the canonical one. - htref = &e; - } - } - } - // cp_band.done(); + assert(refTag == CONSTANT_Utf8); + cp_band.setIndexByTag(refTag); + cp_band.readData(len); + int indexTag = (cp_band.bn == e_cp_Class) ? CONSTANT_Class : 0; + for (int i = 0; i < len; i++) + { + entry &e = cpMap[i]; + e.refs = U_NEW(entry *, e.nrefs = 1); + entry *utf = cp_band.getRef(); + e.refs[0] = utf; + e.value.b = utf->value.b; // copy value of Utf8 string to self + if (indexTag != 0) + { + // Maintain cross-reference: + entry *&htref = cp.hashTabRef(indexTag, e.value.b); + if (htref == nullptr) + { + // Note that if two identical classes are transmitted, + // the first is taken to be the canonical one. + htref = &e; + } + } + } + // cp_band.done(); } void unpacker::read_double_refs(band &cp_band, byte ref1Tag, byte ref2Tag, entry *cpMap, - int len) + int len) { - band &cp_band1 = cp_band; - band &cp_band2 = cp_band.nextBand(); - cp_band1.setIndexByTag(ref1Tag); - cp_band2.setIndexByTag(ref2Tag); - cp_band1.readData(len); - cp_band2.readData(len); - for (int i = 0; i < len; i++) - { - entry &e = cpMap[i]; - e.refs = U_NEW(entry *, e.nrefs = 2); - e.refs[0] = cp_band1.getRef(); - e.refs[1] = cp_band2.getRef(); - } - // cp_band1.done(); - // cp_band2.done(); + band &cp_band1 = cp_band; + band &cp_band2 = cp_band.nextBand(); + cp_band1.setIndexByTag(ref1Tag); + cp_band2.setIndexByTag(ref2Tag); + cp_band1.readData(len); + cp_band2.readData(len); + for (int i = 0; i < len; i++) + { + entry &e = cpMap[i]; + e.refs = U_NEW(entry *, e.nrefs = 2); + e.refs[0] = cp_band1.getRef(); + e.refs[1] = cp_band2.getRef(); + } + // cp_band1.done(); + // cp_band2.done(); } // Cf. PackageReader.readSignatureBands void unpacker::read_signature_values(entry *cpMap, int len) { - cp_Signature_form.setIndexByTag(CONSTANT_Utf8); - cp_Signature_form.readData(len); - int ncTotal = 0; - int i; - for (i = 0; i < len; i++) - { - entry &e = cpMap[i]; - entry &form = *cp_Signature_form.getRef(); - int nc = 0; - - for (const char *ncp = form.utf8String(); *ncp; ncp++) - { - if (*ncp == 'L') - nc++; - } - - ncTotal += nc; - e.refs = U_NEW(entry *, cpMap[i].nrefs = 1 + nc); - e.refs[0] = &form; - } - // cp_Signature_form.done(); - cp_Signature_classes.setIndexByTag(CONSTANT_Class); - cp_Signature_classes.readData(ncTotal); - for (i = 0; i < len; i++) - { - entry &e = cpMap[i]; - for (int j = 1; j < e.nrefs; j++) - { - e.refs[j] = cp_Signature_classes.getRef(); - } - } - // cp_Signature_classes.done(); + cp_Signature_form.setIndexByTag(CONSTANT_Utf8); + cp_Signature_form.readData(len); + int ncTotal = 0; + int i; + for (i = 0; i < len; i++) + { + entry &e = cpMap[i]; + entry &form = *cp_Signature_form.getRef(); + int nc = 0; + + for (const char *ncp = form.utf8String(); *ncp; ncp++) + { + if (*ncp == 'L') + nc++; + } + + ncTotal += nc; + e.refs = U_NEW(entry *, cpMap[i].nrefs = 1 + nc); + e.refs[0] = &form; + } + // cp_Signature_form.done(); + cp_Signature_classes.setIndexByTag(CONSTANT_Class); + cp_Signature_classes.readData(ncTotal); + for (i = 0; i < len; i++) + { + entry &e = cpMap[i]; + for (int j = 1; j < e.nrefs; j++) + { + e.refs[j] = cp_Signature_classes.getRef(); + } + } + // cp_Signature_classes.done(); } // Cf. PackageReader.readConstantPool void unpacker::read_cp() { - int i; - - for (int k = 0; k < (int)N_TAGS_IN_ORDER; k++) - { - byte tag = TAGS_IN_ORDER[k]; - int len = cp.tag_count[tag]; - int base = cp.tag_base[tag]; - - entry *cpMap = &cp.entries[base]; - for (i = 0; i < len; i++) - { - cpMap[i].tag = tag; - cpMap[i].inord = i; - } - - switch (tag) - { - case CONSTANT_Utf8: - read_Utf8_values(cpMap, len); - break; - case CONSTANT_Integer: - read_single_words(cp_Int, cpMap, len); - break; - case CONSTANT_Float: - read_single_words(cp_Float, cpMap, len); - break; - case CONSTANT_Long: - read_double_words(cp_Long_hi /*& cp_Long_lo*/, cpMap, len); - break; - case CONSTANT_Double: - read_double_words(cp_Double_hi /*& cp_Double_lo*/, cpMap, len); - break; - case CONSTANT_String: - read_single_refs(cp_String, CONSTANT_Utf8, cpMap, len); - break; - case CONSTANT_Class: - read_single_refs(cp_Class, CONSTANT_Utf8, cpMap, len); - break; - case CONSTANT_Signature: - read_signature_values(cpMap, len); - break; - case CONSTANT_NameandType: - read_double_refs(cp_Descr_name /*& cp_Descr_type*/, CONSTANT_Utf8, - CONSTANT_Signature, cpMap, len); - break; - case CONSTANT_Fieldref: - read_double_refs(cp_Field_class /*& cp_Field_desc*/, CONSTANT_Class, - CONSTANT_NameandType, cpMap, len); - break; - case CONSTANT_Methodref: - read_double_refs(cp_Method_class /*& cp_Method_desc*/, CONSTANT_Class, - CONSTANT_NameandType, cpMap, len); - break; - case CONSTANT_InterfaceMethodref: - read_double_refs(cp_Imethod_class /*& cp_Imethod_desc*/, CONSTANT_Class, - CONSTANT_NameandType, cpMap, len); - break; - default: - assert(false); - break; - } - } - - cp.expandSignatures(); - cp.initMemberIndexes(); + int i; + + for (int k = 0; k < (int)N_TAGS_IN_ORDER; k++) + { + byte tag = TAGS_IN_ORDER[k]; + int len = cp.tag_count[tag]; + int base = cp.tag_base[tag]; + + entry *cpMap = &cp.entries[base]; + for (i = 0; i < len; i++) + { + cpMap[i].tag = tag; + cpMap[i].inord = i; + } + + switch (tag) + { + case CONSTANT_Utf8: + read_Utf8_values(cpMap, len); + break; + case CONSTANT_Integer: + read_single_words(cp_Int, cpMap, len); + break; + case CONSTANT_Float: + read_single_words(cp_Float, cpMap, len); + break; + case CONSTANT_Long: + read_double_words(cp_Long_hi /*& cp_Long_lo*/, cpMap, len); + break; + case CONSTANT_Double: + read_double_words(cp_Double_hi /*& cp_Double_lo*/, cpMap, len); + break; + case CONSTANT_String: + read_single_refs(cp_String, CONSTANT_Utf8, cpMap, len); + break; + case CONSTANT_Class: + read_single_refs(cp_Class, CONSTANT_Utf8, cpMap, len); + break; + case CONSTANT_Signature: + read_signature_values(cpMap, len); + break; + case CONSTANT_NameandType: + read_double_refs(cp_Descr_name /*& cp_Descr_type*/, CONSTANT_Utf8, + CONSTANT_Signature, cpMap, len); + break; + case CONSTANT_Fieldref: + read_double_refs(cp_Field_class /*& cp_Field_desc*/, CONSTANT_Class, + CONSTANT_NameandType, cpMap, len); + break; + case CONSTANT_Methodref: + read_double_refs(cp_Method_class /*& cp_Method_desc*/, CONSTANT_Class, + CONSTANT_NameandType, cpMap, len); + break; + case CONSTANT_InterfaceMethodref: + read_double_refs(cp_Imethod_class /*& cp_Imethod_desc*/, CONSTANT_Class, + CONSTANT_NameandType, cpMap, len); + break; + default: + assert(false); + break; + } + } + + cp.expandSignatures(); + cp.initMemberIndexes(); #define SNAME(n, s) #s "\0" - const char *symNames = (ALL_ATTR_DO(SNAME) ""); + const char *symNames = (ALL_ATTR_DO(SNAME) ""); #undef SNAME - for (int sn = 0; sn < constant_pool::s_LIMIT; sn++) - { - assert(symNames[0] >= '0' && symNames[0] <= 'Z'); // sanity - bytes name; - name.set(symNames); - if (name.len > 0 && name.ptr[0] != '0') - { - cp.sym[sn] = cp.ensureUtf8(name); - } - symNames += name.len + 1; // skip trailing nullptr to next name - } - - band::initIndexes(this); + for (int sn = 0; sn < constant_pool::s_LIMIT; sn++) + { + assert(symNames[0] >= '0' && symNames[0] <= 'Z'); // sanity + bytes name; + name.set(symNames); + if (name.len > 0 && name.ptr[0] != '0') + { + cp.sym[sn] = cp.ensureUtf8(name); + } + symNames += name.len + 1; // skip trailing nullptr to next name + } + + band::initIndexes(this); } static band *no_bands[] = {nullptr}; // shared empty body inline band &unpacker::attr_definitions::fixed_band(int e_class_xxx) { - return u->all_bands[xxx_flags_hi_bn + (e_class_xxx - e_class_flags_hi)]; + return u->all_bands[xxx_flags_hi_bn + (e_class_xxx - e_class_flags_hi)]; } inline band &unpacker::attr_definitions::xxx_flags_hi() { - return fixed_band(e_class_flags_hi); + return fixed_band(e_class_flags_hi); } inline band &unpacker::attr_definitions::xxx_flags_lo() { - return fixed_band(e_class_flags_lo); + return fixed_band(e_class_flags_lo); } inline band &unpacker::attr_definitions::xxx_attr_count() { - return fixed_band(e_class_attr_count); + return fixed_band(e_class_attr_count); } inline band &unpacker::attr_definitions::xxx_attr_indexes() { - return fixed_band(e_class_attr_indexes); + return fixed_band(e_class_attr_indexes); } inline band &unpacker::attr_definitions::xxx_attr_calls() { - return fixed_band(e_class_attr_calls); + return fixed_band(e_class_attr_calls); } inline unpacker::layout_definition * unpacker::attr_definitions::defineLayout(int idx, entry *nameEntry, const char *layout) { - const char *name = nameEntry->value.b.strval(); - layout_definition *lo = defineLayout(idx, name, layout); - lo->nameEntry = nameEntry; - return lo; + const char *name = nameEntry->value.b.strval(); + layout_definition *lo = defineLayout(idx, name, layout); + lo->nameEntry = nameEntry; + return lo; } unpacker::layout_definition *unpacker::attr_definitions::defineLayout(int idx, const char *name, - const char *layout) + const char *layout) { - assert(flag_limit != 0); // must be set up already - if (idx >= 0) - { - // Fixed attr. - if (idx >= (int)flag_limit) - unpack_abort("attribute index too large"); - if (isRedefined(idx)) - unpack_abort("redefined attribute index"); - redef |= ((uint64_t)1 << idx); - } - else - { - idx = flag_limit + overflow_count.length(); - overflow_count.add(0); // make a new counter - } - layout_definition *lo = U_NEW(layout_definition, 1); - lo->idx = idx; - lo->name = name; - lo->layout = layout; - for (int adds = (idx + 1) - layouts.length(); adds > 0; adds--) - { - layouts.add(nullptr); - } - layouts.get(idx) = lo; - return lo; + assert(flag_limit != 0); // must be set up already + if (idx >= 0) + { + // Fixed attr. + if (idx >= (int)flag_limit) + unpack_abort("attribute index too large"); + if (isRedefined(idx)) + unpack_abort("redefined attribute index"); + redef |= ((uint64_t)1 << idx); + } + else + { + idx = flag_limit + overflow_count.length(); + overflow_count.add(0); // make a new counter + } + layout_definition *lo = U_NEW(layout_definition, 1); + lo->idx = idx; + lo->name = name; + lo->layout = layout; + for (int adds = (idx + 1) - layouts.length(); adds > 0; adds--) + { + layouts.add(nullptr); + } + layouts.get(idx) = lo; + return lo; } band **unpacker::attr_definitions::buildBands(unpacker::layout_definition *lo) { - int i; - if (lo->elems != nullptr) - return lo->bands(); - if (lo->layout[0] == '\0') - { - lo->elems = no_bands; - } - else - { - // Create bands for this attribute by parsing the layout. - bool hasCallables = lo->hasCallables(); - bands_made = 0x10000; // base number for bands made - const char *lp = lo->layout; - lp = parseLayout(lp, lo->elems, -1); - if (lp[0] != '\0' || band_stack.length() > 0) - { - unpack_abort("garbage at end of layout"); - } - band_stack.popTo(0); - - // Fix up callables to point at their callees. - band **bands = lo->elems; - assert(bands == lo->bands()); - int num_callables = 0; - if (hasCallables) - { - while (bands[num_callables] != nullptr) - { - if (bands[num_callables]->le_kind != EK_CBLE) - { - unpack_abort("garbage mixed with callables"); - break; - } - num_callables += 1; - } - } - for (i = 0; i < calls_to_link.length(); i++) - { - band &call = *(band *)calls_to_link.get(i); - assert(call.le_kind == EK_CALL); - // Determine the callee. - int call_num = call.le_len; - if (call_num < 0 || call_num >= num_callables) - { - unpack_abort("bad call in layout"); - break; - } - band &cble = *bands[call_num]; - // Link the call to it. - call.le_body[0] = &cble; - // Distinguish backward calls and callables: - assert(cble.le_kind == EK_CBLE); - // FIXME: hit this one - // assert(cble.le_len == call_num); - cble.le_back |= call.le_back; - } - calls_to_link.popTo(0); - } - return lo->elems; + int i; + if (lo->elems != nullptr) + return lo->bands(); + if (lo->layout[0] == '\0') + { + lo->elems = no_bands; + } + else + { + // Create bands for this attribute by parsing the layout. + bool hasCallables = lo->hasCallables(); + bands_made = 0x10000; // base number for bands made + const char *lp = lo->layout; + lp = parseLayout(lp, lo->elems, -1); + if (lp[0] != '\0' || band_stack.length() > 0) + { + unpack_abort("garbage at end of layout"); + } + band_stack.popTo(0); + + // Fix up callables to point at their callees. + band **bands = lo->elems; + assert(bands == lo->bands()); + int num_callables = 0; + if (hasCallables) + { + while (bands[num_callables] != nullptr) + { + if (bands[num_callables]->le_kind != EK_CBLE) + { + unpack_abort("garbage mixed with callables"); + break; + } + num_callables += 1; + } + } + for (i = 0; i < calls_to_link.length(); i++) + { + band &call = *(band *)calls_to_link.get(i); + assert(call.le_kind == EK_CALL); + // Determine the callee. + int call_num = call.le_len; + if (call_num < 0 || call_num >= num_callables) + { + unpack_abort("bad call in layout"); + break; + } + band &cble = *bands[call_num]; + // Link the call to it. + call.le_body[0] = &cble; + // Distinguish backward calls and callables: + assert(cble.le_kind == EK_CBLE); + // FIXME: hit this one + // assert(cble.le_len == call_num); + cble.le_back |= call.le_back; + } + calls_to_link.popTo(0); + } + return lo->elems; } /* attribute layout language parser attribute_layout: - ( layout_element )* | ( callable )+ + ( layout_element )* | ( callable )+ layout_element: - ( integral | replication | union | call | reference ) + ( integral | replication | union | call | reference ) callable: - '[' body ']' + '[' body ']' body: - ( layout_element )+ + ( layout_element )+ integral: - ( unsigned_int | signed_int | bc_index | bc_offset | flag ) + ( unsigned_int | signed_int | bc_index | bc_offset | flag ) unsigned_int: - uint_type + uint_type signed_int: - 'S' uint_type + 'S' uint_type any_int: - ( unsigned_int | signed_int ) + ( unsigned_int | signed_int ) bc_index: - ( 'P' uint_type | 'PO' uint_type ) + ( 'P' uint_type | 'PO' uint_type ) bc_offset: - 'O' any_int + 'O' any_int flag: - 'F' uint_type + 'F' uint_type uint_type: - ( 'B' | 'H' | 'I' | 'V' ) + ( 'B' | 'H' | 'I' | 'V' ) replication: - 'N' uint_type '[' body ']' + 'N' uint_type '[' body ']' union: - 'T' any_int (union_case)* '(' ')' '[' (body)? ']' + 'T' any_int (union_case)* '(' ')' '[' (body)? ']' union_case: - '(' union_case_tag (',' union_case_tag)* ')' '[' (body)? ']' + '(' union_case_tag (',' union_case_tag)* ')' '[' (body)? ']' union_case_tag: - ( numeral | numeral '-' numeral ) + ( numeral | numeral '-' numeral ) call: - '(' numeral ')' + '(' numeral ')' reference: - reference_type ( 'N' )? uint_type + reference_type ( 'N' )? uint_type reference_type: - ( constant_ref | schema_ref | utf8_ref | untyped_ref ) + ( constant_ref | schema_ref | utf8_ref | untyped_ref ) constant_ref: - ( 'KI' | 'KJ' | 'KF' | 'KD' | 'KS' | 'KQ' ) + ( 'KI' | 'KJ' | 'KF' | 'KD' | 'KS' | 'KQ' ) schema_ref: - ( 'RC' | 'RS' | 'RD' | 'RF' | 'RM' | 'RI' ) + ( 'RC' | 'RS' | 'RD' | 'RF' | 'RM' | 'RI' ) utf8_ref: - 'RU' + 'RU' untyped_ref: - 'RQ' + 'RQ' numeral: - '(' ('-')? (digit)+ ')' + '(' ('-')? (digit)+ ')' digit: - ( '0' | '1' | '2' | '3' | '4' | '5' | '6' | '7' | '8' | '9' ) + ( '0' | '1' | '2' | '3' | '4' | '5' | '6' | '7' | '8' | '9' ) */ const char *unpacker::attr_definitions::parseIntLayout(const char *lp, band *&res, byte le_kind, - bool can_be_signed) + bool can_be_signed) { - band *b = U_NEW(band, 1); - char le = *lp++; - int spec = UNSIGNED5_spec; - if (le == 'S' && can_be_signed) - { - // Note: This is the last use of sign. There is no 'EF_SIGN'. - spec = SIGNED5_spec; - le = *lp++; - } - else if (le == 'B') - { - spec = BYTE1_spec; // unsigned byte - } - b->init(u, bands_made++, spec); - b->le_kind = le_kind; - int le_len = 0; - switch (le) - { - case 'B': - le_len = 1; - break; - case 'H': - le_len = 2; - break; - case 'I': - le_len = 4; - break; - case 'V': - le_len = 0; - break; - default: - unpack_abort("bad layout element"); - } - b->le_len = le_len; - band_stack.add(b); - res = b; - return lp; + band *b = U_NEW(band, 1); + char le = *lp++; + int spec = UNSIGNED5_spec; + if (le == 'S' && can_be_signed) + { + // Note: This is the last use of sign. There is no 'EF_SIGN'. + spec = SIGNED5_spec; + le = *lp++; + } + else if (le == 'B') + { + spec = BYTE1_spec; // unsigned byte + } + b->init(u, bands_made++, spec); + b->le_kind = le_kind; + int le_len = 0; + switch (le) + { + case 'B': + le_len = 1; + break; + case 'H': + le_len = 2; + break; + case 'I': + le_len = 4; + break; + case 'V': + le_len = 0; + break; + default: + unpack_abort("bad layout element"); + } + b->le_len = le_len; + band_stack.add(b); + res = b; + return lp; } const char *unpacker::attr_definitions::parseNumeral(const char *lp, int &res) { - bool sgn = false; - if (*lp == '0') - { - res = 0; - return lp + 1; - } // special case '0' - if (*lp == '-') - { - sgn = true; - lp++; - } - const char *dp = lp; - int con = 0; - while (*dp >= '0' && *dp <= '9') - { - int con0 = con; - con *= 10; - con += (*dp++) - '0'; - if (con <= con0) - { - con = -1; - break; - } // numeral overflow - } - if (lp == dp) - { - unpack_abort("missing numeral in layout"); - } - lp = dp; - if (con < 0 && !(sgn && con == -con)) - { - // (Portability note: Misses the error if int is not 32 bits.) - unpack_abort("numeral overflow"); - } - if (sgn) - con = -con; - res = con; - return lp; + bool sgn = false; + if (*lp == '0') + { + res = 0; + return lp + 1; + } // special case '0' + if (*lp == '-') + { + sgn = true; + lp++; + } + const char *dp = lp; + int con = 0; + while (*dp >= '0' && *dp <= '9') + { + int con0 = con; + con *= 10; + con += (*dp++) - '0'; + if (con <= con0) + { + con = -1; + break; + } // numeral overflow + } + if (lp == dp) + { + unpack_abort("missing numeral in layout"); + } + lp = dp; + if (con < 0 && !(sgn && con == -con)) + { + // (Portability note: Misses the error if int is not 32 bits.) + unpack_abort("numeral overflow"); + } + if (sgn) + con = -con; + res = con; + return lp; } band **unpacker::attr_definitions::popBody(int bs_base) { - // Return everything that was pushed, as a nullptr-terminated pointer array. - int bs_limit = band_stack.length(); - if (bs_base == bs_limit) - { - return no_bands; - } - else - { - int nb = bs_limit - bs_base; - band **res = U_NEW(band *, add_size(nb, 1)); - for (int i = 0; i < nb; i++) - { - band *b = (band *)band_stack.get(bs_base + i); - res[i] = b; - } - band_stack.popTo(bs_base); - return res; - } + // Return everything that was pushed, as a nullptr-terminated pointer array. + int bs_limit = band_stack.length(); + if (bs_base == bs_limit) + { + return no_bands; + } + else + { + int nb = bs_limit - bs_base; + band **res = U_NEW(band *, add_size(nb, 1)); + for (int i = 0; i < nb; i++) + { + band *b = (band *)band_stack.get(bs_base + i); + res[i] = b; + } + band_stack.popTo(bs_base); + return res; + } } const char *unpacker::attr_definitions::parseLayout(const char *lp, band **&res, int curCble) { - int bs_base = band_stack.length(); - bool top_level = (bs_base == 0); - band *b; - enum - { - can_be_signed = true - }; // optional arg to parseIntLayout - - for (bool done = false; !done;) - { - switch (*lp++) - { - case 'B': - case 'H': - case 'I': - case 'V': // unsigned_int - case 'S': // signed_int - --lp; // reparse - case 'F': - lp = parseIntLayout(lp, b, EK_INT); - break; - case 'P': - { - int le_bci = EK_BCI; - if (*lp == 'O') - { - ++lp; - le_bci = EK_BCID; - } - assert(*lp != 'S'); // no PSH, etc. - lp = parseIntLayout(lp, b, EK_INT); - b->le_bci = le_bci; - if (le_bci == EK_BCI) - b->defc = coding::findBySpec(BCI5_spec); - else - b->defc = coding::findBySpec(BRANCH5_spec); - } - break; - case 'O': - lp = parseIntLayout(lp, b, EK_INT, can_be_signed); - b->le_bci = EK_BCO; - b->defc = coding::findBySpec(BRANCH5_spec); - break; - case 'N': // replication: 'N' uint32_t '[' elem ... ']' - lp = parseIntLayout(lp, b, EK_REPL); - assert(*lp == '['); - ++lp; - lp = parseLayout(lp, b->le_body, curCble); - break; - case 'T': // union: 'T' any_int union_case* '(' ')' '[' body ']' - lp = parseIntLayout(lp, b, EK_UN, can_be_signed); - { - int union_base = band_stack.length(); - for (;;) - { // for each case - band &k_case = *U_NEW(band, 1); - band_stack.add(&k_case); - k_case.le_kind = EK_CASE; - k_case.bn = bands_made++; - if (*lp++ != '(') - { - unpack_abort("bad union case"); - return ""; - } - if (*lp++ != ')') - { - --lp; // reparse - // Read some case values. (Use band_stack for temp. storage.) - int case_base = band_stack.length(); - for (;;) - { - int caseval = 0; - lp = parseNumeral(lp, caseval); - band_stack.add((void *)(size_t)caseval); - if (*lp == '-') - { - // new in version 160, allow (1-5) for (1,2,3,4,5) - if (u->majver < JAVA6_PACKAGE_MAJOR_VERSION) - { - unpack_abort( - "bad range in union case label (old archive format)"); - return ""; - } - int caselimit = caseval; - lp++; - lp = parseNumeral(lp, caselimit); - if (caseval >= caselimit || - (uint32_t)(caselimit - caseval) > 0x10000) - { - // Note: 0x10000 is arbitrary implementation restriction. - // We can remove it later if it's important to. - unpack_abort("bad range in union case label"); - } - for (;;) - { - ++caseval; - band_stack.add((void *)(size_t)caseval); - if (caseval == caselimit) - break; - } - } - if (*lp != ',') - break; - lp++; - } - if (*lp++ != ')') - { - unpack_abort("bad case label"); - } - // save away the case labels - int ntags = band_stack.length() - case_base; - int *tags = U_NEW(int, add_size(ntags, 1)); - k_case.le_casetags = tags; - *tags++ = ntags; - for (int i = 0; i < ntags; i++) - { - *tags++ = ptrlowbits(band_stack.get(case_base + i)); - } - band_stack.popTo(case_base); - } - // Got le_casetags. Now grab the body. - assert(*lp == '['); - ++lp; - lp = parseLayout(lp, k_case.le_body, curCble); - if (k_case.le_casetags == nullptr) - break; // done - } - b->le_body = popBody(union_base); - } - break; - case '(': // call: '(' -?NN* ')' - { - band &call = *U_NEW(band, 1); - band_stack.add(&call); - call.le_kind = EK_CALL; - call.bn = bands_made++; - call.le_body = U_NEW(band *, 2); // fill in later - int call_num = 0; - lp = parseNumeral(lp, call_num); - call.le_back = (call_num <= 0); - call_num += curCble; // numeral is self-relative offset - call.le_len = call_num; // use le_len as scratch - calls_to_link.add(&call); - if (*lp++ != ')') - { - unpack_abort("bad call label"); - } - } - break; - case 'K': // reference_type: constant_ref - case 'R': // reference_type: schema_ref - { - int ixTag = CONSTANT_None; - if (lp[-1] == 'K') - { - switch (*lp++) - { - case 'I': - ixTag = CONSTANT_Integer; - break; - case 'J': - ixTag = CONSTANT_Long; - break; - case 'F': - ixTag = CONSTANT_Float; - break; - case 'D': - ixTag = CONSTANT_Double; - break; - case 'S': - ixTag = CONSTANT_String; - break; - case 'Q': - ixTag = CONSTANT_Literal; - break; - } - } - else - { - switch (*lp++) - { - case 'C': - ixTag = CONSTANT_Class; - break; - case 'S': - ixTag = CONSTANT_Signature; - break; - case 'D': - ixTag = CONSTANT_NameandType; - break; - case 'F': - ixTag = CONSTANT_Fieldref; - break; - case 'M': - ixTag = CONSTANT_Methodref; - break; - case 'I': - ixTag = CONSTANT_InterfaceMethodref; - break; - case 'U': - ixTag = CONSTANT_Utf8; - break; // utf8_ref - case 'Q': - ixTag = CONSTANT_All; - break; // untyped_ref - } - } - if (ixTag == CONSTANT_None) - { - unpack_abort("bad reference layout"); - break; - } - bool nullOK = false; - if (*lp == 'N') - { - nullOK = true; - lp++; - } - lp = parseIntLayout(lp, b, EK_REF); - b->defc = coding::findBySpec(UNSIGNED5_spec); - b->initRef(ixTag, nullOK); - } - break; - case '[': - { - // [callable1][callable2]... - if (!top_level) - { - unpack_abort("bad nested callable"); - break; - } - curCble += 1; - band &cble = *U_NEW(band, 1); - band_stack.add(&cble); - cble.le_kind = EK_CBLE; - cble.bn = bands_made++; - lp = parseLayout(lp, cble.le_body, curCble); - } - break; - case ']': - // Hit a closing brace. This ends whatever body we were in. - done = true; - break; - case '\0': - // Hit a nullptr. Also ends the (top-level) body. - --lp; // back up, so caller can see the nullptr also - done = true; - break; - default: - unpack_abort("bad layout"); - } - } - - // Return the accumulated bands: - res = popBody(bs_base); - return lp; + int bs_base = band_stack.length(); + bool top_level = (bs_base == 0); + band *b; + enum + { + can_be_signed = true + }; // optional arg to parseIntLayout + + for (bool done = false; !done;) + { + switch (*lp++) + { + case 'B': + case 'H': + case 'I': + case 'V': // unsigned_int + case 'S': // signed_int + --lp; // reparse + case 'F': + lp = parseIntLayout(lp, b, EK_INT); + break; + case 'P': + { + int le_bci = EK_BCI; + if (*lp == 'O') + { + ++lp; + le_bci = EK_BCID; + } + assert(*lp != 'S'); // no PSH, etc. + lp = parseIntLayout(lp, b, EK_INT); + b->le_bci = le_bci; + if (le_bci == EK_BCI) + b->defc = coding::findBySpec(BCI5_spec); + else + b->defc = coding::findBySpec(BRANCH5_spec); + } + break; + case 'O': + lp = parseIntLayout(lp, b, EK_INT, can_be_signed); + b->le_bci = EK_BCO; + b->defc = coding::findBySpec(BRANCH5_spec); + break; + case 'N': // replication: 'N' uint32_t '[' elem ... ']' + lp = parseIntLayout(lp, b, EK_REPL); + assert(*lp == '['); + ++lp; + lp = parseLayout(lp, b->le_body, curCble); + break; + case 'T': // union: 'T' any_int union_case* '(' ')' '[' body ']' + lp = parseIntLayout(lp, b, EK_UN, can_be_signed); + { + int union_base = band_stack.length(); + for (;;) + { // for each case + band &k_case = *U_NEW(band, 1); + band_stack.add(&k_case); + k_case.le_kind = EK_CASE; + k_case.bn = bands_made++; + if (*lp++ != '(') + { + unpack_abort("bad union case"); + return ""; + } + if (*lp++ != ')') + { + --lp; // reparse + // Read some case values. (Use band_stack for temp. storage.) + int case_base = band_stack.length(); + for (;;) + { + int caseval = 0; + lp = parseNumeral(lp, caseval); + band_stack.add((void *)(size_t)caseval); + if (*lp == '-') + { + // new in version 160, allow (1-5) for (1,2,3,4,5) + if (u->majver < JAVA6_PACKAGE_MAJOR_VERSION) + { + unpack_abort( + "bad range in union case label (old archive format)"); + return ""; + } + int caselimit = caseval; + lp++; + lp = parseNumeral(lp, caselimit); + if (caseval >= caselimit || + (uint32_t)(caselimit - caseval) > 0x10000) + { + // Note: 0x10000 is arbitrary implementation restriction. + // We can remove it later if it's important to. + unpack_abort("bad range in union case label"); + } + for (;;) + { + ++caseval; + band_stack.add((void *)(size_t)caseval); + if (caseval == caselimit) + break; + } + } + if (*lp != ',') + break; + lp++; + } + if (*lp++ != ')') + { + unpack_abort("bad case label"); + } + // save away the case labels + int ntags = band_stack.length() - case_base; + int *tags = U_NEW(int, add_size(ntags, 1)); + k_case.le_casetags = tags; + *tags++ = ntags; + for (int i = 0; i < ntags; i++) + { + *tags++ = ptrlowbits(band_stack.get(case_base + i)); + } + band_stack.popTo(case_base); + } + // Got le_casetags. Now grab the body. + assert(*lp == '['); + ++lp; + lp = parseLayout(lp, k_case.le_body, curCble); + if (k_case.le_casetags == nullptr) + break; // done + } + b->le_body = popBody(union_base); + } + break; + case '(': // call: '(' -?NN* ')' + { + band &call = *U_NEW(band, 1); + band_stack.add(&call); + call.le_kind = EK_CALL; + call.bn = bands_made++; + call.le_body = U_NEW(band *, 2); // fill in later + int call_num = 0; + lp = parseNumeral(lp, call_num); + call.le_back = (call_num <= 0); + call_num += curCble; // numeral is self-relative offset + call.le_len = call_num; // use le_len as scratch + calls_to_link.add(&call); + if (*lp++ != ')') + { + unpack_abort("bad call label"); + } + } + break; + case 'K': // reference_type: constant_ref + case 'R': // reference_type: schema_ref + { + int ixTag = CONSTANT_None; + if (lp[-1] == 'K') + { + switch (*lp++) + { + case 'I': + ixTag = CONSTANT_Integer; + break; + case 'J': + ixTag = CONSTANT_Long; + break; + case 'F': + ixTag = CONSTANT_Float; + break; + case 'D': + ixTag = CONSTANT_Double; + break; + case 'S': + ixTag = CONSTANT_String; + break; + case 'Q': + ixTag = CONSTANT_Literal; + break; + } + } + else + { + switch (*lp++) + { + case 'C': + ixTag = CONSTANT_Class; + break; + case 'S': + ixTag = CONSTANT_Signature; + break; + case 'D': + ixTag = CONSTANT_NameandType; + break; + case 'F': + ixTag = CONSTANT_Fieldref; + break; + case 'M': + ixTag = CONSTANT_Methodref; + break; + case 'I': + ixTag = CONSTANT_InterfaceMethodref; + break; + case 'U': + ixTag = CONSTANT_Utf8; + break; // utf8_ref + case 'Q': + ixTag = CONSTANT_All; + break; // untyped_ref + } + } + if (ixTag == CONSTANT_None) + { + unpack_abort("bad reference layout"); + break; + } + bool nullOK = false; + if (*lp == 'N') + { + nullOK = true; + lp++; + } + lp = parseIntLayout(lp, b, EK_REF); + b->defc = coding::findBySpec(UNSIGNED5_spec); + b->initRef(ixTag, nullOK); + } + break; + case '[': + { + // [callable1][callable2]... + if (!top_level) + { + unpack_abort("bad nested callable"); + break; + } + curCble += 1; + band &cble = *U_NEW(band, 1); + band_stack.add(&cble); + cble.le_kind = EK_CBLE; + cble.bn = bands_made++; + lp = parseLayout(lp, cble.le_body, curCble); + } + break; + case ']': + // Hit a closing brace. This ends whatever body we were in. + done = true; + break; + case '\0': + // Hit a nullptr. Also ends the (top-level) body. + --lp; // back up, so caller can see the nullptr also + done = true; + break; + default: + unpack_abort("bad layout"); + } + } + + // Return the accumulated bands: + res = popBody(bs_base); + return lp; } void unpacker::read_attr_defs() { - int i; - - // Tell each AD which attrc it is and where its fixed flags are: - attr_defs[ATTR_CONTEXT_CLASS].attrc = ATTR_CONTEXT_CLASS; - attr_defs[ATTR_CONTEXT_CLASS].xxx_flags_hi_bn = e_class_flags_hi; - attr_defs[ATTR_CONTEXT_FIELD].attrc = ATTR_CONTEXT_FIELD; - attr_defs[ATTR_CONTEXT_FIELD].xxx_flags_hi_bn = e_field_flags_hi; - attr_defs[ATTR_CONTEXT_METHOD].attrc = ATTR_CONTEXT_METHOD; - attr_defs[ATTR_CONTEXT_METHOD].xxx_flags_hi_bn = e_method_flags_hi; - attr_defs[ATTR_CONTEXT_CODE].attrc = ATTR_CONTEXT_CODE; - attr_defs[ATTR_CONTEXT_CODE].xxx_flags_hi_bn = e_code_flags_hi; - - // Decide whether bands for the optional high flag words are present. - attr_defs[ATTR_CONTEXT_CLASS] - .setHaveLongFlags((archive_options & AO_HAVE_CLASS_FLAGS_HI) != 0); - attr_defs[ATTR_CONTEXT_FIELD] - .setHaveLongFlags((archive_options & AO_HAVE_FIELD_FLAGS_HI) != 0); - attr_defs[ATTR_CONTEXT_METHOD] - .setHaveLongFlags((archive_options & AO_HAVE_METHOD_FLAGS_HI) != 0); - attr_defs[ATTR_CONTEXT_CODE] - .setHaveLongFlags((archive_options & AO_HAVE_CODE_FLAGS_HI) != 0); - - // Set up built-in attrs. - // (The simple ones are hard-coded. The metadata layouts are not.) - const char *md_layout = ( + int i; + + // Tell each AD which attrc it is and where its fixed flags are: + attr_defs[ATTR_CONTEXT_CLASS].attrc = ATTR_CONTEXT_CLASS; + attr_defs[ATTR_CONTEXT_CLASS].xxx_flags_hi_bn = e_class_flags_hi; + attr_defs[ATTR_CONTEXT_FIELD].attrc = ATTR_CONTEXT_FIELD; + attr_defs[ATTR_CONTEXT_FIELD].xxx_flags_hi_bn = e_field_flags_hi; + attr_defs[ATTR_CONTEXT_METHOD].attrc = ATTR_CONTEXT_METHOD; + attr_defs[ATTR_CONTEXT_METHOD].xxx_flags_hi_bn = e_method_flags_hi; + attr_defs[ATTR_CONTEXT_CODE].attrc = ATTR_CONTEXT_CODE; + attr_defs[ATTR_CONTEXT_CODE].xxx_flags_hi_bn = e_code_flags_hi; + + // Decide whether bands for the optional high flag words are present. + attr_defs[ATTR_CONTEXT_CLASS] + .setHaveLongFlags((archive_options & AO_HAVE_CLASS_FLAGS_HI) != 0); + attr_defs[ATTR_CONTEXT_FIELD] + .setHaveLongFlags((archive_options & AO_HAVE_FIELD_FLAGS_HI) != 0); + attr_defs[ATTR_CONTEXT_METHOD] + .setHaveLongFlags((archive_options & AO_HAVE_METHOD_FLAGS_HI) != 0); + attr_defs[ATTR_CONTEXT_CODE] + .setHaveLongFlags((archive_options & AO_HAVE_CODE_FLAGS_HI) != 0); + + // Set up built-in attrs. + // (The simple ones are hard-coded. The metadata layouts are not.) + const char *md_layout = ( // parameter annotations: #define MDL0 "[NB[(1)]]" - MDL0 + MDL0 // annotations: #define MDL1 \ - "[NH[(1)]]" \ - "[RSHNH[RUH(1)]]" - MDL1 - // member_value: - "[TB" - "(66,67,73,83,90)[KIH]" - "(68)[KDH]" - "(70)[KFH]" - "(74)[KJH]" - "(99)[RSH]" - "(101)[RSHRUH]" - "(115)[RUH]" - "(91)[NH[(0)]]" - "(64)[" - // nested annotation: - "RSH" - "NH[RUH(0)]" - "]" - "()[]" - "]"); - - const char *md_layout_P = md_layout; - const char *md_layout_A = md_layout + strlen(MDL0); - const char *md_layout_V = md_layout + strlen(MDL0 MDL1); - assert(0 == strncmp(&md_layout_A[-3], ")]][", 4)); - assert(0 == strncmp(&md_layout_V[-3], ")]][", 4)); - - for (i = 0; i < ATTR_CONTEXT_LIMIT; i++) - { - attr_definitions &ad = attr_defs[i]; - ad.defineLayout(X_ATTR_RuntimeVisibleAnnotations, "RuntimeVisibleAnnotations", - md_layout_A); - ad.defineLayout(X_ATTR_RuntimeInvisibleAnnotations, "RuntimeInvisibleAnnotations", - md_layout_A); - if (i != ATTR_CONTEXT_METHOD) - continue; - ad.defineLayout(METHOD_ATTR_RuntimeVisibleParameterAnnotations, - "RuntimeVisibleParameterAnnotations", md_layout_P); - ad.defineLayout(METHOD_ATTR_RuntimeInvisibleParameterAnnotations, - "RuntimeInvisibleParameterAnnotations", md_layout_P); - ad.defineLayout(METHOD_ATTR_AnnotationDefault, "AnnotationDefault", md_layout_V); - } - - attr_definition_headers.readData(attr_definition_count); - attr_definition_name.readData(attr_definition_count); - attr_definition_layout.readData(attr_definition_count); + "[NH[(1)]]" \ + "[RSHNH[RUH(1)]]" + MDL1 + // member_value: + "[TB" + "(66,67,73,83,90)[KIH]" + "(68)[KDH]" + "(70)[KFH]" + "(74)[KJH]" + "(99)[RSH]" + "(101)[RSHRUH]" + "(115)[RUH]" + "(91)[NH[(0)]]" + "(64)[" + // nested annotation: + "RSH" + "NH[RUH(0)]" + "]" + "()[]" + "]"); + + const char *md_layout_P = md_layout; + const char *md_layout_A = md_layout + strlen(MDL0); + const char *md_layout_V = md_layout + strlen(MDL0 MDL1); + assert(0 == strncmp(&md_layout_A[-3], ")]][", 4)); + assert(0 == strncmp(&md_layout_V[-3], ")]][", 4)); + + for (i = 0; i < ATTR_CONTEXT_LIMIT; i++) + { + attr_definitions &ad = attr_defs[i]; + ad.defineLayout(X_ATTR_RuntimeVisibleAnnotations, "RuntimeVisibleAnnotations", + md_layout_A); + ad.defineLayout(X_ATTR_RuntimeInvisibleAnnotations, "RuntimeInvisibleAnnotations", + md_layout_A); + if (i != ATTR_CONTEXT_METHOD) + continue; + ad.defineLayout(METHOD_ATTR_RuntimeVisibleParameterAnnotations, + "RuntimeVisibleParameterAnnotations", md_layout_P); + ad.defineLayout(METHOD_ATTR_RuntimeInvisibleParameterAnnotations, + "RuntimeInvisibleParameterAnnotations", md_layout_P); + ad.defineLayout(METHOD_ATTR_AnnotationDefault, "AnnotationDefault", md_layout_V); + } + + attr_definition_headers.readData(attr_definition_count); + attr_definition_name.readData(attr_definition_count); + attr_definition_layout.readData(attr_definition_count); // Initialize correct predef bits, to distinguish predefs from new defs. #define ORBIT(n, s) | ((uint64_t)1 << n) - attr_defs[ATTR_CONTEXT_CLASS].predef = (0 X_ATTR_DO(ORBIT) CLASS_ATTR_DO(ORBIT)); - attr_defs[ATTR_CONTEXT_FIELD].predef = (0 X_ATTR_DO(ORBIT) FIELD_ATTR_DO(ORBIT)); - attr_defs[ATTR_CONTEXT_METHOD].predef = (0 X_ATTR_DO(ORBIT) METHOD_ATTR_DO(ORBIT)); - attr_defs[ATTR_CONTEXT_CODE].predef = (0 O_ATTR_DO(ORBIT) CODE_ATTR_DO(ORBIT)); + attr_defs[ATTR_CONTEXT_CLASS].predef = (0 X_ATTR_DO(ORBIT) CLASS_ATTR_DO(ORBIT)); + attr_defs[ATTR_CONTEXT_FIELD].predef = (0 X_ATTR_DO(ORBIT) FIELD_ATTR_DO(ORBIT)); + attr_defs[ATTR_CONTEXT_METHOD].predef = (0 X_ATTR_DO(ORBIT) METHOD_ATTR_DO(ORBIT)); + attr_defs[ATTR_CONTEXT_CODE].predef = (0 O_ATTR_DO(ORBIT) CODE_ATTR_DO(ORBIT)); #undef ORBIT - // Clear out the redef bits, folding them back into predef. - for (i = 0; i < ATTR_CONTEXT_LIMIT; i++) - { - attr_defs[i].predef |= attr_defs[i].redef; - attr_defs[i].redef = 0; - } - - // Now read the transmitted locally defined attrs. - // This will set redef bits again. - for (i = 0; i < attr_definition_count; i++) - { - int header = attr_definition_headers.getByte(); - int attrc = ADH_BYTE_CONTEXT(header); - int idx = ADH_BYTE_INDEX(header); - entry *name = attr_definition_name.getRef(); - entry *layout = attr_definition_layout.getRef(); - attr_defs[attrc].defineLayout(idx, name, layout->value.b.strval()); - } + // Clear out the redef bits, folding them back into predef. + for (i = 0; i < ATTR_CONTEXT_LIMIT; i++) + { + attr_defs[i].predef |= attr_defs[i].redef; + attr_defs[i].redef = 0; + } + + // Now read the transmitted locally defined attrs. + // This will set redef bits again. + for (i = 0; i < attr_definition_count; i++) + { + int header = attr_definition_headers.getByte(); + int attrc = ADH_BYTE_CONTEXT(header); + int idx = ADH_BYTE_INDEX(header); + entry *name = attr_definition_name.getRef(); + entry *layout = attr_definition_layout.getRef(); + attr_defs[attrc].defineLayout(idx, name, layout->value.b.strval()); + } } #define NO_ENTRY_YET ((entry *)-1) static bool isDigitString(bytes &x, int beg, int end) { - if (beg == end) - return false; // nullptr string - byte *xptr = x.ptr; - for (int i = beg; i < end; i++) - { - char ch = xptr[i]; - if (!(ch >= '0' && ch <= '9')) - return false; - } - return true; + if (beg == end) + return false; // nullptr string + byte *xptr = x.ptr; + for (int i = beg; i < end; i++) + { + char ch = xptr[i]; + if (!(ch >= '0' && ch <= '9')) + return false; + } + return true; } enum { // constants for parsing class names - SLASH_MIN = '.', - SLASH_MAX = '/', - DOLLAR_MIN = 0, - DOLLAR_MAX = '-'}; + SLASH_MIN = '.', + SLASH_MAX = '/', + DOLLAR_MIN = 0, + DOLLAR_MAX = '-'}; static int lastIndexOf(int chmin, int chmax, bytes &x, int pos) { - byte *ptr = x.ptr; - for (byte *cp = ptr + pos; --cp >= ptr;) - { - assert(x.inBounds(cp)); - if (*cp >= chmin && *cp <= chmax) - return (int)(cp - ptr); - } - return -1; + byte *ptr = x.ptr; + for (byte *cp = ptr + pos; --cp >= ptr;) + { + assert(x.inBounds(cp)); + if (*cp >= chmin && *cp <= chmax) + return (int)(cp - ptr); + } + return -1; } inner_class *constant_pool::getIC(entry *inner) { - if (inner == nullptr) - return nullptr; - assert(inner->tag == CONSTANT_Class); - if (inner->inord == NO_INORD) - return nullptr; - inner_class *ic = ic_index[inner->inord]; - assert(ic == nullptr || ic->inner == inner); - return ic; + if (inner == nullptr) + return nullptr; + assert(inner->tag == CONSTANT_Class); + if (inner->inord == NO_INORD) + return nullptr; + inner_class *ic = ic_index[inner->inord]; + assert(ic == nullptr || ic->inner == inner); + return ic; } inner_class *constant_pool::getFirstChildIC(entry *outer) { - if (outer == nullptr) - return nullptr; - assert(outer->tag == CONSTANT_Class); - if (outer->inord == NO_INORD) - return nullptr; - inner_class *ic = ic_child_index[outer->inord]; - assert(ic == nullptr || ic->outer == outer); - return ic; + if (outer == nullptr) + return nullptr; + assert(outer->tag == CONSTANT_Class); + if (outer->inord == NO_INORD) + return nullptr; + inner_class *ic = ic_child_index[outer->inord]; + assert(ic == nullptr || ic->outer == outer); + return ic; } inner_class *constant_pool::getNextChildIC(inner_class *child) { - inner_class *ic = child->next_sibling; - assert(ic == nullptr || ic->outer == child->outer); - return ic; + inner_class *ic = child->next_sibling; + assert(ic == nullptr || ic->outer == child->outer); + return ic; } void unpacker::read_ics() { - int i; - int index_size = cp.tag_count[CONSTANT_Class]; - inner_class **ic_index = U_NEW(inner_class *, index_size); - inner_class **ic_child_index = U_NEW(inner_class *, index_size); - cp.ic_index = ic_index; - cp.ic_child_index = ic_child_index; - ics = U_NEW(inner_class, ic_count); - ic_this_class.readData(ic_count); - ic_flags.readData(ic_count); - // Scan flags to get count of long-form bands. - int long_forms = 0; - for (i = 0; i < ic_count; i++) - { - int flags = ic_flags.getInt(); // may be long form! - if ((flags & ACC_IC_LONG_FORM) != 0) - { - long_forms += 1; - ics[i].name = NO_ENTRY_YET; - } - flags &= ~ACC_IC_LONG_FORM; - entry *inner = ic_this_class.getRef(); - uint32_t inord = inner->inord; - assert(inord < (uint32_t)cp.tag_count[CONSTANT_Class]); - if (ic_index[inord] != nullptr) - { - unpack_abort("identical inner class"); - break; - } - ic_index[inord] = &ics[i]; - ics[i].inner = inner; - ics[i].flags = flags; - assert(cp.getIC(inner) == &ics[i]); - } - // ic_this_class.done(); - // ic_flags.done(); - ic_outer_class.readData(long_forms); - ic_name.readData(long_forms); - for (i = 0; i < ic_count; i++) - { - if (ics[i].name == NO_ENTRY_YET) - { - // Long form. - ics[i].outer = ic_outer_class.getRefN(); - ics[i].name = ic_name.getRefN(); - } - else - { - // Fill in outer and name based on inner. - bytes &n = ics[i].inner->value.b; - bytes pkgOuter; - bytes number; - bytes name; - // Parse n into pkgOuter and name (and number). - int dollar1, dollar2; // pointers to $ in the pattern - // parse n = (/)*($)?($)? - int nlen = (int)n.len; - int pkglen = lastIndexOf(SLASH_MIN, SLASH_MAX, n, nlen) + 1; - dollar2 = lastIndexOf(DOLLAR_MIN, DOLLAR_MAX, n, nlen); - if (dollar2 < 0) - { - unpack_abort(); - } - assert(dollar2 >= pkglen); - if (isDigitString(n, dollar2 + 1, nlen)) - { - // n = (/)*$ - number = n.slice(dollar2 + 1, nlen); - name.set(nullptr, 0); - dollar1 = dollar2; - } - else if (pkglen < (dollar1 = lastIndexOf(DOLLAR_MIN, DOLLAR_MAX, n, dollar2 - 1)) && - isDigitString(n, dollar1 + 1, dollar2)) - { - // n = (/)*$$ - number = n.slice(dollar1 + 1, dollar2); - name = n.slice(dollar2 + 1, nlen); - } - else - { - // n = (/)*$ - dollar1 = dollar2; - number.set(nullptr, 0); - name = n.slice(dollar2 + 1, nlen); - } - if (number.ptr == nullptr) - pkgOuter = n.slice(0, dollar1); - else - pkgOuter.set(nullptr, 0); - - if (pkgOuter.ptr != nullptr) - ics[i].outer = cp.ensureClass(pkgOuter); - - if (name.ptr != nullptr) - ics[i].name = cp.ensureUtf8(name); - } - - // update child/sibling list - if (ics[i].outer != nullptr) - { - uint32_t outord = ics[i].outer->inord; - if (outord != NO_INORD) - { - assert(outord < (uint32_t)cp.tag_count[CONSTANT_Class]); - ics[i].next_sibling = ic_child_index[outord]; - ic_child_index[outord] = &ics[i]; - } - } - } - // ic_outer_class.done(); - // ic_name.done(); + int i; + int index_size = cp.tag_count[CONSTANT_Class]; + inner_class **ic_index = U_NEW(inner_class *, index_size); + inner_class **ic_child_index = U_NEW(inner_class *, index_size); + cp.ic_index = ic_index; + cp.ic_child_index = ic_child_index; + ics = U_NEW(inner_class, ic_count); + ic_this_class.readData(ic_count); + ic_flags.readData(ic_count); + // Scan flags to get count of long-form bands. + int long_forms = 0; + for (i = 0; i < ic_count; i++) + { + int flags = ic_flags.getInt(); // may be long form! + if ((flags & ACC_IC_LONG_FORM) != 0) + { + long_forms += 1; + ics[i].name = NO_ENTRY_YET; + } + flags &= ~ACC_IC_LONG_FORM; + entry *inner = ic_this_class.getRef(); + uint32_t inord = inner->inord; + assert(inord < (uint32_t)cp.tag_count[CONSTANT_Class]); + if (ic_index[inord] != nullptr) + { + unpack_abort("identical inner class"); + break; + } + ic_index[inord] = &ics[i]; + ics[i].inner = inner; + ics[i].flags = flags; + assert(cp.getIC(inner) == &ics[i]); + } + // ic_this_class.done(); + // ic_flags.done(); + ic_outer_class.readData(long_forms); + ic_name.readData(long_forms); + for (i = 0; i < ic_count; i++) + { + if (ics[i].name == NO_ENTRY_YET) + { + // Long form. + ics[i].outer = ic_outer_class.getRefN(); + ics[i].name = ic_name.getRefN(); + } + else + { + // Fill in outer and name based on inner. + bytes &n = ics[i].inner->value.b; + bytes pkgOuter; + bytes number; + bytes name; + // Parse n into pkgOuter and name (and number). + int dollar1, dollar2; // pointers to $ in the pattern + // parse n = (/)*($)?($)? + int nlen = (int)n.len; + int pkglen = lastIndexOf(SLASH_MIN, SLASH_MAX, n, nlen) + 1; + dollar2 = lastIndexOf(DOLLAR_MIN, DOLLAR_MAX, n, nlen); + if (dollar2 < 0) + { + unpack_abort(); + } + assert(dollar2 >= pkglen); + if (isDigitString(n, dollar2 + 1, nlen)) + { + // n = (/)*$ + number = n.slice(dollar2 + 1, nlen); + name.set(nullptr, 0); + dollar1 = dollar2; + } + else if (pkglen < (dollar1 = lastIndexOf(DOLLAR_MIN, DOLLAR_MAX, n, dollar2 - 1)) && + isDigitString(n, dollar1 + 1, dollar2)) + { + // n = (/)*$$ + number = n.slice(dollar1 + 1, dollar2); + name = n.slice(dollar2 + 1, nlen); + } + else + { + // n = (/)*$ + dollar1 = dollar2; + number.set(nullptr, 0); + name = n.slice(dollar2 + 1, nlen); + } + if (number.ptr == nullptr) + pkgOuter = n.slice(0, dollar1); + else + pkgOuter.set(nullptr, 0); + + if (pkgOuter.ptr != nullptr) + ics[i].outer = cp.ensureClass(pkgOuter); + + if (name.ptr != nullptr) + ics[i].name = cp.ensureUtf8(name); + } + + // update child/sibling list + if (ics[i].outer != nullptr) + { + uint32_t outord = ics[i].outer->inord; + if (outord != NO_INORD) + { + assert(outord < (uint32_t)cp.tag_count[CONSTANT_Class]); + ics[i].next_sibling = ic_child_index[outord]; + ic_child_index[outord] = &ics[i]; + } + } + } + // ic_outer_class.done(); + // ic_name.done(); } void unpacker::read_classes() { - class_this.readData(class_count); - class_super.readData(class_count); - class_interface_count.readData(class_count); - class_interface.readData(class_interface_count.getIntTotal()); + class_this.readData(class_count); + class_super.readData(class_count); + class_interface_count.readData(class_count); + class_interface.readData(class_interface_count.getIntTotal()); #if 0 int i; @@ -2256,1218 +2250,1220 @@ void unpacker::read_classes() class_super.rewind(); #endif - // Members. - class_field_count.readData(class_count); - class_method_count.readData(class_count); + // Members. + class_field_count.readData(class_count); + class_method_count.readData(class_count); - int field_count = class_field_count.getIntTotal(); - int method_count = class_method_count.getIntTotal(); + int field_count = class_field_count.getIntTotal(); + int method_count = class_method_count.getIntTotal(); - field_descr.readData(field_count); - read_attrs(ATTR_CONTEXT_FIELD, field_count); - method_descr.readData(method_count); - read_attrs(ATTR_CONTEXT_METHOD, method_count); - read_attrs(ATTR_CONTEXT_CLASS, class_count); - read_code_headers(); + field_descr.readData(field_count); + read_attrs(ATTR_CONTEXT_FIELD, field_count); + method_descr.readData(method_count); + read_attrs(ATTR_CONTEXT_METHOD, method_count); + read_attrs(ATTR_CONTEXT_CLASS, class_count); + read_code_headers(); } int unpacker::attr_definitions::predefCount(uint32_t idx) { - return isPredefined(idx) ? flag_count[idx] : 0; + return isPredefined(idx) ? flag_count[idx] : 0; } void unpacker::read_attrs(int attrc, int obj_count) { - attr_definitions &ad = attr_defs[attrc]; - assert(ad.attrc == attrc); - - int i, idx, count; - - bool haveLongFlags = ad.haveLongFlags(); - - band &xxx_flags_hi = ad.xxx_flags_hi(); - if (haveLongFlags) - xxx_flags_hi.readData(obj_count); - - band &xxx_flags_lo = ad.xxx_flags_lo(); - xxx_flags_lo.readData(obj_count); - - // pre-scan flags, counting occurrences of each index bit - uint64_t indexMask = ad.flagIndexMask(); // which flag bits are index bits? - for (i = 0; i < obj_count; i++) - { - uint64_t indexBits = xxx_flags_hi.getLong(xxx_flags_lo, haveLongFlags); - if ((indexBits & ~indexMask) > (ushort) - 1) - { - unpack_abort("undefined attribute flag bit"); - return; - } - indexBits &= indexMask; // ignore classfile flag bits - for (idx = 0; indexBits != 0; idx++, indexBits >>= 1) - { - ad.flag_count[idx] += (int)(indexBits & 1); - } - } - // we'll scan these again later for output: - xxx_flags_lo.rewind(); - xxx_flags_hi.rewind(); - - band &xxx_attr_count = ad.xxx_attr_count(); - // There is one count element for each 1<<16 bit set in flags: - xxx_attr_count.readData(ad.predefCount(X_ATTR_OVERFLOW)); - - band &xxx_attr_indexes = ad.xxx_attr_indexes(); - int overflowIndexCount = xxx_attr_count.getIntTotal(); - xxx_attr_indexes.readData(overflowIndexCount); - // pre-scan attr indexes, counting occurrences of each value - for (i = 0; i < overflowIndexCount; i++) - { - idx = xxx_attr_indexes.getInt(); - if (!ad.isIndex(idx)) - { - unpack_abort("attribute index out of bounds"); - return; - } - ad.getCount(idx) += 1; - } - xxx_attr_indexes.rewind(); // we'll scan it again later for output - - // We will need a backward call count for each used backward callable. - int backwardCounts = 0; - for (idx = 0; idx < ad.layouts.length(); idx++) - { - layout_definition *lo = ad.getLayout(idx); - if (lo != nullptr && ad.getCount(idx) != 0) - { - // Build the bands lazily, only when they are used. - band **bands = ad.buildBands(lo); - if (lo->hasCallables()) - { - for (i = 0; bands[i] != nullptr; i++) - { - if (bands[i]->le_back) - { - assert(bands[i]->le_kind == EK_CBLE); - backwardCounts += 1; - } - } - } - } - } - ad.xxx_attr_calls().readData(backwardCounts); - - // Read built-in bands. - // Mostly, these are hand-coded equivalents to readBandData(). - switch (attrc) - { - case ATTR_CONTEXT_CLASS: - - count = ad.predefCount(CLASS_ATTR_SourceFile); - class_SourceFile_RUN.readData(count); - - count = ad.predefCount(CLASS_ATTR_EnclosingMethod); - class_EnclosingMethod_RC.readData(count); - class_EnclosingMethod_RDN.readData(count); - - count = ad.predefCount(X_ATTR_Signature); - class_Signature_RS.readData(count); - - ad.readBandData(X_ATTR_RuntimeVisibleAnnotations); - ad.readBandData(X_ATTR_RuntimeInvisibleAnnotations); - - count = ad.predefCount(CLASS_ATTR_InnerClasses); - class_InnerClasses_N.readData(count); - - count = class_InnerClasses_N.getIntTotal(); - class_InnerClasses_RC.readData(count); - class_InnerClasses_F.readData(count); - - // Drop remaining columns wherever flags are zero: - count -= class_InnerClasses_F.getIntCount(0); - class_InnerClasses_outer_RCN.readData(count); - class_InnerClasses_name_RUN.readData(count); - - count = ad.predefCount(CLASS_ATTR_ClassFile_version); - class_ClassFile_version_minor_H.readData(count); - class_ClassFile_version_major_H.readData(count); - break; - - case ATTR_CONTEXT_FIELD: - - count = ad.predefCount(FIELD_ATTR_ConstantValue); - field_ConstantValue_KQ.readData(count); - - count = ad.predefCount(X_ATTR_Signature); - field_Signature_RS.readData(count); - - ad.readBandData(X_ATTR_RuntimeVisibleAnnotations); - ad.readBandData(X_ATTR_RuntimeInvisibleAnnotations); - break; - - case ATTR_CONTEXT_METHOD: - - code_count = ad.predefCount(METHOD_ATTR_Code); - // Code attrs are handled very specially below... - - count = ad.predefCount(METHOD_ATTR_Exceptions); - method_Exceptions_N.readData(count); - count = method_Exceptions_N.getIntTotal(); - method_Exceptions_RC.readData(count); - - count = ad.predefCount(X_ATTR_Signature); - method_Signature_RS.readData(count); - - ad.readBandData(X_ATTR_RuntimeVisibleAnnotations); - ad.readBandData(X_ATTR_RuntimeInvisibleAnnotations); - ad.readBandData(METHOD_ATTR_RuntimeVisibleParameterAnnotations); - ad.readBandData(METHOD_ATTR_RuntimeInvisibleParameterAnnotations); - ad.readBandData(METHOD_ATTR_AnnotationDefault); - break; - - case ATTR_CONTEXT_CODE: - // (keep this code aligned with its brother in unpacker::write_attrs) - count = ad.predefCount(CODE_ATTR_StackMapTable); - // disable this feature in old archives! - if (count != 0 && majver < JAVA6_PACKAGE_MAJOR_VERSION) - { - unpack_abort("undefined StackMapTable attribute (old archive format)"); - return; - } - code_StackMapTable_N.readData(count); - count = code_StackMapTable_N.getIntTotal(); - code_StackMapTable_frame_T.readData(count); - // the rest of it depends in a complicated way on frame tags - { - int fat_frame_count = 0; - int offset_count = 0; - int type_count = 0; - for (int k = 0; k < count; k++) - { - int tag = code_StackMapTable_frame_T.getByte(); - if (tag <= 127) - { - // (64-127) [(2)] - if (tag >= 64) - type_count++; - } - else if (tag <= 251) - { - // (247) [(1)(2)] - // (248-251) [(1)] - if (tag >= 247) - offset_count++; - if (tag == 247) - type_count++; - } - else if (tag <= 254) - { - // (252) [(1)(2)] - // (253) [(1)(2)(2)] - // (254) [(1)(2)(2)(2)] - offset_count++; - type_count += (tag - 251); - } - else - { - // (255) [(1)NH[(2)]NH[(2)]] - fat_frame_count++; - } - } - - // done pre-scanning frame tags: - code_StackMapTable_frame_T.rewind(); - - // deal completely with fat frames: - offset_count += fat_frame_count; - code_StackMapTable_local_N.readData(fat_frame_count); - type_count += code_StackMapTable_local_N.getIntTotal(); - code_StackMapTable_stack_N.readData(fat_frame_count); - type_count += code_StackMapTable_stack_N.getIntTotal(); - // read the rest: - code_StackMapTable_offset.readData(offset_count); - code_StackMapTable_T.readData(type_count); - // (7) [RCH] - count = code_StackMapTable_T.getIntCount(7); - code_StackMapTable_RC.readData(count); - // (8) [PH] - count = code_StackMapTable_T.getIntCount(8); - code_StackMapTable_P.readData(count); - } - - count = ad.predefCount(CODE_ATTR_LineNumberTable); - code_LineNumberTable_N.readData(count); - count = code_LineNumberTable_N.getIntTotal(); - code_LineNumberTable_bci_P.readData(count); - code_LineNumberTable_line.readData(count); - - count = ad.predefCount(CODE_ATTR_LocalVariableTable); - code_LocalVariableTable_N.readData(count); - count = code_LocalVariableTable_N.getIntTotal(); - code_LocalVariableTable_bci_P.readData(count); - code_LocalVariableTable_span_O.readData(count); - code_LocalVariableTable_name_RU.readData(count); - code_LocalVariableTable_type_RS.readData(count); - code_LocalVariableTable_slot.readData(count); - - count = ad.predefCount(CODE_ATTR_LocalVariableTypeTable); - code_LocalVariableTypeTable_N.readData(count); - count = code_LocalVariableTypeTable_N.getIntTotal(); - code_LocalVariableTypeTable_bci_P.readData(count); - code_LocalVariableTypeTable_span_O.readData(count); - code_LocalVariableTypeTable_name_RU.readData(count); - code_LocalVariableTypeTable_type_RS.readData(count); - code_LocalVariableTypeTable_slot.readData(count); - break; - } - - // Read compressor-defined bands. - for (idx = 0; idx < ad.layouts.length(); idx++) - { - if (ad.getLayout(idx) == nullptr) - continue; // none at this fixed index <32 - if (idx < (int)ad.flag_limit && ad.isPredefined(idx)) - continue; // already handled - if (ad.getCount(idx) == 0) - continue; // no attributes of this type (then why transmit layouts?) - ad.readBandData(idx); - } + attr_definitions &ad = attr_defs[attrc]; + assert(ad.attrc == attrc); + + int i, idx, count; + + bool haveLongFlags = ad.haveLongFlags(); + + band &xxx_flags_hi = ad.xxx_flags_hi(); + if (haveLongFlags) + xxx_flags_hi.readData(obj_count); + + band &xxx_flags_lo = ad.xxx_flags_lo(); + xxx_flags_lo.readData(obj_count); + + // pre-scan flags, counting occurrences of each index bit + uint64_t indexMask = ad.flagIndexMask(); // which flag bits are index bits? + for (i = 0; i < obj_count; i++) + { + uint64_t indexBits = xxx_flags_hi.getLong(xxx_flags_lo, haveLongFlags); + if ((indexBits & ~indexMask) > (ushort) - 1) + { + unpack_abort("undefined attribute flag bit"); + return; + } + indexBits &= indexMask; // ignore classfile flag bits + for (idx = 0; indexBits != 0; idx++, indexBits >>= 1) + { + ad.flag_count[idx] += (int)(indexBits & 1); + } + } + // we'll scan these again later for output: + xxx_flags_lo.rewind(); + xxx_flags_hi.rewind(); + + band &xxx_attr_count = ad.xxx_attr_count(); + // There is one count element for each 1<<16 bit set in flags: + xxx_attr_count.readData(ad.predefCount(X_ATTR_OVERFLOW)); + + band &xxx_attr_indexes = ad.xxx_attr_indexes(); + int overflowIndexCount = xxx_attr_count.getIntTotal(); + xxx_attr_indexes.readData(overflowIndexCount); + // pre-scan attr indexes, counting occurrences of each value + for (i = 0; i < overflowIndexCount; i++) + { + idx = xxx_attr_indexes.getInt(); + if (!ad.isIndex(idx)) + { + unpack_abort("attribute index out of bounds"); + return; + } + ad.getCount(idx) += 1; + } + xxx_attr_indexes.rewind(); // we'll scan it again later for output + + // We will need a backward call count for each used backward callable. + int backwardCounts = 0; + for (idx = 0; idx < ad.layouts.length(); idx++) + { + layout_definition *lo = ad.getLayout(idx); + if (lo != nullptr && ad.getCount(idx) != 0) + { + // Build the bands lazily, only when they are used. + band **bands = ad.buildBands(lo); + if (lo->hasCallables()) + { + for (i = 0; bands[i] != nullptr; i++) + { + if (bands[i]->le_back) + { + assert(bands[i]->le_kind == EK_CBLE); + backwardCounts += 1; + } + } + } + } + } + ad.xxx_attr_calls().readData(backwardCounts); + + // Read built-in bands. + // Mostly, these are hand-coded equivalents to readBandData(). + switch (attrc) + { + case ATTR_CONTEXT_CLASS: + + count = ad.predefCount(CLASS_ATTR_SourceFile); + class_SourceFile_RUN.readData(count); + + count = ad.predefCount(CLASS_ATTR_EnclosingMethod); + class_EnclosingMethod_RC.readData(count); + class_EnclosingMethod_RDN.readData(count); + + count = ad.predefCount(X_ATTR_Signature); + class_Signature_RS.readData(count); + + ad.readBandData(X_ATTR_RuntimeVisibleAnnotations); + ad.readBandData(X_ATTR_RuntimeInvisibleAnnotations); + + count = ad.predefCount(CLASS_ATTR_InnerClasses); + class_InnerClasses_N.readData(count); + + count = class_InnerClasses_N.getIntTotal(); + class_InnerClasses_RC.readData(count); + class_InnerClasses_F.readData(count); + + // Drop remaining columns wherever flags are zero: + count -= class_InnerClasses_F.getIntCount(0); + class_InnerClasses_outer_RCN.readData(count); + class_InnerClasses_name_RUN.readData(count); + + count = ad.predefCount(CLASS_ATTR_ClassFile_version); + class_ClassFile_version_minor_H.readData(count); + class_ClassFile_version_major_H.readData(count); + break; + + case ATTR_CONTEXT_FIELD: + + count = ad.predefCount(FIELD_ATTR_ConstantValue); + field_ConstantValue_KQ.readData(count); + + count = ad.predefCount(X_ATTR_Signature); + field_Signature_RS.readData(count); + + ad.readBandData(X_ATTR_RuntimeVisibleAnnotations); + ad.readBandData(X_ATTR_RuntimeInvisibleAnnotations); + break; + + case ATTR_CONTEXT_METHOD: + + code_count = ad.predefCount(METHOD_ATTR_Code); + // Code attrs are handled very specially below... + + count = ad.predefCount(METHOD_ATTR_Exceptions); + method_Exceptions_N.readData(count); + count = method_Exceptions_N.getIntTotal(); + method_Exceptions_RC.readData(count); + + count = ad.predefCount(X_ATTR_Signature); + method_Signature_RS.readData(count); + + ad.readBandData(X_ATTR_RuntimeVisibleAnnotations); + ad.readBandData(X_ATTR_RuntimeInvisibleAnnotations); + ad.readBandData(METHOD_ATTR_RuntimeVisibleParameterAnnotations); + ad.readBandData(METHOD_ATTR_RuntimeInvisibleParameterAnnotations); + ad.readBandData(METHOD_ATTR_AnnotationDefault); + break; + + case ATTR_CONTEXT_CODE: + // (keep this code aligned with its brother in unpacker::write_attrs) + count = ad.predefCount(CODE_ATTR_StackMapTable); + // disable this feature in old archives! + if (count != 0 && majver < JAVA6_PACKAGE_MAJOR_VERSION) + { + unpack_abort("undefined StackMapTable attribute (old archive format)"); + return; + } + code_StackMapTable_N.readData(count); + count = code_StackMapTable_N.getIntTotal(); + code_StackMapTable_frame_T.readData(count); + // the rest of it depends in a complicated way on frame tags + { + int fat_frame_count = 0; + int offset_count = 0; + int type_count = 0; + for (int k = 0; k < count; k++) + { + int tag = code_StackMapTable_frame_T.getByte(); + if (tag <= 127) + { + // (64-127) [(2)] + if (tag >= 64) + type_count++; + } + else if (tag <= 251) + { + // (247) [(1)(2)] + // (248-251) [(1)] + if (tag >= 247) + offset_count++; + if (tag == 247) + type_count++; + } + else if (tag <= 254) + { + // (252) [(1)(2)] + // (253) [(1)(2)(2)] + // (254) [(1)(2)(2)(2)] + offset_count++; + type_count += (tag - 251); + } + else + { + // (255) [(1)NH[(2)]NH[(2)]] + fat_frame_count++; + } + } + + // done pre-scanning frame tags: + code_StackMapTable_frame_T.rewind(); + + // deal completely with fat frames: + offset_count += fat_frame_count; + code_StackMapTable_local_N.readData(fat_frame_count); + type_count += code_StackMapTable_local_N.getIntTotal(); + code_StackMapTable_stack_N.readData(fat_frame_count); + type_count += code_StackMapTable_stack_N.getIntTotal(); + // read the rest: + code_StackMapTable_offset.readData(offset_count); + code_StackMapTable_T.readData(type_count); + // (7) [RCH] + count = code_StackMapTable_T.getIntCount(7); + code_StackMapTable_RC.readData(count); + // (8) [PH] + count = code_StackMapTable_T.getIntCount(8); + code_StackMapTable_P.readData(count); + } + + count = ad.predefCount(CODE_ATTR_LineNumberTable); + code_LineNumberTable_N.readData(count); + count = code_LineNumberTable_N.getIntTotal(); + code_LineNumberTable_bci_P.readData(count); + code_LineNumberTable_line.readData(count); + + count = ad.predefCount(CODE_ATTR_LocalVariableTable); + code_LocalVariableTable_N.readData(count); + count = code_LocalVariableTable_N.getIntTotal(); + code_LocalVariableTable_bci_P.readData(count); + code_LocalVariableTable_span_O.readData(count); + code_LocalVariableTable_name_RU.readData(count); + code_LocalVariableTable_type_RS.readData(count); + code_LocalVariableTable_slot.readData(count); + + count = ad.predefCount(CODE_ATTR_LocalVariableTypeTable); + code_LocalVariableTypeTable_N.readData(count); + count = code_LocalVariableTypeTable_N.getIntTotal(); + code_LocalVariableTypeTable_bci_P.readData(count); + code_LocalVariableTypeTable_span_O.readData(count); + code_LocalVariableTypeTable_name_RU.readData(count); + code_LocalVariableTypeTable_type_RS.readData(count); + code_LocalVariableTypeTable_slot.readData(count); + break; + } + + // Read compressor-defined bands. + for (idx = 0; idx < ad.layouts.length(); idx++) + { + if (ad.getLayout(idx) == nullptr) + continue; // none at this fixed index <32 + if (idx < (int)ad.flag_limit && ad.isPredefined(idx)) + continue; // already handled + if (ad.getCount(idx) == 0) + continue; // no attributes of this type (then why transmit layouts?) + ad.readBandData(idx); + } } void unpacker::attr_definitions::readBandData(int idx) { - int j; - uint32_t count = getCount(idx); - if (count == 0) - return; - layout_definition *lo = getLayout(idx); - bool hasCallables = lo->hasCallables(); - band **bands = lo->bands(); - if (!hasCallables) - { - // Read through the rest of the bands in a regular way. - readBandData(bands, count); - } - else - { - // Deal with the callables. - // First set up the forward entry count for each callable. - // This is stored on band::length of the callable. - bands[0]->expectMoreLength(count); - for (j = 0; bands[j] != nullptr; j++) - { - band &j_cble = *bands[j]; - assert(j_cble.le_kind == EK_CBLE); - if (j_cble.le_back) - { - // Add in the predicted effects of backward calls, too. - int back_calls = xxx_attr_calls().getInt(); - j_cble.expectMoreLength(back_calls); - // In a moment, more forward calls may increment j_cble.length. - } - } - // Now consult whichever callables have non-zero entry counts. - readBandData(bands, (uint32_t) - 1); - } + int j; + uint32_t count = getCount(idx); + if (count == 0) + return; + layout_definition *lo = getLayout(idx); + bool hasCallables = lo->hasCallables(); + band **bands = lo->bands(); + if (!hasCallables) + { + // Read through the rest of the bands in a regular way. + readBandData(bands, count); + } + else + { + // Deal with the callables. + // First set up the forward entry count for each callable. + // This is stored on band::length of the callable. + bands[0]->expectMoreLength(count); + for (j = 0; bands[j] != nullptr; j++) + { + band &j_cble = *bands[j]; + assert(j_cble.le_kind == EK_CBLE); + if (j_cble.le_back) + { + // Add in the predicted effects of backward calls, too. + int back_calls = xxx_attr_calls().getInt(); + j_cble.expectMoreLength(back_calls); + // In a moment, more forward calls may increment j_cble.length. + } + } + // Now consult whichever callables have non-zero entry counts. + readBandData(bands, (uint32_t) - 1); + } } // Recursive helper to the previous function: void unpacker::attr_definitions::readBandData(band **body, uint32_t count) { - int j, k; - for (j = 0; body[j] != nullptr; j++) - { - band &b = *body[j]; - if (b.defc != nullptr) - { - // It has data, so read it. - b.readData(count); - } - switch (b.le_kind) - { - case EK_REPL: - { - int reps = b.getIntTotal(); - readBandData(b.le_body, reps); - } - break; - case EK_UN: - { - int remaining = count; - for (k = 0; b.le_body[k] != nullptr; k++) - { - band &k_case = *b.le_body[k]; - int k_count = 0; - if (k_case.le_casetags == nullptr) - { - k_count = remaining; // last (empty) case - } - else - { - int *tags = k_case.le_casetags; - int ntags = *tags++; // 1st element is length (why not?) - while (ntags-- > 0) - { - int tag = *tags++; - k_count += b.getIntCount(tag); - } - } - readBandData(k_case.le_body, k_count); - remaining -= k_count; - } - assert(remaining == 0); - } - break; - case EK_CALL: - // Push the count forward, if it is not a backward call. - if (!b.le_back) - { - band &cble = *b.le_body[0]; - assert(cble.le_kind == EK_CBLE); - cble.expectMoreLength(count); - } - break; - case EK_CBLE: - assert((int)count == -1); // incoming count is meaningless - k = b.length; - assert(k >= 0); - // This is intended and required for non production mode. - assert((b.length = -1)); // make it unable to accept more calls now. - readBandData(b.le_body, k); - break; - } - } + int j, k; + for (j = 0; body[j] != nullptr; j++) + { + band &b = *body[j]; + if (b.defc != nullptr) + { + // It has data, so read it. + b.readData(count); + } + switch (b.le_kind) + { + case EK_REPL: + { + int reps = b.getIntTotal(); + readBandData(b.le_body, reps); + } + break; + case EK_UN: + { + int remaining = count; + for (k = 0; b.le_body[k] != nullptr; k++) + { + band &k_case = *b.le_body[k]; + int k_count = 0; + if (k_case.le_casetags == nullptr) + { + k_count = remaining; // last (empty) case + } + else + { + int *tags = k_case.le_casetags; + int ntags = *tags++; // 1st element is length (why not?) + while (ntags-- > 0) + { + int tag = *tags++; + k_count += b.getIntCount(tag); + } + } + readBandData(k_case.le_body, k_count); + remaining -= k_count; + } + assert(remaining == 0); + } + break; + case EK_CALL: + // Push the count forward, if it is not a backward call. + if (!b.le_back) + { + band &cble = *b.le_body[0]; + assert(cble.le_kind == EK_CBLE); + cble.expectMoreLength(count); + } + break; + case EK_CBLE: + assert((int)count == -1); // incoming count is meaningless + k = b.length; + assert(k >= 0); + // This is intended and required for non production mode. + assert((b.length = -1)); // make it unable to accept more calls now. + readBandData(b.le_body, k); + break; + } + } } static inline band **findMatchingCase(int matchTag, band **cases) { - for (int k = 0; cases[k] != nullptr; k++) - { - band &k_case = *cases[k]; - if (k_case.le_casetags != nullptr) - { - // If it has tags, it must match a tag. - int *tags = k_case.le_casetags; - int ntags = *tags++; // 1st element is length - for (; ntags > 0; ntags--) - { - int tag = *tags++; - if (tag == matchTag) - break; - } - if (ntags == 0) - continue; // does not match - } - return k_case.le_body; - } - return nullptr; + for (int k = 0; cases[k] != nullptr; k++) + { + band &k_case = *cases[k]; + if (k_case.le_casetags != nullptr) + { + // If it has tags, it must match a tag. + int *tags = k_case.le_casetags; + int ntags = *tags++; // 1st element is length + for (; ntags > 0; ntags--) + { + int tag = *tags++; + if (tag == matchTag) + break; + } + if (ntags == 0) + continue; // does not match + } + return k_case.le_body; + } + return nullptr; } // write attribute band data: void unpacker::putlayout(band **body) { - int i; - int prevBII = -1; - int prevBCI = -1; - if (body == NULL) - { - unpack_abort("putlayout: unexpected NULL for body"); - return; - } - for (i = 0; body[i] != nullptr; i++) - { - band &b = *body[i]; - byte le_kind = b.le_kind; - - // Handle scalar part, if any. - int x = 0; - entry *e = nullptr; - if (b.defc != nullptr) - { - // It has data, so unparse an element. - if (b.ixTag != CONSTANT_None) - { - assert(le_kind == EK_REF); - if (b.ixTag == CONSTANT_Literal) - e = b.getRefUsing(cp.getKQIndex()); - else - e = b.getRefN(); - switch (b.le_len) - { - case 0: - break; - case 1: - putu1ref(e); - break; - case 2: - putref(e); - break; - case 4: - putu2(0); - putref(e); - break; - default: - assert(false); - } - } - else - { - assert(le_kind == EK_INT || le_kind == EK_REPL || le_kind == EK_UN); - x = b.getInt(); - - assert(!b.le_bci || prevBCI == (int)to_bci(prevBII)); - switch (b.le_bci) - { - case EK_BCI: // PH: transmit R(bci), store bci - x = to_bci(prevBII = x); - prevBCI = x; - break; - case EK_BCID: // POH: transmit D(R(bci)), store bci - x = to_bci(prevBII += x); - prevBCI = x; - break; - case EK_BCO: // OH: transmit D(R(bci)), store D(bci) - x = to_bci(prevBII += x) - prevBCI; - prevBCI += x; - break; - } - assert(!b.le_bci || prevBCI == (int)to_bci(prevBII)); - - switch (b.le_len) - { - case 0: - break; - case 1: - putu1(x); - break; - case 2: - putu2(x); - break; - case 4: - putu4(x); - break; - default: - assert(false); - } - } - } - - // Handle subparts, if any. - switch (le_kind) - { - case EK_REPL: - // x is the repeat count - while (x-- > 0) - { - putlayout(b.le_body); - } - break; - case EK_UN: - // x is the tag - putlayout(findMatchingCase(x, b.le_body)); - break; - case EK_CALL: - { - band &cble = *b.le_body[0]; - assert(cble.le_kind == EK_CBLE); - // FIXME: hit this one - // assert(cble.le_len == b.le_len); - putlayout(cble.le_body); - } - break; - - case EK_CBLE: - case EK_CASE: - assert(false); // should not reach here - } - } + int i; + int prevBII = -1; + int prevBCI = -1; + if (body == NULL) + { + unpack_abort("putlayout: unexpected NULL for body"); + return; + } + for (i = 0; body[i] != nullptr; i++) + { + band &b = *body[i]; + byte le_kind = b.le_kind; + + // Handle scalar part, if any. + int x = 0; + entry *e = nullptr; + if (b.defc != nullptr) + { + // It has data, so unparse an element. + if (b.ixTag != CONSTANT_None) + { + assert(le_kind == EK_REF); + if (b.ixTag == CONSTANT_Literal) + e = b.getRefUsing(cp.getKQIndex()); + else + e = b.getRefN(); + switch (b.le_len) + { + case 0: + break; + case 1: + putu1ref(e); + break; + case 2: + putref(e); + break; + case 4: + putu2(0); + putref(e); + break; + default: + assert(false); + } + } + else + { + assert(le_kind == EK_INT || le_kind == EK_REPL || le_kind == EK_UN); + x = b.getInt(); + + assert(!b.le_bci || prevBCI == (int)to_bci(prevBII)); + switch (b.le_bci) + { + case EK_BCI: // PH: transmit R(bci), store bci + x = to_bci(prevBII = x); + prevBCI = x; + break; + case EK_BCID: // POH: transmit D(R(bci)), store bci + x = to_bci(prevBII += x); + prevBCI = x; + break; + case EK_BCO: // OH: transmit D(R(bci)), store D(bci) + x = to_bci(prevBII += x) - prevBCI; + prevBCI += x; + break; + } + assert(!b.le_bci || prevBCI == (int)to_bci(prevBII)); + + switch (b.le_len) + { + case 0: + break; + case 1: + putu1(x); + break; + case 2: + putu2(x); + break; + case 4: + putu4(x); + break; + default: + assert(false); + } + } + } + + // Handle subparts, if any. + switch (le_kind) + { + case EK_REPL: + // x is the repeat count + while (x-- > 0) + { + putlayout(b.le_body); + } + break; + case EK_UN: + // x is the tag + putlayout(findMatchingCase(x, b.le_body)); + break; + case EK_CALL: + { + band &cble = *b.le_body[0]; + assert(cble.le_kind == EK_CBLE); + // FIXME: hit this one + // assert(cble.le_len == b.le_len); + putlayout(cble.le_body); + } + break; + + case EK_CBLE: + case EK_CASE: + assert(false); // should not reach here + } + } } void unpacker::read_files() { - file_name.readData(file_count); - if ((archive_options & AO_HAVE_FILE_SIZE_HI) != 0) - file_size_hi.readData(file_count); - file_size_lo.readData(file_count); - if ((archive_options & AO_HAVE_FILE_MODTIME) != 0) - file_modtime.readData(file_count); - int allFiles = file_count + class_count; - if ((archive_options & AO_HAVE_FILE_OPTIONS) != 0) - { - file_options.readData(file_count); - // FO_IS_CLASS_STUB might be set, causing overlap between classes and files - for (int i = 0; i < file_count; i++) - { - if ((file_options.getInt() & FO_IS_CLASS_STUB) != 0) - { - allFiles -= 1; // this one counts as both class and file - } - } - file_options.rewind(); - } - assert((default_file_options & FO_IS_CLASS_STUB) == 0); - files_remaining = allFiles; + file_name.readData(file_count); + if ((archive_options & AO_HAVE_FILE_SIZE_HI) != 0) + file_size_hi.readData(file_count); + file_size_lo.readData(file_count); + if ((archive_options & AO_HAVE_FILE_MODTIME) != 0) + file_modtime.readData(file_count); + int allFiles = file_count + class_count; + if ((archive_options & AO_HAVE_FILE_OPTIONS) != 0) + { + file_options.readData(file_count); + // FO_IS_CLASS_STUB might be set, causing overlap between classes and files + for (int i = 0; i < file_count; i++) + { + if ((file_options.getInt() & FO_IS_CLASS_STUB) != 0) + { + allFiles -= 1; // this one counts as both class and file + } + } + file_options.rewind(); + } + assert((default_file_options & FO_IS_CLASS_STUB) == 0); + files_remaining = allFiles; } void unpacker::get_code_header(int &max_stack, int &max_na_locals, int &handler_count, - int &cflags) + int &cflags) { - int sc = code_headers.getByte(); - if (sc == 0) - { - max_stack = max_na_locals = handler_count = cflags = -1; - return; - } - // Short code header is the usual case: - int nh; - int mod; - if (sc < 1 + 12 * 12) - { - sc -= 1; - nh = 0; - mod = 12; - } - else if (sc < 1 + 12 * 12 + 8 * 8) - { - sc -= 1 + 12 * 12; - nh = 1; - mod = 8; - } - else - { - assert(sc < 1 + 12 * 12 + 8 * 8 + 7 * 7); - sc -= 1 + 12 * 12 + 8 * 8; - nh = 2; - mod = 7; - } - max_stack = sc % mod; - max_na_locals = sc / mod; // caller must add static, siglen - handler_count = nh; - if ((archive_options & AO_HAVE_ALL_CODE_FLAGS) != 0) - cflags = -1; - else - cflags = 0; // this one has no attributes + int sc = code_headers.getByte(); + if (sc == 0) + { + max_stack = max_na_locals = handler_count = cflags = -1; + return; + } + // Short code header is the usual case: + int nh; + int mod; + if (sc < 1 + 12 * 12) + { + sc -= 1; + nh = 0; + mod = 12; + } + else if (sc < 1 + 12 * 12 + 8 * 8) + { + sc -= 1 + 12 * 12; + nh = 1; + mod = 8; + } + else + { + assert(sc < 1 + 12 * 12 + 8 * 8 + 7 * 7); + sc -= 1 + 12 * 12 + 8 * 8; + nh = 2; + mod = 7; + } + max_stack = sc % mod; + max_na_locals = sc / mod; // caller must add static, siglen + handler_count = nh; + if ((archive_options & AO_HAVE_ALL_CODE_FLAGS) != 0) + cflags = -1; + else + cflags = 0; // this one has no attributes } // Cf. PackageReader.readCodeHeaders void unpacker::read_code_headers() { - code_headers.readData(code_count); - int totalHandlerCount = 0; - int totalFlagsCount = 0; - for (int i = 0; i < code_count; i++) - { - int max_stack, max_locals, handler_count, cflags; - get_code_header(max_stack, max_locals, handler_count, cflags); - if (max_stack < 0) - code_max_stack.expectMoreLength(1); - if (max_locals < 0) - code_max_na_locals.expectMoreLength(1); - if (handler_count < 0) - code_handler_count.expectMoreLength(1); - else - totalHandlerCount += handler_count; - if (cflags < 0) - totalFlagsCount += 1; - } - code_headers.rewind(); // replay later during writing - - code_max_stack.readData(); - code_max_na_locals.readData(); - code_handler_count.readData(); - totalHandlerCount += code_handler_count.getIntTotal(); - - // Read handler specifications. - // Cf. PackageReader.readCodeHandlers. - code_handler_start_P.readData(totalHandlerCount); - code_handler_end_PO.readData(totalHandlerCount); - code_handler_catch_PO.readData(totalHandlerCount); - code_handler_class_RCN.readData(totalHandlerCount); - - read_attrs(ATTR_CONTEXT_CODE, totalFlagsCount); + code_headers.readData(code_count); + int totalHandlerCount = 0; + int totalFlagsCount = 0; + for (int i = 0; i < code_count; i++) + { + int max_stack, max_locals, handler_count, cflags; + get_code_header(max_stack, max_locals, handler_count, cflags); + if (max_stack < 0) + code_max_stack.expectMoreLength(1); + if (max_locals < 0) + code_max_na_locals.expectMoreLength(1); + if (handler_count < 0) + code_handler_count.expectMoreLength(1); + else + totalHandlerCount += handler_count; + if (cflags < 0) + totalFlagsCount += 1; + } + code_headers.rewind(); // replay later during writing + + code_max_stack.readData(); + code_max_na_locals.readData(); + code_handler_count.readData(); + totalHandlerCount += code_handler_count.getIntTotal(); + + // Read handler specifications. + // Cf. PackageReader.readCodeHandlers. + code_handler_start_P.readData(totalHandlerCount); + code_handler_end_PO.readData(totalHandlerCount); + code_handler_catch_PO.readData(totalHandlerCount); + code_handler_class_RCN.readData(totalHandlerCount); + + read_attrs(ATTR_CONTEXT_CODE, totalFlagsCount); } static inline bool is_in_range(uint32_t n, uint32_t min, uint32_t max) { - return n - min <= max - min; // unsigned arithmetic! + return n - min <= max - min; // unsigned arithmetic! } static inline bool is_field_op(int bc) { - return is_in_range(bc, bc_getstatic, bc_putfield); + return is_in_range(bc, bc_getstatic, bc_putfield); } static inline bool is_invoke_init_op(int bc) { - return is_in_range(bc, _invokeinit_op, _invokeinit_limit - 1); + return is_in_range(bc, _invokeinit_op, _invokeinit_limit - 1); } static inline bool is_self_linker_op(int bc) { - return is_in_range(bc, _self_linker_op, _self_linker_limit - 1); + return is_in_range(bc, _self_linker_op, _self_linker_limit - 1); } static bool is_branch_op(int bc) { - return is_in_range(bc, bc_ifeq, bc_jsr) || is_in_range(bc, bc_ifnull, bc_jsr_w); + return is_in_range(bc, bc_ifeq, bc_jsr) || is_in_range(bc, bc_ifnull, bc_jsr_w); } static bool is_local_slot_op(int bc) { - return is_in_range(bc, bc_iload, bc_aload) || is_in_range(bc, bc_istore, bc_astore) || - bc == bc_iinc || bc == bc_ret; + return is_in_range(bc, bc_iload, bc_aload) || is_in_range(bc, bc_istore, bc_astore) || + bc == bc_iinc || bc == bc_ret; } band *unpacker::ref_band_for_op(int bc) { - switch (bc) - { - case bc_ildc: - case bc_ildc_w: - return &bc_intref; - case bc_fldc: - case bc_fldc_w: - return &bc_floatref; - case bc_lldc2_w: - return &bc_longref; - case bc_dldc2_w: - return &bc_doubleref; - case bc_aldc: - case bc_aldc_w: - return &bc_stringref; - case bc_cldc: - case bc_cldc_w: - return &bc_classref; - - case bc_getstatic: - case bc_putstatic: - case bc_getfield: - case bc_putfield: - return &bc_fieldref; - - case bc_invokevirtual: - case bc_invokespecial: - case bc_invokestatic: - return &bc_methodref; - case bc_invokeinterface: - return &bc_imethodref; - - case bc_new: - case bc_anewarray: - case bc_checkcast: - case bc_instanceof: - case bc_multianewarray: - return &bc_classref; - } - return nullptr; + switch (bc) + { + case bc_ildc: + case bc_ildc_w: + return &bc_intref; + case bc_fldc: + case bc_fldc_w: + return &bc_floatref; + case bc_lldc2_w: + return &bc_longref; + case bc_dldc2_w: + return &bc_doubleref; + case bc_aldc: + case bc_aldc_w: + return &bc_stringref; + case bc_cldc: + case bc_cldc_w: + return &bc_classref; + + case bc_getstatic: + case bc_putstatic: + case bc_getfield: + case bc_putfield: + return &bc_fieldref; + + case bc_invokevirtual: + case bc_invokespecial: + case bc_invokestatic: + return &bc_methodref; + case bc_invokeinterface: + return &bc_imethodref; + + case bc_new: + case bc_anewarray: + case bc_checkcast: + case bc_instanceof: + case bc_multianewarray: + return &bc_classref; + } + return nullptr; } band *unpacker::ref_band_for_self_op(int bc, bool &isAloadVar, int &origBCVar) { - if (!is_self_linker_op(bc)) - return nullptr; - int idx = (bc - _self_linker_op); - bool isSuper = (idx >= _self_linker_super_flag); - if (isSuper) - idx -= _self_linker_super_flag; - bool isAload = (idx >= _self_linker_aload_flag); - if (isAload) - idx -= _self_linker_aload_flag; - int origBC = _first_linker_op + idx; - bool isField = is_field_op(origBC); - isAloadVar = isAload; - origBCVar = _first_linker_op + idx; - if (!isSuper) - return isField ? &bc_thisfield : &bc_thismethod; - else - return isField ? &bc_superfield : &bc_supermethod; + if (!is_self_linker_op(bc)) + return nullptr; + int idx = (bc - _self_linker_op); + bool isSuper = (idx >= _self_linker_super_flag); + if (isSuper) + idx -= _self_linker_super_flag; + bool isAload = (idx >= _self_linker_aload_flag); + if (isAload) + idx -= _self_linker_aload_flag; + int origBC = _first_linker_op + idx; + bool isField = is_field_op(origBC); + isAloadVar = isAload; + origBCVar = _first_linker_op + idx; + if (!isSuper) + return isField ? &bc_thisfield : &bc_thismethod; + else + return isField ? &bc_superfield : &bc_supermethod; } // Cf. PackageReader.readByteCodes inline // called exactly once => inline - void + void unpacker::read_bcs() { - // read from bc_codes and bc_case_count - fillbytes all_switch_ops; - all_switch_ops.init(); - - // Read directly from rp/rplimit. - // Do this later: bc_codes.readData(...) - byte *rp0 = rp; - - band *bc_which; - byte *opptr = rp; - byte *oplimit = rplimit; - - bool isAload; // passed by ref and then ignored - int junkBC; // passed by ref and then ignored - for (int k = 0; k < code_count; k++) - { - // Scan one method: - for (;;) - { - if (opptr + 2 > oplimit) - { - rp = opptr; - ensure_input(2); - oplimit = rplimit; - rp = rp0; // back up - } - if (opptr == oplimit) - { - unpack_abort(); - } - int bc = *opptr++ & 0xFF; - bool isWide = false; - if (bc == bc_wide) - { - if (opptr == oplimit) - { - unpack_abort(); - } - bc = *opptr++ & 0xFF; - isWide = true; - } - // Adjust expectations of various band sizes. - switch (bc) - { - case bc_tableswitch: - case bc_lookupswitch: - all_switch_ops.addByte(bc); - break; - case bc_iinc: - bc_local.expectMoreLength(1); - bc_which = isWide ? &bc_short : &bc_byte; - bc_which->expectMoreLength(1); - break; - case bc_sipush: - bc_short.expectMoreLength(1); - break; - case bc_bipush: - bc_byte.expectMoreLength(1); - break; - case bc_newarray: - bc_byte.expectMoreLength(1); - break; - case bc_multianewarray: - assert(ref_band_for_op(bc) == &bc_classref); - bc_classref.expectMoreLength(1); - bc_byte.expectMoreLength(1); - break; - case bc_ref_escape: - bc_escrefsize.expectMoreLength(1); - bc_escref.expectMoreLength(1); - break; - case bc_byte_escape: - bc_escsize.expectMoreLength(1); - // bc_escbyte will have to be counted too - break; - default: - if (is_invoke_init_op(bc)) - { - bc_initref.expectMoreLength(1); - break; - } - bc_which = ref_band_for_self_op(bc, isAload, junkBC); - if (bc_which != nullptr) - { - bc_which->expectMoreLength(1); - break; - } - if (is_branch_op(bc)) - { - bc_label.expectMoreLength(1); - break; - } - bc_which = ref_band_for_op(bc); - if (bc_which != nullptr) - { - bc_which->expectMoreLength(1); - assert(bc != bc_multianewarray); // handled elsewhere - break; - } - if (is_local_slot_op(bc)) - { - bc_local.expectMoreLength(1); - break; - } - break; - case bc_end_marker: - // Increment k and test against code_count. - goto doneScanningMethod; - } - } - doneScanningMethod: - { - } - } - - // Go through the formality, so we can use it in a regular fashion later: - assert(rp == rp0); - bc_codes.readData((int)(opptr - rp)); - - int i = 0; - - // To size instruction bands correctly, we need info on switches: - bc_case_count.readData((int)all_switch_ops.size()); - for (i = 0; i < (int)all_switch_ops.size(); i++) - { - int caseCount = bc_case_count.getInt(); - int bc = all_switch_ops.getByte(i); - bc_label.expectMoreLength(1 + caseCount); // default label + cases - bc_case_value.expectMoreLength(bc == bc_tableswitch ? 1 : caseCount); - } - bc_case_count.rewind(); // uses again for output - - all_switch_ops.free(); - - for (i = e_bc_case_value; i <= e_bc_escsize; i++) - { - all_bands[i].readData(); - } - - // The bc_escbyte band is counted by the immediately previous band. - bc_escbyte.readData(bc_escsize.getIntTotal()); + // read from bc_codes and bc_case_count + fillbytes all_switch_ops; + all_switch_ops.init(); + + // Read directly from rp/rplimit. + // Do this later: bc_codes.readData(...) + byte *rp0 = rp; + + band *bc_which; + byte *opptr = rp; + byte *oplimit = rplimit; + + bool isAload; // passed by ref and then ignored + int junkBC; // passed by ref and then ignored + for (int k = 0; k < code_count; k++) + { + // Scan one method: + for (;;) + { + if (opptr + 2 > oplimit) + { + rp = opptr; + ensure_input(2); + oplimit = rplimit; + rp = rp0; // back up + } + if (opptr == oplimit) + { + unpack_abort(); + } + int bc = *opptr++ & 0xFF; + bool isWide = false; + if (bc == bc_wide) + { + if (opptr == oplimit) + { + unpack_abort(); + } + bc = *opptr++ & 0xFF; + isWide = true; + } + // Adjust expectations of various band sizes. + switch (bc) + { + case bc_tableswitch: + case bc_lookupswitch: + all_switch_ops.addByte(bc); + break; + case bc_iinc: + bc_local.expectMoreLength(1); + bc_which = isWide ? &bc_short : &bc_byte; + bc_which->expectMoreLength(1); + break; + case bc_sipush: + bc_short.expectMoreLength(1); + break; + case bc_bipush: + bc_byte.expectMoreLength(1); + break; + case bc_newarray: + bc_byte.expectMoreLength(1); + break; + case bc_multianewarray: + assert(ref_band_for_op(bc) == &bc_classref); + bc_classref.expectMoreLength(1); + bc_byte.expectMoreLength(1); + break; + case bc_ref_escape: + bc_escrefsize.expectMoreLength(1); + bc_escref.expectMoreLength(1); + break; + case bc_byte_escape: + bc_escsize.expectMoreLength(1); + // bc_escbyte will have to be counted too + break; + default: + if (is_invoke_init_op(bc)) + { + bc_initref.expectMoreLength(1); + break; + } + bc_which = ref_band_for_self_op(bc, isAload, junkBC); + if (bc_which != nullptr) + { + bc_which->expectMoreLength(1); + break; + } + if (is_branch_op(bc)) + { + bc_label.expectMoreLength(1); + break; + } + bc_which = ref_band_for_op(bc); + if (bc_which != nullptr) + { + bc_which->expectMoreLength(1); + assert(bc != bc_multianewarray); // handled elsewhere + break; + } + if (is_local_slot_op(bc)) + { + bc_local.expectMoreLength(1); + break; + } + break; + case bc_end_marker: + // Increment k and test against code_count. + goto doneScanningMethod; + } + } + doneScanningMethod: + { + } + } + + // Go through the formality, so we can use it in a regular fashion later: + assert(rp == rp0); + bc_codes.readData((int)(opptr - rp)); + + int i = 0; + + // To size instruction bands correctly, we need info on switches: + bc_case_count.readData((int)all_switch_ops.size()); + for (i = 0; i < (int)all_switch_ops.size(); i++) + { + int caseCount = bc_case_count.getInt(); + int bc = all_switch_ops.getByte(i); + bc_label.expectMoreLength(1 + caseCount); // default label + cases + bc_case_value.expectMoreLength(bc == bc_tableswitch ? 1 : caseCount); + } + bc_case_count.rewind(); // uses again for output + + all_switch_ops.free(); + + for (i = e_bc_case_value; i <= e_bc_escsize; i++) + { + all_bands[i].readData(); + } + + // The bc_escbyte band is counted by the immediately previous band. + bc_escbyte.readData(bc_escsize.getIntTotal()); } void unpacker::read_bands() { - read_file_header(); - - if (cp.nentries == 0) - { - // read_file_header failed to read a CP, because it copied a JAR. - return; - } - - // Do this after the file header has been read: - check_options(); - - read_cp(); - read_attr_defs(); - read_ics(); - read_classes(); - read_bcs(); - read_files(); + read_file_header(); + + if (cp.nentries == 0) + { + // read_file_header failed to read a CP, because it copied a JAR. + return; + } + + // Do this after the file header has been read: + check_options(); + + read_cp(); + read_attr_defs(); + read_ics(); + read_classes(); + read_bcs(); + read_files(); } /// CP routines entry *&constant_pool::hashTabRef(byte tag, bytes &b) { - uint32_t hash = tag + (int)b.len; - for (int i = 0; i < (int)b.len; i++) - { - hash = hash * 31 + (0xFF & b.ptr[i]); - } - entry **ht = hashTab; - int hlen = hashTabLength; - assert((hlen & (hlen - 1)) == 0); // must be power of 2 - uint32_t hash1 = hash & (hlen - 1); // == hash % hlen - uint32_t hash2 = 0; // lazily computed (requires mod op.) - int probes = 0; - while (ht[hash1] != nullptr) - { - entry &e = *ht[hash1]; - if (e.value.b.equals(b) && e.tag == tag) - break; - if (hash2 == 0) - // Note: hash2 must be relatively prime to hlen, hence the "|1". - hash2 = (((hash % 499) & (hlen - 1)) | 1); - hash1 += hash2; - if (hash1 >= (uint32_t)hlen) - hash1 -= hlen; - assert(hash1 < (uint32_t)hlen); - assert(++probes < hlen); - } - return ht[hash1]; + uint32_t hash = tag + (int)b.len; + for (int i = 0; i < (int)b.len; i++) + { + hash = hash * 31 + (0xFF & b.ptr[i]); + } + entry **ht = hashTab; + int hlen = hashTabLength; + assert((hlen & (hlen - 1)) == 0); // must be power of 2 + uint32_t hash1 = hash & (hlen - 1); // == hash % hlen + uint32_t hash2 = 0; // lazily computed (requires mod op.) +#ifndef NDEBUG + int probes = 0; +#endif + while (ht[hash1] != nullptr) + { + entry &e = *ht[hash1]; + if (e.value.b.equals(b) && e.tag == tag) + break; + if (hash2 == 0) + // Note: hash2 must be relatively prime to hlen, hence the "|1". + hash2 = (((hash % 499) & (hlen - 1)) | 1); + hash1 += hash2; + if (hash1 >= (uint32_t)hlen) + hash1 -= hlen; + assert(hash1 < (uint32_t)hlen); + assert(++probes < hlen); + } + return ht[hash1]; } static void insert_extra(entry *e, ptrlist &extras) { - // This ordering helps implement the Pack200 requirement - // of a predictable CP order in the class files produced. - e->inord = NO_INORD; // mark as an "extra" - extras.add(e); - // Note: We will sort the list (by string-name) later. + // This ordering helps implement the Pack200 requirement + // of a predictable CP order in the class files produced. + e->inord = NO_INORD; // mark as an "extra" + extras.add(e); + // Note: We will sort the list (by string-name) later. } entry *constant_pool::ensureUtf8(bytes &b) { - entry *&ix = hashTabRef(CONSTANT_Utf8, b); - if (ix != nullptr) - return ix; - // Make one. - if (nentries == maxentries) - { - unpack_abort("cp utf8 overflow"); - return &entries[tag_base[CONSTANT_Utf8]]; // return something - } - entry &e = entries[nentries++]; - e.tag = CONSTANT_Utf8; - u->saveTo(e.value.b, b); - assert(&e >= first_extra_entry); - insert_extra(&e, tag_extras[CONSTANT_Utf8]); - return ix = &e; + entry *&ix = hashTabRef(CONSTANT_Utf8, b); + if (ix != nullptr) + return ix; + // Make one. + if (nentries == maxentries) + { + unpack_abort("cp utf8 overflow"); + return &entries[tag_base[CONSTANT_Utf8]]; // return something + } + entry &e = entries[nentries++]; + e.tag = CONSTANT_Utf8; + u->saveTo(e.value.b, b); + assert(&e >= first_extra_entry); + insert_extra(&e, tag_extras[CONSTANT_Utf8]); + return ix = &e; } entry *constant_pool::ensureClass(bytes &b) { - entry *&ix = hashTabRef(CONSTANT_Class, b); - if (ix != nullptr) - return ix; - // Make one. - if (nentries == maxentries) - { - unpack_abort("cp class overflow"); - return &entries[tag_base[CONSTANT_Class]]; // return something - } - entry &e = entries[nentries++]; - e.tag = CONSTANT_Class; - e.nrefs = 1; - e.refs = U_NEW(entry *, 1); - ix = &e; // hold my spot in the index - entry *utf = ensureUtf8(b); - e.refs[0] = utf; - e.value.b = utf->value.b; - assert(&e >= first_extra_entry); - insert_extra(&e, tag_extras[CONSTANT_Class]); - return &e; + entry *&ix = hashTabRef(CONSTANT_Class, b); + if (ix != nullptr) + return ix; + // Make one. + if (nentries == maxentries) + { + unpack_abort("cp class overflow"); + return &entries[tag_base[CONSTANT_Class]]; // return something + } + entry &e = entries[nentries++]; + e.tag = CONSTANT_Class; + e.nrefs = 1; + e.refs = U_NEW(entry *, 1); + ix = &e; // hold my spot in the index + entry *utf = ensureUtf8(b); + e.refs[0] = utf; + e.value.b = utf->value.b; + assert(&e >= first_extra_entry); + insert_extra(&e, tag_extras[CONSTANT_Class]); + return &e; } void constant_pool::expandSignatures() { - int i; - int nsigs = 0; - int nreused = 0; - int first_sig = tag_base[CONSTANT_Signature]; - int sig_limit = tag_count[CONSTANT_Signature] + first_sig; - fillbytes buf; - buf.init(1 << 10); - for (i = first_sig; i < sig_limit; i++) - { - entry &e = entries[i]; - assert(e.tag == CONSTANT_Signature); - int refnum = 0; - bytes form = e.refs[refnum++]->asUtf8(); - buf.empty(); - for (int j = 0; j < (int)form.len; j++) - { - int c = form.ptr[j]; - buf.addByte(c); - if (c == 'L') - { - entry *cls = e.refs[refnum++]; - buf.append(cls->className()->asUtf8()); - } - } - assert(refnum == e.nrefs); - bytes &sig = buf.b; - - // try to find a pre-existing Utf8: - entry *&e2 = hashTabRef(CONSTANT_Utf8, sig); - if (e2 != nullptr) - { - assert(e2->isUtf8(sig)); - e.value.b = e2->value.b; - e.refs[0] = e2; - e.nrefs = 1; - nreused++; - } - else - { - // there is no other replacement; reuse this CP entry as a Utf8 - u->saveTo(e.value.b, sig); - e.tag = CONSTANT_Utf8; - e.nrefs = 0; - e2 = &e; - } - nsigs++; - } - buf.free(); - - // go expunge all references to remaining signatures: - for (i = 0; i < (int)nentries; i++) - { - entry &e = entries[i]; - for (int j = 0; j < e.nrefs; j++) - { - entry *&e2 = e.refs[j]; - if (e2 != nullptr && e2->tag == CONSTANT_Signature) - e2 = e2->refs[0]; - } - } + int i; + int nsigs = 0; + int nreused = 0; + int first_sig = tag_base[CONSTANT_Signature]; + int sig_limit = tag_count[CONSTANT_Signature] + first_sig; + fillbytes buf; + buf.init(1 << 10); + for (i = first_sig; i < sig_limit; i++) + { + entry &e = entries[i]; + assert(e.tag == CONSTANT_Signature); + int refnum = 0; + bytes form = e.refs[refnum++]->asUtf8(); + buf.empty(); + for (int j = 0; j < (int)form.len; j++) + { + int c = form.ptr[j]; + buf.addByte(c); + if (c == 'L') + { + entry *cls = e.refs[refnum++]; + buf.append(cls->className()->asUtf8()); + } + } + assert(refnum == e.nrefs); + bytes &sig = buf.b; + + // try to find a pre-existing Utf8: + entry *&e2 = hashTabRef(CONSTANT_Utf8, sig); + if (e2 != nullptr) + { + assert(e2->isUtf8(sig)); + e.value.b = e2->value.b; + e.refs[0] = e2; + e.nrefs = 1; + nreused++; + } + else + { + // there is no other replacement; reuse this CP entry as a Utf8 + u->saveTo(e.value.b, sig); + e.tag = CONSTANT_Utf8; + e.nrefs = 0; + e2 = &e; + } + nsigs++; + } + buf.free(); + + // go expunge all references to remaining signatures: + for (i = 0; i < (int)nentries; i++) + { + entry &e = entries[i]; + for (int j = 0; j < e.nrefs; j++) + { + entry *&e2 = e.refs[j]; + if (e2 != nullptr && e2->tag == CONSTANT_Signature) + e2 = e2->refs[0]; + } + } } void constant_pool::initMemberIndexes() { - // This function does NOT refer to any class schema. - // It is totally internal to the cpool. - int i, j; - - // Get the pre-existing indexes: - int nclasses = tag_count[CONSTANT_Class]; - // entry *classes = tag_base[CONSTANT_Class] + entries; // UNUSED - int nfields = tag_count[CONSTANT_Fieldref]; - entry *fields = tag_base[CONSTANT_Fieldref] + entries; - int nmethods = tag_count[CONSTANT_Methodref]; - entry *methods = tag_base[CONSTANT_Methodref] + entries; - - int *field_counts = T_NEW(int, nclasses); - int *method_counts = T_NEW(int, nclasses); - cpindex *all_indexes = U_NEW(cpindex, nclasses * 2); - entry **field_ix = U_NEW(entry *, add_size(nfields, nclasses)); - entry **method_ix = U_NEW(entry *, add_size(nmethods, nclasses)); - - for (j = 0; j < nfields; j++) - { - entry &f = fields[j]; - i = f.memberClass()->inord; - assert(i < nclasses); - field_counts[i]++; - } - for (j = 0; j < nmethods; j++) - { - entry &m = methods[j]; - i = m.memberClass()->inord; - assert(i < nclasses); - method_counts[i]++; - } - - int fbase = 0, mbase = 0; - for (i = 0; i < nclasses; i++) - { - int fc = field_counts[i]; - int mc = method_counts[i]; - all_indexes[i * 2 + 0].init(fc, field_ix + fbase, CONSTANT_Fieldref + SUBINDEX_BIT); - all_indexes[i * 2 + 1].init(mc, method_ix + mbase, CONSTANT_Methodref + SUBINDEX_BIT); - // reuse field_counts and member_counts as fill pointers: - field_counts[i] = fbase; - method_counts[i] = mbase; - fbase += fc + 1; - mbase += mc + 1; - // (the +1 leaves a space between every subarray) - } - assert(fbase == nfields + nclasses); - assert(mbase == nmethods + nclasses); - - for (j = 0; j < nfields; j++) - { - entry &f = fields[j]; - i = f.memberClass()->inord; - field_ix[field_counts[i]++] = &f; - } - for (j = 0; j < nmethods; j++) - { - entry &m = methods[j]; - i = m.memberClass()->inord; - method_ix[method_counts[i]++] = &m; - } - - member_indexes = all_indexes; - - // Free intermediate buffers. - u->free_temps(); + // This function does NOT refer to any class schema. + // It is totally internal to the cpool. + int i, j; + + // Get the pre-existing indexes: + int nclasses = tag_count[CONSTANT_Class]; + // entry *classes = tag_base[CONSTANT_Class] + entries; // UNUSED + int nfields = tag_count[CONSTANT_Fieldref]; + entry *fields = tag_base[CONSTANT_Fieldref] + entries; + int nmethods = tag_count[CONSTANT_Methodref]; + entry *methods = tag_base[CONSTANT_Methodref] + entries; + + int *field_counts = T_NEW(int, nclasses); + int *method_counts = T_NEW(int, nclasses); + cpindex *all_indexes = U_NEW(cpindex, nclasses * 2); + entry **field_ix = U_NEW(entry *, add_size(nfields, nclasses)); + entry **method_ix = U_NEW(entry *, add_size(nmethods, nclasses)); + + for (j = 0; j < nfields; j++) + { + entry &f = fields[j]; + i = f.memberClass()->inord; + assert(i < nclasses); + field_counts[i]++; + } + for (j = 0; j < nmethods; j++) + { + entry &m = methods[j]; + i = m.memberClass()->inord; + assert(i < nclasses); + method_counts[i]++; + } + + int fbase = 0, mbase = 0; + for (i = 0; i < nclasses; i++) + { + int fc = field_counts[i]; + int mc = method_counts[i]; + all_indexes[i * 2 + 0].init(fc, field_ix + fbase, CONSTANT_Fieldref + SUBINDEX_BIT); + all_indexes[i * 2 + 1].init(mc, method_ix + mbase, CONSTANT_Methodref + SUBINDEX_BIT); + // reuse field_counts and member_counts as fill pointers: + field_counts[i] = fbase; + method_counts[i] = mbase; + fbase += fc + 1; + mbase += mc + 1; + // (the +1 leaves a space between every subarray) + } + assert(fbase == nfields + nclasses); + assert(mbase == nmethods + nclasses); + + for (j = 0; j < nfields; j++) + { + entry &f = fields[j]; + i = f.memberClass()->inord; + field_ix[field_counts[i]++] = &f; + } + for (j = 0; j < nmethods; j++) + { + entry &m = methods[j]; + i = m.memberClass()->inord; + method_ix[method_counts[i]++] = &m; + } + + member_indexes = all_indexes; + + // Free intermediate buffers. + u->free_temps(); } void entry::requestOutputIndex(constant_pool &cp, int req) { - assert(outputIndex <= NOT_REQUESTED); // must not have assigned indexes yet - if (tag == CONSTANT_Signature) - { - ref(0)->requestOutputIndex(cp, req); - return; - } - assert(req == REQUESTED || req == REQUESTED_LDC); - if (outputIndex != NOT_REQUESTED) - { - if (req == REQUESTED_LDC) - outputIndex = req; // this kind has precedence - return; - } - outputIndex = req; - // assert(!cp.outputEntries.contains(this)); - assert(tag != CONSTANT_Signature); - cp.outputEntries.add(this); - for (int j = 0; j < nrefs; j++) - { - ref(j)->requestOutputIndex(cp); - } + assert(outputIndex <= NOT_REQUESTED); // must not have assigned indexes yet + if (tag == CONSTANT_Signature) + { + ref(0)->requestOutputIndex(cp, req); + return; + } + assert(req == REQUESTED || req == REQUESTED_LDC); + if (outputIndex != NOT_REQUESTED) + { + if (req == REQUESTED_LDC) + outputIndex = req; // this kind has precedence + return; + } + outputIndex = req; + // assert(!cp.outputEntries.contains(this)); + assert(tag != CONSTANT_Signature); + cp.outputEntries.add(this); + for (int j = 0; j < nrefs; j++) + { + ref(j)->requestOutputIndex(cp); + } } void constant_pool::resetOutputIndexes() { - int i; - int noes = outputEntries.length(); - entry **oes = (entry **)outputEntries.base(); - for (i = 0; i < noes; i++) - { - entry &e = *oes[i]; - e.outputIndex = NOT_REQUESTED; - } - outputIndexLimit = 0; - outputEntries.empty(); + int i; + int noes = outputEntries.length(); + entry **oes = (entry **)outputEntries.base(); + for (i = 0; i < noes; i++) + { + entry &e = *oes[i]; + e.outputIndex = NOT_REQUESTED; + } + outputIndexLimit = 0; + outputEntries.empty(); } static const byte TAG_ORDER[CONSTANT_Limit] = {0, 1, 0, 2, 3, 4, 5, 7, 6, 10, 11, 12, 9, 8}; extern "C" int outputEntry_cmp(const void *e1p, const void *e2p) { - // Sort entries according to the Pack200 rules for deterministic - // constant pool ordering. - // - // The four sort keys as follows, in order of decreasing importance: - // 1. ldc first, then non-ldc guys - // 2. normal cp_All entries by input order (i.e., address order) - // 3. after that, extra entries by lexical order (as in tag_extras[*]) - entry &e1 = *(entry *)*(void **)e1p; - entry &e2 = *(entry *)*(void **)e2p; - int oi1 = e1.outputIndex; - int oi2 = e2.outputIndex; - assert(oi1 == REQUESTED || oi1 == REQUESTED_LDC); - assert(oi2 == REQUESTED || oi2 == REQUESTED_LDC); - if (oi1 != oi2) - { - if (oi1 == REQUESTED_LDC) - return 0 - 1; - if (oi2 == REQUESTED_LDC) - return 1 - 0; - // Else fall through; neither is an ldc request. - } - if (e1.inord != NO_INORD || e2.inord != NO_INORD) - { - // One or both is normal. Use input order. - if (&e1 > &e2) - return 1 - 0; - if (&e1 < &e2) - return 0 - 1; - return 0; // equal pointers - } - // Both are extras. Sort by tag and then by value. - if (e1.tag != e2.tag) - { - return TAG_ORDER[e1.tag] - TAG_ORDER[e2.tag]; - } - // If the tags are the same, use string comparison. - return compare_Utf8_chars(e1.value.b, e2.value.b); + // Sort entries according to the Pack200 rules for deterministic + // constant pool ordering. + // + // The four sort keys as follows, in order of decreasing importance: + // 1. ldc first, then non-ldc guys + // 2. normal cp_All entries by input order (i.e., address order) + // 3. after that, extra entries by lexical order (as in tag_extras[*]) + entry &e1 = *(entry *)*(void **)e1p; + entry &e2 = *(entry *)*(void **)e2p; + int oi1 = e1.outputIndex; + int oi2 = e2.outputIndex; + assert(oi1 == REQUESTED || oi1 == REQUESTED_LDC); + assert(oi2 == REQUESTED || oi2 == REQUESTED_LDC); + if (oi1 != oi2) + { + if (oi1 == REQUESTED_LDC) + return 0 - 1; + if (oi2 == REQUESTED_LDC) + return 1 - 0; + // Else fall through; neither is an ldc request. + } + if (e1.inord != NO_INORD || e2.inord != NO_INORD) + { + // One or both is normal. Use input order. + if (&e1 > &e2) + return 1 - 0; + if (&e1 < &e2) + return 0 - 1; + return 0; // equal pointers + } + // Both are extras. Sort by tag and then by value. + if (e1.tag != e2.tag) + { + return TAG_ORDER[e1.tag] - TAG_ORDER[e2.tag]; + } + // If the tags are the same, use string comparison. + return compare_Utf8_chars(e1.value.b, e2.value.b); } void constant_pool::computeOutputIndexes() { - int i; - - int noes = outputEntries.length(); - entry **oes = (entry **)outputEntries.base(); - - // Sort the output constant pool into the order required by Pack200. - PTRLIST_QSORT(outputEntries, outputEntry_cmp); - - // Allocate a new index for each entry that needs one. - // We do this in two passes, one for LDC entries and one for the rest. - int nextIndex = 1; // always skip index #0 in output cpool - for (i = 0; i < noes; i++) - { - entry &e = *oes[i]; - assert(e.outputIndex == REQUESTED || e.outputIndex == REQUESTED_LDC); - e.outputIndex = nextIndex++; - if (e.isDoubleWord()) - nextIndex++; // do not use the next index - } - outputIndexLimit = nextIndex; + int i; + + int noes = outputEntries.length(); + entry **oes = (entry **)outputEntries.base(); + + // Sort the output constant pool into the order required by Pack200. + PTRLIST_QSORT(outputEntries, outputEntry_cmp); + + // Allocate a new index for each entry that needs one. + // We do this in two passes, one for LDC entries and one for the rest. + int nextIndex = 1; // always skip index #0 in output cpool + for (i = 0; i < noes; i++) + { + entry &e = *oes[i]; + assert(e.outputIndex == REQUESTED || e.outputIndex == REQUESTED_LDC); + e.outputIndex = nextIndex++; + if (e.isDoubleWord()) + nextIndex++; // do not use the next index + } + outputIndexLimit = nextIndex; } // Unpacker Start @@ -3477,64 +3473,63 @@ void constant_pool::computeOutputIndexes() // Do not reset any unpack options. void unpacker::reset() { - bytes_read_before_reset += bytes_read; - bytes_written_before_reset += bytes_written; - files_written_before_reset += files_written; - classes_written_before_reset += classes_written; - segments_read_before_reset += 1; - if (verbose >= 2) - { - fprintf(stderr, "After segment %d, " LONG_LONG_FORMAT - " bytes read and " LONG_LONG_FORMAT " bytes written.\n", - segments_read_before_reset - 1, bytes_read_before_reset, - bytes_written_before_reset); - fprintf(stderr, - "After segment %d, %d files (of which %d are classes) written to output.\n", - segments_read_before_reset - 1, files_written_before_reset, - classes_written_before_reset); - if (archive_next_count != 0) - { - fprintf(stderr, "After segment %d, %d segment%s remaining (estimated).\n", - segments_read_before_reset - 1, archive_next_count, - archive_next_count == 1 ? "" : "s"); - } - } - - unpacker save_u = (*this); // save bytewise image - infileptr = nullptr; // make asserts happy - jarout = nullptr; // do not close the output jar - gzin = nullptr; // do not close the input gzip stream - this->free(); - this->init(read_input_fn); - - // restore selected interface state: - infileptr = save_u.infileptr; - inbytes = save_u.inbytes; - jarout = save_u.jarout; - gzin = save_u.gzin; - verbose = save_u.verbose; - deflate_hint_or_zero = save_u.deflate_hint_or_zero; - modification_time_or_zero = save_u.modification_time_or_zero; - bytes_read_before_reset = save_u.bytes_read_before_reset; - bytes_written_before_reset = save_u.bytes_written_before_reset; - files_written_before_reset = save_u.files_written_before_reset; - classes_written_before_reset = save_u.classes_written_before_reset; - segments_read_before_reset = save_u.segments_read_before_reset; - // Note: If we use strip_names, watch out: They get nuked here. + bytes_read_before_reset += bytes_read; + bytes_written_before_reset += bytes_written; + files_written_before_reset += files_written; + classes_written_before_reset += classes_written; + segments_read_before_reset += 1; + if (verbose >= 2) + { + fprintf(stderr, "After segment %d, %" PRIu64 " bytes read and %" PRIu64 " bytes written.\n", + segments_read_before_reset - 1, bytes_read_before_reset, + bytes_written_before_reset); + fprintf(stderr, + "After segment %d, %d files (of which %d are classes) written to output.\n", + segments_read_before_reset - 1, files_written_before_reset, + classes_written_before_reset); + if (archive_next_count != 0) + { + fprintf(stderr, "After segment %d, %d segment%s remaining (estimated).\n", + segments_read_before_reset - 1, archive_next_count, + archive_next_count == 1 ? "" : "s"); + } + } + + unpacker save_u = (*this); // save bytewise image + infileptr = nullptr; // make asserts happy + jarout = nullptr; // do not close the output jar + gzin = nullptr; // do not close the input gzip stream + this->free(); + this->init(read_input_fn); + + // restore selected interface state: + infileptr = save_u.infileptr; + inbytes = save_u.inbytes; + jarout = save_u.jarout; + gzin = save_u.gzin; + verbose = save_u.verbose; + deflate_hint_or_zero = save_u.deflate_hint_or_zero; + modification_time_or_zero = save_u.modification_time_or_zero; + bytes_read_before_reset = save_u.bytes_read_before_reset; + bytes_written_before_reset = save_u.bytes_written_before_reset; + files_written_before_reset = save_u.files_written_before_reset; + classes_written_before_reset = save_u.classes_written_before_reset; + segments_read_before_reset = save_u.segments_read_before_reset; + // Note: If we use strip_names, watch out: They get nuked here. } void unpacker::init(read_input_fn_t input_fn) { - int i; - BYTES_OF(*this).clear(); - this->u = this; // self-reference for U_NEW macro - read_input_fn = input_fn; - all_bands = band::makeBands(this); - // Make a default jar buffer; caller may safely overwrite it. - jarout = U_NEW(jar, 1); - jarout->init(this); - for (i = 0; i < ATTR_CONTEXT_LIMIT; i++) - attr_defs[i].u = u; // set up outer ptr + int i; + BYTES_OF(*this).clear(); + this->u = this; // self-reference for U_NEW macro + read_input_fn = input_fn; + all_bands = band::makeBands(this); + // Make a default jar buffer; caller may safely overwrite it. + jarout = U_NEW(jar, 1); + jarout->init(this); + for (i = 0; i < ATTR_CONTEXT_LIMIT; i++) + attr_defs[i].u = u; // set up outer ptr } // Usage: unpack a byte buffer @@ -3543,1251 +3538,1253 @@ void unpacker::init(read_input_fn_t input_fn) // If nullptr, the callback is used to fill an internal buffer. void unpacker::start(void *packptr, size_t len) { - if (packptr != nullptr && len != 0) - { - inbytes.set((byte *)packptr, len); - } - read_bands(); + if (packptr != nullptr && len != 0) + { + inbytes.set((byte *)packptr, len); + } + read_bands(); } void unpacker::check_options() { - if (deflate_hint_or_zero != 0) - { - bool force_deflate_hint = (deflate_hint_or_zero > 0); - if (force_deflate_hint) - default_file_options |= FO_DEFLATE_HINT; - else - default_file_options &= ~FO_DEFLATE_HINT; - // Turn off per-file deflate hint by force. - suppress_file_options |= FO_DEFLATE_HINT; - } - if (modification_time_or_zero != 0) - { - default_file_modtime = modification_time_or_zero; - // Turn off per-file modtime by force. - archive_options &= ~AO_HAVE_FILE_MODTIME; - } + if (deflate_hint_or_zero != 0) + { + bool force_deflate_hint = (deflate_hint_or_zero > 0); + if (force_deflate_hint) + default_file_options |= FO_DEFLATE_HINT; + else + default_file_options &= ~FO_DEFLATE_HINT; + // Turn off per-file deflate hint by force. + suppress_file_options |= FO_DEFLATE_HINT; + } + if (modification_time_or_zero != 0) + { + default_file_modtime = modification_time_or_zero; + // Turn off per-file modtime by force. + archive_options &= ~AO_HAVE_FILE_MODTIME; + } } // classfile writing void unpacker::reset_cur_classfile() { - // set defaults - cur_class_minver = default_class_minver; - cur_class_majver = default_class_majver; - - // reset constant pool state - cp.resetOutputIndexes(); - - // reset fixups - class_fixup_type.empty(); - class_fixup_offset.empty(); - class_fixup_ref.empty(); - requested_ics.empty(); + // set defaults + cur_class_minver = default_class_minver; + cur_class_majver = default_class_majver; + + // reset constant pool state + cp.resetOutputIndexes(); + + // reset fixups + class_fixup_type.empty(); + class_fixup_offset.empty(); + class_fixup_ref.empty(); + requested_ics.empty(); } cpindex *constant_pool::getKQIndex() { - char ch = '?'; - if (u->cur_descr != nullptr) - { - entry *type = u->cur_descr->descrType(); - ch = type->value.b.ptr[0]; - } - byte tag = CONSTANT_Integer; - switch (ch) - { - case 'L': - tag = CONSTANT_String; - break; - case 'I': - tag = CONSTANT_Integer; - break; - case 'J': - tag = CONSTANT_Long; - break; - case 'F': - tag = CONSTANT_Float; - break; - case 'D': - tag = CONSTANT_Double; - break; - case 'B': - case 'S': - case 'C': - case 'Z': - tag = CONSTANT_Integer; - break; - default: - unpack_abort("bad KQ reference"); - break; - } - return getIndex(tag); + char ch = '?'; + if (u->cur_descr != nullptr) + { + entry *type = u->cur_descr->descrType(); + ch = type->value.b.ptr[0]; + } + byte tag = CONSTANT_Integer; + switch (ch) + { + case 'L': + tag = CONSTANT_String; + break; + case 'I': + tag = CONSTANT_Integer; + break; + case 'J': + tag = CONSTANT_Long; + break; + case 'F': + tag = CONSTANT_Float; + break; + case 'D': + tag = CONSTANT_Double; + break; + case 'B': + case 'S': + case 'C': + case 'Z': + tag = CONSTANT_Integer; + break; + default: + unpack_abort("bad KQ reference"); + break; + } + return getIndex(tag); } uint32_t unpacker::to_bci(uint32_t bii) { - uint32_t len = bcimap.length(); - uint32_t *map = (uint32_t *)bcimap.base(); - assert(len > 0); // must be initialized before using to_bci - if (bii < len) - return map[bii]; - // Else it's a fractional or out-of-range BCI. - uint32_t key = bii - len; - for (int i = len;; i--) - { - if (map[i - 1] - (i - 1) <= key) - break; - else - --bii; - } - return bii; + uint32_t len = bcimap.length(); + uint32_t *map = (uint32_t *)bcimap.base(); + assert(len > 0); // must be initialized before using to_bci + if (bii < len) + return map[bii]; + // Else it's a fractional or out-of-range BCI. + uint32_t key = bii - len; + for (int i = len;; i--) + { + if (map[i - 1] - (i - 1) <= key) + break; + else + --bii; + } + return bii; } void unpacker::put_stackmap_type() { - int tag = code_StackMapTable_T.getByte(); - putu1(tag); - switch (tag) - { - case 7: // (7) [RCH] - putref(code_StackMapTable_RC.getRef()); - break; - case 8: // (8) [PH] - putu2(to_bci(code_StackMapTable_P.getInt())); - break; - } + int tag = code_StackMapTable_T.getByte(); + putu1(tag); + switch (tag) + { + case 7: // (7) [RCH] + putref(code_StackMapTable_RC.getRef()); + break; + case 8: // (8) [PH] + putu2(to_bci(code_StackMapTable_P.getInt())); + break; + } } // Functions for writing code. void unpacker::put_label(int curIP, int size) { - code_fixup_type.addByte(size); - code_fixup_offset.add((int)put_empty(size)); - code_fixup_source.add(curIP); + code_fixup_type.addByte(size); + code_fixup_offset.add((int)put_empty(size)); + code_fixup_source.add(curIP); } inline // called exactly once => inline - void + void unpacker::write_bc_ops() { - bcimap.empty(); - code_fixup_type.empty(); - code_fixup_offset.empty(); - code_fixup_source.empty(); - - band *bc_which; - - byte *opptr = bc_codes.curRP(); - // No need for oplimit, since the codes are pre-counted. - - size_t codeBase = wpoffset(); - - bool isAload; // copy-out result - int origBC; - - entry *thisClass = cur_class; - entry *superClass = cur_super; - entry *newClass = nullptr; // class of last _new opcode - - // overwrite any prior index on these bands; it changes w/ current class: - bc_thisfield.setIndex(cp.getFieldIndex(thisClass)); - bc_thismethod.setIndex(cp.getMethodIndex(thisClass)); - if (superClass != nullptr) - { - bc_superfield.setIndex(cp.getFieldIndex(superClass)); - bc_supermethod.setIndex(cp.getMethodIndex(superClass)); - } - - for (int curIP = 0;; curIP++) - { - int curPC = (int)(wpoffset() - codeBase); - bcimap.add(curPC); - ensure_put_space(10); // covers most instrs w/o further bounds check - int bc = *opptr++ & 0xFF; - - putu1_fast(bc); - // Note: See '--wp' below for pseudo-bytecodes like bc_end_marker. - - bool isWide = false; - if (bc == bc_wide) - { - bc = *opptr++ & 0xFF; - putu1_fast(bc); - isWide = true; - } - switch (bc) - { - case bc_end_marker: - --wp; // not really part of the code - assert(opptr <= bc_codes.maxRP()); - bc_codes.curRP() = opptr; // advance over this in bc_codes - goto doneScanningMethod; - case bc_tableswitch: // apc: (df, lo, hi, (hi-lo+1)*(label)) - case bc_lookupswitch: // apc: (df, nc, nc*(case, label)) - { - int caseCount = bc_case_count.getInt(); - while (((wpoffset() - codeBase) % 4) != 0) - putu1_fast(0); - ensure_put_space(30 + caseCount * 8); - put_label(curIP, 4); // int df = bc_label.getInt(); - if (bc == bc_tableswitch) - { - int lo = bc_case_value.getInt(); - int hi = lo + caseCount - 1; - putu4(lo); - putu4(hi); - for (int j = 0; j < caseCount; j++) - { - put_label(curIP, 4); // int lVal = bc_label.getInt(); - // int cVal = lo + j; - } - } - else - { - putu4(caseCount); - for (int j = 0; j < caseCount; j++) - { - int cVal = bc_case_value.getInt(); - putu4(cVal); - put_label(curIP, 4); // int lVal = bc_label.getInt(); - } - } - assert((int)to_bci(curIP) == curPC); - continue; - } - case bc_iinc: - { - int local = bc_local.getInt(); - int delta = (isWide ? bc_short : bc_byte).getInt(); - if (isWide) - { - putu2(local); - putu2(delta); - } - else - { - putu1_fast(local); - putu1_fast(delta); - } - continue; - } - case bc_sipush: - { - int val = bc_short.getInt(); - putu2(val); - continue; - } - case bc_bipush: - case bc_newarray: - { - int val = bc_byte.getByte(); - putu1_fast(val); - continue; - } - case bc_ref_escape: - { - // Note that insnMap has one entry for this. - --wp; // not really part of the code - int size = bc_escrefsize.getInt(); - entry *ref = bc_escref.getRefN(); - switch (size) - { - case 1: - putu1ref(ref); - break; - case 2: - putref(ref); - break; - default: - assert(false); - } - continue; - } - case bc_byte_escape: - { - // Note that insnMap has one entry for all these bytes. - --wp; // not really part of the code - int size = bc_escsize.getInt(); - ensure_put_space(size); - for (int j = 0; j < size; j++) - putu1_fast(bc_escbyte.getByte()); - continue; - } - default: - if (is_invoke_init_op(bc)) - { - origBC = bc_invokespecial; - entry *classRef; - switch (bc - _invokeinit_op) - { - case _invokeinit_self_option: - classRef = thisClass; - break; - case _invokeinit_super_option: - classRef = superClass; - break; - default: - assert(bc == _invokeinit_op + _invokeinit_new_option); - case _invokeinit_new_option: - classRef = newClass; - break; - } - wp[-1] = origBC; // overwrite with origBC - int coding = bc_initref.getInt(); - // Find the nth overloading of in classRef. - entry *ref = nullptr; - cpindex *ix = (classRef == nullptr) ? nullptr : cp.getMethodIndex(classRef); - for (int j = 0, which_init = 0;; j++) - { - ref = (ix == nullptr) ? nullptr : ix->get(j); - if (ref == nullptr) - break; // oops, bad input - assert(ref->tag == CONSTANT_Methodref); - if (ref->memberDescr()->descrName() == cp.sym[constant_pool::s_lt_init_gt]) - { - if (which_init++ == coding) - break; - } - } - putref(ref); - continue; - } - bc_which = ref_band_for_self_op(bc, isAload, origBC); - if (bc_which != nullptr) - { - if (!isAload) - { - wp[-1] = origBC; // overwrite with origBC - } - else - { - wp[-1] = bc_aload_0; // overwrite with _aload_0 - // Note: insnMap keeps the _aload_0 separate. - bcimap.add(++curPC); - ++curIP; - putu1_fast(origBC); - } - entry *ref = bc_which->getRef(); - putref(ref); - continue; - } - if (is_branch_op(bc)) - { - // int lVal = bc_label.getInt(); - if (bc < bc_goto_w) - { - put_label(curIP, 2); // putu2(lVal & 0xFFFF); - } - else - { - assert(bc <= bc_jsr_w); - put_label(curIP, 4); // putu4(lVal); - } - assert((int)to_bci(curIP) == curPC); - continue; - } - bc_which = ref_band_for_op(bc); - if (bc_which != nullptr) - { - entry *ref = bc_which->getRefCommon(bc_which->ix, bc_which->nullOK); - if (ref == nullptr && bc_which == &bc_classref) - { - // Shorthand for class self-references. - ref = thisClass; - } - origBC = bc; - switch (bc) - { - case bc_ildc: - case bc_cldc: - case bc_fldc: - case bc_aldc: - origBC = bc_ldc; - break; - case bc_ildc_w: - case bc_cldc_w: - case bc_fldc_w: - case bc_aldc_w: - origBC = bc_ldc_w; - break; - case bc_lldc2_w: - case bc_dldc2_w: - origBC = bc_ldc2_w; - break; - case bc_new: - newClass = ref; - break; - } - wp[-1] = origBC; // overwrite with origBC - if (origBC == bc_ldc) - { - putu1ref(ref); - } - else - { - putref(ref); - } - if (origBC == bc_multianewarray) - { - // Copy the trailing byte also. - int val = bc_byte.getByte(); - putu1_fast(val); - } - else if (origBC == bc_invokeinterface) - { - int argSize = ref->memberDescr()->descrType()->typeSize(); - putu1_fast(1 + argSize); - putu1_fast(0); - } - continue; - } - if (is_local_slot_op(bc)) - { - int local = bc_local.getInt(); - if (isWide) - { - putu2(local); - if (bc == bc_iinc) - { - int iVal = bc_short.getInt(); - putu2(iVal); - } - } - else - { - putu1_fast(local); - if (bc == bc_iinc) - { - int iVal = bc_byte.getByte(); - putu1_fast(iVal); - } - } - continue; - } - // Random bytecode. Just copy it. - assert(bc < bc_bytecode_limit); - } - } + bcimap.empty(); + code_fixup_type.empty(); + code_fixup_offset.empty(); + code_fixup_source.empty(); + + band *bc_which; + + byte *opptr = bc_codes.curRP(); + // No need for oplimit, since the codes are pre-counted. + + size_t codeBase = wpoffset(); + + bool isAload; // copy-out result + int origBC; + + entry *thisClass = cur_class; + entry *superClass = cur_super; + entry *newClass = nullptr; // class of last _new opcode + + // overwrite any prior index on these bands; it changes w/ current class: + bc_thisfield.setIndex(cp.getFieldIndex(thisClass)); + bc_thismethod.setIndex(cp.getMethodIndex(thisClass)); + if (superClass != nullptr) + { + bc_superfield.setIndex(cp.getFieldIndex(superClass)); + bc_supermethod.setIndex(cp.getMethodIndex(superClass)); + } + + for (int curIP = 0;; curIP++) + { + int curPC = (int)(wpoffset() - codeBase); + bcimap.add(curPC); + ensure_put_space(10); // covers most instrs w/o further bounds check + int bc = *opptr++ & 0xFF; + + putu1_fast(bc); + // Note: See '--wp' below for pseudo-bytecodes like bc_end_marker. + + bool isWide = false; + if (bc == bc_wide) + { + bc = *opptr++ & 0xFF; + putu1_fast(bc); + isWide = true; + } + switch (bc) + { + case bc_end_marker: + --wp; // not really part of the code + assert(opptr <= bc_codes.maxRP()); + bc_codes.curRP() = opptr; // advance over this in bc_codes + goto doneScanningMethod; + case bc_tableswitch: // apc: (df, lo, hi, (hi-lo+1)*(label)) + case bc_lookupswitch: // apc: (df, nc, nc*(case, label)) + { + int caseCount = bc_case_count.getInt(); + while (((wpoffset() - codeBase) % 4) != 0) + putu1_fast(0); + ensure_put_space(30 + caseCount * 8); + put_label(curIP, 4); // int df = bc_label.getInt(); + if (bc == bc_tableswitch) + { + int lo = bc_case_value.getInt(); + int hi = lo + caseCount - 1; + putu4(lo); + putu4(hi); + for (int j = 0; j < caseCount; j++) + { + put_label(curIP, 4); // int lVal = bc_label.getInt(); + // int cVal = lo + j; + } + } + else + { + putu4(caseCount); + for (int j = 0; j < caseCount; j++) + { + int cVal = bc_case_value.getInt(); + putu4(cVal); + put_label(curIP, 4); // int lVal = bc_label.getInt(); + } + } + assert((int)to_bci(curIP) == curPC); + continue; + } + case bc_iinc: + { + int local = bc_local.getInt(); + int delta = (isWide ? bc_short : bc_byte).getInt(); + if (isWide) + { + putu2(local); + putu2(delta); + } + else + { + putu1_fast(local); + putu1_fast(delta); + } + continue; + } + case bc_sipush: + { + int val = bc_short.getInt(); + putu2(val); + continue; + } + case bc_bipush: + case bc_newarray: + { + int val = bc_byte.getByte(); + putu1_fast(val); + continue; + } + case bc_ref_escape: + { + // Note that insnMap has one entry for this. + --wp; // not really part of the code + int size = bc_escrefsize.getInt(); + entry *ref = bc_escref.getRefN(); + switch (size) + { + case 1: + putu1ref(ref); + break; + case 2: + putref(ref); + break; + default: + assert(false); + } + continue; + } + case bc_byte_escape: + { + // Note that insnMap has one entry for all these bytes. + --wp; // not really part of the code + int size = bc_escsize.getInt(); + ensure_put_space(size); + for (int j = 0; j < size; j++) + putu1_fast(bc_escbyte.getByte()); + continue; + } + default: + if (is_invoke_init_op(bc)) + { + origBC = bc_invokespecial; + entry *classRef; + switch (bc - _invokeinit_op) + { + case _invokeinit_self_option: + classRef = thisClass; + break; + case _invokeinit_super_option: + classRef = superClass; + break; + default: + assert(bc == _invokeinit_op + _invokeinit_new_option); + case _invokeinit_new_option: + classRef = newClass; + break; + } + wp[-1] = origBC; // overwrite with origBC + int coding = bc_initref.getInt(); + // Find the nth overloading of in classRef. + entry *ref = nullptr; + cpindex *ix = (classRef == nullptr) ? nullptr : cp.getMethodIndex(classRef); + for (int j = 0, which_init = 0;; j++) + { + ref = (ix == nullptr) ? nullptr : ix->get(j); + if (ref == nullptr) + break; // oops, bad input + assert(ref->tag == CONSTANT_Methodref); + if (ref->memberDescr()->descrName() == cp.sym[constant_pool::s_lt_init_gt]) + { + if (which_init++ == coding) + break; + } + } + putref(ref); + continue; + } + bc_which = ref_band_for_self_op(bc, isAload, origBC); + if (bc_which != nullptr) + { + if (!isAload) + { + wp[-1] = origBC; // overwrite with origBC + } + else + { + wp[-1] = bc_aload_0; // overwrite with _aload_0 + // Note: insnMap keeps the _aload_0 separate. + bcimap.add(++curPC); + ++curIP; + putu1_fast(origBC); + } + entry *ref = bc_which->getRef(); + putref(ref); + continue; + } + if (is_branch_op(bc)) + { + // int lVal = bc_label.getInt(); + if (bc < bc_goto_w) + { + put_label(curIP, 2); // putu2(lVal & 0xFFFF); + } + else + { + assert(bc <= bc_jsr_w); + put_label(curIP, 4); // putu4(lVal); + } + assert((int)to_bci(curIP) == curPC); + continue; + } + bc_which = ref_band_for_op(bc); + if (bc_which != nullptr) + { + entry *ref = bc_which->getRefCommon(bc_which->ix, bc_which->nullOK); + if (ref == nullptr && bc_which == &bc_classref) + { + // Shorthand for class self-references. + ref = thisClass; + } + origBC = bc; + switch (bc) + { + case bc_ildc: + case bc_cldc: + case bc_fldc: + case bc_aldc: + origBC = bc_ldc; + break; + case bc_ildc_w: + case bc_cldc_w: + case bc_fldc_w: + case bc_aldc_w: + origBC = bc_ldc_w; + break; + case bc_lldc2_w: + case bc_dldc2_w: + origBC = bc_ldc2_w; + break; + case bc_new: + newClass = ref; + break; + } + wp[-1] = origBC; // overwrite with origBC + if (origBC == bc_ldc) + { + putu1ref(ref); + } + else + { + putref(ref); + } + if (origBC == bc_multianewarray) + { + // Copy the trailing byte also. + int val = bc_byte.getByte(); + putu1_fast(val); + } + else if (origBC == bc_invokeinterface) + { + int argSize = ref->memberDescr()->descrType()->typeSize(); + putu1_fast(1 + argSize); + putu1_fast(0); + } + continue; + } + if (is_local_slot_op(bc)) + { + int local = bc_local.getInt(); + if (isWide) + { + putu2(local); + if (bc == bc_iinc) + { + int iVal = bc_short.getInt(); + putu2(iVal); + } + } + else + { + putu1_fast(local); + if (bc == bc_iinc) + { + int iVal = bc_byte.getByte(); + putu1_fast(iVal); + } + } + continue; + } + // Random bytecode. Just copy it. + assert(bc < bc_bytecode_limit); + } + } doneScanningMethod: { } - // bcimap.add(curPC); // PC limit is already also in map, from bc_end_marker - - // Armed with a bcimap, we can now fix up all the labels. - for (int i = 0; i < (int)code_fixup_type.size(); i++) - { - int type = code_fixup_type.getByte(i); - byte *bp = wp_at(code_fixup_offset.get(i)); - int curIP = code_fixup_source.get(i); - int destIP = curIP + bc_label.getInt(); - int span = to_bci(destIP) - to_bci(curIP); - switch (type) - { - case 2: - putu2_at(bp, (ushort)span); - break; - case 4: - putu4_at(bp, span); - break; - default: - assert(false); - } - } + // bcimap.add(curPC); // PC limit is already also in map, from bc_end_marker + + // Armed with a bcimap, we can now fix up all the labels. + for (int i = 0; i < (int)code_fixup_type.size(); i++) + { + int type = code_fixup_type.getByte(i); + byte *bp = wp_at(code_fixup_offset.get(i)); + int curIP = code_fixup_source.get(i); + int destIP = curIP + bc_label.getInt(); + int span = to_bci(destIP) - to_bci(curIP); + switch (type) + { + case 2: + putu2_at(bp, (ushort)span); + break; + case 4: + putu4_at(bp, span); + break; + default: + assert(false); + } + } } inline // called exactly once => inline - void + void unpacker::write_code() { - int j; - - int max_stack, max_locals, handler_count, cflags; - get_code_header(max_stack, max_locals, handler_count, cflags); - - if (max_stack < 0) - max_stack = code_max_stack.getInt(); - if (max_locals < 0) - max_locals = code_max_na_locals.getInt(); - if (handler_count < 0) - handler_count = code_handler_count.getInt(); - - int siglen = cur_descr->descrType()->typeSize(); - if ((cur_descr_flags & ACC_STATIC) == 0) - siglen++; - max_locals += siglen; - - putu2(max_stack); - putu2(max_locals); - size_t bcbase = put_empty(4); - - // Write the bytecodes themselves. - write_bc_ops(); - - byte *bcbasewp = wp_at(bcbase); - putu4_at(bcbasewp, (int)(wp - (bcbasewp + 4))); // size of code attr - - putu2(handler_count); - for (j = 0; j < handler_count; j++) - { - int bii = code_handler_start_P.getInt(); - putu2(to_bci(bii)); - bii += code_handler_end_PO.getInt(); - putu2(to_bci(bii)); - bii += code_handler_catch_PO.getInt(); - putu2(to_bci(bii)); - putref(code_handler_class_RCN.getRefN()); - } - - uint64_t indexBits = cflags; - if (cflags < 0) - { - bool haveLongFlags = attr_defs[ATTR_CONTEXT_CODE].haveLongFlags(); - indexBits = code_flags_hi.getLong(code_flags_lo, haveLongFlags); - } - write_attrs(ATTR_CONTEXT_CODE, indexBits); + int j; + + int max_stack, max_locals, handler_count, cflags; + get_code_header(max_stack, max_locals, handler_count, cflags); + + if (max_stack < 0) + max_stack = code_max_stack.getInt(); + if (max_locals < 0) + max_locals = code_max_na_locals.getInt(); + if (handler_count < 0) + handler_count = code_handler_count.getInt(); + + int siglen = cur_descr->descrType()->typeSize(); + if ((cur_descr_flags & ACC_STATIC) == 0) + siglen++; + max_locals += siglen; + + putu2(max_stack); + putu2(max_locals); + size_t bcbase = put_empty(4); + + // Write the bytecodes themselves. + write_bc_ops(); + + byte *bcbasewp = wp_at(bcbase); + putu4_at(bcbasewp, (int)(wp - (bcbasewp + 4))); // size of code attr + + putu2(handler_count); + for (j = 0; j < handler_count; j++) + { + int bii = code_handler_start_P.getInt(); + putu2(to_bci(bii)); + bii += code_handler_end_PO.getInt(); + putu2(to_bci(bii)); + bii += code_handler_catch_PO.getInt(); + putu2(to_bci(bii)); + putref(code_handler_class_RCN.getRefN()); + } + + uint64_t indexBits = cflags; + if (cflags < 0) + { + bool haveLongFlags = attr_defs[ATTR_CONTEXT_CODE].haveLongFlags(); + indexBits = code_flags_hi.getLong(code_flags_lo, haveLongFlags); + } + write_attrs(ATTR_CONTEXT_CODE, indexBits); } int unpacker::write_attrs(int attrc, uint64_t indexBits) { - if (indexBits == 0) - { - // Quick short-circuit. - putu2(0); - return 0; - } - - attr_definitions &ad = attr_defs[attrc]; - - int i, j, j2, idx, count; - - int oiCount = 0; - if (ad.isPredefined(X_ATTR_OVERFLOW) && (indexBits & ((uint64_t)1 << X_ATTR_OVERFLOW)) != 0) - { - indexBits -= ((uint64_t)1 << X_ATTR_OVERFLOW); - oiCount = ad.xxx_attr_count().getInt(); - } - - int bitIndexes[X_ATTR_LIMIT_FLAGS_HI]; - int biCount = 0; - - // Fill bitIndexes with index bits, in order. - for (idx = 0; indexBits != 0; idx++, indexBits >>= 1) - { - if ((indexBits & 1) != 0) - bitIndexes[biCount++] = idx; - } - assert(biCount <= (int)lengthof(bitIndexes)); - - // Write a provisional attribute count, perhaps to be corrected later. - int naOffset = (int)wpoffset(); - int na0 = biCount + oiCount; - putu2(na0); - - int na = 0; - for (i = 0; i < na0; i++) - { - if (i < biCount) - idx = bitIndexes[i]; - else - idx = ad.xxx_attr_indexes().getInt(); - assert(ad.isIndex(idx)); - entry *aname = nullptr; - entry *ref; // scratch - size_t abase = put_empty(2 + 4); - if (idx < (int)ad.flag_limit && ad.isPredefined(idx)) - { - // Switch on the attrc and idx simultaneously. - switch (ADH_BYTE(attrc, idx)) - { - - case ADH_BYTE(ATTR_CONTEXT_CLASS, X_ATTR_OVERFLOW) : - case ADH_BYTE(ATTR_CONTEXT_FIELD, X_ATTR_OVERFLOW) : - case ADH_BYTE(ATTR_CONTEXT_METHOD, X_ATTR_OVERFLOW) : - case ADH_BYTE(ATTR_CONTEXT_CODE, X_ATTR_OVERFLOW) : - // no attribute at all, so back up on this one - wp = wp_at(abase); - continue; - - case ADH_BYTE(ATTR_CONTEXT_CLASS, CLASS_ATTR_ClassFile_version) : - cur_class_minver = class_ClassFile_version_minor_H.getInt(); - cur_class_majver = class_ClassFile_version_major_H.getInt(); - // back up; not a real attribute - wp = wp_at(abase); - continue; - - case ADH_BYTE(ATTR_CONTEXT_CLASS, CLASS_ATTR_InnerClasses) : - // note the existence of this attr, but save for later - if (cur_class_has_local_ics) - unpack_abort("too many InnerClasses attrs"); - cur_class_has_local_ics = true; - wp = wp_at(abase); - continue; - - case ADH_BYTE(ATTR_CONTEXT_CLASS, CLASS_ATTR_SourceFile) : - aname = cp.sym[constant_pool::s_SourceFile]; - ref = class_SourceFile_RUN.getRefN(); - if (ref == nullptr) - { - bytes &n = cur_class->ref(0)->value.b; - // parse n = (/)*?($)* - int pkglen = lastIndexOf(SLASH_MIN, SLASH_MAX, n, (int)n.len) + 1; - bytes prefix = n.slice(pkglen, n.len); - for (;;) - { - // Work backwards, finding all '$', '#', etc. - int dollar = - lastIndexOf(DOLLAR_MIN, DOLLAR_MAX, prefix, (int)prefix.len); - if (dollar < 0) - break; - prefix = prefix.slice(0, dollar); - } - const char *suffix = ".java"; - int len = (int)(prefix.len + strlen(suffix)); - bytes name; - name.set(T_NEW(byte, add_size(len, 1)), len); - name.strcat(prefix).strcat(suffix); - ref = cp.ensureUtf8(name); - } - putref(ref); - break; - - case ADH_BYTE(ATTR_CONTEXT_CLASS, CLASS_ATTR_EnclosingMethod) : - aname = cp.sym[constant_pool::s_EnclosingMethod]; - putref(class_EnclosingMethod_RC.getRefN()); - putref(class_EnclosingMethod_RDN.getRefN()); - break; - - case ADH_BYTE(ATTR_CONTEXT_FIELD, FIELD_ATTR_ConstantValue) : - aname = cp.sym[constant_pool::s_ConstantValue]; - putref(field_ConstantValue_KQ.getRefUsing(cp.getKQIndex())); - break; - - case ADH_BYTE(ATTR_CONTEXT_METHOD, METHOD_ATTR_Code) : - aname = cp.sym[constant_pool::s_Code]; - write_code(); - break; - - case ADH_BYTE(ATTR_CONTEXT_METHOD, METHOD_ATTR_Exceptions) : - aname = cp.sym[constant_pool::s_Exceptions]; - putu2(count = method_Exceptions_N.getInt()); - for (j = 0; j < count; j++) - { - putref(method_Exceptions_RC.getRefN()); - } - break; - - case ADH_BYTE(ATTR_CONTEXT_CODE, CODE_ATTR_StackMapTable) : - aname = cp.sym[constant_pool::s_StackMapTable]; - // (keep this code aligned with its brother in unpacker::read_attrs) - putu2(count = code_StackMapTable_N.getInt()); - for (j = 0; j < count; j++) - { - int tag = code_StackMapTable_frame_T.getByte(); - putu1(tag); - if (tag <= 127) - { - // (64-127) [(2)] - if (tag >= 64) - put_stackmap_type(); - } - else if (tag <= 251) - { - // (247) [(1)(2)] - // (248-251) [(1)] - if (tag >= 247) - putu2(code_StackMapTable_offset.getInt()); - if (tag == 247) - put_stackmap_type(); - } - else if (tag <= 254) - { - // (252) [(1)(2)] - // (253) [(1)(2)(2)] - // (254) [(1)(2)(2)(2)] - putu2(code_StackMapTable_offset.getInt()); - for (int k = (tag - 251); k > 0; k--) - { - put_stackmap_type(); - } - } - else - { - // (255) [(1)NH[(2)]NH[(2)]] - putu2(code_StackMapTable_offset.getInt()); - putu2(j2 = code_StackMapTable_local_N.getInt()); - while (j2-- > 0) - put_stackmap_type(); - putu2(j2 = code_StackMapTable_stack_N.getInt()); - while (j2-- > 0) - put_stackmap_type(); - } - } - break; - - case ADH_BYTE(ATTR_CONTEXT_CODE, CODE_ATTR_LineNumberTable) : - aname = cp.sym[constant_pool::s_LineNumberTable]; - putu2(count = code_LineNumberTable_N.getInt()); - for (j = 0; j < count; j++) - { - putu2(to_bci(code_LineNumberTable_bci_P.getInt())); - putu2(code_LineNumberTable_line.getInt()); - } - break; - - case ADH_BYTE(ATTR_CONTEXT_CODE, CODE_ATTR_LocalVariableTable) : - aname = cp.sym[constant_pool::s_LocalVariableTable]; - putu2(count = code_LocalVariableTable_N.getInt()); - for (j = 0; j < count; j++) - { - int bii = code_LocalVariableTable_bci_P.getInt(); - int bci = to_bci(bii); - putu2(bci); - bii += code_LocalVariableTable_span_O.getInt(); - putu2(to_bci(bii) - bci); - putref(code_LocalVariableTable_name_RU.getRefN()); - putref(code_LocalVariableTable_type_RS.getRefN()); - putu2(code_LocalVariableTable_slot.getInt()); - } - break; - - case ADH_BYTE(ATTR_CONTEXT_CODE, CODE_ATTR_LocalVariableTypeTable) : - aname = cp.sym[constant_pool::s_LocalVariableTypeTable]; - putu2(count = code_LocalVariableTypeTable_N.getInt()); - for (j = 0; j < count; j++) - { - int bii = code_LocalVariableTypeTable_bci_P.getInt(); - int bci = to_bci(bii); - putu2(bci); - bii += code_LocalVariableTypeTable_span_O.getInt(); - putu2(to_bci(bii) - bci); - putref(code_LocalVariableTypeTable_name_RU.getRefN()); - putref(code_LocalVariableTypeTable_type_RS.getRefN()); - putu2(code_LocalVariableTypeTable_slot.getInt()); - } - break; - - case ADH_BYTE(ATTR_CONTEXT_CLASS, X_ATTR_Signature) : - aname = cp.sym[constant_pool::s_Signature]; - putref(class_Signature_RS.getRefN()); - break; - - case ADH_BYTE(ATTR_CONTEXT_FIELD, X_ATTR_Signature) : - aname = cp.sym[constant_pool::s_Signature]; - putref(field_Signature_RS.getRefN()); - break; - - case ADH_BYTE(ATTR_CONTEXT_METHOD, X_ATTR_Signature) : - aname = cp.sym[constant_pool::s_Signature]; - putref(method_Signature_RS.getRefN()); - break; - - case ADH_BYTE(ATTR_CONTEXT_CLASS, X_ATTR_Deprecated) : - case ADH_BYTE(ATTR_CONTEXT_FIELD, X_ATTR_Deprecated) : - case ADH_BYTE(ATTR_CONTEXT_METHOD, X_ATTR_Deprecated) : - aname = cp.sym[constant_pool::s_Deprecated]; - // no data - break; - } - } - - if (aname == nullptr) - { - // Unparse a compressor-defined attribute. - layout_definition *lo = ad.getLayout(idx); - if (lo == nullptr) - { - unpack_abort("bad layout index"); - break; - } - assert((int)lo->idx == idx); - aname = lo->nameEntry; - if (aname == nullptr) - { - bytes nameb; - nameb.set(lo->name); - aname = cp.ensureUtf8(nameb); - // Cache the name entry for next time. - lo->nameEntry = aname; - } - // Execute all the layout elements. - band **bands = lo->bands(); - if (lo->hasCallables()) - { - band &cble = *bands[0]; - assert(cble.le_kind == EK_CBLE); - bands = cble.le_body; - } - putlayout(bands); - } - - if (aname == nullptr) - unpack_abort("bad attribute index"); - - byte *wp1 = wp; - wp = wp_at(abase); - - // DTRT if this attr is on the strip-list. - // (Note that we emptied the data out of the band first.) - if (ad.strip_names.contains(aname)) - { - continue; - } - - // patch the name and length - putref(aname); - putu4((int)(wp1 - (wp + 4))); // put the attr size - wp = wp1; - na++; // count the attrs actually written - } - - if (na != na0) - // Refresh changed count. - putu2_at(wp_at(naOffset), na); - return na; + if (indexBits == 0) + { + // Quick short-circuit. + putu2(0); + return 0; + } + + attr_definitions &ad = attr_defs[attrc]; + + int i, j, j2, idx, count; + + int oiCount = 0; + if (ad.isPredefined(X_ATTR_OVERFLOW) && (indexBits & ((uint64_t)1 << X_ATTR_OVERFLOW)) != 0) + { + indexBits -= ((uint64_t)1 << X_ATTR_OVERFLOW); + oiCount = ad.xxx_attr_count().getInt(); + } + + int bitIndexes[X_ATTR_LIMIT_FLAGS_HI]; + int biCount = 0; + + // Fill bitIndexes with index bits, in order. + for (idx = 0; indexBits != 0; idx++, indexBits >>= 1) + { + if ((indexBits & 1) != 0) + bitIndexes[biCount++] = idx; + } + assert(biCount <= (int)lengthof(bitIndexes)); + + // Write a provisional attribute count, perhaps to be corrected later. + int naOffset = (int)wpoffset(); + int na0 = biCount + oiCount; + putu2(na0); + + int na = 0; + for (i = 0; i < na0; i++) + { + if (i < biCount) + idx = bitIndexes[i]; + else + idx = ad.xxx_attr_indexes().getInt(); + assert(ad.isIndex(idx)); + entry *aname = nullptr; + entry *ref; // scratch + size_t abase = put_empty(2 + 4); + if (idx < (int)ad.flag_limit && ad.isPredefined(idx)) + { + // Switch on the attrc and idx simultaneously. + switch (ADH_BYTE(attrc, idx)) + { + + case ADH_BYTE(ATTR_CONTEXT_CLASS, X_ATTR_OVERFLOW) : + case ADH_BYTE(ATTR_CONTEXT_FIELD, X_ATTR_OVERFLOW) : + case ADH_BYTE(ATTR_CONTEXT_METHOD, X_ATTR_OVERFLOW) : + case ADH_BYTE(ATTR_CONTEXT_CODE, X_ATTR_OVERFLOW) : + // no attribute at all, so back up on this one + wp = wp_at(abase); + continue; + + case ADH_BYTE(ATTR_CONTEXT_CLASS, CLASS_ATTR_ClassFile_version) : + cur_class_minver = class_ClassFile_version_minor_H.getInt(); + cur_class_majver = class_ClassFile_version_major_H.getInt(); + // back up; not a real attribute + wp = wp_at(abase); + continue; + + case ADH_BYTE(ATTR_CONTEXT_CLASS, CLASS_ATTR_InnerClasses) : + // note the existence of this attr, but save for later + if (cur_class_has_local_ics) + unpack_abort("too many InnerClasses attrs"); + cur_class_has_local_ics = true; + wp = wp_at(abase); + continue; + + case ADH_BYTE(ATTR_CONTEXT_CLASS, CLASS_ATTR_SourceFile) : + aname = cp.sym[constant_pool::s_SourceFile]; + ref = class_SourceFile_RUN.getRefN(); + if (ref == nullptr) + { + bytes &n = cur_class->ref(0)->value.b; + // parse n = (/)*?($)* + int pkglen = lastIndexOf(SLASH_MIN, SLASH_MAX, n, (int)n.len) + 1; + bytes prefix = n.slice(pkglen, n.len); + for (;;) + { + // Work backwards, finding all '$', '#', etc. + int dollar = + lastIndexOf(DOLLAR_MIN, DOLLAR_MAX, prefix, (int)prefix.len); + if (dollar < 0) + break; + prefix = prefix.slice(0, dollar); + } + const char *suffix = ".java"; + int len = (int)(prefix.len + strlen(suffix)); + bytes name; + name.set(T_NEW(byte, add_size(len, 1)), len); + name.strcat(prefix).strcat(suffix); + ref = cp.ensureUtf8(name); + } + putref(ref); + break; + + case ADH_BYTE(ATTR_CONTEXT_CLASS, CLASS_ATTR_EnclosingMethod) : + aname = cp.sym[constant_pool::s_EnclosingMethod]; + putref(class_EnclosingMethod_RC.getRefN()); + putref(class_EnclosingMethod_RDN.getRefN()); + break; + + case ADH_BYTE(ATTR_CONTEXT_FIELD, FIELD_ATTR_ConstantValue) : + aname = cp.sym[constant_pool::s_ConstantValue]; + putref(field_ConstantValue_KQ.getRefUsing(cp.getKQIndex())); + break; + + case ADH_BYTE(ATTR_CONTEXT_METHOD, METHOD_ATTR_Code) : + aname = cp.sym[constant_pool::s_Code]; + write_code(); + break; + + case ADH_BYTE(ATTR_CONTEXT_METHOD, METHOD_ATTR_Exceptions) : + aname = cp.sym[constant_pool::s_Exceptions]; + putu2(count = method_Exceptions_N.getInt()); + for (j = 0; j < count; j++) + { + putref(method_Exceptions_RC.getRefN()); + } + break; + + case ADH_BYTE(ATTR_CONTEXT_CODE, CODE_ATTR_StackMapTable) : + aname = cp.sym[constant_pool::s_StackMapTable]; + // (keep this code aligned with its brother in unpacker::read_attrs) + putu2(count = code_StackMapTable_N.getInt()); + for (j = 0; j < count; j++) + { + int tag = code_StackMapTable_frame_T.getByte(); + putu1(tag); + if (tag <= 127) + { + // (64-127) [(2)] + if (tag >= 64) + put_stackmap_type(); + } + else if (tag <= 251) + { + // (247) [(1)(2)] + // (248-251) [(1)] + if (tag >= 247) + putu2(code_StackMapTable_offset.getInt()); + if (tag == 247) + put_stackmap_type(); + } + else if (tag <= 254) + { + // (252) [(1)(2)] + // (253) [(1)(2)(2)] + // (254) [(1)(2)(2)(2)] + putu2(code_StackMapTable_offset.getInt()); + for (int k = (tag - 251); k > 0; k--) + { + put_stackmap_type(); + } + } + else + { + // (255) [(1)NH[(2)]NH[(2)]] + putu2(code_StackMapTable_offset.getInt()); + putu2(j2 = code_StackMapTable_local_N.getInt()); + while (j2-- > 0) + put_stackmap_type(); + putu2(j2 = code_StackMapTable_stack_N.getInt()); + while (j2-- > 0) + put_stackmap_type(); + } + } + break; + + case ADH_BYTE(ATTR_CONTEXT_CODE, CODE_ATTR_LineNumberTable) : + aname = cp.sym[constant_pool::s_LineNumberTable]; + putu2(count = code_LineNumberTable_N.getInt()); + for (j = 0; j < count; j++) + { + putu2(to_bci(code_LineNumberTable_bci_P.getInt())); + putu2(code_LineNumberTable_line.getInt()); + } + break; + + case ADH_BYTE(ATTR_CONTEXT_CODE, CODE_ATTR_LocalVariableTable) : + aname = cp.sym[constant_pool::s_LocalVariableTable]; + putu2(count = code_LocalVariableTable_N.getInt()); + for (j = 0; j < count; j++) + { + int bii = code_LocalVariableTable_bci_P.getInt(); + int bci = to_bci(bii); + putu2(bci); + bii += code_LocalVariableTable_span_O.getInt(); + putu2(to_bci(bii) - bci); + putref(code_LocalVariableTable_name_RU.getRefN()); + putref(code_LocalVariableTable_type_RS.getRefN()); + putu2(code_LocalVariableTable_slot.getInt()); + } + break; + + case ADH_BYTE(ATTR_CONTEXT_CODE, CODE_ATTR_LocalVariableTypeTable) : + aname = cp.sym[constant_pool::s_LocalVariableTypeTable]; + putu2(count = code_LocalVariableTypeTable_N.getInt()); + for (j = 0; j < count; j++) + { + int bii = code_LocalVariableTypeTable_bci_P.getInt(); + int bci = to_bci(bii); + putu2(bci); + bii += code_LocalVariableTypeTable_span_O.getInt(); + putu2(to_bci(bii) - bci); + putref(code_LocalVariableTypeTable_name_RU.getRefN()); + putref(code_LocalVariableTypeTable_type_RS.getRefN()); + putu2(code_LocalVariableTypeTable_slot.getInt()); + } + break; + + case ADH_BYTE(ATTR_CONTEXT_CLASS, X_ATTR_Signature) : + aname = cp.sym[constant_pool::s_Signature]; + putref(class_Signature_RS.getRefN()); + break; + + case ADH_BYTE(ATTR_CONTEXT_FIELD, X_ATTR_Signature) : + aname = cp.sym[constant_pool::s_Signature]; + putref(field_Signature_RS.getRefN()); + break; + + case ADH_BYTE(ATTR_CONTEXT_METHOD, X_ATTR_Signature) : + aname = cp.sym[constant_pool::s_Signature]; + putref(method_Signature_RS.getRefN()); + break; + + case ADH_BYTE(ATTR_CONTEXT_CLASS, X_ATTR_Deprecated) : + case ADH_BYTE(ATTR_CONTEXT_FIELD, X_ATTR_Deprecated) : + case ADH_BYTE(ATTR_CONTEXT_METHOD, X_ATTR_Deprecated) : + aname = cp.sym[constant_pool::s_Deprecated]; + // no data + break; + } + } + + if (aname == nullptr) + { + // Unparse a compressor-defined attribute. + layout_definition *lo = ad.getLayout(idx); + if (lo == nullptr) + { + unpack_abort("bad layout index"); + break; + } + assert((int)lo->idx == idx); + aname = lo->nameEntry; + if (aname == nullptr) + { + bytes nameb; + nameb.set(lo->name); + aname = cp.ensureUtf8(nameb); + // Cache the name entry for next time. + lo->nameEntry = aname; + } + // Execute all the layout elements. + band **bands = lo->bands(); + if (lo->hasCallables()) + { + band &cble = *bands[0]; + assert(cble.le_kind == EK_CBLE); + bands = cble.le_body; + } + putlayout(bands); + } + + if (aname == nullptr) + unpack_abort("bad attribute index"); + + byte *wp1 = wp; + wp = wp_at(abase); + + // DTRT if this attr is on the strip-list. + // (Note that we emptied the data out of the band first.) + if (ad.strip_names.contains(aname)) + { + continue; + } + + // patch the name and length + putref(aname); + putu4((int)(wp1 - (wp + 4))); // put the attr size + wp = wp1; + na++; // count the attrs actually written + } + + if (na != na0) + // Refresh changed count. + putu2_at(wp_at(naOffset), na); + return na; } void unpacker::write_members(int num, int attrc) { - attr_definitions &ad = attr_defs[attrc]; - band &member_flags_hi = ad.xxx_flags_hi(); - band &member_flags_lo = ad.xxx_flags_lo(); - band &member_descr = (&member_flags_hi)[e_field_descr - e_field_flags_hi]; - bool haveLongFlags = ad.haveLongFlags(); - - putu2(num); - uint64_t indexMask = attr_defs[attrc].flagIndexMask(); - for (int i = 0; i < num; i++) - { - uint64_t mflags = member_flags_hi.getLong(member_flags_lo, haveLongFlags); - entry *mdescr = member_descr.getRef(); - cur_descr = mdescr; - putu2(cur_descr_flags = (ushort)(mflags & ~indexMask)); - putref(mdescr->descrName()); - putref(mdescr->descrType()); - write_attrs(attrc, (mflags & indexMask)); - } - cur_descr = nullptr; + attr_definitions &ad = attr_defs[attrc]; + band &member_flags_hi = ad.xxx_flags_hi(); + band &member_flags_lo = ad.xxx_flags_lo(); + band &member_descr = (&member_flags_hi)[e_field_descr - e_field_flags_hi]; + bool haveLongFlags = ad.haveLongFlags(); + + putu2(num); + uint64_t indexMask = attr_defs[attrc].flagIndexMask(); + for (int i = 0; i < num; i++) + { + uint64_t mflags = member_flags_hi.getLong(member_flags_lo, haveLongFlags); + entry *mdescr = member_descr.getRef(); + cur_descr = mdescr; + putu2(cur_descr_flags = (ushort)(mflags & ~indexMask)); + putref(mdescr->descrName()); + putref(mdescr->descrType()); + write_attrs(attrc, (mflags & indexMask)); + } + cur_descr = nullptr; } extern "C" int raw_address_cmp(const void *p1p, const void *p2p) { - void *p1 = *(void **)p1p; - void *p2 = *(void **)p2p; - return (p1 > p2) ? 1 : (p1 < p2) ? -1 : 0; + void *p1 = *(void **)p1p; + void *p2 = *(void **)p2p; + return (p1 > p2) ? 1 : (p1 < p2) ? -1 : 0; } void unpacker::write_classfile_tail() { - cur_classfile_tail.empty(); - set_output(&cur_classfile_tail); + cur_classfile_tail.empty(); + set_output(&cur_classfile_tail); - int i, num; + int i, num; - attr_definitions &ad = attr_defs[ATTR_CONTEXT_CLASS]; + attr_definitions &ad = attr_defs[ATTR_CONTEXT_CLASS]; - bool haveLongFlags = ad.haveLongFlags(); - uint64_t kflags = class_flags_hi.getLong(class_flags_lo, haveLongFlags); - uint64_t indexMask = ad.flagIndexMask(); + bool haveLongFlags = ad.haveLongFlags(); + uint64_t kflags = class_flags_hi.getLong(class_flags_lo, haveLongFlags); + uint64_t indexMask = ad.flagIndexMask(); - cur_class = class_this.getRef(); - cur_super = class_super.getRef(); + cur_class = class_this.getRef(); + cur_super = class_super.getRef(); - if (cur_super == cur_class) - cur_super = nullptr; - // special representation for java/lang/Object + if (cur_super == cur_class) + cur_super = nullptr; + // special representation for java/lang/Object - putu2((ushort)(kflags & ~indexMask)); - putref(cur_class); - putref(cur_super); + putu2((ushort)(kflags & ~indexMask)); + putref(cur_class); + putref(cur_super); - putu2(num = class_interface_count.getInt()); - for (i = 0; i < num; i++) - { - putref(class_interface.getRef()); - } + putu2(num = class_interface_count.getInt()); + for (i = 0; i < num; i++) + { + putref(class_interface.getRef()); + } - write_members(class_field_count.getInt(), ATTR_CONTEXT_FIELD); - write_members(class_method_count.getInt(), ATTR_CONTEXT_METHOD); + write_members(class_field_count.getInt(), ATTR_CONTEXT_FIELD); + write_members(class_method_count.getInt(), ATTR_CONTEXT_METHOD); - cur_class_has_local_ics = false; // may be set true by write_attrs + cur_class_has_local_ics = false; // may be set true by write_attrs - int naOffset = (int)wpoffset(); - int na = write_attrs(ATTR_CONTEXT_CLASS, (kflags & indexMask)); + int naOffset = (int)wpoffset(); + int na = write_attrs(ATTR_CONTEXT_CLASS, (kflags & indexMask)); // at the very last, choose which inner classes (if any) pertain to k: #ifdef ASSERT - for (i = 0; i < ic_count; i++) - { - assert(!ics[i].requested); - } + for (i = 0; i < ic_count; i++) + { + assert(!ics[i].requested); + } #endif - // First, consult the global table and the local constant pool, - // and decide on the globally implied inner classes. - // (Note that we read the cpool's outputIndex fields, but we - // do not yet write them, since the local IC attribute might - // reverse a global decision to declare an IC.) - assert(requested_ics.length() == 0); // must start out empty - // Always include all members of the current class. - for (inner_class *child = cp.getFirstChildIC(cur_class); child != nullptr; - child = cp.getNextChildIC(child)) - { - child->requested = true; - requested_ics.add(child); - } - // And, for each inner class mentioned in the constant pool, - // include it and all its outers. - int noes = cp.outputEntries.length(); - entry **oes = (entry **)cp.outputEntries.base(); - for (i = 0; i < noes; i++) - { - entry &e = *oes[i]; - if (e.tag != CONSTANT_Class) - continue; // wrong sort - for (inner_class *ic = cp.getIC(&e); ic != nullptr; ic = cp.getIC(ic->outer)) - { - if (ic->requested) - break; // already processed - ic->requested = true; - requested_ics.add(ic); - } - } - int local_ics = requested_ics.length(); - // Second, consult a local attribute (if any) and adjust the global set. - inner_class *extra_ics = nullptr; - int num_extra_ics = 0; - if (cur_class_has_local_ics) - { - // adjust the set of ICs by symmetric set difference w/ the locals - num_extra_ics = class_InnerClasses_N.getInt(); - if (num_extra_ics == 0) - { - // Explicit zero count has an irregular meaning: It deletes the attr. - local_ics = 0; // (short-circuit all tests of requested bits) - } - else - { - extra_ics = T_NEW(inner_class, num_extra_ics); - // Note: extra_ics will be freed up by next call to get_next_file(). - } - } - for (i = 0; i < num_extra_ics; i++) - { - inner_class &extra_ic = extra_ics[i]; - extra_ic.inner = class_InnerClasses_RC.getRef(); - // Find the corresponding equivalent global IC: - inner_class *global_ic = cp.getIC(extra_ic.inner); - int flags = class_InnerClasses_F.getInt(); - if (flags == 0) - { - // The extra IC is simply a copy of a global IC. - if (global_ic == nullptr) - { - unpack_abort("bad reference to inner class"); - break; - } - extra_ic = (*global_ic); // fill in rest of fields - } - else - { - flags &= ~ACC_IC_LONG_FORM; // clear high bit if set to get clean zero - extra_ic.flags = flags; - extra_ic.outer = class_InnerClasses_outer_RCN.getRefN(); - extra_ic.name = class_InnerClasses_name_RUN.getRefN(); - // Detect if this is an exact copy of the global tuple. - if (global_ic != nullptr) - { - if (global_ic->flags != extra_ic.flags || global_ic->outer != extra_ic.outer || - global_ic->name != extra_ic.name) - { - global_ic = nullptr; // not really the same, so break the link - } - } - } - if (global_ic != nullptr && global_ic->requested) - { - // This local repetition reverses the globally implied request. - global_ic->requested = false; - extra_ic.requested = false; - local_ics -= 1; - } - else - { - // The global either does not exist, or is not yet requested. - extra_ic.requested = true; - local_ics += 1; - } - } - // Finally, if there are any that survived, put them into an attribute. - // (Note that a zero-count attribute is always deleted.) - // The putref calls below will tell the constant pool to add any - // necessary local CP references to support the InnerClasses attribute. - // This step must be the last round of additions to the local CP. - if (local_ics > 0) - { - // append the new attribute: - putref(cp.sym[constant_pool::s_InnerClasses]); - putu4(2 + 2 * 4 * local_ics); - putu2(local_ics); - PTRLIST_QSORT(requested_ics, raw_address_cmp); - int num_global_ics = requested_ics.length(); - for (i = -num_global_ics; i < num_extra_ics; i++) - { - inner_class *ic; - if (i < 0) - ic = (inner_class *)requested_ics.get(num_global_ics + i); - else - ic = &extra_ics[i]; - if (ic->requested) - { - putref(ic->inner); - putref(ic->outer); - putref(ic->name); - putu2(ic->flags); - } - } - putu2_at(wp_at(naOffset), ++na); // increment class attr count - } - - // Tidy up global 'requested' bits: - for (i = requested_ics.length(); --i >= 0;) - { - inner_class *ic = (inner_class *)requested_ics.get(i); - ic->requested = false; - } - requested_ics.empty(); - - close_output(); - - // rewrite CP references in the tail - cp.computeOutputIndexes(); - int nextref = 0; - for (i = 0; i < (int)class_fixup_type.size(); i++) - { - int type = class_fixup_type.getByte(i); - byte *fixp = wp_at(class_fixup_offset.get(i)); - entry *e = (entry *)class_fixup_ref.get(nextref++); - int idx = e->getOutputIndex(); - switch (type) - { - case 1: - putu1_at(fixp, idx); - break; - case 2: - putu2_at(fixp, idx); - break; - default: - assert(false); // should not reach here - } - } + // First, consult the global table and the local constant pool, + // and decide on the globally implied inner classes. + // (Note that we read the cpool's outputIndex fields, but we + // do not yet write them, since the local IC attribute might + // reverse a global decision to declare an IC.) + assert(requested_ics.length() == 0); // must start out empty + // Always include all members of the current class. + for (inner_class *child = cp.getFirstChildIC(cur_class); child != nullptr; + child = cp.getNextChildIC(child)) + { + child->requested = true; + requested_ics.add(child); + } + // And, for each inner class mentioned in the constant pool, + // include it and all its outers. + int noes = cp.outputEntries.length(); + entry **oes = (entry **)cp.outputEntries.base(); + for (i = 0; i < noes; i++) + { + entry &e = *oes[i]; + if (e.tag != CONSTANT_Class) + continue; // wrong sort + for (inner_class *ic = cp.getIC(&e); ic != nullptr; ic = cp.getIC(ic->outer)) + { + if (ic->requested) + break; // already processed + ic->requested = true; + requested_ics.add(ic); + } + } + int local_ics = requested_ics.length(); + // Second, consult a local attribute (if any) and adjust the global set. + inner_class *extra_ics = nullptr; + int num_extra_ics = 0; + if (cur_class_has_local_ics) + { + // adjust the set of ICs by symmetric set difference w/ the locals + num_extra_ics = class_InnerClasses_N.getInt(); + if (num_extra_ics == 0) + { + // Explicit zero count has an irregular meaning: It deletes the attr. + local_ics = 0; // (short-circuit all tests of requested bits) + } + else + { + extra_ics = T_NEW(inner_class, num_extra_ics); + // Note: extra_ics will be freed up by next call to get_next_file(). + } + } + for (i = 0; i < num_extra_ics; i++) + { + inner_class &extra_ic = extra_ics[i]; + extra_ic.inner = class_InnerClasses_RC.getRef(); + // Find the corresponding equivalent global IC: + inner_class *global_ic = cp.getIC(extra_ic.inner); + int flags = class_InnerClasses_F.getInt(); + if (flags == 0) + { + // The extra IC is simply a copy of a global IC. + if (global_ic == nullptr) + { + unpack_abort("bad reference to inner class"); + break; + } + extra_ic = (*global_ic); // fill in rest of fields + } + else + { + flags &= ~ACC_IC_LONG_FORM; // clear high bit if set to get clean zero + extra_ic.flags = flags; + extra_ic.outer = class_InnerClasses_outer_RCN.getRefN(); + extra_ic.name = class_InnerClasses_name_RUN.getRefN(); + // Detect if this is an exact copy of the global tuple. + if (global_ic != nullptr) + { + if (global_ic->flags != extra_ic.flags || global_ic->outer != extra_ic.outer || + global_ic->name != extra_ic.name) + { + global_ic = nullptr; // not really the same, so break the link + } + } + } + if (global_ic != nullptr && global_ic->requested) + { + // This local repetition reverses the globally implied request. + global_ic->requested = false; + extra_ic.requested = false; + local_ics -= 1; + } + else + { + // The global either does not exist, or is not yet requested. + extra_ic.requested = true; + local_ics += 1; + } + } + // Finally, if there are any that survived, put them into an attribute. + // (Note that a zero-count attribute is always deleted.) + // The putref calls below will tell the constant pool to add any + // necessary local CP references to support the InnerClasses attribute. + // This step must be the last round of additions to the local CP. + if (local_ics > 0) + { + // append the new attribute: + putref(cp.sym[constant_pool::s_InnerClasses]); + putu4(2 + 2 * 4 * local_ics); + putu2(local_ics); + PTRLIST_QSORT(requested_ics, raw_address_cmp); + int num_global_ics = requested_ics.length(); + for (i = -num_global_ics; i < num_extra_ics; i++) + { + inner_class *ic; + if (i < 0) + ic = (inner_class *)requested_ics.get(num_global_ics + i); + else + ic = &extra_ics[i]; + if (ic->requested) + { + putref(ic->inner); + putref(ic->outer); + putref(ic->name); + putu2(ic->flags); + } + } + putu2_at(wp_at(naOffset), ++na); // increment class attr count + } + + // Tidy up global 'requested' bits: + for (i = requested_ics.length(); --i >= 0;) + { + inner_class *ic = (inner_class *)requested_ics.get(i); + ic->requested = false; + } + requested_ics.empty(); + + close_output(); + + // rewrite CP references in the tail + cp.computeOutputIndexes(); + int nextref = 0; + for (i = 0; i < (int)class_fixup_type.size(); i++) + { + int type = class_fixup_type.getByte(i); + byte *fixp = wp_at(class_fixup_offset.get(i)); + entry *e = (entry *)class_fixup_ref.get(nextref++); + int idx = e->getOutputIndex(); + switch (type) + { + case 1: + putu1_at(fixp, idx); + break; + case 2: + putu2_at(fixp, idx); + break; + default: + assert(false); // should not reach here + } + } } void unpacker::write_classfile_head() { - cur_classfile_head.empty(); - set_output(&cur_classfile_head); - - putu4(JAVA_MAGIC); - putu2(cur_class_minver); - putu2(cur_class_majver); - putu2(cp.outputIndexLimit); - - int checkIndex = 1; - int noes = cp.outputEntries.length(); - entry **oes = (entry **)cp.outputEntries.base(); - for (int i = 0; i < noes; i++) - { - entry &e = *oes[i]; - assert(e.getOutputIndex() == checkIndex++); - byte tag = e.tag; - assert(tag != CONSTANT_Signature); - putu1(tag); - switch (tag) - { - case CONSTANT_Utf8: - putu2((int)e.value.b.len); - put_bytes(e.value.b); - break; - case CONSTANT_Integer: - case CONSTANT_Float: - putu4(e.value.i); - break; - case CONSTANT_Long: - case CONSTANT_Double: - putu8(e.value.l); - assert(checkIndex++); - break; - case CONSTANT_Class: - case CONSTANT_String: - // just write the ref - putu2(e.refs[0]->getOutputIndex()); - break; - case CONSTANT_Fieldref: - case CONSTANT_Methodref: - case CONSTANT_InterfaceMethodref: - case CONSTANT_NameandType: - putu2(e.refs[0]->getOutputIndex()); - putu2(e.refs[1]->getOutputIndex()); - break; - default: - unpack_abort(ERROR_INTERNAL); - } - } - close_output(); + cur_classfile_head.empty(); + set_output(&cur_classfile_head); + + putu4(JAVA_MAGIC); + putu2(cur_class_minver); + putu2(cur_class_majver); + putu2(cp.outputIndexLimit); + +#ifndef NDEBUG + int checkIndex = 1; +#endif + int noes = cp.outputEntries.length(); + entry **oes = (entry **)cp.outputEntries.base(); + for (int i = 0; i < noes; i++) + { + entry &e = *oes[i]; + assert(e.getOutputIndex() == checkIndex++); + byte tag = e.tag; + assert(tag != CONSTANT_Signature); + putu1(tag); + switch (tag) + { + case CONSTANT_Utf8: + putu2((int)e.value.b.len); + put_bytes(e.value.b); + break; + case CONSTANT_Integer: + case CONSTANT_Float: + putu4(e.value.i); + break; + case CONSTANT_Long: + case CONSTANT_Double: + putu8(e.value.l); + assert(checkIndex++); + break; + case CONSTANT_Class: + case CONSTANT_String: + // just write the ref + putu2(e.refs[0]->getOutputIndex()); + break; + case CONSTANT_Fieldref: + case CONSTANT_Methodref: + case CONSTANT_InterfaceMethodref: + case CONSTANT_NameandType: + putu2(e.refs[0]->getOutputIndex()); + putu2(e.refs[1]->getOutputIndex()); + break; + default: + unpack_abort(ERROR_INTERNAL); + } + } + close_output(); } unpacker::file *unpacker::get_next_file() { - free_temps(); - if (files_remaining == 0) - { - // Leave a clue that we're exhausted. - cur_file.name = nullptr; - cur_file.size = 0; - if (archive_size != 0) - { - uint64_t predicted_size = unsized_bytes_read + archive_size; - if (predicted_size != bytes_read) - unpack_abort("archive header had incorrect size"); - } - return nullptr; - } - files_remaining -= 1; - assert(files_written < file_count || classes_written < class_count); - cur_file.name = ""; - cur_file.size = 0; - cur_file.modtime = default_file_modtime; - cur_file.options = default_file_options; - cur_file.data[0].set(nullptr, 0); - cur_file.data[1].set(nullptr, 0); - if (files_written < file_count) - { - entry *e = file_name.getRef(); - cur_file.name = e->utf8String(); - bool haveLongSize = ((archive_options & AO_HAVE_FILE_SIZE_HI) != 0); - cur_file.size = file_size_hi.getLong(file_size_lo, haveLongSize); - if ((archive_options & AO_HAVE_FILE_MODTIME) != 0) - cur_file.modtime += file_modtime.getInt(); // relative to archive modtime - if ((archive_options & AO_HAVE_FILE_OPTIONS) != 0) - cur_file.options |= file_options.getInt() & ~suppress_file_options; - } - else if (classes_written < class_count) - { - // there is a class for a missing file record - cur_file.options |= FO_IS_CLASS_STUB; - } - if ((cur_file.options & FO_IS_CLASS_STUB) != 0) - { - assert(classes_written < class_count); - classes_written += 1; - if (cur_file.size != 0) - { - unpack_abort("class file size transmitted"); - } - reset_cur_classfile(); - - // write the meat of the classfile: - write_classfile_tail(); - cur_file.data[1] = cur_classfile_tail.b; - - // write the CP of the classfile, second: - write_classfile_head(); - cur_file.data[0] = cur_classfile_head.b; - - cur_file.size += cur_file.data[0].len; - cur_file.size += cur_file.data[1].len; - if (cur_file.name[0] == '\0') - { - bytes &prefix = cur_class->ref(0)->value.b; - const char *suffix = ".class"; - int len = (int)(prefix.len + strlen(suffix)); - bytes name; - name.set(T_NEW(byte, add_size(len, 1)), len); - cur_file.name = name.strcat(prefix).strcat(suffix).strval(); - } - } - else - { - // If there is buffered file data, produce a pointer to it. - if (cur_file.size != (size_t)cur_file.size) - { - // Silly size specified. - unpack_abort("resource file too large"); - } - size_t rpleft = input_remaining(); - if (rpleft > 0) - { - if (rpleft > cur_file.size) - rpleft = (size_t)cur_file.size; - cur_file.data[0].set(rp, rpleft); - rp += rpleft; - } - if (rpleft < cur_file.size) - { - // Caller must read the rest. - size_t fleft = (size_t)cur_file.size - rpleft; - bytes_read += fleft; // Credit it to the overall archive size. - } - } - bytes_written += cur_file.size; - files_written += 1; - return &cur_file; + free_temps(); + if (files_remaining == 0) + { + // Leave a clue that we're exhausted. + cur_file.name = nullptr; + cur_file.size = 0; + if (archive_size != 0) + { + uint64_t predicted_size = unsized_bytes_read + archive_size; + if (predicted_size != bytes_read) + unpack_abort("archive header had incorrect size"); + } + return nullptr; + } + files_remaining -= 1; + assert(files_written < file_count || classes_written < class_count); + cur_file.name = ""; + cur_file.size = 0; + cur_file.modtime = default_file_modtime; + cur_file.options = default_file_options; + cur_file.data[0].set(nullptr, 0); + cur_file.data[1].set(nullptr, 0); + if (files_written < file_count) + { + entry *e = file_name.getRef(); + cur_file.name = e->utf8String(); + bool haveLongSize = ((archive_options & AO_HAVE_FILE_SIZE_HI) != 0); + cur_file.size = file_size_hi.getLong(file_size_lo, haveLongSize); + if ((archive_options & AO_HAVE_FILE_MODTIME) != 0) + cur_file.modtime += file_modtime.getInt(); // relative to archive modtime + if ((archive_options & AO_HAVE_FILE_OPTIONS) != 0) + cur_file.options |= file_options.getInt() & ~suppress_file_options; + } + else if (classes_written < class_count) + { + // there is a class for a missing file record + cur_file.options |= FO_IS_CLASS_STUB; + } + if ((cur_file.options & FO_IS_CLASS_STUB) != 0) + { + assert(classes_written < class_count); + classes_written += 1; + if (cur_file.size != 0) + { + unpack_abort("class file size transmitted"); + } + reset_cur_classfile(); + + // write the meat of the classfile: + write_classfile_tail(); + cur_file.data[1] = cur_classfile_tail.b; + + // write the CP of the classfile, second: + write_classfile_head(); + cur_file.data[0] = cur_classfile_head.b; + + cur_file.size += cur_file.data[0].len; + cur_file.size += cur_file.data[1].len; + if (cur_file.name[0] == '\0') + { + bytes &prefix = cur_class->ref(0)->value.b; + const char *suffix = ".class"; + int len = (int)(prefix.len + strlen(suffix)); + bytes name; + name.set(T_NEW(byte, add_size(len, 1)), len); + cur_file.name = name.strcat(prefix).strcat(suffix).strval(); + } + } + else + { + // If there is buffered file data, produce a pointer to it. + if (cur_file.size != (size_t)cur_file.size) + { + // Silly size specified. + unpack_abort("resource file too large"); + } + size_t rpleft = input_remaining(); + if (rpleft > 0) + { + if (rpleft > cur_file.size) + rpleft = (size_t)cur_file.size; + cur_file.data[0].set(rp, rpleft); + rp += rpleft; + } + if (rpleft < cur_file.size) + { + // Caller must read the rest. + size_t fleft = (size_t)cur_file.size - rpleft; + bytes_read += fleft; // Credit it to the overall archive size. + } + } + bytes_written += cur_file.size; + files_written += 1; + return &cur_file; } // Write a file to jarout. void unpacker::write_file_to_jar(unpacker::file *f) { - size_t htsize = f->data[0].len + f->data[1].len; - uint64_t fsize = f->size; - if (htsize == fsize) - { - jarout->addJarEntry(f->name, f->deflate_hint(), f->modtime, f->data[0], f->data[1]); - } - else - { - assert(input_remaining() == 0); - bytes part1, part2; - part1.len = f->data[0].len; - part1.set(T_NEW(byte, part1.len), part1.len); - part1.copyFrom(f->data[0]); - assert(f->data[1].len == 0); - part2.set(nullptr, 0); - size_t fleft = (size_t)fsize - part1.len; - assert(bytes_read > fleft); // part2 already credited by get_next_file - bytes_read -= fleft; - if (fleft > 0) - { - // Must read some more. - if (live_input) - { - // Stop using the input buffer. Make a new one: - if (free_input) - input.free(); - input.init(fleft > (1 << 12) ? fleft : (1 << 12)); - free_input = true; - live_input = false; - } - else - { - // Make it large enough. - assert(free_input); // must be reallocable - input.ensureSize(fleft); - } - rplimit = rp = input.base(); - input.setLimit(rp + fleft); - if (!ensure_input(fleft)) - unpack_abort("EOF reading resource file"); - part2.ptr = input_scan(); - part2.len = input_remaining(); - rplimit = rp = input.base(); - } - jarout->addJarEntry(f->name, f->deflate_hint(), f->modtime, part1, part2); - } - if (verbose >= 3) - { - fprintf(stderr, "Wrote " LONG_LONG_FORMAT " bytes to: %s\n", fsize, f->name); - } + size_t htsize = f->data[0].len + f->data[1].len; + uint64_t fsize = f->size; + if (htsize == fsize) + { + jarout->addJarEntry(f->name, f->deflate_hint(), f->modtime, f->data[0], f->data[1]); + } + else + { + assert(input_remaining() == 0); + bytes part1, part2; + part1.len = f->data[0].len; + part1.set(T_NEW(byte, part1.len), part1.len); + part1.copyFrom(f->data[0]); + assert(f->data[1].len == 0); + part2.set(nullptr, 0); + size_t fleft = (size_t)fsize - part1.len; + assert(bytes_read > fleft); // part2 already credited by get_next_file + bytes_read -= fleft; + if (fleft > 0) + { + // Must read some more. + if (live_input) + { + // Stop using the input buffer. Make a new one: + if (free_input) + input.free(); + input.init(fleft > (1 << 12) ? fleft : (1 << 12)); + free_input = true; + live_input = false; + } + else + { + // Make it large enough. + assert(free_input); // must be reallocable + input.ensureSize(fleft); + } + rplimit = rp = input.base(); + input.setLimit(rp + fleft); + if (!ensure_input(fleft)) + unpack_abort("EOF reading resource file"); + part2.ptr = input_scan(); + part2.len = input_remaining(); + rplimit = rp = input.base(); + } + jarout->addJarEntry(f->name, f->deflate_hint(), f->modtime, part1, part2); + } + if (verbose >= 3) + { + fprintf(stderr, "Wrote %" PRIu64 " bytes to: %s\n", fsize, f->name); + } } diff --git a/libraries/pack200/src/unpack.h b/libraries/pack200/src/unpack.h index 0100700d..cc5dd60a 100644 --- a/libraries/pack200/src/unpack.h +++ b/libraries/pack200/src/unpack.h @@ -23,6 +23,8 @@ * questions. */ +#pragma once + // Global Structures struct jar; struct gunzip; @@ -33,106 +35,108 @@ struct cpindex; struct inner_class; struct value_stream; +typedef int64_t (*read_input_fn_t)(unpacker *self, void *buf, int64_t minlen, int64_t maxlen); + struct cpindex { - uint32_t len; - entry *base1; // base of primary index - entry **base2; // base of secondary index - byte ixTag; // type of entries (!= CONSTANT_None), plus 64 if sub-index - enum - { - SUB_TAG = 64 - }; - - entry *get(uint32_t i); - - void init(int len_, entry *base1_, int ixTag_) - { - len = len_; - base1 = base1_; - base2 = nullptr; - ixTag = ixTag_; - } - void init(int len_, entry **base2_, int ixTag_) - { - len = len_; - base1 = nullptr; - base2 = base2_; - ixTag = ixTag_; - } + uint32_t len; + entry *base1; // base of primary index + entry **base2; // base of secondary index + byte ixTag; // type of entries (!= CONSTANT_None), plus 64 if sub-index + enum + { + SUB_TAG = 64 + }; + + entry *get(uint32_t i); + + void init(int len_, entry *base1_, int ixTag_) + { + len = len_; + base1 = base1_; + base2 = nullptr; + ixTag = ixTag_; + } + void init(int len_, entry **base2_, int ixTag_) + { + len = len_; + base1 = nullptr; + base2 = base2_; + ixTag = ixTag_; + } }; struct constant_pool { - uint32_t nentries; - entry *entries; - entry *first_extra_entry; - uint32_t maxentries; // total allocated size of entries - - // Position and size of each homogeneous subrange: - int tag_count[CONSTANT_Limit]; - int tag_base[CONSTANT_Limit]; - cpindex tag_index[CONSTANT_Limit]; - ptrlist tag_extras[CONSTANT_Limit]; - - cpindex *member_indexes; // indexed by 2*CONSTANT_Class.inord - cpindex *getFieldIndex(entry *classRef); - cpindex *getMethodIndex(entry *classRef); - - inner_class **ic_index; - inner_class **ic_child_index; - inner_class *getIC(entry *inner); - inner_class *getFirstChildIC(entry *outer); - inner_class *getNextChildIC(inner_class *child); - - int outputIndexLimit; // index limit after renumbering - ptrlist outputEntries; // list of entry* needing output idx assigned - - entry **hashTab; - uint32_t hashTabLength; - entry *&hashTabRef(byte tag, bytes &b); - entry *ensureUtf8(bytes &b); - entry *ensureClass(bytes &b); - - // Well-known Utf8 symbols. - enum - { + uint32_t nentries; + entry *entries; + entry *first_extra_entry; + uint32_t maxentries; // total allocated size of entries + + // Position and size of each homogeneous subrange: + int tag_count[CONSTANT_Limit]; + int tag_base[CONSTANT_Limit]; + cpindex tag_index[CONSTANT_Limit]; + ptrlist tag_extras[CONSTANT_Limit]; + + cpindex *member_indexes; // indexed by 2*CONSTANT_Class.inord + cpindex *getFieldIndex(entry *classRef); + cpindex *getMethodIndex(entry *classRef); + + inner_class **ic_index; + inner_class **ic_child_index; + inner_class *getIC(entry *inner); + inner_class *getFirstChildIC(entry *outer); + inner_class *getNextChildIC(inner_class *child); + + int outputIndexLimit; // index limit after renumbering + ptrlist outputEntries; // list of entry* needing output idx assigned + + entry **hashTab; + uint32_t hashTabLength; + entry *&hashTabRef(byte tag, bytes &b); + entry *ensureUtf8(bytes &b); + entry *ensureClass(bytes &b); + + // Well-known Utf8 symbols. + enum + { #define SNAME(n, s) s_##s, - ALL_ATTR_DO(SNAME) + ALL_ATTR_DO(SNAME) #undef SNAME - s_lt_init_gt, // - s_LIMIT - }; - entry *sym[s_LIMIT]; - - // read counts from hdr, allocate main arrays - enum - { - NUM_COUNTS = 12 - }; - void init(unpacker *u, int counts[NUM_COUNTS]); - - // pointer to outer unpacker, for error checks etc. - unpacker *u; - - int getCount(byte tag) - { - assert((uint32_t)tag < CONSTANT_Limit); - return tag_count[tag]; - } - cpindex *getIndex(byte tag) - { - assert((uint32_t)tag < CONSTANT_Limit); - return &tag_index[tag]; - } - cpindex *getKQIndex(); // uses cur_descr - - void expandSignatures(); - void initMemberIndexes(); - - void computeOutputOrder(); - void computeOutputIndexes(); - void resetOutputIndexes(); + s_lt_init_gt, // + s_LIMIT + }; + entry *sym[s_LIMIT]; + + // read counts from hdr, allocate main arrays + enum + { + NUM_COUNTS = 12 + }; + void init(unpacker *u, int counts[NUM_COUNTS]); + + // pointer to outer unpacker, for error checks etc. + unpacker *u; + + int getCount(byte tag) + { + assert((uint32_t)tag < CONSTANT_Limit); + return tag_count[tag]; + } + cpindex *getIndex(byte tag) + { + assert((uint32_t)tag < CONSTANT_Limit); + return &tag_index[tag]; + } + cpindex *getKQIndex(); // uses cur_descr + + void expandSignatures(); + void initMemberIndexes(); + + void computeOutputOrder(); + void computeOutputIndexes(); + void resetOutputIndexes(); }; /* @@ -141,407 +145,405 @@ struct constant_pool */ struct unpacker { - // One element of the resulting JAR. - struct file - { - const char *name; - uint64_t size; - int modtime; - int options; - bytes data[2]; - // Note: If Sum(data[*].len) < size, - // remaining bytes must be read directly from the input stream. - bool deflate_hint() - { - return ((options & FO_DEFLATE_HINT) != 0); - } - }; - - // if running Unix-style, here are the inputs and outputs - FILE *infileptr; // buffered - bytes inbytes; // direct - gunzip *gzin; // gunzip filter, if any - jar *jarout; // output JAR file - - // pointer to self, for U_NEW macro - unpacker *u; - - ptrlist mallocs; // list of guys to free when we are all done - ptrlist tmallocs; // list of guys to free on next client request - fillbytes smallbuf; // supplies small alloc requests - fillbytes tsmallbuf; // supplies temporary small alloc requests - - // option management members - int verbose; // verbose level, 0 means no output - int deflate_hint_or_zero; // ==0 means not set, otherwise -1 or 1 - int modification_time_or_zero; - - // input stream - fillbytes input; // the whole block (size is predicted, has slop too) - bool live_input; // is the data in this block live? - bool free_input; // must the input buffer be freed? - byte *rp; // read pointer (< rplimit <= input.limit()) - byte *rplimit; // how much of the input block has been read? - uint64_t bytes_read; - int unsized_bytes_read; - - // callback to read at least one byte, up to available input - typedef int64_t (*read_input_fn_t)(unpacker *self, void *buf, int64_t minlen, - int64_t maxlen); - read_input_fn_t read_input_fn; - - // archive header fields - int magic, minver, majver; - size_t archive_size; - int archive_next_count, archive_options, archive_modtime; - int band_headers_size; - int file_count, attr_definition_count, ic_count, class_count; - int default_class_minver, default_class_majver; - int default_file_options, suppress_file_options; // not header fields - int default_archive_modtime, default_file_modtime; // not header fields - int code_count; // not a header field - int files_remaining; // not a header field - - // engine state - band *all_bands; // indexed by band_number - byte *meta_rp; // read-pointer into (copy of) band_headers - constant_pool cp; // all constant pool information - inner_class *ics; // InnerClasses - - // output stream - bytes output; // output block (either classfile head or tail) - byte *wp; // write pointer (< wplimit == output.limit()) - byte *wpbase; // write pointer starting address (<= wp) - byte *wplimit; // how much of the output block has been written? - - // output state - file cur_file; - entry *cur_class; // CONSTANT_Class entry - entry *cur_super; // CONSTANT_Class entry or nullptr - entry *cur_descr; // CONSTANT_NameandType entry - int cur_descr_flags; // flags corresponding to cur_descr - int cur_class_minver, cur_class_majver; - bool cur_class_has_local_ics; - fillbytes cur_classfile_head; - fillbytes cur_classfile_tail; - int files_written; // also tells which file we're working on - int classes_written; // also tells which class we're working on - uint64_t bytes_written; - intlist bcimap; - fillbytes class_fixup_type; - intlist class_fixup_offset; - ptrlist class_fixup_ref; - fillbytes code_fixup_type; // which format of branch operand? - intlist code_fixup_offset; // location of operand needing fixup - intlist code_fixup_source; // encoded ID of branch insn - ptrlist requested_ics; // which ics need output? - - // stats pertaining to multiple segments (updated on reset) - uint64_t bytes_read_before_reset; - uint64_t bytes_written_before_reset; - int files_written_before_reset; - int classes_written_before_reset; - int segments_read_before_reset; - - // attribute state - struct layout_definition - { - uint32_t idx; // index (0..31...) which identifies this layout - const char *name; // name of layout - entry *nameEntry; - const char *layout; // string of layout (not yet parsed) - band **elems; // array of top-level layout elems (or callables) - - bool hasCallables() - { - return layout[0] == '['; - } - band **bands() - { - assert(elems != nullptr); - return elems; - } - }; - struct attr_definitions - { - unpacker *u; // pointer to self, for U_NEW macro - int xxx_flags_hi_bn; // locator for flags, count, indexes, calls bands - int attrc; // ATTR_CONTEXT_CLASS, etc. - uint32_t flag_limit; // 32 or 63, depending on archive_options bit - uint64_t predef; // mask of built-in definitions - uint64_t redef; // mask of local flag definitions or redefinitions - ptrlist layouts; // local (compressor-defined) defs, in index order - int flag_count[X_ATTR_LIMIT_FLAGS_HI]; - intlist overflow_count; - ptrlist strip_names; // what attribute names are being stripped? - ptrlist band_stack; // Temp., used during layout parsing. - ptrlist calls_to_link; // (ditto) - int bands_made; // (ditto) - - void free() - { - layouts.free(); - overflow_count.free(); - strip_names.free(); - band_stack.free(); - calls_to_link.free(); - } - - // Locate the five fixed bands. - band &xxx_flags_hi(); - band &xxx_flags_lo(); - band &xxx_attr_count(); - band &xxx_attr_indexes(); - band &xxx_attr_calls(); - band &fixed_band(int e_class_xxx); - - // Register a new layout, and make bands for it. - layout_definition *defineLayout(int idx, const char *name, const char *layout); - layout_definition *defineLayout(int idx, entry *nameEntry, const char *layout); - band **buildBands(layout_definition *lo); - - // Parse a layout string or part of one, recursively if necessary. - const char *parseLayout(const char *lp, band **&res, int curCble); - const char *parseNumeral(const char *lp, int &res); - const char *parseIntLayout(const char *lp, band *&res, byte le_kind, - bool can_be_signed = false); - band **popBody(int band_stack_base); // pops a body off band_stack - - // Read data into the bands of the idx-th layout. - void readBandData(int idx); // parse layout, make bands, read data - void readBandData(band **body, uint32_t count); // recursive helper - - layout_definition *getLayout(uint32_t idx) - { - if (idx >= (uint32_t)layouts.length()) - return nullptr; - return (layout_definition *)layouts.get(idx); - } - - void setHaveLongFlags(bool z) - { - assert(flag_limit == 0); // not set up yet - flag_limit = (z ? X_ATTR_LIMIT_FLAGS_HI : X_ATTR_LIMIT_NO_FLAGS_HI); - } - bool haveLongFlags() - { - assert(flag_limit == X_ATTR_LIMIT_NO_FLAGS_HI || - flag_limit == X_ATTR_LIMIT_FLAGS_HI); - return flag_limit == X_ATTR_LIMIT_FLAGS_HI; - } - - // Return flag_count if idx is predef and not redef, else zero. - int predefCount(uint32_t idx); - - bool isRedefined(uint32_t idx) - { - if (idx >= flag_limit) - return false; - return (bool)((redef >> idx) & 1); - } - bool isPredefined(uint32_t idx) - { - if (idx >= flag_limit) - return false; - return (bool)(((predef & ~redef) >> idx) & 1); - } - uint64_t flagIndexMask() - { - return (predef | redef); - } - bool isIndex(uint32_t idx) - { - assert(flag_limit != 0); // must be set up already - if (idx < flag_limit) - return (bool)(((predef | redef) >> idx) & 1); - else - return (idx - flag_limit < (uint32_t)overflow_count.length()); - } - int &getCount(uint32_t idx) - { - assert(isIndex(idx)); - if (idx < flag_limit) - return flag_count[idx]; - else - return overflow_count.get(idx - flag_limit); - } - }; - - attr_definitions attr_defs[ATTR_CONTEXT_LIMIT]; - - // Initialization - void init(read_input_fn_t input_fn = nullptr); - // Resets to a known sane state - void reset(); - // Deallocates all storage. - void free(); - // Deallocates temporary storage (volatile after next client call). - void free_temps() - { - tsmallbuf.init(); - tmallocs.freeAll(); - } - - // Option management methods - bool set_option(const char *option, const char *value); - const char *get_option(const char *option); - - // Fetching input. - bool ensure_input(int64_t more); - byte *input_scan() - { - return rp; - } - size_t input_remaining() - { - return rplimit - rp; - } - size_t input_consumed() - { - return rp - input.base(); - } - - // Entry points to the unpack engine - static int run(int argc, char **argv); // Unix-style entry point. - void check_options(); - void start(void *packptr = nullptr, size_t len = 0); - void write_file_to_jar(file *f); - void finish(); - - // Public post unpack methods - int get_files_remaining() - { - return files_remaining; - } - int get_segments_remaining() - { - return archive_next_count; - } - file *get_next_file(); // returns nullptr on last file - - // General purpose methods - void *alloc(size_t size) - { - return alloc_heap(size, true); - } - void *temp_alloc(size_t size) - { - return alloc_heap(size, true, true); - } - void *alloc_heap(size_t size, bool smallOK = false, bool temp = false); - void saveTo(bytes &b, const char *str) - { - saveTo(b, (byte *)str, strlen(str)); - } - void saveTo(bytes &b, bytes &data) - { - saveTo(b, data.ptr, data.len); - } - void saveTo(bytes &b, byte *ptr, size_t len); //{ b.ptr = U_NEW...} - const char *saveStr(const char *str) - { - bytes buf; - saveTo(buf, str); - return buf.strval(); - } - const char *saveIntStr(int num) - { - char buf[30]; - sprintf(buf, "%d", num); - return saveStr(buf); - } - static unpacker *current(); // find current instance - - // Output management - void set_output(fillbytes *which) - { - assert(wp == nullptr); - which->ensureSize(1 << 12); // covers the average classfile - wpbase = which->base(); - wp = which->limit(); - wplimit = which->end(); - } - fillbytes *close_output(fillbytes *which = nullptr); // inverse of set_output - - // These take an implicit parameter of wp/wplimit, and resize as necessary: - byte *put_space(size_t len); // allocates space at wp, returns pointer - size_t put_empty(size_t s) - { - byte *p = put_space(s); - return p - wpbase; - } - void ensure_put_space(size_t len); - void put_bytes(bytes &b) - { - b.writeTo(put_space(b.len)); - } - void putu1(int n) - { - putu1_at(put_space(1), n); - } - void putu1_fast(int n) - { - putu1_at(wp++, n); - } - void putu2(int n); // { putu2_at(put_space(2), n); } - void putu4(int n); // { putu4_at(put_space(4), n); } - void putu8(int64_t n); // { putu8_at(put_space(8), n); } - void putref(entry *e); // { putu2_at(put_space(2), putref_index(e, 2)); } - void putu1ref(entry *e); // { putu1_at(put_space(1), putref_index(e, 1)); } - int putref_index(entry *e, int size); // size in [1..2] - void put_label(int curIP, int size); // size in {2,4} - void putlayout(band **body); - void put_stackmap_type(); - - size_t wpoffset() - { - return (size_t)(wp - wpbase); - } // (unvariant across overflow) - byte *wp_at(size_t offset) - { - return wpbase + offset; - } - uint32_t to_bci(uint32_t bii); - void get_code_header(int &max_stack, int &max_na_locals, int &handler_count, int &cflags); - band *ref_band_for_self_op(int bc, bool &isAloadVar, int &origBCVar); - band *ref_band_for_op(int bc); - - // Definitions of standard classfile int formats: - static void putu1_at(byte *wp, int n) - { - assert(n == (n & 0xFF)); - wp[0] = n; - } - static void putu2_at(byte *wp, int n); - static void putu4_at(byte *wp, int n); - static void putu8_at(byte *wp, int64_t n); - - // Private stuff - void reset_cur_classfile(); - void write_classfile_tail(); - void write_classfile_head(); - void write_code(); - void write_bc_ops(); - void write_members(int num, int attrc); // attrc=ATTR_CONTEXT_FIELD/METHOD - int write_attrs(int attrc, uint64_t indexBits); - - // The readers - void read_bands(); - void read_file_header(); - void read_cp(); - void read_cp_counts(value_stream &hdr); - void read_attr_defs(); - void read_ics(); - void read_attrs(int attrc, int obj_count); - void read_classes(); - void read_code_headers(); - void read_bcs(); - void read_bc_ops(); - void read_files(); - void read_Utf8_values(entry *cpMap, int len); - void read_single_words(band &cp_band, entry *cpMap, int len); - void read_double_words(band &cp_bands, entry *cpMap, int len); - void read_single_refs(band &cp_band, byte refTag, entry *cpMap, int len); - void read_double_refs(band &cp_band, byte ref1Tag, byte ref2Tag, entry *cpMap, int len); - void read_signature_values(entry *cpMap, int len); + // One element of the resulting JAR. + struct file + { + const char *name; + uint64_t size; + int modtime; + int options; + bytes data[2]; + // Note: If Sum(data[*].len) < size, + // remaining bytes must be read directly from the input stream. + bool deflate_hint() + { + return ((options & FO_DEFLATE_HINT) != 0); + } + }; + + // if running Unix-style, here are the inputs and outputs + FILE *infileptr; // buffered + bytes inbytes; // direct + gunzip *gzin; // gunzip filter, if any + jar *jarout; // output JAR file + + // pointer to self, for U_NEW macro + unpacker *u; + + ptrlist mallocs; // list of guys to free when we are all done + ptrlist tmallocs; // list of guys to free on next client request + fillbytes smallbuf; // supplies small alloc requests + fillbytes tsmallbuf; // supplies temporary small alloc requests + + // option management members + int verbose; // verbose level, 0 means no output + int deflate_hint_or_zero; // ==0 means not set, otherwise -1 or 1 + int modification_time_or_zero; + + // input stream + fillbytes input; // the whole block (size is predicted, has slop too) + bool live_input; // is the data in this block live? + bool free_input; // must the input buffer be freed? + byte *rp; // read pointer (< rplimit <= input.limit()) + byte *rplimit; // how much of the input block has been read? + uint64_t bytes_read; + int unsized_bytes_read; + + // callback to read at least one byte, up to available input + read_input_fn_t read_input_fn; + + // archive header fields + int magic, minver, majver; + size_t archive_size; + int archive_next_count, archive_options, archive_modtime; + int band_headers_size; + int file_count, attr_definition_count, ic_count, class_count; + int default_class_minver, default_class_majver; + int default_file_options, suppress_file_options; // not header fields + int default_archive_modtime, default_file_modtime; // not header fields + int code_count; // not a header field + int files_remaining; // not a header field + + // engine state + band *all_bands; // indexed by band_number + byte *meta_rp; // read-pointer into (copy of) band_headers + constant_pool cp; // all constant pool information + inner_class *ics; // InnerClasses + + // output stream + bytes output; // output block (either classfile head or tail) + byte *wp; // write pointer (< wplimit == output.limit()) + byte *wpbase; // write pointer starting address (<= wp) + byte *wplimit; // how much of the output block has been written? + + // output state + file cur_file; + entry *cur_class; // CONSTANT_Class entry + entry *cur_super; // CONSTANT_Class entry or nullptr + entry *cur_descr; // CONSTANT_NameandType entry + int cur_descr_flags; // flags corresponding to cur_descr + int cur_class_minver, cur_class_majver; + bool cur_class_has_local_ics; + fillbytes cur_classfile_head; + fillbytes cur_classfile_tail; + int files_written; // also tells which file we're working on + int classes_written; // also tells which class we're working on + uint64_t bytes_written; + intlist bcimap; + fillbytes class_fixup_type; + intlist class_fixup_offset; + ptrlist class_fixup_ref; + fillbytes code_fixup_type; // which format of branch operand? + intlist code_fixup_offset; // location of operand needing fixup + intlist code_fixup_source; // encoded ID of branch insn + ptrlist requested_ics; // which ics need output? + + // stats pertaining to multiple segments (updated on reset) + uint64_t bytes_read_before_reset; + uint64_t bytes_written_before_reset; + int files_written_before_reset; + int classes_written_before_reset; + int segments_read_before_reset; + + // attribute state + struct layout_definition + { + uint32_t idx; // index (0..31...) which identifies this layout + const char *name; // name of layout + entry *nameEntry; + const char *layout; // string of layout (not yet parsed) + band **elems; // array of top-level layout elems (or callables) + + bool hasCallables() + { + return layout[0] == '['; + } + band **bands() + { + assert(elems != nullptr); + return elems; + } + }; + struct attr_definitions + { + unpacker *u; // pointer to self, for U_NEW macro + int xxx_flags_hi_bn; // locator for flags, count, indexes, calls bands + int attrc; // ATTR_CONTEXT_CLASS, etc. + uint32_t flag_limit; // 32 or 63, depending on archive_options bit + uint64_t predef; // mask of built-in definitions + uint64_t redef; // mask of local flag definitions or redefinitions + ptrlist layouts; // local (compressor-defined) defs, in index order + int flag_count[X_ATTR_LIMIT_FLAGS_HI]; + intlist overflow_count; + ptrlist strip_names; // what attribute names are being stripped? + ptrlist band_stack; // Temp., used during layout parsing. + ptrlist calls_to_link; // (ditto) + int bands_made; // (ditto) + + void free() + { + layouts.free(); + overflow_count.free(); + strip_names.free(); + band_stack.free(); + calls_to_link.free(); + } + + // Locate the five fixed bands. + band &xxx_flags_hi(); + band &xxx_flags_lo(); + band &xxx_attr_count(); + band &xxx_attr_indexes(); + band &xxx_attr_calls(); + band &fixed_band(int e_class_xxx); + + // Register a new layout, and make bands for it. + layout_definition *defineLayout(int idx, const char *name, const char *layout); + layout_definition *defineLayout(int idx, entry *nameEntry, const char *layout); + band **buildBands(layout_definition *lo); + + // Parse a layout string or part of one, recursively if necessary. + const char *parseLayout(const char *lp, band **&res, int curCble); + const char *parseNumeral(const char *lp, int &res); + const char *parseIntLayout(const char *lp, band *&res, byte le_kind, + bool can_be_signed = false); + band **popBody(int band_stack_base); // pops a body off band_stack + + // Read data into the bands of the idx-th layout. + void readBandData(int idx); // parse layout, make bands, read data + void readBandData(band **body, uint32_t count); // recursive helper + + layout_definition *getLayout(uint32_t idx) + { + if (idx >= (uint32_t)layouts.length()) + return nullptr; + return (layout_definition *)layouts.get(idx); + } + + void setHaveLongFlags(bool z) + { + assert(flag_limit == 0); // not set up yet + flag_limit = (z ? X_ATTR_LIMIT_FLAGS_HI : X_ATTR_LIMIT_NO_FLAGS_HI); + } + bool haveLongFlags() + { + assert(flag_limit == X_ATTR_LIMIT_NO_FLAGS_HI || + flag_limit == X_ATTR_LIMIT_FLAGS_HI); + return flag_limit == X_ATTR_LIMIT_FLAGS_HI; + } + + // Return flag_count if idx is predef and not redef, else zero. + int predefCount(uint32_t idx); + + bool isRedefined(uint32_t idx) + { + if (idx >= flag_limit) + return false; + return (bool)((redef >> idx) & 1); + } + bool isPredefined(uint32_t idx) + { + if (idx >= flag_limit) + return false; + return (bool)(((predef & ~redef) >> idx) & 1); + } + uint64_t flagIndexMask() + { + return (predef | redef); + } + bool isIndex(uint32_t idx) + { + assert(flag_limit != 0); // must be set up already + if (idx < flag_limit) + return (bool)(((predef | redef) >> idx) & 1); + else + return (idx - flag_limit < (uint32_t)overflow_count.length()); + } + int &getCount(uint32_t idx) + { + assert(isIndex(idx)); + if (idx < flag_limit) + return flag_count[idx]; + else + return overflow_count.get(idx - flag_limit); + } + }; + + attr_definitions attr_defs[ATTR_CONTEXT_LIMIT]; + + // Initialization + void init(read_input_fn_t input_fn = nullptr); + // Resets to a known sane state + void reset(); + // Deallocates all storage. + void free(); + // Deallocates temporary storage (volatile after next client call). + void free_temps() + { + tsmallbuf.init(); + tmallocs.freeAll(); + } + + // Option management methods + bool set_option(const char *option, const char *value); + const char *get_option(const char *option); + + // Fetching input. + bool ensure_input(int64_t more); + byte *input_scan() + { + return rp; + } + size_t input_remaining() + { + return rplimit - rp; + } + size_t input_consumed() + { + return rp - input.base(); + } + + // Entry points to the unpack engine + static int run(int argc, char **argv); // Unix-style entry point. + void check_options(); + void start(void *packptr = nullptr, size_t len = 0); + void write_file_to_jar(file *f); + void finish(); + + // Public post unpack methods + int get_files_remaining() + { + return files_remaining; + } + int get_segments_remaining() + { + return archive_next_count; + } + file *get_next_file(); // returns nullptr on last file + + // General purpose methods + void *alloc(size_t size) + { + return alloc_heap(size, true); + } + void *temp_alloc(size_t size) + { + return alloc_heap(size, true, true); + } + void *alloc_heap(size_t size, bool smallOK = false, bool temp = false); + void saveTo(bytes &b, const char *str) + { + saveTo(b, (byte *)str, strlen(str)); + } + void saveTo(bytes &b, bytes &data) + { + saveTo(b, data.ptr, data.len); + } + void saveTo(bytes &b, byte *ptr, size_t len); //{ b.ptr = U_NEW...} + const char *saveStr(const char *str) + { + bytes buf; + saveTo(buf, str); + return buf.strval(); + } + const char *saveIntStr(int num) + { + char buf[30]; + sprintf(buf, "%d", num); + return saveStr(buf); + } + static unpacker *current(); // find current instance + + // Output management + void set_output(fillbytes *which) + { + assert(wp == nullptr); + which->ensureSize(1 << 12); // covers the average classfile + wpbase = which->base(); + wp = which->limit(); + wplimit = which->end(); + } + fillbytes *close_output(fillbytes *which = nullptr); // inverse of set_output + + // These take an implicit parameter of wp/wplimit, and resize as necessary: + byte *put_space(size_t len); // allocates space at wp, returns pointer + size_t put_empty(size_t s) + { + byte *p = put_space(s); + return p - wpbase; + } + void ensure_put_space(size_t len); + void put_bytes(bytes &b) + { + b.writeTo(put_space(b.len)); + } + void putu1(int n) + { + putu1_at(put_space(1), n); + } + void putu1_fast(int n) + { + putu1_at(wp++, n); + } + void putu2(int n); // { putu2_at(put_space(2), n); } + void putu4(int n); // { putu4_at(put_space(4), n); } + void putu8(int64_t n); // { putu8_at(put_space(8), n); } + void putref(entry *e); // { putu2_at(put_space(2), putref_index(e, 2)); } + void putu1ref(entry *e); // { putu1_at(put_space(1), putref_index(e, 1)); } + int putref_index(entry *e, int size); // size in [1..2] + void put_label(int curIP, int size); // size in {2,4} + void putlayout(band **body); + void put_stackmap_type(); + + size_t wpoffset() + { + return (size_t)(wp - wpbase); + } // (unvariant across overflow) + byte *wp_at(size_t offset) + { + return wpbase + offset; + } + uint32_t to_bci(uint32_t bii); + void get_code_header(int &max_stack, int &max_na_locals, int &handler_count, int &cflags); + band *ref_band_for_self_op(int bc, bool &isAloadVar, int &origBCVar); + band *ref_band_for_op(int bc); + + // Definitions of standard classfile int formats: + static void putu1_at(byte *wp, int n) + { + assert(n == (n & 0xFF)); + wp[0] = n; + } + static void putu2_at(byte *wp, int n); + static void putu4_at(byte *wp, int n); + static void putu8_at(byte *wp, int64_t n); + + // Private stuff + void reset_cur_classfile(); + void write_classfile_tail(); + void write_classfile_head(); + void write_code(); + void write_bc_ops(); + void write_members(int num, int attrc); // attrc=ATTR_CONTEXT_FIELD/METHOD + int write_attrs(int attrc, uint64_t indexBits); + + // The readers + void read_bands(); + void read_file_header(); + void read_cp(); + void read_cp_counts(value_stream &hdr); + void read_attr_defs(); + void read_ics(); + void read_attrs(int attrc, int obj_count); + void read_classes(); + void read_code_headers(); + void read_bcs(); + void read_bc_ops(); + void read_files(); + void read_Utf8_values(entry *cpMap, int len); + void read_single_words(band &cp_band, entry *cpMap, int len); + void read_double_words(band &cp_bands, entry *cpMap, int len); + void read_single_refs(band &cp_band, byte refTag, entry *cpMap, int len); + void read_double_refs(band &cp_band, byte ref1Tag, byte ref2Tag, entry *cpMap, int len); + void read_signature_values(entry *cpMap, int len); }; diff --git a/libraries/pack200/src/unpack200.cpp b/libraries/pack200/src/unpack200.cpp index 22b7f3b0..e8826f28 100644 --- a/libraries/pack200/src/unpack200.cpp +++ b/libraries/pack200/src/unpack200.cpp @@ -45,118 +45,118 @@ // Callback for fetching data, Unix style. static int64_t read_input_via_stdio(unpacker *u, void *buf, int64_t minlen, int64_t maxlen) { - assert(u->infileptr != nullptr); - assert(minlen <= maxlen); // don't talk nonsense - int64_t numread = 0; - char *bufptr = (char *)buf; - while (numread < minlen) - { - // read available input, up to buf.length or maxlen - int readlen = (1 << 16); - if (readlen > (maxlen - numread)) - readlen = (int)(maxlen - numread); - int nr = 0; + assert(u->infileptr != nullptr); + assert(minlen <= maxlen); // don't talk nonsense + int64_t numread = 0; + char *bufptr = (char *)buf; + while (numread < minlen) + { + // read available input, up to buf.length or maxlen + int readlen = (1 << 16); + if (readlen > (maxlen - numread)) + readlen = (int)(maxlen - numread); + int nr = 0; - nr = (int)fread(bufptr, 1, readlen, u->infileptr); - if (nr <= 0) - { - if (errno != EINTR) - break; - nr = 0; - } - numread += nr; - bufptr += nr; - assert(numread <= maxlen); - } - return numread; + nr = (int)fread(bufptr, 1, readlen, u->infileptr); + if (nr <= 0) + { + if (errno != EINTR) + break; + nr = 0; + } + numread += nr; + bufptr += nr; + assert(numread <= maxlen); + } + return numread; } enum { - EOF_MAGIC = 0, - BAD_MAGIC = -1 + EOF_MAGIC = 0, + BAD_MAGIC = -1 }; static int read_magic(unpacker *u, char peek[], int peeklen) { - assert(peeklen == 4); // magic numbers are always 4 bytes - int64_t nr = (u->read_input_fn)(u, peek, peeklen, peeklen); - if (nr != peeklen) - { - return (nr == 0) ? EOF_MAGIC : BAD_MAGIC; - } - int magic = 0; - for (int i = 0; i < peeklen; i++) - { - magic <<= 8; - magic += peek[i] & 0xFF; - } - return magic; + assert(peeklen == 4); // magic numbers are always 4 bytes + int64_t nr = (u->read_input_fn)(u, peek, peeklen, peeklen); + if (nr != peeklen) + { + return (nr == 0) ? EOF_MAGIC : BAD_MAGIC; + } + int magic = 0; + for (int i = 0; i < peeklen; i++) + { + magic <<= 8; + magic += peek[i] & 0xFF; + } + return magic; } void unpack_200(FILE *input, FILE *output) { - unpacker u; - u.init(read_input_via_stdio); + unpacker u; + u.init(read_input_via_stdio); - // initialize jar output - // the output takes ownership of the file handle - jar jarout; - jarout.init(&u); - jarout.jarfp = output; + // initialize jar output + // the output takes ownership of the file handle + jar jarout; + jarout.init(&u); + jarout.jarfp = output; - // the input doesn't - u.infileptr = input; + // the input doesn't + u.infileptr = input; - // read the magic! - char peek[4]; - int magic; - magic = read_magic(&u, peek, (int)sizeof(peek)); + // read the magic! + char peek[4]; + int magic; + magic = read_magic(&u, peek, (int)sizeof(peek)); - // if it is a gzip encoded file, we need an extra gzip input filter - if ((magic & GZIP_MAGIC_MASK) == GZIP_MAGIC) - { - gunzip *gzin = NEW(gunzip, 1); - gzin->init(&u); - // FIXME: why the side effects? WHY? - u.gzin->start(magic); - u.start(); - } - else - { - // otherwise, feed the bytes to the unpacker directly - u.start(peek, sizeof(peek)); - } + // if it is a gzip encoded file, we need an extra gzip input filter + if ((magic & GZIP_MAGIC_MASK) == GZIP_MAGIC) + { + gunzip *gzin = NEW(gunzip, 1); + gzin->init(&u); + // FIXME: why the side effects? WHY? + u.gzin->start(magic); + u.start(); + } + else + { + // otherwise, feed the bytes to the unpacker directly + u.start(peek, sizeof(peek)); + } - // Note: The checks to u.aborting() are necessary to gracefully - // terminate processing when the first segment throws an error. - for (;;) - { - // Each trip through this loop unpacks one segment - // and then resets the unpacker. - for (unpacker::file *filep; (filep = u.get_next_file()) != nullptr;) - { - u.write_file_to_jar(filep); - } + // Note: The checks to u.aborting() are necessary to gracefully + // terminate processing when the first segment throws an error. + for (;;) + { + // Each trip through this loop unpacks one segment + // and then resets the unpacker. + for (unpacker::file *filep; (filep = u.get_next_file()) != nullptr;) + { + u.write_file_to_jar(filep); + } - // Peek ahead for more data. - magic = read_magic(&u, peek, (int)sizeof(peek)); - if (magic != (int)JAVA_PACKAGE_MAGIC) - { - // we do not feel strongly about this kind of thing... - /* - if (magic != EOF_MAGIC) - unpack_abort("garbage after end of pack archive"); - */ - break; // all done - } + // Peek ahead for more data. + magic = read_magic(&u, peek, (int)sizeof(peek)); + if (magic != (int)JAVA_PACKAGE_MAGIC) + { + // we do not feel strongly about this kind of thing... + /* + if (magic != EOF_MAGIC) + unpack_abort("garbage after end of pack archive"); + */ + break; // all done + } - // Release all storage from parsing the old segment. - u.reset(); - // Restart, beginning with the peek-ahead. - u.start(peek, sizeof(peek)); - } - u.finish(); - u.free(); // tidy up malloc blocks - fclose(input); + // Release all storage from parsing the old segment. + u.reset(); + // Restart, beginning with the peek-ahead. + u.start(peek, sizeof(peek)); + } + u.finish(); + u.free(); // tidy up malloc blocks + fclose(input); } diff --git a/libraries/pack200/src/utils.cpp b/libraries/pack200/src/utils.cpp index 0b7d91ca..fd6dad60 100644 --- a/libraries/pack200/src/utils.cpp +++ b/libraries/pack200/src/utils.cpp @@ -50,22 +50,22 @@ void *must_malloc(size_t size) { - size_t msize = size; - void *ptr = (msize > PSIZE_MAX) ? nullptr : malloc(msize); - if (ptr != nullptr) - { - memset(ptr, 0, size); - } - else - { - throw std::runtime_error(ERROR_ENOMEM); - } - return ptr; + size_t msize = size; + void *ptr = (msize > PSIZE_MAX) ? nullptr : malloc(msize); + if (ptr != nullptr) + { + memset(ptr, 0, size); + } + else + { + throw std::runtime_error(ERROR_ENOMEM); + } + return ptr; } void unpack_abort(const char *msg) { - if (msg == nullptr) - msg = "corrupt pack file or internal error"; - throw std::runtime_error(msg); + if (msg == nullptr) + msg = "corrupt pack file or internal error"; + throw std::runtime_error(msg); } diff --git a/libraries/pack200/src/utils.h b/libraries/pack200/src/utils.h index 5a3dc8f6..3bd2dae7 100644 --- a/libraries/pack200/src/utils.h +++ b/libraries/pack200/src/utils.h @@ -35,17 +35,17 @@ void *must_malloc(size_t size); inline size_t scale_size(size_t size, size_t scale) { - return (size > PSIZE_MAX / scale) ? OVERFLOW : size * scale; + return (size > PSIZE_MAX / scale) ? OVERFLOW : size * scale; } inline size_t add_size(size_t size1, size_t size2) { - return ((size1 | size2 | (size1 + size2)) > PSIZE_MAX) ? OVERFLOW : size1 + size2; + return ((size1 | size2 | (size1 + size2)) > PSIZE_MAX) ? OVERFLOW : size1 + size2; } inline size_t add_size(size_t size1, size_t size2, int size3) { - return add_size(add_size(size1, size2), size3); + return add_size(add_size(size1, size2), size3); } struct unpacker; diff --git a/libraries/pack200/src/zip.cpp b/libraries/pack200/src/zip.cpp index 32e8bd50..e776510b 100644 --- a/libraries/pack200/src/zip.cpp +++ b/libraries/pack200/src/zip.cpp @@ -52,7 +52,7 @@ inline uint32_t jar::get_crc32(uint32_t c, uchar *ptr, uint32_t len) { - return crc32(c, ptr, len); + return crc32(c, ptr, len); } // FIXME: this is bullshit. Do real endianness detection. @@ -68,175 +68,175 @@ inline uint32_t jar::get_crc32(uint32_t c, uchar *ptr, uint32_t len) void jar::init(unpacker *u_) { - BYTES_OF(*this).clear(); - u = u_; - u->jarout = this; + BYTES_OF(*this).clear(); + u = u_; + u->jarout = this; } // Write data to the ZIP output stream. void jar::write_data(void *buff, int len) { - while (len > 0) - { - int rc = (int)fwrite(buff, 1, len, jarfp); - if (rc <= 0) - { - fprintf(stderr, "Error: write on output file failed err=%d\n", errno); - exit(1); // Called only from the native standalone unpacker - } - output_file_offset += rc; - buff = ((char *)buff) + rc; - len -= rc; - } + while (len > 0) + { + int rc = (int)fwrite(buff, 1, len, jarfp); + if (rc <= 0) + { + fprintf(stderr, "Error: write on output file failed err=%d\n", errno); + exit(1); // Called only from the native standalone unpacker + } + output_file_offset += rc; + buff = ((char *)buff) + rc; + len -= rc; + } } void jar::add_to_jar_directory(const char *fname, bool store, int modtime, int len, int clen, - uint32_t crc) + uint32_t crc) { - uint32_t fname_length = (uint32_t)strlen(fname); - ushort header[23]; - if (modtime == 0) - modtime = default_modtime; - uint32_t dostime = get_dostime(modtime); - - header[0] = (ushort)SWAP_BYTES(0x4B50); - header[1] = (ushort)SWAP_BYTES(0x0201); - header[2] = (ushort)SWAP_BYTES(0xA); - - // required version - header[3] = (ushort)SWAP_BYTES(0xA); - - // flags 02 = maximum sub-compression flag - header[4] = (store) ? 0x0 : SWAP_BYTES(0x2); - - // Compression method 8=deflate. - header[5] = (store) ? 0x0 : SWAP_BYTES(0x08); - - // Last modified date and time. - header[6] = (ushort)GET_INT_LO(dostime); - header[7] = (ushort)GET_INT_HI(dostime); - - // CRC - header[8] = (ushort)GET_INT_LO(crc); - header[9] = (ushort)GET_INT_HI(crc); - - // Compressed length: - header[10] = (ushort)GET_INT_LO(clen); - header[11] = (ushort)GET_INT_HI(clen); - - // Uncompressed length. - header[12] = (ushort)GET_INT_LO(len); - header[13] = (ushort)GET_INT_HI(len); - - // Filename length - header[14] = (ushort)SWAP_BYTES(fname_length); - // So called "extra field" length. - header[15] = 0; - // So called "comment" length. - header[16] = 0; - // Disk number start - header[17] = 0; - // File flags => binary - header[18] = 0; - // More file flags - header[19] = 0; - header[20] = 0; - // Offset within ZIP file. - header[21] = (ushort)GET_INT_LO(output_file_offset); - header[22] = (ushort)GET_INT_HI(output_file_offset); - - // Copy the whole thing into the central directory. - central_directory.append(header, sizeof(header)); - - // Copy the fname to the header. - central_directory.append(fname, fname_length); - - central_directory_count++; + uint32_t fname_length = (uint32_t)strlen(fname); + ushort header[23]; + if (modtime == 0) + modtime = default_modtime; + uint32_t dostime = get_dostime(modtime); + + header[0] = (ushort)SWAP_BYTES(0x4B50); + header[1] = (ushort)SWAP_BYTES(0x0201); + header[2] = (ushort)SWAP_BYTES(0xA); + + // required version + header[3] = (ushort)SWAP_BYTES(0xA); + + // flags 02 = maximum sub-compression flag + header[4] = (store) ? 0x0 : SWAP_BYTES(0x2); + + // Compression method 8=deflate. + header[5] = (store) ? 0x0 : SWAP_BYTES(0x08); + + // Last modified date and time. + header[6] = (ushort)GET_INT_LO(dostime); + header[7] = (ushort)GET_INT_HI(dostime); + + // CRC + header[8] = (ushort)GET_INT_LO(crc); + header[9] = (ushort)GET_INT_HI(crc); + + // Compressed length: + header[10] = (ushort)GET_INT_LO(clen); + header[11] = (ushort)GET_INT_HI(clen); + + // Uncompressed length. + header[12] = (ushort)GET_INT_LO(len); + header[13] = (ushort)GET_INT_HI(len); + + // Filename length + header[14] = (ushort)SWAP_BYTES(fname_length); + // So called "extra field" length. + header[15] = 0; + // So called "comment" length. + header[16] = 0; + // Disk number start + header[17] = 0; + // File flags => binary + header[18] = 0; + // More file flags + header[19] = 0; + header[20] = 0; + // Offset within ZIP file. + header[21] = (ushort)GET_INT_LO(output_file_offset); + header[22] = (ushort)GET_INT_HI(output_file_offset); + + // Copy the whole thing into the central directory. + central_directory.append(header, sizeof(header)); + + // Copy the fname to the header. + central_directory.append(fname, fname_length); + + central_directory_count++; } void jar::write_jar_header(const char *fname, bool store, int modtime, int len, int clen, - uint32_t crc) + uint32_t crc) { - uint32_t fname_length = (uint32_t)strlen(fname); - ushort header[15]; - if (modtime == 0) - modtime = default_modtime; - uint32_t dostime = get_dostime(modtime); + uint32_t fname_length = (uint32_t)strlen(fname); + ushort header[15]; + if (modtime == 0) + modtime = default_modtime; + uint32_t dostime = get_dostime(modtime); - // ZIP LOC magic. - header[0] = (ushort)SWAP_BYTES(0x4B50); - header[1] = (ushort)SWAP_BYTES(0x0403); + // ZIP LOC magic. + header[0] = (ushort)SWAP_BYTES(0x4B50); + header[1] = (ushort)SWAP_BYTES(0x0403); - // Version - header[2] = (ushort)SWAP_BYTES(0xA); + // Version + header[2] = (ushort)SWAP_BYTES(0xA); - // flags 02 = maximum sub-compression flag - header[3] = (store) ? 0x0 : SWAP_BYTES(0x2); + // flags 02 = maximum sub-compression flag + header[3] = (store) ? 0x0 : SWAP_BYTES(0x2); - // Compression method = deflate - header[4] = (store) ? 0x0 : SWAP_BYTES(0x08); + // Compression method = deflate + header[4] = (store) ? 0x0 : SWAP_BYTES(0x08); - // Last modified date and time. - header[5] = (ushort)GET_INT_LO(dostime); - header[6] = (ushort)GET_INT_HI(dostime); + // Last modified date and time. + header[5] = (ushort)GET_INT_LO(dostime); + header[6] = (ushort)GET_INT_HI(dostime); - // CRC - header[7] = (ushort)GET_INT_LO(crc); - header[8] = (ushort)GET_INT_HI(crc); + // CRC + header[7] = (ushort)GET_INT_LO(crc); + header[8] = (ushort)GET_INT_HI(crc); - // Compressed length: - header[9] = (ushort)GET_INT_LO(clen); - header[10] = (ushort)GET_INT_HI(clen); + // Compressed length: + header[9] = (ushort)GET_INT_LO(clen); + header[10] = (ushort)GET_INT_HI(clen); - // Uncompressed length. - header[11] = (ushort)GET_INT_LO(len); - header[12] = (ushort)GET_INT_HI(len); + // Uncompressed length. + header[11] = (ushort)GET_INT_LO(len); + header[12] = (ushort)GET_INT_HI(len); - // Filename length - header[13] = (ushort)SWAP_BYTES(fname_length); - // So called "extra field" length. - header[14] = 0; + // Filename length + header[13] = (ushort)SWAP_BYTES(fname_length); + // So called "extra field" length. + header[14] = 0; - // Write the LOC header to the output file. - write_data(header, (int)sizeof(header)); + // Write the LOC header to the output file. + write_data(header, (int)sizeof(header)); - // Copy the fname to the header. - write_data((char *)fname, (int)fname_length); + // Copy the fname to the header. + write_data((char *)fname, (int)fname_length); } void jar::write_central_directory() { - bytes mc; - mc.set("PACK200"); - - ushort header[11]; - - // Create the End of Central Directory structure. - header[0] = (ushort)SWAP_BYTES(0x4B50); - header[1] = (ushort)SWAP_BYTES(0x0605); - // disk numbers - header[2] = 0; - header[3] = 0; - // Number of entries in central directory. - header[4] = (ushort)SWAP_BYTES(central_directory_count); - header[5] = (ushort)SWAP_BYTES(central_directory_count); - // Size of the central directory} - header[6] = (ushort)GET_INT_LO((int)central_directory.size()); - header[7] = (ushort)GET_INT_HI((int)central_directory.size()); - // Offset of central directory within disk. - header[8] = (ushort)GET_INT_LO(output_file_offset); - header[9] = (ushort)GET_INT_HI(output_file_offset); - // zipfile comment length; - header[10] = (ushort)SWAP_BYTES((int)mc.len); - - // Write the central directory. - write_data(central_directory.b); - - // Write the End of Central Directory structure. - write_data(header, (int)sizeof(header)); - - // Write the comment. - write_data(mc); + bytes mc; + mc.set("PACK200"); + + ushort header[11]; + + // Create the End of Central Directory structure. + header[0] = (ushort)SWAP_BYTES(0x4B50); + header[1] = (ushort)SWAP_BYTES(0x0605); + // disk numbers + header[2] = 0; + header[3] = 0; + // Number of entries in central directory. + header[4] = (ushort)SWAP_BYTES(central_directory_count); + header[5] = (ushort)SWAP_BYTES(central_directory_count); + // Size of the central directory} + header[6] = (ushort)GET_INT_LO((int)central_directory.size()); + header[7] = (ushort)GET_INT_HI((int)central_directory.size()); + // Offset of central directory within disk. + header[8] = (ushort)GET_INT_LO(output_file_offset); + header[9] = (ushort)GET_INT_HI(output_file_offset); + // zipfile comment length; + header[10] = (ushort)SWAP_BYTES((int)mc.len); + + // Write the central directory. + write_data(central_directory.b); + + // Write the End of Central Directory structure. + write_data(header, (int)sizeof(header)); + + // Write the comment. + write_data(mc); } // Public API @@ -244,74 +244,74 @@ void jar::write_central_directory() // Open a Jar file and initialize. void jar::openJarFile(const char *fname) { - if (!jarfp) - { - jarfp = fopen(fname, "wb"); - if (!jarfp) - { - fprintf(stderr, "Error: Could not open jar file: %s\n", fname); - exit(3); // Called only from the native standalone unpacker - } - } + if (!jarfp) + { + jarfp = fopen(fname, "wb"); + if (!jarfp) + { + fprintf(stderr, "Error: Could not open jar file: %s\n", fname); + exit(3); // Called only from the native standalone unpacker + } + } } // Add a ZIP entry and copy the file data void jar::addJarEntry(const char *fname, bool deflate_hint, int modtime, bytes &head, - bytes &tail) + bytes &tail) { - int len = (int)(head.len + tail.len); - int clen = 0; - - uint32_t crc = get_crc32(0, Z_NULL, 0); - if (head.len != 0) - crc = get_crc32(crc, (uchar *)head.ptr, (uint32_t)head.len); - if (tail.len != 0) - crc = get_crc32(crc, (uchar *)tail.ptr, (uint32_t)tail.len); - - bool deflate = (deflate_hint && len > 0); - - if (deflate) - { - if (deflate_bytes(head, tail) == false) - { - deflate = false; - } - } - clen = (int)((deflate) ? deflated.size() : len); - add_to_jar_directory(fname, !deflate, modtime, len, clen, crc); - write_jar_header(fname, !deflate, modtime, len, clen, crc); - - if (deflate) - { - write_data(deflated.b); - } - else - { - write_data(head); - write_data(tail); - } + int len = (int)(head.len + tail.len); + int clen = 0; + + uint32_t crc = get_crc32(0, Z_NULL, 0); + if (head.len != 0) + crc = get_crc32(crc, (uchar *)head.ptr, (uint32_t)head.len); + if (tail.len != 0) + crc = get_crc32(crc, (uchar *)tail.ptr, (uint32_t)tail.len); + + bool deflate = (deflate_hint && len > 0); + + if (deflate) + { + if (deflate_bytes(head, tail) == false) + { + deflate = false; + } + } + clen = (int)((deflate) ? deflated.size() : len); + add_to_jar_directory(fname, !deflate, modtime, len, clen, crc); + write_jar_header(fname, !deflate, modtime, len, clen, crc); + + if (deflate) + { + write_data(deflated.b); + } + else + { + write_data(head); + write_data(tail); + } } // Add a ZIP entry for a directory name no data void jar::addDirectoryToJarFile(const char *dir_name) { - bool store = true; - add_to_jar_directory((const char *)dir_name, store, default_modtime, 0, 0, 0); - write_jar_header((const char *)dir_name, store, default_modtime, 0, 0, 0); + bool store = true; + add_to_jar_directory((const char *)dir_name, store, default_modtime, 0, 0, 0); + write_jar_header((const char *)dir_name, store, default_modtime, 0, 0, 0); } // Write out the central directory and close the jar file. void jar::closeJarFile(bool central) { - if (jarfp) - { - fflush(jarfp); - if (central) - write_central_directory(); - fflush(jarfp); - fclose(jarfp); - } - reset(); + if (jarfp) + { + fflush(jarfp); + if (central) + write_central_directory(); + fflush(jarfp); + fclose(jarfp); + } + reset(); } /* Convert the date y/n/d and time h:m:s to a four byte DOS date and @@ -320,9 +320,9 @@ void jar::closeJarFile(bool central) */ inline uint32_t jar::dostime(int y, int n, int d, int h, int m, int s) { - return y < 1980 ? dostime(1980, 1, 1, 0, 0, 0) - : (((uint32_t)y - 1980) << 25) | ((uint32_t)n << 21) | ((uint32_t)d << 16) | - ((uint32_t)h << 11) | ((uint32_t)m << 5) | ((uint32_t)s >> 1); + return y < 1980 ? dostime(1980, 1, 1, 0, 0, 0) + : (((uint32_t)y - 1980) << 25) | ((uint32_t)n << 21) | ((uint32_t)d << 16) | + ((uint32_t)h << 11) | ((uint32_t)m << 5) | ((uint32_t)s >> 1); } /* #ifdef _REENTRANT // solaris @@ -336,20 +336,20 @@ extern "C" struct tm *gmtime_r(const time_t *, struct tm *); */ uint32_t jar::get_dostime(int modtime) { - // see defines.h - if (modtime != 0 && modtime == modtime_cache) - return dostime_cache; - if (modtime != 0 && default_modtime == 0) - default_modtime = modtime; // catch a reasonable default - time_t t = modtime; - struct tm sbuf; - (void)memset((void *)&sbuf, 0, sizeof(sbuf)); - struct tm *s = gmtime_r(&t, &sbuf); - modtime_cache = modtime; - dostime_cache = - dostime(s->tm_year + 1900, s->tm_mon + 1, s->tm_mday, s->tm_hour, s->tm_min, s->tm_sec); - // printf("modtime %d => %d\n", modtime_cache, dostime_cache); - return dostime_cache; + // see defines.h + if (modtime != 0 && modtime == modtime_cache) + return dostime_cache; + if (modtime != 0 && default_modtime == 0) + default_modtime = modtime; // catch a reasonable default + time_t t = modtime; + struct tm sbuf; + (void)memset((void *)&sbuf, 0, sizeof(sbuf)); + struct tm *s = gmtime_r(&t, &sbuf); + modtime_cache = modtime; + dostime_cache = + dostime(s->tm_year + 1900, s->tm_mon + 1, s->tm_mday, s->tm_hour, s->tm_min, s->tm_sec); + // printf("modtime %d => %d\n", modtime_cache, dostime_cache); + return dostime_cache; } /* Returns true on success, and will set the clen to the compressed @@ -358,232 +358,232 @@ uint32_t jar::get_dostime(int modtime) */ bool jar::deflate_bytes(bytes &head, bytes &tail) { - int len = (int)(head.len + tail.len); - - z_stream zs; - BYTES_OF(zs).clear(); - - // NOTE: the window size should always be -MAX_WBITS normally -15. - // unzip/zipup.c and java/Deflater.c - - int error = - deflateInit2(&zs, Z_BEST_COMPRESSION, Z_DEFLATED, -MAX_WBITS, 8, Z_DEFAULT_STRATEGY); - if (error != Z_OK) - { - /* - switch (error) - { - case Z_MEM_ERROR: - PRINTCR((2, "Error: deflate error : Out of memory \n")); - break; - case Z_STREAM_ERROR: - PRINTCR((2, "Error: deflate error : Invalid compression level \n")); - break; - case Z_VERSION_ERROR: - PRINTCR((2, "Error: deflate error : Invalid version\n")); - break; - default: - PRINTCR((2, "Error: Internal deflate error error = %d\n", error)); - } - */ - return false; - } - - deflated.empty(); - zs.next_out = (uchar *)deflated.grow(len + (len / 2)); - zs.avail_out = (int)deflated.size(); - - zs.next_in = (uchar *)head.ptr; - zs.avail_in = (int)head.len; - - bytes *first = &head; - bytes *last = &tail; - if (last->len == 0) - { - first = nullptr; - last = &head; - } - else if (first->len == 0) - { - first = nullptr; - } - - if (first != nullptr && error == Z_OK) - { - zs.next_in = (uchar *)first->ptr; - zs.avail_in = (int)first->len; - error = deflate(&zs, Z_NO_FLUSH); - } - if (error == Z_OK) - { - zs.next_in = (uchar *)last->ptr; - zs.avail_in = (int)last->len; - error = deflate(&zs, Z_FINISH); - } - if (error == Z_STREAM_END) - { - if (len > (int)zs.total_out) - { - deflated.b.len = zs.total_out; - deflateEnd(&zs); - return true; - } - deflateEnd(&zs); - return false; - } - - deflateEnd(&zs); - return false; + int len = (int)(head.len + tail.len); + + z_stream zs; + BYTES_OF(zs).clear(); + + // NOTE: the window size should always be -MAX_WBITS normally -15. + // unzip/zipup.c and java/Deflater.c + + int error = + deflateInit2(&zs, Z_BEST_COMPRESSION, Z_DEFLATED, -MAX_WBITS, 8, Z_DEFAULT_STRATEGY); + if (error != Z_OK) + { + /* + switch (error) + { + case Z_MEM_ERROR: + PRINTCR((2, "Error: deflate error : Out of memory \n")); + break; + case Z_STREAM_ERROR: + PRINTCR((2, "Error: deflate error : Invalid compression level \n")); + break; + case Z_VERSION_ERROR: + PRINTCR((2, "Error: deflate error : Invalid version\n")); + break; + default: + PRINTCR((2, "Error: Internal deflate error error = %d\n", error)); + } + */ + return false; + } + + deflated.empty(); + zs.next_out = (uchar *)deflated.grow(len + (len / 2)); + zs.avail_out = (int)deflated.size(); + + zs.next_in = (uchar *)head.ptr; + zs.avail_in = (int)head.len; + + bytes *first = &head; + bytes *last = &tail; + if (last->len == 0) + { + first = nullptr; + last = &head; + } + else if (first->len == 0) + { + first = nullptr; + } + + if (first != nullptr && error == Z_OK) + { + zs.next_in = (uchar *)first->ptr; + zs.avail_in = (int)first->len; + error = deflate(&zs, Z_NO_FLUSH); + } + if (error == Z_OK) + { + zs.next_in = (uchar *)last->ptr; + zs.avail_in = (int)last->len; + error = deflate(&zs, Z_FINISH); + } + if (error == Z_STREAM_END) + { + if (len > (int)zs.total_out) + { + deflated.b.len = zs.total_out; + deflateEnd(&zs); + return true; + } + deflateEnd(&zs); + return false; + } + + deflateEnd(&zs); + return false; } // Callback for fetching data from a GZIP input stream static int64_t read_input_via_gzip(unpacker *u, void *buf, int64_t minlen, int64_t maxlen) { - assert(minlen <= maxlen); // don't talk nonsense - int64_t numread = 0; - char *bufptr = (char *)buf; - char *inbuf = u->gzin->inbuf; - size_t inbuflen = sizeof(u->gzin->inbuf); - unpacker::read_input_fn_t read_gzin_fn = (unpacker::read_input_fn_t)u->gzin->read_input_fn; - z_stream &zs = *(z_stream *)u->gzin->zstream; - while (numread < minlen) - { - int readlen = (1 << 16); // pretty arbitrary - if (readlen > (maxlen - numread)) - readlen = (int)(maxlen - numread); - zs.next_out = (uchar *)bufptr; - zs.avail_out = readlen; - if (zs.avail_in == 0) - { - zs.avail_in = (int)read_gzin_fn(u, inbuf, 1, inbuflen); - zs.next_in = (uchar *)inbuf; - } - int error = inflate(&zs, Z_NO_FLUSH); - if (error != Z_OK && error != Z_STREAM_END) - { - unpack_abort("error inflating input"); - break; - } - int nr = readlen - zs.avail_out; - numread += nr; - bufptr += nr; - assert(numread <= maxlen); - if (error == Z_STREAM_END) - { - enum - { - TRAILER_LEN = 8 - }; - // skip 8-byte trailer - if (zs.avail_in >= TRAILER_LEN) - { - zs.avail_in -= TRAILER_LEN; - } - else - { - // Bug: 5023768,we read past the TRAILER_LEN to see if there is - // any extraneous data, as we dont support concatenated .gz - // files just yet. - int extra = (int)read_gzin_fn(u, inbuf, 1, inbuflen); - zs.avail_in += extra - TRAILER_LEN; - } - // %%% should check final CRC and length here - // %%% should check for concatenated *.gz files here - if (zs.avail_in > 0) - unpack_abort("garbage after end of deflated input stream"); - // pop this filter off: - u->gzin->free(); - break; - } - } - - // fprintf(u->errstrm, "readInputFn(%d,%d) => %d (gunzip)\n", - // (int)minlen, (int)maxlen, (int)numread); - return numread; + assert(minlen <= maxlen); // don't talk nonsense + int64_t numread = 0; + char *bufptr = (char *)buf; + char *inbuf = u->gzin->inbuf; + size_t inbuflen = sizeof(u->gzin->inbuf); + read_input_fn_t read_gzin_fn = u->gzin->read_input_fn; + z_stream &zs = *(z_stream *)u->gzin->zstream; + while (numread < minlen) + { + int readlen = (1 << 16); // pretty arbitrary + if (readlen > (maxlen - numread)) + readlen = (int)(maxlen - numread); + zs.next_out = (uchar *)bufptr; + zs.avail_out = readlen; + if (zs.avail_in == 0) + { + zs.avail_in = (int)read_gzin_fn(u, inbuf, 1, inbuflen); + zs.next_in = (uchar *)inbuf; + } + int error = inflate(&zs, Z_NO_FLUSH); + if (error != Z_OK && error != Z_STREAM_END) + { + unpack_abort("error inflating input"); + break; + } + int nr = readlen - zs.avail_out; + numread += nr; + bufptr += nr; + assert(numread <= maxlen); + if (error == Z_STREAM_END) + { + enum + { + TRAILER_LEN = 8 + }; + // skip 8-byte trailer + if (zs.avail_in >= TRAILER_LEN) + { + zs.avail_in -= TRAILER_LEN; + } + else + { + // Bug: 5023768,we read past the TRAILER_LEN to see if there is + // any extraneous data, as we dont support concatenated .gz + // files just yet. + int extra = (int)read_gzin_fn(u, inbuf, 1, inbuflen); + zs.avail_in += extra - TRAILER_LEN; + } + // %%% should check final CRC and length here + // %%% should check for concatenated *.gz files here + if (zs.avail_in > 0) + unpack_abort("garbage after end of deflated input stream"); + // pop this filter off: + u->gzin->free(); + break; + } + } + + // fprintf(u->errstrm, "readInputFn(%d,%d) => %d (gunzip)\n", + // (int)minlen, (int)maxlen, (int)numread); + return numread; } void gunzip::init(unpacker *u_) { - BYTES_OF(*this).clear(); - u = u_; - assert(u->gzin == nullptr); // once only, please - read_input_fn = (void *)u->read_input_fn; - zstream = NEW(z_stream, 1); - u->gzin = this; - u->read_input_fn = read_input_via_gzip; + BYTES_OF(*this).clear(); + u = u_; + assert(u->gzin == nullptr); // once only, please + read_input_fn = u->read_input_fn; + zstream = NEW(z_stream, 1); + u->gzin = this; + u->read_input_fn = read_input_via_gzip; } void gunzip::start(int magic) { - assert((magic & GZIP_MAGIC_MASK) == GZIP_MAGIC); - int gz_flg = (magic & 0xFF); // keep "flg", discard other 3 bytes - enum - { - FHCRC = (1 << 1), - FEXTRA = (1 << 2), - FNAME = (1 << 3), - FCOMMENT = (1 << 4) - }; - char gz_mtime[4]; - char gz_xfl[1]; - char gz_os[1]; - char gz_extra_len[2]; - char gz_hcrc[2]; - char gz_ignore; - // do not save extra, name, comment - read_fixed_field(gz_mtime, sizeof(gz_mtime)); - read_fixed_field(gz_xfl, sizeof(gz_xfl)); - read_fixed_field(gz_os, sizeof(gz_os)); - if (gz_flg & FEXTRA) - { - read_fixed_field(gz_extra_len, sizeof(gz_extra_len)); - int extra_len = gz_extra_len[0] & 0xFF; - extra_len += (gz_extra_len[1] & 0xFF) << 8; - for (; extra_len > 0; extra_len--) - { - read_fixed_field(&gz_ignore, 1); - } - } - int null_terms = 0; - if (gz_flg & FNAME) - null_terms++; - if (gz_flg & FCOMMENT) - null_terms++; - for (; null_terms; null_terms--) - { - for (;;) - { - gz_ignore = 0; - read_fixed_field(&gz_ignore, 1); - if (gz_ignore == 0) - break; - } - } - if (gz_flg & FHCRC) - read_fixed_field(gz_hcrc, sizeof(gz_hcrc)); - - // now the input stream is ready to read into the inflater - int error = inflateInit2((z_stream *)zstream, -MAX_WBITS); - if (error != Z_OK) - { - unpack_abort("cannot create input"); - } + assert((magic & GZIP_MAGIC_MASK) == GZIP_MAGIC); + int gz_flg = (magic & 0xFF); // keep "flg", discard other 3 bytes + enum + { + FHCRC = (1 << 1), + FEXTRA = (1 << 2), + FNAME = (1 << 3), + FCOMMENT = (1 << 4) + }; + char gz_mtime[4]; + char gz_xfl[1]; + char gz_os[1]; + char gz_extra_len[2]; + char gz_hcrc[2]; + char gz_ignore; + // do not save extra, name, comment + read_fixed_field(gz_mtime, sizeof(gz_mtime)); + read_fixed_field(gz_xfl, sizeof(gz_xfl)); + read_fixed_field(gz_os, sizeof(gz_os)); + if (gz_flg & FEXTRA) + { + read_fixed_field(gz_extra_len, sizeof(gz_extra_len)); + int extra_len = gz_extra_len[0] & 0xFF; + extra_len += (gz_extra_len[1] & 0xFF) << 8; + for (; extra_len > 0; extra_len--) + { + read_fixed_field(&gz_ignore, 1); + } + } + int null_terms = 0; + if (gz_flg & FNAME) + null_terms++; + if (gz_flg & FCOMMENT) + null_terms++; + for (; null_terms; null_terms--) + { + for (;;) + { + gz_ignore = 0; + read_fixed_field(&gz_ignore, 1); + if (gz_ignore == 0) + break; + } + } + if (gz_flg & FHCRC) + read_fixed_field(gz_hcrc, sizeof(gz_hcrc)); + + // now the input stream is ready to read into the inflater + int error = inflateInit2((z_stream *)zstream, -MAX_WBITS); + if (error != Z_OK) + { + unpack_abort("cannot create input"); + } } void gunzip::free() { - assert(u->gzin == this); - u->gzin = nullptr; - u->read_input_fn = (unpacker::read_input_fn_t) this->read_input_fn; - inflateEnd((z_stream *)zstream); - ::free(zstream); - zstream = nullptr; - ::free(this); + assert(u->gzin == this); + u->gzin = nullptr; + u->read_input_fn = this->read_input_fn; + inflateEnd((z_stream *)zstream); + ::free(zstream); + zstream = nullptr; + ::free(this); } void gunzip::read_fixed_field(char *buf, size_t buflen) { - int64_t nr = ((unpacker::read_input_fn_t)read_input_fn)(u, buf, buflen, buflen); - if ((size_t)nr != buflen) - unpack_abort("short stream header"); + int64_t nr = read_input_fn(u, buf, buflen, buflen); + if ((size_t)nr != buflen) + unpack_abort("short stream header"); } diff --git a/libraries/pack200/src/zip.h b/libraries/pack200/src/zip.h index 67ec24da..ff973a86 100644 --- a/libraries/pack200/src/zip.h +++ b/libraries/pack200/src/zip.h @@ -22,89 +22,91 @@ * or visit www.oracle.com if you need additional information or have any * questions. */ +#pragma once + #include typedef unsigned short ushort; typedef unsigned int uint32_t; typedef unsigned char uchar; -struct unpacker; +#include "unpack.h" struct jar { - // JAR file writer - FILE *jarfp; - int default_modtime; - - // Used by unix2dostime: - int modtime_cache; - uint32_t dostime_cache; - - // Private members - fillbytes central_directory; - ushort central_directory_count; - uint32_t output_file_offset; - fillbytes deflated; // temporary buffer - - // pointer to outer unpacker, for error checks etc. - unpacker *u; - - // Public Methods - void openJarFile(const char *fname); - void addJarEntry(const char *fname, bool deflate_hint, int modtime, bytes &head, - bytes &tail); - void addDirectoryToJarFile(const char *dir_name); - void closeJarFile(bool central); - - void init(unpacker *u_); - - void free() - { - central_directory.free(); - deflated.free(); - } - - void reset() - { - free(); - init(u); - } - - // Private Methods - void write_data(void *ptr, int len); - void write_data(bytes &b) - { - write_data(b.ptr, (int)b.len); - } - void add_to_jar_directory(const char *fname, bool store, int modtime, int len, int clen, - uint32_t crc); - void write_jar_header(const char *fname, bool store, int modtime, int len, int clen, - unsigned int crc); - void write_central_directory(); - uint32_t dostime(int y, int n, int d, int h, int m, int s); - uint32_t get_dostime(int modtime); - - // The definitions of these depend on the NO_ZLIB option: - bool deflate_bytes(bytes &head, bytes &tail); - static uint32_t get_crc32(uint32_t c, unsigned char *ptr, uint32_t len); + // JAR file writer + FILE *jarfp; + int default_modtime; + + // Used by unix2dostime: + int modtime_cache; + uint32_t dostime_cache; + + // Private members + fillbytes central_directory; + ushort central_directory_count; + uint32_t output_file_offset; + fillbytes deflated; // temporary buffer + + // pointer to outer unpacker, for error checks etc. + unpacker *u; + + // Public Methods + void openJarFile(const char *fname); + void addJarEntry(const char *fname, bool deflate_hint, int modtime, bytes &head, + bytes &tail); + void addDirectoryToJarFile(const char *dir_name); + void closeJarFile(bool central); + + void init(unpacker *u_); + + void free() + { + central_directory.free(); + deflated.free(); + } + + void reset() + { + free(); + init(u); + } + + // Private Methods + void write_data(void *ptr, int len); + void write_data(bytes &b) + { + write_data(b.ptr, (int)b.len); + } + void add_to_jar_directory(const char *fname, bool store, int modtime, int len, int clen, + uint32_t crc); + void write_jar_header(const char *fname, bool store, int modtime, int len, int clen, + unsigned int crc); + void write_central_directory(); + uint32_t dostime(int y, int n, int d, int h, int m, int s); + uint32_t get_dostime(int modtime); + + // The definitions of these depend on the NO_ZLIB option: + bool deflate_bytes(bytes &head, bytes &tail); + static uint32_t get_crc32(uint32_t c, unsigned char *ptr, uint32_t len); }; struct gunzip { - // optional gzip input stream control block + // optional gzip input stream control block - // pointer to outer unpacker, for error checks etc. - unpacker *u; + // pointer to outer unpacker, for error checks etc. + unpacker *u; - void *read_input_fn; // underlying \bchar\b stream - void *zstream; // inflater state - char inbuf[1 << 14]; // input buffer + read_input_fn_t read_input_fn; // underlying \bchar\b stream + void *zstream; // inflater state + char inbuf[1 << 14]; // input buffer - void init(unpacker *u_); // pushes new value on u->read_input_fn + void init(unpacker *u_); // pushes new value on u->read_input_fn - void free(); + void free(); - void start(int magic); + void start(int magic); - // private stuff - void read_fixed_field(char *buf, size_t buflen); + // private stuff + void read_fixed_field(char *buf, size_t buflen); }; -- cgit v1.2.3