1 files changed, 59 insertions, 3 deletions

diff --git a/third_party/aom/av1/encoder/hash.c b/third_party/aom/av1/encoder/hash.c
index 89c5bd8a3..180115d9f 100644
--- a/third_party/aom/av1/encoder/hash.c
+++ b/third_party/aom/av1/encoder/hash.c
@@ -22,7 +22,7 @@ static void crc_calculator_process_data(CRC_CALCULATOR *p_crc_calculator,
   }
 }
 
-void crc_calculator_reset(CRC_CALCULATOR *p_crc_calculator) {
+static void crc_calculator_reset(CRC_CALCULATOR *p_crc_calculator) {
   p_crc_calculator->remainder = 0;
 }
 
@@ -61,9 +61,65 @@ void av1_crc_calculator_init(CRC_CALCULATOR *p_crc_calculator, uint32_t bits,
   crc_calculator_init_table(p_crc_calculator);
 }
 
-uint32_t av1_get_crc_value(CRC_CALCULATOR *p_crc_calculator, uint8_t *p,
-                           int length) {
+uint32_t av1_get_crc_value(void *crc_calculator, uint8_t *p, int length) {
+  CRC_CALCULATOR *p_crc_calculator = (CRC_CALCULATOR *)crc_calculator;
   crc_calculator_reset(p_crc_calculator);
   crc_calculator_process_data(p_crc_calculator, p, length);
   return crc_calculator_get_crc(p_crc_calculator);
 }
+
+/* CRC-32C (iSCSI) polynomial in reversed bit order. */
+#define POLY 0x82f63b78
+
+/* Construct table for software CRC-32C calculation. */
+void av1_crc32c_calculator_init(CRC32C *p_crc32c) {
+  uint32_t crc;
+
+  for (int n = 0; n < 256; n++) {
+    crc = n;
+    crc = crc & 1 ? (crc >> 1) ^ POLY : crc >> 1;
+    crc = crc & 1 ? (crc >> 1) ^ POLY : crc >> 1;
+    crc = crc & 1 ? (crc >> 1) ^ POLY : crc >> 1;
+    crc = crc & 1 ? (crc >> 1) ^ POLY : crc >> 1;
+    crc = crc & 1 ? (crc >> 1) ^ POLY : crc >> 1;
+    crc = crc & 1 ? (crc >> 1) ^ POLY : crc >> 1;
+    crc = crc & 1 ? (crc >> 1) ^ POLY : crc >> 1;
+    crc = crc & 1 ? (crc >> 1) ^ POLY : crc >> 1;
+    p_crc32c->table[0][n] = crc;
+  }
+  for (int n = 0; n < 256; n++) {
+    crc = p_crc32c->table[0][n];
+    for (int k = 1; k < 8; k++) {
+      crc = p_crc32c->table[0][crc & 0xff] ^ (crc >> 8);
+      p_crc32c->table[k][n] = crc;
+    }
+  }
+}
+
+/* Table-driven software version as a fall-back.  This is about 15 times slower
+ than using the hardware instructions.  This assumes little-endian integers,
+ as is the case on Intel processors that the assembler code here is for. */
+uint32_t av1_get_crc32c_value_c(CRC32C *p, uint8_t *buf, size_t len) {
+  const uint8_t *next = (const uint8_t *)(buf);
+  uint64_t crc;
+
+  crc = 0 ^ 0xffffffff;
+  while (len && ((uintptr_t)next & 7) != 0) {
+    crc = p->table[0][(crc ^ *next++) & 0xff] ^ (crc >> 8);
+    len--;
+  }
+  while (len >= 8) {
+    crc ^= *(uint64_t *)next;
+    crc = p->table[7][crc & 0xff] ^ p->table[6][(crc >> 8) & 0xff] ^
+          p->table[5][(crc >> 16) & 0xff] ^ p->table[4][(crc >> 24) & 0xff] ^
+          p->table[3][(crc >> 32) & 0xff] ^ p->table[2][(crc >> 40) & 0xff] ^
+          p->table[1][(crc >> 48) & 0xff] ^ p->table[0][crc >> 56];
+    next += 8;
+    len -= 8;
+  }
+  while (len) {
+    crc = p->table[0][(crc ^ *next++) & 0xff] ^ (crc >> 8);
+    len--;
+  }
+  return (uint32_t)crc ^ 0xffffffff;
+}