Re: [Maria-developers] [Commits] f28af02: Removed HW acceleration for big endian checksum
Looks OK to me. This can only slightly punish the very few people who are upgrading from a buggy MySQL 5.6 or MariaDB 10.0 or 10.1 installation on a big-endian machine (such as SPARC) into a newer version. And those users could invoke innochecksum to rewrite the page checksums. On Thu, Nov 23, 2017 at 12:57 PM, Sergey Vojtovich <svoj@mariadb.org> wrote:
revision-id: f28af027a0439d760c6d26eb67a932e97eb665d7 (mariadb-10.3.2-66-gf28af02) parent(s): b2571e534f9eed873cc9b1ce5f48b109c1ee0fcc committer: Sergey Vojtovich timestamp: 2017-11-23 14:53:16 +0400 message:
Removed HW acceleration for big endian checksum
Big endian checksum is needed to workaround 10+ years old bug, where checksum was calculated incorrectly on big endian hardware. We can still checksum such tablespaces using software implementation of CRC32.
--- storage/innobase/include/ut0crc32.h | 4 +- storage/innobase/ut/ut0crc32.cc | 89 +----------------------------- ------- 2 files changed, 3 insertions(+), 90 deletions(-)
diff --git a/storage/innobase/include/ut0crc32.h b/storage/innobase/include/ut0crc32.h index 509ac09..527a3e81 100644 --- a/storage/innobase/include/ut0crc32.h +++ b/storage/innobase/include/ut0crc32.h @@ -47,9 +47,9 @@ typedef uint32_t (*ut_crc32_func_t)(const byte* ptr, ulint len); /** Pointer to CRC32 calculation function. */ extern ut_crc32_func_t ut_crc32;
-/** Pointer to CRC32 calculation function, which uses big-endian byte order +/** CRC32 calculation function, which uses big-endian byte order when converting byte strings to integers internally. */ -extern ut_crc32_func_t ut_crc32_legacy_big_endian; +extern uint32_t ut_crc32_legacy_big_endian(const byte* buf, ulint len);
extern const char* ut_crc32_implementation;
diff --git a/storage/innobase/ut/ut0crc32.cc b/storage/innobase/ut/ ut0crc32.cc index 6383e6d..36c52b7 100644 --- a/storage/innobase/ut/ut0crc32.cc +++ b/storage/innobase/ut/ut0crc32.cc @@ -89,10 +89,6 @@ mysys/my_perf.c, contributed by Facebook under the following license. /** Pointer to CRC32 calculation function. */ ut_crc32_func_t ut_crc32;
-/** Pointer to CRC32 calculation function, which uses big-endian byte order -when converting byte strings to integers internally. */ -ut_crc32_func_t ut_crc32_legacy_big_endian; - /** Text description of CRC32 implementation */ const char* ut_crc32_implementation;
@@ -238,37 +234,6 @@ ut_crc32_64_hw( *len -= 8; }
-/** Calculate CRC32 over 64-bit byte string using a hardware/CPU instruction. -The byte string is converted to a 64-bit integer using big endian byte order. -@param[in,out] crc crc32 checksum so far when this function is called, -when the function ends it will contain the new checksum -@param[in,out] data data to be checksummed, the pointer will be advanced -with 8 bytes -@param[in,out] len remaining bytes, it will be decremented with 8 */ -inline -void -ut_crc32_64_legacy_big_endian_hw( - uint32_t* crc, - const byte** data, - ulint* len) -{ - uint64_t data_int = *reinterpret_cast<const uint64_t*>(*data); - -#ifndef WORDS_BIGENDIAN - data_int = ut_crc32_swap_byteorder(data_int); -#else - /* Currently we only support x86_64 (little endian) CPUs. In case - some big endian CPU supports a CRC32 instruction, then maybe we will - NOT need a byte order swap here. */ -#error Dont know how to handle big endian CPUs -#endif /* WORDS_BIGENDIAN */ - - *crc = ut_crc32_64_low_hw(*crc, data_int); - - *data += 8; - *len -= 8; -} - /** Calculates CRC32 using hardware/CPU instructions. @param[in] buf data over which to calculate CRC32 @param[in] len data length @@ -355,56 +320,6 @@ ut_crc32_hw(
return(~crc); } - -/** Calculates CRC32 using hardware/CPU instructions. -This function uses big endian byte ordering when converting byte sequence to -integers. -@param[in] buf data over which to calculate CRC32 -@param[in] len data length -@return CRC-32C (polynomial 0x11EDC6F41) */ -uint32_t -ut_crc32_legacy_big_endian_hw( - const byte* buf, - ulint len) -{ - uint32_t crc = 0xFFFFFFFFU; - - /* Calculate byte-by-byte up to an 8-byte aligned address. After - this consume the input 8-bytes at a time. */ - while (len > 0 && (reinterpret_cast<uintptr_t>(buf) & 7) != 0) { - ut_crc32_8_hw(&crc, &buf, &len); - } - - while (len >= 128) { - /* This call is repeated 16 times. 16 * 8 = 128. */ - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - } - - while (len >= 8) { - ut_crc32_64_legacy_big_endian_hw(&crc, &buf, &len); - } - - while (len > 0) { - ut_crc32_8_hw(&crc, &buf, &len); - } - - return(~crc); -} #endif /* defined(__GNUC__) && defined(__x86_64__) */
/* CRC32 software implementation. */ @@ -600,7 +515,7 @@ integers. @param[in] len data length @return CRC-32C (polynomial 0x11EDC6F41) */ uint32_t -ut_crc32_legacy_big_endian_sw( +ut_crc32_legacy_big_endian( const byte* buf, ulint len) { @@ -654,7 +569,6 @@ ut_crc32_init() { ut_crc32_slice8_table_init(); ut_crc32 = ut_crc32_sw; - ut_crc32_legacy_big_endian = ut_crc32_legacy_big_endian_sw; ut_crc32_implementation = "Using generic crc32 instructions";
#if defined(__GNUC__) && defined(__x86_64__) @@ -687,7 +601,6 @@ ut_crc32_init()
if (features_ecx & 1 << 20) { ut_crc32 = ut_crc32_hw; - ut_crc32_legacy_big_endian = ut_crc32_legacy_big_endian_hw; ut_crc32_implementation = "Using SSE2 crc32 instructions"; }
_______________________________________________ commits mailing list commits@mariadb.org https://lists.askmonty.org/cgi-bin/mailman/listinfo/commits
-- Marko Mäkelä, Lead Developer InnoDB MariaDB Corporation DON’T MISS M|18 MariaDB User Conference February 26 - 27, 2018 New York City https://m18.mariadb.com/
participants (1)
-
Marko Mäkelä