1
0
mirror of https://github.com/postgres/postgres.git synced 2025-12-09 02:08:45 +03:00

Use ARMv8 CRC instructions where available.

ARMv8 introduced special CPU instructions for calculating CRC-32C. Use
them, when available, for speed.

Like with the similar Intel CRC instructions, several factors affect
whether the instructions can be used. The compiler intrinsics for them must
be supported by the compiler, and the instructions must be supported by the
target architecture. If the compilation target architecture does not
support the instructions, but adding "-march=armv8-a+crc" makes them
available, then we compile the code with a runtime check to determine if
the host we're running on supports them or not.

For the runtime check, use glibc getauxval() function. Unfortunately,
that's not very portable, but I couldn't find any more portable way to do
it. If getauxval() is not available, the CRC instructions will still be
used if the target architecture supports them without any additional
compiler flags, but the runtime check will not be available.

Original patch by Yuqi Gu, heavily modified by me. Reviewed by Andres
Freund, Thomas Munro.

Discussion: https://www.postgresql.org/message-id/HE1PR0801MB1323D171938EABC04FFE7FA9E3110%40HE1PR0801MB1323.eurprd08.prod.outlook.com
This commit is contained in:
Heikki Linnakangas
2018-04-04 12:22:45 +03:00
parent 638a199fa9
commit f044d71e33
11 changed files with 456 additions and 41 deletions

View File

@@ -0,0 +1,72 @@
/*-------------------------------------------------------------------------
*
* pg_crc32c_armv8.c
* Compute CRC-32C checksum using ARMv8 CRC Extension instructions
*
* Portions Copyright (c) 1996-2018, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
*
* IDENTIFICATION
* src/port/pg_crc32c_armv8.c
*
*-------------------------------------------------------------------------
*/
#include "c.h"
#include "port/pg_crc32c.h"
#include <arm_acle.h>
pg_crc32c
pg_comp_crc32c_armv8(pg_crc32c crc, const void *data, size_t len)
{
const unsigned char *p = data;
const unsigned char *pend = p + len;
/*
* ARMv8 doesn't require alignment, but aligned memory access is
* significantly faster. Process leading bytes so that the loop below
* starts with a pointer aligned to eight bytes.
*/
if (!PointerIsAligned(p, uint16) && p < pend)
{
crc = __crc32cb(crc, *p);
p += 1;
}
if (!PointerIsAligned(p, uint32) && p < pend)
{
crc = __crc32ch(crc, *(uint16 *) p);
p += 2;
}
if (!PointerIsAligned(p, uint64) && p < pend)
{
crc = __crc32cw(crc, *(uint32 *) p);
p += 4;
}
/* Process eight bytes at a time, as far as we can. */
while (p + 8 <= pend)
{
crc = __crc32cd(crc, *(uint64 *) p);
p += 8;
}
/* Process remaining 0-7 bytes. */
if (p + 4 <= pend)
{
crc = __crc32cw(crc, *(uint32 *) p);
p += 4;
}
if (p + 2 <= pend)
{
crc = __crc32ch(crc, *(uint16 *) p);
p += 2;
}
if (p < pend)
{
crc = __crc32cb(crc, *p);
}
return crc;
}