upstream/postgresql
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2026-06-11 01:30:11 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Refactor detection of x86 ZMM registers

Browse source 
- Call _xgetbv within x86_set_runtime_features rather than in a
  separate function
- Use symbols for XCR mask bits rather than a magic constant

A future commit will build on this to detect YMM registers without
code duplication.

Reviewed-by: Zsolt Parragi <zsolt.parragi@percona.com>
Discussion: https://postgr.es/m/CANWCAZbgEUFw7LuYSVeJ=Tj98R5HoOB1Ffeqk3aLvbw5rU5NTw@mail.gmail.com
This commit is contained in:
John Naylor 2026-02-28 16:28:09 +07:00
parent 3f98862980
commit 51bb4a58ed
1 changed files with 27 additions and 16 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

43
src/port/pg_cpu_x86.c
View file

@ -31,32 +31,34 @@
#include "port/pg_cpu.h"
/*
* XSAVE state component bits that we need
*
* https://www.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-vol-1-manual.pdf
* Chapter "MANAGING STATE USING THE XSAVE FEATURE SET"
*/
#define XMM (1<<1)
#define YMM (1<<2)
#define OPMASK (1<<5)
#define ZMM0_15 (1<<6)
#define ZMM16_31 (1<<7)
/* array indexed by enum X86FeatureId */
bool X86Features[X86FeaturesSize] = {0};
/*
* Does XGETBV say the ZMM registers are enabled?
*
* NB: Caller is responsible for verifying that osxsave is available
* before calling this.
*/
#ifdef HAVE_XSAVE_INTRINSICS
pg_attribute_target("xsave")
#endif
static bool
zmm_regs_available(void)
mask_available(uint32 value, uint32 mask)
{
#ifdef HAVE_XSAVE_INTRINSICS
return (_xgetbv(0) & 0xe6) == 0xe6;
#else
return false;
#endif
return (value & mask) == mask;
}
/*
* Parse the CPU ID info for runtime checks.
*/
#ifdef HAVE_XSAVE_INTRINSICS
pg_attribute_target("xsave")
#endif
void
set_x86_features(void)
{
@ -76,6 +78,8 @@ set_x86_features(void)
/* All these features depend on OSXSAVE */
if (exx[2] & (1 << 27))
{
uint32 xcr0_val = 0;
/* second cpuid call on leaf 7 to check extended AVX-512 support */
memset(exx, 0, 4 * sizeof(exx[0]));
@ -86,7 +90,14 @@ set_x86_features(void)
__cpuidex(exx, 7, 0);
#endif
if (zmm_regs_available())
#ifdef HAVE_XSAVE_INTRINSICS
/* get value of Extended Control Register */
xcr0_val = _xgetbv(0);
#endif
/* Are ZMM registers enabled? */
if (mask_available(xcr0_val, XMM | YMM |
OPMASK | ZMM0_15 | ZMM16_31))
{
X86Features[PG_AVX512_BW] = exx[1] >> 30 & 1;
X86Features[PG_AVX512_VL] = exx[1] >> 31 & 1;