glibc/sysdeps/unix/sysv/linux/aarch64/cpu-features.c
Naohiro Tamura fa527f345c aarch64: Added optimized memcpy and memmove for A64FX
This patch optimizes the performance of memcpy/memmove for A64FX [1]
which implements ARMv8-A SVE and has L1 64KB cache per core and L2 8MB
cache per NUMA node.

The performance optimization makes use of Scalable Vector Register
with several techniques such as loop unrolling, memory access
alignment, cache zero fill, and software pipelining.

SVE assembler code for memcpy/memmove is implemented as Vector Length
Agnostic code so theoretically it can be run on any SOC which supports
ARMv8-A SVE standard.

We confirmed that all testcases have been passed by running 'make
check' and 'make xcheck' not only on A64FX but also on ThunderX2.

And also we confirmed that the SVE 512 bit vector register performance
is roughly 4 times better than Advanced SIMD 128 bit register and 8
times better than scalar 64 bit register by running 'make bench'.

[1] https://github.com/fujitsu/A64FX

Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
Reviewed-by: Szabolcs Nagy <Szabolcs.Nagy@arm.com>
2021-05-27 09:47:53 +01:00

124 lines
3.7 KiB
C

/* Initialize CPU feature data. AArch64 version.
This file is part of the GNU C Library.
Copyright (C) 2017-2021 Free Software Foundation, Inc.
The GNU C Library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
The GNU C Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with the GNU C Library; if not, see
<https://www.gnu.org/licenses/>. */
#include <cpu-features.h>
#include <sys/auxv.h>
#include <elf/dl-hwcaps.h>
#include <sys/prctl.h>
#define DCZID_DZP_MASK (1 << 4)
#define DCZID_BS_MASK (0xf)
/* The maximal set of permitted tags that the MTE random tag generation
instruction may use. We exclude tag 0 because a) we want to reserve
that for the libc heap structures and b) because it makes it easier
to see when pointer have been correctly tagged. */
#define MTE_ALLOWED_TAGS (0xfffe << PR_MTE_TAG_SHIFT)
#if HAVE_TUNABLES
struct cpu_list
{
const char *name;
uint64_t midr;
};
static struct cpu_list cpu_list[] = {
{"falkor", 0x510FC000},
{"thunderxt88", 0x430F0A10},
{"thunderx2t99", 0x431F0AF0},
{"thunderx2t99p1", 0x420F5160},
{"phecda", 0x680F0000},
{"ares", 0x411FD0C0},
{"emag", 0x503F0001},
{"kunpeng920", 0x481FD010},
{"a64fx", 0x460F0010},
{"generic", 0x0}
};
static uint64_t
get_midr_from_mcpu (const char *mcpu)
{
for (int i = 0; i < sizeof (cpu_list) / sizeof (struct cpu_list); i++)
if (strcmp (mcpu, cpu_list[i].name) == 0)
return cpu_list[i].midr;
return UINT64_MAX;
}
#endif
static inline void
init_cpu_features (struct cpu_features *cpu_features)
{
register uint64_t midr = UINT64_MAX;
#if HAVE_TUNABLES
/* Get the tunable override. */
const char *mcpu = TUNABLE_GET (glibc, cpu, name, const char *, NULL);
if (mcpu != NULL)
midr = get_midr_from_mcpu (mcpu);
#endif
/* If there was no useful tunable override, query the MIDR if the kernel
allows it. */
if (midr == UINT64_MAX)
{
if (GLRO (dl_hwcap) & HWCAP_CPUID)
asm volatile ("mrs %0, midr_el1" : "=r"(midr));
else
midr = 0;
}
cpu_features->midr_el1 = midr;
/* Check if ZVA is enabled. */
unsigned dczid;
asm volatile ("mrs %0, dczid_el0" : "=r"(dczid));
if ((dczid & DCZID_DZP_MASK) == 0)
cpu_features->zva_size = 4 << (dczid & DCZID_BS_MASK);
/* Check if BTI is supported. */
cpu_features->bti = GLRO (dl_hwcap2) & HWCAP2_BTI;
/* Setup memory tagging support if the HW and kernel support it, and if
the user has requested it. */
cpu_features->mte_state = 0;
#ifdef USE_MTAG
# if HAVE_TUNABLES
int mte_state = TUNABLE_GET (glibc, mem, tagging, unsigned, 0);
cpu_features->mte_state = (GLRO (dl_hwcap2) & HWCAP2_MTE) ? mte_state : 0;
/* If we lack the MTE feature, disable the tunable, since it will
otherwise cause instructions that won't run on this CPU to be used. */
TUNABLE_SET (glibc, mem, tagging, cpu_features->mte_state);
# endif
if (cpu_features->mte_state & 2)
__prctl (PR_SET_TAGGED_ADDR_CTRL,
(PR_TAGGED_ADDR_ENABLE | PR_MTE_TCF_SYNC | MTE_ALLOWED_TAGS),
0, 0, 0);
else if (cpu_features->mte_state)
__prctl (PR_SET_TAGGED_ADDR_CTRL,
(PR_TAGGED_ADDR_ENABLE | PR_MTE_TCF_ASYNC | MTE_ALLOWED_TAGS),
0, 0, 0);
#endif
/* Check if SVE is supported. */
cpu_features->sve = GLRO (dl_hwcap) & HWCAP_SVE;
}