[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20240222183937.981D71D9@davehans-spike.ostc.intel.com>
Date: Thu, 22 Feb 2024 10:39:37 -0800
From: Dave Hansen <dave.hansen@...ux.intel.com>
To: linux-kernel@...r.kernel.org
Cc: kirill.shutemov@...ux.intel.com,pbonzini@...hat.com,tglx@...utronix.de,x86@...nel.org,bp@...en8.de,Dave Hansen <dave.hansen@...ux.intel.com>,mpatocka@...hat.com,snitzer@...hat.com,daniel@...ll.ch
Subject: [RFC][PATCH 08/34] x86/cpu: Add CLFLUSH size helper
From: Dave Hansen <dave.hansen@...ux.intel.com>
This is probably starting to look familiar: The size of a cacheline is
fundamental system-wide information. Keeping it per-cpu is just silly.
Introduce a system-wide helper for looking up the cacheline size and
use it.
This does one slightly odd looking thing, it stops setting
c->x86_clflush_size on all but the boot CPU. This is functionally OK
because all readers of the secondary CPU values also go away. It also
makes it explicit that the 'boot_cpu_data' is the one true system-wide
value.
Signed-off-by: Dave Hansen <dave.hansen@...ux.intel.com>
Cc: Mikulas Patocka <mpatocka@...hat.com>
Cc: Mike Snitzer <snitzer@...hat.com>
Cc: Daniel Vetter <daniel@...ll.ch>
---
b/arch/x86/include/asm/processor.h | 5 +++++
b/arch/x86/kernel/cpu/centaur.c | 2 +-
b/arch/x86/kernel/cpu/common.c | 9 +++++----
b/arch/x86/kernel/cpu/intel.c | 2 +-
b/arch/x86/kernel/cpu/proc.c | 2 +-
b/arch/x86/lib/usercopy_64.c | 7 +++----
b/arch/x86/mm/pat/set_memory.c | 2 +-
b/arch/x86/pci/common.c | 2 +-
b/drivers/gpu/drm/drm_cache.c | 4 ++--
b/drivers/gpu/drm/i915/i915_cmd_parser.c | 3 +--
b/drivers/gpu/drm/i915/i915_gem.c | 2 +-
b/drivers/md/dm-writecache.c | 2 +-
12 files changed, 23 insertions(+), 19 deletions(-)
diff -puN arch/x86/include/asm/processor.h~x86_clflush_size-func arch/x86/include/asm/processor.h
--- a/arch/x86/include/asm/processor.h~x86_clflush_size-func 2024-02-22 10:08:52.112596720 -0800
+++ b/arch/x86/include/asm/processor.h 2024-02-22 10:08:52.132597505 -0800
@@ -777,4 +777,9 @@ static inline u8 x86_virt_bits(void)
return boot_cpu_data.x86_virt_bits;
}
+static inline u8 x86_clflush_size(void)
+{
+ return boot_cpu_data.x86_clflush_size;
+}
+
#endif /* _ASM_X86_PROCESSOR_H */
diff -puN arch/x86/kernel/cpu/centaur.c~x86_clflush_size-func arch/x86/kernel/cpu/centaur.c
--- a/arch/x86/kernel/cpu/centaur.c~x86_clflush_size-func 2024-02-22 10:08:52.112596720 -0800
+++ b/arch/x86/kernel/cpu/centaur.c 2024-02-22 10:08:52.132597505 -0800
@@ -62,7 +62,7 @@ static void init_c3(struct cpuinfo_x86 *
set_cpu_cap(c, X86_FEATURE_3DNOW);
#endif
if (c->x86 == 0x6 && c->x86_model >= 0xf) {
- c->x86_cache_alignment = c->x86_clflush_size * 2;
+ c->x86_cache_alignment = x86_clflush_size() * 2;
set_cpu_cap(c, X86_FEATURE_REP_GOOD);
}
diff -puN arch/x86/kernel/cpu/common.c~x86_clflush_size-func arch/x86/kernel/cpu/common.c
--- a/arch/x86/kernel/cpu/common.c~x86_clflush_size-func 2024-02-22 10:08:52.112596720 -0800
+++ b/arch/x86/kernel/cpu/common.c 2024-02-22 10:08:52.132597505 -0800
@@ -954,8 +954,9 @@ void cpu_detect(struct cpuinfo_x86 *c)
c->x86_stepping = x86_stepping(tfms);
if (cap0 & (1<<19)) {
- c->x86_clflush_size = ((misc >> 8) & 0xff) * 8;
- c->x86_cache_alignment = c->x86_clflush_size;
+ if (c == &boot_cpu_data)
+ c->x86_clflush_size = ((misc >> 8) & 0xff) * 8;
+ c->x86_cache_alignment = x86_clflush_size();
}
}
}
@@ -1123,7 +1124,7 @@ void get_cpu_address_sizes(struct cpuinf
}
}
c->x86_cache_bits = c->x86_phys_bits;
- c->x86_cache_alignment = c->x86_clflush_size;
+ c->x86_cache_alignment = x86_clflush_size();
}
static void identify_cpu_without_cpuid(struct cpuinfo_x86 *c)
@@ -1831,7 +1832,7 @@ static void identify_cpu(struct cpuinfo_
c->x86_phys_bits = 32;
c->x86_virt_bits = 32;
#endif
- c->x86_cache_alignment = c->x86_clflush_size;
+ c->x86_cache_alignment = x86_clflush_size();
memset(&c->x86_capability, 0, sizeof(c->x86_capability));
#ifdef CONFIG_X86_VMX_FEATURE_NAMES
memset(&c->vmx_capability, 0, sizeof(c->vmx_capability));
diff -puN arch/x86/kernel/cpu/intel.c~x86_clflush_size-func arch/x86/kernel/cpu/intel.c
--- a/arch/x86/kernel/cpu/intel.c~x86_clflush_size-func 2024-02-22 10:08:52.116596877 -0800
+++ b/arch/x86/kernel/cpu/intel.c 2024-02-22 10:08:52.132597505 -0800
@@ -653,7 +653,7 @@ static void init_intel(struct cpuinfo_x8
#ifdef CONFIG_X86_64
if (c->x86 == 15)
- c->x86_cache_alignment = c->x86_clflush_size * 2;
+ c->x86_cache_alignment = x86_clflush_size() * 2;
if (c->x86 == 6)
set_cpu_cap(c, X86_FEATURE_REP_GOOD);
#else
diff -puN arch/x86/kernel/cpu/proc.c~x86_clflush_size-func arch/x86/kernel/cpu/proc.c
--- a/arch/x86/kernel/cpu/proc.c~x86_clflush_size-func 2024-02-22 10:08:52.116596877 -0800
+++ b/arch/x86/kernel/cpu/proc.c 2024-02-22 10:08:52.132597505 -0800
@@ -130,7 +130,7 @@ static int show_cpuinfo(struct seq_file
if (c->x86_tlbsize > 0)
seq_printf(m, "TLB size\t: %d 4K pages\n", c->x86_tlbsize);
#endif
- seq_printf(m, "clflush size\t: %u\n", c->x86_clflush_size);
+ seq_printf(m, "clflush size\t: %u\n", x86_clflush_size());
seq_printf(m, "cache_alignment\t: %d\n", c->x86_cache_alignment);
seq_printf(m, "address sizes\t: %u bits physical, %u bits virtual\n",
x86_phys_bits(), x86_virt_bits());
diff -puN arch/x86/lib/usercopy_64.c~x86_clflush_size-func arch/x86/lib/usercopy_64.c
--- a/arch/x86/lib/usercopy_64.c~x86_clflush_size-func 2024-02-22 10:08:52.120597034 -0800
+++ b/arch/x86/lib/usercopy_64.c 2024-02-22 10:08:52.132597505 -0800
@@ -27,13 +27,12 @@
*/
static void clean_cache_range(void *addr, size_t size)
{
- u16 x86_clflush_size = boot_cpu_data.x86_clflush_size;
- unsigned long clflush_mask = x86_clflush_size - 1;
+ unsigned long clflush_mask = x86_clflush_size() - 1;
void *vend = addr + size;
void *p;
for (p = (void *)((unsigned long)addr & ~clflush_mask);
- p < vend; p += x86_clflush_size)
+ p < vend; p += x86_clflush_size())
clwb(p);
}
@@ -65,7 +64,7 @@ long __copy_user_flushcache(void *dst, c
clean_cache_range(dst, size);
} else {
if (!IS_ALIGNED(dest, 8)) {
- dest = ALIGN(dest, boot_cpu_data.x86_clflush_size);
+ dest = ALIGN(dest, x86_clflush_size());
clean_cache_range(dst, 1);
}
diff -puN arch/x86/mm/pat/set_memory.c~x86_clflush_size-func arch/x86/mm/pat/set_memory.c
--- a/arch/x86/mm/pat/set_memory.c~x86_clflush_size-func 2024-02-22 10:08:52.120597034 -0800
+++ b/arch/x86/mm/pat/set_memory.c 2024-02-22 10:08:52.132597505 -0800
@@ -314,7 +314,7 @@ static unsigned long __cpa_addr(struct c
static void clflush_cache_range_opt(void *vaddr, unsigned int size)
{
- const unsigned long clflush_size = boot_cpu_data.x86_clflush_size;
+ const unsigned long clflush_size = x86_clflush_size();
void *p = (void *)((unsigned long)vaddr & ~(clflush_size - 1));
void *vend = vaddr + size;
diff -puN arch/x86/pci/common.c~x86_clflush_size-func arch/x86/pci/common.c
--- a/arch/x86/pci/common.c~x86_clflush_size-func 2024-02-22 10:08:52.120597034 -0800
+++ b/arch/x86/pci/common.c 2024-02-22 10:08:52.132597505 -0800
@@ -480,7 +480,7 @@ void pcibios_scan_root(int busnum)
void __init pcibios_set_cache_line_size(void)
{
- pci_dfl_cache_line_size = boot_cpu_data.x86_clflush_size >> 2;
+ pci_dfl_cache_line_size = x86_clflush_size() >> 2;
printk(KERN_DEBUG "PCI: pci_cache_line_size set to %d bytes\n",
pci_dfl_cache_line_size << 2);
}
diff -puN drivers/gpu/drm/drm_cache.c~x86_clflush_size-func drivers/gpu/drm/drm_cache.c
--- a/drivers/gpu/drm/drm_cache.c~x86_clflush_size-func 2024-02-22 10:08:52.124597191 -0800
+++ b/drivers/gpu/drm/drm_cache.c 2024-02-22 10:08:52.132597505 -0800
@@ -52,7 +52,7 @@ drm_clflush_page(struct page *page)
{
uint8_t *page_virtual;
unsigned int i;
- const int size = boot_cpu_data.x86_clflush_size;
+ const int size = x86_clflush_size();
if (unlikely(page == NULL))
return;
@@ -160,7 +160,7 @@ drm_clflush_virt_range(void *addr, unsig
{
#if defined(CONFIG_X86)
if (static_cpu_has(X86_FEATURE_CLFLUSH)) {
- const int size = boot_cpu_data.x86_clflush_size;
+ const int size = x86_clflush_size();
void *end = addr + length;
addr = (void *)(((unsigned long)addr) & -size);
diff -puN drivers/gpu/drm/i915/i915_cmd_parser.c~x86_clflush_size-func drivers/gpu/drm/i915/i915_cmd_parser.c
--- a/drivers/gpu/drm/i915/i915_cmd_parser.c~x86_clflush_size-func 2024-02-22 10:08:52.124597191 -0800
+++ b/drivers/gpu/drm/i915/i915_cmd_parser.c 2024-02-22 10:08:52.132597505 -0800
@@ -1203,8 +1203,7 @@ static u32 *copy_batch(struct drm_i915_g
*/
remain = length;
if (dst_needs_clflush & CLFLUSH_BEFORE)
- remain = round_up(remain,
- boot_cpu_data.x86_clflush_size);
+ remain = round_up(remain, x86_clflush_size());
ptr = dst;
x = offset_in_page(offset);
diff -puN drivers/gpu/drm/i915/i915_gem.c~x86_clflush_size-func drivers/gpu/drm/i915/i915_gem.c
--- a/drivers/gpu/drm/i915/i915_gem.c~x86_clflush_size-func 2024-02-22 10:08:52.124597191 -0800
+++ b/drivers/gpu/drm/i915/i915_gem.c 2024-02-22 10:08:52.132597505 -0800
@@ -696,7 +696,7 @@ i915_gem_shmem_pwrite(struct drm_i915_ge
*/
partial_cacheline_write = 0;
if (needs_clflush & CLFLUSH_BEFORE)
- partial_cacheline_write = boot_cpu_data.x86_clflush_size - 1;
+ partial_cacheline_write = x86_clflush_size() - 1;
user_data = u64_to_user_ptr(args->data_ptr);
remain = args->size;
diff -puN drivers/md/dm-writecache.c~x86_clflush_size-func drivers/md/dm-writecache.c
--- a/drivers/md/dm-writecache.c~x86_clflush_size-func 2024-02-22 10:08:52.128597348 -0800
+++ b/drivers/md/dm-writecache.c 2024-02-22 10:08:52.132597505 -0800
@@ -1229,7 +1229,7 @@ static void memcpy_flushcache_optimized(
*/
#ifdef CONFIG_X86
if (static_cpu_has(X86_FEATURE_CLFLUSHOPT) &&
- likely(boot_cpu_data.x86_clflush_size == 64) &&
+ likely(x86_clflush_size() == 64) &&
likely(size >= 768)) {
do {
memcpy((void *)dest, (void *)source, 64);
_
Powered by blists - more mailing lists