Skip to contentSkip to conversation

TLB flush optimization

PATCH Superseded Alex Shi 8 participants 33 messages May 8, 2012 – May 10, 2012

Subsystems (16)
Memory management x86 MM core Tracing +12 more subsystems

v3 of 10 Superseded Latest is v10 →

lore
Share & tools
lore

Alex Shi

author
#0 lore raw

PATCH v3 1/1 TLB flush optimization

This version just added 2 patches, the 6th and 7th. The first 5 patches are same as v2.

Seems it is a little hard for me to find a real case to show performance gain. But the advantage is there ture.

Any comments are appreciated!

Alex

[PATCH v3 1/7] x86/tlb_info: get last level TLB entry number of CPU [PATCH v3 2/7] x86/flush_tlb: try flush_tlb_single one by one in [PATCH v3 3/7] x86/tlb: fall back to flush all when meet a THP large [PATCH v3 4/7] x86/tlb: add tlb flush all factor for specific CPUs [PATCH v3 5/7] x86/tlb: remove comments for tlb_flush_range [PATCH v3 6/7] x86/tlb: optimizing flush_tlb_mm
[PATCH v3 7/7] x86/tlb: add tlb_flushall_factor into sysfs for user

author
#1 lore raw

PATCH v3 1/7 x86/tlb_info: get last level TLB entry number of CPU

For 4KB pages, x86 CPU has 2 or 1 level TLB, first level is data TLB and instruction TLB, second level is shared TLB for both data and instructions.

For hupe page TLB, usually there is just one level and seperated by 2MB/4MB and 1GB.

Although each levels TLB size is important for performance tuning, but for genernal and rude optimizing, last level TLB entry number is suitable. And in fact, last level TLB always has the biggest entry number.

This patch will get the biggest TLB entry number and use it in furture TLB optimizing.

For all kinds of x86 vendor friendly, vendor specific code was moved to its specific files.

---
 arch/x86/include/asm/processor.h |   11 +++
 arch/x86/kernel/cpu/common.c     |   21 ++++++
 arch/x86/kernel/cpu/cpu.h        |    9 +++
 arch/x86/kernel/cpu/intel.c      |  141 ++++++++++++++++++++++++++++++++++++++
 4 files changed, 182 insertions(+), 0 deletions(-)
arch/x86/include/asm/processor.h +11 -0
index 4fa7dcc..797faca 100644
@@ -61,6 +61,17 @@ static inline void *current_text_addr(void)
 # define ARCH_MIN_MMSTRUCT_ALIGN	0
 #endif
 
+enum tlb_infos {
+	ENTRIES,
+	NR_INFO
+};
+
+extern u16 __read_mostly tlb_lli_4k[NR_INFO];
+extern u16 __read_mostly tlb_lli_2m[NR_INFO];
+extern u16 __read_mostly tlb_lli_4m[NR_INFO];
+extern u16 __read_mostly tlb_lld_4k[NR_INFO];
+extern u16 __read_mostly tlb_lld_2m[NR_INFO];
+extern u16 __read_mostly tlb_lld_4m[NR_INFO];
 /*
  *  CPU type and hardware bug flags. Kept separately for each CPU.
  *  Members of this structure are referenced in head.S, so think twice
arch/x86/kernel/cpu/common.c +21 -0
index cf79302..0152082 100644
@@ -452,6 +452,25 @@ void __cpuinit cpu_detect_cache_sizes(struct cpuinfo_x86 *c)
 	c->x86_cache_size = l2size;
 }
 
+u16 __read_mostly tlb_lli_4k[NR_INFO];
+u16 __read_mostly tlb_lli_2m[NR_INFO];
+u16 __read_mostly tlb_lli_4m[NR_INFO];
+u16 __read_mostly tlb_lld_4k[NR_INFO];
+u16 __read_mostly tlb_lld_2m[NR_INFO];
+u16 __read_mostly tlb_lld_4m[NR_INFO];
+
+void __cpuinit cpu_detect_tlb(struct cpuinfo_x86 *c)
+{
+	if (c->x86_vendor == X86_VENDOR_INTEL)
+		intel_cpu_detect_tlb(c);
+
+	printk(KERN_INFO "Last level iTLB entries: 4KB %d, 2MB %d, 4MB %d\n" \
+		"Last level dTLB entries: 4KB %d, 2MB %d, 4MB %d\n",
+		tlb_lli_4k[ENTRIES], tlb_lli_2m[ENTRIES],
+		tlb_lli_4m[ENTRIES], tlb_lld_4k[ENTRIES],
+		tlb_lld_2m[ENTRIES], tlb_lld_4m[ENTRIES]);
+}
+
 void __cpuinit detect_ht(struct cpuinfo_x86 *c)
 {
 #ifdef CONFIG_X86_HT
@@ -911,6 +930,8 @@ void __init identify_boot_cpu(void)
 #else
 	vgetcpu_set_mode();
 #endif
+	if (boot_cpu_data.cpuid_level >= 2)
+		cpu_detect_tlb(&boot_cpu_data);
 }
 
 void __cpuinit identify_secondary_cpu(struct cpuinfo_x86 *c)
arch/x86/kernel/cpu/cpu.h +9 -0
index 8bacc78..c8dc726 100644
@@ -24,6 +24,14 @@ struct cpu_dev {
 	int		c_x86_vendor;
 };
 
+struct _tlb_table {
+	unsigned char descriptor;
+	char tlb_type;
+	unsigned int entries;
+	/* unsigned int ways; */
+	char info[128];
+};
+
 #define cpu_dev_register(cpu_devX) \
 	static const struct cpu_dev *const __cpu_dev_##cpu_devX __used \
 	__attribute__((__section__(".x86_cpu_dev.init"))) = \
@@ -34,4 +42,5 @@ extern const struct cpu_dev *const __x86_cpu_dev_start[],
 
 extern void get_cpu_cap(struct cpuinfo_x86 *c);
 extern void cpu_detect_cache_sizes(struct cpuinfo_x86 *c);
+extern void intel_cpu_detect_tlb(struct cpuinfo_x86 *c);
 #endif /* ARCH_X86_CPU_H */
arch/x86/kernel/cpu/intel.c +141 -0

147 lines not on this page · raw message

Signature · 1 line
1.7.5.4
author
#2 lore raw

PATCH v3 2/7 x86/flush_tlb: try flush_tlb_single one by one in flush_tlb_range

In reply to Alex Shi

x86 has no flush_tlb_range support in instruction level. Currently the flush_tlb_range just implemented by flushing all page table. That is not the best solution for all scenarios. In fact, if we just use 'invlpg' to flush few lines from TLB, we can get the performance gain from later remain TLB lines accessing.

But the 'invlpg' instruction costs much of time. Its execution time can compete with cr3 rewriting, and even a bit more on SNB CPU.

So, on a 512 4KB TLB entries CPU, the balance points is at:

	(512 - X) * 100ns(assumed TLB refill cost) =
		X(TLB flush entries) * 100ns(assumed invlpg cost)

Here, X is 256, that is 1/2 of 512 entries.

But with the mysterious CPU pre-fetcher and page miss handler Unit, the assumed TLB refill cost is far lower then 100ns in sequential access. And 2 HT siblings in one core makes the memory access more faster if they are accessing the same memory. So, in the patch, I just do the change when the target entries is less than 1/16 of whole active tlb entries. Actually, I have no data support for the percentage '1/16', so any suggestions are welcomed.

As to hugetlb, guess due to smaller page table, and smaller active TLB entries, I didn't see benefit via my benchmark, so no optimizing now.

My macro benchmark show in ideal scenarios, the performance improves 70 percent in reading. And in worst scenario, the reading/writing performance is similar with unpatched 3.4-rc4 kernel.

Here is the reading data on my 2P * 4cores *HT NHM EP machine, with THP 'always':

multi thread testing, '-t' paramter is thread number:
with patch unpatched 3.4-rc4

./mprotect -t 1           14ns		24ns
./mprotect -t 2           13ns		22ns
./mprotect -t 4           12ns		19ns
./mprotect -t 8           14ns		16ns
./mprotect -t 16          28ns		26ns
./mprotect -t 32          54ns		51ns
./mprotect -t 128         200ns		199ns

Single process with sequencial flushing and memory accessing:

with patch unpatched 3.4-rc4
./mprotect 7ns 11ns
./mprotect -p 4096 -l 8 -n 10240

			    21ns		21ns

I also tried other benchmarks on Intel core2/NHM/SNB EP and NHM EX machine. No clear performance change on specjbb2005 with openjdk, and oltp reading.

---
 arch/x86/include/asm/paravirt.h       |    5 +-
 arch/x86/include/asm/paravirt_types.h |    3 +-
 arch/x86/include/asm/tlbflush.h       |   19 +++----
 arch/x86/include/asm/uv/uv.h          |    5 +-
 arch/x86/mm/tlb.c                     |   97 +++++++++++++++++++++++++++------
 arch/x86/platform/uv/tlb_uv.c         |    6 +-
 arch/x86/xen/mmu.c                    |    9 ++--
 include/trace/events/xen.h            |   12 +++--
 8 files changed, 113 insertions(+), 43 deletions(-)
arch/x86/include/asm/paravirt.h +3 -2
index aa0f913..03da4ab 100644
@@ -397,9 +397,10 @@ static inline void __flush_tlb_single(unsigned long addr)
 
 static inline void flush_tlb_others(const struct cpumask *cpumask,
 				    struct mm_struct *mm,
-				    unsigned long va)
+				    unsigned long start,
+				    unsigned long end)
 {
-	PVOP_VCALL3(pv_mmu_ops.flush_tlb_others, cpumask, mm, va);
+	PVOP_VCALL4(pv_mmu_ops.flush_tlb_others, cpumask, mm, start, end);
 }
 
 static inline int paravirt_pgd_alloc(struct mm_struct *mm)
arch/x86/include/asm/paravirt_types.h +2 -1
index 8e8b9a4..600a5fcac9 100644
@@ -250,7 +250,8 @@ struct pv_mmu_ops {
 	void (*flush_tlb_single)(unsigned long addr);
 	void (*flush_tlb_others)(const struct cpumask *cpus,
 				 struct mm_struct *mm,
-				 unsigned long va);
+				 unsigned long start,
+				 unsigned long end);
 
 	/* Hooks for allocating and freeing a pagetable top-level */
 	int  (*pgd_alloc)(struct mm_struct *mm);
arch/x86/include/asm/tlbflush.h +9 -10
index c0e108e..ec30dfb 100644
@@ -77,7 +77,7 @@ static inline void __flush_tlb_one(unsigned long addr)
  *  - flush_tlb_page(vma, vmaddr) flushes one page
  *  - flush_tlb_range(vma, start, end) flushes a range of pages
  *  - flush_tlb_kernel_range(start, end) flushes a range of kernel pages
- *  - flush_tlb_others(cpumask, mm, va) flushes TLBs on other cpus
+ *  - flush_tlb_others(cpumask, mm, start, end) flushes TLBs on other cpus
  *
  * ..but the i386 has somewhat limited tlb flushing capabilities,
  * and page-granular flushes are available only on i486 and up.
@@ -115,7 +115,8 @@ static inline void flush_tlb_range(struct vm_area_struct *vma,
 
 static inline void native_flush_tlb_others(const struct cpumask *cpumask,
 					   struct mm_struct *mm,
-					   unsigned long va)
+					   unsigned long start,
+					   unsigned long end)
 {
 }
 
@@ -133,17 +134,14 @@ extern void flush_tlb_all(void);
 extern void flush_tlb_current_task(void);
 extern void flush_tlb_mm(struct mm_struct *);
 extern void flush_tlb_page(struct vm_area_struct *, unsigned long);
+extern void flush_tlb_range(struct vm_area_struct *vma,
+				   unsigned long start, unsigned long end);
 
 #define flush_tlb()	flush_tlb_current_task()
 
-static inline void flush_tlb_range(struct vm_area_struct *vma,
-				   unsigned long start, unsigned long end)
-{
-	flush_tlb_mm(vma->vm_mm);
-}
-
 void native_flush_tlb_others(const struct cpumask *cpumask,
-			     struct mm_struct *mm, unsigned long va);
+				struct mm_struct *mm,
+				unsigned long start, unsigned long end);
 
 #define TLBSTATE_OK	1
 #define TLBSTATE_LAZY	2
@@ -163,7 +161,8 @@ static inline void reset_lazy_tlbstate(void)
 #endif	/* SMP */
 
 #ifndef CONFIG_PARAVIRT
-#define flush_tlb_others(mask, mm, va)	native_flush_tlb_others(mask, mm, va)
+#define flush_tlb_others(mask, mm, start, end)	\
+	native_flush_tlb_others(mask, mm, start, end)
 #endif
 
 static inline void flush_tlb_kernel_range(unsigned long start,
arch/x86/include/asm/uv/uv.h +3 -2
index 3bb9491..b47c2a8 100644
@@ -15,7 +15,8 @@ extern void uv_nmi_init(void);
 extern void uv_system_init(void);
 extern const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
 						 struct mm_struct *mm,
-						 unsigned long va,
+						 unsigned long start,
+						 unsigned end,
 						 unsigned int cpu);
 
 #else	/* X86_UV */
@@ -26,7 +27,7 @@ static inline void uv_cpu_init(void)	{ }
 static inline void uv_system_init(void)	{ }
 static inline const struct cpumask *
 uv_flush_tlb_others(const struct cpumask *cpumask, struct mm_struct *mm,
-		    unsigned long va, unsigned int cpu)
+		    unsigned long start, unsigned long end, unsigned int cpu)
 { return cpumask; }
 
 #endif	/* X86_UV */
arch/x86/mm/tlb.c +81 -16

168 lines not on this page · raw message

arch/x86/platform/uv/tlb_uv.c +3 -3
index 3ae0e61..0df5ad2 100644
@@ -1068,8 +1068,8 @@ static int set_distrib_bits(struct cpumask *flush_mask, struct bau_control *bcp,
  * done.  The returned pointer is valid till preemption is re-enabled.
  */
 const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
-				struct mm_struct *mm, unsigned long va,
-				unsigned int cpu)
+				struct mm_struct *mm, unsigned long start,
+				unsigned end, unsigned int cpu)
 {
 	int locals = 0;
 	int remotes = 0;
@@ -1112,7 +1112,7 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask,
 
 	record_send_statistics(stat, locals, hubs, remotes, bau_desc);
 
-	bau_desc->payload.address = va;
+	bau_desc->payload.address = start;
 	bau_desc->payload.sending_cpu = cpu;
 	/*
 	 * uv_flush_send_and_wait returns 0 if all cpu's were messaged,
arch/x86/xen/mmu.c +5 -4
index b8e2794..75bab52 100644
@@ -1239,7 +1239,8 @@ static void xen_flush_tlb_single(unsigned long addr)
 }
 
 static void xen_flush_tlb_others(const struct cpumask *cpus,
-				 struct mm_struct *mm, unsigned long va)
+				 struct mm_struct *mm, unsigned long start,
+				 unsigned long end)
 {
 	struct {
 		struct mmuext_op op;
@@ -1251,7 +1252,7 @@ static void xen_flush_tlb_others(const struct cpumask *cpus,
 	} *args;
 	struct multicall_space mcs;
 
-	trace_xen_mmu_flush_tlb_others(cpus, mm, va);
+	trace_xen_mmu_flush_tlb_others(cpus, mm, start, end);
 
 	if (cpumask_empty(cpus))
 		return;		/* nothing to do */
@@ -1264,11 +1265,11 @@ static void xen_flush_tlb_others(const struct cpumask *cpus,
 	cpumask_and(to_cpumask(args->mask), cpus, cpu_online_mask);
 	cpumask_clear_cpu(smp_processor_id(), to_cpumask(args->mask));
 
-	if (va == TLB_FLUSH_ALL) {
+	if (start == TLB_FLUSH_ALL) {
 		args->op.cmd = MMUEXT_TLB_FLUSH_MULTI;
 	} else {
 		args->op.cmd = MMUEXT_INVLPG_MULTI;
-		args->op.arg1.linear_addr = va;
+		args->op.arg1.linear_addr = start;
 	}
 
 	MULTI_mmuext_op(mcs.mc, &args->op, 1, NULL, DOMID_SELF);
include/trace/events/xen.h +7 -5
index 92f1a79..15ba03b 100644
@@ -397,18 +397,20 @@ TRACE_EVENT(xen_mmu_flush_tlb_single,
 
 TRACE_EVENT(xen_mmu_flush_tlb_others,
 	    TP_PROTO(const struct cpumask *cpus, struct mm_struct *mm,
-		     unsigned long addr),
-	    TP_ARGS(cpus, mm, addr),
+		     unsigned long addr, unsigned long end),
+	    TP_ARGS(cpus, mm, addr, end),
 	    TP_STRUCT__entry(
 		    __field(unsigned, ncpus)
 		    __field(struct mm_struct *, mm)
 		    __field(unsigned long, addr)
+		    __field(unsigned long, end)
 		    ),
 	    TP_fast_assign(__entry->ncpus = cpumask_weight(cpus);
 			   __entry->mm = mm;
-			   __entry->addr = addr),
-	    TP_printk("ncpus %d mm %p addr %lx",
-		      __entry->ncpus, __entry->mm, __entry->addr)
+			   __entry->addr = addr,
+			   __entry->end = end),
+	    TP_printk("ncpus %d mm %p addr %lx, end %lx",
+		      __entry->ncpus, __entry->mm, __entry->addr, __entry->end)
 	);
 
 TRACE_EVENT(xen_mmu_write_cr3,
Signature · 1 line
1.7.5.4
author
#3 lore raw

PATCH v3 3/7 x86/tlb: fall back to flush all when meet a THP large page

In reply to Alex Shi

We don't need to flush large pages by PAGE_SIZE step, that just waste time. and actually, large page don't need 'invlpg' optimizing according to our macro benchmark. So, just flush whole TLB is enough for them.

The following result is tested on a 2CPU * 4cores * 2HT NHM EP machine, with THP 'always' setting.

Multi-thread testing, '-t' paramter is thread number:
without this patch with this patch

./mprotect -t 1         14ns                       13ns
./mprotect -t 2         13ns                       13ns
./mprotect -t 4         12ns                       11ns
./mprotect -t 8         14ns                       10ns
./mprotect -t 16        28ns                       28ns
./mprotect -t 32        54ns                       52ns
./mprotect -t 128       200ns                      200ns
---
 arch/x86/mm/tlb.c |   27 +++++++++++++++++++++++++++
 1 files changed, 27 insertions(+), 0 deletions(-)
arch/x86/mm/tlb.c +27 -0
index c4e694d..4f709e6 100644
@@ -316,12 +316,35 @@ void flush_tlb_mm(struct mm_struct *mm)
 
 #define FLUSHALL_BAR	16
 
+static inline int has_large_page(struct mm_struct *mm,
+				 unsigned long start, unsigned long end)
+{
+	pgd_t *pgd;
+	pud_t *pud;
+	pmd_t *pmd;
+	unsigned long addr;
+	for (addr = start; addr <= end; addr += HPAGE_SIZE) {
+		pgd = pgd_offset(mm, addr);
+		if (likely(!pgd_none(*pgd))) {
+			pud = pud_offset(pgd, addr);
+			if (likely(!pud_none(*pud))) {
+				pmd = pmd_offset(pud, addr);
+				if (likely(!pmd_none(*pmd)))
+					if (pmd_large(*pmd))
+						return 1;
+			}
+		}
+	}
+	return 0;
+}
+
 void flush_tlb_range(struct vm_area_struct *vma,
 				   unsigned long start, unsigned long end)
 {
 	struct mm_struct *mm;
 
 	if (!cpu_has_invlpg || vma->vm_flags & VM_HUGETLB) {
+flush_all:
 		flush_tlb_mm(vma->vm_mm);
 		return;
 	}
@@ -344,6 +367,10 @@ void flush_tlb_range(struct vm_area_struct *vma,
 			if ((end - start)/PAGE_SIZE > act_entries/FLUSHALL_BAR)
 				local_flush_tlb();
 			else {
+				if (has_large_page(mm, start, end)) {
+					preempt_enable();
+					goto flush_all;
+				}
 				for (addr = start; addr <= end;
 						addr += PAGE_SIZE)
 					__flush_tlb_single(addr);
Signature · 1 line
1.7.5.4
author
#4 lore raw

PATCH v3 4/7 x86/tlb: add tlb flush all factor for specific CPUs

In reply to Alex Shi

Testing show different CPU type(micro architectures and NUMA mode) has different balance points between the TLB flush all and multiple invlpg. And there maybe has cases the tlb flush change has no any help.

This patch give a interface to let x86 vendor developers have a chance to set different factors for different CPU type.

like some machine in my hands, balance points is 16 entries on Romely-EP; while it is at 8 entries on Bloomfield NHM-EP; but on model 15 core2 Xeon using invlpg has nothing help.

For untested machine, no optimization now.

---
 arch/x86/include/asm/processor.h |    2 ++
 arch/x86/kernel/cpu/common.c     |   13 +++++++++++--
 arch/x86/kernel/cpu/intel.c      |   30 ++++++++++++++++++++++++++++++
 arch/x86/mm/tlb.c                |    8 ++++----
 4 files changed, 47 insertions(+), 6 deletions(-)
arch/x86/include/asm/processor.h +2 -0
index 797faca..6a7e9c3 100644
@@ -72,6 +72,8 @@ extern u16 __read_mostly tlb_lli_4m[NR_INFO];
 extern u16 __read_mostly tlb_lld_4k[NR_INFO];
 extern u16 __read_mostly tlb_lld_2m[NR_INFO];
 extern u16 __read_mostly tlb_lld_4m[NR_INFO];
+extern u16 __read_mostly tlb_flushall_factor;
+
 /*
  *  CPU type and hardware bug flags. Kept separately for each CPU.
  *  Members of this structure are referenced in head.S, so think twice
arch/x86/kernel/cpu/common.c +11 -2
index 0152082..3e2e310 100644
@@ -459,16 +459,25 @@ u16 __read_mostly tlb_lld_4k[NR_INFO];
 u16 __read_mostly tlb_lld_2m[NR_INFO];
 u16 __read_mostly tlb_lld_4m[NR_INFO];
 
+/*
+ * tlb_flushall_factor shows the balance point in replacing cr3 write with
+ * multiple 'invlpg'. Different CPU type has different this value, which
+ * get from a macro benchmark named mproctect.c that published in lkml.
+ */
+u16 __read_mostly tlb_flushall_factor;
+
 void __cpuinit cpu_detect_tlb(struct cpuinfo_x86 *c)
 {
 	if (c->x86_vendor == X86_VENDOR_INTEL)
 		intel_cpu_detect_tlb(c);
 
 	printk(KERN_INFO "Last level iTLB entries: 4KB %d, 2MB %d, 4MB %d\n" \
-		"Last level dTLB entries: 4KB %d, 2MB %d, 4MB %d\n",
+		"Last level dTLB entries: 4KB %d, 2MB %d, 4MB %d\n"	     \
+		"tlb_flushall_factor is 1/%d\n",
 		tlb_lli_4k[ENTRIES], tlb_lli_2m[ENTRIES],
 		tlb_lli_4m[ENTRIES], tlb_lld_4k[ENTRIES],
-		tlb_lld_2m[ENTRIES], tlb_lld_4m[ENTRIES]);
+		tlb_lld_2m[ENTRIES], tlb_lld_4m[ENTRIES],
+		tlb_flushall_factor);
 }
 
 void __cpuinit detect_ht(struct cpuinfo_x86 *c)
arch/x86/kernel/cpu/intel.c +30 -0
index 86e6131..42480c6 100644
@@ -610,6 +610,35 @@ void intel_tlb_lookup(const unsigned char desc)
 	}
 }
 
+void intel_tlb_flushall_factor_set(struct cpuinfo_x86 *c)
+{
+	switch (c->x86_model) {
+	case 15: /* original 65 nm celeron/pentium/core2/xeon, "Merom"/"Conroe" */
+		tlb_flushall_factor = 0;
+		break;
+	case 26: /* 45 nm nehalem, "Bloomfield" */
+	case 30: /* 45 nm nehalem, "Lynnfield" */
+	case 37: /* 32 nm nehalem, "Clarkdale" */
+	case 44: /* 32 nm nehalem, "Gulftown" */
+	case 46: /* 45 nm nehalem-ex, "Beckton" */
+		tlb_flushall_factor = 64;
+		break;
+	case 42: /* SandyBridge */
+	case 45: /* SandyBridge, "Romely-EP" */
+		tlb_flushall_factor = 32;
+		break;
+	case 28: /* Atom */
+	case 47: /* 32 nm Xeon E7 */
+	case 14: /* 65 nm core solo/duo, "Yonah" */
+	case 22: /* single-core 65 nm celeron/core2solo "Merom-L"/"Conroe-L" */
+	case 23: /* current 45 nm celeron/core2/xeon "Penryn"/"Wolfdale" */
+	case 29: /* six-core 45 nm xeon "Dunnington" */
+
+	default:
+		tlb_flushall_factor = 0;
+	}
+}
+
 void intel_cpu_detect_tlb(struct cpuinfo_x86 *c)
 {
 	int i, j, n;
@@ -630,6 +659,7 @@ void intel_cpu_detect_tlb(struct cpuinfo_x86 *c)
 		for (j = 1 ; j < 16 ; j++)
 			intel_tlb_lookup(desc[j]);
 	}
+	intel_tlb_flushall_factor_set(c);
 }
 
 static const struct cpu_dev __cpuinitconst intel_cpu_dev = {
arch/x86/mm/tlb.c +4 -4
index 4f709e6..91896dc 100644
@@ -314,8 +314,6 @@ void flush_tlb_mm(struct mm_struct *mm)
 	preempt_enable();
 }
 
-#define FLUSHALL_BAR	16
-
 static inline int has_large_page(struct mm_struct *mm,
 				 unsigned long start, unsigned long end)
 {
@@ -343,7 +341,8 @@ void flush_tlb_range(struct vm_area_struct *vma,
 {
 	struct mm_struct *mm;
 
-	if (!cpu_has_invlpg || vma->vm_flags & VM_HUGETLB) {
+	if (!cpu_has_invlpg || vma->vm_flags & VM_HUGETLB
+				|| !tlb_flushall_factor) {
 flush_all:
 		flush_tlb_mm(vma->vm_mm);
 		return;
@@ -364,7 +363,8 @@ flush_all:
 			act_entries = tlb_entries > mm->total_vm ?
 					mm->total_vm : tlb_entries;
 
-			if ((end - start)/PAGE_SIZE > act_entries/FLUSHALL_BAR)
+			if ((end - start)/PAGE_SIZE >
+					act_entries/tlb_flushall_factor)
 				local_flush_tlb();
 			else {
 				if (has_large_page(mm, start, end)) {
Signature · 1 line
1.7.5.4
#5 lore raw

Re: [PATCH v3 4/7] x86/tlb: add tlb flush all factor for specific CPUs

arch/x86/kernel/cpu/intel.c void intel_tlb_lookup(const unsigned char desc) · line 618

Quoted 6 lines · Alex Shi
arch/x86/kernel/cpu/intel.c +6 -0
+void intel_tlb_flushall_factor_set(struct cpuinfo_x86 *c)
+{
+       switch (c->x86_model) {
+       case 15: /* original 65 nm celeron/pentium/core2/xeon, "Merom"/"Conroe" */
+               tlb_flushall_factor = 0;
+               break;

Why isn't this is the bottom list of core chips?

arch/x86/kernel/cpu/intel.c void intel_tlb_lookup(const unsigned char desc) · line 631

Quoted 13 lines
arch/x86/kernel/cpu/intel.c +13 -0
+       case 26: /* 45 nm nehalem, "Bloomfield" */
+       case 30: /* 45 nm nehalem, "Lynnfield" */
+       case 37: /* 32 nm nehalem, "Clarkdale" */
+       case 44: /* 32 nm nehalem, "Gulftown" */
+       case 46: /* 45 nm nehalem-ex, "Beckton" */
+               tlb_flushall_factor = 64;
+               break;
+       case 42: /* SandyBridge */
+       case 45: /* SandyBridge, "Romely-EP" */
+               tlb_flushall_factor = 32;
+               break;
+       case 28: /* Atom */
+       case 47: /* 32 nm Xeon E7 */

This is a wsm-ex, right? Why isn't it listed with the other nehalems?

arch/x86/kernel/cpu/intel.c void intel_tlb_lookup(const unsigned char desc) · line 635

Quoted 4 lines
arch/x86/kernel/cpu/intel.c +4 -0
+       case 14: /* 65 nm core solo/duo, "Yonah" */
+       case 22: /* single-core 65 nm celeron/core2solo "Merom-L"/"Conroe-L" */
+       case 23: /* current 45 nm celeron/core2/xeon "Penryn"/"Wolfdale" */
+       case 29: /* six-core 45 nm xeon "Dunnington" */

So never use invlpg for Atom/Core/Core2?

arch/x86/kernel/cpu/intel.c flush_all: · line 367

Quoted 12 lines
arch/x86/kernel/cpu/intel.c +6 -0
+       default:
+               tlb_flushall_factor = 0;
+       }
+}
@@ -364,7 +363,8 @@ flush_all:
                        act_entries = tlb_entries > mm->total_vm ?
                                        mm->total_vm : tlb_entries;
 
+                       if ((end - start)/PAGE_SIZE >
+                                       act_entries/tlb_flushall_factor) 

You're doing an actual full division, wouldn't a shift be better?

author
#6 lore raw

Re: [PATCH v3 4/7] x86/tlb: add tlb flush all factor for specific CPUs

Quoted 9 lines · Peter Zijlstra

arch/x86/kernel/cpu/intel.c void intel_tlb_lookup(const unsigned char desc) · line 618

Quoted 6 lines · Alex Shi
arch/x86/kernel/cpu/intel.c +6 -0
+void intel_tlb_flushall_factor_set(struct cpuinfo_x86 *c)
+{
+       switch (c->x86_model) {
+       case 15: /* original 65 nm celeron/pentium/core2/xeon, "Merom"/"Conroe" */
+               tlb_flushall_factor = 0;
+               break;

Why isn't this is the bottom list of core chips?

It was tested. but the bottom list cpu was not tested.

Quoted 15 lines
Quoted 4 lines
arch/x86/kernel/cpu/intel.c +3 -0
+       case 26: /* 45 nm nehalem, "Bloomfield" */
+       case 30: /* 45 nm nehalem, "Lynnfield" */
+       case 37: /* 32 nm nehalem, "Clarkdale" */

arch/x86/kernel/cpu/intel.c void intel_tlb_lookup(const unsigned char desc) · line 631

Quoted 10 lines
arch/x86/kernel/cpu/intel.c +10 -0
+       case 44: /* 32 nm nehalem, "Gulftown" */
+       case 46: /* 45 nm nehalem-ex, "Beckton" */
+               tlb_flushall_factor = 64;
+               break;
+       case 42: /* SandyBridge */
+       case 45: /* SandyBridge, "Romely-EP" */
+               tlb_flushall_factor = 32;
+               break;
+       case 28: /* Atom */
+       case 47: /* 32 nm Xeon E7 */
Quoted 1 line

This is a wsm-ex, right? Why isn't it listed with the other nehalems?

I don't know this. Thanks for this info

Quoted 7 lines

arch/x86/kernel/cpu/intel.c void intel_tlb_lookup(const unsigned char desc) · line 635

Quoted 5 lines
arch/x86/kernel/cpu/intel.c +4 -0
+       case 14: /* 65 nm core solo/duo, "Yonah" */
+       case 22: /* single-core 65 nm celeron/core2solo "Merom-L"/"Conroe-L" */
+       case 23: /* current 45 nm celeron/core2/xeon "Penryn"/"Wolfdale" */
+       case 29: /* six-core 45 nm xeon "Dunnington" */

So never use invlpg for Atom/Core/Core2?

Uh, I will remove the CPU list if they weren't tested.

Quoted 7 lines
Quoted 5 lines
arch/x86/kernel/cpu/intel.c +4 -0
+       default:
+               tlb_flushall_factor = 0;
+       }
+}
Quoted 8 lines

arch/x86/mm/tlb.c flush_all: · line 367

Quoted 6 lines
arch/x86/mm/tlb.c +2 -0
@@ -364,7 +363,8 @@ flush_all:
                        act_entries = tlb_entries > mm->total_vm ?
                                        mm->total_vm : tlb_entries;
 
+                       if ((end - start)/PAGE_SIZE >
+                                       act_entries/tlb_flushall_factor) 

You're doing an actual full division, wouldn't a shift be better?

Thanks!

#7 lore raw

Re: [PATCH v3 4/7] x86/tlb: add tlb flush all factor for specific CPUs

In reply to Alex Shi
Quoted 25 lines · Alex Shi
arch/x86/kernel/cpu/intel.c +19 -0
@@ -610,6 +610,35 @@ void intel_tlb_lookup(const unsigned char desc)
 	}
 }
 
+void intel_tlb_flushall_factor_set(struct cpuinfo_x86 *c)
+{
+	switch (c->x86_model) {
+	case 15: /* original 65 nm celeron/pentium/core2/xeon, "Merom"/"Conroe" */
+		tlb_flushall_factor = 0;
+		break;
+	case 26: /* 45 nm nehalem, "Bloomfield" */
+	case 30: /* 45 nm nehalem, "Lynnfield" */
+	case 37: /* 32 nm nehalem, "Clarkdale" */
+	case 44: /* 32 nm nehalem, "Gulftown" */
+	case 46: /* 45 nm nehalem-ex, "Beckton" */
+		tlb_flushall_factor = 64;
+		break;
+	case 42: /* SandyBridge */
+	case 45: /* SandyBridge, "Romely-EP" */
+		tlb_flushall_factor = 32;
+		break;
+	case 28: /* Atom */
+	case 47: /* 32 nm Xeon E7 */

arch/x86/kernel/cpu/intel.c void intel_tlb_lookup(const unsigned char desc) · line 641

Quoted 10 lines
arch/x86/kernel/cpu/intel.c +10 -0
+	case 14: /* 65 nm core solo/duo, "Yonah" */
+	case 22: /* single-core 65 nm celeron/core2solo "Merom-L"/"Conroe-L" */
+	case 23: /* current 45 nm celeron/core2/xeon "Penryn"/"Wolfdale" */
+	case 29: /* six-core 45 nm xeon "Dunnington" */
+
+	default:
+		tlb_flushall_factor = 0;
+	}
+}
+

This uses x86_model without qualifying it x86 (family), however that is meaningless. All the CPUs you are dealing with above have c->x86 == 6, but you need to handle others correctly (even if that just means defaulting it to zero.)

One way to do that is to do:

switch ((c->x86 << 8) + c->x86_model) {

... and use numbers like 0x62d instead of 45.

-hpa

Signature · 2 lines
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.
author
#8 lore raw

Re: [PATCH v3 4/7] x86/tlb: add tlb flush all factor for specific CPUs

Quoted 7 lines

This uses x86_model without qualifying it x86 (family), however that is meaningless. All the CPUs you are dealing with above have c->x86 == 6, but you need to handle others correctly (even if that just means defaulting it to zero.)

Quoted 5 lines

One way to do that is to do:

switch ((c->x86 << 8) + c->x86_model) {

... and use numbers like 0x62d instead of 45.

Yes, this is better, thanks!

Quoted 4 lines

-hpa

#9 lore raw

Re: [PATCH v3 4/7] x86/tlb: add tlb flush all factor for specific CPUs

In reply to Alex Shi
Quoted 7 lines · Alex Shi

like some machine in my hands, balance points is 16 entries on Romely-EP; while it is at 8 entries on Bloomfield NHM-EP; but on model 15 core2 Xeon using invlpg has nothing help.

For untested machine, no optimization now.

This seems a bit concerning from a forward compatibility point of view.
It would make more sense to assume that future processors would behave
more like the most recent (Sandy Bridge in your case) unless there is evidence to the contrary.

arch/x86/mm/tlb.c flush_all: · line 368

Quoted 5 lines
arch/x86/mm/tlb.c +2 -1
-			if ((end - start)/PAGE_SIZE > act_entries/FLUSHALL_BAR)
+			if ((end - start)/PAGE_SIZE >
+					act_entries/tlb_flushall_factor)
 				local_flush_tlb();

You are doing a dynamic division (and on unknown/older CPUs you're dividing by zero!!) but all your divisors are powers of two. Make it a shift instead.

-hpa

Signature · 2 lines
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.
#10 lore raw

Re: [PATCH v3 4/7] x86/tlb: add tlb flush all factor for specific CPUs

Quoted 5 lines

This seems a bit concerning from a forward compatibility point of view.
It would make more sense to assume that future processors would behave
more like the most recent (Sandy Bridge in your case) unless there is evidence to the contrary.

This would not work on new Atoms, which inter mix model numbers.

In the past I did a hack in other software to check the cache sizes to distinguish the two for unknown models It's not pretty however.

-Andi

#11 lore raw

Re: [PATCH v3 4/7] x86/tlb: add tlb flush all factor for specific CPUs

Quoted 7 lines · Andi Kleen
Quoted 5 lines

This seems a bit concerning from a forward compatibility point of view.
It would make more sense to assume that future processors would behave
more like the most recent (Sandy Bridge in your case) unless there is evidence to the contrary.

This would not work on new Atoms, which inter mix model numbers.

That is another problem, but there isn't much we (as the Linux community) can do about that, so we can just do the optimistic thing...

Quoted 3 lines

In the past I did a hack in other software to check the cache sizes to distinguish the two
for unknown models It's not pretty however.

Ick.

-hpa

#12 lore raw

Re: [PATCH v3 4/7] x86/tlb: add tlb flush all factor for specific CPUs

In reply to Alex Shi

arch/x86/mm/tlb.c void flush_tlb_range(struct vm_area_struct *vma, · line 345

Quoted 4 lines · Alex Shi
arch/x86/mm/tlb.c +2 -1
-	if (!cpu_has_invlpg || vma->vm_flags & VM_HUGETLB) {
+	if (!cpu_has_invlpg || vma->vm_flags & VM_HUGETLB
+				|| !tlb_flushall_factor) {

OK nevermind on the division by zero, but this yet another test being performed.

If nothing else you could fold these tests and make tlb_flushall_factor zero if !cpu_has_invlpg.

-hpa

Signature · 2 lines
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.
author
#13 lore raw

Re: [PATCH v3 4/7] x86/tlb: add tlb flush all factor for specific CPUs

Quoted 11 lines · H. Peter Anvin

arch/x86/mm/tlb.c void flush_tlb_range(struct vm_area_struct *vma, · line 345

Quoted 4 lines · Alex Shi
arch/x86/mm/tlb.c +2 -1
-	if (!cpu_has_invlpg || vma->vm_flags & VM_HUGETLB) {
+	if (!cpu_has_invlpg || vma->vm_flags & VM_HUGETLB
+				|| !tlb_flushall_factor) {

OK nevermind on the division by zero, but this yet another test being performed.

If nothing else you could fold these tests and make tlb_flushall_factor zero if !cpu_has_invlpg.

Yes, should be.

Quoted 3 lines

-hpa

author
#14 lore raw

PATCH v3 5/7 x86/tlb: remove comments for tlb_flush_range implement suggestion

In reply to Alex Shi

Thanks to this comments that inspired me using invlpg to replace cr3 rewrite.
Now, it is time to remove from code to avoid reader confusing.

---
 arch/x86/include/asm/tlbflush.h |    4 ----
 1 files changed, 0 insertions(+), 4 deletions(-)
arch/x86/include/asm/tlbflush.h +0 -4
index ec30dfb..51f8b1c 100644
@@ -81,10 +81,6 @@ static inline void __flush_tlb_one(unsigned long addr)
  *
  * ..but the i386 has somewhat limited tlb flushing capabilities,
  * and page-granular flushes are available only on i486 and up.
- *
- * x86-64 can only flush individual pages or full VMs. For a range flush
- * we always do the full VM. Might be worth trying if for a small
- * range a few INVLPGs in a row are a win.
  */
 
 #ifndef CONFIG_SMP
Signature · 1 line
1.7.5.4
maintainer
#15 lore raw

Re: [PATCH v3 5/7] x86/tlb: remove comments for tlb_flush_range implement suggestion

Quoted 3 lines · Alex Shi

Thanks to this comments that inspired me using invlpg to replace cr3 rewrite.
Now, it is time to remove from code to avoid reader confusing.

Nitpick, but this should just be folded into the patch which does the invlpg conversion.

Quoted 29 lines

---
 arch/x86/include/asm/tlbflush.h |    4 ----

 1 files changed, 0 insertions(+), 4 deletions(-)
arch/x86/include/asm/tlbflush.h +0 -5
index ec30dfb..51f8b1c 100644
@@ -81,10 +81,6 @@ static inline void __flush_tlb_one(unsigned long addr)
  *
  * ..but the i386 has somewhat limited tlb flushing capabilities,
  * and page-granular flushes are available only on i486 and up.
- *
- * x86-64 can only flush individual pages or full VMs. For a range flush
- * we always do the full VM. Might be worth trying if for a small
- * range a few INVLPGs in a row are a win.
  */
 
  #ifndef CONFIG_SMP
--

1.7.5.4

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

author
#16 lore raw

PATCH v3 6/7 x86/tlb: optimizing flush_tlb_mm

In reply to Alex Shi

Not every flush_tlb_mm execution moment is really need to evacuate all TLB entries, like in munmap, just few 'invlpg' is better for whole process performance, since it leaves most of TLB entries for later accessing.

This patch is changing flush_tlb_mm(mm) to flush_tlb_mm(mm, start, end) in cases.

---
 arch/x86/include/asm/tlb.h      |    2 +-
 arch/x86/include/asm/tlbflush.h |    5 ++-
 arch/x86/mm/pgtable.c           |    2 +-
 arch/x86/mm/tlb.c               |   65 ++++++++++++++++++--------------------
 fs/proc/task_mmu.c              |    2 +-
 include/asm-generic/tlb.h       |    4 +-
 include/asm-generic/tlbflush.h  |    3 +-
 kernel/fork.c                   |    2 +-
 mm/memory.c                     |    9 +++--
 9 files changed, 47 insertions(+), 47 deletions(-)
arch/x86/include/asm/tlb.h +1 -1
index 829215f..505fdfe 100644
@@ -4,7 +4,7 @@
 #define tlb_start_vma(tlb, vma) do { } while (0)
 #define tlb_end_vma(tlb, vma) do { } while (0)
 #define __tlb_remove_tlb_entry(tlb, ptep, address) do { } while (0)
-#define tlb_flush(tlb) flush_tlb_mm((tlb)->mm)
+#define tlb_flush(tlb, start, end) flush_tlb_mm((tlb)->mm, start, end)
 
 #include <asm-generic/tlb.h>
 
arch/x86/include/asm/tlbflush.h +3 -2
index 51f8b1c..d1baeac 100644
@@ -89,7 +89,8 @@ static inline void __flush_tlb_one(unsigned long addr)
 #define flush_tlb_all() __flush_tlb_all()
 #define local_flush_tlb() __flush_tlb()
 
-static inline void flush_tlb_mm(struct mm_struct *mm)
+static inline void flush_tlb_mm(struct mm_struct *mm,
+			unsigned long start, unsigned long end)
 {
 	if (mm == current->active_mm)
 		__flush_tlb();
@@ -128,7 +129,7 @@ static inline void reset_lazy_tlbstate(void)
 
 extern void flush_tlb_all(void);
 extern void flush_tlb_current_task(void);
-extern void flush_tlb_mm(struct mm_struct *);
+extern void flush_tlb_mm(struct mm_struct *, unsigned long, unsigned long);
 extern void flush_tlb_page(struct vm_area_struct *, unsigned long);
 extern void flush_tlb_range(struct vm_area_struct *vma,
 				   unsigned long start, unsigned long end);
arch/x86/mm/pgtable.c +1 -1
index 8573b83..5aea5b0 100644
@@ -168,7 +168,7 @@ void pud_populate(struct mm_struct *mm, pud_t *pudp, pmd_t *pmd)
 	 * section 8.1: in PAE mode we explicitly have to flush the
 	 * TLB via cr3 if the top-level pgd is changed...
 	 */
-	flush_tlb_mm(mm);
+	flush_tlb_mm(mm, 0UL, -1UL);
 }
 #else  /* !CONFIG_X86_PAE */
 
arch/x86/mm/tlb.c +31 -34

103 lines not on this page · raw message

fs/proc/task_mmu.c +1 -1
index 2d60492..5728c8f 100644
@@ -660,7 +660,7 @@ static ssize_t clear_refs_write(struct file *file, const char __user *buf,
 			walk_page_range(vma->vm_start, vma->vm_end,
 					&clear_refs_walk);
 		}
-		flush_tlb_mm(mm);
+		flush_tlb_mm(mm, 0UL, -1UL);
 		up_read(&mm->mmap_sem);
 		mmput(mm);
 	}
include/asm-generic/tlb.h +2 -2
index f96a5b5..24e205d 100644
@@ -112,7 +112,7 @@ static inline int tlb_fast_mode(struct mmu_gather *tlb)
 }
 
 void tlb_gather_mmu(struct mmu_gather *tlb, struct mm_struct *mm, bool fullmm);
-void tlb_flush_mmu(struct mmu_gather *tlb);
+void tlb_flush_mmu(struct mmu_gather *tlb, unsigned long start, unsigned long end);
 void tlb_finish_mmu(struct mmu_gather *tlb, unsigned long start, unsigned long end);
 int __tlb_remove_page(struct mmu_gather *tlb, struct page *page);
 
@@ -123,7 +123,7 @@ int __tlb_remove_page(struct mmu_gather *tlb, struct page *page);
 static inline void tlb_remove_page(struct mmu_gather *tlb, struct page *page)
 {
 	if (!__tlb_remove_page(tlb, page))
-		tlb_flush_mmu(tlb);
+		tlb_flush_mmu(tlb, 0UL, -1UL);
 }
 
 /**
include/asm-generic/tlbflush.h +2 -1
index d6d0a88..db1d4bb 100644
@@ -11,7 +11,8 @@
 
 #include <linux/bug.h>
 
-static inline void flush_tlb_mm(struct mm_struct *mm)
+static inline void flush_tlb_mm(struct mm_struct *mm,
+			unsigned long start, unsigned long end)
 {
 	BUG();
 }
kernel/fork.c +1 -1
index b9372a0..a4f0c64 100644
@@ -427,7 +427,7 @@ static int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm)
 	retval = 0;
 out:
 	up_write(&mm->mmap_sem);
-	flush_tlb_mm(oldmm);
+	flush_tlb_mm(oldmm, 0UL, -1UL);
 	up_write(&oldmm->mmap_sem);
 	return retval;
 fail_nomem_anon_vma_fork:
mm/memory.c +5 -4
index 6105f47..05e2c2e 100644
@@ -218,14 +218,15 @@ void tlb_gather_mmu(struct mmu_gather *tlb, struct mm_struct *mm, bool fullmm)
 #endif
 }
 
-void tlb_flush_mmu(struct mmu_gather *tlb)
+void tlb_flush_mmu(struct mmu_gather *tlb,
+			unsigned long start, unsigned long end)
 {
 	struct mmu_gather_batch *batch;
 
 	if (!tlb->need_flush)
 		return;
 	tlb->need_flush = 0;
-	tlb_flush(tlb);
+	tlb_flush(tlb, start, end);
 #ifdef CONFIG_HAVE_RCU_TABLE_FREE
 	tlb_table_flush(tlb);
 #endif
@@ -248,7 +249,7 @@ void tlb_finish_mmu(struct mmu_gather *tlb, unsigned long start, unsigned long e
 {
 	struct mmu_gather_batch *batch, *next;
 
-	tlb_flush_mmu(tlb);
+	tlb_flush_mmu(tlb, start, end);
 
 	/* keep the page table cache within bounds */
 	check_pgt_cache();
@@ -1204,7 +1205,7 @@ again:
 	 */
 	if (force_flush) {
 		force_flush = 0;
-		tlb_flush_mmu(tlb);
+		tlb_flush_mmu(tlb, addr, end);
 		if (addr != end)
 			goto again;
 	}
Signature · 1 line
1.7.5.4
#17 lore raw

Re: [PATCH v3 6/7] x86/tlb: optimizing flush_tlb_mm

Quoted 3 lines · Alex Shi

This patch is changing flush_tlb_mm(mm) to flush_tlb_mm(mm, start, end)
in cases.

Uhm, no please don't. That's called flush_tlb_range().

author
#18 lore raw

PATCH v3 7/7 x86/tlb: add tlb_flushall_factor into sysfs for user change

In reply to Alex Shi

kernel will replace cr3 rewrite with invlpg when
tlb flush entries <= 1/tlb_flushall_factor
if tlb_flushall_factor is 0, kernel won't do this replace.

User can modify its value according to specific applications.

---
 arch/x86/kernel/cpu/intel.c |   35 +++++++++++++++++++++++++++++++++++
 drivers/base/cpu.c          |    2 ++
 include/linux/cpu.h         |    2 ++
 3 files changed, 39 insertions(+), 0 deletions(-)
arch/x86/kernel/cpu/intel.c +35 -0
index 42480c6..8a14081 100644
@@ -662,6 +662,41 @@ void intel_cpu_detect_tlb(struct cpuinfo_x86 *c)
 	intel_tlb_flushall_factor_set(c);
 }
 
+static ssize_t __tlb_flushall_factor_store(const char *buf,
+					size_t count, int smt)
+{
+	unsigned int factor = 0;
+
+	if (sscanf(buf, "%u", &factor) != 1)
+		return -EINVAL;
+
+	tlb_flushall_factor = factor;
+
+	return count;
+}
+
+static ssize_t tlb_flushall_factor_show(struct device *dev,
+					   struct device_attribute *attr,
+					   char *buf)
+{
+	return sprintf(buf, "%u\n", tlb_flushall_factor);
+}
+static ssize_t tlb_flushall_factor_store(struct device *dev,
+					    struct device_attribute *attr,
+					    const char *buf, size_t count)
+{
+	return __tlb_flushall_factor_store(buf, count, 0);
+}
+
+static DEVICE_ATTR(tlb_flushall_factor, 0644,
+		   tlb_flushall_factor_show,
+		   tlb_flushall_factor_store);
+
+int __init create_sysfs_tlb_flushall_factor(struct device *dev)
+{
+	return device_create_file(dev, &dev_attr_tlb_flushall_factor);
+}
+
 static const struct cpu_dev __cpuinitconst intel_cpu_dev = {
 	.c_vendor	= "Intel",
 	.c_ident	= { "GenuineIntel" },
drivers/base/cpu.c +2 -0
index adf937b..b46e1b9 100644
@@ -331,6 +331,8 @@ void __init cpu_dev_init(void)
 
 	cpu_dev_register_generic();
 
+	create_sysfs_tlb_flushall_factor(cpu_subsys.dev_root);
+
 #if defined(CONFIG_SCHED_MC) || defined(CONFIG_SCHED_SMT)
 	sched_create_sysfs_power_savings_entries(cpu_subsys.dev_root);
 #endif
include/linux/cpu.h +2 -0
index ee28844..5c953b0 100644
@@ -36,6 +36,8 @@ extern void cpu_remove_dev_attr(struct device_attribute *attr);
 extern int cpu_add_dev_attr_group(struct attribute_group *attrs);
 extern void cpu_remove_dev_attr_group(struct attribute_group *attrs);
 
+extern int create_sysfs_tlb_flushall_factor(struct device *dev);
+
 extern int sched_create_sysfs_power_savings_entries(struct device *dev);
 
 #ifdef CONFIG_HOTPLUG_CPU
Signature · 1 line
1.7.5.4
#19 lore raw

Re: [PATCH v3 7/7] x86/tlb: add tlb_flushall_factor into sysfs for user change

Quoted 5 lines · Alex Shi

kernel will replace cr3 rewrite with invlpg when
tlb flush entries <= 1/tlb_flushall_factor
if tlb_flushall_factor is 0, kernel won't do this replace.

User can modify its value according to specific applications.

Yay!, another useless config knob :-(

#20 lore raw

Re: [PATCH v3 7/7] x86/tlb: add tlb_flushall_factor into sysfs for user change

Quoted 9 lines · Peter Zijlstra
Quoted 5 lines · Alex Shi

kernel will replace cr3 rewrite with invlpg when
tlb flush entries <= 1/tlb_flushall_factor
if tlb_flushall_factor is 0, kernel won't do this replace.

User can modify its value according to specific applications.

Yay!, another useless config knob :-(

I think this one is mostly for benchmarking new silicon to find out the proper value, I'm guessing?

-hpa

#21 lore raw

Re: [PATCH v3 7/7] x86/tlb: add tlb_flushall_factor into sysfs for user change

Quoted 6 lines · H. Peter Anvin
Quoted 5 lines · Peter Zijlstra
Quoted 4 lines · Alex Shi

kernel will replace cr3 rewrite with invlpg when
tlb flush entries <= 1/tlb_flushall_factor
if tlb_flushall_factor is 0, kernel won't do this replace.

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

User can modify its value according to specific applications.

Yay!, another useless config knob :-(

I think this one is mostly for benchmarking new silicon to find out the proper value, I'm guessing?

Yep, this was supposed to be for pinpointing the proper entries count. I believe we don't have this thing of having code intree but not really intree - only for testing new silicon. Maybe a debug module or add this to debugfs?

Signature · 8 lines
Regards/Gruss,
Boris.

Advanced Micro Devices GmbH
Einsteinring 24, 85609 Dornach
GM: Alberto Bozzo
Reg: Dornach, Landkreis Muenchen
HRB Nr. 43632 WEEE Registernr: 129 19551
#22 lore raw

Re: [PATCH v3 7/7] x86/tlb: add tlb_flushall_factor into sysfs for user change

Quoted 6 lines · Borislav Petkov

Yep, this was supposed to be for pinpointing the proper entries count. I believe we don't have this thing of having code intree but not really intree - only for testing new silicon. Maybe a debug module or add this to debugfs?

We have a lot of that kind of code, it is usually hidden under CONFIG_*DEBUG* something.

-hpa

Signature · 2 lines
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.
#23 lore raw

Re: [PATCH v3 7/7] x86/tlb: add tlb_flushall_factor into sysfs for user change

Quoted 10 lines · H. Peter Anvin
Quoted 6 lines · Borislav Petkov

Yep, this was supposed to be for pinpointing the proper entries count. I believe we don't have this thing of having code intree but not really intree - only for testing new silicon. Maybe a debug module or add this to debugfs?

We have a lot of that kind of code, it is usually hidden under CONFIG_*DEBUG* something.

Then maybe this is where this knob should go, along with adding it to debugfs instead of sysfs...

Signature · 8 lines
Regards/Gruss,
Boris.

Advanced Micro Devices GmbH
Einsteinring 24, 85609 Dornach
GM: Alberto Bozzo
Reg: Dornach, Landkreis Muenchen
HRB Nr. 43632 WEEE Registernr: 129 19551
#24 lore raw

Re: [PATCH v3 7/7] x86/tlb: add tlb_flushall_factor into sysfs for user change

Quoted 9 lines · Borislav Petkov
Quoted 8 lines · H. Peter Anvin
Quoted 6 lines · Borislav Petkov

Yep, this was supposed to be for pinpointing the proper entries count. I believe we don't have this thing of having code intree but not really intree - only for testing new silicon. Maybe a debug module or add this to debugfs?

Quoted 6 lines
Quoted 2 lines

We have a lot of that kind of code, it is usually hidden under CONFIG_*DEBUG* something.

Then maybe this is where this knob should go, along with adding it to debugfs instead of sysfs...

I don't see a problem with it in sysfs.

Signature · 2 lines
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.
maintainer
#25 lore raw

Re: [PATCH v3 7/7] x86/tlb: add tlb_flushall_factor into sysfs for user change

Quoted 10 lines · H. Peter Anvin
Quoted 9 lines · Borislav Petkov
Quoted 8 lines · H. Peter Anvin
Quoted 6 lines · Borislav Petkov

Yep, this was supposed to be for pinpointing the proper entries count. I believe we don't have this thing of having code intree but not really intree - only for testing new silicon. Maybe a debug module or add this to debugfs?

Quoted 8 lines
Quoted 6 lines
Quoted 2 lines

We have a lot of that kind of code, it is usually hidden under CONFIG_*DEBUG* something.

Then maybe this is where this knob should go, along with adding it to debugfs instead of sysfs...

I don't see a problem with it in sysfs.

If it's in sysfs, it better be documented in Documentation/ABI.

#26 lore raw

Re: [PATCH v3 7/7] x86/tlb: add tlb_flushall_factor into sysfs for user change

Quoted 6 lines · Greg KH
Quoted 4 lines
Quoted 2 lines

Then maybe this is where this knob should go, along with adding it to debugfs instead of sysfs...

I don't see a problem with it in sysfs.

If it's in sysfs, it better be documented in Documentation/ABI.

See what I'm sayin'? And besides, debugfs fits better the purpose of this, maybe somewhere under <debugfs>/x86/...

Signature · 8 lines
Regards/Gruss,
Boris.

Advanced Micro Devices GmbH
Einsteinring 24, 85609 Dornach
GM: Alberto Bozzo
Reg: Dornach, Landkreis Muenchen
HRB Nr. 43632 WEEE Registernr: 129 19551
author
#27 lore raw

Re: [PATCH v3 7/7] x86/tlb: add tlb_flushall_factor into sysfs for user change

In reply to H. Peter Anvin
Quoted 10 lines · H. Peter Anvin
Quoted 9 lines · Borislav Petkov
Quoted 8 lines · H. Peter Anvin
Quoted 6 lines · Borislav Petkov

Yep, this was supposed to be for pinpointing the proper entries count. I believe we don't have this thing of having code intree but not really intree - only for testing new silicon. Maybe a debug module or add this to debugfs?

Quoted 9 lines
Quoted 6 lines
Quoted 2 lines

We have a lot of that kind of code, it is usually hidden under CONFIG_*DEBUG* something.

Then maybe this is where this knob should go, along with adding it to debugfs instead of sysfs...

I don't see a problem with it in sysfs.

Ok. I will modify it according to comments.

#28 lore raw
In reply to Alex Shi
Quoted 10 lines · Alex Shi

This version just added 2 patches, the 6th and 7th. The first 5 patches are same as v2.

Seems it is a little hard for me to find a real case to show performance gain. But the advantage is there ture.

Any comments are appreciated!

Alex

Quoted 7 lines

[PATCH v3 1/7] x86/tlb_info: get last level TLB entry number of CPU [PATCH v3 2/7] x86/flush_tlb: try flush_tlb_single one by one in [PATCH v3 3/7] x86/tlb: fall back to flush all when meet a THP large [PATCH v3 4/7] x86/tlb: add tlb flush all factor for specific CPUs [PATCH v3 5/7] x86/tlb: remove comments for tlb_flush_range [PATCH v3 6/7] x86/tlb: optimizing flush_tlb_mm
[PATCH v3 7/7] x86/tlb: add tlb_flushall_factor into sysfs for user

Have you tried what happens if you get rid of the funny multi-vector-ipi scheme and use the generic smp_call functions?

author
#29 lore raw
Quoted 11 lines
Quoted 8 lines

[PATCH v3 1/7] x86/tlb_info: get last level TLB entry number of CPU [PATCH v3 2/7] x86/flush_tlb: try flush_tlb_single one by one in [PATCH v3 3/7] x86/tlb: fall back to flush all when meet a THP large [PATCH v3 4/7] x86/tlb: add tlb flush all factor for specific CPUs [PATCH v3 5/7] x86/tlb: remove comments for tlb_flush_range [PATCH v3 6/7] x86/tlb: optimizing flush_tlb_mm
[PATCH v3 7/7] x86/tlb: add tlb_flushall_factor into sysfs for user

Have you tried what happens if you get rid of the funny multi-vector-ipi scheme and use the generic smp_call functions?

Not yet.

#30 lore raw
In reply to Peter Zijlstra
Quoted 2 lines

Have you tried what happens if you get rid of the funny multi-vector-ipi scheme and use the generic smp_call functions?

Yes we did. It's much faster on larger systems.

But haven't sent the patch yet because wasn't sure if it wasn't slower on small systems.

-Andi

Signature · 1 line
[email protected] -- Speaking for myself only
author
#31 lore raw
Quoted 7 lines · Andi Kleen
Quoted 2 lines

Have you tried what happens if you get rid of the funny multi-vector-ipi scheme and use the generic smp_call functions?

Yes we did. It's much faster on larger systems.

But haven't sent the patch yet because wasn't sure if it wasn't slower on small systems.

Why you worried about the small system? are there some bad example there?

It should be helpful after multiple invlpg enabled. I like to take a look this later. :)

Quoted 4 lines

-Andi

#32 lore raw
In reply to Andi Kleen

From: Andi Kleen <[email protected]>
Subject: Re: [PATCH v3] TLB flush optimization
Date: Wed, 9 May 2012 16:45:12 -0700

Quoted 9 lines
Quoted 2 lines

Have you tried what happens if you get rid of the funny multi-vector-ipi scheme and use the generic smp_call functions?

Yes we did. It's much faster on larger systems.

But haven't sent the patch yet because wasn't sure if it wasn't slower on small systems.

-Andi

I'm not sure the reason of performance gain. I'm guessing that the performance gain depends on waiting time of specific multi-vector-ipi vs wasting time of generic code consumed for the processing actually unnecessary for TLB flushing, and on small systems the specific one is shorter than the generic one, and on large systems the converse holds. Is this correct?

Thanks.
HATAYAMA, Daisuke