mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Andi Kleen <ak@suse.de>
To: linux-kernel@vger.kernel.org, jbeulich@novell.com, mingo@elte.hu,
	tglx@linutronix.de
Subject: [PATCH] [15/31] CPA: Don't flush the caches when the CPU supports self-snoop
Date: Mon, 14 Jan 2008 23:16:47 +0100 (CET)	[thread overview]
Message-ID: <20080114221647.AF1FF14F83@wotan.suse.de> (raw)
In-Reply-To: <200801141116.534682000@suse.de>


When the self-snoop CPUID bit is set change_page_attr() only needs to flush
TLBs, but not the caches.

The description of self-snoop in the Intel manuals is a bit vague
but I got confirmation that this is what SS really means.

This should improve c_p_a() performance significantly on newer
Intel CPUs.

Note: the line > 80 characters will be modified again in a followup

Signed-off-by: Andi Kleen <ak@suse.de>

---
 arch/x86/mm/pageattr_32.c    |    5 +++--
 arch/x86/mm/pageattr_64.c    |    4 ++--
 include/asm-x86/cpufeature.h |    1 +
 3 files changed, 6 insertions(+), 4 deletions(-)

Index: linux/arch/x86/mm/pageattr_32.c
===================================================================
--- linux.orig/arch/x86/mm/pageattr_32.c
+++ linux/arch/x86/mm/pageattr_32.c
@@ -97,10 +97,11 @@ static void flush_kernel_map(void *arg)
 	struct flush_arg *a = (struct flush_arg *)arg;
 	struct flush *f;
 
-	if ((!cpu_has_clflush || a->full_flush) && boot_cpu_data.x86_model >= 4)
+	if ((!cpu_has_clflush || a->full_flush) && boot_cpu_data.x86_model >= 4 &&
+		!cpu_has_ss)
 		wbinvd();
 	list_for_each_entry(f, &a->l, l) {
-		if (!a->full_flush)
+		if (!a->full_flush && !cpu_has_ss)
 			clflush_cache_range((void *)f->addr, PAGE_SIZE);
 		if (!a->full_flush)
 			__flush_tlb_one(f->addr);
Index: linux/include/asm-x86/cpufeature.h
===================================================================
--- linux.orig/include/asm-x86/cpufeature.h
+++ linux/include/asm-x86/cpufeature.h
@@ -167,6 +167,7 @@
 #define cpu_has_pebs		boot_cpu_has(X86_FEATURE_PEBS)
 #define cpu_has_clflush		boot_cpu_has(X86_FEATURE_CLFLSH)
 #define cpu_has_bts		boot_cpu_has(X86_FEATURE_BTS)
+#define cpu_has_ss		boot_cpu_has(X86_FEATURE_SELFSNOOP)
 
 #if defined(CONFIG_X86_INVLPG) || defined(CONFIG_X86_64)
 # define cpu_has_invlpg		1
Index: linux/arch/x86/mm/pageattr_64.c
===================================================================
--- linux.orig/arch/x86/mm/pageattr_64.c
+++ linux/arch/x86/mm/pageattr_64.c
@@ -94,10 +94,10 @@ static void flush_kernel_map(void *arg)
 
 	/* When clflush is available always use it because it is
 	   much cheaper than WBINVD. */
-	if (a->full_flush || !cpu_has_clflush)
+	if ((a->full_flush || !cpu_has_clflush) && !cpu_has_ss)
 		asm volatile("wbinvd" ::: "memory");
 	list_for_each_entry(f, &a->l, l) {
-		if (!a->full_flush)
+		if (!a->full_flush && !cpu_has_ss)
 			clflush_cache_range((void *)f->addr, PAGE_SIZE);
 		if (!a->full_flush)
 			__flush_tlb_one(f->addr);

  parent reply	other threads:[~2008-01-14 22:22 UTC|newest]

Thread overview: 56+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-01-14 22:16 [PATCH] [0/31] Great change_page_attr patch series v2 Andi Kleen
2008-01-14 22:16 ` [PATCH] [1/31] Shrink __PAGE_KERNEL/__PAGE_KERNEL_EXEC on non PAE kernels Andi Kleen
2008-01-14 22:16 ` [PATCH] [2/31] CPA: Do a simple self test at boot Andi Kleen
2008-01-15  8:47   ` Harvey Harrison
2008-01-15  9:59     ` Andi Kleen
2008-01-15 10:07       ` Harvey Harrison
2008-01-14 22:16 ` [PATCH] [3/31] Add pte accessors for the global bit Andi Kleen
2008-01-14 22:16 ` [PATCH] [4/31] Add pte_pgprot on i386 Andi Kleen
2008-01-15 13:00   ` Johannes Weiner
2008-01-14 22:16 ` [PATCH] [5/31] Don't drop NX bit in pte modifier functions for 32bit Andi Kleen
2008-01-14 22:16 ` [PATCH] [6/31] CPA: Undo white space changes Andi Kleen
2008-01-14 22:16 ` [PATCH] [7/31] Extract page table dumping code from i386 fault handler into dump_pagetable() Andi Kleen
2008-01-15  8:56   ` Harvey Harrison
2008-01-15 10:00     ` Andi Kleen
2008-01-15 10:05       ` Harvey Harrison
2008-01-14 22:16 ` [PATCH] [8/31] CPA: Return the page table level in lookup_address() Andi Kleen
2008-01-14 22:16 ` [PATCH] [9/31] CPA: Add simple self test at boot Andi Kleen
2008-01-15 10:37   ` Harvey Harrison
2008-01-14 22:16 ` [PATCH] [10/31] CPA: Change kernel_map_pages to not use c_p_a() Andi Kleen
2008-01-14 22:16 ` [PATCH] [11/31] CPA: Change 32bit back to init_mm semaphore locking Andi Kleen
2008-01-14 22:16 ` [PATCH] [12/31] CPA: CLFLUSH support in change_page_attr() Andi Kleen
2008-01-15  8:40   ` Jan Beulich
2008-01-15  9:57     ` Andi Kleen
2008-01-14 22:16 ` [PATCH] [13/31] CPA: Use macros to modify the PG_arch_1 page flags in change_page_attr Andi Kleen
2008-01-15  9:29   ` Harvey Harrison
2008-01-15 10:06     ` Andi Kleen
2008-01-15 10:15       ` Harvey Harrison
2008-01-15 10:25         ` Andi Kleen
2008-01-14 22:16 ` [PATCH] [14/31] CPA: Use page granuality TLB flushing " Andi Kleen
2008-01-14 22:16 ` Andi Kleen [this message]
2008-01-14 22:16 ` [PATCH] [16/31] CPA: Use wbinvd() macro instead of inline assembly in 64bit c_p_a() Andi Kleen
2008-01-14 22:16 ` [PATCH] [17/31] CPA: Reorder TLB / cache flushes to follow Intel recommendation Andi Kleen
2008-01-14 22:16 ` [PATCH] [18/31] CPA: Make change_page_attr() more robust against use of PAT bits Andi Kleen
2008-01-14 22:16 ` [PATCH] [19/31] CPA: Limit cache flushing to pages that really change caching Andi Kleen
2008-01-15  8:46   ` Jan Beulich
2008-01-14 22:16 ` [PATCH] [20/31] CPA: Fix inaccurate comments in 64bit change_page_attr() Andi Kleen
2008-01-14 22:16 ` [PATCH] [21/31] CPA: Dump pagetable when inconsistency is detected Andi Kleen
2008-01-14 22:16 ` [PATCH] [22/31] CPA: Only queue actually unused page table pages for freeing Andi Kleen
2008-01-14 22:16 ` [PATCH] [23/31] CPA: Remove unnecessary masking of address Andi Kleen
2008-01-14 22:16 ` [PATCH] [24/31] CPA: Only unmap kernel init pages in text mapping when CONFIG_DEBUG_RODATA is set Andi Kleen
2008-01-14 22:16 ` [PATCH] [25/31] CPA: Always do full TLB flush when splitting large pages Andi Kleen
2008-01-14 22:16 ` [PATCH] [26/31] CPA: Fix reference counting when changing already changed pages Andi Kleen
2008-01-15  9:05   ` Jan Beulich
2008-01-15 10:04     ` Andi Kleen
2008-01-15 12:00       ` Jan Beulich
2008-01-14 22:17 ` [PATCH] [27/31] CPA: Change comments of external interfaces to kerneldoc format Andi Kleen
2008-01-14 22:50   ` Randy Dunlap
2008-01-15  0:49     ` Andi Kleen
2008-01-14 22:17 ` [PATCH] [28/31] CPA: Make kernel_text test match boot mapping initialization Andi Kleen
2008-01-14 22:17 ` [PATCH] [29/31] CPA: Add a BUG_ON checking for someone setting the kernel text NX Andi Kleen
2008-01-14 22:17 ` [PATCH] [30/31] Remove set_kernel_exec Andi Kleen
2008-01-14 22:17 ` [PATCH] [31/31] Clean up pte_exec Andi Kleen
2008-01-15  9:11 ` [PATCH] [0/31] Great change_page_attr patch series v2 Jan Beulich
2008-01-15 10:06   ` Andi Kleen
2008-01-15 11:55     ` Jan Beulich
2008-01-15 12:43       ` Andi Kleen

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20080114221647.AF1FF14F83@wotan.suse.de \
    --to=ak@suse.de \
    --cc=jbeulich@novell.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=tglx@linutronix.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

Powered by JetHome