Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1628794 > unrolled thread

[PATCH v3 0/4] x86/mm: Straightforward TLB flush fixes/cleanups

Back to article view | Back to linux.kernel

Contents

  [PATCH v3 0/4] x86/mm: Straightforward TLB flush fixes/cleanups Andy Lutomirski <luto@kernel.org> - 2017-04-22 09:10 +0200
    [PATCH v3 4/4] x86/mm: Fix flush_tlb_page() on Xen Andy Lutomirski <luto@kernel.org> - 2017-04-22 09:10 +0200
    [PATCH v3 2/4] x86/mm: Remove flush_tlb() and flush_tlb_current_task() Andy Lutomirski <luto@kernel.org> - 2017-04-22 09:10 +0200
    [PATCH v3 3/4] x86/mm: Make flush_tlb_mm_range() more predictable Andy Lutomirski <luto@kernel.org> - 2017-04-22 09:10 +0200
    [PATCH v3 1/4] x86/vm86/32: Switch to flush_tlb_mm_range() in mark_screen_rdonly() Andy Lutomirski <luto@kernel.org> - 2017-04-22 09:10 +0200

#1628794 — [PATCH v3 0/4] x86/mm: Straightforward TLB flush fixes/cleanups

I have some less straightforward cleanups coming, but these three
are easy and might even be okay for 4.12 assuming that someone feels
like reviewing them.

Changes from v2:
 - Added patch 1 (I suck at grep)
 - Delete one more copy of flush_tlb() in patch 2 (ditto)

Changes from v1:
 - Fixed the Cc list on patch 3.  No code changes at all.

Andy Lutomirski (4):
  x86/vm86/32: Switch to flush_tlb_mm_range() in mark_screen_rdonly()
  x86/mm: Remove flush_tlb() and flush_tlb_current_task()
  x86/mm: Make flush_tlb_mm_range() more predictable
  x86/mm: Fix flush_tlb_page() on Xen

 arch/x86/include/asm/tlbflush.h |  9 ---------
 arch/x86/kernel/vm86_32.c       |  2 +-
 arch/x86/mm/tlb.c               | 33 ++++++++-------------------------
 3 files changed, 9 insertions(+), 35 deletions(-)

-- 
2.9.3

[toc] | [next] | [standalone]

#1628795 — [PATCH v3 4/4] x86/mm: Fix flush_tlb_page() on Xen

flush_tlb_page() passes a bogus range to flush_tlb_others() and
expects the latter to fix it up.  native_flush_tlb_others() has the
fixup but Xen's version doesn't.  Move the fixup to
flush_tlb_others().

AFAICS the only real effect is that, without this fix, Xen would
flush everything instead of just the one page on remote vCPUs in
when flush_tlb_page() was called.

Cc: Rik van Riel <riel@redhat.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Nadav Amit <namit@vmware.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Juergen Gross <jgross@suse.com>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Fixes: e7b52ffd45a6 ("x86/flush_tlb: try flush_tlb_single one by one in flush_tlb_range")
Signed-off-by: Andy Lutomirski <luto@kernel.org>
---
 arch/x86/mm/tlb.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/arch/x86/mm/tlb.c b/arch/x86/mm/tlb.c
index 9db9260a5e9f..6e7bedf69af7 100644
--- a/arch/x86/mm/tlb.c
+++ b/arch/x86/mm/tlb.c
@@ -263,8 +263,6 @@ void native_flush_tlb_others(const struct cpumask *cpumask,
 {
 	struct flush_tlb_info info;
 
-	if (end == 0)
-		end = start + PAGE_SIZE;
 	info.flush_mm = mm;
 	info.flush_start = start;
 	info.flush_end = end;
@@ -378,7 +376,7 @@ void flush_tlb_page(struct vm_area_struct *vma, unsigned long start)
 	}
 
 	if (cpumask_any_but(mm_cpumask(mm), smp_processor_id()) < nr_cpu_ids)
-		flush_tlb_others(mm_cpumask(mm), mm, start, 0UL);
+		flush_tlb_others(mm_cpumask(mm), mm, start, start + PAGE_SIZE);
 
 	preempt_enable();
 }
-- 
2.9.3

[toc] | [prev] | [next] | [standalone]

#1628796 — [PATCH v3 2/4] x86/mm: Remove flush_tlb() and flush_tlb_current_task()

I was trying to figure out what how flush_tlb_current_task() would
possibly work correctly if current->mm != current->active_mm, but I
realized I could spare myself the effort: it has no callers except
the unused flush_tlb() macro.

Cc: Rik van Riel <riel@redhat.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Nadav Amit <namit@vmware.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andy Lutomirski <luto@kernel.org>
---
 arch/x86/include/asm/tlbflush.h |  9 ---------
 arch/x86/mm/tlb.c               | 17 -----------------
 2 files changed, 26 deletions(-)

diff --git a/arch/x86/include/asm/tlbflush.h b/arch/x86/include/asm/tlbflush.h
index fc5abff9b7fd..8e7ae9e6c59a 100644
--- a/arch/x86/include/asm/tlbflush.h
+++ b/arch/x86/include/asm/tlbflush.h
@@ -205,7 +205,6 @@ static inline void __flush_tlb_one(unsigned long addr)
 /*
  * TLB flushing:
  *
- *  - flush_tlb() flushes the current mm struct TLBs
  *  - flush_tlb_all() flushes all processes TLBs
  *  - flush_tlb_mm(mm) flushes the specified mm context TLB's
  *  - flush_tlb_page(vma, vmaddr) flushes one page
@@ -237,11 +236,6 @@ static inline void flush_tlb_all(void)
 	__flush_tlb_all();
 }
 
-static inline void flush_tlb(void)
-{
-	__flush_tlb_up();
-}
-
 static inline void local_flush_tlb(void)
 {
 	__flush_tlb_up();
@@ -303,14 +297,11 @@ static inline void flush_tlb_kernel_range(unsigned long start,
 		flush_tlb_mm_range(vma->vm_mm, start, end, vma->vm_flags)
 
 extern void flush_tlb_all(void);
-extern void flush_tlb_current_task(void);
 extern void flush_tlb_page(struct vm_area_struct *, unsigned long);
 extern void flush_tlb_mm_range(struct mm_struct *mm, unsigned long start,
 				unsigned long end, unsigned long vmflag);
 extern void flush_tlb_kernel_range(unsigned long start, unsigned long end);
 
-#define flush_tlb()	flush_tlb_current_task()
-
 void native_flush_tlb_others(const struct cpumask *cpumask,
 				struct mm_struct *mm,
 				unsigned long start, unsigned long end);
diff --git a/arch/x86/mm/tlb.c b/arch/x86/mm/tlb.c
index a7655f6caf7d..92ec37f517ab 100644
--- a/arch/x86/mm/tlb.c
+++ b/arch/x86/mm/tlb.c
@@ -289,23 +289,6 @@ void native_flush_tlb_others(const struct cpumask *cpumask,
 	smp_call_function_many(cpumask, flush_tlb_func, &info, 1);
 }
 
-void flush_tlb_current_task(void)
-{
-	struct mm_struct *mm = current->mm;
-
-	preempt_disable();
-
-	count_vm_tlb_event(NR_TLB_LOCAL_FLUSH_ALL);
-
-	/* This is an implicit full barrier that synchronizes with switch_mm. */
-	local_flush_tlb();
-
-	trace_tlb_flush(TLB_LOCAL_SHOOTDOWN, TLB_FLUSH_ALL);
-	if (cpumask_any_but(mm_cpumask(mm), smp_processor_id()) < nr_cpu_ids)
-		flush_tlb_others(mm_cpumask(mm), mm, 0UL, TLB_FLUSH_ALL);
-	preempt_enable();
-}
-
 /*
  * See Documentation/x86/tlb.txt for details.  We choose 33
  * because it is large enough to cover the vast majority (at
-- 
2.9.3

[toc] | [prev] | [next] | [standalone]

#1628797 — [PATCH v3 3/4] x86/mm: Make flush_tlb_mm_range() more predictable

I'm about to rewrite the function almost completely, but first I
want to get a functional change out of the way.  Currently, if
flush_tlb_mm_range() does not flush the local TLB at all, it will
never do individual page flushes on remote CPUs.  This seems to be
an accident, and preserving it will be awkward.  Let's change it
first so that any regressions in the rewrite will be easier to
bisect and so that the rewrite can attempt to change no visible
behavior at all.

The fix is simple: we can simply avoid short-circuiting the
calculation of base_pages_to_flush.

As a side effect, this also eliminates a potential corner case: if
tlb_single_page_flush_ceiling == TLB_FLUSH_ALL, flush_tlb_mm_range()
could have ended up flushing the entire address space one page at a
time.

Cc: Rik van Riel <riel@redhat.com>
Cc: Nadav Amit <namit@vmware.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Acked-by: Dave Hansen <dave.hansen@intel.com>
Signed-off-by: Andy Lutomirski <luto@kernel.org>
---
 arch/x86/mm/tlb.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/arch/x86/mm/tlb.c b/arch/x86/mm/tlb.c
index 92ec37f517ab..9db9260a5e9f 100644
--- a/arch/x86/mm/tlb.c
+++ b/arch/x86/mm/tlb.c
@@ -309,6 +309,12 @@ void flush_tlb_mm_range(struct mm_struct *mm, unsigned long start,
 	unsigned long base_pages_to_flush = TLB_FLUSH_ALL;
 
 	preempt_disable();
+
+	if ((end != TLB_FLUSH_ALL) && !(vmflag & VM_HUGETLB))
+		base_pages_to_flush = (end - start) >> PAGE_SHIFT;
+	if (base_pages_to_flush > tlb_single_page_flush_ceiling)
+		base_pages_to_flush = TLB_FLUSH_ALL;
+
 	if (current->active_mm != mm) {
 		/* Synchronize with switch_mm. */
 		smp_mb();
@@ -325,15 +331,11 @@ void flush_tlb_mm_range(struct mm_struct *mm, unsigned long start,
 		goto out;
 	}
 
-	if ((end != TLB_FLUSH_ALL) && !(vmflag & VM_HUGETLB))
-		base_pages_to_flush = (end - start) >> PAGE_SHIFT;
-
 	/*
 	 * Both branches below are implicit full barriers (MOV to CR or
 	 * INVLPG) that synchronize with switch_mm.
 	 */
-	if (base_pages_to_flush > tlb_single_page_flush_ceiling) {
-		base_pages_to_flush = TLB_FLUSH_ALL;
+	if (base_pages_to_flush == TLB_FLUSH_ALL) {
 		count_vm_tlb_event(NR_TLB_LOCAL_FLUSH_ALL);
 		local_flush_tlb();
 	} else {
-- 
2.9.3

[toc] | [prev] | [next] | [standalone]

#1628798 — [PATCH v3 1/4] x86/vm86/32: Switch to flush_tlb_mm_range() in mark_screen_rdonly()

mark_screen_rdonly() is the last remaining caller of flush_tlb().
flush_tlb_mm_range() is potentially faster and isn't obsolete.

Compile-tested only because I don't know whether software that uses
this mechanism even exists.

Cc: Rik van Riel <riel@redhat.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Nadav Amit <namit@vmware.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Sasha Levin <sasha.levin@oracle.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Andy Lutomirski <luto@kernel.org>
---
 arch/x86/kernel/vm86_32.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/kernel/vm86_32.c b/arch/x86/kernel/vm86_32.c
index 23ee89ce59a9..3eda76b3c835 100644
--- a/arch/x86/kernel/vm86_32.c
+++ b/arch/x86/kernel/vm86_32.c
@@ -193,7 +193,7 @@ static void mark_screen_rdonly(struct mm_struct *mm)
 	pte_unmap_unlock(pte, ptl);
 out:
 	up_write(&mm->mmap_sem);
-	flush_tlb();
+	flush_tlb_mm_range(mm, 0xA0000, 0xA0000 + 32*PAGE_SIZE, 0UL);
 }
 
 
-- 
2.9.3

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web