x86: Flush TLB if PGD entry is changed in i386 PAE mode

commit 4981d01eada5354d81c8929d5b2836829ba3df7b upstream.

According to intel CPU manual, every time PGD entry is changed in i386 PAE
mode, we need do a full TLB flush. Current code follows this and there is
comment for this too in the code.

But current code misses the multi-threaded case. A changed page table
might be used by several CPUs, every such CPU should flush TLB. Usually
this isn't a problem, because we prepopulate all PGD entries at process
fork. But when the process does munmap and follows new mmap, this issue
will be triggered.

When it happens, some CPUs keep doing page faults:

  http://marc.info/?l=linux-kernel&m=129915020508238&w=2

Reported-by: Yasunori Goto<y-goto@jp.fujitsu.com>
Tested-by: Yasunori Goto<y-goto@jp.fujitsu.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Shaohua Li<shaohua.li@intel.com>
Cc: Mallick Asit K <asit.k.mallick@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: linux-mm <linux-mm@kvack.org>
LKML-Reference: <1300246649.2337.95.camel@sli10-conroe>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

diff --git a/arch/x86/include/asm/pgtable-3level.h b/arch/x86/include/asm/pgtable-3level.h
index 177b0165ea01fa8aeeb1f8749252d422de52b3a6..33927d283c921b888b2d092011a0e61476ee9dc4 100644 (file)
--- a/arch/x86/include/asm/pgtable-3level.h
+++ b/arch/x86/include/asm/pgtable-3level.h
@@ -69,8 +69,6 @@ static inline void native_pmd_clear(pmd_t *pmd)
 
 static inline void pud_clear(pud_t *pudp)
 {
-       unsigned long pgd;
-
        set_pud(pudp, __pud(0));
 
        /*
@@ -79,13 +77,10 @@ static inline void pud_clear(pud_t *pudp)
         * section 8.1: in PAE mode we explicitly have to flush the
         * TLB via cr3 if the top-level pgd is changed...
         *
-        * Make sure the pud entry we're updating is within the
-        * current pgd to avoid unnecessary TLB flushes.
+        * Currently all places where pud_clear() is called either have
+        * flush_tlb_mm() followed or don't need TLB flush (x86_64 code or
+        * pud_clear_bad()), so we don't need TLB flush here.
         */
-       pgd = read_cr3();
-       if (__pa(pudp) >= pgd && __pa(pudp) <
-           (pgd + sizeof(pgd_t)*PTRS_PER_PGD))
-               write_cr3(pgd);
 }
 
 #ifdef CONFIG_SMP

diff --git a/arch/x86/mm/pgtable.c b/arch/x86/mm/pgtable.c
index 5c4ee422590e5dc23aec0071e642bf246b627565..2a72049068e73f67eaf05b108b13e026c9b6139d 100644 (file)
--- a/arch/x86/mm/pgtable.c
+++ b/arch/x86/mm/pgtable.c
@@ -160,8 +160,7 @@ 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...
         */
-       if (mm == current->active_mm)
-               write_cr3(read_cr3());
+       flush_tlb_mm(mm);
 }
 #else  /* !CONFIG_X86_PAE */