Subject: [merged] mm-numa-recheck-for-transhuge-pages-under-lock-during-protection-changes.patch removed from -mm tree
To: mgorman@xxxxxxx,riel@xxxxxxxxxx,sasha.levin@xxxxxxxxxx,mm-commits@xxxxxxxxxxxxxxx
From: akpm@xxxxxxxxxxxxxxxxxxxx
Date: Tue, 08 Apr 2014 13:32:11 -0700
The patch titled
Subject: mm: numa: recheck for transhuge pages under lock during protection changes
has been removed from the -mm tree. Its filename was
mm-numa-recheck-for-transhuge-pages-under-lock-during-protection-changes.patch
This patch was dropped because it was merged into mainline or a subsystem tree
------------------------------------------------------
From: Mel Gorman <mgorman@xxxxxxx>
Subject: mm: numa: recheck for transhuge pages under lock during protection changes
Sasha reported the following bug using trinity
[ 886.745765] kernel BUG at mm/mprotect.c:149!
[ 886.746831] invalid opcode: 0000 [#1] PREEMPT SMP DEBUG_PAGEALLOC
[ 886.748511] Dumping ftrace buffer:
[ 886.749998] (ftrace buffer empty)
[ 886.750110] Modules linked in:
[ 886.751396] CPU: 20 PID: 26219 Comm: trinity-c216 Tainted: G W 3.14.0-rc5-next-20140305-sasha-00011-ge06f5f3-dirty #105
[ 886.751396] task: ffff8800b6c80000 ti: ffff880228436000 task.ti: ffff880228436000
[ 886.751396] RIP: 0010:[<ffffffff812aab33>] [<ffffffff812aab33>] change_protection_range+0x3b3/0x500
[ 886.751396] RSP: 0000:ffff880228437da8 EFLAGS: 00010282
[ 886.751396] RAX: 8000000527c008e5 RBX: 00007f647916e000 RCX: 0000000000000000
[ 886.751396] RDX: ffff8802ef488e40 RSI: 00007f6479000000 RDI: 8000000527c008e5
[ 886.751396] RBP: ffff880228437e78 R08: 0000000000000000 R09: 0000000000000000
[ 886.751396] R10: 0000000000000001 R11: 0000000000000000 R12: ffff8802ef488e40
[ 886.751396] R13: 00007f6479000000 R14: 00007f647916e000 R15: 00007f646e34e000
[ 886.751396] FS: 00007f64b28d4700(0000) GS:ffff88052ba00000(0000) knlGS:0000000000000000
[ 886.751396] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[ 886.751396] CR2: 00007f64aed83af8 CR3: 0000000206e52000 CR4: 00000000000006a0
[ 886.751396] Stack:
[ 886.751396] ffff880200000001 ffffffff8447f152 ffff880228437dd8 ffff880228af3000
[ 886.769142] 00007f646e34e000 00007f647916dfff 0000000000000000 00007f647916e000
[ 886.769142] ffff880206e527f0 00007f647916dfff 0000000000000000 0000000000000000
[ 886.769142] Call Trace:
[ 886.769142] [<ffffffff8447f152>] ? preempt_count_sub+0xe2/0x120
[ 886.769142] [<ffffffff812aaca5>] change_protection+0x25/0x30
[ 886.769142] [<ffffffff812c3eeb>] change_prot_numa+0x1b/0x30
[ 886.769142] [<ffffffff8118df49>] task_numa_work+0x279/0x360
[ 886.769142] [<ffffffff8116c57e>] task_work_run+0xae/0xf0
[ 886.769142] [<ffffffff8106ffbe>] do_notify_resume+0x8e/0xe0
[ 886.769142] [<ffffffff8447a93b>] retint_signal+0x4d/0x92
[ 886.769142] Code: 49 8b 3c 24 48 83 3d fc 2e ba 04 00 75 12 0f 0b 0f 1f 84 00 00 00 00 00 eb fe 66 0f 1f 44 00 00 48 89 f8 66 66 66 90 84 c0 79 0d <0f> 0b 0f 1f 00 eb fe 66 0f 1f 44 00 00 8b 4d 9c 44 8b 4d 98
+89
[ 886.769142] RIP [<ffffffff812aab33>] change_protection_range+0x3b3/0x500
[ 886.769142] RSP <ffff880228437da8>
The VM_BUG_ON was added by the patch "mm,numa: reorganize
change_pmd_range". The race existed without the patch but was just harder
to hit.
The problem is that a transhuge check is made without holding the PTL.
It's possible at the time of the check that a parallel fault clears the
pmd and inserts a new one which then triggers the VM_BUG_ON check. This
patch removes the VM_BUG_ON but fixes the race by rechecking transhuge
under the PTL when marking page tables for NUMA hinting and bailing if a
race occurred. It is not a problem for calls to mprotect() as they hold
mmap_sem for write.
Signed-off-by: Mel Gorman <mgorman@xxxxxxx>
Reported-by: Sasha Levin <sasha.levin@xxxxxxxxxx>
Reviewed-by: Rik van Riel <riel@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
mm/mprotect.c | 36 ++++++++++++++++++++++++++++++++++--
1 file changed, 34 insertions(+), 2 deletions(-)
diff -puN mm/mprotect.c~mm-numa-recheck-for-transhuge-pages-under-lock-during-protection-changes mm/mprotect.c
--- a/mm/mprotect.c~mm-numa-recheck-for-transhuge-pages-under-lock-during-protection-changes
+++ a/mm/mprotect.c
@@ -36,6 +36,34 @@ static inline pgprot_t pgprot_modify(pgp
}
#endif
+/*
+ * For a prot_numa update we only hold mmap_sem for read so there is a
+ * potential race with faulting where a pmd was temporarily none. This
+ * function checks for a transhuge pmd under the appropriate lock. It
+ * returns a pte if it was successfully locked or NULL if it raced with
+ * a transhuge insertion.
+ */
+static pte_t *lock_pte_protection(struct vm_area_struct *vma, pmd_t *pmd,
+ unsigned long addr, int prot_numa, spinlock_t **ptl)
+{
+ pte_t *pte;
+ spinlock_t *pmdl;
+
+ /* !prot_numa is protected by mmap_sem held for write */
+ if (!prot_numa)
+ return pte_offset_map_lock(vma->vm_mm, pmd, addr, ptl);
+
+ pmdl = pmd_lock(vma->vm_mm, pmd);
+ if (unlikely(pmd_trans_huge(*pmd) || pmd_none(*pmd))) {
+ spin_unlock(pmdl);
+ return NULL;
+ }
+
+ pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, ptl);
+ spin_unlock(pmdl);
+ return pte;
+}
+
static unsigned long change_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
unsigned long addr, unsigned long end, pgprot_t newprot,
int dirty_accountable, int prot_numa)
@@ -45,7 +73,10 @@ static unsigned long change_pte_range(st
spinlock_t *ptl;
unsigned long pages = 0;
- pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
+ pte = lock_pte_protection(vma, pmd, addr, prot_numa, &ptl);
+ if (!pte)
+ return 0;
+
arch_enter_lazy_mmu_mode();
do {
oldpte = *pte;
@@ -132,12 +163,13 @@ static inline unsigned long change_pmd_r
pages += HPAGE_PMD_NR;
nr_huge_updates++;
}
+
+ /* huge pmd was handled */
continue;
}
}
/* fall through, the trans huge pmd just split */
}
- VM_BUG_ON(pmd_trans_huge(*pmd));
this_pages = change_pte_range(vma, pmd, addr, next, newprot,
dirty_accountable, prot_numa);
pages += this_pages;
_
Patches currently in -mm which might be from mgorman@xxxxxxx are
origin.patch
mm-vmscan-do-not-swap-anon-pages-just-because-freefile-is-low.patch
mm-introduce-do_shared_fault-and-drop-do_fault-fix-fix.patch
mm-compactionc-isolate_freepages_block-small-tuneup.patch
mm-only-force-scan-in-reclaim-when-none-of-the-lrus-are-big-enough.patch
do_shared_fault-check-that-mmap_sem-is-held.patch
--
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
