Re: Linux 2.6.35.2

[Linus Torvalds]

On Sat, Aug 14, 2010 at 5:37 AM, Sedat Dilek <sedat.dilek@xxxxxxxxxxxxxx> wrote:
>
> root@tbox:~# grep "Aug 14 13:00:5" /var/log/syslog
> Aug 14 13:00:53 tbox kernel: [ 1084.804086] BUG: scheduling while atomic: bash/3256/0x10000001

Gaah. Locking problem. We really should have let this one simmer in
-devel a bit longer before rushing it into -stable.

Attached potential patch. Untested, but I forgot about the locking in
the first version of the patch, and then the "trivial one-liner fix"
missed the deeper problem.

Greg - please don't use this patch yet. I'll test it. And wait for
verification that it fixes other peoblems too.

                    Linus
 mm/memory.c |   13 ++++++-------
 1 files changed, 6 insertions(+), 7 deletions(-)

diff --git a/mm/memory.c b/mm/memory.c
index 9b3b73f..b6e5fd2 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -2792,24 +2792,23 @@ static int do_anonymous_page(struct mm_struct *mm, struct vm_area_struct *vma,
 	spinlock_t *ptl;
 	pte_t entry;
 
-	if (check_stack_guard_page(vma, address) < 0) {
-		pte_unmap(page_table);
+	pte_unmap(page_table);
+
+	/* Check if we need to add a guard page to the stack */
+	if (check_stack_guard_page(vma, address) < 0)
 		return VM_FAULT_SIGBUS;
-	}
 
+	/* Use the zero-page for reads */
 	if (!(flags & FAULT_FLAG_WRITE)) {
 		entry = pte_mkspecial(pfn_pte(my_zero_pfn(address),
 						vma->vm_page_prot));
-		ptl = pte_lockptr(mm, pmd);
-		spin_lock(ptl);
+		page_table = pte_offset_map_lock(mm, pmd, address, &ptl);
 		if (!pte_none(*page_table))
 			goto unlock;
 		goto setpte;
 	}
 
 	/* Allocate our own private page. */
-	pte_unmap(page_table);
-
 	if (unlikely(anon_vma_prepare(vma)))
 		goto oom;
 	page = alloc_zeroed_user_highpage_movable(vma, address);