Re: general protection fault in khugepaged

[Kirill A. Shutemov]

On Fri, Jul 17, 2020 at 03:40:04PM -0700, syzbot wrote:
> syzbot has bisected this issue to:
> 
> commit ffe945e633b527d5a4577b42cbadec3c7cbcf096
> Author: Kirill A. Shutemov <kirill.shutemov@xxxxxxxxxxxxxxx>
> Date:   Wed Jun 3 23:00:09 2020 +0000
> 
>     khugepaged: do not stop collapse if less than half PTEs are referenced
> 
> bisection log:  https://syzkaller.appspot.com/x/bisect.txt?x=17356c2f100000
> start commit:   e44f65fd xen-netfront: remove redundant assignment to vari..
> git tree:       net-next
> final oops:     https://syzkaller.appspot.com/x/report.txt?x=14b56c2f100000
> console output: https://syzkaller.appspot.com/x/log.txt?x=10b56c2f100000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=829871134ca5e230
> dashboard link: https://syzkaller.appspot.com/bug?extid=ed318e8b790ca72c5ad0
> syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=113406a7100000
> C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=175597d3100000
> 
> Reported-by: syzbot+ed318e8b790ca72c5ad0@xxxxxxxxxxxxxxxxxxxxxxxxx
> Fixes: ffe945e633b5 ("khugepaged: do not stop collapse if less than half PTEs are referenced")
> 
> For information about bisection process see: https://goo.gl/tpsmEJ#bisection
> 

My guess is that anonymous VMA got replaced by file VMA under us.

Could you reproduce the issue with the patch below? I have hard time
triggering it myself.

diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index b043c40a21d4..f8ec58a1e72b 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -1107,6 +1107,8 @@ static void collapse_huge_page(struct mm_struct *mm,
 	if (mm_find_pmd(mm, address) != pmd)
 		goto out;
 
+	if (!vma->anon_vma)
+		dump_vma(vma);
 	anon_vma_lock_write(vma->anon_vma);
 
 	mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, NULL, mm,
-- 
 Kirill A. Shutemov