Re: general protection fault in khugepaged

["Kirill A. Shutemov" <kirill@shutemov.name>]

On Sat, Jul 18, 2020 at 04:54:09PM +0800, Hillf Danton wrote:
> 
> Wed, 08 Jul 2020 23:43:19 -0700
> > syzbot found the following crash on:
> > 
> > HEAD commit:    e44f65fd xen-netfront: remove redundant assignment to vari..
> > git tree:       net-next
> > console output: https://syzkaller.appspot.com/x/log.txt?x=15de54a7100000
> > kernel config:  https://syzkaller.appspot.com/x/.config?x=829871134ca5e230
> > dashboard link: https://syzkaller.appspot.com/bug?extid=ed318e8b790ca72c5ad0
> > compiler:       gcc (GCC) 10.1.0-syz 20200507
> > syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=113406a7100000
> > C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=175597d3100000
> > 
> > IMPORTANT: if you fix the bug, please add the following tag to the commit:
> > Reported-by: syzbot+ed318e8b790ca72c5ad0@syzkaller.appspotmail.com
> > 
> > general protection fault, probably for non-canonical address 0xdffffc0000000000: 0000 [#1] PREEMPT SMP KASAN
> > KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007]
> > CPU: 1 PID: 1155 Comm: khugepaged Not tainted 5.8.0-rc2-syzkaller #0
> > Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
> > RIP: 0010:anon_vma_lock_write include/linux/rmap.h:120 [inline]
> > RIP: 0010:collapse_huge_page mm/khugepaged.c:1110 [inline]
> > RIP: 0010:khugepaged_scan_pmd mm/khugepaged.c:1349 [inline]
> > RIP: 0010:khugepaged_scan_mm_slot mm/khugepaged.c:2110 [inline]
> > RIP: 0010:khugepaged_do_scan mm/khugepaged.c:2193 [inline]
> > RIP: 0010:khugepaged+0x3bba/0x5a10 mm/khugepaged.c:2238
> > Code: 01 00 00 48 8d bb 88 00 00 00 48 89 f8 48 c1 e8 03 42 80 3c 30 00 0f 85 fa 0f 00 00 48 8b 9b 88 00 00 00 48 89 d8 48 c1 e8 03 <42> 80 3c 30 00 0f 85 d4 0f 00 00 48 8b 3b 48 83 c7 08 e8 9f ff 30
> > RSP: 0018:ffffc90004be7c80 EFLAGS: 00010246
> > RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffffffff81a72d8b
> > RDX: ffff8880a69d8100 RSI: ffffffff81b7606b RDI: ffff88809f0577c0
> > RBP: 0000000000000000 R08: 0000000000000000 R09: ffff8881ff213a7f
> > R10: 0000000000000080 R11: 0000000000000000 R12: ffffffff8aae6110
> > R13: ffffc90004be7de0 R14: dffffc0000000000 R15: 0000000020000000
> > FS:  0000000000000000(0000) GS:ffff8880ae700000(0000) knlGS:0000000000000000
> > CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> > CR2: 0000000000000000 CR3: 00000001fe0cf000 CR4: 00000000001406e0
> > DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> > DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
> > Call Trace:
> >  kthread+0x3b5/0x4a0 kernel/kthread.c:291
> >  ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:293
> 
> Fixes: 99cb0dbd47a1 ("mm,thp: add read-only THP support for (non-shmem) FS")

It looks sane, but I think it's easier to check in
hugepage_vma_revalidate(). Something like this (untested):

diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index b043c40a21d4..700f5160f3e4 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -958,6 +958,9 @@ static int hugepage_vma_revalidate(struct mm_struct *mm, unsigned long address,
 		return SCAN_ADDRESS_RANGE;
 	if (!hugepage_vma_check(vma, vma->vm_flags))
 		return SCAN_VMA_CHECK;
+	/* Anon VMA expected */
+	if (!vma->anon_vma || vma->vm_ops)
+		return SCAN_VMA_CHECK;
 	return 0;
 }
 
-- 
 Kirill A. Shutemov