[syzbot] [tmpfs?] kernel BUG in folio_flags

[syzbot]

Hello,

syzbot found the following issue on:

HEAD commit: 0a093b2893c7 Add linux-next specific files for 20230112
git tree: linux-next
console output: https://syzkaller.appspot.com/x/log.txt?x=15c67c5e480000
kernel config: https://syzkaller.appspot.com/x/.config?x=835f3591019836d5
dashboard link: https://syzkaller.appspot.com/bug?extid=fa6f2ee9039b87ef86c4
compiler: gcc (Debian 10.2.1-6) 10.2.1 20210110, GNU ld (GNU Binutils for Debian) 2.35.2

Unfortunately, I don't have any reproducer for this issue yet.

Downloadable assets:
disk image: https://storage.googleapis.com/syzbot-assets/8111a570d6cb/disk-0a093b28.raw.xz
vmlinux: https://storage.googleapis.com/syzbot-assets/ecc135b7fc9a/vmlinux-0a093b28.xz
kernel image: https://storage.googleapis.com/syzbot-assets/ca8d73b446ea/bzImage-0a093b28.xz

IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: syzbot+fa6f2e...@syzkaller.appspotmail.com

dentry_kill fs/dcache.c:745 [inline]
dput+0x6ac/0xe10 fs/dcache.c:913
__fput+0x3cc/0xa90 fs/file_table.c:329
task_work_run+0x16f/0x270 kernel/task_work.c:179
exit_task_work include/linux/task_work.h:38 [inline]
do_exit+0xb17/0x2a90 kernel/exit.c:867
do_group_exit+0xd4/0x2a0 kernel/exit.c:1012
get_signal+0x225f/0x24f0 kernel/signal.c:2859
------------[ cut here ]------------
kernel BUG at include/linux/page-flags.h:317!
invalid opcode: 0000 [#1] PREEMPT SMP KASAN
CPU: 0 PID: 10782 Comm: syz-executor.1 Not tainted 6.2.0-rc3-next-20230112-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/26/2022
RIP: 0010:folio_flags.constprop.0+0x122/0x150 include/linux/page-flags.h:317
Code: 52 ff ff ff e8 ff f2 cd ff 48 8d 43 ff 49 39 c4 0f 84 40 ff ff ff e8 ed f2 cd ff 48 c7 c6 40 0b 57 8a 4c 89 e7 e8 2e e4 05 00 <0f> 0b e8 57 c2 1b 00 e9 fe fe ff ff e8 4d c2 1b 00 eb a9 4c 89 e7
RSP: 0018:ffffc90005437b00 EFLAGS: 00010246
RAX: 0000000000040000 RBX: 0000000000000001 RCX: ffffc90014e22000
RDX: 0000000000040000 RSI: ffffffff81b3ca02 RDI: 0000000000000000
RBP: 0000000000000004 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000001 R11: 0000000000000001 R12: ffffea0000b28100
R13: 0000000000000000 R14: dffffc0000000000 R15: 1ffffd400016502c
FS: 00007f1e422fd700(0000) GS:ffff8880b9800000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00005642ce910130 CR3: 000000001cef4000 CR4: 00000000003506f0
Call Trace:
<TASK>
folio_test_head include/linux/page-flags.h:780 [inline]
folio_test_large include/linux/page-flags.h:801 [inline]
folio_nr_pages include/linux/mm.h:1735 [inline]
compound_nr include/linux/mm.h:1747 [inline]
zero_user_segments.constprop.0+0x262/0x350 include/linux/highmem.h:288
shmem_write_end+0x684/0x780 mm/shmem.c:2600
generic_perform_write+0x316/0x570 mm/filemap.c:3783
__generic_file_write_iter+0x2ae/0x500 mm/filemap.c:3900
generic_file_write_iter+0xe3/0x350 mm/filemap.c:3932
call_write_iter include/linux/fs.h:1852 [inline]
new_sync_write fs/read_write.c:491 [inline]
vfs_write+0x9ed/0xe10 fs/read_write.c:584
ksys_write+0x12b/0x250 fs/read_write.c:637
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x39/0xb0 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x63/0xcd
RIP: 0033:0x7f1e4168c0c9
Code: 28 00 00 00 75 05 48 83 c4 28 c3 e8 f1 19 00 00 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f1e422fd168 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
RAX: ffffffffffffffda RBX: 00007f1e417abf80 RCX: 00007f1e4168c0c9
RDX: 0000000000000004 RSI: 00000000200000c0 RDI: 0000000000000003
RBP: 00007f1e416e7ae9 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
R13: 00007ffc7e6ac66f R14: 00007f1e422fd300 R15: 0000000000022000
</TASK>
Modules linked in:
---[ end trace 0000000000000000 ]---
RIP: 0010:folio_flags.constprop.0+0x122/0x150 include/linux/page-flags.h:317
Code: 52 ff ff ff e8 ff f2 cd ff 48 8d 43 ff 49 39 c4 0f 84 40 ff ff ff e8 ed f2 cd ff 48 c7 c6 40 0b 57 8a 4c 89 e7 e8 2e e4 05 00 <0f> 0b e8 57 c2 1b 00 e9 fe fe ff ff e8 4d c2 1b 00 eb a9 4c 89 e7
RSP: 0018:ffffc90005437b00 EFLAGS: 00010246
RAX: 0000000000040000 RBX: 0000000000000001 RCX: ffffc90014e22000
RDX: 0000000000040000 RSI: ffffffff81b3ca02 RDI: 0000000000000000
RBP: 0000000000000004 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000001 R11: 0000000000000001 R12: ffffea0000b28100
R13: 0000000000000000 R14: dffffc0000000000 R15: 1ffffd400016502c
FS: 00007f1e422fd700(0000) GS:ffff8880b9900000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00005642ce910130 CR3: 000000001cef4000 CR4: 00000000003506e0


---
This report is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at syzk...@googlegroups.com.

syzbot will keep track of this issue. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.

[Matthew Wilcox]

Oh, good one. shmem isn't actually doing anything wrong here, but
it'll be nice to fix it. (net -10 lines)

What's happening is that shmem is deliberately calling
zero_user_segments() on a tail page. zero_user_segments() calls
compound_nr() which used to return 1 when called on a tail page but now
trips this assert. I didn't intend this change of behaviour, so
let's start by fixing that.

Andrew, this is a fix for "mm: Reimplement compound_nr()". I'll send
the shmem change separately since it's not a bug fix.

diff --git a/include/linux/mm.h b/include/linux/mm.h
index 57d702fc8677..78f751f3ba5b 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -1741,10 +1741,22 @@ static inline long folio_nr_pages(struct folio *folio)
#endif
}

-/* Returns the number of pages in this potentially compound page. */
+/*
+ * compound_nr() returns the number of pages in this potentially compound
+ * page. compound_nr() can be called on a tail page, and is defined to
+ * return 1 in that case.
+ */
static inline unsigned long compound_nr(struct page *page)
{
- return folio_nr_pages((struct folio *)page);
+ struct folio *folio = (struct folio *)page;
+
+ if (!test_bit(PG_head, &folio->flags))
+ return 1;
+#ifdef CONFIG_64BIT
+ return folio->_folio_nr_pages;
+#else
+ return 1L << folio->_folio_order;
+#endif
}

/**
@@ -1753,8 +1765,7 @@ static inline unsigned long compound_nr(struct page *page)
*/
static inline int thp_nr_pages(struct page *page)
{
- VM_BUG_ON_PGFLAGS(PageTail(page), page);
- return compound_nr(page);
+ return folio_nr_pages((struct folio *)page);
}

/**