Re: [syzbot] [mm?] possible deadlock in f2fs_file_write_iter

[Eric Biggers]

On Mon, Jun 26, 2023 at 08:10:45AM -0700, syzbot wrote:
> Hello,
> 
> syzbot found the following issue on:
> 
> HEAD commit:    c87d46a9e8eb Add linux-next specific files for 20230622
> git tree:       linux-next
> console output: https://syzkaller.appspot.com/x/log.txt?x=129edf4b280000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=d8ac8dd33677e8e0
> dashboard link: https://syzkaller.appspot.com/bug?extid=b67b270114065129f12b
> 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/bdf5ea2db527/disk-c87d46a9.raw.xz
> vmlinux: https://storage.googleapis.com/syzbot-assets/3201e576e2ad/vmlinux-c87d46a9.xz
> kernel image: https://storage.googleapis.com/syzbot-assets/d011d9fa4e63/bzImage-c87d46a9.xz
> 
> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> Reported-by: syzbot+b67b270114065129f12b@xxxxxxxxxxxxxxxxxxxxxxxxx
> 
> F2FS-fs (loop5): Mounted with checkpoint version = 48b305e5
> ======================================================
> WARNING: possible circular locking dependency detected
> 6.4.0-rc7-next-20230622-syzkaller #0 Not tainted
> ------------------------------------------------------
> syz-executor.5/16876 is trying to acquire lock:
> ffff88802a033120 (&mm->mmap_lock){++++}-{3:3}, at: percpu_ref_put include/linux/percpu-refcount.h:351 [inline]
> ffff88802a033120 (&mm->mmap_lock){++++}-{3:3}, at: put_dev_pagemap include/linux/memremap.h:251 [inline]
> ffff88802a033120 (&mm->mmap_lock){++++}-{3:3}, at: gup_pte_range mm/gup.c:2583 [inline]
> ffff88802a033120 (&mm->mmap_lock){++++}-{3:3}, at: gup_pmd_range mm/gup.c:2950 [inline]
> ffff88802a033120 (&mm->mmap_lock){++++}-{3:3}, at: gup_pud_range mm/gup.c:2978 [inline]
> ffff88802a033120 (&mm->mmap_lock){++++}-{3:3}, at: gup_p4d_range mm/gup.c:3003 [inline]
> ffff88802a033120 (&mm->mmap_lock){++++}-{3:3}, at: gup_pgd_range mm/gup.c:3031 [inline]
> ffff88802a033120 (&mm->mmap_lock){++++}-{3:3}, at: lockless_pages_from_mm mm/gup.c:3084 [inline]
> ffff88802a033120 (&mm->mmap_lock){++++}-{3:3}, at: internal_get_user_pages_fast+0x135f/0x3370 mm/gup.c:3133
> 
> but task is already holding lock:
> ffff88808b20ea30 (&fi->i_gc_rwsem[WRITE]){++++}-{3:3}, at: f2fs_down_read fs/f2fs/f2fs.h:2107 [inline]
> ffff88808b20ea30 (&fi->i_gc_rwsem[WRITE]){++++}-{3:3}, at: f2fs_dio_write_iter fs/f2fs/file.c:4643 [inline]
> ffff88808b20ea30 (&fi->i_gc_rwsem[WRITE]){++++}-{3:3}, at: f2fs_file_write_iter+0x1161/0x2500 fs/f2fs/file.c:4766
> 

This was caused by the patch
"f2fs: fix to avoid mmap vs set_compress_option case"
(https://lore.kernel.org/linux-f2fs-devel/20230529104709.2560779-1-chao@xxxxxxxxxx/)
which has been dropped.

#syz set subsystems: f2fs
#syz invalid

- Eric