Re: [PATCH] squashfs: fix leaked page locks when page actor allocation fails

[Phillip Lougher <phillip@squashfs.org.uk>]

On 27/09/2026 07:05, Nguyen Ngoc Thang wrote:

> Hi Phillip,
>
> syzbot reports a hang in evict() during umount of a squashfs mount
> (INFO: task hung in evict (2)):
>
>    https://syzkaller.appspot.com/bug?extid=65b1e2d8f2d618a93e96
>
> Root cause: squashfs_readpage_block() grabs and locks the sibling pages
> of the block with grab_cache_page_nowait(), and only afterwards allocates
> the page actor.  If squashfs_page_actor_init_special() fails, the code
> does "goto out", which only kfree()s the page array. The locked, referenced
> sibling pages are never unlocked or put. A later truncate_inode_pages_range()
> from evict() blocks on the leaked folio lock forever.
>
> The ordering was introduced by f268eedddf35 ("squashfs: extend "page actor"
> to handle missing pages"); before it the actor was allocated first.
>
> Fix: use "goto mark_errored", which already unlocks and puts every page
> except the target page (owned by the caller, which unlocks it via
> folio_end_read()). res is still -ENOMEM at that point.
>
> Testing: QEMU (KVM, KASAN, lockdep, IMA tcb policy) running the syzbot C
> reproducer, with fail-nth swept over 1..400 in openat().  The failslab
> hit lands in squashfs_page_actor_init_special() from
> squashfs_readpage_block() via ima_calc_file_hash().
>    - before: hung task in evict()/truncate_inode_pages_range() after ~12s,
>      same trace as the syzbot report
>    - after: 170s of runs, 174 injected failures at that site, no hang
>
> The patch is below / attached.
>
> ----- patch -----
>  From f686b39f997dc20222f455c5ac6421f6a5d6c8d8 Mon Sep 17 00:00:00 2001
> From: Nguyen Ngoc Thang <ngocthang2710.1999@gmail.com>
> Date: Sun, 27 Sep 2026 12:40:12 +0700
> Subject: [PATCH] squashfs: fix leaked page locks when page actor allocation
>   fails
>
> squashfs_readpage_block() grabs and locks the sibling pages of the
> target block before it allocates the page actor.  If that allocation
> fails it jumps to "out", which only frees the page array, so the locked
> pages are never unlocked or released.
>
> A later truncate_inode_pages_range() on the inode, e.g. from evict() at
> umount, then waits forever on the leaked folio lock and the task hangs.
>
> Go through mark_errored instead, which unlocks and puts every page
> except the target, which the caller handles.
>
> Fixes: f268eedddf35 ("squashfs: extend "page actor" to handle missing pages")
> Reported-by: syzbot+65b1e2d8f2d618a93e96@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=65b1e2d8f2d618a93e96
> Cc: stable@vger.kernel.org
> Signed-off-by: Nguyen Ngoc Thang <ngocthang2710.1999@gmail.com>
> ---
>   fs/squashfs/file_direct.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/squashfs/file_direct.c b/fs/squashfs/file_direct.c
> index 2c3e809d6891..177357443543 100644
> --- a/fs/squashfs/file_direct.c
> +++ b/fs/squashfs/file_direct.c
> @@ -70,7 +70,7 @@ int squashfs_readpage_block(struct folio *folio, u64 block, int bsize,
>   	actor = squashfs_page_actor_init_special(msblk, page, pages, expected,
>   						start_index << PAGE_SHIFT);
>   	if (actor == NULL)
> -		goto out;
> +		goto mark_errored;
>   
>   	/* Decompress directly into the page cache buffers */
>   	res = squashfs_read_data(inode->i_sb, block, bsize, NULL, actor);

Sorry for the delay, I'm currently snowed under with AI reports and bugs;

But, it looks good.

Reviewed-by: Phillip Lougher <phillip@squashfs.org.uk 
<https://squashfs.org.uk>>