2021-02-22 17:18:45

by Jan Kara

Subject: [PATCH] ext4: Add checks to xattr code that we have appropriate reclaim context

Syzbot is reporting that ext4 can enter fs reclaim from kvmalloc() while
the transaction is started like:

fs_reclaim_acquire+0x117/0x150 mm/page_alloc.c:4340
might_alloc include/linux/sched/mm.h:193 [inline]
slab_pre_alloc_hook mm/slab.h:493 [inline]
slab_alloc_node mm/slub.c:2817 [inline]
__kmalloc_node+0x5f/0x430 mm/slub.c:4015
kmalloc_node include/linux/slab.h:575 [inline]
kvmalloc_node+0x61/0xf0 mm/util.c:587
kvmalloc include/linux/mm.h:781 [inline]
ext4_xattr_inode_cache_find fs/ext4/xattr.c:1465 [inline]
ext4_xattr_inode_lookup_create fs/ext4/xattr.c:1508 [inline]
ext4_xattr_set_entry+0x1ce6/0x3780 fs/ext4/xattr.c:1649
ext4_xattr_ibody_set+0x78/0x2b0 fs/ext4/xattr.c:2224
ext4_xattr_set_handle+0x8f4/0x13e0 fs/ext4/xattr.c:2380
ext4_xattr_set+0x13a/0x340 fs/ext4/xattr.c:2493

This should be impossible since transaction start sets PF_MEMALLOC_NOFS.
Add some assertions to the code to catch if something isn't working as
expected early.

Link: https://lore.kernel.org/linux-ext4/[email protected]/
Signed-off-by: Jan Kara <[email protected]>
---
fs/ext4/xattr.c | 4 ++++
1 file changed, 4 insertions(+)

Hi Ted!

This is the debug patch we talked about on our last ext4 call. I hope we can
carry it for some time and see if syzbot hits any of the warnings.

Honza

diff --git a/fs/ext4/xattr.c b/fs/ext4/xattr.c
index 372208500f4e..083c95126781 100644
--- a/fs/ext4/xattr.c
+++ b/fs/ext4/xattr.c
@@ -1462,6 +1462,9 @@ ext4_xattr_inode_cache_find(struct inode *inode, const void *value,
if (!ce)
return NULL;

+ WARN_ON_ONCE(ext4_handle_valid(journal_current_handle()) &&
+ !(current->flags & PF_MEMALLOC_NOFS));
+
ea_data = kvmalloc(value_len, GFP_KERNEL);
if (!ea_data) {
mb_cache_entry_put(ea_inode_cache, ce);
@@ -2327,6 +2330,7 @@ ext4_xattr_set_handle(handle_t *handle, struct inode *inode, int name_index,
error = -ENOSPC;
goto cleanup;
}
+ WARN_ON_ONCE(!(current->flags & PF_MEMALLOC_NOFS));
}

error = ext4_reserve_inode_write(handle, inode, &is.iloc);
--
2.26.2

2021-03-05 14:30:56

by Theodore Ts'o

Subject: Re: [PATCH] ext4: Add checks to xattr code that we have appropriate reclaim context

On Mon, Feb 22, 2021 at 06:16:26PM +0100, Jan Kara wrote:
> Syzbot is reporting that ext4 can enter fs reclaim from kvmalloc() while
> the transaction is started like:
>
> fs_reclaim_acquire+0x117/0x150 mm/page_alloc.c:4340
> might_alloc include/linux/sched/mm.h:193 [inline]
> slab_pre_alloc_hook mm/slab.h:493 [inline]
> slab_alloc_node mm/slub.c:2817 [inline]
> __kmalloc_node+0x5f/0x430 mm/slub.c:4015
> kmalloc_node include/linux/slab.h:575 [inline]
> kvmalloc_node+0x61/0xf0 mm/util.c:587
> kvmalloc include/linux/mm.h:781 [inline]
> ext4_xattr_inode_cache_find fs/ext4/xattr.c:1465 [inline]
> ext4_xattr_inode_lookup_create fs/ext4/xattr.c:1508 [inline]
> ext4_xattr_set_entry+0x1ce6/0x3780 fs/ext4/xattr.c:1649
> ext4_xattr_ibody_set+0x78/0x2b0 fs/ext4/xattr.c:2224
> ext4_xattr_set_handle+0x8f4/0x13e0 fs/ext4/xattr.c:2380
> ext4_xattr_set+0x13a/0x340 fs/ext4/xattr.c:2493
>
> This should be impossible since transaction start sets PF_MEMALLOC_NOFS.
> Add some assertions to the code to catch if something isn't working as
> expected early.
>
> Link: https://lore.kernel.org/linux-ext4/[email protected]/
> Signed-off-by: Jan Kara <[email protected]>

Thanks, applied.

- Ted