[PATCH v2] btrfs: don't BUG_ON() NOCOW ordered-extents with checksum list

Johannes Thumshirn posted 1 patch 1 month, 3 weeks ago
There is a newer version of this series
fs/btrfs/inode.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
[PATCH v2] btrfs: don't BUG_ON() NOCOW ordered-extents with checksum list
Posted by Johannes Thumshirn 1 month, 3 weeks ago
From: Johannes Thumshirn <johannes.thumshirn@wdc.com>

Currently we BUG_ON() in btrfs_finish_one_ordered() if we finishing an
ordered-extent that is flagged as NOCOW, but it's checsum list is non-empty.

This is clearly a logic error which we can recover from by aborting the
transaction.

For developer builds which enable CONFIG_BTRFS_ASSERT, also ASSERT() that the
list is empty.

Suggested-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
---
Changes to v1:
* Fixup if () and ASSERT() (Qu)
* Fix spelling of 'Currently'
---
 fs/btrfs/inode.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 103ec917ca9d..e57b73943ab8 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -3088,7 +3088,10 @@ int btrfs_finish_one_ordered(struct btrfs_ordered_extent *ordered_extent)
 
 	if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) {
 		/* Logic error */
-		BUG_ON(!list_empty(&ordered_extent->list));
+		if (!list_empty(&ordered_extent->list)) {
+			ASSERT(list_empty(&ordered_extent->list));
+			btrfs_abort_transaction(trans, -EINVAL);
+		}
 
 		btrfs_inode_safe_disk_i_size_write(inode, 0);
 		ret = btrfs_update_inode_fallback(trans, inode);
-- 
2.43.0
Re: [PATCH v2] btrfs: don't BUG_ON() NOCOW ordered-extents with checksum list
Posted by Filipe Manana 1 month, 3 weeks ago
On Fri, Oct 4, 2024 at 11:53 AM Johannes Thumshirn <jth@kernel.org> wrote:
>
> From: Johannes Thumshirn <johannes.thumshirn@wdc.com>
>
> Currently we BUG_ON() in btrfs_finish_one_ordered() if we finishing an
> ordered-extent that is flagged as NOCOW, but it's checsum list is non-empty.
>
> This is clearly a logic error which we can recover from by aborting the
> transaction.
>
> For developer builds which enable CONFIG_BTRFS_ASSERT, also ASSERT() that the
> list is empty.
>
> Suggested-by: Filipe Manana <fdmanana@suse.com>
> Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
> ---
> Changes to v1:
> * Fixup if () and ASSERT() (Qu)
> * Fix spelling of 'Currently'
> ---
>  fs/btrfs/inode.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
> index 103ec917ca9d..e57b73943ab8 100644
> --- a/fs/btrfs/inode.c
> +++ b/fs/btrfs/inode.c
> @@ -3088,7 +3088,10 @@ int btrfs_finish_one_ordered(struct btrfs_ordered_extent *ordered_extent)
>
>         if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) {
>                 /* Logic error */
> -               BUG_ON(!list_empty(&ordered_extent->list));
> +               if (!list_empty(&ordered_extent->list)) {
> +                       ASSERT(list_empty(&ordered_extent->list));

I find this confusing and not so easy to grasp immediately. It's the
same as older places where we have:

if (unexpected condition) {
   ASSERT(0);
   (...)
}

I find it more natural and less confusing to just do:

ASSERT(list_empty(&ordered_extent->list));
if (unlikely(!list_empty(&ordered_extent->list))) {
    ret = -EINVAL;
    btrfs_abort_transaction(trans, ret);
    goto out;
}

> +                       btrfs_abort_transaction(trans, -EINVAL);
> +               }

This also misses setting 'ret' to the error and the goto into the
label 'out', as I've placed in the example above.

Thanks.

>
>                 btrfs_inode_safe_disk_i_size_write(inode, 0);
>                 ret = btrfs_update_inode_fallback(trans, inode);
> --
> 2.43.0
>
>
Re: [PATCH v2] btrfs: don't BUG_ON() NOCOW ordered-extents with checksum list
Posted by Qu Wenruo 1 month, 3 weeks ago

在 2024/10/4 20:23, Johannes Thumshirn 写道:
> From: Johannes Thumshirn <johannes.thumshirn@wdc.com>
>
> Currently we BUG_ON() in btrfs_finish_one_ordered() if we finishing an
> ordered-extent that is flagged as NOCOW, but it's checsum list is non-empty.
>
> This is clearly a logic error which we can recover from by aborting the
> transaction.
>
> For developer builds which enable CONFIG_BTRFS_ASSERT, also ASSERT() that the
> list is empty.
>
> Suggested-by: Filipe Manana <fdmanana@suse.com>
> Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>

Reviewed-by: Qu Wenruo <wqu@suse.com>

Thanks,
Qu
> ---
> Changes to v1:
> * Fixup if () and ASSERT() (Qu)
> * Fix spelling of 'Currently'
> ---
>   fs/btrfs/inode.c | 5 ++++-
>   1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
> index 103ec917ca9d..e57b73943ab8 100644
> --- a/fs/btrfs/inode.c
> +++ b/fs/btrfs/inode.c
> @@ -3088,7 +3088,10 @@ int btrfs_finish_one_ordered(struct btrfs_ordered_extent *ordered_extent)
>
>   	if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) {
>   		/* Logic error */
> -		BUG_ON(!list_empty(&ordered_extent->list));
> +		if (!list_empty(&ordered_extent->list)) {
> +			ASSERT(list_empty(&ordered_extent->list));
> +			btrfs_abort_transaction(trans, -EINVAL);
> +		}
>
>   		btrfs_inode_safe_disk_i_size_write(inode, 0);
>   		ret = btrfs_update_inode_fallback(trans, inode);