[PATCH v2] ext2: Make sbi an proper alias for EXT2_SB(sb) in ext2_fill_super()

[PATCH v2] ext2: Make sbi an proper alias for EXT2_SB(sb) in ext2_fill_super()

[Nathan Chancellor]

A recent LLVM change to fix parsing and validating __guarded_by on
structure members [1] reveals a couple of warnings in ext2_fill_super()
from the annotations added by commit 023051c2d0f1 ("ext2: mark
s_mount_state as guarded by s_lock") and commit c23746524faf ("ext2:
annotate s_rsv_window_root as requiring s_rsv_window_lock"):

  fs/ext2/super.c:1040:7: error: writing variable 's_mount_state' requires holding spinlock '_res->s_lock' exclusively [-Werror,-Wthread-safety-precise]
   1040 |         sbi->s_mount_state = le16_to_cpu(es->s_state);
        |              ^
  fs/ext2/super.c:1040:7: note: found near match 'EXT2_SB(sb).s_lock'
  fs/ext2/super.c:1138:8: error: writing variable 's_rsv_window_root' requires holding spinlock '_res->s_rsv_window_lock' exclusively [-Werror,-Wthread-safety-precise]
   1138 |                 sbi->s_rsv_window_root = RB_ROOT;
        |                      ^
  fs/ext2/super.c:1138:8: note: found near match 'EXT2_SB(sb).s_rsv_window_lock'

Context analysis does very minimal alias detection [2], so it is not
able to figure out that sbi == sb->s_fs_info == EXT2_SB(sb). The only
reason there were no warnings before the above LLVM change is because
the annotation was not parsed and verified correctly.

Rework the allocation and assignment of sbi to have a direct assignment
of EXT2_SB(sb) to sbi so that the analysis knows that sbi and
EXT2_SB(sb) refer to the same object, clearing up the warnings.

Fixes: 9708538ba204 ("ext2: enable context analysis support for ext2 filesystem")
Link: https://github.com/llvm/llvm-project/commit/217e229a1f32785fccdd3b3f51429f0d309dea45 [1]
Link: https://clang.llvm.org/docs/ThreadSafetyAnalysis.html#no-alias-analysis [2]
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---
Changes in v2:
- Fix allocation size (Sashiko). I thought that I checked that they
  were equivalent but I guess not :/
- Link to v1: https://patch.msgid.link/20261023-ext2-fix-context-analysis-clang-24-v1-1-2b35c33cd0fa@kernel.org
---
 fs/ext2/super.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/fs/ext2/super.c b/fs/ext2/super.c
index fb217ef2d69f..bde5b433776c 100644
--- a/fs/ext2/super.c
+++ b/fs/ext2/super.c
@@ -883,17 +883,18 @@ static int ext2_fill_super(struct super_block *sb, struct fs_context *fc)
 	__le32 features;
 	int err;
 
-	sbi = kzalloc_obj(*sbi);
-	if (!sbi)
+	sb->s_fs_info = kzalloc_obj(*sbi);
+	if (!sb->s_fs_info)
 		return -ENOMEM;
+	sbi = EXT2_SB(sb);
 
 	sbi->s_blockgroup_lock =
 		kzalloc_obj(struct blockgroup_lock);
 	if (!sbi->s_blockgroup_lock) {
-		kfree(sbi);
+		kfree(sb->s_fs_info);
+		sb->s_fs_info = NULL;
 		return -ENOMEM;
 	}
-	sb->s_fs_info = sbi;
 	sbi->s_sb_block = sb_block;
 
 	guard(spinlock_init)(&EXT2_SB(sb)->s_lock);

---
base-commit: 782654a3e9c4065dea2e9f10539899d348b7a1fc
base-tree: git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
change-id: 20261023-ext2-fix-context-analysis-clang-24-504b5dd53038

Best regards,
--  
Cheers,
Nathan

Re: [PATCH v2] ext2: Make sbi an proper alias for EXT2_SB(sb) in ext2_fill_super()

[Marco Elver]

On Fri, 23 Oct 2026 at 23:25, Nathan Chancellor <nathan@kernel.org> wrote:
>
> A recent LLVM change to fix parsing and validating __guarded_by on
> structure members [1] reveals a couple of warnings in ext2_fill_super()
> from the annotations added by commit 023051c2d0f1 ("ext2: mark
> s_mount_state as guarded by s_lock") and commit c23746524faf ("ext2:
> annotate s_rsv_window_root as requiring s_rsv_window_lock"):
>
>   fs/ext2/super.c:1040:7: error: writing variable 's_mount_state' requires holding spinlock '_res->s_lock' exclusively [-Werror,-Wthread-safety-precise]
>    1040 |         sbi->s_mount_state = le16_to_cpu(es->s_state);
>         |              ^
>   fs/ext2/super.c:1040:7: note: found near match 'EXT2_SB(sb).s_lock'
>   fs/ext2/super.c:1138:8: error: writing variable 's_rsv_window_root' requires holding spinlock '_res->s_rsv_window_lock' exclusively [-Werror,-Wthread-safety-precise]
>    1138 |                 sbi->s_rsv_window_root = RB_ROOT;
>         |                      ^
>   fs/ext2/super.c:1138:8: note: found near match 'EXT2_SB(sb).s_rsv_window_lock'
>
> Context analysis does very minimal alias detection [2], so it is not
> able to figure out that sbi == sb->s_fs_info == EXT2_SB(sb). The only

Right, though function-local aliases are resolved. We just aren't able
to figure out aliases that propagate through non-local variables.

> reason there were no warnings before the above LLVM change is because
> the annotation was not parsed and verified correctly.
>
> Rework the allocation and assignment of sbi to have a direct assignment
> of EXT2_SB(sb) to sbi so that the analysis knows that sbi and
> EXT2_SB(sb) refer to the same object, clearing up the warnings.
>
> Fixes: 9708538ba204 ("ext2: enable context analysis support for ext2 filesystem")
> Link: https://github.com/llvm/llvm-project/commit/217e229a1f32785fccdd3b3f51429f0d309dea45 [1]
> Link: https://clang.llvm.org/docs/ThreadSafetyAnalysis.html#no-alias-analysis [2]
> Signed-off-by: Nathan Chancellor <nathan@kernel.org>

Reviewed-by: Marco Elver <elver@google.com>

Additional note: This was uncovered by an unstable Clang 24
(unreleased) change, which was meant to make the analysis more precise
[1], yet stumbled over this. This change is needed to prepare for
Clang 24.

Thanks!

> ---
> Changes in v2:
> - Fix allocation size (Sashiko). I thought that I checked that they
>   were equivalent but I guess not :/
> - Link to v1: https://patch.msgid.link/20261023-ext2-fix-context-analysis-clang-24-v1-1-2b35c33cd0fa@kernel.org
> ---
>  fs/ext2/super.c | 9 +++++----
>  1 file changed, 5 insertions(+), 4 deletions(-)
>
> diff --git a/fs/ext2/super.c b/fs/ext2/super.c
> index fb217ef2d69f..bde5b433776c 100644
> --- a/fs/ext2/super.c
> +++ b/fs/ext2/super.c
> @@ -883,17 +883,18 @@ static int ext2_fill_super(struct super_block *sb, struct fs_context *fc)
>         __le32 features;
>         int err;
>
> -       sbi = kzalloc_obj(*sbi);
> -       if (!sbi)
> +       sb->s_fs_info = kzalloc_obj(*sbi);
> +       if (!sb->s_fs_info)
>                 return -ENOMEM;
> +       sbi = EXT2_SB(sb);
>
>         sbi->s_blockgroup_lock =
>                 kzalloc_obj(struct blockgroup_lock);
>         if (!sbi->s_blockgroup_lock) {
> -               kfree(sbi);
> +               kfree(sb->s_fs_info);
> +               sb->s_fs_info = NULL;
>                 return -ENOMEM;
>         }
> -       sb->s_fs_info = sbi;
>         sbi->s_sb_block = sb_block;
>
>         guard(spinlock_init)(&EXT2_SB(sb)->s_lock);
>
> ---
> base-commit: 782654a3e9c4065dea2e9f10539899d348b7a1fc
> base-tree: git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
> change-id: 20261023-ext2-fix-context-analysis-clang-24-504b5dd53038
>
> Best regards,
> --
> Cheers,
> Nathan
>

Re: [PATCH v2] ext2: Make sbi an proper alias for EXT2_SB(sb) in ext2_fill_super()

[Nathan Chancellor]

On Mon, Oct 26, 2026 at 11:08:30AM +0100, Marco Elver wrote:
> Right, though function-local aliases are resolved. We just aren't able
> to figure out aliases that propagate through non-local variables.

Ah okay thanks, I'll keep that in mind in case it comes up in the
future.

-- 
Cheers,
Nathan

Re: [PATCH v2] ext2: Make sbi an proper alias for EXT2_SB(sb) in ext2_fill_super()

[Jan Kara]

On Fri 23-10-26 14:25:02, Nathan Chancellor wrote:
> A recent LLVM change to fix parsing and validating __guarded_by on
> structure members [1] reveals a couple of warnings in ext2_fill_super()
> from the annotations added by commit 023051c2d0f1 ("ext2: mark
> s_mount_state as guarded by s_lock") and commit c23746524faf ("ext2:
> annotate s_rsv_window_root as requiring s_rsv_window_lock"):
> 
>   fs/ext2/super.c:1040:7: error: writing variable 's_mount_state' requires holding spinlock '_res->s_lock' exclusively [-Werror,-Wthread-safety-precise]
>    1040 |         sbi->s_mount_state = le16_to_cpu(es->s_state);
>         |              ^
>   fs/ext2/super.c:1040:7: note: found near match 'EXT2_SB(sb).s_lock'
>   fs/ext2/super.c:1138:8: error: writing variable 's_rsv_window_root' requires holding spinlock '_res->s_rsv_window_lock' exclusively [-Werror,-Wthread-safety-precise]
>    1138 |                 sbi->s_rsv_window_root = RB_ROOT;
>         |                      ^
>   fs/ext2/super.c:1138:8: note: found near match 'EXT2_SB(sb).s_rsv_window_lock'
> 
> Context analysis does very minimal alias detection [2], so it is not
> able to figure out that sbi == sb->s_fs_info == EXT2_SB(sb). The only
> reason there were no warnings before the above LLVM change is because
> the annotation was not parsed and verified correctly.
> 
> Rework the allocation and assignment of sbi to have a direct assignment
> of EXT2_SB(sb) to sbi so that the analysis knows that sbi and
> EXT2_SB(sb) refer to the same object, clearing up the warnings.
> 
> Fixes: 9708538ba204 ("ext2: enable context analysis support for ext2 filesystem")
> Link: https://github.com/llvm/llvm-project/commit/217e229a1f32785fccdd3b3f51429f0d309dea45 [1]
> Link: https://clang.llvm.org/docs/ThreadSafetyAnalysis.html#no-alias-analysis [2]
> Signed-off-by: Nathan Chancellor <nathan@kernel.org>

OK, if this helps you :). Added to my tree, I plan to push it to Linus for
rc2.

								Honza

> ---
> Changes in v2:
> - Fix allocation size (Sashiko). I thought that I checked that they
>   were equivalent but I guess not :/
> - Link to v1: https://patch.msgid.link/20261023-ext2-fix-context-analysis-clang-24-v1-1-2b35c33cd0fa@kernel.org
> ---
>  fs/ext2/super.c | 9 +++++----
>  1 file changed, 5 insertions(+), 4 deletions(-)
> 
> diff --git a/fs/ext2/super.c b/fs/ext2/super.c
> index fb217ef2d69f..bde5b433776c 100644
> --- a/fs/ext2/super.c
> +++ b/fs/ext2/super.c
> @@ -883,17 +883,18 @@ static int ext2_fill_super(struct super_block *sb, struct fs_context *fc)
>  	__le32 features;
>  	int err;
>  
> -	sbi = kzalloc_obj(*sbi);
> -	if (!sbi)
> +	sb->s_fs_info = kzalloc_obj(*sbi);
> +	if (!sb->s_fs_info)
>  		return -ENOMEM;
> +	sbi = EXT2_SB(sb);
>  
>  	sbi->s_blockgroup_lock =
>  		kzalloc_obj(struct blockgroup_lock);
>  	if (!sbi->s_blockgroup_lock) {
> -		kfree(sbi);
> +		kfree(sb->s_fs_info);
> +		sb->s_fs_info = NULL;
>  		return -ENOMEM;
>  	}
> -	sb->s_fs_info = sbi;
>  	sbi->s_sb_block = sb_block;
>  
>  	guard(spinlock_init)(&EXT2_SB(sb)->s_lock);
> 
> ---
> base-commit: 782654a3e9c4065dea2e9f10539899d348b7a1fc
> base-tree: git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
> change-id: 20261023-ext2-fix-context-analysis-clang-24-504b5dd53038
> 
> Best regards,
> --  
> Cheers,
> Nathan
> 
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

Re: [PATCH v2] ext2: Make sbi an proper alias for EXT2_SB(sb) in ext2_fill_super()

[Nathan Chancellor]

On Mon, Oct 26, 2026 at 01:32:34PM +0100, Jan Kara wrote:
> OK, if this helps you :). Added to my tree, I plan to push it to Linus for
> rc2.

Great, that would help a lot, thanks!

-- 
Cheers,
Nathan