Re: [PATCH v4 01/16] fs: Add case sensitivity info to file_kattr

[Christian Brauner <brauner@kernel.org>]

On Wed, Jan 14, 2026 at 12:01:30PM -0800, Darrick J. Wong wrote:
> On Wed, Jan 14, 2026 at 02:01:14PM -0500, Chuck Lever wrote:
> > 
> > 
> > On Wed, Jan 14, 2026, at 1:11 PM, Jan Kara wrote:
> > > On Wed 14-01-26 09:28:44, Chuck Lever wrote:
> > >> From: Chuck Lever <chuck.lever@oracle.com>
> > >> 
> > >> Enable upper layers such as NFSD to retrieve case sensitivity
> > >> information from file systems by adding case_insensitive and
> > >> case_nonpreserving boolean fields to struct file_kattr.
> > >> 
> > >> These fields default to false (POSIX semantics: case-sensitive and
> > >> case-preserving), allowing filesystems to set them only when
> > >> behavior differs from the default.
> > >> 
> > >> Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
> > > ...
> > >> diff --git a/include/uapi/linux/fs.h b/include/uapi/linux/fs.h
> > >> index 66ca526cf786..07286d34b48b 100644
> > >> --- a/include/uapi/linux/fs.h
> > >> +++ b/include/uapi/linux/fs.h
> > >> @@ -229,10 +229,20 @@ struct file_attr {
> > >>  	__u32 fa_nextents;	/* nextents field value (get)   */
> > >>  	__u32 fa_projid;	/* project identifier (get/set) */
> > >>  	__u32 fa_cowextsize;	/* CoW extsize field value (get/set) */
> > >> +	/* VER1 additions: */
> > >> +	__u32 fa_case_behavior;	/* case sensitivity (get) */
> > >> +	__u32 fa_reserved;
> > >>  };
> > >>  
> > >>  #define FILE_ATTR_SIZE_VER0 24
> > >> -#define FILE_ATTR_SIZE_LATEST FILE_ATTR_SIZE_VER0
> > >> +#define FILE_ATTR_SIZE_VER1 32
> > >> +#define FILE_ATTR_SIZE_LATEST FILE_ATTR_SIZE_VER1
> > >> +
> > >> +/*
> > >> + * Case sensitivity flags for fa_case_behavior
> > >> + */
> > >> +#define FS_CASE_INSENSITIVE	0x00000001	/* case-insensitive lookups */
> > >> +#define FS_CASE_NONPRESERVING	0x00000002	/* case not preserved */
> > >
> > > This is a matter of taste so not sure what others think about it but
> > > file_attr already have fa_xflags field and there is already one flag which
> > > doesn't directly correspond to on-disk representation (FS_XFLAG_HASATTR) so
> > > we could also put the two new flags in there... I have hard time imagining
> > > fa_case_behavior would grow past the two flags you've introduced so u32
> > > seems a bit wasteful.
> > 
> > No problem. I'll wait for additional guidance on this.
> 
> Sounds like a better use of space in struct file_attr than adding
> another pair of u32.

Fine by me as well.