* [PATCH v2] xfs: compile out v4 support if disabled
@ 2024-03-25 3:13 Christoph Hellwig
2024-03-26 0:02 ` Darrick J. Wong
0 siblings, 1 reply; 4+ messages in thread
From: Christoph Hellwig @ 2024-03-25 3:13 UTC (permalink / raw)
To: chandan.babu, djwong; +Cc: linux-xfs
Add a few strategic IS_ENABLED statements to let the compiler eliminate
unused code when CONFIG_XFS_SUPPORT_V4 is disabled.
This saves multiple kilobytes of .text in my .config:
$ size xfs.o.*
text data bss dec hex filename
1363633 294836 592 1659061 1950b5 xfs.o.new
1371453 294868 592 1666913 196f61 xfs.o.old
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
Changes since v1:
- compile out the right bits, and more of them
fs/xfs/xfs_mount.h | 40 +++++++++++++++++++++++++++++++---------
fs/xfs/xfs_super.c | 22 +++++++++++++---------
2 files changed, 44 insertions(+), 18 deletions(-)
diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
index 0e8d7779c0a561..85e327bf05041f 100644
--- a/fs/xfs/xfs_mount.h
+++ b/fs/xfs/xfs_mount.h
@@ -337,19 +337,10 @@ static inline void xfs_add_ ## name (struct xfs_mount *mp) \
__XFS_ADD_FEAT(attr, ATTR)
__XFS_HAS_FEAT(nlink, NLINK)
__XFS_ADD_FEAT(quota, QUOTA)
-__XFS_HAS_FEAT(align, ALIGN)
__XFS_HAS_FEAT(dalign, DALIGN)
-__XFS_HAS_FEAT(logv2, LOGV2)
__XFS_HAS_FEAT(sector, SECTOR)
-__XFS_HAS_FEAT(extflg, EXTFLG)
__XFS_HAS_FEAT(asciici, ASCIICI)
-__XFS_HAS_FEAT(lazysbcount, LAZYSBCOUNT)
-__XFS_ADD_FEAT(attr2, ATTR2)
__XFS_HAS_FEAT(parent, PARENT)
-__XFS_ADD_FEAT(projid32, PROJID32)
-__XFS_HAS_FEAT(crc, CRC)
-__XFS_HAS_FEAT(v3inodes, V3INODES)
-__XFS_HAS_FEAT(pquotino, PQUOTINO)
__XFS_HAS_FEAT(ftype, FTYPE)
__XFS_HAS_FEAT(finobt, FINOBT)
__XFS_HAS_FEAT(rmapbt, RMAPBT)
@@ -362,6 +353,37 @@ __XFS_HAS_FEAT(bigtime, BIGTIME)
__XFS_HAS_FEAT(needsrepair, NEEDSREPAIR)
__XFS_HAS_FEAT(large_extent_counts, NREXT64)
+/*
+ * Some features are always on for v5 file systems, allow the compiler to
+ * eliminiate dead code when building without v4 support.
+ */
+#define __XFS_HAS_V4_FEAT(name, NAME) \
+static inline bool xfs_has_ ## name (struct xfs_mount *mp) \
+{ \
+ return !IS_ENABLED(CONFIG_XFS_SUPPORT_V4) || \
+ (mp->m_features & XFS_FEAT_ ## NAME); \
+}
+
+#define __XFS_ADD_V4_FEAT(name, NAME) \
+ __XFS_HAS_V4_FEAT(name, NAME); \
+static inline void xfs_add_ ## name (struct xfs_mount *mp) \
+{ \
+ if (IS_ENABLED(CONFIG_XFS_SUPPORT_V4)) { \
+ mp->m_features |= XFS_FEAT_ ## NAME; \
+ xfs_sb_version_add ## name(&mp->m_sb); \
+ } \
+}
+
+__XFS_HAS_V4_FEAT(align, ALIGN)
+__XFS_HAS_V4_FEAT(logv2, LOGV2)
+__XFS_HAS_V4_FEAT(extflg, EXTFLG)
+__XFS_HAS_V4_FEAT(lazysbcount, LAZYSBCOUNT)
+__XFS_ADD_V4_FEAT(attr2, ATTR2)
+__XFS_ADD_V4_FEAT(projid32, PROJID32)
+__XFS_HAS_V4_FEAT(v3inodes, V3INODES)
+__XFS_HAS_V4_FEAT(crc, CRC)
+__XFS_HAS_V4_FEAT(pquotino, PQUOTINO)
+
/*
* Mount features
*
diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
index 71732457583370..f325a2d1b3a902 100644
--- a/fs/xfs/xfs_super.c
+++ b/fs/xfs/xfs_super.c
@@ -1589,17 +1589,21 @@ xfs_fs_fill_super(
if (error)
goto out_free_sb;
- /* V4 support is undergoing deprecation. */
- if (!xfs_has_crc(mp)) {
-#ifdef CONFIG_XFS_SUPPORT_V4
+ /*
+ * V4 support is undergoing deprecation.
+ *
+ * Note: this has to use an open coded m_features check as xfs_has_crc
+ * always returns false for !CONFIG_XFS_SUPPORT_V4.
+ */
+ if (!(mp->m_features & XFS_FEAT_CRC)) {
+ if (!IS_ENABLED(CONFIG_XFS_SUPPORT_V4)) {
+ xfs_warn(mp,
+ "Deprecated V4 format (crc=0) not supported by kernel.");
+ error = -EINVAL;
+ goto out_free_sb;
+ }
xfs_warn_once(mp,
"Deprecated V4 format (crc=0) will not be supported after September 2030.");
-#else
- xfs_warn(mp,
- "Deprecated V4 format (crc=0) not supported by kernel.");
- error = -EINVAL;
- goto out_free_sb;
-#endif
}
/* ASCII case insensitivity is undergoing deprecation. */
--
2.39.2
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH v2] xfs: compile out v4 support if disabled
2024-03-25 3:13 [PATCH v2] xfs: compile out v4 support if disabled Christoph Hellwig
@ 2024-03-26 0:02 ` Darrick J. Wong
2024-03-26 5:50 ` Christoph Hellwig
0 siblings, 1 reply; 4+ messages in thread
From: Darrick J. Wong @ 2024-03-26 0:02 UTC (permalink / raw)
To: Christoph Hellwig; +Cc: chandan.babu, linux-xfs
On Mon, Mar 25, 2024 at 11:13:18AM +0800, Christoph Hellwig wrote:
> Add a few strategic IS_ENABLED statements to let the compiler eliminate
> unused code when CONFIG_XFS_SUPPORT_V4 is disabled.
>
> This saves multiple kilobytes of .text in my .config:
>
> $ size xfs.o.*
> text data bss dec hex filename
> 1363633 294836 592 1659061 1950b5 xfs.o.new
> 1371453 294868 592 1666913 196f61 xfs.o.old
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---
>
> Changes since v1:
> - compile out the right bits, and more of them
>
> fs/xfs/xfs_mount.h | 40 +++++++++++++++++++++++++++++++---------
> fs/xfs/xfs_super.c | 22 +++++++++++++---------
> 2 files changed, 44 insertions(+), 18 deletions(-)
>
> diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
> index 0e8d7779c0a561..85e327bf05041f 100644
> --- a/fs/xfs/xfs_mount.h
> +++ b/fs/xfs/xfs_mount.h
> @@ -337,19 +337,10 @@ static inline void xfs_add_ ## name (struct xfs_mount *mp) \
> __XFS_ADD_FEAT(attr, ATTR)
> __XFS_HAS_FEAT(nlink, NLINK)
> __XFS_ADD_FEAT(quota, QUOTA)
> -__XFS_HAS_FEAT(align, ALIGN)
> __XFS_HAS_FEAT(dalign, DALIGN)
> -__XFS_HAS_FEAT(logv2, LOGV2)
> __XFS_HAS_FEAT(sector, SECTOR)
> -__XFS_HAS_FEAT(extflg, EXTFLG)
> __XFS_HAS_FEAT(asciici, ASCIICI)
> -__XFS_HAS_FEAT(lazysbcount, LAZYSBCOUNT)
> -__XFS_ADD_FEAT(attr2, ATTR2)
> __XFS_HAS_FEAT(parent, PARENT)
> -__XFS_ADD_FEAT(projid32, PROJID32)
> -__XFS_HAS_FEAT(crc, CRC)
> -__XFS_HAS_FEAT(v3inodes, V3INODES)
> -__XFS_HAS_FEAT(pquotino, PQUOTINO)
> __XFS_HAS_FEAT(ftype, FTYPE)
> __XFS_HAS_FEAT(finobt, FINOBT)
> __XFS_HAS_FEAT(rmapbt, RMAPBT)
> @@ -362,6 +353,37 @@ __XFS_HAS_FEAT(bigtime, BIGTIME)
> __XFS_HAS_FEAT(needsrepair, NEEDSREPAIR)
> __XFS_HAS_FEAT(large_extent_counts, NREXT64)
>
> +/*
> + * Some features are always on for v5 file systems, allow the compiler to
> + * eliminiate dead code when building without v4 support.
> + */
> +#define __XFS_HAS_V4_FEAT(name, NAME) \
Shouldn't this be called __XFS_HAS_V5_FEAT?
> +static inline bool xfs_has_ ## name (struct xfs_mount *mp) \
> +{ \
> + return !IS_ENABLED(CONFIG_XFS_SUPPORT_V4) || \
> + (mp->m_features & XFS_FEAT_ ## NAME); \
> +}
> +
> +#define __XFS_ADD_V4_FEAT(name, NAME) \
> + __XFS_HAS_V4_FEAT(name, NAME); \
This too?
The rest of the patch looks ok. :)
--D
> +static inline void xfs_add_ ## name (struct xfs_mount *mp) \
> +{ \
> + if (IS_ENABLED(CONFIG_XFS_SUPPORT_V4)) { \
> + mp->m_features |= XFS_FEAT_ ## NAME; \
> + xfs_sb_version_add ## name(&mp->m_sb); \
> + } \
> +}
> +
> +__XFS_HAS_V4_FEAT(align, ALIGN)
> +__XFS_HAS_V4_FEAT(logv2, LOGV2)
> +__XFS_HAS_V4_FEAT(extflg, EXTFLG)
> +__XFS_HAS_V4_FEAT(lazysbcount, LAZYSBCOUNT)
> +__XFS_ADD_V4_FEAT(attr2, ATTR2)
> +__XFS_ADD_V4_FEAT(projid32, PROJID32)
> +__XFS_HAS_V4_FEAT(v3inodes, V3INODES)
> +__XFS_HAS_V4_FEAT(crc, CRC)
> +__XFS_HAS_V4_FEAT(pquotino, PQUOTINO)
> +
> /*
> * Mount features
> *
> diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
> index 71732457583370..f325a2d1b3a902 100644
> --- a/fs/xfs/xfs_super.c
> +++ b/fs/xfs/xfs_super.c
> @@ -1589,17 +1589,21 @@ xfs_fs_fill_super(
> if (error)
> goto out_free_sb;
>
> - /* V4 support is undergoing deprecation. */
> - if (!xfs_has_crc(mp)) {
> -#ifdef CONFIG_XFS_SUPPORT_V4
> + /*
> + * V4 support is undergoing deprecation.
> + *
> + * Note: this has to use an open coded m_features check as xfs_has_crc
> + * always returns false for !CONFIG_XFS_SUPPORT_V4.
> + */
> + if (!(mp->m_features & XFS_FEAT_CRC)) {
> + if (!IS_ENABLED(CONFIG_XFS_SUPPORT_V4)) {
> + xfs_warn(mp,
> + "Deprecated V4 format (crc=0) not supported by kernel.");
> + error = -EINVAL;
> + goto out_free_sb;
> + }
> xfs_warn_once(mp,
> "Deprecated V4 format (crc=0) will not be supported after September 2030.");
> -#else
> - xfs_warn(mp,
> - "Deprecated V4 format (crc=0) not supported by kernel.");
> - error = -EINVAL;
> - goto out_free_sb;
> -#endif
> }
>
> /* ASCII case insensitivity is undergoing deprecation. */
> --
> 2.39.2
>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v2] xfs: compile out v4 support if disabled
2024-03-26 0:02 ` Darrick J. Wong
@ 2024-03-26 5:50 ` Christoph Hellwig
2024-03-26 17:24 ` Darrick J. Wong
0 siblings, 1 reply; 4+ messages in thread
From: Christoph Hellwig @ 2024-03-26 5:50 UTC (permalink / raw)
To: Darrick J. Wong; +Cc: Christoph Hellwig, chandan.babu, linux-xfs
On Mon, Mar 25, 2024 at 05:02:37PM -0700, Darrick J. Wong wrote:
> > +/*
> > + * Some features are always on for v5 file systems, allow the compiler to
> > + * eliminiate dead code when building without v4 support.
> > + */
> > +#define __XFS_HAS_V4_FEAT(name, NAME) \
>
> Shouldn't this be called __XFS_HAS_V5_FEAT?
They are features for v4 and unconditional for v5. Given that Dave
came up with the same names independently they can't be all bad,
although I don't really care very strongly.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v2] xfs: compile out v4 support if disabled
2024-03-26 5:50 ` Christoph Hellwig
@ 2024-03-26 17:24 ` Darrick J. Wong
0 siblings, 0 replies; 4+ messages in thread
From: Darrick J. Wong @ 2024-03-26 17:24 UTC (permalink / raw)
To: Christoph Hellwig; +Cc: chandan.babu, linux-xfs
On Tue, Mar 26, 2024 at 06:50:47AM +0100, Christoph Hellwig wrote:
> On Mon, Mar 25, 2024 at 05:02:37PM -0700, Darrick J. Wong wrote:
> > > +/*
> > > + * Some features are always on for v5 file systems, allow the compiler to
> > > + * eliminiate dead code when building without v4 support.
> > > + */
> > > +#define __XFS_HAS_V4_FEAT(name, NAME) \
> >
> > Shouldn't this be called __XFS_HAS_V5_FEAT?
>
> They are features for v4 and unconditional for v5. Given that Dave
> came up with the same names independently they can't be all bad,
> although I don't really care very strongly.
Ok, you've both persuaded me then.
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
--D
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2024-03-26 17:24 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-03-25 3:13 [PATCH v2] xfs: compile out v4 support if disabled Christoph Hellwig
2024-03-26 0:02 ` Darrick J. Wong
2024-03-26 5:50 ` Christoph Hellwig
2024-03-26 17:24 ` Darrick J. Wong
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox