2008-10-16 10:03:04

by Manish Katiyar

Subject: [PATCH] ext4: Remove compilation warnings in fs/ext4/mballoc.c

Hi Ted,

Below patch fixes the following compilation warnings
fs/ext4/mballoc.c:2717: warning: unused variable 'proc'
fs/ext4/mballoc.c:2715: warning: unused variable 'mode'

I think a cleaner fix would be to call these functions only when
CONFIG_PROC_FS is set and cleanup EXT4_PROC_HANDLER to be always
defined . Also since s_proc in sb only makes sense when procfs is
configured should that also be ifdef'd in the structure itself ??
Let me know if these cleanup can be done in a better way (or the way
you would prefer :-) . I can do this.


Signed-off-by: Manish Katiyar <[email protected]>
---
fs/ext4/mballoc.c | 4 ++++
1 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index b580714..c88b5ab 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -2712,12 +2712,14 @@ ext4_mb_free_committed_blocks(struct super_block *sb)

static int ext4_mb_init_per_dev_proc(struct super_block *sb)
{
+#ifdef CONFIG_PROC_FS
mode_t mode = S_IFREG | S_IRUGO | S_IWUSR;
struct ext4_sb_info *sbi = EXT4_SB(sb);
struct proc_dir_entry *proc;

if (sbi->s_proc == NULL)
return -EINVAL;
+#endif

EXT4_PROC_HANDLER(EXT4_MB_STATS_NAME, mb_stats);
EXT4_PROC_HANDLER(EXT4_MB_MAX_TO_SCAN_NAME, mb_max_to_scan);
@@ -2739,10 +2741,12 @@ err_out:

static int ext4_mb_destroy_per_dev_proc(struct super_block *sb)
{
+#ifdef CONFIG_PROC_FS
struct ext4_sb_info *sbi = EXT4_SB(sb);

if (sbi->s_proc == NULL)
return -EINVAL;
+#endif

remove_proc_entry(EXT4_MB_GROUP_PREALLOC, sbi->s_proc);
remove_proc_entry(EXT4_MB_STREAM_REQ, sbi->s_proc);
--
1.5.4.3


Thanks -
Manish

2008-10-16 15:12:18

by Theodore Ts'o

Subject: Re: [PATCH] ext4: Remove compilation warnings in fs/ext4/mballoc.c

On Thu, Oct 16, 2008 at 03:33:02PM +0530, Manish Katiyar wrote:
> I think a cleaner fix would be to call these functions only when
> CONFIG_PROC_FS is set and cleanup EXT4_PROC_HANDLER to be always
> defined . Also since s_proc in sb only makes sense when procfs is
> configured should that also be ifdef'd in the structure itself ??
> Let me know if these cleanup can be done in a better way (or the way
> you would prefer :-) . I can do this.

What I put into the patch queue comments out all of
ext4_mb_init_per_dev_proc() and ext4_mb_destroy_per_dev_proc() if
CONFIG_PROC_FS is not defined. This not the common case, (even "make
allnoconfig" leaves CONFIG_PROC_FS defined), so don't think it's worth
messing with #ifdef'ing the structure, and adding to add even more
#ifdef's into the code. For a similar reason it's not worth it to add
#ifdef's to the call sites of these functions. It just clutters the
code more, and gcc should be able to optimize out a call to an empty
static function anyway.

- Ted