2011-06-23 15:08:27

by Maarten Lankhorst

Subject: [PATCH] xfs: Silence bounds checking compiler warning

gcc with -Warray-bounds generates a false positive on this
since xfs defines the struct with u8 name[1]; to be able to
add a tag at the end.

Signed-off-by: Maarten Lankhorst <[email protected]>
---
fs/xfs/xfs_dir2_block.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/xfs/xfs_dir2_block.c b/fs/xfs/xfs_dir2_block.c
index 580d99c..09b4aed 100644
--- a/fs/xfs/xfs_dir2_block.c
+++ b/fs/xfs/xfs_dir2_block.c
@@ -1148,7 +1148,7 @@ xfs_dir2_sf_to_block(
((char *)block + XFS_DIR2_DATA_DOTDOT_OFFSET);
dep->inumber = cpu_to_be64(xfs_dir2_sf_get_inumber(sfp, &sfp->hdr.parent));
dep->namelen = 2;
- dep->name[0] = dep->name[1] = '.';
+ dep->name[0] = ((__u8 *)dep->name)[1] = '.';
tagp = xfs_dir2_data_entry_tag_p(dep);
*tagp = cpu_to_be16((char *)dep - (char *)block);
xfs_dir2_data_log_entry(tp, bp, dep);
--
1.7.5.4

2011-06-23 15:33:55

by Maarten Lankhorst

Subject: Re: [PATCH] xfs: Silence bounds checking compiler warning

On 06/23/2011 05:08 PM, Maarten Lankhorst wrote:
> gcc with -Warray-bounds generates a false positive on this
> since xfs defines the struct with u8 name[1]; to be able to
> add a tag at the end.
>
> Signed-off-by: Maarten Lankhorst <[email protected]>
> ---
> fs/xfs/xfs_dir2_block.c | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/fs/xfs/xfs_dir2_block.c b/fs/xfs/xfs_dir2_block.c
> index 580d99c..09b4aed 100644
> --- a/fs/xfs/xfs_dir2_block.c
> +++ b/fs/xfs/xfs_dir2_block.c
> @@ -1148,7 +1148,7 @@ xfs_dir2_sf_to_block(
> ((char *)block + XFS_DIR2_DATA_DOTDOT_OFFSET);
> dep->inumber = cpu_to_be64(xfs_dir2_sf_get_inumber(sfp, &sfp->hdr.parent));
> dep->namelen = 2;
> - dep->name[0] = dep->name[1] = '.';
> + dep->name[0] = ((__u8 *)dep->name)[1] = '.';
> tagp = xfs_dir2_data_entry_tag_p(dep);
> *tagp = cpu_to_be16((char *)dep - (char *)block);
> xfs_dir2_data_log_entry(tp, bp, dep);
Oops, that still gives a warning, I used (u8*) first which works, but it seems the type
conversion was what made it shut up. Is there any type I can cast __u8 to safely,
or is memset preferred? gcc was too smart and saw through my __u8 cast.

~Maarten