2023-09-29 06:18:25

by Deepak R Varma

Subject: [PATCH] gfs2: Set error on error path

Set the error variable inside the error path on failure. Saves
unnecessary variable assignment during normal execution.

Signed-off-by: Deepak R Varma <[email protected]>
---
fs/gfs2/bmap.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/fs/gfs2/bmap.c b/fs/gfs2/bmap.c
index ef7017fb6951..93bd8ea34444 100644
--- a/fs/gfs2/bmap.c
+++ b/fs/gfs2/bmap.c
@@ -162,9 +162,10 @@ int gfs2_unstuff_dinode(struct gfs2_inode *ip)

down_write(&ip->i_rw_mutex);
page = grab_cache_page(inode->i_mapping, 0);
- error = -ENOMEM;
- if (!page)
+ if (!page) {
+ error = -ENOMEM;
goto out;
+ }
error = __gfs2_unstuff_inode(ip, page);
unlock_page(page);
put_page(page);
--
2.39.2

2023-09-29 07:34:25

by Deepak R Varma

Subject: Re: [PATCH] gfs2: Set error on error path

On Fri, 2023-09-29 at 10:14 +0300, Dan Carpenter wrote:
> On Fri, Sep 29, 2023 at 11:44:40AM +0530, Deepak R Varma wrote:
> > Set the error variable inside the error path on failure. Saves
> > unnecessary variable assignment during normal execution.
> >
> > Signed-off-by: Deepak R Varma <[email protected]>
> > ---
> >  fs/gfs2/bmap.c | 5 +++--
> >  1 file changed, 3 insertions(+), 2 deletions(-)
> >
> > diff --git a/fs/gfs2/bmap.c b/fs/gfs2/bmap.c
> > index ef7017fb6951..93bd8ea34444 100644
> > --- a/fs/gfs2/bmap.c
> > +++ b/fs/gfs2/bmap.c
> > @@ -162,9 +162,10 @@ int gfs2_unstuff_dinode(struct gfs2_inode *ip)
> >  
> >         down_write(&ip->i_rw_mutex);
> >         page = grab_cache_page(inode->i_mapping, 0);
> > -       error = -ENOMEM;
> > -       if (!page)
> > +       if (!page) {
> > +               error = -ENOMEM;
> >                 goto out;
> > +       }
>
> It used to be that the other way was faster?  I'm pretty sure the
> compiler can figure it out these days.  Please don't randomly start
> changing people's prefered style unless it's drivers/staging/ code.
>
Hi Dan,
Thank you for the feedback. I did not know about the compiler
optimization resulting in original code being faster.
I will let the maintainers comment on breaking the style.

regards,
deepak.

> regards,
> dan carpenter
>

2023-09-29 23:31:54

by Dan Carpenter

Subject: Re: [PATCH] gfs2: Set error on error path

On Fri, Sep 29, 2023 at 11:44:40AM +0530, Deepak R Varma wrote:
> Set the error variable inside the error path on failure. Saves
> unnecessary variable assignment during normal execution.
>
> Signed-off-by: Deepak R Varma <[email protected]>
> ---
> fs/gfs2/bmap.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/fs/gfs2/bmap.c b/fs/gfs2/bmap.c
> index ef7017fb6951..93bd8ea34444 100644
> --- a/fs/gfs2/bmap.c
> +++ b/fs/gfs2/bmap.c
> @@ -162,9 +162,10 @@ int gfs2_unstuff_dinode(struct gfs2_inode *ip)
>
> down_write(&ip->i_rw_mutex);
> page = grab_cache_page(inode->i_mapping, 0);
> - error = -ENOMEM;
> - if (!page)
> + if (!page) {
> + error = -ENOMEM;
> goto out;
> + }

It used to be that the other way was faster? I'm pretty sure the
compiler can figure it out these days. Please don't randomly start
changing people's prefered style unless it's drivers/staging/ code.

regards,
dan carpenter