kconfig: fix missing fclose() on error paths

[PATCH] kconfig: fix missing fclose() on error paths

Posted by Masahiro Yamada 4 years, 4 months ago

The file is not closed when ferror() fails.

Fixes: 00d674cb3536 ("kconfig: refactor conf_write_dep()")
Fixes: 57ddd07c4560 ("kconfig: refactor conf_write_autoconf()")
Reported-by: Ryan Cai <ycaibb@gmail.com>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
---

 scripts/kconfig/confdata.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/scripts/kconfig/confdata.c b/scripts/kconfig/confdata.c
index 59717be31210..16897cb8cefd 100644
--- a/scripts/kconfig/confdata.c
+++ b/scripts/kconfig/confdata.c
@@ -979,10 +979,10 @@ static int conf_write_autoconf_cmd(const char *autoconf_name)
 
 	fprintf(out, "\n$(deps_config): ;\n");
 
-	if (ferror(out)) /* error check for all fprintf() calls */
-		return -1;
-
+	ret = ferror(out); /* error check for all fprintf() calls */
 	fclose(out);
+	if (ret)
+		return -1;
 
 	if (rename(tmp, name)) {
 		perror("rename");
@@ -1093,10 +1093,10 @@ static int __conf_write_autoconf(const char *filename,
 			print_symbol(file, sym);
 
 	/* check possible errors in conf_write_heading() and print_symbol() */
-	if (ferror(file))
-		return -1;
-
+	ret = ferror(file);
 	fclose(file);
+	if (ret)
+		return -1;
 
 	if (rename(tmp, filename)) {
 		perror("rename");
-- 
2.32.0

Re: [PATCH] kconfig: fix missing fclose() on error paths

Posted by Masahiro Yamada 4 years, 4 months ago

On Tue, Feb 8, 2022 at 3:26 PM Masahiro Yamada <masahiroy@kernel.org> wrote:
>
> The file is not closed when ferror() fails.
>
> Fixes: 00d674cb3536 ("kconfig: refactor conf_write_dep()")
> Fixes: 57ddd07c4560 ("kconfig: refactor conf_write_autoconf()")
> Reported-by: Ryan Cai <ycaibb@gmail.com>
> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
> ---


Applied to linux-kbuild/fixes.


>
>  scripts/kconfig/confdata.c | 12 ++++++------
>  1 file changed, 6 insertions(+), 6 deletions(-)
>
> diff --git a/scripts/kconfig/confdata.c b/scripts/kconfig/confdata.c
> index 59717be31210..16897cb8cefd 100644
> --- a/scripts/kconfig/confdata.c
> +++ b/scripts/kconfig/confdata.c
> @@ -979,10 +979,10 @@ static int conf_write_autoconf_cmd(const char *autoconf_name)
>
>         fprintf(out, "\n$(deps_config): ;\n");
>
> -       if (ferror(out)) /* error check for all fprintf() calls */
> -               return -1;
> -
> +       ret = ferror(out); /* error check for all fprintf() calls */
>         fclose(out);
> +       if (ret)
> +               return -1;
>
>         if (rename(tmp, name)) {
>                 perror("rename");
> @@ -1093,10 +1093,10 @@ static int __conf_write_autoconf(const char *filename,
>                         print_symbol(file, sym);
>
>         /* check possible errors in conf_write_heading() and print_symbol() */
> -       if (ferror(file))
> -               return -1;
> -
> +       ret = ferror(file);
>         fclose(file);
> +       if (ret)
> +               return -1;
>
>         if (rename(tmp, filename)) {
>                 perror("rename");
> --
> 2.32.0
>


-- 
Best Regards
Masahiro Yamada

RE: [PATCH] kconfig: fix missing fclose() on error paths

Posted by David Laight 4 years, 4 months ago

From: Masahiro Yamada
> Sent: 10 February 2022 00:54
> >
> > The file is not closed when ferror() fails.
> >
> > Fixes: 00d674cb3536 ("kconfig: refactor conf_write_dep()")
> > Fixes: 57ddd07c4560 ("kconfig: refactor conf_write_autoconf()")
> > Reported-by: Ryan Cai <ycaibb@gmail.com>
> > Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
> > ---
> 
> 
> Applied to linux-kbuild/fixes.
> 
> 
> >
> >  scripts/kconfig/confdata.c | 12 ++++++------
> >  1 file changed, 6 insertions(+), 6 deletions(-)
> >
> > diff --git a/scripts/kconfig/confdata.c b/scripts/kconfig/confdata.c
> > index 59717be31210..16897cb8cefd 100644
> > --- a/scripts/kconfig/confdata.c
> > +++ b/scripts/kconfig/confdata.c
> > @@ -979,10 +979,10 @@ static int conf_write_autoconf_cmd(const char *autoconf_name)
> >
> >         fprintf(out, "\n$(deps_config): ;\n");
> >
> > -       if (ferror(out)) /* error check for all fprintf() calls */
> > -               return -1;
> > -
> > +       ret = ferror(out); /* error check for all fprintf() calls */
> >         fclose(out);
> > +       if (ret)
> > +               return -1;

There's not much point calling ferror() unless you call fflush() first.

Similarly there's about zero point in checking the return from fprintf().
At least I've never found any distro adding 'must_check' to fprintf().
('must_not_check' would be more appropriate!)

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)

Re: [PATCH] kconfig: fix missing fclose() on error paths

Posted by Masahiro Yamada 4 years, 4 months ago

On Fri, Feb 11, 2022 at 8:47 PM David Laight <David.Laight@aculab.com> wrote:
>
> From: Masahiro Yamada
> > Sent: 10 February 2022 00:54
> > >
> > > The file is not closed when ferror() fails.
> > >
> > > Fixes: 00d674cb3536 ("kconfig: refactor conf_write_dep()")
> > > Fixes: 57ddd07c4560 ("kconfig: refactor conf_write_autoconf()")
> > > Reported-by: Ryan Cai <ycaibb@gmail.com>
> > > Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
> > > ---
> >
> >
> > Applied to linux-kbuild/fixes.
> >
> >
> > >
> > >  scripts/kconfig/confdata.c | 12 ++++++------
> > >  1 file changed, 6 insertions(+), 6 deletions(-)
> > >
> > > diff --git a/scripts/kconfig/confdata.c b/scripts/kconfig/confdata.c
> > > index 59717be31210..16897cb8cefd 100644
> > > --- a/scripts/kconfig/confdata.c
> > > +++ b/scripts/kconfig/confdata.c
> > > @@ -979,10 +979,10 @@ static int conf_write_autoconf_cmd(const char *autoconf_name)
> > >
> > >         fprintf(out, "\n$(deps_config): ;\n");
> > >
> > > -       if (ferror(out)) /* error check for all fprintf() calls */
> > > -               return -1;
> > > -
> > > +       ret = ferror(out); /* error check for all fprintf() calls */
> > >         fclose(out);
> > > +       if (ret)
> > > +               return -1;
>
> There's not much point calling ferror() unless you call fflush() first.
>
> Similarly there's about zero point in checking the return from fprintf().
> At least I've never found any distro adding 'must_check' to fprintf().
> ('must_not_check' would be more appropriate!)


Thanks for pointing it out. You are right.

I assume you are not opposed to this patch itself.

I will insert fflush() before ferror()
as a separate patch.





>         David
>
> -
> Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
> Registration No: 1397386 (Wales)



-- 
Best Regards
Masahiro Yamada