Re: [PATCH v2] xfs_scrub: Use POSIX-conformant strerror_r

["Darrick J. Wong" <djwong@kernel.org>]

On Fri, Sep 12, 2025 at 08:00:44AM -0700, Darrick J. Wong wrote:
> On Sat, Sep 06, 2025 at 03:12:07AM -0500, A. Wilcox wrote:
> > When building xfsprogs with musl libc, strerror_r returns int as
> > specified in POSIX.  This differs from the glibc extension that returns
> > char*.  Successful calls will return 0, which will be dereferenced as a
> > NULL pointer by (v)fprintf.
> > 
> > Signed-off-by: A. Wilcox <AWilcox@Wilcox-Tech.com>
> 
> Isn't C fun?
> Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>

Ohh yes it is, this patch broke the build for me:

common.c: In function ‘__str_out’:
common.c:129:17: error: ignoring return value of ‘strerror_r’ declared with attribute ‘warn_unused_result’ [-Werror=unused-result]
  129 |                 strerror_r(error, buf, DESCR_BUFSZ);
      |                 ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

<sigh> xfsprogs can't get the XSI version because it defines GNU_SOURCE,
and you can't shut up gcc by casting the whole expression to void.

Do you folks remove the -D_GNU_SOURCE from builddefs.in when building
against musl?  Or do you leave the definition alone, taking advantage of
the fact that #define'ing a symbol is not a guarantee of functionality?

--D

> --D
> 
> > ---
> >  scrub/common.c | 3 ++-
> >  1 file changed, 2 insertions(+), 1 deletion(-)
> > 
> > diff --git a/scrub/common.c b/scrub/common.c
> > index 14cd677b..9437d0ab 100644
> > --- a/scrub/common.c
> > +++ b/scrub/common.c
> > @@ -126,7 +126,8 @@ __str_out(
> >  	fprintf(stream, "%s%s: %s: ", stream_start(stream),
> >  			_(err_levels[level].string), descr);
> >  	if (error) {
> > -		fprintf(stream, _("%s."), strerror_r(error, buf, DESCR_BUFSZ));
> > +		strerror_r(error, buf, DESCR_BUFSZ);
> > +		fprintf(stream, _("%s."), buf);
> >  	} else {
> >  		va_start(args, format);
> >  		vfprintf(stream, format, args);
> > -- 
> > 2.49.0
> > 
> > 
>