diff mbox series

fstests: workaround for gcc-15

Message ID 20250117043709.2941857-1-zlang@kernel.org (mailing list archive)
State New
Headers show
Series fstests: workaround for gcc-15 | expand

Commit Message

Zorro Lang Jan. 17, 2025, 4:37 a.m. UTC
GCC-15 does a big change, it changes the default language version for
C compilation from -std=gnu17 to -std=gnu23. That cause lots of "old
style" C codes hit build errors. On the other word, current xfstests
can't be used with GCC-15. So -std=gnu17 can help that.

Signed-off-by: Zorro Lang <zlang@kernel.org>
---

Hi,

I send this patch just for talking about this issue. The upcoming gcc-15
does lots of changes, a big change is using C23 by default:

  https://gcc.gnu.org/gcc-15/porting_to.html

xfstests has many old style C codes, they hard to be built with gcc-15.
So we have to either add -std=$old_version (likes this patch), or port
the code to C23.

This patch is just a workaround (and a reminder for someone might hit
this issue with gcc-15 too). If you have any good suggestions or experience
(for this kind of issue) to share, feel free to reply.

Thanks,
Zorro

 include/builddefs.in | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Darrick J. Wong Jan. 17, 2025, 5:27 p.m. UTC | #1
On Fri, Jan 17, 2025 at 12:37:09PM +0800, Zorro Lang wrote:
> GCC-15 does a big change, it changes the default language version for
> C compilation from -std=gnu17 to -std=gnu23. That cause lots of "old
> style" C codes hit build errors. On the other word, current xfstests
> can't be used with GCC-15. So -std=gnu17 can help that.
> 
> Signed-off-by: Zorro Lang <zlang@kernel.org>
> ---
> 
> Hi,
> 
> I send this patch just for talking about this issue. The upcoming gcc-15
> does lots of changes, a big change is using C23 by default:
> 
>   https://gcc.gnu.org/gcc-15/porting_to.html
> 
> xfstests has many old style C codes, they hard to be built with gcc-15.
> So we have to either add -std=$old_version (likes this patch), or port
> the code to C23.
> 
> This patch is just a workaround (and a reminder for someone might hit
> this issue with gcc-15 too). If you have any good suggestions or experience
> (for this kind of issue) to share, feel free to reply.

-std=gnu11 to match the kernel and xfsprogs?

--D

> Thanks,
> Zorro
> 
>  include/builddefs.in | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/include/builddefs.in b/include/builddefs.in
> index 5b5864278..ef124bb87 100644
> --- a/include/builddefs.in
> +++ b/include/builddefs.in
> @@ -75,7 +75,7 @@ HAVE_RLIMIT_NOFILE = @have_rlimit_nofile@
>  NEED_INTERNAL_XFS_IOC_EXCHANGE_RANGE = @need_internal_xfs_ioc_exchange_range@
>  HAVE_FICLONE = @have_ficlone@
>  
> -GCCFLAGS = -funsigned-char -fno-strict-aliasing -Wall
> +GCCFLAGS = -funsigned-char -fno-strict-aliasing -std=gnu17 -Wall
>  SANITIZER_CFLAGS += @autovar_init_cflags@
>  
>  ifeq ($(PKG_PLATFORM),linux)
> -- 
> 2.47.1
> 
>
Zorro Lang Jan. 18, 2025, 2:32 p.m. UTC | #2
On Fri, Jan 17, 2025 at 09:27:36AM -0800, Darrick J. Wong wrote:
> On Fri, Jan 17, 2025 at 12:37:09PM +0800, Zorro Lang wrote:
> > GCC-15 does a big change, it changes the default language version for
> > C compilation from -std=gnu17 to -std=gnu23. That cause lots of "old
> > style" C codes hit build errors. On the other word, current xfstests
> > can't be used with GCC-15. So -std=gnu17 can help that.
> > 
> > Signed-off-by: Zorro Lang <zlang@kernel.org>
> > ---
> > 
> > Hi,
> > 
> > I send this patch just for talking about this issue. The upcoming gcc-15
> > does lots of changes, a big change is using C23 by default:
> > 
> >   https://gcc.gnu.org/gcc-15/porting_to.html
> > 
> > xfstests has many old style C codes, they hard to be built with gcc-15.
> > So we have to either add -std=$old_version (likes this patch), or port
> > the code to C23.
> > 
> > This patch is just a workaround (and a reminder for someone might hit
> > this issue with gcc-15 too). If you have any good suggestions or experience
> > (for this kind of issue) to share, feel free to reply.
> 
> -std=gnu11 to match the kernel and xfsprogs?

So you prefer using a settled "-std=xxx" to changing codes to match "gnu23"?

> 
> --D
> 
> > Thanks,
> > Zorro
> > 
> >  include/builddefs.in | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/include/builddefs.in b/include/builddefs.in
> > index 5b5864278..ef124bb87 100644
> > --- a/include/builddefs.in
> > +++ b/include/builddefs.in
> > @@ -75,7 +75,7 @@ HAVE_RLIMIT_NOFILE = @have_rlimit_nofile@
> >  NEED_INTERNAL_XFS_IOC_EXCHANGE_RANGE = @need_internal_xfs_ioc_exchange_range@
> >  HAVE_FICLONE = @have_ficlone@
> >  
> > -GCCFLAGS = -funsigned-char -fno-strict-aliasing -Wall
> > +GCCFLAGS = -funsigned-char -fno-strict-aliasing -std=gnu17 -Wall
> >  SANITIZER_CFLAGS += @autovar_init_cflags@
> >  
> >  ifeq ($(PKG_PLATFORM),linux)
> > -- 
> > 2.47.1
> > 
> > 
>
Darrick J. Wong Jan. 18, 2025, 8:14 p.m. UTC | #3
On Sat, Jan 18, 2025 at 10:32:14PM +0800, Zorro Lang wrote:
> On Fri, Jan 17, 2025 at 09:27:36AM -0800, Darrick J. Wong wrote:
> > On Fri, Jan 17, 2025 at 12:37:09PM +0800, Zorro Lang wrote:
> > > GCC-15 does a big change, it changes the default language version for
> > > C compilation from -std=gnu17 to -std=gnu23. That cause lots of "old
> > > style" C codes hit build errors. On the other word, current xfstests
> > > can't be used with GCC-15. So -std=gnu17 can help that.
> > > 
> > > Signed-off-by: Zorro Lang <zlang@kernel.org>
> > > ---
> > > 
> > > Hi,
> > > 
> > > I send this patch just for talking about this issue. The upcoming gcc-15
> > > does lots of changes, a big change is using C23 by default:
> > > 
> > >   https://gcc.gnu.org/gcc-15/porting_to.html
> > > 
> > > xfstests has many old style C codes, they hard to be built with gcc-15.
> > > So we have to either add -std=$old_version (likes this patch), or port
> > > the code to C23.
> > > 
> > > This patch is just a workaround (and a reminder for someone might hit
> > > this issue with gcc-15 too). If you have any good suggestions or experience
> > > (for this kind of issue) to share, feel free to reply.
> > 
> > -std=gnu11 to match the kernel and xfsprogs?
> 
> So you prefer using a settled "-std=xxx" to changing codes to match "gnu23"?

Only gcc 14 and 15 support gnu23, which will break the builds on ancient
distros such as Debian 12 and RHEL9.  Also, there's the human cost that
now fs developers have to know /two/ C dialects -- gnu11 for the kernel,
and gnu23 for fstests.

So yes, I prefer using -std=gnu11 for better consistency and tooling
support.

--D

> > 
> > --D
> > 
> > > Thanks,
> > > Zorro
> > > 
> > >  include/builddefs.in | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/include/builddefs.in b/include/builddefs.in
> > > index 5b5864278..ef124bb87 100644
> > > --- a/include/builddefs.in
> > > +++ b/include/builddefs.in
> > > @@ -75,7 +75,7 @@ HAVE_RLIMIT_NOFILE = @have_rlimit_nofile@
> > >  NEED_INTERNAL_XFS_IOC_EXCHANGE_RANGE = @need_internal_xfs_ioc_exchange_range@
> > >  HAVE_FICLONE = @have_ficlone@
> > >  
> > > -GCCFLAGS = -funsigned-char -fno-strict-aliasing -Wall
> > > +GCCFLAGS = -funsigned-char -fno-strict-aliasing -std=gnu17 -Wall
> > >  SANITIZER_CFLAGS += @autovar_init_cflags@
> > >  
> > >  ifeq ($(PKG_PLATFORM),linux)
> > > -- 
> > > 2.47.1
> > > 
> > > 
> > 
>
diff mbox series

Patch

diff --git a/include/builddefs.in b/include/builddefs.in
index 5b5864278..ef124bb87 100644
--- a/include/builddefs.in
+++ b/include/builddefs.in
@@ -75,7 +75,7 @@  HAVE_RLIMIT_NOFILE = @have_rlimit_nofile@
 NEED_INTERNAL_XFS_IOC_EXCHANGE_RANGE = @need_internal_xfs_ioc_exchange_range@
 HAVE_FICLONE = @have_ficlone@
 
-GCCFLAGS = -funsigned-char -fno-strict-aliasing -Wall
+GCCFLAGS = -funsigned-char -fno-strict-aliasing -std=gnu17 -Wall
 SANITIZER_CFLAGS += @autovar_init_cflags@
 
 ifeq ($(PKG_PLATFORM),linux)