[PATCH] seccomp, s390: fix build for syscall type change

[PATCH] seccomp, s390: fix build for syscall type change

[Tycho Andersen]

A recent patch landed in the security tree [1] that changed the type of the
seccomp syscall. Unfortunately, I didn't quite get every instance of the
forward declarations, and thus there is a build failure. Here's the last
one that I could find, for s390. It should go through the security tree,
although hopefully some s390 people can check and make sure it looks
reasonable?

The only oddity is the trailing semicolon; some lines around this patch
have it, and some lines don't. I've left this one as-is.

[1]: https://lore.kernel.org/lkml/20181212231630.GA31584@beast/T/#u

Signed-off-by: Tycho Andersen <tycho@tycho.ws>
---
 arch/s390/kernel/compat_wrapper.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/s390/kernel/compat_wrapper.c b/arch/s390/kernel/compat_wrapper.c
index 2ce28bf0c5ec..48c4ce668244 100644
--- a/arch/s390/kernel/compat_wrapper.c
+++ b/arch/s390/kernel/compat_wrapper.c
@@ -164,7 +164,7 @@ COMPAT_SYSCALL_WRAP3(finit_module, int, fd, const char __user *, uargs, int, fla
 COMPAT_SYSCALL_WRAP3(sched_setattr, pid_t, pid, struct sched_attr __user *, attr, unsigned int, flags);
 COMPAT_SYSCALL_WRAP4(sched_getattr, pid_t, pid, struct sched_attr __user *, attr, unsigned int, size, unsigned int, flags);
 COMPAT_SYSCALL_WRAP5(renameat2, int, olddfd, const char __user *, oldname, int, newdfd, const char __user *, newname, unsigned int, flags);
-COMPAT_SYSCALL_WRAP3(seccomp, unsigned int, op, unsigned int, flags, const char __user *, uargs)
+COMPAT_SYSCALL_WRAP3(seccomp, unsigned int, op, unsigned int, flags, void __user *, uargs)
 COMPAT_SYSCALL_WRAP3(getrandom, char __user *, buf, size_t, count, unsigned int, flags)
 COMPAT_SYSCALL_WRAP2(memfd_create, const char __user *, uname, unsigned int, flags)
 COMPAT_SYSCALL_WRAP3(bpf, int, cmd, union bpf_attr *, attr, unsigned int, size);
-- 
2.19.1

Re: [PATCH] seccomp, s390: fix build for syscall type change

[Kees Cook]

On Thu, Dec 13, 2018 at 12:10 PM Tycho Andersen <tycho@tycho.ws> wrote:
>
> A recent patch landed in the security tree [1] that changed the type of the
> seccomp syscall. Unfortunately, I didn't quite get every instance of the
> forward declarations, and thus there is a build failure. Here's the last
> one that I could find, for s390. It should go through the security tree,
> although hopefully some s390 people can check and make sure it looks
> reasonable?
>
> The only oddity is the trailing semicolon; some lines around this patch
> have it, and some lines don't. I've left this one as-is.
>
> [1]: https://lore.kernel.org/lkml/20181212231630.GA31584@beast/T/#u
>
> Signed-off-by: Tycho Andersen <tycho@tycho.ws>

Thanks! Added to my -next tree.

-Kees

> ---
>  arch/s390/kernel/compat_wrapper.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/s390/kernel/compat_wrapper.c b/arch/s390/kernel/compat_wrapper.c
> index 2ce28bf0c5ec..48c4ce668244 100644
> --- a/arch/s390/kernel/compat_wrapper.c
> +++ b/arch/s390/kernel/compat_wrapper.c
> @@ -164,7 +164,7 @@ COMPAT_SYSCALL_WRAP3(finit_module, int, fd, const char __user *, uargs, int, fla
>  COMPAT_SYSCALL_WRAP3(sched_setattr, pid_t, pid, struct sched_attr __user *, attr, unsigned int, flags);
>  COMPAT_SYSCALL_WRAP4(sched_getattr, pid_t, pid, struct sched_attr __user *, attr, unsigned int, size, unsigned int, flags);
>  COMPAT_SYSCALL_WRAP5(renameat2, int, olddfd, const char __user *, oldname, int, newdfd, const char __user *, newname, unsigned int, flags);
> -COMPAT_SYSCALL_WRAP3(seccomp, unsigned int, op, unsigned int, flags, const char __user *, uargs)
> +COMPAT_SYSCALL_WRAP3(seccomp, unsigned int, op, unsigned int, flags, void __user *, uargs)
>  COMPAT_SYSCALL_WRAP3(getrandom, char __user *, buf, size_t, count, unsigned int, flags)
>  COMPAT_SYSCALL_WRAP2(memfd_create, const char __user *, uname, unsigned int, flags)
>  COMPAT_SYSCALL_WRAP3(bpf, int, cmd, union bpf_attr *, attr, unsigned int, size);
> --
> 2.19.1
>


-- 
Kees Cook

Re: [PATCH] seccomp, s390: fix build for syscall type change

[Heiko Carstens]

On Thu, Dec 13, 2018 at 04:50:06PM -0800, Kees Cook wrote:
> On Thu, Dec 13, 2018 at 12:10 PM Tycho Andersen <tycho@tycho.ws> wrote:
> >
> > A recent patch landed in the security tree [1] that changed the type of the
> > seccomp syscall. Unfortunately, I didn't quite get every instance of the
> > forward declarations, and thus there is a build failure. Here's the last
> > one that I could find, for s390. It should go through the security tree,
> > although hopefully some s390 people can check and make sure it looks
> > reasonable?
> >
> > The only oddity is the trailing semicolon; some lines around this patch
> > have it, and some lines don't. I've left this one as-is.
> >
> > [1]: https://lore.kernel.org/lkml/20181212231630.GA31584@beast/T/#u
> >
> > Signed-off-by: Tycho Andersen <tycho@tycho.ws>
> 
> Thanks! Added to my -next tree.
> 
> -Kees

Hi Kees,

did you forget to push your tree? At least today's linux-next kernel
fails to compile on s390 because of this missing patch.

> >  arch/s390/kernel/compat_wrapper.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/arch/s390/kernel/compat_wrapper.c b/arch/s390/kernel/compat_wrapper.c
> > index 2ce28bf0c5ec..48c4ce668244 100644
> > --- a/arch/s390/kernel/compat_wrapper.c
> > +++ b/arch/s390/kernel/compat_wrapper.c
> > @@ -164,7 +164,7 @@ COMPAT_SYSCALL_WRAP3(finit_module, int, fd, const char __user *, uargs, int, fla
> >  COMPAT_SYSCALL_WRAP3(sched_setattr, pid_t, pid, struct sched_attr __user *, attr, unsigned int, flags);
> >  COMPAT_SYSCALL_WRAP4(sched_getattr, pid_t, pid, struct sched_attr __user *, attr, unsigned int, size, unsigned int, flags);
> >  COMPAT_SYSCALL_WRAP5(renameat2, int, olddfd, const char __user *, oldname, int, newdfd, const char __user *, newname, unsigned int, flags);
> > -COMPAT_SYSCALL_WRAP3(seccomp, unsigned int, op, unsigned int, flags, const char __user *, uargs)
> > +COMPAT_SYSCALL_WRAP3(seccomp, unsigned int, op, unsigned int, flags, void __user *, uargs)
> >  COMPAT_SYSCALL_WRAP3(getrandom, char __user *, buf, size_t, count, unsigned int, flags)
> >  COMPAT_SYSCALL_WRAP2(memfd_create, const char __user *, uname, unsigned int, flags)
> >  COMPAT_SYSCALL_WRAP3(bpf, int, cmd, union bpf_attr *, attr, unsigned int, size);

Re: [PATCH] seccomp, s390: fix build for syscall type change

[Kees Cook]

On Mon, Dec 17, 2018 at 4:42 AM Heiko Carstens
<heiko.carstens@de.ibm.com> wrote:
>
> On Thu, Dec 13, 2018 at 04:50:06PM -0800, Kees Cook wrote:
> > On Thu, Dec 13, 2018 at 12:10 PM Tycho Andersen <tycho@tycho.ws> wrote:
> > >
> > > A recent patch landed in the security tree [1] that changed the type of the
> > > seccomp syscall. Unfortunately, I didn't quite get every instance of the
> > > forward declarations, and thus there is a build failure. Here's the last
> > > one that I could find, for s390. It should go through the security tree,
> > > although hopefully some s390 people can check and make sure it looks
> > > reasonable?
> > >
> > > The only oddity is the trailing semicolon; some lines around this patch
> > > have it, and some lines don't. I've left this one as-is.
> > >
> > > [1]: https://lore.kernel.org/lkml/20181212231630.GA31584@beast/T/#u
> > >
> > > Signed-off-by: Tycho Andersen <tycho@tycho.ws>
> >
> > Thanks! Added to my -next tree.
> >
> > -Kees
>
> Hi Kees,
>
> did you forget to push your tree? At least today's linux-next kernel
> fails to compile on s390 because of this missing patch.

I sent a pull request to James for this. James, can you pick this up please?

-Kees

>
> > >  arch/s390/kernel/compat_wrapper.c | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/arch/s390/kernel/compat_wrapper.c b/arch/s390/kernel/compat_wrapper.c
> > > index 2ce28bf0c5ec..48c4ce668244 100644
> > > --- a/arch/s390/kernel/compat_wrapper.c
> > > +++ b/arch/s390/kernel/compat_wrapper.c
> > > @@ -164,7 +164,7 @@ COMPAT_SYSCALL_WRAP3(finit_module, int, fd, const char __user *, uargs, int, fla
> > >  COMPAT_SYSCALL_WRAP3(sched_setattr, pid_t, pid, struct sched_attr __user *, attr, unsigned int, flags);
> > >  COMPAT_SYSCALL_WRAP4(sched_getattr, pid_t, pid, struct sched_attr __user *, attr, unsigned int, size, unsigned int, flags);
> > >  COMPAT_SYSCALL_WRAP5(renameat2, int, olddfd, const char __user *, oldname, int, newdfd, const char __user *, newname, unsigned int, flags);
> > > -COMPAT_SYSCALL_WRAP3(seccomp, unsigned int, op, unsigned int, flags, const char __user *, uargs)
> > > +COMPAT_SYSCALL_WRAP3(seccomp, unsigned int, op, unsigned int, flags, void __user *, uargs)
> > >  COMPAT_SYSCALL_WRAP3(getrandom, char __user *, buf, size_t, count, unsigned int, flags)
> > >  COMPAT_SYSCALL_WRAP2(memfd_create, const char __user *, uname, unsigned int, flags)
> > >  COMPAT_SYSCALL_WRAP3(bpf, int, cmd, union bpf_attr *, attr, unsigned int, size);
>


-- 
Kees Cook