On Tue, Jun 21, 2016 at 07:59:27PM +0800, Zhangjian (Bamvor) wrote:
> Hi,
>
> On 2016/6/21 15:58, Andreas Schwab wrote:
> >Yury Norov <yno...@caviumnetworks.com> writes:
> >
> >>diff --git a/sysdeps/unix/sysv/linux/sysdep-vdso.h
> >>b/sysdeps/unix/sysv/linux/sysdep-vdso.h
> >>index e8c4a7b..351d6bb 100644
> >>--- a/sysdeps/unix/sysv/linux/sysdep-vdso.h
> >>+++ b/sysdeps/unix/sysv/linux/sysdep-vdso.h
> >>@@ -37,7 +37,7 @@
> >> __label__ out;
> >> \
> >> __label__ iserr;
> >> \
> >> INTERNAL_SYSCALL_DECL (sc_err);
> >> \
> >>- long int sc_ret;
> >> \
> >>+ __syscall_slong_t sc_ret;
> >> \
> >> \
> >> __typeof (__vdso_##name) vdsop = __vdso_##name;
> >> \
> >> PTR_DEMANGLE (vdsop); \
> >>@@ -64,7 +64,7 @@
> >> # define INTERNAL_VSYSCALL(name, err, nr, args...)
> >> \
> >> ({
> >> \
> >> __label__ out;
> >> \
> >>- long v_ret;
> >> \
> >>+ __syscall_slong_t v_ret;
> >> \
> >> \
> >> __typeof (__vdso_##name) vdsop = __vdso_##name;
> >> \
> >> PTR_DEMANGLE (vdsop); \
> >
> >That part should have no effect now that __SYSCALL_SLONG_TYPE remains to
> >be long.
> Agree. According to the discussion in "[PATCH 11/27] [AARCH64] Syscalls for
> ILP32 are passed
> always via 64bit values.". We should update this to long long too.
>
I think comment to patch 11 is outdated. Andreas right, this chunk is
useless. I'll remove it in next submission.
