Re: [PATCH v6 18/19] linux-user/s390x: Add vdso

[Thomas Huth]

On 05/10/2023 18.14, Philippe Mathieu-Daudé wrote:
> On 4/10/23 01:01, Richard Henderson wrote:
> > On 10/3/23 06:07, Philippe Mathieu-Daudé wrote:
> > > > +++ b/linux-user/s390x/vdso.ld
> > > > @@ -0,0 +1,69 @@
> > > > +/*
> > > > + * Linker script for linux x86-64 replacement vdso.

Should that be s390x instead of x86-64 ?

> > > > + * Copyright 2023 Linaro, Ltd.
> > > > + *
> > > > + * SPDX-License-Identifier: GPL-2.0-or-later
> > > > + */
> > > > +
> > > > +VERSION {
> > > > +        LINUX_2.6.29 {
> > > > +        global:
> > > > +                __kernel_gettimeofday;
> > > > +                __kernel_clock_gettime;
> > > > +                __kernel_clock_getres;
> > > > +                __kernel_getcpu;
> > > > +                __kernel_restart_syscall;
> > >
> > > Where is __kernel_restart_syscall defined?
> >
> > It isn't, and I guess isn't needed for qemu (we have QEMU_ERESTARTSYS).
> > I'm not sure why it's an exported symbol for the s390x vdso for the 
> > kernel, even if they do require some user-space code.
> >
> > Anyway, I'll remove the symbol here.
>
> OK.
>
> I dare to give:
>
> Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>
> But still I'm Cc'ing David / Thomas who better know s390x :)

Uh, oh, vdso ... I remember some dark magic here on the real kernels on 
s390x, involving access registers ... good that we don't need those here :-)

I quickly glanced through the code, and apart from the typo above, it looks 
good to me, so:

Acked-by: Thomas Huth <thuth@redhat.com>