[RFC PATCH] Fix Oops in rtas_stop_self()
Li Zhong
zhong at linux.vnet.ibm.com
Mon Apr 28 10:26:45 EST 2014
On Fri, 2014-04-25 at 22:18 +1000, Anton Blanchard wrote:
> Hi,
>
> > When trying offline cpus, I noticed following Oops in
> > rtas_stop_self(), and it seems caused by commit 41dd03a9. The Oops
> > disappears after reverting this commit.
> >
> > After reading the code, I guess it might be caused by moving the
> > rtas_args to stack. Still need some more time to read enter_rtas to
> > understand why it happens, but the problem seems could be solved by
> > moving the rtas_args away from stack by adding static before it.
>
> Nice catch. RTAS is 32bit and if your box has more than 4GB RAM then
> your stack could easily be outside 32bit range.
Ah, yes, the stack here is obviously at a much higher address than 4GB.
>
> You can add:
>
> Signed-off-by: Anton Blanchard <anton at samba.org>
>
> And also:
>
> Cc: stable at vger.kernel.org # 3.14+
OK,
Thanks, Zhong
>
> > Signed-off-by: Li Zhong <zhong at linux.vnet.ibm.com>
> > ---
> > arch/powerpc/platforms/pseries/hotplug-cpu.c | 5 +++--
> > 1 file changed, 3 insertions(+), 2 deletions(-)
> >
> > diff --git a/arch/powerpc/platforms/pseries/hotplug-cpu.c
> > b/arch/powerpc/platforms/pseries/hotplug-cpu.c index 9b8e050..20d6297
> > 100644 --- a/arch/powerpc/platforms/pseries/hotplug-cpu.c
> > +++ b/arch/powerpc/platforms/pseries/hotplug-cpu.c
> > @@ -88,13 +88,14 @@ void set_default_offline_state(int cpu)
> >
> > static void rtas_stop_self(void)
> > {
> > - struct rtas_args args = {
> > - .token = cpu_to_be32(rtas_stop_self_token),
> > + static struct rtas_args args = {
> > .nargs = 0,
> > .nret = 1,
> > .rets = &args.args[0],
> > };
> >
> > + args.token = cpu_to_be32(rtas_stop_self_token);
> > +
> > local_irq_disable();
> >
> > BUG_ON(rtas_stop_self_token == RTAS_UNKNOWN_SERVICE);
> >
> >
> > _______________________________________________
> > Linuxppc-dev mailing list
> > Linuxppc-dev at lists.ozlabs.org
> > https://lists.ozlabs.org/listinfo/linuxppc-dev
>
More information about the Linuxppc-dev
mailing list