Re: [PATCH 7/12] IPMI: add poll delay

["Jesper Juhl" <jesper.juhl@gmail.com>]

On 05/12/06, Corey Minyard <minyard@acm.org> wrote:
> Andrew Morton wrote:
> > On Fri, 1 Dec 2006 22:35:20 -0600
> > Corey Minyard <minyard@acm.org> wrote:
> >
> >
> >> Make sure to delay a little in the IPMI poll routine so we can pass in
> >> a timeout time and thus time things out.
> >>
> >> Signed-off-by: Corey Minyard <minyard@acm.org>
> >>
> >> Index: linux-2.6.19-rc6/drivers/char/ipmi/ipmi_si_intf.c
> >> ===================================================================
> >> --- linux-2.6.19-rc6.orig/drivers/char/ipmi/ipmi_si_intf.c
> >> +++ linux-2.6.19-rc6/drivers/char/ipmi/ipmi_si_intf.c
> >> @@ -807,7 +807,12 @@ static void poll(void *send_info)
> >>  {
> >>      struct smi_info *smi_info = send_info;
> >>
> >> -    smi_event_handler(smi_info, 0);
> >> +    /*
> >> +     * Make sure there is some delay in the poll loop so we can
> >> +     * drive time forward and timeout things.
> >> +     */
> >> +    udelay(10);
> >> +    smi_event_handler(smi_info, 10);
> >>  }
> >>
> >
> > I don't understand what this patch is doing.  It looks fishy.  More
> > details, please?
> >
> Yeah, it does look a little fishy.  This is a poll routine that is only
> called at panic
> time; it is used to force things to happen in the driver without
> scheduling or
> timers.  The driver does this so it can set watchdog parameters and store
> panic information in the event log at panic time.
>
> Without this change, if something goes wrong in the BMC the driver will
> never
> time out the operation since it doesn't see time being driven forward.
> So this
> makes sure the driver sees time advancing as it should.
>

Hmm, I wonder if this could explain why some of my IBM servers become
unreachable via IPMI after a kernel crash.

-- 
Jesper Juhl <jesper.juhl@gmail.com>
Don't top-post  http://www.catb.org/~esr/jargon/html/T/top-post.html
Plain text mails only, please      http://www.expita.com/nomime.html