Re: [PATCH] panic.c: export panic_on_oops

[Artem Bityutskiy <dedekind1@gmail.com>]

On Mon, 2009-10-12 at 13:15 +0200, ext Ingo Molnar wrote:
> > We use the mtdoops module which stores oopses on the Flash partition,
> > in order to make it possible to analyze them later. And mtdoops needs
> > to know whether 'panic_on_oops' is on or off. Thus, we need this
> > variable to be exported.
> 
> hm, why does it need it? Could you send the patch please that makes use 
> of it (as an easy way to explain the need for the export) - current 
> upstream drivers/mtd/mtdoops.c doesnt need it so it must be some pending 
> change.

Yes, current drivers/mtd/mtdoops.c does not need it, but we have this
patch:

http://lists.infradead.org/pipermail/linux-mtd/2009-October/027450.html

which makes sure mtdoops writes the oops log immediately, because we
have 'panic_on_oops' set so this is our last chance to save the oops.

Here is the patch inlined as well:

Author: Aaro Koskinen <aaro.koskinen@nokia.com>
Date:   Thu Oct 1 18:16:55 2009 +0300

    mtd: mtdoops: do not schedule work if we are going to die

    If panic_on_oops is set, the log should be written right away
    because this is not going to be a second chance.

    Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com>
    Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>

diff --git a/drivers/mtd/mtdoops.c b/drivers/mtd/mtdoops.c
index 1060337..ac67833 100644
--- a/drivers/mtd/mtdoops.c
+++ b/drivers/mtd/mtdoops.c
@@ -335,7 +335,7 @@ static void mtdoops_console_sync(void)
        cxt->ready = 0;
        spin_unlock_irqrestore(&cxt->writecount_lock, flags);

-       if (mtd->panic_write && in_interrupt())
+       if (mtd->panic_write && (in_interrupt() || panic_on_oops))
                /* Interrupt context, we're going to panic so try and log */
                mtdoops_write(cxt, 1);
        else

-- 
Best Regards,
Artem Bityutskiy (Артём Битюцкий)