Re: [Dle-develop] [RFC][PATCH v4 -next 1/4] Move kmsg_dump(KMSG_DUMP_PANIC) below smp_send_stop()

[Don Z. <dz...@re...>]

On Thu, Jan 05, 2012 at 03:10:25PM -0500, Seiji Aguchi wrote:
> 
> >Aren't you worried about the comment about smp_send_stop() not
> >being hardened to work in a panic situation?
> > 	/*
> > 	 * Note smp_send_stop is the usual smp shutdown function, which
> > 	 * unfortunately means it may not be hardened to work in a panic
> 
> This comment is wrong because Don improved smp_send_stop() by switching REBOOT_VECTOR to NMI.
> And his patch has already merged to linux-next tree.

I only fixed x86.  Who knows what the other arches do..

I don't know how to prove something is hardened other than not seeing any
hangs or false reboots on in that piece of code.

Cheers,
Don

> 
> http://git.kernel.org/?p=linux/kernel/git/next/linux-next.git;a=commitdiff;h=3603a2512f9e69dc87914ba922eb4a0812b21cd6
> 
> So, current smp_send_stop() is hardened to work in a panic situation.
> 
> I will remove this wrong comment.
> 
> Seiji
> 
> >-----Original Message-----
> >From: Luck, Tony [mailto:ton...@in...]
> >Sent: Thursday, January 05, 2012 2:07 PM
> >To: Seiji Aguchi; Don Zickus
> >Cc: lin...@vg...; Matthew Garrett; Vivek Goyal; Chen, Gong; ak...@li...; Brown, Len;
> >'yin...@in...'; 'ak...@li...'; 'hu...@ch...'; 'mi...@el...'; jm...@na...;
> >a.p...@ch...; nam...@gm...; dle...@li...; Satoru Moriya
> >Subject: RE: [RFC][PATCH v4 -next 1/4] Move kmsg_dump(KMSG_DUMP_PANIC) below smp_send_stop()
> >
> >-	kmsg_dump(KMSG_DUMP_PANIC);
> >-
> > 	/*
> > 	 * Note smp_send_stop is the usual smp shutdown function, which
> > 	 * unfortunately means it may not be hardened to work in a panic
> >@@ -117,6 +115,8 @@ void panic(const char *fmt, ...)
> > 	 */
> > 	smp_send_stop();
> >
> >+	kmsg_dump(KMSG_DUMP_PANIC);
> >+
> >
> >Aren't you worried about the comment about smp_send_stop() not
> >being hardened to work in a panic situation?
> >
> >If it does work - we are clearly much better off moving the
> >kmsg_dump() call down like this. It makes life much simpler
> >and cleaner to work with just one running cpu.
> >
> >But if something goes wrong - we might not see the dump at all!
> >
> >How do we compare these cases and decide that it is better to
> >trust that smp_send_stop() will return?
> >
> >-Tony