Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1758149Ab1EZRiW (ORCPT ); Thu, 26 May 2011 13:38:22 -0400 Received: from mail-vx0-f174.google.com ([209.85.220.174]:54389 "EHLO mail-vx0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757559Ab1EZRiV convert rfc822-to-8bit (ORCPT ); Thu, 26 May 2011 13:38:21 -0400 DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type:content-transfer-encoding; b=mopXDDPnxHszWPEYK6YIYAjojyvMCysO10s9aahZXali85r9AVLS7dsWDbcb44H9OU fLuQVCv5IhkImEWHbEy9bJUXUZKeOx7DlyXFgkdIDyZGQV+WQvME2S9FoN+rNjMvqmNB hHfQLYNwepUOWoAy1Q2onb1AoLG7IL69GEGvw= MIME-Version: 1.0 In-Reply-To: <20110526171209.GA17988@in.ibm.com> References: <20110526170722.GB23266@in.ibm.com> <20110526171209.GA17988@in.ibm.com> Date: Thu, 26 May 2011 19:38:19 +0200 Message-ID: Subject: Re: [Patch 1/6] XPANIC: Add extended panic interface From: richard -rw- weinberger To: prasad@linux.vnet.ibm.com Cc: Linux Kernel Mailing List , Andi Kleen , "Luck, Tony" , Vivek Goyal , kexec@lists.infradead.org, "Eric W. Biederman" , anderson@redhat.com Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 6139 Lines: 176 On Thu, May 26, 2011 at 7:12 PM, K.Prasad wrote: > > commit e668fa1aea7844ac4c7ea09030a2f3e647a4adb1 > Author: Andi Kleen > Date: ? Fri Nov 19 18:36:44 2010 +0100 > > ? ?XPANIC: Add extended panic interface > > ? ?One panic size doesn't fit all. > > ? ?Machine check has some special requirements on panic, like: > ? ?- It wants to do a reboot timeout by default so that the machine > ? ?check event can be logged to disk after warm reboot. > ? ?- For memory errors it usually doesn't want to do crash dumps because > ? ?that can lead to crash loops (dumping corrupted memory would > ? ?lead to another machine check) > ? ?- It doesn't want to do backtraces because machine checks > ? ?are not a kernel bug. > > ? ?In a earlier patch this was done with various adhoc hacks, > ? ?but it's cleaner to extend panic to a 'xpanic' that directly > ? ?gets a flag and timeout argument. > > ? ?The only user right now will be x86 machine checks, but I consider > ? ?it likely that other users will switch to this too. > > ? ?For example one obvious candidate would be the "no root > ? ?found" panic which doesn't really deserve a backtrace. > > ? ?I exported a vpanic() interface too as a global. That's not > ? ?needed by the current user, but the interface has to exist > ? ?internally anyways and I could see how other code would > ? ?find a v* variant of panic useful. > > ? ?Originally based on a suggestion by H. Peter Anvin. > ? ?Signed-off-by: Andi Kleen > > --- > ?include/linux/kernel.h | ? 11 +++++++++++ > ?kernel/panic.c ? ? ? ? | ? 41 +++++++++++++++++++++++++++++++---------- > ?2 files changed, 42 insertions(+), 10 deletions(-) > > Index: linux-2.6.slim_kdump/include/linux/kernel.h > =================================================================== > --- linux-2.6.slim_kdump.orig/include/linux/kernel.h > +++ linux-2.6.slim_kdump/include/linux/kernel.h > @@ -175,10 +175,21 @@ static inline void might_fault(void) > ?} > ?#endif > > +enum panic_flags { > + ? ? ? PANIC_NO_KEXEC ? ? = (1 << 0), > + ? ? ? PANIC_NO_BACKTRACE = (1 << 1), > +}; > + > ?extern struct atomic_notifier_head panic_notifier_list; > ?extern long (*panic_blink)(int state); > ?NORET_TYPE void panic(const char * fmt, ...) > ? ? ? ?__attribute__ ((NORET_AND format (printf, 1, 2))) __cold; > +NORET_TYPE void xpanic(enum panic_flags flags, int timeout, > + ? ? ? ? ? ? ? ? ? ? ?const char *fmt, ...) > + ? ? ? __attribute__ ((NORET_AND format (printf, 3, 4))) __cold; > +NORET_TYPE void vpanic(enum panic_flags flags, int timeout, > + ? ? ? ? ? ? ? ? ? ? ?const char *fmt, > + ? ? ? ? ? ? ? ? ? ? ?va_list ap) __noreturn __cold; > ?extern void oops_enter(void); > ?extern void oops_exit(void); > ?void print_oops_end_marker(void); > Index: linux-2.6.slim_kdump/kernel/panic.c > =================================================================== > --- linux-2.6.slim_kdump.orig/kernel/panic.c > +++ linux-2.6.slim_kdump/kernel/panic.c > @@ -57,14 +57,37 @@ EXPORT_SYMBOL(panic_blink); > ?* > ?* ? ? This function never returns. > ?*/ > -NORET_TYPE void panic(const char * fmt, ...) > +NORET_TYPE void panic(const char *fmt, ...) > +{ > + ? ? ? va_list ap; > + ? ? ? va_start(ap, fmt); > + ? ? ? vpanic(0, 0, fmt, ap); > +} > +EXPORT_SYMBOL(panic); > + > +NORET_TYPE void xpanic(enum panic_flags flags, int timeout, > + ? ? ? ? ? ? ? ? ? ? ? const char *fmt, ...) > +{ > + ? ? ? va_list ap; > + ? ? ? va_start(ap, fmt); > + ? ? ? xpanic(flags, timeout, fmt, ap); Why are you calling xpanic() here again? I guess you meant vpanic(). > +} > +EXPORT_SYMBOL(xpanic); > + > +NORET_TYPE void vpanic(enum panic_flags flags, int timeout, > + ? ? ? ? ? ? ? ? ? ? ?const char * fmt, va_list args) > ?{ > ? ? ? ?static char buf[1024]; > - ? ? ? va_list args; > ? ? ? ?long i, i_next = 0; > ? ? ? ?int state = 0; > > ? ? ? ?/* > + ? ? ? ?* Let user always override panic_timeout. > + ? ? ? ?*/ > + ? ? ? if (panic_timeout > 0) > + ? ? ? ? ? ? ? timeout = panic_timeout; > + > + ? ? ? /* > ? ? ? ? * It's possible to come here directly from a panic-assertion and > ? ? ? ? * not have preempt disabled. Some functions called from here want > ? ? ? ? * preempt to be disabled. No point enabling it later though... > @@ -73,12 +96,11 @@ NORET_TYPE void panic(const char * fmt, > > ? ? ? ?console_verbose(); > ? ? ? ?bust_spinlocks(1); > - ? ? ? va_start(args, fmt); > ? ? ? ?vsnprintf(buf, sizeof(buf), fmt, args); > - ? ? ? va_end(args); > ? ? ? ?printk(KERN_EMERG "Kernel panic - not syncing: %s\n",buf); > ?#ifdef CONFIG_DEBUG_BUGVERBOSE > - ? ? ? dump_stack(); > + ? ? ? if (!(flags & PANIC_NO_BACKTRACE)) > + ? ? ? ? ? ? ? dump_stack(); > ?#endif > > ? ? ? ?/* > @@ -86,7 +108,8 @@ NORET_TYPE void panic(const char * fmt, > ? ? ? ? * everything else. > ? ? ? ? * Do we want to call this before we try to display a message? > ? ? ? ? */ > - ? ? ? crash_kexec(NULL); > + ? ? ? if (!(flags & PANIC_NO_KEXEC)) > + ? ? ? ? ? ? ? crash_kexec(NULL); > > ? ? ? ?kmsg_dump(KMSG_DUMP_PANIC); > > @@ -104,7 +127,7 @@ NORET_TYPE void panic(const char * fmt, > ? ? ? ?if (!panic_blink) > ? ? ? ? ? ? ? ?panic_blink = no_blink; > > - ? ? ? if (panic_timeout > 0) { > + ? ? ? if (timeout > 0) { > ? ? ? ? ? ? ? ?/* > ? ? ? ? ? ? ? ? * Delay timeout seconds before rebooting the machine. > ? ? ? ? ? ? ? ? * We can't use the "normal" timers since we just panicked. > @@ -152,9 +175,7 @@ NORET_TYPE void panic(const char * fmt, > ? ? ? ? ? ? ? ?mdelay(PANIC_TIMER_STEP); > ? ? ? ?} > ?} > - > -EXPORT_SYMBOL(panic); > - > +EXPORT_SYMBOL(vpanic); > > ?struct tnt { > ? ? ? ?u8 ? ? ?bit; > -- > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at ?http://vger.kernel.org/majordomo-info.html > Please read the FAQ at ?http://www.tux.org/lkml/ > -- Thanks, //richard -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/