linux-kernel - Re: [PATCH] let WARN_ON() output the condition

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20061206092714.GA31980@elte.hu>
Date:	Wed, 6 Dec 2006 10:27:15 +0100
From:	Ingo Molnar <mingo@...e.hu>
To:	Jiri Kosina <jikos@...os.cz>
Cc:	Andrew Morton <akpm@...l.org>, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] let WARN_ON() output the condition


* Jiri Kosina <jikos@...os.cz> wrote:

> On Wed, 6 Dec 2006, Ingo Molnar wrote:
> 
> > i'll probably ack such a patch, it can be useful even when the line 
> > number is unique: if someone reports a WARN_ON() from an old kernel i 
> > dont have to dig up the exact source but can see it right from the 
> > condition what happened. Useful redundancy in bug output can be quite 
> > useful at times. Please post it and we'll see whether it's acceptable.
> 
> OK, thanks, I will send it later today.
> 
> BTW I still don't see how to distinguish it easily ... for example:
> 
> WARNING at kernel/mutex.c:132 __mutex_lock_common()
>  [<c0103d70>] dump_trace+0x68/0x1b5
>  [<c0103ed5>] show_trace_log_lvl+0x18/0x2c
>  [<c010445b>] show_trace+0xf/0x11
>  [<c01044cd>] dump_stack+0x12/0x14
>  [<c037523f>] __mutex_lock_slowpath+0xc6/0x261
>  [<c0199c61>] create_dir+0x24/0x1ba
>  [<c019a30b>] sysfs_create_dir+0x45/0x5f
>  [<c01f302b>] kobject_add+0xd6/0x18d
>  [<c01f31fb>] kobject_register+0x19/0x30
>  [<c02e771a>] md_probe+0x11a/0x124
>  [<c0267fa4>] kobj_lookup+0xe6/0x122
>  [<c01ec12e>] get_gendisk+0xe/0x1b
>  [<c018590e>] do_open+0x2e/0x298
>  [<c0185d0f>] blkdev_open+0x25/0x4d
>  [<c016451b>] __dentry_open+0xc3/0x17e
>  [<c0164650>] nameidata_to_filp+0x24/0x33
>  [<c0164691>] do_filp_open+0x32/0x39
>  [<c01646da>] do_sys_open+0x42/0xbe
>  [<c016478f>] sys_open+0x1c/0x1e
>  [<c0102dbc>] syscall_call+0x7/0xb
> 
> How can you see immediately which one of the two WARN_ONs in 
> spin_lock_mutex() triggered?

yeah, i can tell that even without assembly or gdb, just from the 
offset:

>  [<c037523f>] __mutex_lock_slowpath+0xc6/0x261

there are 4 WARN_ON()s in __mutex_lock_slowpath(), distributed roughly 
equally. Which makes the above one the second out of the four 
WARN_ON()s, i.e.:

                DEBUG_LOCKS_WARN_ON(l->magic != l);     \

Did i get it right? (but then again i guess i've got an unfair advantage 
in interpreting locking related bug messages ;-)

but it can also be told semantically, it cannot be the in_interrupt() 
assert because this is clearly not IRQ context:

>  [<c037523f>] __mutex_lock_slowpath+0xc6/0x261
>  [<c0199c61>] create_dir+0x24/0x1ba
>  [<c019a30b>] sysfs_create_dir+0x45/0x5f

but in such cases i'd rather suggest the use of inline functions instead 
of macros and then it's a simple gdb lookup to figure out the call site. 
So, which clown added that macro to mutex-debug.h ... oh, never mind.

	Ingo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/