linux-kernel - Re: [PATCH 2/3] printk: Add %ptc to safely print a task's comm

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

Date:	Tue, 10 May 2011 18:10:46 -0700
From:	John Stultz <john.stultz@...aro.org>
To:	Joe Perches <joe@...ches.com>
Cc:	LKML <linux-kernel@...r.kernel.org>, "Ted Ts'o" <tytso@....edu>,
	KOSAKI Motohiro <kosaki.motohiro@...fujitsu.com>,
	David Rientjes <rientjes@...gle.com>,
	Dave Hansen <dave@...ux.vnet.ibm.com>,
	Andrew Morton <akpm@...ux-foundation.org>, linux-mm@...ck.org
Subject: Re: [PATCH 2/3] printk: Add %ptc to safely print a task's comm

On Tue, 2011-05-10 at 17:51 -0700, Joe Perches wrote:
> On Tue, 2011-05-10 at 17:23 -0700, John Stultz wrote:
> > Acessing task->comm requires proper locking. However in the past
> > access to current->comm could be done without locking. This
> > is no longer the case, so all comm access needs to be done
> > while holding the comm_lock.
> > 
> > In my attempt to clean up unprotected comm access, I've noticed
> > most comm access is done for printk output. To simpify correct
> > locking in these cases, I've introduced a new %ptc format,
> > which will safely print the corresponding task's comm.
> 
> Hi John.
> 
> Couple of tyops for Accessing and simplify in your commit message
> and a few comments on the patch.

Ah. Yes. Thanks!

> Could misuse of %ptc (not using current) cause system lockup?

It very well could. Although I don't see other %p options tring to
handle invalid pointers. Any suggestions on how to best handle this?


> > Example use:
> > printk("%ptc: unaligned epc - sending SIGBUS.\n", current);
> 
> 
> > diff --git a/lib/vsprintf.c b/lib/vsprintf.c
> > index bc0ac6b..b9c97b8 100644
> > --- a/lib/vsprintf.c
> > +++ b/lib/vsprintf.c
> > @@ -797,6 +797,26 @@ char *uuid_string(char *buf, char *end, const u8 *addr,
> >  	return string(buf, end, uuid, spec);
> >  }
> >  
> > +static noinline_for_stack
> > +char *task_comm_string(char *buf, char *end, u8 *addr,
> > +			 struct printf_spec spec, const char *fmt)
> 
> addr should be void * not u8 *
> 
> > +{
> > +	struct task_struct *tsk = (struct task_struct *) addr;
> 
> no cast.
> 
> Maybe it'd be better to use current inside this routine and not
> pass the pointer at all.

That sounds reasonable. Most users are current, so forcing the more rare
non-current users to copy it to a buffer first and use the normal %s
would not be of much impact.

Although I'm not sure if there's precedent for a %p value that didn't
take a argument. Thoughts on that? Anyone else have an opinion here?

Thanks so much for the review and feedback!
-john

--
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/