linux-kernel - Re: [PATCH] kmsg: Honor dmesg

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:	Wed, 27 Feb 2013 13:01:45 -0500
From:	Josh Boyer <jwboyer@...hat.com>
To:	Kees Cook <keescook@...omium.org>
Cc:	Andrew Morton <akpm@...ux-foundation.org>,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	eparis@...hat.com, Christian Kujau <lists@...dbynature.de>,
	stable@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] kmsg: Honor dmesg_restrict sysctl on /dev/kmsg

On Wed, Feb 27, 2013 at 09:54:27AM -0800, Kees Cook wrote:
> On Fri, Feb 22, 2013 at 01:18:57PM -0500, Josh Boyer wrote:
> > Originally, the addition of dmesg_restrict covered both the syslog
> > method of accessing dmesg, as well as /dev/kmsg itself.  This was done
> > indirectly by security_syslog calling cap_syslog before doing any LSM
> > checks.
> > 
> > However, commit 12b3052c3ee (capabilities/syslog: open code cap_syslog
> > logic to fix build failure) moved the code around and pushed the checks
> > into the caller itself.  That seems to have inadvertently dropped the
> > checks for dmesg_restrict on /dev/kmsg.  Most people haven't noticed
> > because util-linux dmesg(1) defaults to using the syslog method for
> > access in older versions.  With util-linux 2.22 and a kernel newer than
> > 3.5, dmesg(1) defaults to reading directly from /dev/kmsg.
> > 
> > Fix this by making an explicit check in the devkmsg_open function.
> > 
> > This fixes https://bugzilla.redhat.com/show_bug.cgi?id=903192
> > 
> > Reported-by: Christian Kujau <lists@...dbynature.de>
> > CC: stable@...r.kernel.org
> > Signed-off-by: Josh Boyer <jwboyer@...hat.com>
> > ---
> >  kernel/printk.c | 3 +++
> >  1 file changed, 3 insertions(+)
> > 
> > diff --git a/kernel/printk.c b/kernel/printk.c
> > index f24633a..398ef9a 100644
> > --- a/kernel/printk.c
> > +++ b/kernel/printk.c
> > @@ -615,6 +615,9 @@ static int devkmsg_open(struct inode *inode, struct file *file)
> >  	struct devkmsg_user *user;
> >  	int err;
> >  
> > +	if (dmesg_restrict && !capable(CAP_SYSLOG))
> > +		return -EACCES;
> > +
> 
> I think this should use check_syslog_permissions() instead, as done for
> /proc/kmsg and the syslog syscall.
> 
> 	err = check_syslog_permissions(SYSLOG_ACTION_OPTION, SYSLOG_FROM_FILE);

Did you mean SYSLOG_ACTION_OPEN?

I didn't code it that way because the comment in that function about the
capability checks already being done seem pretty off to me.  I could
have just misread the /proc code though.  I can resend with the change
you suggest if everyone thinks that's a better way.

Also, the LSM hooks aren't doing any capability checks at all that I can
see, which may or may not be a bug in and of itself but I have no idea.
I was hoping Eric would speak up about that.

> 	if (err)
> 		return err;
> 
> And going forward we should probably think about dropping the CAP_SYS_ADMIN
> backward-compat code in check_syslog_permissions.

Sure, but that's a separate commit.

josh
--
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/