From: Torsten Duwe <duwe@caldera.de>
To: Gregory Maxwell <greg@linuxpower.cx>
Cc: linux-kernel@vger.kernel.org
Subject: Modprobe local root exploit
Date: Mon, 13 Nov 2000 17:26:00 +0100 (CET) [thread overview]
Message-ID: <14864.5656.706778.275865@ns.caldera.de> (raw)
In-Reply-To: <20001113093727.C1918@xi.linuxpower.cx>
In-Reply-To: <20001113093727.C1918@xi.linuxpower.cx>
>>>>> "Gregory" == Gregory Maxwell <greg@linuxpower.cx> writes:
Gregory> After seeing the modprobe local root exploit today, I asked
Gregory> myself why kmod executes modprobe with full root and doesn't
Gregory> drop some capabilities first.
Gregory> Why? It wouldn't close the hole, but it would narrow it down.
This might also be a good idea; but my suggestion is to not allow arbitrary
strings as module names in the first place. As far as I can see, all valid
strings for KMOD requests consist of alphanumeric chars plus dash and
underscore. Anybody with autoloaded modules that don't fit this pattern even
after /etc/modules.conf translation please object !
Here's the patch...
Torsten
--- linux/kernel/kmod.c.orig Tue Sep 26 01:18:55 2000
+++ linux/kernel/kmod.c Mon Nov 13 16:57:02 2000
@@ -168,6 +168,22 @@
static atomic_t kmod_concurrent = ATOMIC_INIT(0);
#define MAX_KMOD_CONCURRENT 50 /* Completely arbitrary value - KAO */
static int kmod_loop_msg;
+ const char * p;
+
+ /* For security reasons ensure the requested name consists
+ * only of allowed characters. Especially whitespace and
+ * shell metacharacters might confuse modprobe.
+ */
+ for (p = module_name; *p; p++)
+ {
+ if ((*p & 0xdf) >= 'a' && (*p & 0xdf) <= 'z')
+ continue;
+ if (*p >= '0' && *p <= '9')
+ continue;
+ if (*p == '_' || *p == '-')
+ continue;
+ return -EINVAL;
+ }
/* Don't allow request_module() before the root fs is mounted! */
if ( ! current->fs->root ) {
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/
next prev parent reply other threads:[~2000-11-13 16:26 UTC|newest]
Thread overview: 28+ messages / expand[flat|nested] mbox.gz Atom feed top
2000-11-13 14:37 Modprobe local root exploit Gregory Maxwell
2000-11-13 16:26 ` Torsten Duwe [this message]
2000-11-13 16:44 ` Francis Galiegue
2000-11-13 16:45 ` Torsten Duwe
2000-11-13 16:56 ` Chris Evans
2000-11-13 17:21 ` Jan Dvorak
2000-11-13 18:11 ` Torsten Duwe
2000-11-14 5:02 ` Peter Samuelson
2000-11-14 5:50 ` Keith Owens
2000-11-14 9:19 ` Florian Weimer
2000-11-14 10:42 ` Malcolm Beattie
2000-11-14 10:54 ` Jakub Jelinek
2000-11-14 11:58 ` Chris Evans
2000-11-14 10:58 ` Keith Owens
2000-11-14 12:28 ` Nick Holloway
2000-11-14 14:01 ` David Woodhouse
2000-11-14 1:35 ` Horst von Brand
2000-11-13 19:46 ` Peter Samuelson
2000-11-14 11:29 ` Daniel Phillips
2000-11-14 14:23 ` Daniel Phillips
2000-11-14 16:25 ` David Relson
2000-11-15 4:09 ` Horst von Brand
2000-11-16 5:22 ` Alan Cox
2000-11-16 6:04 ` H. Peter Anvin
2000-11-16 6:14 ` Keith Owens
2000-11-16 6:16 ` H. Peter Anvin
2000-11-16 14:12 ` Torsten Duwe
2000-11-16 15:07 ` Alan Cox
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=14864.5656.706778.275865@ns.caldera.de \
--to=duwe@caldera.de \
--cc=Torsten.Duwe@caldera.de \
--cc=greg@linuxpower.cx \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox