linux-kernel - Re: [PATCH] exec: do not leave bprm->interp on stack

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]

Message-ID: <alpine.LFD.2.02.1211240004060.15883@wniryva.cad.erqung.pbz>
Date:	Sat, 24 Nov 2012 00:13:45 +0530 (IST)
From:	P J P <ppandit@...hat.com>
To:	Kees Cook <keescook@...omium.org>
cc:	Al Viro <viro@...iv.linux.org.uk>, linux-kernel@...r.kernel.org,
	Andrew Morton <akpm@...ux-foundation.org>,
	Josh Triplett <josh@...htriplett.org>,
	Serge Hallyn <serge.hallyn@...onical.com>,
	linux-fsdevel@...r.kernel.org, halfdog <me@...fdog.net>
Subject: Re: [PATCH] exec: do not leave bprm->interp on stack


   Hello Kees, all,

Please have a look at a *NEW* patch at the end of this mail. It seems to fix 
both the issues, stack disclosure + undue recursions.

It uses modprobe "--first-time" option which returns an error code when trying 
to load a module which is already present or unload one which is not present.

Checking this return value at - request_module() - and breaking the loop in 
case of an error seems to be doing the trick.

+-- On Mon, 19 Nov 2012, Kees Cook wrote --+
| I think to avoid the explosion of request_module calls in the abusive case, 
| we could simply return ELOOP instead of ENOEXEC on max recursion.

 -> http://www.spinics.net/lists/mm-commits/msg92433.html

1. returning -ELOOP has a side effect of not reaching to request_module()
   ever, for:
==
#ifdef CONFIG_MODULES
1415                 if (retval != -ENOEXEC || bprm->mm == NULL) {
1416                         break;
1417                 } else {
                        ...
==

2. above patch does not seem to fix the 2^6(64) recursions issue, for:
==
+                       bprm->recursion_depth = depth + 1;
                        retval = fn(bprm);
                        bprm->recursion_depth = depth;
==

setting - recursion_dept = depth - again and the outer for(try=0;try<2...) 
loop seems to be causing the 2^6 recursions.


Please have a look at this *NEW* patch to fix both the issues, stack 
disclosure + undue recursions.

===
diff --git a/fs/exec.c b/fs/exec.c
index 0039055..dec467f 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -1423,7 +1423,14 @@ int search_binary_handler(struct linux_binprm *bprm,struct pt_regs *regs)
 				break; /* -ENOEXEC */
 			if (try)
 				break; /* -ENOEXEC */
-			request_module("binfmt-%04x", *(unsigned short *)(&bprm->buf[2]));
+            if (request_module("binfmt-%04x",
+                                *(unsigned short *)(&bprm->buf[2])))
+            {
+                printk(KERN_WARNING
+                        "request_module: failed to load: binfmt-%04x",
+                         *(unsigned short *)(&bprm->buf[2]));
+                break;
+            }
 		}
 #else
 		break;
diff --git a/kernel/kmod.c b/kernel/kmod.c
index 1c317e3..7ec0e3e 100644
--- a/kernel/kmod.c
+++ b/kernel/kmod.c
@@ -83,7 +83,7 @@ static int call_modprobe(char *module_name, int wait)
 		NULL
 	};
 
-	char **argv = kmalloc(sizeof(char *[5]), GFP_KERNEL);
+	char **argv = kmalloc(sizeof(char *[6]), GFP_KERNEL);
 	if (!argv)
 		goto out;
 
@@ -93,9 +93,10 @@ static int call_modprobe(char *module_name, int wait)
 
 	argv[0] = modprobe_path;
 	argv[1] = "-q";
-	argv[2] = "--";
-	argv[3] = module_name;	/* check free_modprobe_argv() */
-	argv[4] = NULL;
+	argv[2] = "--first-time";
+	argv[3] = "--";
+	argv[4] = module_name;	/* check free_modprobe_argv() */
+	argv[5] = NULL;
 
 	return call_usermodehelper_fns(modprobe_path, argv, envp,
 		wait | UMH_KILLABLE, NULL, free_modprobe_argv, NULL);
===



Thank you.
--
Prasad J Pandit / Red Hat Security Response Team
DB7A 84C5 D3F9 7CD1 B5EB  C939 D048 7860 3655 602B
--
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/