linux-kernel - Re: [PATCH] UBI: dereference after kfree in create

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: <463BC019.40305@gmail.com>
Date:	Fri, 04 May 2007 19:22:01 -0400
From:	Florin Malita <fmalita@...il.com>
To:	Satyam Sharma <satyam.sharma@...il.com>
CC:	dedekind@...radead.org, linux-mtd@...ts.infradead.org,
	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] UBI: dereference after kfree in create_vtbl

Hi Satyam,

Satyam Sharma wrote:
> Eeks ... no, wait. You found a (two, actually) bug alright, but fixed
> it wrong. When we fail a write, we *must* add it to the corrupted list
> and _then_ attempt to retry. So, the "if (++tries <= 5)" applies to
> "if (!err) goto retry;" and not to the ubi_scan_add_to_list(). The
> difference is quite subtle here ...

Not being familiar with the code, I was specifically trying to preserve 
the old semantics and only address the use-after-free issue. So if there 
was another bug... well, I guess I succeeded at preserving it ;)

> The correct fix should actually be as follows: (Artem, this is diffed
> on the original vtbl.c)
[snip]
> +    err = ubi_scan_add_to_list(si, new_seb->pnum, new_seb->ec, 
> &si->corr);
> +    kfree(new_seb);
> +    if (++tries <= 5)
>         if (!err)
>             goto retry;

There's a side effect to this change: by unconditionally overwriting err 
we lose the original error code. Then if we're exceeding the number of 
retries we'll end up returning 0 which is probably not what you want.

Return code aside, it seems the only thing ubi_scan_add_to_list() is 
doing is allocate a new struct ubi_scan_leb, initialize some fields with 
values passed from new_seb and then add it to the desired list. But 
copying new_seb to a newly allocated structure and then immediately 
freeing the old one seems redundant - why not just add new_seb to the 
corrupted list and be done? Then we don't have to deal with allocation 
failures in an error path anymore - something like this (diff against 
the original code):

Signed-off-by: Florin Malita <fmalita@...il.com>
---

diff --git a/drivers/mtd/ubi/vtbl.c b/drivers/mtd/ubi/vtbl.c
index b6fd6bb..2ad2d59 100644
--- a/drivers/mtd/ubi/vtbl.c
+++ b/drivers/mtd/ubi/vtbl.c
@@ -317,14 +317,10 @@ retry:
 	return err;

 write_error:
-	kfree(new_seb);
-	/* May be this physical eraseblock went bad, try to pick another one */
-	if (++tries <= 5) {
-		err = ubi_scan_add_to_list(si, new_seb->pnum, new_seb->ec,
-					   &si->corr);
-		if (!err)
-			goto retry;
-	}
+	/* Maybe this physical eraseblock went bad, try to pick another one */
+	list_add_tail(&new_seb->u.list, &si->corr);
+	if (++tries <= 5)
+		goto retry;
 out_free:
 	ubi_free_vid_hdr(ubi, vid_hdr);
 	return err;

-
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/