linux-kernel - Re: [PATCH v1] fs/super.c: Add NULL check for type in iterate_supers

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>] [day] [month] [year] [list]

Message-ID: <20250403024756.GL2023217@ZenIV>
Date: Thu, 3 Apr 2025 03:47:56 +0100
From: Al Viro <viro@...iv.linux.org.uk>
To: Xiaole He <hexiaole1994@....com>
Cc: James Bottomley <James.Bottomley@...senpartnership.com>,
	"brauner@...nel.org" <brauner@...nel.org>,
	"jack@...e.cz" <jack@...e.cz>,
	"linux-fsdevel@...r.kernel.org" <linux-fsdevel@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v1] fs/super.c: Add NULL check for type in
 iterate_supers_type

On Thu, Apr 03, 2025 at 10:37:17AM +0800, Xiaole He wrote:
> Thank you for your feedback.
> While I acknowledge your points, I would like to clarify the rationale
> behind submitting this patch.
> During my experimentation with an external module interacting with the
> superblock, I utilized iterate_supers_type (from fs/super.c) as it is
> an exported symbol. However, I observed a potential vulnerability in
> its implementation: the type argument can be passed as NULL, leading
> to a null pointer dereference. To verify this, I deliberately triggered
> a scenario where type was set to NULL, resulting in the following dmesg
> output:

> After this observasion, I worry about if this vulnerability can cause
> the whole kernel crash if the type argument is passed by a
> unintentional NULL in the kernel code rather than in the external
> module.
> Thus I submitted the patch to address the missing null-check.
> Thank you for your review.

You do realize that passing it NULL as the second (function pointer) argument
would also oops, right?  Passing (void (*)(struct super_block *))kfree
there would do even more unpleasant things, etc.

Sure, it's exported - so's strlen().  While we are at it, checking just for
NULL is not the limit - what if the caller gives it ERR_PTR(...) as argument?