linux-kernel - Re: [PATCH] lkdtm: Add kfence read after free crash type

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, 29 Nov 2023 12:22:27 -0800
From:   Kees Cook <keescook@...omium.org>
To:     Stephen Boyd <swboyd@...omium.org>
Cc:     linux-kernel@...r.kernel.org, patches@...ts.linux.dev,
        Arnd Bergmann <arnd@...db.de>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Subject: Re: [PATCH] lkdtm: Add kfence read after free crash type

On Mon, Nov 27, 2023 at 03:49:45PM -0800, Stephen Boyd wrote:
> Add the ability to allocate memory from kfence and trigger a read after
> free on that memory to validate that kfence is working properly. This is
> used by ChromeOS integration tests to validate that kfence errors can be
> collected on user devices and parsed properly.

This looks really good; thanks for adding this!

> 
> Signed-off-by: Stephen Boyd <swboyd@...omium.org>
> ---
>  drivers/misc/lkdtm/heap.c | 64 +++++++++++++++++++++++++++++++++++++++
>  1 file changed, 64 insertions(+)
> 
> diff --git a/drivers/misc/lkdtm/heap.c b/drivers/misc/lkdtm/heap.c
> index 0ce4cbf6abda..608872bcc7e0 100644
> --- a/drivers/misc/lkdtm/heap.c
> +++ b/drivers/misc/lkdtm/heap.c
> @@ -4,6 +4,7 @@
>   * page allocation and slab allocations.
>   */
>  #include "lkdtm.h"
> +#include <linux/kfence.h>
>  #include <linux/slab.h>
>  #include <linux/vmalloc.h>
>  #include <linux/sched.h>
> @@ -132,6 +133,66 @@ static void lkdtm_READ_AFTER_FREE(void)
>  	kfree(val);
>  }
>  
> +#if IS_ENABLED(CONFIG_KFENCE)

I really try hard to avoid having tests disappear depending on configs,
and instead report the expected failure case (as you have). Can this be
built without the IS_ENABLED() tests?

-- 
Kees Cook