linux-kernel - Re: [PATCH V10] devcoredump: add context check in dev

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

Date:   Mon, 26 Sep 2022 10:36:13 +0200
From:   Greg KH <gregkh@...uxfoundation.org>
To:     Duoming Zhou <duoming@....edu.cn>
Cc:     linux-kernel@...r.kernel.org, johannes@...solutions.net,
        rafael@...nel.org
Subject: Re: [PATCH V10] devcoredump: add context check in dev_coredumpm

On Mon, Sep 26, 2022 at 02:51:26PM +0800, Duoming Zhou wrote:
> The dev_coredumpm(), dev_coredumpv() and dev_coredumpsg() could not
> be used in atomic context, because they call kvasprintf_const() and
> kstrdup() with GFP_KERNEL parameter. The process is shown below:
> 
> dev_coredumpv(.., gfp_t gfp)
>   dev_coredumpm(.., gfp_t gfp)
>     dev_set_name
>       kobject_set_name_vargs
>         kvasprintf_const(GFP_KERNEL, ...); //may sleep
>           kstrdup(s, GFP_KERNEL); //may sleep
> 
> This patch adds context check in dev_coredumpm() in order to show
> dev_coredumpm() and its callers could not be used in atomic context.
> 
> What's more, this change can allow the api to evolve and will not
> influence the users that call this api.
> 
> Signed-off-by: Duoming Zhou <duoming@....edu.cn>
> ---
> Changes since v9:
>   - v9: Add context check in dev_coredumpm().
>   - v10: Remove the "Fixes" tag.
> 
>  drivers/base/devcoredump.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/base/devcoredump.c b/drivers/base/devcoredump.c
> index f4d794d6bb8..806ee872f5f 100644
> --- a/drivers/base/devcoredump.c
> +++ b/drivers/base/devcoredump.c
> @@ -255,6 +255,9 @@ void dev_coredumpm(struct device *dev, struct module *owner,
>  	struct devcd_entry *devcd;
>  	struct device *existing;
>  
> +	if (!gfpflags_normal_context(gfp))
> +		goto free;

I still do not know why this is needed at all as there are no in-kernel
abusers of this api.

We do not add code to the kernel for out-of-tree users, or "future
users" at all, otherwise it would be unmaintable.

thanks,

greg k-h