Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 6 Oct 2022 16:27:04 -0700
From:   Kees Cook <keescook@chromium.org>
To:     "Guilherme G. Piccoli" <gpiccoli@igalia.com>
Cc:     linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org,
        kernel-dev@igalia.com, kernel@gpiccoli.net, anton@enomsg.org,
        ccross@android.com, tony.luck@intel.com
Subject: Re: [PATCH 4/8] pstore: Alert on backend write error
Message-ID: <202210061625.950B43C119@keescook>
References: <20221006224212.569555-1-gpiccoli@igalia.com>
 <20221006224212.569555-5-gpiccoli@igalia.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20221006224212.569555-5-gpiccoli@igalia.com>
Precedence: bulk

On Thu, Oct 06, 2022 at 07:42:08PM -0300, Guilherme G. Piccoli wrote:
> The pstore dump function doesn't alert at all on errors - despite
> pstore is usually a last resource and if it fails users won't be
> able to read the kernel log, this is not the case for server users
> with serial access, for example.
> 
> So, let's at least attempt to inform such advanced users on the first
> backend writing error detected during the kmsg dump - this is also
> very useful for pstore debugging purposes.
> 
> Signed-off-by: Guilherme G. Piccoli <gpiccoli@igalia.com>
> ---
>  fs/pstore/platform.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/fs/pstore/platform.c b/fs/pstore/platform.c
> index 06c2c66af332..ee50812fdd2e 100644
> --- a/fs/pstore/platform.c
> +++ b/fs/pstore/platform.c
> @@ -463,6 +463,9 @@ static void pstore_dump(struct kmsg_dumper *dumper,
>  		if (ret == 0 && reason == KMSG_DUMP_OOPS) {
>  			pstore_new_entry = 1;
>  			pstore_timer_kick();
> +		} else {
> +			pr_err_once("backend (%s) writing error (%d)\n",
> +				    psinfo->name, ret);

We're holding a spinlock here, so doing a pr_*() call isn't a great
idea. It's kind of not a great idea to try to write to the log in the
middle of a dump either, but we do attempt it at the start.

Perhaps keep a saved_ret or something and send it after the spin lock is
released?

-- 
Kees Cook