linux-kernel - Re: [RFC 2/2] remoteproc: core: keep rproc in crash state in case of recovery failure

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:   Tue, 5 May 2020 19:05:40 -0700
From:   Bjorn Andersson <bjorn.andersson@...aro.org>
To:     Loic Pallardy <loic.pallardy@...com>
Cc:     ohad@...ery.com, mathieu.poirier@...aro.org,
        linux-remoteproc@...r.kernel.org, linux-kernel@...r.kernel.org,
        arnaud.pouliquen@...com, benjamin.gaignard@...aro.org,
        fabien.dessenne@...com, s-anna@...com
Subject: Re: [RFC 2/2] remoteproc: core: keep rproc in crash state in case of
 recovery failure

On Wed 11 Mar 03:54 PDT 2020, Loic Pallardy wrote:

> When an error occurs during recovery procedure, internal rproc
> variables may be unaligned:
> - state is set to RPROC_OFFLINE
> - power atomic not equal to 0
> which is normal as only rproc_stop() has been executed and not
> rproc_shutdown()
> 
> In such case, rproc_boot() can be re-executed by client to
> reboot co-processor.
> 
> This patch proposes to keep rproc in RPROC_CRASHED state in case
> of recovery failure to be coherent with recovery disabled mode.
> 
> Signed-off-by: Loic Pallardy <loic.pallardy@...com>
> ---
>  drivers/remoteproc/remoteproc_core.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/remoteproc/remoteproc_core.c b/drivers/remoteproc/remoteproc_core.c
> index 7ac87a75cd1b..def4f9fc881d 100644
> --- a/drivers/remoteproc/remoteproc_core.c
> +++ b/drivers/remoteproc/remoteproc_core.c
> @@ -1679,6 +1679,12 @@ int rproc_trigger_recovery(struct rproc *rproc)
>  	release_firmware(firmware_p);
>  
>  unlock_mutex:
> +	/*
> +	 * In case of error during recovery sequence restore rproc
> +	 * state in CRASHED
> +	 */
> +	if (ret)
> +		rproc->state = RPROC_CRASHED;

Got back to this after looking at Mathieu's synchronization series, I
think it would be cleaner if we move the rproc->state update out of
rproc_start() and rproc_stop().

That way we would leave the state in CRASHED state throughout the
recovery process, which I think makes it easier to reason about the
various states and their transitions.

Regards,
Bjorn

>  	mutex_unlock(&rproc->lock);
>  	return ret;
>  }
> -- 
> 2.7.4
>