Re: [PATCH] remoteproc: core: check state in rproc_boot

[Bjorn Andersson <bjorn.andersson@linaro.org>]

On Thu 19 May 01:41 CDT 2022, Peng Fan (OSS) wrote:

> From: Peng Fan <peng.fan@nxp.com>
> 
> If remote processor has already been in RUNNING or ATTACHED
> state, report it. Not just increment the power counter and return
> success.
> 
> Without this patch, if m7 is in RUNNING state, and start it again,
> nothing output to console.
> If wanna to stop the m7, we need write twice 'stop'.
> 
> This patch is to improve that the 2nd start would show some useful
> info.
> 
> Signed-off-by: Peng Fan <peng.fan@nxp.com>
> ---
> 
> Not sure to keep power counter or not.
> 

I did discuss this with Mathieu, whom argued in favor of keeping the
refcount mechanism.

I can see that there could be a scenario where multiple user-space
components keep the remotproc running while they are, and if there is
any such user this ABI change would be a breakage.

That said, it's more than once that I accidentally have bumped the
refcount and then assumed that a single stop would tear down the
remoteproc...

>  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 02a04ab34a23..f37e0758c096 100644
> --- a/drivers/remoteproc/remoteproc_core.c
> +++ b/drivers/remoteproc/remoteproc_core.c
> @@ -2005,6 +2005,12 @@ int rproc_boot(struct rproc *rproc)
>  		goto unlock_mutex;
>  	}
>  
> +	if (rproc->state == RPROC_RUNNING || rproc->state == RPROC_ATTACHED) {

If we were to do this would it make sense to boot it out of anything but
RPROC_OFFLINE?

Regards,
Bjorn

> +		ret = -EINVAL;
> +		dev_err(dev, "%s already booted\n", rproc->name);
> +		goto unlock_mutex;
> +	}
> +
>  	/* skip the boot or attach process if rproc is already powered up */
>  	if (atomic_inc_return(&rproc->power) > 1) {
>  		ret = 0;
> -- 
> 2.25.1
>