[PATCH 2/2] remoteproc: imx_dsp_rproc: Wait for suspend ACK only if WAIT_FW_READY is set

[Bjorn Andersson]

On Tue, Nov 25, 2025 at 02:49:03PM +0200, Iuliana Prodan (OSS) wrote:
> From: Iuliana Prodan <iuliana.prodan at nxp.com>
> 
> The DSP suspend path currently waits unconditionally
> for a suspend ack from the firmware.
> This breaks firmwares that do not implement the
> mailbox-based READY handshake, as the DSP never
> responds and system suspend fails with -EBUSY.
> 

But if the firmware doesn't implement "mailbox-based READY handshake",
do you still want to send the RP_MBOX_SUSPEND_SYSTEM message?

If so, can you clarify here in the commit message that the firmware
expects the mailbox-based message, and only the "handshake" part should
be omitted.

If that part isn't implemented either, then I think you should fix the
code to not poke the mailbox in the first place.


Also, wrap your commit message at 75 characters, please.

> The driver already uses the WAIT_FW_READY flag to
> indicate that the firmware supports the READY
> handshake at boot. Apply the same logic during
> suspend: only wait for the suspend ack when the
> firmware is expected to support it.
> 
> Signed-off-by: Iuliana Prodan <iuliana.prodan at nxp.com>
> ---
>  drivers/remoteproc/imx_dsp_rproc.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/remoteproc/imx_dsp_rproc.c b/drivers/remoteproc/imx_dsp_rproc.c
> index fc0470aa72c1..e25dbe32ef79 100644
> --- a/drivers/remoteproc/imx_dsp_rproc.c
> +++ b/drivers/remoteproc/imx_dsp_rproc.c
> @@ -1327,10 +1327,11 @@ static int imx_dsp_suspend(struct device *dev)
>  	}
>  
>  	/*
> -	 * DSP need to save the context at suspend.
> -	 * Here waiting the response for DSP, then power can be disabled.
> +	 * The DSP must save its context during suspend.

Please double check that this comment reflect above conclusion.

Regards,
Bjorn

> +	 * Wait for a response from the DSP if required before disabling power.
>  	 */
> -	if (!wait_for_completion_timeout(&priv->pm_comp, msecs_to_jiffies(100)))
> +	if (priv->flags & WAIT_FW_READY &&
> +	    !wait_for_completion_timeout(&priv->pm_comp, msecs_to_jiffies(100)))
>  		return -EBUSY;
>  
>  out:
> -- 
> 2.34.1
>