Message ID | 20240314-pinctrl-scmi-v5-0-b19576e557f2@nxp.com |
---|---|
Headers | show |
Series | firmware: arm_scmi: Add SCMI v3.2 pincontrol protocol basic support | expand |
On Fri, Mar 15, 2024 at 12:31:51AM +0000, Peng Fan wrote: > > Subject: Re: [PATCH v5 0/4] firmware: arm_scmi: Add SCMI v3.2 pincontrol > > protocol basic support > > > > On Thu, Mar 14, 2024 at 09:35:17PM +0800, Peng Fan (OSS) wrote: > > > Since SCMI 3.2 Spec is released, and this patchset has got R-b/T-b, is > > > it ok to land this patchset? > > > > > > > I'll have a look at this last version and a spin on my test setup. > > > > ...but has this V5 change at all since the Reviewed-by tags due to the latest > > spec changes ? > > The tags are same as V4. I only did a rebase, no more changes. > > Ok. > > ...IOW does this V5 include the latest small bits spec-changes or those latest > > gpio-related spec-changes are just not needed at the level of the Linux pinctrl > > support as of now and can be added later on when a Linux gpio driver will be > > built on top of this ? > > In my current test, I no need the gpio related changes, so I would add that later > if you are ok. > I COULD have agreed with this, since AFAIK there is currently an effort to add support for GPIO on top of SCMI Pinctrl BUT not in Linux, so no reason to block this series for gpio-related missing features, that should only be additions not breaking backward compatibility... ....BUT, I've just wrapped my head again around the latest public release of v3.2 spec (which has gone through so many changes and additions that I had lost track O_o) AND beside the above mentioned GPIO changes there are indeed also BREAKING changes around the commands PINCTRL_SETTINGS_GET and PINCTRL_SETTINGS_CONFIGURE (which were the old PINCTRL_CONFIG_GET/SET), that now also get/set the selected function: so that, at the end the payload itself of those commands/replies has also changed IN SIZE, so the driver needs definitely to be updated (and whatever you use to test on the backend server too, if you want to test this...) I think these changes (which I forgot being there) were in since last month, so already V4 was broken in these regards (which I have not looked at) I'll leave some comments along the series and test all of this again next week... ...since too many things has changed and I want to re-verify all on my side. Thanks, Cristian
On Fri, Mar 15, 2024 at 04:53:10PM +0000, Cristian Marussi wrote: > On Fri, Mar 15, 2024 at 12:31:51AM +0000, Peng Fan wrote: > (and whatever you use to test on the backend server too, if you want > to test this...) > What are people using to test this, btw? regards, dan carpenter
On Fri, Mar 15, 2024 at 09:20:02PM +0300, Dan Carpenter wrote: > On Fri, Mar 15, 2024 at 04:53:10PM +0000, Cristian Marussi wrote: > > On Fri, Mar 15, 2024 at 12:31:51AM +0000, Peng Fan wrote: > > (and whatever you use to test on the backend server too, if you want > > to test this...) > > > > What are people using to test this, btw? Hi Dan, I think NXP has their own SCMI server embedded in TF-A (Peng posted a link at the public repo a while ago I think...) supporting Pinctrl; additionally Oleksii/EPAM (the original author of this series) have their own distinct proprietary SCMI server implementation with Pinctrl (not sure where it run from in this case but it is a Xen based setup if I remember right..). Beside these, there are at least a couple more Vendor proprietary incarnations of SCMI servers that I am aware of (that most probably did not support Pinctrl anyway as of now...) On top of this, on the other side there is, of course, the official SCP/SCMI server reference implementation, that can live in a number of different places thanks to the the virtualized environment built by Linaro, but this latter SCP/SCMI server does not have any official Pinctrl support either as of now. Thanks, Cristian