Re: [PATCH] firmware: arm_scmi: optee: Drop the support for the OPTEE shared dynamic buffer

[Etienne Carriere]

Hi Sudeep,

(sorry, previous mail was empty)

On Fri, 26 Nov 2021 at 15:28, Sudeep Holla <sudeep.holla@xxxxxxx> wrote:
>
> On Fri, Nov 26, 2021 at 08:59:45AM +0100, Etienne Carriere wrote:
> > Hello Sudeep,
> >
> > On Thu, 25 Nov 2021 at 19:25, Cristian Marussi <cristian.marussi@xxxxxxx> wrote:
> > >
> > > On Thu, Nov 25, 2021 at 03:07:30PM +0000, Sudeep Holla wrote:
> > > > The shared memory buffer allocated by the optee driver is normal cached
> > > > memory and can't be used with IOMEM APIs used in shmem_*.
> > > >
> > > > We currently support only IO memory for shared memory and supporting
> > > > normal cached memory needs more changes and needs to be thought through
> > > > properly. So for now, let us drop the support for this OPTEE shared buffer.
> > > >
> > > > Cc: Cristian Marussi <cristian.marussi@xxxxxxx>
> > > > Cc: Etienne Carriere <etienne.carriere@xxxxxxxxxx>
> > > > Signed-off-by: Sudeep Holla <sudeep.holla@xxxxxxx>
> > > > ---
> > >
> > > Hi,
> > >
> > >
> > > >  drivers/firmware/arm_scmi/optee.c | 19 +------------------
> > > >  1 file changed, 1 insertion(+), 18 deletions(-)
> > > >
> > > > diff --git a/drivers/firmware/arm_scmi/optee.c b/drivers/firmware/arm_scmi/optee.c
> > > > index 901737c9f5f8..175b39bcd470 100644
> > > > --- a/drivers/firmware/arm_scmi/optee.c
> > > > +++ b/drivers/firmware/arm_scmi/optee.c
> > > > @@ -282,23 +282,6 @@ static void scmi_optee_clear_channel(struct scmi_chan_info *cinfo)
> > > >       shmem_clear_channel(channel->shmem);
> > > >  }
> > > >
> > > > -static int setup_dynamic_shmem(struct device *dev, struct scmi_optee_channel *channel)
> > > > -{
> > > > -     const size_t msg_size = SCMI_OPTEE_MAX_MSG_SIZE;
> > > > -
> > > > -     channel->tee_shm = tee_shm_alloc_kernel_buf(scmi_optee_private->tee_ctx, msg_size);
> > > > -     if (IS_ERR(channel->tee_shm)) {
> > > > -             dev_err(channel->cinfo->dev, "shmem allocation failed\n");
> > > > -             return -ENOMEM;
> > > > -     }
> > > > -
> > > > -     channel->shmem = (void *)tee_shm_get_va(channel->tee_shm, 0);
> > > > -     memset(channel->shmem, 0, msg_size);
> > > > -     shmem_clear_channel(channel->shmem);
> > > > -
> > > > -     return 0;
> > > > -}
> > > > -
> > > >  static int setup_static_shmem(struct device *dev, struct scmi_chan_info *cinfo,
> > > >                             struct scmi_optee_channel *channel)
> > > >  {
> > > > @@ -342,7 +325,7 @@ static int setup_shmem(struct device *dev, struct scmi_chan_info *cinfo,
> > > >       if (of_find_property(cinfo->dev->of_node, "shmem", NULL))
> > > >               return setup_static_shmem(dev, cinfo, channel);
> > > >       else
> > > > -             return setup_dynamic_shmem(dev, channel);
> > > > +             return -ENOMEM;
> > > >  }
> > > >
> >
> > I would rather find an alternate way to support tee shared memory.
>
> Sure
>
> > I think OP-TEE could use msg.c format when handling tee memory.
>
> Okay
>
> > Linux and OP-TEE Scmi transport discovery negotiate the channel type and
> > support for msg format could allow OP-TEE to use its shm management.
>
> I am fine with that, just that what we have in for-next/scmi is broken and
> I want to remove the support just because it is buggy and not because I
> disagree with the requirement.
>
> > I will prepare an implementation but if you prefer the current remove
> > support and later introduce back tee shm support, I'm fine.
> >
>
> Sure, we may need to support this in a generic way. I mean in a way, other
> transport can also use them if they need it. I remember someone else had
> asked this in the past.
>
> So yes, I am happy to merge the support for tee shm when that is ready.
> What we have now is buggy and needs to be dropped. Sorry for not identifying
> it early.
>

Fine. Sorry for the burden.
`Reviewed-by: Etienne Carriere <etienne.carriere@xxxxxxxxxx>` for the changes
By the way Yes I tested tee shm setup on read hardware, with tee
cached shared memory. shmem.c uses ioread()/iowrite() that adds some
memory barriers. I don't think it hurts to do so when accessing cached
memory.

br,
etienne


> --
> Regards,
> Sudeep