Re: [PATCH v20 2/4] remoteproc/mediatek: add SCP support for mt8183

[Pi-Hsun Shih]

Hi,
Thanks for the review, I'll address them in the next version. Some
inline comment below.

On Tue, Nov 12, 2019 at 6:53 AM Bjorn Andersson
<bjorn.andersson@xxxxxxxxxx> wrote:
>
> On Mon 14 Oct 00:58 PDT 2019, Pi-Hsun Shih wrote:
> > diff --git a/drivers/remoteproc/mtk_scp.c b/drivers/remoteproc/mtk_scp.c
> [..]
> > +struct platform_device *scp_get_pdev(struct platform_device *pdev)
>
> I'm unable to find a patch that calls this, but I assume you're only
> using the returned struct platform_device * in order to call the other
> exported functions in this driver.

Some more information:

Patches for drivers that are using this function includes:
https://patchwork.kernel.org/patch/11126059/
https://patchwork.kernel.org/patch/11134913/
https://patchwork.kernel.org/patch/11135073/
https://patchwork.kernel.org/patch/11138511/
https://patchwork.kernel.org/patch/11140755/

The returned platform_device are used either:
* As a pointer passing back to the scp_ipi_{register,unregister,send} APIs
  # This is the case above.
* Use the ->dev field for either passing to dma_alloc_coherent
(11134913, 11138511), or logging
(https://patchwork.kernel.org/patch/11140755/ mdp_vpu_register).
  # Probably would need to export another function for mtk_scp* ->
device* if going for this change.

A particular problematic patch for this change is
https://patchwork.kernel.org/patch/11135073/, which stores both
platform_device from SCP or VPU in the same field, but it can be
changed to two different fields.

>
> If this is the case I would suggest that you return a struct mtk_scp *
> instead, as this makes your API cleaner and prevents confusion about
> what platform_device could/should be passed in.
>
> Note that you don't need to disclose the struct mtk_scp to your clients,
> just add a "struct mtk_scp;" in include/remoteproc/mtk_scp.h and your
> clients can pass this pointer around.

Ok I'll change to this.

> > +             return -ENOMEM;
> > +     }
> > +
> > +     /* Reserved SCP code size */
> > +     scp->dram_size = MAX_CODE_SIZE;
> > +     scp->cpu_addr = dma_alloc_coherent(scp->dev, scp->dram_size,
> > +                                        &scp->phys_addr, GFP_KERNEL);
>
> Don't you have a problem with that the reserved memory region must be
> 8MB for this allocation to succeed? If so, consider using devm_ioremap
> or similar to map the region.

Yes the reserved memory need to be large enough.
There are other drivers (https://patchwork.kernel.org/patch/11134913/,
https://patchwork.kernel.org/patch/11138511/) that also use
dma_alloc_coherent on the same reserved memory, so we need to use
dma_alloc_coherent here too.

It seems to be problematic if this dma_alloc_coherent is not called
before the other two dma_alloc_coherent, I'll check this.

> [...]
> Regards,
> Bjorn