Re: [PATCH v4 0/3] Introduce support for T-head TH1520 Mailbox

[Drew Fustini <dfustini@tenstorrent.com>]

On Wed, Oct 16, 2024 at 12:03:05AM +0200, Michal Wilczynski wrote:
> 
> 
> On 10/14/24 20:31, Drew Fustini wrote:
> > On Mon, Oct 14, 2024 at 02:33:11PM +0200, Michal Wilczynski wrote:
> >> The T-head TH1520 SoC supports a hardware mailbox that enables two cores
> >> within the SoC to communicate and coordinate [1]. One example of such
> >> coordination would be cooperation with the T-Head E902 core, which is
> >> responsible for power, clock, and resource management. For example, in
> >> the specific case of the BXM-4-64 GPU, it needs to be powered on by the
> >> E902 core, and the kernel running on the E910 needs to 'ask' the
> >> firmware running on the E902 core to enable power to the GPU island.
> >> Given recent advancements in work on the upstream GPU driver [2], there
> >> is an emerging need to get this code in the mainline kernel.
> >>
> >> Link: https://protect2.fireeye.com/v1/url?k=2021d256-7fbdfb7c-20205919-000babe598f7-ca654d1a9bc866ac&q=1&e=11e97355-e6e9-4aac-a996-cc475156b3c8&u=https%3A%2F%2Fopenbeagle.org%2Fbeaglev-ahead%2Fbeaglev-ahead%2F-%2Fblob%2Fmain%2Fdocs%2FTH1520%2520System%2520User%2520Manual.pdf [1]
> >> Link: https://gitlab.freedesktop.org/imagination/linux-firmware/-/issues/1 [2]
> >>
> >> Thanks, Krzysztof and Rob, for your review! Since this series is gaining
> >> some interest, I've dropped the RFC prefix with the v3 update.
> > 
> > I've applied this series and booted okay. I see the driver loaded:
> > 
> >  /sys/devices/platform/soc/ffffc38000.mailbox/driver points to
> >  /sys/bus/platform/drivers/th1520-mbox
> > 
> > How do you test that the communication with the E902 is working
> > correctly?
> 
> Thank you for your interest. To test this, I've prepared a diff that
> includes the missing drivers utilizing the mailbox and enabled the GPU
> node in the device tree to use the drm/imagination driver.
> 
> I've observed that when the power was turned off through the E902 core
> using the mailbox, the drm/imagination driver would hang in
> pvr_load_gpu_id() while attempting to read its BVNC from the register.
> However, when the GPU was turned on via the mailbox, the BVNC could be
> read correctly. Still, the firmware fails to boot due to some missing
> programming in the drm/imagination driver, which is currently being
> worked on. I've briefly explained this in the first commit of this
> series.
> 
> If you'd like to try this yourself, I'd be happy to push these setups to
> a GitHub repository and provide you with a link, so you can see the
> setup in action.

I think that would be helpful for myself and others to be able to see
the interaction.

Thanks,
Drew