Re: [PATCH v2 0/7] Add dts for mt8183 GPU (and misc panfrost patches)

[Robin Murphy]

On 09/01/2020 12:01 pm, Steven Price wrote:
> On 08/01/2020 05:23, Nicolas Boichat wrote:
> > Hi!
> >
> > Sorry for the long delay since 
> > https://patchwork.kernel.org/patch/11132381/,
> > finally got around to give this a real try.
> >
> > The main purpose of this series is to upstream the dts change and the 
> > binding
> > document, but I wanted to see how far I could probe the GPU, to check 
> > that the
> > binding is indeed correct. The rest of the patches are 
> > RFC/work-in-progress, but
> > I think some of them could already be picked up.
> >
> > So this is tested on MT8183 with a chromeos-4.19 kernel, and a ton of
> > backports to get the latest panfrost driver (I should probably try on
> > linux-next at some point but this was the path of least resistance).
> >
> > I tested it as a module as it's more challenging (originally probing 
> > would
> > work built-in, on boot, but not as a module, as I didn't have the power
> > domain changes, and all power domains are on by default during boot).
> >
> > Probing logs looks like this, currently:
> > [Â 221.867726] panfrost 13040000.gpu: clock rate = 511999970
> > [Â 221.867929] panfrost 13040000.gpu: Linked as a consumer to 
> > regulator.14
> > [Â 221.868600] panfrost 13040000.gpu: Linked as a consumer to 
> > regulator.31
> > [Â 221.870586] panfrost 13040000.gpu: Linked as a consumer to 
> > genpd:0:13040000.gpu
> > [Â 221.871492] panfrost 13040000.gpu: Linked as a consumer to 
> > genpd:1:13040000.gpu
> > [Â 221.871866] panfrost 13040000.gpu: Linked as a consumer to 
> > genpd:2:13040000.gpu
> > [Â 221.872427] panfrost 13040000.gpu: mali-g72 id 0x6221 major 0x0 
> > minor 0x3 status 0x0
> > [Â 221.872439] panfrost 13040000.gpu: features: 00000000,13de77ff, 
> > issues: 00000000,00000400
> > [Â 221.872445] panfrost 13040000.gpu: Features: L2:0x07120206 
> > Shader:0x00000000 Tiler:0x00000809 Mem:0x1 MMU:0x00002830 AS:0xff JS:0x7
> > [Â 221.872449] panfrost 13040000.gpu: shader_present=0x7 l2_present=0x1
> > [Â 221.873526] panfrost 13040000.gpu: error powering up gpu stack
> > [Â 221.878088] [drm] Initialized panfrost 1.1.0 20180908 for 
> > 13040000.gpu on minor 2
> > [Â 221.940817] panfrost 13040000.gpu: error powering up gpu stack
> > [Â 222.018233] panfrost 13040000.gpu: error powering up gpu stack
> > (repeated)
>
> It's interesting that it's only the stack that is failing. In hardware 
> there's a dependency: L2->stack->shader - so in theory the shader cores 
> shouldn't be able to power up either. There are some known hardware bugs 
> here though[1]:
>
>  ÂÂÂÂMODULE_PARM_DESC(corestack_driver_control,
>  ÂÂÂÂÂÂÂÂÂÂÂ "Let the driver power on/off the GPU core stack 
> independently "
>  ÂÂÂÂÂÂÂÂÂÂÂ "without involving the Power Domain Controller. This should "
>  ÂÂÂÂÂÂÂÂÂÂÂ "only be enabled on platforms for which integration of the 
> PDC "
>  ÂÂÂÂÂÂÂÂÂÂÂ "to the Mali GPU is known to be problematic.");
>
> [1] 
> https://github.com/ianmacd/d2s/blob/master/drivers/gpu/arm/b_r16p0/backend/gpu/mali_kbase_pm_driver.c#L57 
>
>
> It might be worth just dropping the code for powering up/down stacks and 
> let the GPU's own dependency management handle it.

FWIW I remember digging into that same message a while back (although 
I've forgotten which particular GPU I was playing with at the time), and 
concluded that the STACK_PWRON/STACK_READY registers might just not be 
implemented on some GPUs, and/or this easy-to-overlook register bit 
could be some kind of enable for the functionality:

https://github.com/ianmacd/d2s/blob/master/drivers/gpu/arm/b_r16p0/backend/gpu/mali_kbase_pm_driver.c#L1631

Since even in kbase this is all behind an 'expert' config option, I'm 
inclined to agree that just dropping it from panfrost unless and until 
it proves necessary is probably preferable to adding more logic and 
inscrutable register-magic.

Robin.

> Steve
>
> > So the GPU is probed, but there's an issue when powering up the STACK, 
> > not
> > quite sure why, I'll try to have a deeper look, at some point.
> >
> > Thanks!
> >
> > Nicolas
> >
> > v2:
> > Â - Use sram instead of mali_sram as SRAM supply name.
> > Â - Rename mali@ to gpu@.
> > Â - Add dt-bindings changes
> > Â - Stacking patches after the device tree change that allow basic
> > ÂÂÂ probing (still incomplete and broken).
> >
> > Nicolas Boichat (7):
> > ÂÂ dt-bindings: gpu: mali-bifrost: Add Mediatek MT8183
> > ÂÂ arm64: dts: mt8183: Add node for the Mali GPU
> > ÂÂ drm/panfrost: Improve error reporting in panfrost_gpu_power_on
> > ÂÂ drm/panfrost: Add support for a second regulator for the GPU
> > ÂÂ drm/panfrost: Add support for multiple power domain support
> > ÂÂ RFC: drm/panfrost: Add bifrost compatible string
> > ÂÂ RFC: drm/panfrost: devfreq: Add support for 2 regulators
> >
> > Â .../bindings/gpu/arm,mali-bifrost.yamlÂÂÂÂÂÂÂ |Â 20 ++++
> > Â arch/arm64/boot/dts/mediatek/mt8183-evb.dtsÂÂ |ÂÂ 7 ++
> > Â arch/arm64/boot/dts/mediatek/mt8183.dtsiÂÂÂÂÂ | 104 +++++++++++++++++
> > Â drivers/gpu/drm/panfrost/panfrost_devfreq.cÂÂ |Â 18 +++
> > Â drivers/gpu/drm/panfrost/panfrost_device.cÂÂÂ | 108 ++++++++++++++++--
> > Â drivers/gpu/drm/panfrost/panfrost_device.hÂÂÂ |ÂÂ 7 ++
> > Â drivers/gpu/drm/panfrost/panfrost_drv.cÂÂÂÂÂÂ |ÂÂ 1 +
> > Â drivers/gpu/drm/panfrost/panfrost_gpu.cÂÂÂÂÂÂ |Â 15 ++-
> > Â 8 files changed, 267 insertions(+), 13 deletions(-)
>
> _______________________________________________
> dri-devel mailing list
> dri-devel@xxxxxxxxxxxxxxxxxxxxx
> https://lists.freedesktop.org/mailman/listinfo/dri-devel