[PATCH v3 01/29] armv8: ls1028a: add IOMMU stream ID to vivante node

[Vladimir Oltean]

On Thu, Sep 02, 2021 at 06:45:30PM +0200, Michael Walle wrote:
> The fixup is done for the "fsl,ls1028a-gpu" which isn't any official
> device tree binding. Don't break it, but instead add a fixup for another
> compatible "vivante,gc" which is the offical one for the GPU on the
> LS1028A.
> 
> Signed-off-by: Michael Walle <michael at walle.cc>
> ---
>  arch/arm/cpu/armv8/fsl-layerscape/ls1028_ids.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/arch/arm/cpu/armv8/fsl-layerscape/ls1028_ids.c b/arch/arm/cpu/armv8/fsl-layerscape/ls1028_ids.c
> index 49df8b3790..d93a793f39 100644
> --- a/arch/arm/cpu/armv8/fsl-layerscape/ls1028_ids.c
> +++ b/arch/arm/cpu/armv8/fsl-layerscape/ls1028_ids.c
> @@ -18,6 +18,7 @@ struct icid_id_table icid_tbl[] = {
>  	SET_SATA_ICID(1, "fsl,ls1028a-ahci", FSL_SATA1_STREAM_ID),
>  	SET_EDMA_ICID(FSL_EDMA_STREAM_ID),
>  	SET_QDMA_ICID("fsl,ls1028a-qdma", FSL_DMA_STREAM_ID),
> +	SET_GPU_ICID("vivante,gc", FSL_GPU_STREAM_ID),
>  	SET_GPU_ICID("fsl,ls1028a-gpu", FSL_GPU_STREAM_ID),
>  	SET_DISPLAY_ICID(FSL_DISPLAY_STREAM_ID),
>  #ifdef CONFIG_FSL_CAAM
> -- 
> 2.30.2
> 

So I have zero tangency with the LS1028A GPU (I've just ordered a
DisplayPort to HDMI adapter so I can test it from now on with my HDMI
capture card), but I think that you know that the "fsl,ls1028a-gpu"
compatible string is for this driver:
https://source.codeaurora.org/external/qoriq/qoriq-components/linux/tree/drivers/mxc/gpu-viv/hal/os/linux/kernel/platform/freescale/gc_hal_kernel_platform_imx.c?h=LSDK-21.08
which yes, has no upstream equivalent, and will probably never be upstream.

And by the way, every time I boot an upstream kernel with upstream
U-Boot, I get this warning due to the attempted fixup, which is really annoying:

WARNING could not find node fsl,ls1028a-gpu: FDT_ERR_NOTFOUND.

So I personally would not mind if we would just delete anything having
to do with "fsl,ls1028a-gpu" from mainline U-Boot (at least until
something meaningful gets added to Linux which is specific to the
LS1028A) :)

Let me add Andy Tang to the discussion just to make sure that he's aware
of what you're up to:
https://www.spinics.net/lists/kernel/msg4060045.html