| Message ID | 20210921071807.30978-2-narmstrong@baylibre.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | ARM: dts: dra7: enable BB2D module | expand |
* Neil Armstrong <narmstrong@baylibre.com> [210921 10:18]: > From: Gowtham Tammana <g-tammana@ti.com> > > BB2D is a Vivante GC 2D Accelerator. > This adds the node to the dts file within a target module node. > Crossbar index number is used for interrupt mapping. Thanks applying into omap-for-v5.16/dt. Tony
Hi, Neil, Tony
# sorry for the confusion if you have received the mail which I sent
before, which is not in plain text mode.
We have one out of tree change to enable the SGX544 gpu for the
beagleboard-X15 Android build,
https://android-review.linaro.org/c/kernel/common/+/20521/11/arch/arm/boot/dts/dra7.dtsi
and that seems to conflict with this BB2D enabling change,
Could you please help give some suggestions on how we should update our patch
to make it work with BB2D, without the revert of this change?
On Wed, 6 Oct 2021 at 15:47, Tony Lindgren <tony@atomide.com> wrote:
>
> * Neil Armstrong <narmstrong@baylibre.com> [210921 10:18]:
> > From: Gowtham Tammana <g-tammana@ti.com>
> >
> > BB2D is a Vivante GC 2D Accelerator.
> > This adds the node to the dts file within a target module node.
> > Crossbar index number is used for interrupt mapping.
>
> Thanks applying into omap-for-v5.16/dt.
>
> Tony
Hi, On 23/11/2021 13:17, Yongqin Liu wrote: > Hi, Neil, Tony > > # sorry for the confusion if you have received the mail which I sent > before, which is not in plain text mode. > > We have one out of tree change to enable the SGX544 gpu for the > beagleboard-X15 Android build, > https://android-review.linaro.org/c/kernel/common/+/20521/11/arch/arm/boot/dts/dra7.dtsi > > and that seems to conflict with this BB2D enabling change, > Could you please help give some suggestions on how we should update our patch > to make it work with BB2D, without the revert of this change? This BB2D patch alters the target-module@59000000 while your SGX change alters the target-module@56000000. Please rebase your patches. Neil > > On Wed, 6 Oct 2021 at 15:47, Tony Lindgren <tony@atomide.com> wrote: >> >> * Neil Armstrong <narmstrong@baylibre.com> [210921 10:18]: >>> From: Gowtham Tammana <g-tammana@ti.com> >>> >>> BB2D is a Vivante GC 2D Accelerator. >>> This adds the node to the dts file within a target module node. >>> Crossbar index number is used for interrupt mapping. >> >> Thanks applying into omap-for-v5.16/dt. >> >> Tony >
Hi, Neil On Tue, 23 Nov 2021 at 20:47, Neil Armstrong <narmstrong@baylibre.com> wrote: > > Hi, > > On 23/11/2021 13:17, Yongqin Liu wrote: > > Hi, Neil, Tony > > > > # sorry for the confusion if you have received the mail which I sent > > before, which is not in plain text mode. > > > > We have one out of tree change to enable the SGX544 gpu for the > > beagleboard-X15 Android build, > > https://android-review.linaro.org/c/kernel/common/+/20521/11/arch/arm/boot/dts/dra7.dtsi > > > > and that seems to conflict with this BB2D enabling change, > > Could you please help give some suggestions on how we should update our patch > > to make it work with BB2D, without the revert of this change? > > This BB2D patch alters the target-module@59000000 while your SGX > change alters the target-module@56000000. > > Please rebase your patches. I am sorry if the "conflict" I used previously caused the confusion. What I meant with the "conflict" word is the feature conflict, not the patch merge conflict. I could merge my SGX change with the BB2D change there, but then my build could not boot successfully to the homescreen, I need to revert the BB2D change to have it boot to the homescreen successfully. Here are the serial console output in case you want to check: https://pastebin.com/RY472b96 work with the BB2D change reverted https://pastebin.com/aP97r7rJ does not work with the BB2D change. Not sure if you have any idea about the problem. Please help give some suggestions on what I could try. Thanks, Yongqin > > > > > On Wed, 6 Oct 2021 at 15:47, Tony Lindgren <tony@atomide.com> wrote: > >> > >> * Neil Armstrong <narmstrong@baylibre.com> [210921 10:18]: > >>> From: Gowtham Tammana <g-tammana@ti.com> > >>> > >>> BB2D is a Vivante GC 2D Accelerator. > >>> This adds the node to the dts file within a target module node. > >>> Crossbar index number is used for interrupt mapping. > >> > >> Thanks applying into omap-for-v5.16/dt. > >> > >> Tony > > >
* Yongqin Liu <yongqin.liu@linaro.org> [211123 17:27]: > Hi, Neil > > On Tue, 23 Nov 2021 at 20:47, Neil Armstrong <narmstrong@baylibre.com> wrote: > > > > Hi, > > > > On 23/11/2021 13:17, Yongqin Liu wrote: > > > Hi, Neil, Tony > > > > > > # sorry for the confusion if you have received the mail which I sent > > > before, which is not in plain text mode. > > > > > > We have one out of tree change to enable the SGX544 gpu for the > > > beagleboard-X15 Android build, > > > https://android-review.linaro.org/c/kernel/common/+/20521/11/arch/arm/boot/dts/dra7.dtsi > > > > > > and that seems to conflict with this BB2D enabling change, > > > Could you please help give some suggestions on how we should update our patch > > > to make it work with BB2D, without the revert of this change? > > > > This BB2D patch alters the target-module@59000000 while your SGX > > change alters the target-module@56000000. > > > > Please rebase your patches. > I am sorry if the "conflict" I used previously caused the confusion. > What I meant with the "conflict" word is the feature conflict, not the > patch merge conflict. > > I could merge my SGX change with the BB2D change there, but then my > build could not boot successfully to the homescreen, > I need to revert the BB2D change to have it boot to the homescreen successfully. > > Here are the serial console output in case you want to check: > https://pastebin.com/RY472b96 work with the BB2D change reverted > https://pastebin.com/aP97r7rJ does not work with the BB2D change. > > Not sure if you have any idea about the problem. > Please help give some suggestions on what I could try. Sounds like your out of tree SGX patches also tinker with the BB2D module. To me it sounds like you can just tag the new target-module@59000000 added by this patch with status = "disabled" in your SGX patch until you have updated the driver code. I'm adding this patch into omap-for-v5.17/dt as for 2D acceleration there is the etnaviv driver in the mainline kernel that should be usable to some extent. Regards, Tony
* Tony Lindgren <tony@atomide.com> [211209 07:08]: > Sounds like your out of tree SGX patches also tinker with the BB2D module. > > To me it sounds like you can just tag the new target-module@59000000 added > by this patch with status = "disabled" in your SGX patch until you have > updated the driver code. The above still applies.. > I'm adding this patch into omap-for-v5.17/dt as for 2D acceleration there > is the etnaviv driver in the mainline kernel that should be usable to > some extent. ..but looks like we already merged this patch earlier so no need to reapply :) Sorry I still had the patch tagged in my inbox. Regards, Tony
diff --git a/arch/arm/boot/dts/dra7.dtsi b/arch/arm/boot/dts/dra7.dtsi index dfc1ef8ef6ae..6b485cbed8d5 100644 --- a/arch/arm/boot/dts/dra7.dtsi +++ b/arch/arm/boot/dts/dra7.dtsi @@ -965,6 +965,25 @@ hdmi: encoder@0 { }; }; + target-module@59000000 { + compatible = "ti,sysc-omap4", "ti,sysc"; + reg = <0x59000020 0x4>; + reg-names = "rev"; + clocks = <&dss_clkctrl DRA7_DSS_BB2D_CLKCTRL 0>; + clock-names = "fck"; + #address-cells = <1>; + #size-cells = <1>; + ranges = <0x0 0x59000000 0x1000>; + + bb2d: gpu@0 { + compatible = "vivante,gc"; + reg = <0x0 0x700>; + interrupts = <GIC_SPI 120 IRQ_TYPE_LEVEL_HIGH>; + clocks = <&dss_clkctrl DRA7_BB2D_CLKCTRL 0>; + clock-names = "core"; + }; + }; + aes1_target: target-module@4b500000 { compatible = "ti,sysc-omap2", "ti,sysc"; reg = <0x4b500080 0x4>,