From: Felipe Balbi <[email protected]>
Hi folks,
I'm trying to enable the framebuffer on Microsoft Surface Duo. Looking
through some internal docs, it came to my attention that the bootloader
will fill up the framebuffer address and size to a memory node names
splash_region. Adding the node, I can see the address of the
framebuffer. Creating the relevant framebuffer device using
simple-framebuffer, I can't see it working. Tried dd if=/dev/urandom
of=/dev/fb0 and fb-test. None of which manage to get rid of what's
already on the screen, put there by the bootloader (platform Logo).
Wondering if any of you have seen a behavior such as this and how did
you manage to get framebuffer working on SM8150 (I see at least Sony
Xperia has the node).
Felipe Balbi (2):
arm64: boot: dts: qcom: sm8150: add a label for reserved-memory
arm64: boot: dts: qcom: surface duo: add minimal framebuffer
.../dts/qcom/sm8150-microsoft-surface-duo.dts | 19 +++++++++++++++++++
arch/arm64/boot/dts/qcom/sm8150.dtsi | 2 +-
2 files changed, 20 insertions(+), 1 deletion(-)
--
2.34.1
From: Felipe Balbi <[email protected]>
With this label, we can let boards append board-specific memory
regions which should not be mapped.
Signed-off-by: Felipe Balbi <[email protected]>
---
arch/arm64/boot/dts/qcom/sm8150.dtsi | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/arm64/boot/dts/qcom/sm8150.dtsi b/arch/arm64/boot/dts/qcom/sm8150.dtsi
index 15348512cf9e..26f9eb8e9b42 100644
--- a/arch/arm64/boot/dts/qcom/sm8150.dtsi
+++ b/arch/arm64/boot/dts/qcom/sm8150.dtsi
@@ -660,7 +660,7 @@ CLUSTER_PD: cpu-cluster0 {
};
};
- reserved-memory {
+ reserved_memory: reserved-memory {
#address-cells = <2>;
#size-cells = <2>;
ranges;
--
2.34.1
From: Felipe Balbi <[email protected]>
Add a minimal framebuffer device so we can use the display on the
Surface Duo device.
Signed-off-by: Felipe Balbi <[email protected]>
---
.../dts/qcom/sm8150-microsoft-surface-duo.dts | 19 +++++++++++++++++++
1 file changed, 19 insertions(+)
diff --git a/arch/arm64/boot/dts/qcom/sm8150-microsoft-surface-duo.dts b/arch/arm64/boot/dts/qcom/sm8150-microsoft-surface-duo.dts
index a73317e1a824..c629ec115fec 100644
--- a/arch/arm64/boot/dts/qcom/sm8150-microsoft-surface-duo.dts
+++ b/arch/arm64/boot/dts/qcom/sm8150-microsoft-surface-duo.dts
@@ -25,6 +25,17 @@ chosen {
stdout-path = "serial0:115200n8";
};
+ framebuffer0: framebuffer@9c000000 {
+ compatible = "simple-framebuffer";
+ reg = <0 0x9c000000 0 0x02400000>;
+ status = "okay";
+
+ height = <1800>;
+ width = <1350>;
+ stride = <(1350 * 4)>;
+ format = "a8r8g8b8";
+ };
+
vph_pwr: vph-pwr-regulator {
compatible = "regulator-fixed";
regulator-name = "vph_pwr";
@@ -472,6 +483,14 @@ &pon_resin {
linux,code = <KEY_VOLUMEDOWN>;
};
+&reserved_memory {
+ splash_region: splash_region@9c000000 {
+ /* We expect the bootloader to fill in the size */
+ reg = <0x0 0x9c000000 0x0 0x0>;
+ no-map;
+ };
+};
+
&tlmm {
gpio-reserved-ranges = <126 4>;
--
2.34.1
On 17.12.2021 13:57, Felipe Balbi wrote:
> From: Felipe Balbi <[email protected]>
>
> Hi folks,
>
> I'm trying to enable the framebuffer on Microsoft Surface Duo. Looking
> through some internal docs, it came to my attention that the bootloader
> will fill up the framebuffer address and size to a memory node names
> splash_region. Adding the node, I can see the address of the
> framebuffer. Creating the relevant framebuffer device using
> simple-framebuffer, I can't see it working. Tried dd if=/dev/urandom
> of=/dev/fb0 and fb-test. None of which manage to get rid of what's
> already on the screen, put there by the bootloader (platform Logo).
>
> Wondering if any of you have seen a behavior such as this and how did
> you manage to get framebuffer working on SM8150 (I see at least Sony
> Xperia has the node).
>
> Felipe Balbi (2):
> arm64: boot: dts: qcom: sm8150: add a label for reserved-memory
> arm64: boot: dts: qcom: surface duo: add minimal framebuffer
>
> .../dts/qcom/sm8150-microsoft-surface-duo.dts | 19 +++++++++++++++++++
> arch/arm64/boot/dts/qcom/sm8150.dtsi | 2 +-
> 2 files changed, 20 insertions(+), 1 deletion(-)
Hi,
this issue is totally unique to the Duo and your bootloader configuration.
Gus (CCd, co-author of Lumia 950/XL patches) and I were dissecting this precise
issue (albeit for a different usecase) and in our testing it turned out that XBL likely
kills the display stack upon exiting Boot Services and jumping to LinuxLoader. This may
be a bug that comes from the legacy of this device, as exiting Boot Services would
be rather undesirable in that scenario..
One fix would be to ask the bootloader team to look into it and fix it from there,
otherwise you'd have to bring up the display using the DPU1 driver, or perhaps in a
third-stage-bootloader (pls don't do it for the sanity of us all :D)
You can see the thing happening during the boot animation when the Microsoft
logo goes black for a split second and then it reappears with the "Powered by Android"
splash a short while after.
P.S I don't have this device and I'm only speaking on behalf of what we discovered
on Gus's one.
Konrad
On 17.12.2021 13:57, Felipe Balbi wrote:
> From: Felipe Balbi <[email protected]>
>
> Add a minimal framebuffer device so we can use the display on the
> Surface Duo device.
>
> Signed-off-by: Felipe Balbi <[email protected]>
> ---
> .../dts/qcom/sm8150-microsoft-surface-duo.dts | 19 +++++++++++++++++++
> 1 file changed, 19 insertions(+)
>
> diff --git a/arch/arm64/boot/dts/qcom/sm8150-microsoft-surface-duo.dts b/arch/arm64/boot/dts/qcom/sm8150-microsoft-surface-duo.dts
> index a73317e1a824..c629ec115fec 100644
> --- a/arch/arm64/boot/dts/qcom/sm8150-microsoft-surface-duo.dts
> +++ b/arch/arm64/boot/dts/qcom/sm8150-microsoft-surface-duo.dts
> @@ -25,6 +25,17 @@ chosen {
> stdout-path = "serial0:115200n8";
> };
>
> + framebuffer0: framebuffer@9c000000 {
> + compatible = "simple-framebuffer";
> + reg = <0 0x9c000000 0 0x02400000>;
You can remove the leading 0 in the size cell.
> + status = "okay";
This line is unnecessary, as it's enabled by default
> +
> + height = <1800>;
> + width = <1350>;
> + stride = <(1350 * 4)>;
> + format = "a8r8g8b8";
> + };
> +
> vph_pwr: vph-pwr-regulator {
> compatible = "regulator-fixed";
> regulator-name = "vph_pwr";
> @@ -472,6 +483,14 @@ &pon_resin {
> linux,code = <KEY_VOLUMEDOWN>;
> };
>
> +&reserved_memory {
> + splash_region: splash_region@9c000000 {
> + /* We expect the bootloader to fill in the size */
Would it be different than the framebuffer size?
> + reg = <0x0 0x9c000000 0x0 0x0>;
> + no-map;
> + };
> +};
> +
> &tlmm {
> gpio-reserved-ranges = <126 4>;
>
Konrad
Hi,
On Fri, 17 Dec 2021 at 15:58, Felipe Balbi <[email protected]> wrote:
>
> From: Felipe Balbi <[email protected]>
>
> Hi folks,
>
> I'm trying to enable the framebuffer on Microsoft Surface Duo. Looking
> through some internal docs, it came to my attention that the bootloader
> will fill up the framebuffer address and size to a memory node names
> splash_region. Adding the node, I can see the address of the
> framebuffer. Creating the relevant framebuffer device using
> simple-framebuffer, I can't see it working. Tried dd if=/dev/urandom
> of=/dev/fb0 and fb-test. None of which manage to get rid of what's
> already on the screen, put there by the bootloader (platform Logo).
>
> Wondering if any of you have seen a behavior such as this and how did
> you manage to get framebuffer working on SM8150 (I see at least Sony
> Xperia has the node).
What issues do you have with the DRM_MSM/DPU1 driver? I think it supports 8150.
>
> Felipe Balbi (2):
> arm64: boot: dts: qcom: sm8150: add a label for reserved-memory
> arm64: boot: dts: qcom: surface duo: add minimal framebuffer
>
> .../dts/qcom/sm8150-microsoft-surface-duo.dts | 19 +++++++++++++++++++
> arch/arm64/boot/dts/qcom/sm8150.dtsi | 2 +-
> 2 files changed, 20 insertions(+), 1 deletion(-)
>
> --
> 2.34.1
--
With best wishes
Dmitry
Hi,
Alexey Minnekhanov <[email protected]> writes:
> @@ -472,6 +483,14 @@ &pon_resin {
> linux,code = <KEY_VOLUMEDOWN>;
> };
>
> +&reserved_memory {
> + splash_region: splash_region@9c000000 {
> + /* We expect the bootloader to fill in the size */
> + reg = <0x0 0x9c000000 0x0 0x0>;
>
> 0x2400000 should be here as size, maybe?
bootloader is filling that up, but sure I only used this to get the size
and base address from bootloader. I could just hardcode the values now
that I know them :-)
>
> cont_splash_mem: memory@9c000000 {
> reg = <0x0 0x9c000000 0x0 0x2400000>;
> no-map;
> };
>
> FWIW, example above (with similar simple-framebuffer
> node) worked fine for my sm8150 based device.
Oh okay. So it's likely to be something specific to our bootloader.
--
balbi
Hi,
Dmitry Baryshkov <[email protected]> writes:
> Hi,
>
> On Fri, 17 Dec 2021 at 15:58, Felipe Balbi <[email protected]> wrote:
>>
>> From: Felipe Balbi <[email protected]>
>>
>> Hi folks,
>>
>> I'm trying to enable the framebuffer on Microsoft Surface Duo. Looking
>> through some internal docs, it came to my attention that the bootloader
>> will fill up the framebuffer address and size to a memory node names
>> splash_region. Adding the node, I can see the address of the
>> framebuffer. Creating the relevant framebuffer device using
>> simple-framebuffer, I can't see it working. Tried dd if=/dev/urandom
>> of=/dev/fb0 and fb-test. None of which manage to get rid of what's
>> already on the screen, put there by the bootloader (platform Logo).
>>
>> Wondering if any of you have seen a behavior such as this and how did
>> you manage to get framebuffer working on SM8150 (I see at least Sony
>> Xperia has the node).
>
> What issues do you have with the DRM_MSM/DPU1 driver? I think it supports 8150.
I haven't tried using it. I'll give it a shot next week.
--
balbi
Hi Konrad,
Konrad Dybcio <[email protected]> writes:
> On 17.12.2021 13:57, Felipe Balbi wrote:
>> From: Felipe Balbi <[email protected]>
>>
>> Hi folks,
>>
>> I'm trying to enable the framebuffer on Microsoft Surface Duo. Looking
>> through some internal docs, it came to my attention that the bootloader
>> will fill up the framebuffer address and size to a memory node names
>> splash_region. Adding the node, I can see the address of the
>> framebuffer. Creating the relevant framebuffer device using
>> simple-framebuffer, I can't see it working. Tried dd if=/dev/urandom
>> of=/dev/fb0 and fb-test. None of which manage to get rid of what's
>> already on the screen, put there by the bootloader (platform Logo).
>>
>> Wondering if any of you have seen a behavior such as this and how did
>> you manage to get framebuffer working on SM8150 (I see at least Sony
>> Xperia has the node).
>>
>> Felipe Balbi (2):
>> arm64: boot: dts: qcom: sm8150: add a label for reserved-memory
>> arm64: boot: dts: qcom: surface duo: add minimal framebuffer
>>
>> .../dts/qcom/sm8150-microsoft-surface-duo.dts | 19 +++++++++++++++++++
>> arch/arm64/boot/dts/qcom/sm8150.dtsi | 2 +-
>> 2 files changed, 20 insertions(+), 1 deletion(-)
>
> Hi,
>
>
> this issue is totally unique to the Duo and your bootloader configuration.
>
>
> Gus (CCd, co-author of Lumia 950/XL patches) and I were dissecting
> this precise issue (albeit for a different usecase) and in our testing
> it turned out that XBL likely kills the display stack upon exiting
> Boot Services and jumping to LinuxLoader. This may be a bug that comes
> from the legacy of this device, as exiting Boot Services would be
> rather undesirable in that scenario..
This is very nice background information which I didn't have. Thanks :-)
> One fix would be to ask the bootloader team to look into it and fix it
> from there, otherwise you'd have to bring up the display using the
> DPU1 driver, or perhaps in a third-stage-bootloader (pls don't do it
I'll give DPU1 a shot, thanks for the pointer
> for the sanity of us all :D)
no 3rd stages :-)
--
balbi
Hi,
Konrad Dybcio <[email protected]> writes:
> On 17.12.2021 13:57, Felipe Balbi wrote:
>> From: Felipe Balbi <[email protected]>
>>
>> Add a minimal framebuffer device so we can use the display on the
>> Surface Duo device.
>>
>> Signed-off-by: Felipe Balbi <[email protected]>
>> ---
>> .../dts/qcom/sm8150-microsoft-surface-duo.dts | 19 +++++++++++++++++++
>> 1 file changed, 19 insertions(+)
>>
>> diff --git a/arch/arm64/boot/dts/qcom/sm8150-microsoft-surface-duo.dts b/arch/arm64/boot/dts/qcom/sm8150-microsoft-surface-duo.dts
>> index a73317e1a824..c629ec115fec 100644
>> --- a/arch/arm64/boot/dts/qcom/sm8150-microsoft-surface-duo.dts
>> +++ b/arch/arm64/boot/dts/qcom/sm8150-microsoft-surface-duo.dts
>> @@ -25,6 +25,17 @@ chosen {
>> stdout-path = "serial0:115200n8";
>> };
>>
>> + framebuffer0: framebuffer@9c000000 {
>> + compatible = "simple-framebuffer";
>> + reg = <0 0x9c000000 0 0x02400000>;
>
> You can remove the leading 0 in the size cell.
okay
>> + status = "okay";
>
> This line is unnecessary, as it's enabled by default
understood
>> @@ -472,6 +483,14 @@ &pon_resin {
>> linux,code = <KEY_VOLUMEDOWN>;
>> };
>>
>> +&reserved_memory {
>> + splash_region: splash_region@9c000000 {
>> + /* We expect the bootloader to fill in the size */
>
> Would it be different than the framebuffer size?
honestly, I used this mainly to get the correct base address and size. I
had 0 everywhere and bootloader filled the entire thing for me. I can
just hardcode with the expected values anyway.
--
balbi
On 18.12.2021 08:17, Felipe Balbi wrote:
> Hi Konrad,
>
> Konrad Dybcio <[email protected]> writes:
>
>> On 17.12.2021 13:57, Felipe Balbi wrote:
>>> From: Felipe Balbi <[email protected]>
>>>
>>> Hi folks,
>>>
>>> I'm trying to enable the framebuffer on Microsoft Surface Duo. Looking
>>> through some internal docs, it came to my attention that the bootloader
>>> will fill up the framebuffer address and size to a memory node names
>>> splash_region. Adding the node, I can see the address of the
>>> framebuffer. Creating the relevant framebuffer device using
>>> simple-framebuffer, I can't see it working. Tried dd if=/dev/urandom
>>> of=/dev/fb0 and fb-test. None of which manage to get rid of what's
>>> already on the screen, put there by the bootloader (platform Logo).
>>>
>>> Wondering if any of you have seen a behavior such as this and how did
>>> you manage to get framebuffer working on SM8150 (I see at least Sony
>>> Xperia has the node).
>>>
>>> Felipe Balbi (2):
>>> arm64: boot: dts: qcom: sm8150: add a label for reserved-memory
>>> arm64: boot: dts: qcom: surface duo: add minimal framebuffer
>>>
>>> .../dts/qcom/sm8150-microsoft-surface-duo.dts | 19 +++++++++++++++++++
>>> arch/arm64/boot/dts/qcom/sm8150.dtsi | 2 +-
>>> 2 files changed, 20 insertions(+), 1 deletion(-)
>> Hi,
>>
>>
>> this issue is totally unique to the Duo and your bootloader configuration.
>>
>>
>> Gus (CCd, co-author of Lumia 950/XL patches) and I were dissecting
>> this precise issue (albeit for a different usecase) and in our testing
>> it turned out that XBL likely kills the display stack upon exiting
>> Boot Services and jumping to LinuxLoader. This may be a bug that comes
>> from the legacy of this device, as exiting Boot Services would be
>> rather undesirable in that scenario..
> This is very nice background information which I didn't have. Thanks :-)
>
>> One fix would be to ask the bootloader team to look into it and fix it
>> from there, otherwise you'd have to bring up the display using the
>> DPU1 driver, or perhaps in a third-stage-bootloader (pls don't do it
> I'll give DPU1 a shot, thanks for the pointer
Won't work yet. Your display (well, displays.. hehe) are CMD mode and
have DSC. Both of which are unsupported on SM8150. DSC patches by Vinod
seem to only work on 845 (or at least don't work on 8250 for me) and Marijn
from SoMainline (added to CC) is working on cleaning up CMD mode
support, as Qualcomm moved the pingpong functionality (basically
the way of SoC telling the panel HEY IT'S TIME TO DRAW A FRAME)
to a different hardware block and that needs some care in code).
>> for the sanity of us all :D)
> no 3rd stages :-)
>
Good :P
Konrad