[PATCH 0/2] qcom: dts: qrb4210-rb2 usb/emmc fixes

[PATCH 0/2] qcom: dts: qrb4210-rb2 usb/emmc fixes

[Caleb Connolly]

Add the missing iommus property on the first sdhc, and set a default
mode for USB role switch. The iommus property in particular is necessary
when running U-Boot as a first-stage bootloader, since it can't rely on
EDK2 to configure the SMMU and crashes with:

[020b9ed0d]TLB Dump Message:Not a valid memory address to dump TLBs for device APPS_TCU or no memory available.
[020b9f533]SMMU GLOBAL TCU NON-SEC FAULT: bit mask=0x00002000
[020b9fb23]SMMU:>> APPS_TCU NonSec Global Fault:
[020b9ff9f] NSGFSR=0x80000002
[020ba0404] NSGFAR=0x00000000bf6c6800
[020ba0909] NSGFSYNR0=0x00000000
[020ba0d1b] NSGFSYNR1=0x00c000c0
[020ba117d] NSGFSYNR2=0x00000000
[020ba15a9] NSCR0=0x00200406

The GFSYNR1 register in particular clues us in to the stream ID of the
peripherals, in this case 0x00c0 corresponds to the first SDHC.

---
Caleb Connolly (2):
      arm64: dts: qcom: sm6115: add iommu for sdhc_1
      arm64: dts: qcom: qrb4210-rb2: set role-switch-default-mode

 arch/arm64/boot/dts/qcom/qrb4210-rb2.dts | 4 ++++
 arch/arm64/boot/dts/qcom/sm6115.dtsi     | 1 +
 2 files changed, 5 insertions(+)
---
change-id: 20240619-rb2-fixes-94f24a0908d1
base-commit: 63072efe1495d9d42546b10b46f2c063ec8db3ce

// Caleb (they/them)

[PATCH 1/2] arm64: dts: qcom: sm6115: add iommu for sdhc_1

[Caleb Connolly]

The first SDHC can do DMA like most other peripherals, add the missing
iommus entry which is required to set this up.

This may have been working on Linux before since the bootloader
configures it and it may not be full torn down. But other software like
U-Boot needs this to initialize the eMMC properly.

Fixes: 97e563bf5ba1 ("arm64: dts: qcom: sm6115: Add basic soc dtsi")
Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
---
 arch/arm64/boot/dts/qcom/sm6115.dtsi | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/arm64/boot/dts/qcom/sm6115.dtsi b/arch/arm64/boot/dts/qcom/sm6115.dtsi
index aca0a87092e4..9ed062150aaf 100644
--- a/arch/arm64/boot/dts/qcom/sm6115.dtsi
+++ b/arch/arm64/boot/dts/qcom/sm6115.dtsi
@@ -1089,8 +1089,9 @@ sdhc_1: mmc@4744000 {
 			clock-names = "iface", "core", "xo", "ice";
 
 			power-domains = <&rpmpd SM6115_VDDCX>;
 			operating-points-v2 = <&sdhc1_opp_table>;
+			iommus = <&apps_smmu 0x00c0 0x0>;
 			interconnects = <&system_noc MASTER_SDCC_1 RPM_ALWAYS_TAG
 					 &bimc SLAVE_EBI_CH0 RPM_ALWAYS_TAG>,
 					<&bimc MASTER_AMPSS_M0 RPM_ALWAYS_TAG
 					 &config_noc SLAVE_SDCC_1 RPM_ALWAYS_TAG>;

-- 
2.45.0

[PATCH 2/2] arm64: dts: qcom: qrb4210-rb2: set role-switch-default-mode

[Caleb Connolly]

Give a hint to the OS which role we prefer. Host mode generally makes
the most sense.

Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
---
 arch/arm64/boot/dts/qcom/qrb4210-rb2.dts | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/arch/arm64/boot/dts/qcom/qrb4210-rb2.dts b/arch/arm64/boot/dts/qcom/qrb4210-rb2.dts
index 2c39bb1b97db..e9a63956b8b7 100644
--- a/arch/arm64/boot/dts/qcom/qrb4210-rb2.dts
+++ b/arch/arm64/boot/dts/qcom/qrb4210-rb2.dts
@@ -648,8 +648,12 @@ &uart4 {
 &usb {
 	status = "okay";
 };
 
+&usb_dwc3 {
+	role-switch-default-mode = "host";
+};
+
 &usb_dwc3_hs {
 	remote-endpoint = <&pmi632_hs_in>;
 };
 

-- 
2.45.0

Re: [PATCH 1/2] arm64: dts: qcom: sm6115: add iommu for sdhc_1

[Dmitry Baryshkov]

On Wed, 19 Jun 2024 at 23:33, Caleb Connolly <caleb.connolly@linaro.org> wrote:
>
> The first SDHC can do DMA like most other peripherals, add the missing
> iommus entry which is required to set this up.
>
> This may have been working on Linux before since the bootloader
> configures it and it may not be full torn down. But other software like
> U-Boot needs this to initialize the eMMC properly.
>
> Fixes: 97e563bf5ba1 ("arm64: dts: qcom: sm6115: Add basic soc dtsi")
> Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
> ---
>  arch/arm64/boot/dts/qcom/sm6115.dtsi | 1 +
>  1 file changed, 1 insertion(+)

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>

-- 
With best wishes
Dmitry

Re: [PATCH 2/2] arm64: dts: qcom: qrb4210-rb2: set role-switch-default-mode

[Dmitry Baryshkov]

On Wed, 19 Jun 2024 at 23:33, Caleb Connolly <caleb.connolly@linaro.org> wrote:
>
> Give a hint to the OS which role we prefer. Host mode generally makes
> the most sense.

Why?

>
> Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
> ---
>  arch/arm64/boot/dts/qcom/qrb4210-rb2.dts | 4 ++++
>  1 file changed, 4 insertions(+)

Would it make sense to set this for all the RB and HDK boards?

-- 
With best wishes
Dmitry

Re: [PATCH 2/2] arm64: dts: qcom: qrb4210-rb2: set role-switch-default-mode

[Caleb Connolly]

On 20/06/2024 15:15, Dmitry Baryshkov wrote:
> On Wed, 19 Jun 2024 at 23:33, Caleb Connolly <caleb.connolly@linaro.org> wrote:
>>
>> Give a hint to the OS which role we prefer. Host mode generally makes
>> the most sense.
> 
> Why?

I guess this is subjective, but on these boards the more common usecase 
is host mode (before we had role switching we forced them to host mode...).
> 
>>
>> Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
>> ---
>>   arch/arm64/boot/dts/qcom/qrb4210-rb2.dts | 4 ++++
>>   1 file changed, 4 insertions(+)
> 
> Would it make sense to set this for all the RB and HDK boards?

The rb1/2 are the only boards which lack multiple USB controllers. For 
others it's fine to leave the default (peripheral mode).
> 

-- 
// Caleb (they/them)

Re: [PATCH 2/2] arm64: dts: qcom: qrb4210-rb2: set role-switch-default-mode

[Dmitry Baryshkov]

On Thu, Jun 20, 2024 at 03:30:29PM GMT, Caleb Connolly wrote:
> 
> 
> On 20/06/2024 15:15, Dmitry Baryshkov wrote:
> > On Wed, 19 Jun 2024 at 23:33, Caleb Connolly <caleb.connolly@linaro.org> wrote:
> > > 
> > > Give a hint to the OS which role we prefer. Host mode generally makes
> > > the most sense.
> > 
> > Why?
> 
> I guess this is subjective, but on these boards the more common usecase is
> host mode (before we had role switching we forced them to host mode...).
> > 
> > > 
> > > Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
> > > ---
> > >   arch/arm64/boot/dts/qcom/qrb4210-rb2.dts | 4 ++++
> > >   1 file changed, 4 insertions(+)
> > 
> > Would it make sense to set this for all the RB and HDK boards?
> 
> The rb1/2 are the only boards which lack multiple USB controllers. For
> others it's fine to leave the default (peripheral mode).

SM8450-HDK and SM8650-HDK also have just a single USB-C port. My logic
was slightly different. We consider these devices to be SBCs, so I'd
expect that they act as hosts _by_default_. If somebody plugs RB board
into a laptop, then it's logical that it should work as a device, but
between the phone and the RB board the RB is a host.

-- 
With best wishes
Dmitry

Re: [PATCH 2/2] arm64: dts: qcom: qrb4210-rb2: set role-switch-default-mode

[Caleb Connolly]

On 20/06/2024 17:07, Dmitry Baryshkov wrote:
> On Thu, Jun 20, 2024 at 03:30:29PM GMT, Caleb Connolly wrote:
>>
>>
>> On 20/06/2024 15:15, Dmitry Baryshkov wrote:
>>> On Wed, 19 Jun 2024 at 23:33, Caleb Connolly <caleb.connolly@linaro.org> wrote:
>>>>
>>>> Give a hint to the OS which role we prefer. Host mode generally makes
>>>> the most sense.
>>>
>>> Why?
>>
>> I guess this is subjective, but on these boards the more common usecase is
>> host mode (before we had role switching we forced them to host mode...).
>>>
>>>>
>>>> Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
>>>> ---
>>>>    arch/arm64/boot/dts/qcom/qrb4210-rb2.dts | 4 ++++
>>>>    1 file changed, 4 insertions(+)
>>>
>>> Would it make sense to set this for all the RB and HDK boards?
>>
>> The rb1/2 are the only boards which lack multiple USB controllers. For
>> others it's fine to leave the default (peripheral mode).
> 
> SM8450-HDK and SM8650-HDK also have just a single USB-C port. My logic
> was slightly different. We consider these devices to be SBCs, so I'd
> expect that they act as hosts _by_default_. If somebody plugs RB board
> into a laptop, then it's logical that it should work as a device, but
> between the phone and the RB board the RB is a host.

Ahh I see, then yes perhaps it makes sense. I can send v2 with patches 
for other boards too.

* qrb2210-rb1
* qrb4210-rb2
* sm8450-hdk
* sm8650-hdk

Any others?
> 

-- 
// Caleb (they/them)

Re: [PATCH 2/2] arm64: dts: qcom: qrb4210-rb2: set role-switch-default-mode

[Dmitry Baryshkov]

On Thu, 20 Jun 2024 at 19:49, Caleb Connolly <caleb.connolly@linaro.org> wrote:
>
>
>
> On 20/06/2024 17:07, Dmitry Baryshkov wrote:
> > On Thu, Jun 20, 2024 at 03:30:29PM GMT, Caleb Connolly wrote:
> >>
> >>
> >> On 20/06/2024 15:15, Dmitry Baryshkov wrote:
> >>> On Wed, 19 Jun 2024 at 23:33, Caleb Connolly <caleb.connolly@linaro.org> wrote:
> >>>>
> >>>> Give a hint to the OS which role we prefer. Host mode generally makes
> >>>> the most sense.
> >>>
> >>> Why?
> >>
> >> I guess this is subjective, but on these boards the more common usecase is
> >> host mode (before we had role switching we forced them to host mode...).
> >>>
> >>>>
> >>>> Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
> >>>> ---
> >>>>    arch/arm64/boot/dts/qcom/qrb4210-rb2.dts | 4 ++++
> >>>>    1 file changed, 4 insertions(+)
> >>>
> >>> Would it make sense to set this for all the RB and HDK boards?
> >>
> >> The rb1/2 are the only boards which lack multiple USB controllers. For
> >> others it's fine to leave the default (peripheral mode).
> >
> > SM8450-HDK and SM8650-HDK also have just a single USB-C port. My logic
> > was slightly different. We consider these devices to be SBCs, so I'd
> > expect that they act as hosts _by_default_. If somebody plugs RB board
> > into a laptop, then it's logical that it should work as a device, but
> > between the phone and the RB board the RB is a host.
>
> Ahh I see, then yes perhaps it makes sense. I can send v2 with patches
> for other boards too.
>
> * qrb2210-rb1
> * qrb4210-rb2
> * sm8450-hdk
> * sm8650-hdk
>
> Any others?

I checked, sm8550-hdk also doesn't seem to have the USB-A port. I
think that's it for now.

-- 
With best wishes
Dmitry

Re: [PATCH 2/2] arm64: dts: qcom: qrb4210-rb2: set role-switch-default-mode

[Bjorn Andersson]

On Thu, Jun 20, 2024 at 06:49:46PM GMT, Caleb Connolly wrote:
> 
> 
> On 20/06/2024 17:07, Dmitry Baryshkov wrote:
> > On Thu, Jun 20, 2024 at 03:30:29PM GMT, Caleb Connolly wrote:
> > > 
> > > 
> > > On 20/06/2024 15:15, Dmitry Baryshkov wrote:
> > > > On Wed, 19 Jun 2024 at 23:33, Caleb Connolly <caleb.connolly@linaro.org> wrote:
> > > > > 
> > > > > Give a hint to the OS which role we prefer. Host mode generally makes
> > > > > the most sense.
> > > > 
> > > > Why?
> > > 
> > > I guess this is subjective, but on these boards the more common usecase is
> > > host mode (before we had role switching we forced them to host mode...).
> > > > 
> > > > > 
> > > > > Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
> > > > > ---
> > > > >    arch/arm64/boot/dts/qcom/qrb4210-rb2.dts | 4 ++++
> > > > >    1 file changed, 4 insertions(+)
> > > > 
> > > > Would it make sense to set this for all the RB and HDK boards?
> > > 
> > > The rb1/2 are the only boards which lack multiple USB controllers. For
> > > others it's fine to leave the default (peripheral mode).
> > 
> > SM8450-HDK and SM8650-HDK also have just a single USB-C port. My logic
> > was slightly different. We consider these devices to be SBCs, so I'd
> > expect that they act as hosts _by_default_. If somebody plugs RB board
> > into a laptop, then it's logical that it should work as a device, but
> > between the phone and the RB board the RB is a host.
> 
> Ahh I see, then yes perhaps it makes sense. I can send v2 with patches for
> other boards too.
> 
> * qrb2210-rb1
> * qrb4210-rb2
> * sm8450-hdk
> * sm8650-hdk
> 
> Any others?

qcs6490-rb3gen2 please.

I'm picking patch 1 for v6.10, no need to repost it.

Regards,
Bjorn

> > 
> 
> -- 
> // Caleb (they/them)

Re: (subset) [PATCH 0/2] qcom: dts: qrb4210-rb2 usb/emmc fixes

[Bjorn Andersson]

On Wed, 19 Jun 2024 22:33:37 +0200, Caleb Connolly wrote:
> Add the missing iommus property on the first sdhc, and set a default
> mode for USB role switch. The iommus property in particular is necessary
> when running U-Boot as a first-stage bootloader, since it can't rely on
> EDK2 to configure the SMMU and crashes with:
> 
> [020b9ed0d]TLB Dump Message:Not a valid memory address to dump TLBs for device APPS_TCU or no memory available.
> [020b9f533]SMMU GLOBAL TCU NON-SEC FAULT: bit mask=0x00002000
> [020b9fb23]SMMU:>> APPS_TCU NonSec Global Fault:
> [020b9ff9f] NSGFSR=0x80000002
> [020ba0404] NSGFAR=0x00000000bf6c6800
> [020ba0909] NSGFSYNR0=0x00000000
> [020ba0d1b] NSGFSYNR1=0x00c000c0
> [020ba117d] NSGFSYNR2=0x00000000
> [020ba15a9] NSCR0=0x00200406
> 
> [...]

Applied, thanks!

[1/2] arm64: dts: qcom: sm6115: add iommu for sdhc_1
      commit: 94ea124aeefe1ef271263f176634034e22c49311

Best regards,
-- 
Bjorn Andersson <andersson@kernel.org>

Re: [PATCH 2/2] arm64: dts: qcom: qrb4210-rb2: set role-switch-default-mode

[Dmitry Baryshkov]

On Fri, 21 Jun 2024 at 08:32, Bjorn Andersson <andersson@kernel.org> wrote:
>
> On Thu, Jun 20, 2024 at 06:49:46PM GMT, Caleb Connolly wrote:
> >
> >
> > On 20/06/2024 17:07, Dmitry Baryshkov wrote:
> > > On Thu, Jun 20, 2024 at 03:30:29PM GMT, Caleb Connolly wrote:
> > > >
> > > >
> > > > On 20/06/2024 15:15, Dmitry Baryshkov wrote:
> > > > > On Wed, 19 Jun 2024 at 23:33, Caleb Connolly <caleb.connolly@linaro.org> wrote:
> > > > > >
> > > > > > Give a hint to the OS which role we prefer. Host mode generally makes
> > > > > > the most sense.
> > > > >
> > > > > Why?
> > > >
> > > > I guess this is subjective, but on these boards the more common usecase is
> > > > host mode (before we had role switching we forced them to host mode...).
> > > > >
> > > > > >
> > > > > > Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
> > > > > > ---
> > > > > >    arch/arm64/boot/dts/qcom/qrb4210-rb2.dts | 4 ++++
> > > > > >    1 file changed, 4 insertions(+)
> > > > >
> > > > > Would it make sense to set this for all the RB and HDK boards?
> > > >
> > > > The rb1/2 are the only boards which lack multiple USB controllers. For
> > > > others it's fine to leave the default (peripheral mode).
> > >
> > > SM8450-HDK and SM8650-HDK also have just a single USB-C port. My logic
> > > was slightly different. We consider these devices to be SBCs, so I'd
> > > expect that they act as hosts _by_default_. If somebody plugs RB board
> > > into a laptop, then it's logical that it should work as a device, but
> > > between the phone and the RB board the RB is a host.
> >
> > Ahh I see, then yes perhaps it makes sense. I can send v2 with patches for
> > other boards too.
> >
> > * qrb2210-rb1
> > * qrb4210-rb2
> > * sm8450-hdk
> > * sm8650-hdk
> >
> > Any others?
>
> qcs6490-rb3gen2 please.

Seeing the list grow on and on. What about making a policy decision
that all non-phone-factor ARCH_QCOM devices should use the host role
by default? So HDK, IDP, RBn, DragonBoard, laptops => host. QRD,
phones => peripheral. Not sure about the tablets. Probably peripheral
too (unless it's a convertible / 2-in-1).

>
> I'm picking patch 1 for v6.10, no need to repost it.
>
> Regards,
> Bjorn
>
> > >
> >
> > --
> > // Caleb (they/them)



-- 
With best wishes
Dmitry