When booting with Device Tree, the current default boot configuration
table option, the request to boot via 'host mode' comes from the
"dr_mode" property. A property of the same name can be used inside
ACPI tables too. However it is missing from the SDM845's ACPI tables
so we have to supply this information using Platform Device Properites
instead.
Signed-off-by: Lee Jones <[email protected]>
---
drivers/usb/dwc3/dwc3-qcom.c | 12 ++++++++++++
1 file changed, 12 insertions(+)
diff --git a/drivers/usb/dwc3/dwc3-qcom.c b/drivers/usb/dwc3/dwc3-qcom.c
index 349bf549ee44..f21fdd6cdd1a 100644
--- a/drivers/usb/dwc3/dwc3-qcom.c
+++ b/drivers/usb/dwc3/dwc3-qcom.c
@@ -468,6 +468,11 @@ static const struct acpi_device_id dwc3_qcom_acpi_match[] = {
};
MODULE_DEVICE_TABLE(acpi, dwc3_qcom_acpi_match);
+static const struct property_entry dwc3_qcom_acpi_properties[] = {
+ PROPERTY_ENTRY_STRING("dr_mode", "host"),
+ {}
+};
+
static int dwc3_qcom_probe(struct platform_device *pdev)
{
struct device_node *np = pdev->dev.of_node, *dwc3_np;
@@ -603,6 +608,13 @@ static int dwc3_qcom_probe(struct platform_device *pdev)
goto platform_unalloc;
}
+ ret = platform_device_add_properties(qcom->dwc3,
+ dwc3_qcom_acpi_properties);
+ if (ret < 0) {
+ dev_err(&pdev->dev, "failed to add properties\n");
+ goto platform_unalloc;
+ }
+
ret = platform_device_add(qcom->dwc3);
if (ret) {
dev_err(&pdev->dev, "failed to add device\n");
--
2.17.1
On Tue 04 Jun 03:44 PDT 2019, Lee Jones wrote:
> When booting with Device Tree, the current default boot configuration
> table option, the request to boot via 'host mode' comes from the
> "dr_mode" property.
This has been the default on the MTP, but this is changing as this is
causing issues when connected downstream from a hub (the typical
development case for the primary USB port of a phone like device) and
more importantly we don't have support for the PMIC blocks that control
VBUS.
Once these issues are resolved the dr_mode would be "otg".
> A property of the same name can be used inside
> ACPI tables too. However it is missing from the SDM845's ACPI tables
> so we have to supply this information using Platform Device Properites
> instead.
>
Afaict this would install a fall-back property, so in the case that we
have specified dr_mode in DT (or ACPI) that would take precedence. So
the commit message should reflect that this redefines the default choice
to be "host", rather than "otg".
Which is in conflict with what's described for dr_mode in
Documentation/devicetree/bindings/usb/generic.txt
And this driver is used on a range of different Qualcomm platforms, so I
don't think this is SDM845 specific.
Regards,
Bjorn
> Signed-off-by: Lee Jones <[email protected]>
> ---
> drivers/usb/dwc3/dwc3-qcom.c | 12 ++++++++++++
> 1 file changed, 12 insertions(+)
>
> diff --git a/drivers/usb/dwc3/dwc3-qcom.c b/drivers/usb/dwc3/dwc3-qcom.c
> index 349bf549ee44..f21fdd6cdd1a 100644
> --- a/drivers/usb/dwc3/dwc3-qcom.c
> +++ b/drivers/usb/dwc3/dwc3-qcom.c
> @@ -468,6 +468,11 @@ static const struct acpi_device_id dwc3_qcom_acpi_match[] = {
> };
> MODULE_DEVICE_TABLE(acpi, dwc3_qcom_acpi_match);
>
> +static const struct property_entry dwc3_qcom_acpi_properties[] = {
> + PROPERTY_ENTRY_STRING("dr_mode", "host"),
> + {}
> +};
> +
> static int dwc3_qcom_probe(struct platform_device *pdev)
> {
> struct device_node *np = pdev->dev.of_node, *dwc3_np;
> @@ -603,6 +608,13 @@ static int dwc3_qcom_probe(struct platform_device *pdev)
> goto platform_unalloc;
> }
>
> + ret = platform_device_add_properties(qcom->dwc3,
> + dwc3_qcom_acpi_properties);
> + if (ret < 0) {
> + dev_err(&pdev->dev, "failed to add properties\n");
> + goto platform_unalloc;
> + }
> +
> ret = platform_device_add(qcom->dwc3);
> if (ret) {
> dev_err(&pdev->dev, "failed to add device\n");
> --
> 2.17.1
>
On Wed, 05 Jun 2019, Bjorn Andersson wrote:
> On Tue 04 Jun 03:44 PDT 2019, Lee Jones wrote:
>
> > When booting with Device Tree, the current default boot configuration
> > table option, the request to boot via 'host mode' comes from the
> > "dr_mode" property.
>
> This has been the default on the MTP, but this is changing as this is
> causing issues when connected downstream from a hub (the typical
> development case for the primary USB port of a phone like device) and
> more importantly we don't have support for the PMIC blocks that control
> VBUS.
My point is not about which mode is currently chosen. It's more about
the capability of choosing which mode is appropriate for a given
system via DT.
> Once these issues are resolved the dr_mode would be "otg".
OTG doesn't work on this H/W, so we need to specify "host" mode.
> > A property of the same name can be used inside
> > ACPI tables too. However it is missing from the SDM845's ACPI tables
> > so we have to supply this information using Platform Device Properites
> > instead.
> >
>
> Afaict this would install a fall-back property, so in the case that we
> have specified dr_mode in DT (or ACPI) that would take precedence. So
That's correct.
> the commit message should reflect that this redefines the default choice
> to be "host", rather than "otg".
No problem.
> Which is in conflict with what's described for dr_mode in
> Documentation/devicetree/bindings/usb/generic.txt
This implementation only affects ACPI based platforms. When booting
with DT, the description in that DT related document is still
accurate.
> And this driver is used on a range of different Qualcomm platforms, so I
> don't think this is SDM845 specific.
ACPI based platforms?
All the ones I've seen use the XHCI USB driver directly ("PNP0D10").
> > Signed-off-by: Lee Jones <[email protected]>
> > ---
> > drivers/usb/dwc3/dwc3-qcom.c | 12 ++++++++++++
> > 1 file changed, 12 insertions(+)
> >
> > diff --git a/drivers/usb/dwc3/dwc3-qcom.c b/drivers/usb/dwc3/dwc3-qcom.c
> > index 349bf549ee44..f21fdd6cdd1a 100644
> > --- a/drivers/usb/dwc3/dwc3-qcom.c
> > +++ b/drivers/usb/dwc3/dwc3-qcom.c
> > @@ -468,6 +468,11 @@ static const struct acpi_device_id dwc3_qcom_acpi_match[] = {
> > };
> > MODULE_DEVICE_TABLE(acpi, dwc3_qcom_acpi_match);
> >
> > +static const struct property_entry dwc3_qcom_acpi_properties[] = {
> > + PROPERTY_ENTRY_STRING("dr_mode", "host"),
> > + {}
> > +};
> > +
> > static int dwc3_qcom_probe(struct platform_device *pdev)
> > {
> > struct device_node *np = pdev->dev.of_node, *dwc3_np;
> > @@ -603,6 +608,13 @@ static int dwc3_qcom_probe(struct platform_device *pdev)
> > goto platform_unalloc;
> > }
> >
> > + ret = platform_device_add_properties(qcom->dwc3,
> > + dwc3_qcom_acpi_properties);
> > + if (ret < 0) {
> > + dev_err(&pdev->dev, "failed to add properties\n");
> > + goto platform_unalloc;
> > + }
> > +
> > ret = platform_device_add(qcom->dwc3);
> > if (ret) {
> > dev_err(&pdev->dev, "failed to add device\n");
--
Lee Jones [李琼斯]
Linaro Services Technical Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog
On 6/5/2019 2:34 AM, Lee Jones wrote:
> On Wed, 05 Jun 2019, Bjorn Andersson wrote:
>
>> On Tue 04 Jun 03:44 PDT 2019, Lee Jones wrote:
>>
>>> When booting with Device Tree, the current default boot configuration
>>> table option, the request to boot via 'host mode' comes from the
>>> "dr_mode" property.
>>
>> This has been the default on the MTP, but this is changing as this is
>> causing issues when connected downstream from a hub (the typical
>> development case for the primary USB port of a phone like device) and
>> more importantly we don't have support for the PMIC blocks that control
>> VBUS.
>
> My point is not about which mode is currently chosen. It's more about
> the capability of choosing which mode is appropriate for a given
> system via DT.
>
>> Once these issues are resolved the dr_mode would be "otg".
>
> OTG doesn't work on this H/W, so we need to specify "host" mode.
How have you made that determination?
>
>>> A property of the same name can be used inside
>>> ACPI tables too. However it is missing from the SDM845's ACPI tables
>>> so we have to supply this information using Platform Device Properites
>>> instead.
>>>
>>
>> Afaict this would install a fall-back property, so in the case that we
>> have specified dr_mode in DT (or ACPI) that would take precedence. So
>
> That's correct.
>
>> the commit message should reflect that this redefines the default choice
>> to be "host", rather than "otg".
>
> No problem. >
>> Which is in conflict with what's described for dr_mode in
>> Documentation/devicetree/bindings/usb/generic.txt
>
> This implementation only affects ACPI based platforms. When booting
> with DT, the description in that DT related document is still
> accurate.
>
>> And this driver is used on a range of different Qualcomm platforms, so I
>> don't think this is SDM845 specific.
>
> ACPI based platforms?
>
> All the ones I've seen use the XHCI USB driver directly ("PNP0D10").
>
>>> Signed-off-by: Lee Jones <[email protected]>
>>> ---
>>> drivers/usb/dwc3/dwc3-qcom.c | 12 ++++++++++++
>>> 1 file changed, 12 insertions(+)
>>>
>>> diff --git a/drivers/usb/dwc3/dwc3-qcom.c b/drivers/usb/dwc3/dwc3-qcom.c
>>> index 349bf549ee44..f21fdd6cdd1a 100644
>>> --- a/drivers/usb/dwc3/dwc3-qcom.c
>>> +++ b/drivers/usb/dwc3/dwc3-qcom.c
>>> @@ -468,6 +468,11 @@ static const struct acpi_device_id dwc3_qcom_acpi_match[] = {
>>> };
>>> MODULE_DEVICE_TABLE(acpi, dwc3_qcom_acpi_match);
>>>
>>> +static const struct property_entry dwc3_qcom_acpi_properties[] = {
>>> + PROPERTY_ENTRY_STRING("dr_mode", "host"),
>>> + {}
>>> +};
>>> +
>>> static int dwc3_qcom_probe(struct platform_device *pdev)
>>> {
>>> struct device_node *np = pdev->dev.of_node, *dwc3_np;
>>> @@ -603,6 +608,13 @@ static int dwc3_qcom_probe(struct platform_device *pdev)
>>> goto platform_unalloc;
>>> }
>>>
>>> + ret = platform_device_add_properties(qcom->dwc3,
>>> + dwc3_qcom_acpi_properties);
>>> + if (ret < 0) {
>>> + dev_err(&pdev->dev, "failed to add properties\n");
>>> + goto platform_unalloc;
>>> + }
>>> +
>>> ret = platform_device_add(qcom->dwc3);
>>> if (ret) {
>>> dev_err(&pdev->dev, "failed to add device\n");
>
--
Jeffrey Hugo
Qualcomm Datacenter Technologies as an affiliate of Qualcomm
Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.
On Wed, 05 Jun 2019, Jeffrey Hugo wrote:
> On 6/5/2019 2:34 AM, Lee Jones wrote:
> > On Wed, 05 Jun 2019, Bjorn Andersson wrote:
> >
> > > On Tue 04 Jun 03:44 PDT 2019, Lee Jones wrote:
> > >
> > > > When booting with Device Tree, the current default boot configuration
> > > > table option, the request to boot via 'host mode' comes from the
> > > > "dr_mode" property.
> > >
> > > This has been the default on the MTP, but this is changing as this is
> > > causing issues when connected downstream from a hub (the typical
> > > development case for the primary USB port of a phone like device) and
> > > more importantly we don't have support for the PMIC blocks that control
> > > VBUS.
> >
> > My point is not about which mode is currently chosen. It's more about
> > the capability of choosing which mode is appropriate for a given
> > system via DT.
> >
> > > Once these issues are resolved the dr_mode would be "otg".
> >
> > OTG doesn't work on this H/W, so we need to specify "host" mode.
>
> How have you made that determination?
I enabled OTG and plugged in all of the things.
--
Lee Jones [李琼斯]
Linaro Services Technical Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog
On Wed 05 Jun 01:34 PDT 2019, Lee Jones wrote:
> On Wed, 05 Jun 2019, Bjorn Andersson wrote:
>
> > On Tue 04 Jun 03:44 PDT 2019, Lee Jones wrote:
> >
> > > When booting with Device Tree, the current default boot configuration
> > > table option, the request to boot via 'host mode' comes from the
> > > "dr_mode" property.
> >
> > This has been the default on the MTP, but this is changing as this is
> > causing issues when connected downstream from a hub (the typical
> > development case for the primary USB port of a phone like device) and
> > more importantly we don't have support for the PMIC blocks that control
> > VBUS.
>
> My point is not about which mode is currently chosen. It's more about
> the capability of choosing which mode is appropriate for a given
> system via DT.
>
> > Once these issues are resolved the dr_mode would be "otg".
>
> OTG doesn't work on this H/W, so we need to specify "host" mode.
>
My objection is that when you say "this H/W" you mean a particular
product, but you're making this decision for all SDM845 based products
using ACPI.
I don't know if there is a Windows phone based on SDM845, but if there
is then I don't think forcing it to host would be correct.
> > > A property of the same name can be used inside
> > > ACPI tables too. However it is missing from the SDM845's ACPI tables
> > > so we have to supply this information using Platform Device Properites
> > > instead.
> > >
> >
> > Afaict this would install a fall-back property, so in the case that we
> > have specified dr_mode in DT (or ACPI) that would take precedence. So
>
> That's correct.
>
> > the commit message should reflect that this redefines the default choice
> > to be "host", rather than "otg".
>
> No problem.
>
> > Which is in conflict with what's described for dr_mode in
> > Documentation/devicetree/bindings/usb/generic.txt
>
> This implementation only affects ACPI based platforms. When booting
> with DT, the description in that DT related document is still
> accurate.
>
You're right, I got lost between the patches and the sprinkled if
(ACPI_HANDLE()) in the probe. This is only added for ACPI.
> > And this driver is used on a range of different Qualcomm platforms, so I
> > don't think this is SDM845 specific.
>
> ACPI based platforms?
>
> All the ones I've seen use the XHCI USB driver directly ("PNP0D10").
>
MSM8998 (835) has the same controller, so this should affect those
laptops as well.
Regards,
Bjorn
> > > Signed-off-by: Lee Jones <[email protected]>
> > > ---
> > > drivers/usb/dwc3/dwc3-qcom.c | 12 ++++++++++++
> > > 1 file changed, 12 insertions(+)
> > >
> > > diff --git a/drivers/usb/dwc3/dwc3-qcom.c b/drivers/usb/dwc3/dwc3-qcom.c
> > > index 349bf549ee44..f21fdd6cdd1a 100644
> > > --- a/drivers/usb/dwc3/dwc3-qcom.c
> > > +++ b/drivers/usb/dwc3/dwc3-qcom.c
> > > @@ -468,6 +468,11 @@ static const struct acpi_device_id dwc3_qcom_acpi_match[] = {
> > > };
> > > MODULE_DEVICE_TABLE(acpi, dwc3_qcom_acpi_match);
> > >
> > > +static const struct property_entry dwc3_qcom_acpi_properties[] = {
> > > + PROPERTY_ENTRY_STRING("dr_mode", "host"),
> > > + {}
> > > +};
> > > +
> > > static int dwc3_qcom_probe(struct platform_device *pdev)
> > > {
> > > struct device_node *np = pdev->dev.of_node, *dwc3_np;
> > > @@ -603,6 +608,13 @@ static int dwc3_qcom_probe(struct platform_device *pdev)
> > > goto platform_unalloc;
> > > }
> > >
> > > + ret = platform_device_add_properties(qcom->dwc3,
> > > + dwc3_qcom_acpi_properties);
> > > + if (ret < 0) {
> > > + dev_err(&pdev->dev, "failed to add properties\n");
> > > + goto platform_unalloc;
> > > + }
> > > +
> > > ret = platform_device_add(qcom->dwc3);
> > > if (ret) {
> > > dev_err(&pdev->dev, "failed to add device\n");
>
> --
> Lee Jones [?????????]
> Linaro Services Technical Lead
> Linaro.org ??? Open source software for ARM SoCs
> Follow Linaro: Facebook | Twitter | Blog
On Wed, 05 Jun 2019, Bjorn Andersson wrote:
> On Wed 05 Jun 01:34 PDT 2019, Lee Jones wrote:
>
> > On Wed, 05 Jun 2019, Bjorn Andersson wrote:
> >
> > > On Tue 04 Jun 03:44 PDT 2019, Lee Jones wrote:
> > >
> > > > When booting with Device Tree, the current default boot configuration
> > > > table option, the request to boot via 'host mode' comes from the
> > > > "dr_mode" property.
> > >
> > > This has been the default on the MTP, but this is changing as this is
> > > causing issues when connected downstream from a hub (the typical
> > > development case for the primary USB port of a phone like device) and
> > > more importantly we don't have support for the PMIC blocks that control
> > > VBUS.
> >
> > My point is not about which mode is currently chosen. It's more about
> > the capability of choosing which mode is appropriate for a given
> > system via DT.
> >
> > > Once these issues are resolved the dr_mode would be "otg".
> >
> > OTG doesn't work on this H/W, so we need to specify "host" mode.
>
> My objection is that when you say "this H/W" you mean a particular
> product, but you're making this decision for all SDM845 based products
> using ACPI.
>
> I don't know if there is a Windows phone based on SDM845, but if there
> is then I don't think forcing it to host would be correct.
You mean if someone wanted to boot Linux on a Windows phone? Not sure
how likely that is, but even if a) there is an SDM845 based Windows
phone and b) someone is crazy enough to run Linux on it, it should be
trivial for them to conduct some device matching and choose a
different property based on the result.
[...]
> > > And this driver is used on a range of different Qualcomm platforms, so I
> > > don't think this is SDM845 specific.
> >
> > ACPI based platforms?
> >
> > All the ones I've seen use the XHCI USB driver directly ("PNP0D10").
>
> MSM8998 (835) has the same controller, so this should affect those
> laptops as well.
This would also be the correct configuration for them too. OTG
doesn't make much sense for a laptop form factor.
--
Lee Jones [李琼斯]
Linaro Services Technical Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog