[PATCH 0/2] arm64: dts: qcom: msm8996: fix issues with probing of USB hosts

[PATCH 0/2] arm64: dts: qcom: msm8996: fix issues with probing of USB hosts

[Dmitry Baryshkov]

The lack of the interconnects for the USB2 host might result in the
register writes for the PHY not reaching the PHY and thus the PLL not
locking up, resulting in -EBUSY errors for the USB host on the platform.
Add missing interconnect link and add interconnect paths to the USB2
device.

Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
---
Dmitry Baryshkov (2):
      interconnect: qcom: msm8996: add missing link to SLAVE_USB_HS
      arm64: dts: qcom: msm8996: add interconnect paths to USB2 controller

 arch/arm64/boot/dts/qcom/msm8996.dtsi | 3 +++
 drivers/interconnect/qcom/msm8996.c   | 1 +
 2 files changed, 4 insertions(+)
---
base-commit: bf2602a3cb2381fb1a04bf1c39a290518d2538d1
change-id: 20251002-fix-msm8996-icc-ce1df61ad702

Best regards,
-- 
With best wishes
Dmitry

[PATCH 1/2] interconnect: qcom: msm8996: add missing link to SLAVE_USB_HS

[Dmitry Baryshkov]

From the initial submission the interconnect driver missed the link from
SNOC_PNOC to the USB 2 configuration space. Add missing link in order to
let the platform configure and utilize this path.

Fixes: 7add937f5222 ("interconnect: qcom: Add MSM8996 interconnect provider driver")
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
---
 drivers/interconnect/qcom/msm8996.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/interconnect/qcom/msm8996.c b/drivers/interconnect/qcom/msm8996.c
index b73566c9b21f9dd275878419e030ab07163ef8dd..84cfafb22aa17dba0d5fc4074613546c3dd95625 100644
--- a/drivers/interconnect/qcom/msm8996.c
+++ b/drivers/interconnect/qcom/msm8996.c
@@ -552,6 +552,7 @@ static struct qcom_icc_node mas_venus_vmem = {
 static const u16 mas_snoc_pnoc_links[] = {
 	MSM8996_SLAVE_BLSP_1,
 	MSM8996_SLAVE_BLSP_2,
+	MSM8996_SLAVE_USB_HS,
 	MSM8996_SLAVE_SDCC_1,
 	MSM8996_SLAVE_SDCC_2,
 	MSM8996_SLAVE_SDCC_4,

-- 
2.47.3

[PATCH 2/2] arm64: dts: qcom: msm8996: add interconnect paths to USB2 controller

[Dmitry Baryshkov]

Add the missing interconnects to the USB2 host. The Fixes tag points to
the commit which broke probing of the USB host on that platform.

Fixes: 130733a10079 ("interconnect: qcom: msm8996: Promote to core_initcall")
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>

---
Note: without the previous patch applying this one can result in the
kernel stuck at booting because of the EPROBE_DEFER loop. I suggest
applying them through the same tree in order to make sure that the tree
is not broken (or using an immutable tag for the icc commit).
---
 arch/arm64/boot/dts/qcom/msm8996.dtsi | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/arch/arm64/boot/dts/qcom/msm8996.dtsi b/arch/arm64/boot/dts/qcom/msm8996.dtsi
index c75b522f6eba66afeb71be5d81624183641bde71..33608b1d7d060b0a614357929eb3404bab46ecb6 100644
--- a/arch/arm64/boot/dts/qcom/msm8996.dtsi
+++ b/arch/arm64/boot/dts/qcom/msm8996.dtsi
@@ -3496,6 +3496,9 @@ usb2: usb@76f8800 {
 					  <&gcc GCC_USB20_MASTER_CLK>;
 			assigned-clock-rates = <19200000>, <60000000>;
 
+			interconnects = <&pnoc MASTER_USB_HS &bimc SLAVE_EBI_CH0>,
+					<&bimc MASTER_AMPSS_M0 &pnoc SLAVE_USB_HS>;
+			interconnect-names = "usb-ddr", "apps-usb";
 			power-domains = <&gcc USB30_GDSC>;
 			qcom,select-utmi-as-pipe-clk;
 			status = "disabled";

-- 
2.47.3

Re: [PATCH 1/2] interconnect: qcom: msm8996: add missing link to SLAVE_USB_HS

[Konrad Dybcio]

On 10/2/25 10:53 AM, Dmitry Baryshkov wrote:
> From the initial submission the interconnect driver missed the link from
> SNOC_PNOC to the USB 2 configuration space. Add missing link in order to
> let the platform configure and utilize this path.
> 
> Fixes: 7add937f5222 ("interconnect: qcom: Add MSM8996 interconnect provider driver")
> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
> ---

Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>

Konrad

Re: [PATCH 2/2] arm64: dts: qcom: msm8996: add interconnect paths to USB2 controller

[Konrad Dybcio]

On 10/2/25 10:53 AM, Dmitry Baryshkov wrote:
> Add the missing interconnects to the USB2 host. The Fixes tag points to
> the commit which broke probing of the USB host on that platform.
> 
> Fixes: 130733a10079 ("interconnect: qcom: msm8996: Promote to core_initcall")
> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
> 
> ---
> Note: without the previous patch applying this one can result in the
> kernel stuck at booting because of the EPROBE_DEFER loop. I suggest
> applying them through the same tree in order to make sure that the tree
> is not broken (or using an immutable tag for the icc commit).
> ---

with that ^

Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>

Konrad

Re: [PATCH 2/2] arm64: dts: qcom: msm8996: add interconnect paths to USB2 controller

[Konrad Dybcio]

On 10/2/25 10:53 AM, Dmitry Baryshkov wrote:
> Add the missing interconnects to the USB2 host. The Fixes tag points to
> the commit which broke probing of the USB host on that platform.
> 
> Fixes: 130733a10079 ("interconnect: qcom: msm8996: Promote to core_initcall")
> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
> 
> ---
> Note: without the previous patch applying this one can result in the
> kernel stuck at booting because of the EPROBE_DEFER loop. I suggest
> applying them through the same tree in order to make sure that the tree
> is not broken (or using an immutable tag for the icc commit).
> ---

There seems to be syntax for this, at least for stable patches
(which we should probably either use or add +noautosel):

* Specify any additional patch prerequisites for cherry picking::

    xCxc: <stable@vger.kernel.org> # 3.3.x: a1f84a3: sched: Check for idle
    xCxc: <stable@vger.kernel.org> # 3.3.x: 1b9508f: sched: Rate-limit newidle
    xCxc: <stable@vger.kernel.org> # 3.3.x: fd21073: sched: Fix affinity logic
    xCxc: <stable@vger.kernel.org> # 3.3.x
    xSxixgxxnxexd-oxfxf-xbxy: Ingo Molnar <mingo@elte.hu>

(deliberatelly added a bunch of 'x'-es to not interrupt the maintainers'
b4 workflow)

(via https://www.kernel.org/doc/Documentation/process/stable-kernel-rules.rst)

Konrad

Re: [PATCH 2/2] arm64: dts: qcom: msm8996: add interconnect paths to USB2 controller

[Dmitry Baryshkov]

On Mon, Oct 06, 2025 at 11:30:52AM +0200, Konrad Dybcio wrote:
> On 10/2/25 10:53 AM, Dmitry Baryshkov wrote:
> > Add the missing interconnects to the USB2 host. The Fixes tag points to
> > the commit which broke probing of the USB host on that platform.
> > 
> > Fixes: 130733a10079 ("interconnect: qcom: msm8996: Promote to core_initcall")
> > Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
> > 
> > ---
> > Note: without the previous patch applying this one can result in the
> > kernel stuck at booting because of the EPROBE_DEFER loop. I suggest
> > applying them through the same tree in order to make sure that the tree
> > is not broken (or using an immutable tag for the icc commit).
> > ---
> 
> There seems to be syntax for this, at least for stable patches
> (which we should probably either use or add +noautosel):

This syntax is for stable trees. At this point I'm more worried about
those patches being picked up correctly. If the patches are picked up in
a correct order, they can be a subject to autosel (provided that it
maintains the order).

> 
> * Specify any additional patch prerequisites for cherry picking::
> 
>     xCxc: <stable@vger.kernel.org> # 3.3.x: a1f84a3: sched: Check for idle
>     xCxc: <stable@vger.kernel.org> # 3.3.x: 1b9508f: sched: Rate-limit newidle
>     xCxc: <stable@vger.kernel.org> # 3.3.x: fd21073: sched: Fix affinity logic
>     xCxc: <stable@vger.kernel.org> # 3.3.x
>     xSxixgxxnxexd-oxfxf-xbxy: Ingo Molnar <mingo@elte.hu>
> 
> (deliberatelly added a bunch of 'x'-es to not interrupt the maintainers'
> b4 workflow)
> 
> (via https://www.kernel.org/doc/Documentation/process/stable-kernel-rules.rst)
> 
> Konrad

-- 
With best wishes
Dmitry

Re: [PATCH 2/2] arm64: dts: qcom: msm8996: add interconnect paths to USB2 controller

[Dmitry Baryshkov]

On Mon, Oct 06, 2025 at 11:30:52AM +0200, Konrad Dybcio wrote:
> On 10/2/25 10:53 AM, Dmitry Baryshkov wrote:
> > Add the missing interconnects to the USB2 host. The Fixes tag points to
> > the commit which broke probing of the USB host on that platform.
> > 
> > Fixes: 130733a10079 ("interconnect: qcom: msm8996: Promote to core_initcall")
> > Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
> > 
> > ---
> > Note: without the previous patch applying this one can result in the
> > kernel stuck at booting because of the EPROBE_DEFER loop. I suggest
> > applying them through the same tree in order to make sure that the tree
> > is not broken (or using an immutable tag for the icc commit).
> > ---
> 
> There seems to be syntax for this, at least for stable patches
> (which we should probably either use or add +noautosel):
> 
> * Specify any additional patch prerequisites for cherry picking::
> 
>     xCxc: <stable@vger.kernel.org> # 3.3.x: a1f84a3: sched: Check for idle
>     xCxc: <stable@vger.kernel.org> # 3.3.x: 1b9508f: sched: Rate-limit newidle
>     xCxc: <stable@vger.kernel.org> # 3.3.x: fd21073: sched: Fix affinity logic
>     xCxc: <stable@vger.kernel.org> # 3.3.x
>     xSxixgxxnxexd-oxfxf-xbxy: Ingo Molnar <mingo@elte.hu>
> 
> (deliberatelly added a bunch of 'x'-es to not interrupt the maintainers'
> b4 workflow)

Bjorn, Georgi, Konrad, how should we proceed with this series?

-- 
With best wishes
Dmitry

Re: [PATCH 2/2] arm64: dts: qcom: msm8996: add interconnect paths to USB2 controller

[Bjorn Andersson]

On Thu, Oct 30, 2025 at 08:24:29PM +0200, Dmitry Baryshkov wrote:
> On Mon, Oct 06, 2025 at 11:30:52AM +0200, Konrad Dybcio wrote:
> > On 10/2/25 10:53 AM, Dmitry Baryshkov wrote:
> > > Add the missing interconnects to the USB2 host. The Fixes tag points to
> > > the commit which broke probing of the USB host on that platform.
> > > 
> > > Fixes: 130733a10079 ("interconnect: qcom: msm8996: Promote to core_initcall")
> > > Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
> > > 
> > > ---
> > > Note: without the previous patch applying this one can result in the
> > > kernel stuck at booting because of the EPROBE_DEFER loop. I suggest
> > > applying them through the same tree in order to make sure that the tree
> > > is not broken (or using an immutable tag for the icc commit).
> > > ---
> > 
> > There seems to be syntax for this, at least for stable patches
> > (which we should probably either use or add +noautosel):
> > 
> > * Specify any additional patch prerequisites for cherry picking::
> > 
> >     xCxc: <stable@vger.kernel.org> # 3.3.x: a1f84a3: sched: Check for idle
> >     xCxc: <stable@vger.kernel.org> # 3.3.x: 1b9508f: sched: Rate-limit newidle
> >     xCxc: <stable@vger.kernel.org> # 3.3.x: fd21073: sched: Fix affinity logic
> >     xCxc: <stable@vger.kernel.org> # 3.3.x
> >     xSxixgxxnxexd-oxfxf-xbxy: Ingo Molnar <mingo@elte.hu>
> > 
> > (deliberatelly added a bunch of 'x'-es to not interrupt the maintainers'
> > b4 workflow)
> 
> Bjorn, Georgi, Konrad, how should we proceed with this series?
> 

I agree with you on the Cc part, that doesn't affect how these patches
shows up in torvalds/master during the merge window. So merging them in
the same branch would be the way.

@Georgi, I'm fine with doing this in either tree, if you prefer to pick
the two patches:

Acked-by: Bjorn Andersson <andersson@kernel.org>

Regards,
Bjorn

> -- 
> With best wishes
> Dmitry