Re: [PATCH 03/10] mfd: qcom_rpm: add msm8960 QDSS clock resource

[Antony Kurniawan Soemardi]

On 4/18/2026 11:11 PM, Dmitry Baryshkov wrote:
> On Thu, Apr 16, 2026 at 03:49:33PM +0200, Konrad Dybcio wrote:
> > On 4/15/26 5:20 PM, Antony Kurniawan Soemardi wrote:
> > > On 4/14/2026 3:07 PM, Konrad Dybcio wrote:
> > > > On 4/14/26 10:06 AM, Konrad Dybcio wrote:
> > > > > On 4/13/26 8:55 PM, Antony Kurniawan Soemardi via B4 Relay wrote:
> > > > > > From: Antony Kurniawan Soemardi <linux@xxxxxxxxxxxxxx>
> > > > > >
> > > > > > msm8960 uses the same clock descriptor as apq8064 but lacked the
> > > > >
> > > > > This doesn't quite seem to be the case, some fields differ and
> > > > > apq8064 additionally has:
> > > > >
> > > > > QCOM_RPM_PM8821_SMPS1
> > > > > QCOM_RPM_PM8821_SMPS2
> > > > > QCOM_RPM_PM8821_LDO1
> > > > > QCOM_RPM_VDDMIN_GPIO
> > > >
> > > > Ah hmm, the MFD driver seems to provide *all* RPM resources..
> > >
> > > What I meant by "clock descriptor" in the commit message was
> > > specifically the subset corresponding to RPM managed clocks. From what I
> > > can tell based on downstream code, msm8960 and apq8064 seem to share the
> > > same set of RPM clocks, even though the overall resource lists differ.
> > >
> > > Is that understanding correct?
> >
> > If that's struct msm_rpm_map_data on msm-3.x, then I see that 8x60 has:
> >
> > +MSM_RPM_MAP(PLL_4, PLL_4, 1),
> > +MSM_RPM_MAP(SMI_CLK, SMI_CLK, 1),
> >
> > While 8960 has:
> > -MSM_RPM_MAP(QDSS_CLK, QDSS_CLK, 1),
>
> You are comparing 8x60 to 8960, while it should be 8960 to 8064.
>
> I see that there are differences, but the QDSS is the same.

I'm looking at downstream code from my Sony Xperia SP (based on
android-msm-mako-3.4 for LG Nexus 4 / APQ8064). Both apq8064 and msm8960
uses the same clock-8960.c file in that tree, including the RPM clock
descriptions, so I assumed their RPM clock sets were identical.

That's why later commit 1da13533627d ("clk: qcom: clk-rpm: add msm8960
compatible") reuses rpm_clk_apq8064 for qcom,rpmcc-msm8960 compatible.
But the problem is rpm_clk_apq8064 includes QDSS, which is not yet
present on msm8960_rpm_resource_table, causing rpmcc init to fail, hence
this patch.

That said, I agree with Dmitry on the other email reply. I will update
the commit message to match with android-msm-mako-3.4 downstream code
instead of claiming both SoC has the same clock descriptions.

--
Thanks,
Antony K. S.