| Message ID | 20230620-topic-dsiphy_rpm-v2-0-a11a751f34f0@linaro.org |
|---|---|
| Headers | show |
| Series | DSIPHY RPM | expand |
On 20/06/2023 14:43, Konrad Dybcio wrote: > This helper has been introduced to avoid programmer errors (missing > _put calls leading to dangling refcnt) when using pm_runtime_get, use it. > > While at it, start checking the return value. > > Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org> > --- > drivers/gpu/drm/msm/dsi/phy/dsi_phy.c | 4 +++- > 1 file changed, 3 insertions(+), 1 deletion(-) Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org> What about also adding the following tag? Fixes: 5c8290284402 ("drm/msm/dsi: Split PHY drivers to separate files")
On 11/07/2023 17:21, Dmitry Baryshkov wrote: > > On Tue, 20 Jun 2023 13:43:19 +0200, Konrad Dybcio wrote: >> Some recent SoCs use power rails that we model as GENPDs to power the >> DSIPHY. This series attempts to make such configurations suspendable. >> >> Tested on SM6375. >> >> > > Applied, thanks! > > [1/2] drm/msm/dsi: Use pm_runtime_resume_and_get to prevent refcnt leaks > https://gitlab.freedesktop.org/lumag/msm/-/commit/a402e0e61b75 > [2/2] drm/msm/dsi: Enable runtime PM > https://gitlab.freedesktop.org/lumag/msm/-/commit/4e905c2acc9d Unfortunately this series breaks our CI, see [1], [2]. I had to remove these patches for now. I suppose this is either because of a probe deferral or because of having subset of drivers built as module. Konrad, could you please take a look? [1] https://gitlab.freedesktop.org/drm/msm/-/jobs/45271774 [2] https://gitlab.freedesktop.org/drm/msm/-/jobs/45271775
On 11.07.2023 18:31, Dmitry Baryshkov wrote: > On 11/07/2023 17:21, Dmitry Baryshkov wrote: >> >> On Tue, 20 Jun 2023 13:43:19 +0200, Konrad Dybcio wrote: >>> Some recent SoCs use power rails that we model as GENPDs to power the >>> DSIPHY. This series attempts to make such configurations suspendable. >>> >>> Tested on SM6375. >>> >>> >> >> Applied, thanks! >> >> [1/2] drm/msm/dsi: Use pm_runtime_resume_and_get to prevent refcnt leaks >> https://gitlab.freedesktop.org/lumag/msm/-/commit/a402e0e61b75 >> [2/2] drm/msm/dsi: Enable runtime PM >> https://gitlab.freedesktop.org/lumag/msm/-/commit/4e905c2acc9d > > Unfortunately this series breaks our CI, see [1], [2]. I had to remove these patches for now. > > I suppose this is either because of a probe deferral or because of having subset of drivers built as module. Konrad, could you please take a look? I see no reason why it would break things :/ You can see that rpmhpd sync_state has not completed, which means all PDs should still be pinned at max vote.. Can we somehow retest it? If it still fails, can you try enabling runtime pm on dispcc and hooking up vddcx? Konrad > > [1] https://gitlab.freedesktop.org/drm/msm/-/jobs/45271774 > [2] https://gitlab.freedesktop.org/drm/msm/-/jobs/45271775 >
On 15.07.2023 17:25, Konrad Dybcio wrote: > On 11.07.2023 18:31, Dmitry Baryshkov wrote: >> On 11/07/2023 17:21, Dmitry Baryshkov wrote: >>> >>> On Tue, 20 Jun 2023 13:43:19 +0200, Konrad Dybcio wrote: >>>> Some recent SoCs use power rails that we model as GENPDs to power the >>>> DSIPHY. This series attempts to make such configurations suspendable. >>>> >>>> Tested on SM6375. >>>> >>>> >>> >>> Applied, thanks! >>> >>> [1/2] drm/msm/dsi: Use pm_runtime_resume_and_get to prevent refcnt leaks >>> https://gitlab.freedesktop.org/lumag/msm/-/commit/a402e0e61b75 >>> [2/2] drm/msm/dsi: Enable runtime PM >>> https://gitlab.freedesktop.org/lumag/msm/-/commit/4e905c2acc9d >> >> Unfortunately this series breaks our CI, see [1], [2]. I had to remove these patches for now. >> >> I suppose this is either because of a probe deferral or because of having subset of drivers built as module. Konrad, could you please take a look? > I see no reason why it would break things :/ > > You can see that rpmhpd sync_state has not completed, which means all > PDs should still be pinned at max vote.. > > Can we somehow retest it? > > If it still fails, can you try enabling runtime pm on dispcc and hooking > up vddcx? IIRC this was a fluke with the CI, can we retry? Konrad
Some recent SoCs use power rails that we model as GENPDs to power the DSIPHY. This series attempts to make such configurations suspendable. Tested on SM6375. Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org> --- Changes in v2: - Improve patch 1 commit message - Add return value checking in patch 1 - Link to v1: https://lore.kernel.org/r/20230620-topic-dsiphy_rpm-v1-0-446a7fd0ce43@linaro.org --- Konrad Dybcio (2): drm/msm/dsi: Use pm_runtime_resume_and_get to prevent refcnt leaks drm/msm/dsi: Enable runtime PM drivers/gpu/drm/msm/dsi/phy/dsi_phy.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) --- base-commit: 9dbf40840551df336c95ce2a3adbdd25ed53c0ef change-id: 20230620-topic-dsiphy_rpm-ea4817badaab Best regards,