Hi Konrad, Thanks for your patch. On 1/13/2024 8:20 PM, Konrad Dybcio wrote:
On Qualcomm SoCs, certain branch clocks either need to be always-on, or should be if you're interested in touching some part of the hardware. Using CLK_IS_CRITICAL for this purpose sounds like a genius idea, however that messes with the runtime pm handling - if a clock is marked as such, the clock controller device will never enter the "suspended" state, leaving the associated resources online, which in turn breaks SoC-wide suspend.
I am really curious to know a little more about the SoC-Wide Suspend not happening on these targets. Could you add more details here ?
The Resource Power Manager (RPM) is the main aggregator on these targets where the active & sleep votes on XO, shared rails (CX/MX) decide the SoC wide suspend. The High Level OS on our internal platforms never had any suspend issues due to clocks(GCC/GPUCC) or shared rails being kept enabled from the consumers.
This series aims to solve that on a couple SoCs that I could test the changes on and it sprinkles some runtime pm enablement atop these drivers.
As CX is a shared resource/rail on these specific targets we definitely do not achieve any power saving with the runtime pm attached to these clock controllers, but I see a little more SW overhead. Though you could please add your observations/comments.
Removing the CLK_IS_CRITICAL is a good cleanup and moving them to probe is a good way to handle the always-on clocks.
Signed-off-by: Konrad Dybcio <konrad.dybcio@xxxxxxxxxx> --- Changes in v6: - Rebase (next-20240112) - Reorder qcom_branch_set_clk_en calls by register in "*: Unregister critical clocks" (Johan) - Pick up tags - Link to v5: https://lore.kernel.org/r/20230717-topic-branch_aon_cleanup-v5-0-99942e6bf1ba@xxxxxxxxxx Changes in v5: - Change the "Keep the critical clocks always-on" comment to "Keep some clocks always-on" - Add the same comment to commits unregistering clocks on 6115/6375/2290 - Link to v4: https://lore.kernel.org/r/20230717-topic-branch_aon_cleanup-v4-0-32c293ded915@xxxxxxxxxx Changes in v4: - Add and unify the "/* Keep the critical clocks always-on */" comment - Rebase (next-20231222), also include 8650, X1E and 8280camcc drivers - Drop enabling runtime PM on GCC - Improve the commit message of "clk: qcom: gpucc-sm6115: Add runtime PM" - Link to v3: https://lore.kernel.org/r/20230717-topic-branch_aon_cleanup-v3-0-3e31bce9c626@xxxxxxxxxx Changes in v3: - Rebase (next-20231219) - Fix up a copypaste mistake in "gcc-sm6375: Unregister critical clocks" (bod) - Pick up tags - Link to v2: https://lore.kernel.org/r/20230717-topic-branch_aon_cleanup-v2-0-2a583460ef26@xxxxxxxxxx Changes in v2: - Rebase - Pick up tags - Fix up missing pm_runtime_put in SM6375 GCC (Johan) - Clarify the commit message of "Add runtime PM" commits (Johan) - "GPU_CCC" -> "GPU_CC" (oops) - Rebase atop next-20231129 - Also fix up camcc-sm8550 & gcc-sm4450 - Unify and clean up the comment style - Fix missing comments in gcc-sc7180.. - Drop Johan's ack from "clk: qcom: Use qcom_branch_set_clk_en()" - Improve 6115 dt patch commit message (Bjorn) - Link to v1: https://lore.kernel.org/r/20230717-topic-branch_aon_cleanup-v1-0-27784d27a4f4@xxxxxxxxxx --- Konrad Dybcio (12): clk: qcom: branch: Add a helper for setting the enable bit clk: qcom: Use qcom_branch_set_clk_en() clk: qcom: gcc-sm6375: Unregister critical clocks clk: qcom: gpucc-sm6375: Unregister critical clocks clk: qcom: gpucc-sm6115: Unregister critical clocks clk: qcom: gpucc-sm6115: Add runtime PM clk: qcom: gcc-sm6115: Unregister critical clocks clk: qcom: gcc-qcm2290: Unregister critical clocks arm64: dts: qcom: sm6375: Add VDD_CX to GCC arm64: dts: qcom: qcm2290: Add VDD_CX to GCC arm64: dts: qcom: sm6115: Add VDD_CX to GCC arm64: dts: qcom: sm6115: Add VDD_CX to GPU_CC > arch/arm64/boot/dts/qcom/qcm2290.dtsi | 1 + arch/arm64/boot/dts/qcom/sm6115.dtsi | 3 + arch/arm64/boot/dts/qcom/sm6375.dtsi | 1 + drivers/clk/qcom/camcc-sc8280xp.c | 6 +- drivers/clk/qcom/camcc-sm8550.c | 10 +-- drivers/clk/qcom/clk-branch.h | 7 ++ drivers/clk/qcom/dispcc-qcm2290.c | 4 +- drivers/clk/qcom/dispcc-sc7280.c | 7 +- drivers/clk/qcom/dispcc-sc8280xp.c | 4 +- drivers/clk/qcom/dispcc-sm6115.c | 4 +- drivers/clk/qcom/dispcc-sm8250.c | 4 +- drivers/clk/qcom/dispcc-sm8450.c | 7 +- drivers/clk/qcom/dispcc-sm8550.c | 7 +- drivers/clk/qcom/dispcc-sm8650.c | 4 +- drivers/clk/qcom/gcc-qcm2290.c | 106 +++-------------------------- drivers/clk/qcom/gcc-sa8775p.c | 25 +++---- drivers/clk/qcom/gcc-sc7180.c | 22 +++--- drivers/clk/qcom/gcc-sc7280.c | 20 +++--- drivers/clk/qcom/gcc-sc8180x.c | 28 +++----- drivers/clk/qcom/gcc-sc8280xp.c | 25 +++---- drivers/clk/qcom/gcc-sdx55.c | 12 ++-- drivers/clk/qcom/gcc-sdx65.c | 13 ++-- drivers/clk/qcom/gcc-sdx75.c | 10 +-- drivers/clk/qcom/gcc-sm4450.c | 28 +++----- drivers/clk/qcom/gcc-sm6115.c | 124 +++------------------------------- drivers/clk/qcom/gcc-sm6375.c | 105 +++------------------------- drivers/clk/qcom/gcc-sm7150.c | 23 +++---- drivers/clk/qcom/gcc-sm8250.c | 19 ++---- drivers/clk/qcom/gcc-sm8350.c | 20 +++--- drivers/clk/qcom/gcc-sm8450.c | 21 +++--- drivers/clk/qcom/gcc-sm8550.c | 21 +++--- drivers/clk/qcom/gcc-sm8650.c | 16 ++--- drivers/clk/qcom/gcc-x1e80100.c | 16 ++--- drivers/clk/qcom/gpucc-sc7280.c | 9 +-- drivers/clk/qcom/gpucc-sc8280xp.c | 9 +-- drivers/clk/qcom/gpucc-sm6115.c | 53 ++++++--------- drivers/clk/qcom/gpucc-sm6375.c | 34 ++-------- drivers/clk/qcom/gpucc-sm8550.c | 10 +-- drivers/clk/qcom/lpasscorecc-sc7180.c | 7 +- drivers/clk/qcom/videocc-sm8250.c | 6 +- drivers/clk/qcom/videocc-sm8350.c | 10 +-- drivers/clk/qcom/videocc-sm8450.c | 13 ++-- drivers/clk/qcom/videocc-sm8550.c | 13 ++-- 43 files changed, 234 insertions(+), 653 deletions(-) --- base-commit: 8d04a7e2ee3fd6aabb8096b00c64db0d735bc874 change-id: 20230717-topic-branch_aon_cleanup-6976c13fe71c Best regards,
-- Thanks & Regards, Taniya Das.
