On Tue Jul 1, 2025 at 1:16 PM CEST, Dmitry Baryshkov wrote:
> On Mon, Jun 30, 2025 at 10:01:35AM +0200, Luca Weiss wrote:
>> Hi Konrad,
>>
>> On Fri Jun 27, 2025 at 5:14 PM CEST, Luca Weiss wrote:
>> > On Fri Jun 27, 2025 at 5:10 PM CEST, Konrad Dybcio wrote:
>> >> On 6/25/25 11:12 AM, Luca Weiss wrote:
>> >>> Document and add the clock drivers for GCC, CAMCC, DISPCC, GPUCC and
>> >>> VIDEOCC on the SM7635 SoC.
>> >>>
>> >>> Signed-off-by: Luca Weiss <[email protected]>
>> >>> ---
>> >>> Luca Weiss (10):
>> >>> dt-bindings: clock: qcom: document the SM7635 Global Clock
>> >>> Controller
>> >>> clk: qcom: Add Global Clock controller (GCC) driver for SM7635
>> >>> dt-bindings: clock: qcom: document the SM7635 Camera Clock
>> >>> Controller
>> >>> clk: qcom: Add Camera Clock controller (CAMCC) driver for SM7635
>> >>> dt-bindings: clock: qcom: document the SM7635 Display Clock
>> >>> Controller
>> >>> clk: qcom: Add Display Clock controller (DISPCC) driver for SM7635
>> >>> dt-bindings: clock: qcom: document the SM7635 GPU Clock Controller
>> >>> clk: qcom: Add Graphics Clock controller (GPUCC) driver for SM7635
>> >>> dt-bindings: clock: qcom: document the SM7635 Video Clock
>> >>> Controller
>> >>> clk: qcom: Add Video Clock controller (VIDEOCC) driver for SM7635
>> >>
>> >> We had a massive yak shaving patchset go in this season, please move
>> >> the magic settings in .probe to qcom_cc_driver_data {}
>> >
>> > Okay cool, I found them
>> > https://lore.kernel.org/linux-arm-msm/[email protected]/
>>
>> For camcc, gpucc and videocc it seems quite simple to follow these
>> changes.
>>
>> For dispcc I don't know what to do with this line.
>>
>> /* Enable clock gating for MDP clocks */
>> regmap_update_bits(regmap, DISP_CC_MISC_CMD, 0x10, 0x10);
>
> Use clk_regs_configure() callback to set this bit.
Got it, found an example on the lists.
>
>>
>> Do I just keep the regmap references in this probe function and just
>> move the clk_lucid_ole_pll_configure & qcom_branch_set_clk_en to the
>> config struct?
>>
>> And similar for gcc, I can move the qcom_branch_set_clk_en calls there
>> but the qcom_cc_register_rcg_dfs needs to be kept.
>
> Would you mind extnding struct qcom_cc_desc with args to call
> qcom_cc_register_rcg_dfs() and call it from qcom_cc_really_probe()?
Something like this? Not quite sure when (in what order) this should be
called, is that place fine?
I'd include a patch then in the v2 of this series.
diff --git a/drivers/clk/qcom/common.c b/drivers/clk/qcom/common.c
index b3838d885db2..d53f290c6121 100644
--- a/drivers/clk/qcom/common.c
+++ b/drivers/clk/qcom/common.c
@@ -390,6 +390,14 @@ int qcom_cc_really_probe(struct device *dev,
goto put_rpm;
}
+ if (desc->dfs_rcgs && desc->num_dfs_rcgs) {
+ ret = qcom_cc_register_rcg_dfs(regmap,
+ desc->dfs_rcgs,
+ desc->num_dfs_rcgs);
+ if (ret)
+ goto put_rpm;
+ }
+
cc->rclks = rclks;
cc->num_rclks = num_clks;
diff --git a/drivers/clk/qcom/common.h b/drivers/clk/qcom/common.h
index 0f4b2d40c65c..dbe7ebe5b871 100644
--- a/drivers/clk/qcom/common.h
+++ b/drivers/clk/qcom/common.h
@@ -43,6 +43,8 @@ struct qcom_cc_desc {
size_t num_gdscs;
struct clk_hw **clk_hws;
size_t num_clk_hws;
+ const struct clk_rcg_dfs_data *dfs_rcgs;
+ size_t num_dfs_rcgs;
const struct qcom_icc_hws_data *icc_hws;
size_t num_icc_hws;
unsigned int icc_first_node_id;
Regards
Luca
>
>>
>> Does that sound okay, or what exactly is the desired outcome of the
>> requested changes?
>>
>> Regards
>> Luca
>>
>> >
>> >>
>> >> Konrad
>>
