* [PATCH v2 00/22] Restructure RPM SMD ICC
@ 2023-06-09 20:19 Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 01/22] soc: qcom: smd-rpm: Add QCOM_SMD_RPM_STATE_NUM Konrad Dybcio
` (21 more replies)
0 siblings, 22 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Dmitry Baryshkov, Stephan Gerhold
This series reshuffles things around, moving the management of SMD RPM
bus clocks to the interconnect framework where they belong. This helps
us solve a couple of issues:
1. We can work towards unused clk cleanup of RPMCC without worrying
about it killing some NoC bus, resulting in the SoC dying.
Deasserting actually unused RPM clocks (among other things) will
let us achieve "true SoC-wide power collapse states", also known as
VDD_LOW and VDD_MIN.
2. We no longer have to keep tons of quirky bus clock ifs in the icc
driver. You either have a RPM clock and call "rpm set rate" or you
have a single non-RPM clock (like AHB_CLK_SRC) or you don't have any.
3. There's less overhead - instead of going through layers and layers of
the CCF, ratesetting comes down to calling max() and sending a single
RPM message. ICC is very very dynamic so that's a big plus.
The clocks still need to be vaguely described in the clk-smd-rpm driver,
as it gives them an initial kickoff, before actually telling RPM to
enable DVFS scaling. After RPM receives that command, all clocks that
have not been assigned a rate are considered unused and are shut down
in hardware, leading to the same issue as described in point 1.
We can consider marking them __initconst in the future, but this series
is very fat even without that..
Apart from that, it squashes a couple of bugs that really need fixing..
--- MERGING STRATEGY ---
If Stephen and Georgi agree, it would be best to take all of this through
the qcom tree, as it touches on heavily intertwined components and
introduces compile-time dependencies between icc and clk drivers.
Tested on SM6375 (OOT), MSM8998 (OOT), MSM8996.
MSM8974 conversion to common code and modernization will be handled separately.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
Changes in v2:
- Sort entries properly in "Add missing headers in icc-rpm.h"
- Fix the check for no clocks on a given provider
- Replace "Divide clk rate by src node bus width" with a proper fix
- Add "Set correct bandwidth through RPM bw req"
- Split "Add QCOM_SMD_RPM_STATE_NUM" into 2 logical changes
- Move "Separate out interconnect bus clocks" a bit later in the series
- Link to v1: https://lore.kernel.org/r/20230526-topic-smd_icc-v1-0-1bf8e6663c4e@linaro.org
---
Konrad Dybcio (22):
soc: qcom: smd-rpm: Add QCOM_SMD_RPM_STATE_NUM
soc: qcom: smd-rpm: Use tabs for defines
clk: qcom: smd-rpm: Move some RPM resources to the common header
clk: qcom: smd-rpm: Export clock scaling availability
interconnect: qcom: icc-rpm: Introduce keep_alive
interconnect: qcom: icc-rpm: Allow negative QoS offset
interconnect: qcom: Fold smd-rpm.h into icc-rpm.h
interconnect: qcom: smd-rpm: Add rpmcc handling skeleton code
interconnect: qcom: Add missing headers in icc-rpm.h
interconnect: qcom: Define RPM bus clocks
interconnect: qcom: sdm660: Hook up RPM bus clk definitions
interconnect: qcom: msm8996: Hook up RPM bus clk definitions
interconnect: qcom: qcs404: Hook up RPM bus clk definitions
interconnect: qcom: msm8939: Hook up RPM bus clk definitions
interconnect: qcom: msm8916: Hook up RPM bus clk definitions
interconnect: qcom: qcm2290: Hook up RPM bus clk definitions
interconnect: qcom: icc-rpm: Control bus rpmcc from icc
clk: qcom: smd-rpm: Separate out interconnect bus clocks
interconnect: qcom: icc-rpm: Fix bucket number
interconnect: qcom: icc-rpm: Set bandwidth on both contexts
interconnect: qcom: icc-rpm: Set correct bandwidth through RPM bw req
interconnect: qcom: icc-rpm: Fix bandwidth calculations
drivers/clk/qcom/clk-smd-rpm.c | 300 ++++++++++++-----------------
drivers/interconnect/qcom/Makefile | 2 +-
drivers/interconnect/qcom/icc-rpm-clocks.c | 66 +++++++
drivers/interconnect/qcom/icc-rpm.c | 212 ++++++++++----------
drivers/interconnect/qcom/icc-rpm.h | 55 ++++--
drivers/interconnect/qcom/msm8916.c | 4 +-
drivers/interconnect/qcom/msm8939.c | 5 +-
drivers/interconnect/qcom/msm8974.c | 2 +-
drivers/interconnect/qcom/msm8996.c | 9 +-
drivers/interconnect/qcom/qcm2290.c | 7 +-
drivers/interconnect/qcom/qcs404.c | 4 +-
drivers/interconnect/qcom/sdm660.c | 7 +-
drivers/interconnect/qcom/smd-rpm.c | 39 +++-
drivers/interconnect/qcom/smd-rpm.h | 15 --
include/linux/soc/qcom/smd-rpm.h | 22 ++-
15 files changed, 427 insertions(+), 322 deletions(-)
---
base-commit: 53ab6975c12d1ad86c599a8927e8c698b144d669
change-id: 20230526-topic-smd_icc-b8213948a5ed
Best regards,
--
Konrad Dybcio <konrad.dybcio@linaro.org>
^ permalink raw reply [flat|nested] 45+ messages in thread
* [PATCH v2 01/22] soc: qcom: smd-rpm: Add QCOM_SMD_RPM_STATE_NUM
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 02/22] soc: qcom: smd-rpm: Use tabs for defines Konrad Dybcio
` (20 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio
Add a preprocessor define to indicate the number of RPM contexts/states.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
include/linux/soc/qcom/smd-rpm.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/include/linux/soc/qcom/smd-rpm.h b/include/linux/soc/qcom/smd-rpm.h
index 2990f425fdef..e468f94fa323 100644
--- a/include/linux/soc/qcom/smd-rpm.h
+++ b/include/linux/soc/qcom/smd-rpm.h
@@ -6,6 +6,7 @@ struct qcom_smd_rpm;
#define QCOM_SMD_RPM_ACTIVE_STATE 0
#define QCOM_SMD_RPM_SLEEP_STATE 1
+#define QCOM_SMD_RPM_STATE_NUM 2
/*
* Constants used for addressing resources in the RPM.
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 02/22] soc: qcom: smd-rpm: Use tabs for defines
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 01/22] soc: qcom: smd-rpm: Add QCOM_SMD_RPM_STATE_NUM Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 03/22] clk: qcom: smd-rpm: Move some RPM resources to the common header Konrad Dybcio
` (19 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio
Use tabs for defines to make things spaced consistently.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
include/linux/soc/qcom/smd-rpm.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/include/linux/soc/qcom/smd-rpm.h b/include/linux/soc/qcom/smd-rpm.h
index e468f94fa323..99499e4b080e 100644
--- a/include/linux/soc/qcom/smd-rpm.h
+++ b/include/linux/soc/qcom/smd-rpm.h
@@ -4,8 +4,8 @@
struct qcom_smd_rpm;
-#define QCOM_SMD_RPM_ACTIVE_STATE 0
-#define QCOM_SMD_RPM_SLEEP_STATE 1
+#define QCOM_SMD_RPM_ACTIVE_STATE 0
+#define QCOM_SMD_RPM_SLEEP_STATE 1
#define QCOM_SMD_RPM_STATE_NUM 2
/*
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 03/22] clk: qcom: smd-rpm: Move some RPM resources to the common header
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 01/22] soc: qcom: smd-rpm: Add QCOM_SMD_RPM_STATE_NUM Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 02/22] soc: qcom: smd-rpm: Use tabs for defines Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 04/22] clk: qcom: smd-rpm: Export clock scaling availability Konrad Dybcio
` (18 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Dmitry Baryshkov
In preparation for handling the bus clocks in the icc driver, carve out
some defines and a struct definition to the common rpm header.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/clk/qcom/clk-smd-rpm.c | 13 -------------
include/linux/soc/qcom/smd-rpm.h | 15 +++++++++++++++
2 files changed, 15 insertions(+), 13 deletions(-)
diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
index e4de74b68797..937cb1515968 100644
--- a/drivers/clk/qcom/clk-smd-rpm.c
+++ b/drivers/clk/qcom/clk-smd-rpm.c
@@ -18,13 +18,6 @@
#include <dt-bindings/clock/qcom,rpmcc.h>
-#define QCOM_RPM_KEY_SOFTWARE_ENABLE 0x6e657773
-#define QCOM_RPM_KEY_PIN_CTRL_CLK_BUFFER_ENABLE_KEY 0x62636370
-#define QCOM_RPM_SMD_KEY_RATE 0x007a484b
-#define QCOM_RPM_SMD_KEY_ENABLE 0x62616e45
-#define QCOM_RPM_SMD_KEY_STATE 0x54415453
-#define QCOM_RPM_SCALING_ENABLE_ID 0x2
-
#define __DEFINE_CLK_SMD_RPM_PREFIX(_prefix, _name, _active, \
type, r_id, key) \
static struct clk_smd_rpm clk_smd_rpm_##_prefix##_active; \
@@ -171,12 +164,6 @@ struct clk_smd_rpm {
unsigned long rate;
};
-struct clk_smd_rpm_req {
- __le32 key;
- __le32 nbytes;
- __le32 value;
-};
-
struct rpm_smd_clk_desc {
struct clk_smd_rpm **clks;
size_t num_clks;
diff --git a/include/linux/soc/qcom/smd-rpm.h b/include/linux/soc/qcom/smd-rpm.h
index 99499e4b080e..8190878645f9 100644
--- a/include/linux/soc/qcom/smd-rpm.h
+++ b/include/linux/soc/qcom/smd-rpm.h
@@ -2,6 +2,8 @@
#ifndef __QCOM_SMD_RPM_H__
#define __QCOM_SMD_RPM_H__
+#include <linux/types.h>
+
struct qcom_smd_rpm;
#define QCOM_SMD_RPM_ACTIVE_STATE 0
@@ -45,6 +47,19 @@ struct qcom_smd_rpm;
#define QCOM_SMD_RPM_PKA_CLK 0x616b70
#define QCOM_SMD_RPM_MCFG_CLK 0x6766636d
+#define QCOM_RPM_KEY_SOFTWARE_ENABLE 0x6e657773
+#define QCOM_RPM_KEY_PIN_CTRL_CLK_BUFFER_ENABLE_KEY 0x62636370
+#define QCOM_RPM_SMD_KEY_RATE 0x007a484b
+#define QCOM_RPM_SMD_KEY_ENABLE 0x62616e45
+#define QCOM_RPM_SMD_KEY_STATE 0x54415453
+#define QCOM_RPM_SCALING_ENABLE_ID 0x2
+
+struct clk_smd_rpm_req {
+ __le32 key;
+ __le32 nbytes;
+ __le32 value;
+};
+
int qcom_rpm_smd_write(struct qcom_smd_rpm *rpm,
int state,
u32 resource_type, u32 resource_id,
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 04/22] clk: qcom: smd-rpm: Export clock scaling availability
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (2 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 03/22] clk: qcom: smd-rpm: Move some RPM resources to the common header Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-10 11:35 ` Stephan Gerhold
2023-06-09 20:19 ` [PATCH v2 05/22] interconnect: qcom: icc-rpm: Introduce keep_alive Konrad Dybcio
` (17 subsequent siblings)
21 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio
Before we issue a call to RPM through clk_smd_rpm_enable_scaling() the
clock rate requests will not be commited in hardware. This poses a
race threat since we're accessing the bus clocks directly from within
the interconnect framework.
Add a marker to indicate that we're good to go with sending new requests
and export it so that it can be referenced from icc.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/clk/qcom/clk-smd-rpm.c | 9 +++++++++
include/linux/soc/qcom/smd-rpm.h | 2 ++
2 files changed, 11 insertions(+)
diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
index 937cb1515968..482fe30ee6f0 100644
--- a/drivers/clk/qcom/clk-smd-rpm.c
+++ b/drivers/clk/qcom/clk-smd-rpm.c
@@ -151,6 +151,7 @@
#define to_clk_smd_rpm(_hw) container_of(_hw, struct clk_smd_rpm, hw)
static struct qcom_smd_rpm *rpmcc_smd_rpm;
+static bool smd_rpm_clk_scaling;
struct clk_smd_rpm {
const int rpm_res_type;
@@ -385,6 +386,12 @@ static unsigned long clk_smd_rpm_recalc_rate(struct clk_hw *hw,
return r->rate;
}
+bool qcom_smd_rpm_scaling_available(void)
+{
+ return smd_rpm_clk_scaling;
+}
+EXPORT_SYMBOL_GPL(qcom_smd_rpm_scaling_available);
+
static int clk_smd_rpm_enable_scaling(void)
{
int ret;
@@ -410,6 +417,8 @@ static int clk_smd_rpm_enable_scaling(void)
return ret;
}
+ smd_rpm_clk_scaling = true;
+
pr_debug("%s: RPM clock scaling is enabled\n", __func__);
return 0;
}
diff --git a/include/linux/soc/qcom/smd-rpm.h b/include/linux/soc/qcom/smd-rpm.h
index 8190878645f9..420f6c79d535 100644
--- a/include/linux/soc/qcom/smd-rpm.h
+++ b/include/linux/soc/qcom/smd-rpm.h
@@ -65,4 +65,6 @@ int qcom_rpm_smd_write(struct qcom_smd_rpm *rpm,
u32 resource_type, u32 resource_id,
void *buf, size_t count);
+bool qcom_smd_rpm_scaling_available(void);
+
#endif
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 05/22] interconnect: qcom: icc-rpm: Introduce keep_alive
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (3 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 04/22] clk: qcom: smd-rpm: Export clock scaling availability Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 06/22] interconnect: qcom: icc-rpm: Allow negative QoS offset Konrad Dybcio
` (16 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Dmitry Baryshkov
The downstream kernel employs the concept of "keeping the bus alive"
by voting for the minimum (XO/19.2MHz) rate at all times on certain
(well, most) buses. This is a very important thing to have, as if we
either have a lackluster/wrong DT that doesn't specify a (high enough)
vote on a certain bus, we may lose access to the entire bus altogether.
This is very apparent when we only start introducing interconnect
support on a given platform and haven't yet introduced voting on all
peripherals.
The same can happen if we only have a single driver casting a vote on
a certain bus and that driver exits/crashes/suspends.
The keepalive vote is limited to the ACTIVE bucket, as keeping a
permanent vote on the SLEEP one could prevent the platform from properly
entering low power mode states.
Introduce the very same concept, with a slight twist: the vendor
kernel checks whether the rate is zero before setting the minimum
vote, but that's rather silly, as in doing so we're at the mercy
of CCF. Instead, explicitly clamp the rates to always be >= 19.2 MHz
for providers with keep_alive=true.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/icc-rpm.c | 10 ++++++++++
drivers/interconnect/qcom/icc-rpm.h | 3 +++
2 files changed, 13 insertions(+)
diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
index 6acc7686ed38..863e8ba1daa2 100644
--- a/drivers/interconnect/qcom/icc-rpm.c
+++ b/drivers/interconnect/qcom/icc-rpm.c
@@ -50,6 +50,8 @@
#define NOC_QOS_MODE_FIXED_VAL 0x0
#define NOC_QOS_MODE_BYPASS_VAL 0x2
+#define ICC_BUS_CLK_MIN_RATE 19200000ULL
+
static int qcom_icc_set_qnoc_qos(struct icc_node *src)
{
struct icc_provider *provider = src->provider;
@@ -380,6 +382,13 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
do_div(rate, src_qn->buswidth);
rate = min_t(u64, rate, LONG_MAX);
+ /*
+ * Downstream checks whether the requested rate is zero, but it makes little sense
+ * to vote for a value that's below the lower threshold, so let's not do so.
+ */
+ if (bucket == QCOM_ICC_BUCKET_WAKE && qp->keep_alive)
+ rate = max(ICC_BUS_CLK_MIN_RATE, rate);
+
if (qp->bus_clk_rate[i] == rate)
continue;
@@ -453,6 +462,7 @@ int qnoc_probe(struct platform_device *pdev)
for (i = 0; i < qp->num_bus_clks; i++)
qp->bus_clks[i].id = bus_clocks[i];
+ qp->keep_alive = desc->keep_alive;
qp->type = desc->type;
qp->qos_offset = desc->qos_offset;
diff --git a/drivers/interconnect/qcom/icc-rpm.h b/drivers/interconnect/qcom/icc-rpm.h
index ee705edf19dd..d2c04c400cad 100644
--- a/drivers/interconnect/qcom/icc-rpm.h
+++ b/drivers/interconnect/qcom/icc-rpm.h
@@ -33,6 +33,7 @@ enum qcom_icc_type {
* @bus_clk_rate: bus clock rate in Hz
* @bus_clks: the clk_bulk_data table of bus clocks
* @intf_clks: a clk_bulk_data array of interface clocks
+ * @keep_alive: whether to always keep a minimum vote on the bus clocks
* @is_on: whether the bus is powered on
*/
struct qcom_icc_provider {
@@ -45,6 +46,7 @@ struct qcom_icc_provider {
u64 bus_clk_rate[NUM_BUS_CLKS];
struct clk_bulk_data bus_clks[NUM_BUS_CLKS];
struct clk_bulk_data *intf_clks;
+ bool keep_alive;
bool is_on;
};
@@ -102,6 +104,7 @@ struct qcom_icc_desc {
const char * const *bus_clocks;
const char * const *intf_clocks;
size_t num_intf_clocks;
+ bool keep_alive;
bool no_clk_scaling;
enum qcom_icc_type type;
const struct regmap_config *regmap_cfg;
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 06/22] interconnect: qcom: icc-rpm: Allow negative QoS offset
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (4 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 05/22] interconnect: qcom: icc-rpm: Introduce keep_alive Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 07/22] interconnect: qcom: Fold smd-rpm.h into icc-rpm.h Konrad Dybcio
` (15 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio
In some very very very very unfortunate cases, the correct offset of
the QoS registers will be.. negative. One such case is MSM8998, where
The DDR BWMON occupies what-would-be-the-BIMC-base which we usually
take into account with the register calculation, making the actual
BIMC node start at what-would-be-the-BIMC-base+0x300.
In order to keep the calculation code sane, the simplest - however
ugly it may be - solution is to allow the offset to be negative.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/icc-rpm.h | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/interconnect/qcom/icc-rpm.h b/drivers/interconnect/qcom/icc-rpm.h
index d2c04c400cad..ba840a436cc0 100644
--- a/drivers/interconnect/qcom/icc-rpm.h
+++ b/drivers/interconnect/qcom/icc-rpm.h
@@ -29,10 +29,10 @@ enum qcom_icc_type {
* @num_intf_clks: the total number of intf_clks clk_bulk_data entries
* @type: the ICC provider type
* @regmap: regmap for QoS registers read/write access
- * @qos_offset: offset to QoS registers
* @bus_clk_rate: bus clock rate in Hz
* @bus_clks: the clk_bulk_data table of bus clocks
* @intf_clks: a clk_bulk_data array of interface clocks
+ * @qos_offset: offset to QoS registers
* @keep_alive: whether to always keep a minimum vote on the bus clocks
* @is_on: whether the bus is powered on
*/
@@ -42,7 +42,7 @@ struct qcom_icc_provider {
int num_intf_clks;
enum qcom_icc_type type;
struct regmap *regmap;
- unsigned int qos_offset;
+ int qos_offset;
u64 bus_clk_rate[NUM_BUS_CLKS];
struct clk_bulk_data bus_clks[NUM_BUS_CLKS];
struct clk_bulk_data *intf_clks;
@@ -108,7 +108,7 @@ struct qcom_icc_desc {
bool no_clk_scaling;
enum qcom_icc_type type;
const struct regmap_config *regmap_cfg;
- unsigned int qos_offset;
+ int qos_offset;
};
/* Valid for all bus types */
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 07/22] interconnect: qcom: Fold smd-rpm.h into icc-rpm.h
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (5 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 06/22] interconnect: qcom: icc-rpm: Allow negative QoS offset Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 08/22] interconnect: qcom: smd-rpm: Add rpmcc handling skeleton code Konrad Dybcio
` (14 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Dmitry Baryshkov
smd-rpm.h is not very useful as-is and both files are always included
anyway.. Combine them.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/icc-rpm.c | 1 -
drivers/interconnect/qcom/icc-rpm.h | 5 +++++
drivers/interconnect/qcom/msm8916.c | 1 -
drivers/interconnect/qcom/msm8939.c | 1 -
drivers/interconnect/qcom/msm8974.c | 2 +-
drivers/interconnect/qcom/msm8996.c | 1 -
drivers/interconnect/qcom/qcm2290.c | 1 -
drivers/interconnect/qcom/qcs404.c | 1 -
drivers/interconnect/qcom/sdm660.c | 1 -
drivers/interconnect/qcom/smd-rpm.c | 2 +-
drivers/interconnect/qcom/smd-rpm.h | 15 ---------------
11 files changed, 7 insertions(+), 24 deletions(-)
diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
index 863e8ba1daa2..b8ecf9538ab9 100644
--- a/drivers/interconnect/qcom/icc-rpm.c
+++ b/drivers/interconnect/qcom/icc-rpm.c
@@ -14,7 +14,6 @@
#include <linux/regmap.h>
#include <linux/slab.h>
-#include "smd-rpm.h"
#include "icc-common.h"
#include "icc-rpm.h"
diff --git a/drivers/interconnect/qcom/icc-rpm.h b/drivers/interconnect/qcom/icc-rpm.h
index ba840a436cc0..9b4ea4e39b9f 100644
--- a/drivers/interconnect/qcom/icc-rpm.h
+++ b/drivers/interconnect/qcom/icc-rpm.h
@@ -6,6 +6,8 @@
#ifndef __DRIVERS_INTERCONNECT_QCOM_ICC_RPM_H
#define __DRIVERS_INTERCONNECT_QCOM_ICC_RPM_H
+#include <linux/soc/qcom/smd-rpm.h>
+
#include <dt-bindings/interconnect/qcom,icc.h>
#define RPM_BUS_MASTER_REQ 0x73616d62
@@ -121,4 +123,7 @@ enum qos_mode {
int qnoc_probe(struct platform_device *pdev);
int qnoc_remove(struct platform_device *pdev);
+bool qcom_icc_rpm_smd_available(void);
+int qcom_icc_rpm_smd_send(int ctx, int rsc_type, int id, u32 val);
+
#endif
diff --git a/drivers/interconnect/qcom/msm8916.c b/drivers/interconnect/qcom/msm8916.c
index 5c4ba2f37c8e..196b05879896 100644
--- a/drivers/interconnect/qcom/msm8916.c
+++ b/drivers/interconnect/qcom/msm8916.c
@@ -15,7 +15,6 @@
#include <dt-bindings/interconnect/qcom,msm8916.h>
-#include "smd-rpm.h"
#include "icc-rpm.h"
enum {
diff --git a/drivers/interconnect/qcom/msm8939.c b/drivers/interconnect/qcom/msm8939.c
index caf0aefad668..639566dce45a 100644
--- a/drivers/interconnect/qcom/msm8939.c
+++ b/drivers/interconnect/qcom/msm8939.c
@@ -16,7 +16,6 @@
#include <dt-bindings/interconnect/qcom,msm8939.h>
-#include "smd-rpm.h"
#include "icc-rpm.h"
enum {
diff --git a/drivers/interconnect/qcom/msm8974.c b/drivers/interconnect/qcom/msm8974.c
index 1828deaca443..968162213d40 100644
--- a/drivers/interconnect/qcom/msm8974.c
+++ b/drivers/interconnect/qcom/msm8974.c
@@ -38,7 +38,7 @@
#include <linux/platform_device.h>
#include <linux/slab.h>
-#include "smd-rpm.h"
+#include "icc-rpm.h"
enum {
MSM8974_BIMC_MAS_AMPSS_M0 = 1,
diff --git a/drivers/interconnect/qcom/msm8996.c b/drivers/interconnect/qcom/msm8996.c
index 20340fb62fe6..1f7e88a37acd 100644
--- a/drivers/interconnect/qcom/msm8996.c
+++ b/drivers/interconnect/qcom/msm8996.c
@@ -18,7 +18,6 @@
#include <dt-bindings/interconnect/qcom,msm8996.h>
#include "icc-rpm.h"
-#include "smd-rpm.h"
#include "msm8996.h"
static const char * const mm_intf_clocks[] = {
diff --git a/drivers/interconnect/qcom/qcm2290.c b/drivers/interconnect/qcom/qcm2290.c
index a29cdb4fac03..cb636e67a5a4 100644
--- a/drivers/interconnect/qcom/qcm2290.c
+++ b/drivers/interconnect/qcom/qcm2290.c
@@ -19,7 +19,6 @@
#include <linux/slab.h>
#include "icc-rpm.h"
-#include "smd-rpm.h"
enum {
QCM2290_MASTER_APPSS_PROC = 1,
diff --git a/drivers/interconnect/qcom/qcs404.c b/drivers/interconnect/qcom/qcs404.c
index fae155344332..938283ddd0e3 100644
--- a/drivers/interconnect/qcom/qcs404.c
+++ b/drivers/interconnect/qcom/qcs404.c
@@ -13,7 +13,6 @@
#include <linux/of_device.h>
-#include "smd-rpm.h"
#include "icc-rpm.h"
enum {
diff --git a/drivers/interconnect/qcom/sdm660.c b/drivers/interconnect/qcom/sdm660.c
index 7ffaf70d62d3..003fc7d110a7 100644
--- a/drivers/interconnect/qcom/sdm660.c
+++ b/drivers/interconnect/qcom/sdm660.c
@@ -17,7 +17,6 @@
#include <linux/slab.h>
#include "icc-rpm.h"
-#include "smd-rpm.h"
enum {
SDM660_MASTER_IPA = 1,
diff --git a/drivers/interconnect/qcom/smd-rpm.c b/drivers/interconnect/qcom/smd-rpm.c
index dc8ff8d133a9..b0183262ba66 100644
--- a/drivers/interconnect/qcom/smd-rpm.c
+++ b/drivers/interconnect/qcom/smd-rpm.c
@@ -13,7 +13,7 @@
#include <linux/platform_device.h>
#include <linux/soc/qcom/smd-rpm.h>
-#include "smd-rpm.h"
+#include "icc-rpm.h"
#define RPM_KEY_BW 0x00007762
diff --git a/drivers/interconnect/qcom/smd-rpm.h b/drivers/interconnect/qcom/smd-rpm.h
deleted file mode 100644
index ca9d0327b8ac..000000000000
--- a/drivers/interconnect/qcom/smd-rpm.h
+++ /dev/null
@@ -1,15 +0,0 @@
-/* SPDX-License-Identifier: GPL-2.0 */
-/*
- * Copyright (c) 2019, Linaro Ltd.
- * Author: Georgi Djakov <georgi.djakov@linaro.org>
- */
-
-#ifndef __DRIVERS_INTERCONNECT_QCOM_SMD_RPM_H
-#define __DRIVERS_INTERCONNECT_QCOM_SMD_RPM_H
-
-#include <linux/soc/qcom/smd-rpm.h>
-
-bool qcom_icc_rpm_smd_available(void);
-int qcom_icc_rpm_smd_send(int ctx, int rsc_type, int id, u32 val);
-
-#endif
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 08/22] interconnect: qcom: smd-rpm: Add rpmcc handling skeleton code
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (6 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 07/22] interconnect: qcom: Fold smd-rpm.h into icc-rpm.h Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 09/22] interconnect: qcom: Add missing headers in icc-rpm.h Konrad Dybcio
` (13 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio
Introduce qcom_icc_rpm_set_bus_rate() in preparation for handling RPM
clock resources within the interconnect framework. This lets us greatly
simplify all of the code handling, as setting the rate comes down to:
u32 rate_khz = max(clk.sleep_rate, clk.active_rate, clk_a.active_rate)
write_to_rpm(clock.description, rate_khz);
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/icc-rpm.h | 15 +++++++++++++++
drivers/interconnect/qcom/smd-rpm.c | 37 +++++++++++++++++++++++++++++++++++++
2 files changed, 52 insertions(+)
diff --git a/drivers/interconnect/qcom/icc-rpm.h b/drivers/interconnect/qcom/icc-rpm.h
index 9b4ea4e39b9f..aec192321411 100644
--- a/drivers/interconnect/qcom/icc-rpm.h
+++ b/drivers/interconnect/qcom/icc-rpm.h
@@ -22,6 +22,18 @@ enum qcom_icc_type {
QCOM_ICC_QNOC,
};
+/**
+ * struct rpm_clk_resource - RPM bus clock resource
+ * @resource_type: RPM resource type of the clock resource
+ * @clock_id: index of the clock resource of a specific resource type
+ * @branch: whether the resource represents a branch clock
+*/
+struct rpm_clk_resource {
+ u32 resource_type;
+ u32 clock_id;
+ bool branch;
+};
+
#define NUM_BUS_CLKS 2
/**
@@ -47,6 +59,7 @@ struct qcom_icc_provider {
int qos_offset;
u64 bus_clk_rate[NUM_BUS_CLKS];
struct clk_bulk_data bus_clks[NUM_BUS_CLKS];
+ const struct rpm_clk_resource *bus_clk_desc;
struct clk_bulk_data *intf_clks;
bool keep_alive;
bool is_on;
@@ -104,6 +117,7 @@ struct qcom_icc_desc {
struct qcom_icc_node * const *nodes;
size_t num_nodes;
const char * const *bus_clocks;
+ const struct rpm_clk_resource *bus_clk_desc;
const char * const *intf_clocks;
size_t num_intf_clocks;
bool keep_alive;
@@ -125,5 +139,6 @@ int qnoc_remove(struct platform_device *pdev);
bool qcom_icc_rpm_smd_available(void);
int qcom_icc_rpm_smd_send(int ctx, int rsc_type, int id, u32 val);
+int qcom_icc_rpm_set_bus_rate(const struct rpm_clk_resource *clk, u32 active_rate, u32 sleep_rate);
#endif
diff --git a/drivers/interconnect/qcom/smd-rpm.c b/drivers/interconnect/qcom/smd-rpm.c
index b0183262ba66..6c51e346b326 100644
--- a/drivers/interconnect/qcom/smd-rpm.c
+++ b/drivers/interconnect/qcom/smd-rpm.c
@@ -16,6 +16,7 @@
#include "icc-rpm.h"
#define RPM_KEY_BW 0x00007762
+#define QCOM_RPM_SMD_KEY_RATE 0x007a484b
static struct qcom_smd_rpm *icc_smd_rpm;
@@ -44,6 +45,38 @@ int qcom_icc_rpm_smd_send(int ctx, int rsc_type, int id, u32 val)
}
EXPORT_SYMBOL_GPL(qcom_icc_rpm_smd_send);
+int qcom_icc_rpm_set_bus_rate(const struct rpm_clk_resource *clk, u32 active_rate, u32 sleep_rate)
+{
+ struct clk_smd_rpm_req req = {
+ .key = cpu_to_le32(QCOM_RPM_SMD_KEY_RATE),
+ .nbytes = cpu_to_le32(sizeof(u32)),
+ };
+ int ret;
+
+ /* Branch clocks are only on/off */
+ if (clk->branch) {
+ active_rate = !!active_rate;
+ sleep_rate = !!sleep_rate;
+ }
+
+ req.value = cpu_to_le32(active_rate);
+ ret = qcom_rpm_smd_write(icc_smd_rpm,
+ QCOM_SMD_RPM_ACTIVE_STATE,
+ clk->resource_type,
+ clk->clock_id,
+ &req, sizeof(req));
+ if (ret)
+ return ret;
+
+ req.value = cpu_to_le32(sleep_rate);
+ return qcom_rpm_smd_write(icc_smd_rpm,
+ QCOM_SMD_RPM_SLEEP_STATE,
+ clk->resource_type,
+ clk->clock_id,
+ &req, sizeof(req));
+}
+EXPORT_SYMBOL_GPL(qcom_icc_rpm_set_bus_rate);
+
static int qcom_icc_rpm_smd_remove(struct platform_device *pdev)
{
icc_smd_rpm = NULL;
@@ -60,6 +93,10 @@ static int qcom_icc_rpm_smd_probe(struct platform_device *pdev)
return -ENODEV;
}
+ /* We need the clock driver to kick things off first to avoid ugly races */
+ if (!qcom_smd_rpm_scaling_available())
+ return -EPROBE_DEFER;
+
return 0;
}
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 09/22] interconnect: qcom: Add missing headers in icc-rpm.h
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (7 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 08/22] interconnect: qcom: smd-rpm: Add rpmcc handling skeleton code Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 10/22] interconnect: qcom: Define RPM bus clocks Konrad Dybcio
` (12 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Dmitry Baryshkov
Currently the header does not provide all the required dependencies.
Fix it.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/icc-rpm.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/interconnect/qcom/icc-rpm.h b/drivers/interconnect/qcom/icc-rpm.h
index aec192321411..057a1a8b3cb9 100644
--- a/drivers/interconnect/qcom/icc-rpm.h
+++ b/drivers/interconnect/qcom/icc-rpm.h
@@ -6,6 +6,8 @@
#ifndef __DRIVERS_INTERCONNECT_QCOM_ICC_RPM_H
#define __DRIVERS_INTERCONNECT_QCOM_ICC_RPM_H
+#include <linux/interconnect-provider.h>
+#include <linux/platform_device.h>
#include <linux/soc/qcom/smd-rpm.h>
#include <dt-bindings/interconnect/qcom,icc.h>
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 10/22] interconnect: qcom: Define RPM bus clocks
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (8 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 09/22] interconnect: qcom: Add missing headers in icc-rpm.h Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 11/22] interconnect: qcom: sdm660: Hook up RPM bus clk definitions Konrad Dybcio
` (11 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Dmitry Baryshkov
Add the definitions for RPM bus clocks that will be used by many
different platforms.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/Makefile | 2 +-
drivers/interconnect/qcom/icc-rpm-clocks.c | 66 ++++++++++++++++++++++++++++++
drivers/interconnect/qcom/icc-rpm.h | 13 ++++++
3 files changed, 80 insertions(+), 1 deletion(-)
diff --git a/drivers/interconnect/qcom/Makefile b/drivers/interconnect/qcom/Makefile
index ab988926433c..80d9d2da95d1 100644
--- a/drivers/interconnect/qcom/Makefile
+++ b/drivers/interconnect/qcom/Makefile
@@ -29,7 +29,7 @@ qnoc-sm8250-objs := sm8250.o
qnoc-sm8350-objs := sm8350.o
qnoc-sm8450-objs := sm8450.o
qnoc-sm8550-objs := sm8550.o
-icc-smd-rpm-objs := smd-rpm.o icc-rpm.o
+icc-smd-rpm-objs := smd-rpm.o icc-rpm.o icc-rpm-clocks.o
obj-$(CONFIG_INTERCONNECT_QCOM_BCM_VOTER) += icc-bcm-voter.o
obj-$(CONFIG_INTERCONNECT_QCOM_MSM8916) += qnoc-msm8916.o
diff --git a/drivers/interconnect/qcom/icc-rpm-clocks.c b/drivers/interconnect/qcom/icc-rpm-clocks.c
new file mode 100644
index 000000000000..a195bf32aff4
--- /dev/null
+++ b/drivers/interconnect/qcom/icc-rpm-clocks.c
@@ -0,0 +1,66 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2023 Linaro Ltd
+ */
+
+#include <linux/soc/qcom/smd-rpm.h>
+
+#include "icc-rpm.h"
+
+const struct rpm_clk_resource aggre1_clk = {
+ .resource_type = QCOM_SMD_RPM_AGGR_CLK,
+ .clock_id = 1,
+};
+
+const struct rpm_clk_resource aggre2_clk = {
+ .resource_type = QCOM_SMD_RPM_AGGR_CLK,
+ .clock_id = 2,
+};
+
+const struct rpm_clk_resource bimc_clk = {
+ .resource_type = QCOM_SMD_RPM_MEM_CLK,
+ .clock_id = 0,
+};
+
+const struct rpm_clk_resource bus_0_clk = {
+ .resource_type = QCOM_SMD_RPM_BUS_CLK,
+ .clock_id = 0,
+};
+
+const struct rpm_clk_resource bus_1_clk = {
+ .resource_type = QCOM_SMD_RPM_BUS_CLK,
+ .clock_id = 1,
+};
+
+const struct rpm_clk_resource bus_2_clk = {
+ .resource_type = QCOM_SMD_RPM_BUS_CLK,
+ .clock_id = 2,
+};
+
+const struct rpm_clk_resource mmaxi_0_clk = {
+ .resource_type = QCOM_SMD_RPM_MMAXI_CLK,
+ .clock_id = 0,
+};
+
+const struct rpm_clk_resource mmaxi_1_clk = {
+ .resource_type = QCOM_SMD_RPM_MMAXI_CLK,
+ .clock_id = 1,
+};
+
+const struct rpm_clk_resource qup_clk = {
+ .resource_type = QCOM_SMD_RPM_QUP_CLK,
+ .clock_id = 0,
+};
+
+/* Branch clocks */
+const struct rpm_clk_resource aggre1_branch_clk = {
+ .resource_type = QCOM_SMD_RPM_AGGR_CLK,
+ .clock_id = 1,
+ .branch = true,
+};
+
+const struct rpm_clk_resource aggre2_branch_clk = {
+ .resource_type = QCOM_SMD_RPM_AGGR_CLK,
+ .clock_id = 2,
+ .branch = true,
+};
diff --git a/drivers/interconnect/qcom/icc-rpm.h b/drivers/interconnect/qcom/icc-rpm.h
index 057a1a8b3cb9..99e34f684c85 100644
--- a/drivers/interconnect/qcom/icc-rpm.h
+++ b/drivers/interconnect/qcom/icc-rpm.h
@@ -136,6 +136,19 @@ enum qos_mode {
NOC_QOS_MODE_BYPASS,
};
+extern const struct rpm_clk_resource aggre1_clk;
+extern const struct rpm_clk_resource aggre2_clk;
+extern const struct rpm_clk_resource bimc_clk;
+extern const struct rpm_clk_resource bus_0_clk;
+extern const struct rpm_clk_resource bus_1_clk;
+extern const struct rpm_clk_resource bus_2_clk;
+extern const struct rpm_clk_resource mmaxi_0_clk;
+extern const struct rpm_clk_resource mmaxi_1_clk;
+extern const struct rpm_clk_resource qup_clk;
+
+extern const struct rpm_clk_resource aggre1_branch_clk;
+extern const struct rpm_clk_resource aggre2_branch_clk;
+
int qnoc_probe(struct platform_device *pdev);
int qnoc_remove(struct platform_device *pdev);
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 11/22] interconnect: qcom: sdm660: Hook up RPM bus clk definitions
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (9 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 10/22] interconnect: qcom: Define RPM bus clocks Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 12/22] interconnect: qcom: msm8996: " Konrad Dybcio
` (10 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Dmitry Baryshkov
Assign the necessary definitions to migrate to the new bus clock
handling mechanism.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/sdm660.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/drivers/interconnect/qcom/sdm660.c b/drivers/interconnect/qcom/sdm660.c
index 003fc7d110a7..5743ed680e8e 100644
--- a/drivers/interconnect/qcom/sdm660.c
+++ b/drivers/interconnect/qcom/sdm660.c
@@ -1511,6 +1511,7 @@ static const struct qcom_icc_desc sdm660_a2noc = {
.type = QCOM_ICC_NOC,
.nodes = sdm660_a2noc_nodes,
.num_nodes = ARRAY_SIZE(sdm660_a2noc_nodes),
+ .bus_clk_desc = &aggre2_clk,
.intf_clocks = a2noc_intf_clocks,
.num_intf_clocks = ARRAY_SIZE(a2noc_intf_clocks),
.regmap_cfg = &sdm660_a2noc_regmap_config,
@@ -1539,6 +1540,7 @@ static const struct qcom_icc_desc sdm660_bimc = {
.type = QCOM_ICC_BIMC,
.nodes = sdm660_bimc_nodes,
.num_nodes = ARRAY_SIZE(sdm660_bimc_nodes),
+ .bus_clk_desc = &bimc_clk,
.regmap_cfg = &sdm660_bimc_regmap_config,
};
@@ -1593,6 +1595,7 @@ static const struct qcom_icc_desc sdm660_cnoc = {
.type = QCOM_ICC_NOC,
.nodes = sdm660_cnoc_nodes,
.num_nodes = ARRAY_SIZE(sdm660_cnoc_nodes),
+ .bus_clk_desc = &bus_2_clk,
.regmap_cfg = &sdm660_cnoc_regmap_config,
};
@@ -1655,6 +1658,7 @@ static const struct qcom_icc_desc sdm660_mnoc = {
.type = QCOM_ICC_NOC,
.nodes = sdm660_mnoc_nodes,
.num_nodes = ARRAY_SIZE(sdm660_mnoc_nodes),
+ .bus_clk_desc = &mmaxi_0_clk,
.intf_clocks = mm_intf_clocks,
.num_intf_clocks = ARRAY_SIZE(mm_intf_clocks),
.regmap_cfg = &sdm660_mnoc_regmap_config,
@@ -1692,6 +1696,7 @@ static const struct qcom_icc_desc sdm660_snoc = {
.type = QCOM_ICC_NOC,
.nodes = sdm660_snoc_nodes,
.num_nodes = ARRAY_SIZE(sdm660_snoc_nodes),
+ .bus_clk_desc = &bus_1_clk,
.regmap_cfg = &sdm660_snoc_regmap_config,
};
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 12/22] interconnect: qcom: msm8996: Hook up RPM bus clk definitions
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (10 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 11/22] interconnect: qcom: sdm660: Hook up RPM bus clk definitions Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 13/22] interconnect: qcom: qcs404: " Konrad Dybcio
` (9 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio
Assign the necessary definitions to migrate to the new bus clock
handling mechanism.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/msm8996.c | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/drivers/interconnect/qcom/msm8996.c b/drivers/interconnect/qcom/msm8996.c
index 1f7e88a37acd..a596f4035d2e 100644
--- a/drivers/interconnect/qcom/msm8996.c
+++ b/drivers/interconnect/qcom/msm8996.c
@@ -1840,6 +1840,7 @@ static const struct qcom_icc_desc msm8996_a1noc = {
.type = QCOM_ICC_NOC,
.nodes = a1noc_nodes,
.num_nodes = ARRAY_SIZE(a1noc_nodes),
+ .bus_clk_desc = &aggre1_branch_clk,
.regmap_cfg = &msm8996_a1noc_regmap_config
};
@@ -1861,6 +1862,7 @@ static const struct qcom_icc_desc msm8996_a2noc = {
.type = QCOM_ICC_NOC,
.nodes = a2noc_nodes,
.num_nodes = ARRAY_SIZE(a2noc_nodes),
+ .bus_clk_desc = &aggre2_branch_clk,
.intf_clocks = a2noc_intf_clocks,
.num_intf_clocks = ARRAY_SIZE(a2noc_intf_clocks),
.regmap_cfg = &msm8996_a2noc_regmap_config
@@ -1889,6 +1891,7 @@ static const struct qcom_icc_desc msm8996_bimc = {
.type = QCOM_ICC_BIMC,
.nodes = bimc_nodes,
.num_nodes = ARRAY_SIZE(bimc_nodes),
+ .bus_clk_desc = &bimc_clk,
.regmap_cfg = &msm8996_bimc_regmap_config
};
@@ -1947,6 +1950,7 @@ static const struct qcom_icc_desc msm8996_cnoc = {
.type = QCOM_ICC_NOC,
.nodes = cnoc_nodes,
.num_nodes = ARRAY_SIZE(cnoc_nodes),
+ .bus_clk_desc = &bus_2_clk,
.regmap_cfg = &msm8996_cnoc_regmap_config
};
@@ -2000,6 +2004,7 @@ static const struct qcom_icc_desc msm8996_mnoc = {
.type = QCOM_ICC_NOC,
.nodes = mnoc_nodes,
.num_nodes = ARRAY_SIZE(mnoc_nodes),
+ .bus_clk_desc = &mmaxi_0_clk,
.intf_clocks = mm_intf_clocks,
.num_intf_clocks = ARRAY_SIZE(mm_intf_clocks),
.regmap_cfg = &msm8996_mnoc_regmap_config
@@ -2038,6 +2043,7 @@ static const struct qcom_icc_desc msm8996_pnoc = {
.type = QCOM_ICC_NOC,
.nodes = pnoc_nodes,
.num_nodes = ARRAY_SIZE(pnoc_nodes),
+ .bus_clk_desc = &bus_0_clk,
.regmap_cfg = &msm8996_pnoc_regmap_config
};
@@ -2082,6 +2088,7 @@ static const struct qcom_icc_desc msm8996_snoc = {
.type = QCOM_ICC_NOC,
.nodes = snoc_nodes,
.num_nodes = ARRAY_SIZE(snoc_nodes),
+ .bus_clk_desc = &bus_1_clk,
.regmap_cfg = &msm8996_snoc_regmap_config
};
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 13/22] interconnect: qcom: qcs404: Hook up RPM bus clk definitions
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (11 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 12/22] interconnect: qcom: msm8996: " Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 14/22] interconnect: qcom: msm8939: " Konrad Dybcio
` (8 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Dmitry Baryshkov
Assign the necessary definitions to migrate to the new bus clock
handling mechanism.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/qcs404.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drivers/interconnect/qcom/qcs404.c b/drivers/interconnect/qcom/qcs404.c
index 938283ddd0e3..9deb4d81104e 100644
--- a/drivers/interconnect/qcom/qcs404.c
+++ b/drivers/interconnect/qcom/qcs404.c
@@ -984,6 +984,7 @@ static struct qcom_icc_node * const qcs404_bimc_nodes[] = {
};
static const struct qcom_icc_desc qcs404_bimc = {
+ .bus_clk_desc = &bimc_clk,
.nodes = qcs404_bimc_nodes,
.num_nodes = ARRAY_SIZE(qcs404_bimc_nodes),
};
@@ -1038,6 +1039,7 @@ static struct qcom_icc_node * const qcs404_pcnoc_nodes[] = {
};
static const struct qcom_icc_desc qcs404_pcnoc = {
+ .bus_clk_desc = &bus_0_clk,
.nodes = qcs404_pcnoc_nodes,
.num_nodes = ARRAY_SIZE(qcs404_pcnoc_nodes),
};
@@ -1066,6 +1068,7 @@ static struct qcom_icc_node * const qcs404_snoc_nodes[] = {
};
static const struct qcom_icc_desc qcs404_snoc = {
+ .bus_clk_desc = &bus_1_clk,
.nodes = qcs404_snoc_nodes,
.num_nodes = ARRAY_SIZE(qcs404_snoc_nodes),
};
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 14/22] interconnect: qcom: msm8939: Hook up RPM bus clk definitions
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (12 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 13/22] interconnect: qcom: qcs404: " Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-10 12:02 ` Stephan Gerhold
2023-06-09 20:19 ` [PATCH v2 15/22] interconnect: qcom: msm8916: " Konrad Dybcio
` (7 subsequent siblings)
21 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Dmitry Baryshkov
Assign the necessary definitions to migrate to the new bus clock
handling mechanism.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/msm8939.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/drivers/interconnect/qcom/msm8939.c b/drivers/interconnect/qcom/msm8939.c
index 639566dce45a..94b11b590a8e 100644
--- a/drivers/interconnect/qcom/msm8939.c
+++ b/drivers/interconnect/qcom/msm8939.c
@@ -1284,6 +1284,7 @@ static const struct qcom_icc_desc msm8939_snoc = {
.type = QCOM_ICC_NOC,
.nodes = msm8939_snoc_nodes,
.num_nodes = ARRAY_SIZE(msm8939_snoc_nodes),
+ .bus_clk_desc = &bus_1_clk,
.regmap_cfg = &msm8939_snoc_regmap_config,
.qos_offset = 0x7000,
};
@@ -1304,6 +1305,7 @@ static const struct qcom_icc_desc msm8939_snoc_mm = {
.type = QCOM_ICC_NOC,
.nodes = msm8939_snoc_mm_nodes,
.num_nodes = ARRAY_SIZE(msm8939_snoc_mm_nodes),
+ .bus_clk_desc = &bus_2_clk,
.regmap_cfg = &msm8939_snoc_regmap_config,
.qos_offset = 0x7000,
};
@@ -1332,6 +1334,7 @@ static const struct qcom_icc_desc msm8939_bimc = {
.type = QCOM_ICC_BIMC,
.nodes = msm8939_bimc_nodes,
.num_nodes = ARRAY_SIZE(msm8939_bimc_nodes),
+ .bus_clk_desc = &bimc_clk,
.regmap_cfg = &msm8939_bimc_regmap_config,
.qos_offset = 0x8000,
};
@@ -1403,6 +1406,7 @@ static const struct qcom_icc_desc msm8939_pcnoc = {
.type = QCOM_ICC_NOC,
.nodes = msm8939_pcnoc_nodes,
.num_nodes = ARRAY_SIZE(msm8939_pcnoc_nodes),
+ .bus_clk_desc = &bus_0_clk,
.regmap_cfg = &msm8939_pcnoc_regmap_config,
.qos_offset = 0x7000,
};
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 15/22] interconnect: qcom: msm8916: Hook up RPM bus clk definitions
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (13 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 14/22] interconnect: qcom: msm8939: " Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-10 12:02 ` Stephan Gerhold
2023-06-09 20:19 ` [PATCH v2 16/22] interconnect: qcom: qcm2290: " Konrad Dybcio
` (6 subsequent siblings)
21 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Dmitry Baryshkov
Assign the necessary definitions to migrate to the new bus clock
handling mechanism.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/msm8916.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drivers/interconnect/qcom/msm8916.c b/drivers/interconnect/qcom/msm8916.c
index 196b05879896..be2a190a8b52 100644
--- a/drivers/interconnect/qcom/msm8916.c
+++ b/drivers/interconnect/qcom/msm8916.c
@@ -1231,6 +1231,7 @@ static const struct qcom_icc_desc msm8916_snoc = {
.type = QCOM_ICC_NOC,
.nodes = msm8916_snoc_nodes,
.num_nodes = ARRAY_SIZE(msm8916_snoc_nodes),
+ .bus_clk_desc = &bus_1_clk,
.regmap_cfg = &msm8916_snoc_regmap_config,
.qos_offset = 0x7000,
};
@@ -1259,6 +1260,7 @@ static const struct qcom_icc_desc msm8916_bimc = {
.type = QCOM_ICC_BIMC,
.nodes = msm8916_bimc_nodes,
.num_nodes = ARRAY_SIZE(msm8916_bimc_nodes),
+ .bus_clk_desc = &bimc_clk,
.regmap_cfg = &msm8916_bimc_regmap_config,
.qos_offset = 0x8000,
};
@@ -1328,6 +1330,7 @@ static const struct qcom_icc_desc msm8916_pcnoc = {
.type = QCOM_ICC_NOC,
.nodes = msm8916_pcnoc_nodes,
.num_nodes = ARRAY_SIZE(msm8916_pcnoc_nodes),
+ .bus_clk_desc = &bus_0_clk,
.regmap_cfg = &msm8916_pcnoc_regmap_config,
.qos_offset = 0x7000,
};
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 16/22] interconnect: qcom: qcm2290: Hook up RPM bus clk definitions
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (14 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 15/22] interconnect: qcom: msm8916: " Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 17/22] interconnect: qcom: icc-rpm: Control bus rpmcc from icc Konrad Dybcio
` (5 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Dmitry Baryshkov
Assign the necessary definitions to migrate to the new bus clock
handling mechanism.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/qcm2290.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/drivers/interconnect/qcom/qcm2290.c b/drivers/interconnect/qcom/qcm2290.c
index cb636e67a5a4..2c7a76fab83c 100644
--- a/drivers/interconnect/qcom/qcm2290.c
+++ b/drivers/interconnect/qcom/qcm2290.c
@@ -1196,6 +1196,7 @@ static const struct qcom_icc_desc qcm2290_bimc = {
.type = QCOM_ICC_BIMC,
.nodes = qcm2290_bimc_nodes,
.num_nodes = ARRAY_SIZE(qcm2290_bimc_nodes),
+ .bus_clk_desc = &bimc_clk,
.regmap_cfg = &qcm2290_bimc_regmap_config,
/* M_REG_BASE() in vendor msm_bus_bimc_adhoc driver */
.qos_offset = 0x8000,
@@ -1251,6 +1252,7 @@ static const struct qcom_icc_desc qcm2290_cnoc = {
.type = QCOM_ICC_NOC,
.nodes = qcm2290_cnoc_nodes,
.num_nodes = ARRAY_SIZE(qcm2290_cnoc_nodes),
+ .bus_clk_desc = &bus_1_clk,
.regmap_cfg = &qcm2290_cnoc_regmap_config,
};
@@ -1292,6 +1294,7 @@ static const struct qcom_icc_desc qcm2290_snoc = {
.type = QCOM_ICC_QNOC,
.nodes = qcm2290_snoc_nodes,
.num_nodes = ARRAY_SIZE(qcm2290_snoc_nodes),
+ .bus_clk_desc = &bus_2_clk,
.regmap_cfg = &qcm2290_snoc_regmap_config,
/* Vendor DT node fab-sys_noc property 'qcom,base-offset' */
.qos_offset = 0x15000,
@@ -1306,6 +1309,7 @@ static const struct qcom_icc_desc qcm2290_qup_virt = {
.type = QCOM_ICC_QNOC,
.nodes = qcm2290_qup_virt_nodes,
.num_nodes = ARRAY_SIZE(qcm2290_qup_virt_nodes),
+ .bus_clk_desc = &qup_clk,
};
static struct qcom_icc_node * const qcm2290_mmnrt_virt_nodes[] = {
@@ -1319,6 +1323,7 @@ static const struct qcom_icc_desc qcm2290_mmnrt_virt = {
.type = QCOM_ICC_QNOC,
.nodes = qcm2290_mmnrt_virt_nodes,
.num_nodes = ARRAY_SIZE(qcm2290_mmnrt_virt_nodes),
+ .bus_clk_desc = &mmaxi_0_clk,
.regmap_cfg = &qcm2290_snoc_regmap_config,
.qos_offset = 0x15000,
};
@@ -1333,6 +1338,7 @@ static const struct qcom_icc_desc qcm2290_mmrt_virt = {
.type = QCOM_ICC_QNOC,
.nodes = qcm2290_mmrt_virt_nodes,
.num_nodes = ARRAY_SIZE(qcm2290_mmrt_virt_nodes),
+ .bus_clk_desc = &mmaxi_1_clk,
.regmap_cfg = &qcm2290_snoc_regmap_config,
.qos_offset = 0x15000,
};
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 17/22] interconnect: qcom: icc-rpm: Control bus rpmcc from icc
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (15 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 16/22] interconnect: qcom: qcm2290: " Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-10 11:58 ` Stephan Gerhold
2023-06-09 20:19 ` [PATCH v2 18/22] clk: qcom: smd-rpm: Separate out interconnect bus clocks Konrad Dybcio
` (4 subsequent siblings)
21 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio
The sole purpose of bus clocks that were previously registered with
rpmcc was to convey the aggregated bandwidth to RPM. There's no good
reason to keep them outside the interconnect framework, as it only
adds to the plentiful complexity.
Add the required code to handle these clocks from within SMD RPM ICC.
RPM-owned bus clocks are no longer considered a thing, but sadly we
have to allow for the existence of HLOS-owned bus clocks, as some
(mostly older) SoCs (ab)use these for bus scaling (e.g. MSM8998 and
&mmcc AHB_CLK_SRC).
This in turn is trivially solved with a single *clk, which is filled
and used iff qp.bus_clk_desc is absent and we have a "bus" clock-names
entry in the DT node.
This change should(tm) be fully compatible with all sorts of old
Device Trees as far as the interconnect functionality goes (modulo
abusing bus clock handles, but that's a mistake in and of itself).
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/icc-rpm.c | 116 +++++++++++++++++++-----------------
drivers/interconnect/qcom/icc-rpm.h | 13 ++--
drivers/interconnect/qcom/msm8996.c | 1 -
drivers/interconnect/qcom/sdm660.c | 1 -
4 files changed, 66 insertions(+), 65 deletions(-)
diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
index b8ecf9538ab9..6d40815c5401 100644
--- a/drivers/interconnect/qcom/icc-rpm.c
+++ b/drivers/interconnect/qcom/icc-rpm.c
@@ -49,7 +49,7 @@
#define NOC_QOS_MODE_FIXED_VAL 0x0
#define NOC_QOS_MODE_BYPASS_VAL 0x2
-#define ICC_BUS_CLK_MIN_RATE 19200000ULL
+#define ICC_BUS_CLK_MIN_RATE 19200ULL /* kHz */
static int qcom_icc_set_qnoc_qos(struct icc_node *src)
{
@@ -338,11 +338,10 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
struct qcom_icc_node *src_qn = NULL, *dst_qn = NULL;
struct icc_provider *provider;
u64 sum_bw;
- u64 rate;
+ u64 active_rate, sleep_rate;
u64 agg_avg[QCOM_ICC_NUM_BUCKETS], agg_peak[QCOM_ICC_NUM_BUCKETS];
u64 max_agg_avg;
- int ret, i;
- int bucket;
+ int ret;
src_qn = src->data;
if (dst)
@@ -364,49 +363,50 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
return ret;
}
- for (i = 0; i < qp->num_bus_clks; i++) {
- /*
- * Use WAKE bucket for active clock, otherwise, use SLEEP bucket
- * for other clocks. If a platform doesn't set interconnect
- * path tags, by default use sleep bucket for all clocks.
- *
- * Note, AMC bucket is not supported yet.
- */
- if (!strcmp(qp->bus_clks[i].id, "bus_a"))
- bucket = QCOM_ICC_BUCKET_WAKE;
- else
- bucket = QCOM_ICC_BUCKET_SLEEP;
-
- rate = icc_units_to_bps(max(agg_avg[bucket], agg_peak[bucket]));
- do_div(rate, src_qn->buswidth);
- rate = min_t(u64, rate, LONG_MAX);
-
- /*
- * Downstream checks whether the requested rate is zero, but it makes little sense
- * to vote for a value that's below the lower threshold, so let's not do so.
- */
- if (bucket == QCOM_ICC_BUCKET_WAKE && qp->keep_alive)
- rate = max(ICC_BUS_CLK_MIN_RATE, rate);
-
- if (qp->bus_clk_rate[i] == rate)
- continue;
-
- ret = clk_set_rate(qp->bus_clks[i].clk, rate);
- if (ret) {
- pr_err("%s clk_set_rate error: %d\n",
- qp->bus_clks[i].id, ret);
+ /* Some providers don't have a bus clock to scale */
+ if (!qp->bus_clk_desc && !qp->bus_clk)
+ return 0;
+
+ /* Intentionally keep the rates in kHz as that's what RPM accepts */
+ active_rate = max(agg_avg[QCOM_SMD_RPM_ACTIVE_STATE],
+ agg_peak[QCOM_SMD_RPM_ACTIVE_STATE]);
+ do_div(active_rate, src_qn->buswidth);
+
+ sleep_rate = max(agg_avg[QCOM_SMD_RPM_SLEEP_STATE],
+ agg_peak[QCOM_SMD_RPM_SLEEP_STATE]);
+ do_div(sleep_rate, src_qn->buswidth);
+
+ /*
+ * Downstream checks whether the requested rate is zero, but it makes little sense
+ * to vote for a value that's below the lower threshold, so let's not do so.
+ */
+ if (qp->keep_alive)
+ active_rate = max(ICC_BUS_CLK_MIN_RATE, active_rate);
+
+ /* Some providers have a non-RPM-owned bus clock - convert kHz->Hz for the CCF */
+ if (qp->bus_clk)
+ return clk_set_rate(qp->bus_clk, 1000ULL * max(active_rate, sleep_rate));
+
+ /* RPM only accepts <=INT_MAX rates */
+ active_rate = min_t(u32, active_rate, INT_MAX);
+ sleep_rate = min_t(u32, sleep_rate, INT_MAX);
+
+ if ((active_rate != qp->bus_clk_rate[QCOM_SMD_RPM_ACTIVE_STATE]) ||
+ (sleep_rate != qp->bus_clk_rate[QCOM_SMD_RPM_SLEEP_STATE])) {
+ ret = qcom_icc_rpm_set_bus_rate(qp->bus_clk_desc,
+ active_rate,
+ sleep_rate);
+ if (ret)
return ret;
- }
- qp->bus_clk_rate[i] = rate;
}
+ /* Cache the rate after we've successfully commited it to RPM */
+ qp->bus_clk_rate[QCOM_SMD_RPM_ACTIVE_STATE] = active_rate;
+ qp->bus_clk_rate[QCOM_SMD_RPM_SLEEP_STATE] = sleep_rate;
+
return 0;
}
-static const char * const bus_clocks[] = {
- "bus", "bus_a",
-};
-
int qnoc_probe(struct platform_device *pdev)
{
struct device *dev = &pdev->dev;
@@ -448,6 +448,18 @@ int qnoc_probe(struct platform_device *pdev)
if (!qp->intf_clks)
return -ENOMEM;
+ if (desc->bus_clk_desc) {
+ qp->bus_clk_desc = devm_kzalloc(dev, sizeof(*qp->bus_clk_desc),
+ GFP_KERNEL);
+ if (!qp->bus_clk_desc)
+ return -ENOMEM;
+
+ qp->bus_clk_desc = desc->bus_clk_desc;
+ } else if (!IS_ERR(devm_clk_get(dev, "bus"))) {
+ /* Some older SoCs may have a single non-RPM-owned bus clock. */
+ qp->bus_clk = devm_clk_get(dev, "bus");
+ }
+
data = devm_kzalloc(dev, struct_size(data, nodes, num_nodes),
GFP_KERNEL);
if (!data)
@@ -457,10 +469,6 @@ int qnoc_probe(struct platform_device *pdev)
for (i = 0; i < cd_num; i++)
qp->intf_clks[i].id = cds[i];
- qp->num_bus_clks = desc->no_clk_scaling ? 0 : NUM_BUS_CLKS;
- for (i = 0; i < qp->num_bus_clks; i++)
- qp->bus_clks[i].id = bus_clocks[i];
-
qp->keep_alive = desc->keep_alive;
qp->type = desc->type;
qp->qos_offset = desc->qos_offset;
@@ -490,13 +498,11 @@ int qnoc_probe(struct platform_device *pdev)
}
regmap_done:
- ret = devm_clk_bulk_get(dev, qp->num_bus_clks, qp->bus_clks);
- if (ret)
- return ret;
-
- ret = clk_bulk_prepare_enable(qp->num_bus_clks, qp->bus_clks);
- if (ret)
- return ret;
+ if (qp->bus_clk) {
+ ret = clk_prepare_enable(qp->bus_clk);
+ if (ret)
+ return ret;
+ }
ret = devm_clk_bulk_get(dev, qp->num_intf_clks, qp->intf_clks);
if (ret)
@@ -566,7 +572,8 @@ int qnoc_probe(struct platform_device *pdev)
icc_provider_deregister(provider);
err_remove_nodes:
icc_nodes_remove(provider);
- clk_bulk_disable_unprepare(qp->num_bus_clks, qp->bus_clks);
+ if (qp->bus_clk)
+ clk_disable_unprepare(qp->bus_clk);
return ret;
}
@@ -578,7 +585,8 @@ int qnoc_remove(struct platform_device *pdev)
icc_provider_deregister(&qp->provider);
icc_nodes_remove(&qp->provider);
- clk_bulk_disable_unprepare(qp->num_bus_clks, qp->bus_clks);
+ if (qp->bus_clk)
+ clk_disable_unprepare(qp->bus_clk);
return 0;
}
diff --git a/drivers/interconnect/qcom/icc-rpm.h b/drivers/interconnect/qcom/icc-rpm.h
index 99e34f684c85..11e7d503e4d0 100644
--- a/drivers/interconnect/qcom/icc-rpm.h
+++ b/drivers/interconnect/qcom/icc-rpm.h
@@ -36,32 +36,29 @@ struct rpm_clk_resource {
bool branch;
};
-#define NUM_BUS_CLKS 2
-
/**
* struct qcom_icc_provider - Qualcomm specific interconnect provider
* @provider: generic interconnect provider
- * @num_bus_clks: the total number of bus_clks clk_bulk_data entries (0 or 2)
* @num_intf_clks: the total number of intf_clks clk_bulk_data entries
* @type: the ICC provider type
* @regmap: regmap for QoS registers read/write access
* @bus_clk_rate: bus clock rate in Hz
- * @bus_clks: the clk_bulk_data table of bus clocks
+ * @bus_clk_desc: a pointer to a rpm_clk_resource description of bus clocks
* @intf_clks: a clk_bulk_data array of interface clocks
+ * @bus_clk: a pointer to a HLOS-owned bus clock
* @qos_offset: offset to QoS registers
* @keep_alive: whether to always keep a minimum vote on the bus clocks
* @is_on: whether the bus is powered on
*/
struct qcom_icc_provider {
struct icc_provider provider;
- int num_bus_clks;
int num_intf_clks;
enum qcom_icc_type type;
struct regmap *regmap;
int qos_offset;
- u64 bus_clk_rate[NUM_BUS_CLKS];
- struct clk_bulk_data bus_clks[NUM_BUS_CLKS];
+ u32 bus_clk_rate[QCOM_SMD_RPM_STATE_NUM];
const struct rpm_clk_resource *bus_clk_desc;
+ struct clk *bus_clk;
struct clk_bulk_data *intf_clks;
bool keep_alive;
bool is_on;
@@ -118,12 +115,10 @@ struct qcom_icc_node {
struct qcom_icc_desc {
struct qcom_icc_node * const *nodes;
size_t num_nodes;
- const char * const *bus_clocks;
const struct rpm_clk_resource *bus_clk_desc;
const char * const *intf_clocks;
size_t num_intf_clocks;
bool keep_alive;
- bool no_clk_scaling;
enum qcom_icc_type type;
const struct regmap_config *regmap_cfg;
int qos_offset;
diff --git a/drivers/interconnect/qcom/msm8996.c b/drivers/interconnect/qcom/msm8996.c
index a596f4035d2e..8081b3cb1025 100644
--- a/drivers/interconnect/qcom/msm8996.c
+++ b/drivers/interconnect/qcom/msm8996.c
@@ -1818,7 +1818,6 @@ static const struct qcom_icc_desc msm8996_a0noc = {
.num_nodes = ARRAY_SIZE(a0noc_nodes),
.intf_clocks = a0noc_intf_clocks,
.num_intf_clocks = ARRAY_SIZE(a0noc_intf_clocks),
- .no_clk_scaling = true,
.regmap_cfg = &msm8996_a0noc_regmap_config
};
diff --git a/drivers/interconnect/qcom/sdm660.c b/drivers/interconnect/qcom/sdm660.c
index 5743ed680e8e..211fa1fa569c 100644
--- a/drivers/interconnect/qcom/sdm660.c
+++ b/drivers/interconnect/qcom/sdm660.c
@@ -1618,7 +1618,6 @@ static const struct qcom_icc_desc sdm660_gnoc = {
.nodes = sdm660_gnoc_nodes,
.num_nodes = ARRAY_SIZE(sdm660_gnoc_nodes),
.regmap_cfg = &sdm660_gnoc_regmap_config,
- .no_clk_scaling = true,
};
static struct qcom_icc_node * const sdm660_mnoc_nodes[] = {
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 18/22] clk: qcom: smd-rpm: Separate out interconnect bus clocks
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (16 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 17/22] interconnect: qcom: icc-rpm: Control bus rpmcc from icc Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 19/22] interconnect: qcom: icc-rpm: Fix bucket number Konrad Dybcio
` (3 subsequent siblings)
21 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio
The interconnect bus clocks are now handled within the ICC framework.
They still however need to get a kickstart *before* we call
clk_smd_rpm_enable_scaling(), or RPM will assume that they should all
be running at 0 kHz and the system will inevitably die.
Separate them out to ensure such a kickstart can still take place.
As a happy accident, the file got smaller:
Total: Before=41951, After=41555, chg -0.94%
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/clk/qcom/clk-smd-rpm.c | 278 +++++++++++++++++------------------------
1 file changed, 115 insertions(+), 163 deletions(-)
diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
index 482fe30ee6f0..f8f5977408e0 100644
--- a/drivers/clk/qcom/clk-smd-rpm.c
+++ b/drivers/clk/qcom/clk-smd-rpm.c
@@ -168,6 +168,14 @@ struct clk_smd_rpm {
struct rpm_smd_clk_desc {
struct clk_smd_rpm **clks;
size_t num_clks;
+
+ /*
+ * Interconnect clocks are managed by the icc framework, this driver
+ * only kickstarts them so that they don't get gated between
+ * clk_smd_rpm_enable_scaling() and interconnect driver initialization.
+ */
+ struct clk_smd_rpm **icc_clks;
+ size_t num_icc_clks;
bool scaling_before_handover;
};
@@ -507,13 +515,69 @@ DEFINE_CLK_SMD_RPM_XO_BUFFER(div_clk1, 11, 19200000);
DEFINE_CLK_SMD_RPM_XO_BUFFER(div_clk2, 12, 19200000);
DEFINE_CLK_SMD_RPM_XO_BUFFER(div_clk3, 13, 19200000);
+static struct clk_smd_rpm *bimc_pcnoc_icc_clks[] = {
+ &clk_smd_rpm_bimc_clk,
+ &clk_smd_rpm_bus_0_pcnoc_clk,
+};
+
+static struct clk_smd_rpm *bimc_pcnoc_snoc_icc_clks[] = {
+ &clk_smd_rpm_bimc_clk,
+ &clk_smd_rpm_bus_0_pcnoc_clk,
+ &clk_smd_rpm_bus_1_snoc_clk,
+};
+
+static struct clk_smd_rpm *bimc_pcnoc_snoc_smmnoc_icc_clks[] = {
+ &clk_smd_rpm_bimc_clk,
+ &clk_smd_rpm_bus_0_pcnoc_clk,
+ &clk_smd_rpm_bus_1_snoc_clk,
+ &clk_smd_rpm_bus_2_sysmmnoc_clk,
+};
+
+static struct clk_smd_rpm *bimc_pcnoc_snoc_cnoc_ocmem_icc_clks[] = {
+ &clk_smd_rpm_bimc_clk,
+ &clk_smd_rpm_bus_0_pcnoc_clk,
+ &clk_smd_rpm_bus_1_snoc_clk,
+ &clk_smd_rpm_bus_2_cnoc_clk,
+ &clk_smd_rpm_ocmemgx_clk,
+};
+
+static struct clk_smd_rpm *msm8996_icc_clks[] = {
+ &clk_smd_rpm_bimc_clk,
+ &clk_smd_rpm_branch_aggre1_noc_clk,
+ &clk_smd_rpm_branch_aggre2_noc_clk,
+ &clk_smd_rpm_bus_0_pcnoc_clk,
+ &clk_smd_rpm_bus_1_snoc_clk,
+ &clk_smd_rpm_bus_2_cnoc_clk,
+ &clk_smd_rpm_mmssnoc_axi_rpm_clk,
+};
+
+static struct clk_smd_rpm *msm8998_icc_clks[] = {
+ &clk_smd_rpm_aggre1_noc_clk,
+ &clk_smd_rpm_aggre2_noc_clk,
+ &clk_smd_rpm_bimc_clk,
+ &clk_smd_rpm_bus_1_snoc_clk,
+ &clk_smd_rpm_bus_2_cnoc_clk,
+ &clk_smd_rpm_mmssnoc_axi_rpm_clk,
+};
+
+static struct clk_smd_rpm *sdm660_icc_clks[] = {
+ &clk_smd_rpm_aggre2_noc_clk,
+ &clk_smd_rpm_bimc_clk,
+ &clk_smd_rpm_bus_1_snoc_clk,
+ &clk_smd_rpm_bus_2_cnoc_clk,
+ &clk_smd_rpm_mmssnoc_axi_rpm_clk,
+};
+
+static struct clk_smd_rpm *sm_qnoc_icc_clks[] = {
+ &clk_smd_rpm_bimc_clk,
+ &clk_smd_rpm_bus_1_cnoc_clk,
+ &clk_smd_rpm_mmnrt_clk,
+ &clk_smd_rpm_mmrt_clk,
+ &clk_smd_rpm_qup_clk,
+ &clk_smd_rpm_bus_2_snoc_clk,
+};
+
static struct clk_smd_rpm *msm8909_clks[] = {
- [RPM_SMD_PCNOC_CLK] = &clk_smd_rpm_bus_0_pcnoc_clk,
- [RPM_SMD_PCNOC_A_CLK] = &clk_smd_rpm_bus_0_pcnoc_a_clk,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_1_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_1_snoc_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
[RPM_SMD_QPIC_CLK] = &clk_smd_rpm_qpic_clk,
[RPM_SMD_QPIC_CLK_A] = &clk_smd_rpm_qpic_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_qdss_clk,
@@ -539,15 +603,11 @@ static struct clk_smd_rpm *msm8909_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_msm8909 = {
.clks = msm8909_clks,
.num_clks = ARRAY_SIZE(msm8909_clks),
+ .icc_clks = bimc_pcnoc_snoc_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(bimc_pcnoc_snoc_icc_clks),
};
static struct clk_smd_rpm *msm8916_clks[] = {
- [RPM_SMD_PCNOC_CLK] = &clk_smd_rpm_bus_0_pcnoc_clk,
- [RPM_SMD_PCNOC_A_CLK] = &clk_smd_rpm_bus_0_pcnoc_a_clk,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_1_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_1_snoc_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_qdss_clk,
[RPM_SMD_QDSS_A_CLK] = &clk_smd_rpm_qdss_a_clk,
[RPM_SMD_BB_CLK1] = &clk_smd_rpm_bb_clk1,
@@ -571,21 +631,15 @@ static struct clk_smd_rpm *msm8916_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_msm8916 = {
.clks = msm8916_clks,
.num_clks = ARRAY_SIZE(msm8916_clks),
+ .icc_clks = bimc_pcnoc_snoc_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(bimc_pcnoc_snoc_icc_clks),
};
static struct clk_smd_rpm *msm8917_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_PNOC_CLK] = &clk_smd_rpm_bus_0_pcnoc_clk,
- [RPM_SMD_PNOC_A_CLK] = &clk_smd_rpm_bus_0_pcnoc_a_clk,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_1_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_1_snoc_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
[RPM_SMD_BIMC_GPU_CLK] = &clk_smd_rpm_bimc_gpu_clk,
[RPM_SMD_BIMC_GPU_A_CLK] = &clk_smd_rpm_bimc_gpu_a_clk,
- [RPM_SMD_SYSMMNOC_CLK] = &clk_smd_rpm_bus_2_sysmmnoc_clk,
- [RPM_SMD_SYSMMNOC_A_CLK] = &clk_smd_rpm_bus_2_sysmmnoc_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_qdss_clk,
[RPM_SMD_QDSS_A_CLK] = &clk_smd_rpm_qdss_a_clk,
[RPM_SMD_BB_CLK1] = &clk_smd_rpm_bb_clk1,
@@ -605,19 +659,13 @@ static struct clk_smd_rpm *msm8917_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_msm8917 = {
.clks = msm8917_clks,
.num_clks = ARRAY_SIZE(msm8917_clks),
+ .icc_clks = bimc_pcnoc_snoc_smmnoc_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(bimc_pcnoc_snoc_smmnoc_icc_clks),
};
static struct clk_smd_rpm *msm8936_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_PCNOC_CLK] = &clk_smd_rpm_bus_0_pcnoc_clk,
- [RPM_SMD_PCNOC_A_CLK] = &clk_smd_rpm_bus_0_pcnoc_a_clk,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_1_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_1_snoc_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
- [RPM_SMD_SYSMMNOC_CLK] = &clk_smd_rpm_bus_2_sysmmnoc_clk,
- [RPM_SMD_SYSMMNOC_A_CLK] = &clk_smd_rpm_bus_2_sysmmnoc_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_qdss_clk,
[RPM_SMD_QDSS_A_CLK] = &clk_smd_rpm_qdss_a_clk,
[RPM_SMD_BB_CLK1] = &clk_smd_rpm_bb_clk1,
@@ -641,25 +689,17 @@ static struct clk_smd_rpm *msm8936_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_msm8936 = {
.clks = msm8936_clks,
.num_clks = ARRAY_SIZE(msm8936_clks),
+ .icc_clks = bimc_pcnoc_snoc_smmnoc_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(bimc_pcnoc_snoc_smmnoc_icc_clks),
};
static struct clk_smd_rpm *msm8974_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_PNOC_CLK] = &clk_smd_rpm_bus_0_pcnoc_clk,
- [RPM_SMD_PNOC_A_CLK] = &clk_smd_rpm_bus_0_pcnoc_a_clk,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_1_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_1_snoc_a_clk,
- [RPM_SMD_CNOC_CLK] = &clk_smd_rpm_bus_2_cnoc_clk,
- [RPM_SMD_CNOC_A_CLK] = &clk_smd_rpm_bus_2_cnoc_a_clk,
[RPM_SMD_MMSSNOC_AHB_CLK] = &clk_smd_rpm_bus_3_mmssnoc_ahb_clk,
[RPM_SMD_MMSSNOC_AHB_A_CLK] = &clk_smd_rpm_bus_3_mmssnoc_ahb_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
[RPM_SMD_GFX3D_CLK_SRC] = &clk_smd_rpm_gfx3d_clk_src,
[RPM_SMD_GFX3D_A_CLK_SRC] = &clk_smd_rpm_gfx3d_a_clk_src,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
- [RPM_SMD_OCMEMGX_CLK] = &clk_smd_rpm_ocmemgx_clk,
- [RPM_SMD_OCMEMGX_A_CLK] = &clk_smd_rpm_ocmemgx_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_qdss_clk,
[RPM_SMD_QDSS_A_CLK] = &clk_smd_rpm_qdss_a_clk,
[RPM_SMD_CXO_D0] = &clk_smd_rpm_cxo_d0,
@@ -693,20 +733,14 @@ static struct clk_smd_rpm *msm8974_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_msm8974 = {
.clks = msm8974_clks,
.num_clks = ARRAY_SIZE(msm8974_clks),
+ .icc_clks = bimc_pcnoc_snoc_cnoc_ocmem_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(bimc_pcnoc_snoc_cnoc_ocmem_icc_clks),
.scaling_before_handover = true,
};
static struct clk_smd_rpm *msm8976_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_PCNOC_CLK] = &clk_smd_rpm_bus_0_pcnoc_clk,
- [RPM_SMD_PCNOC_A_CLK] = &clk_smd_rpm_bus_0_pcnoc_a_clk,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_1_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_1_snoc_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
- [RPM_SMD_SYSMMNOC_CLK] = &clk_smd_rpm_bus_2_sysmmnoc_clk,
- [RPM_SMD_SYSMMNOC_A_CLK] = &clk_smd_rpm_bus_2_sysmmnoc_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_qdss_clk,
[RPM_SMD_QDSS_A_CLK] = &clk_smd_rpm_qdss_a_clk,
[RPM_SMD_BB_CLK1] = &clk_smd_rpm_bb_clk1,
@@ -727,24 +761,15 @@ static struct clk_smd_rpm *msm8976_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_msm8976 = {
.clks = msm8976_clks,
- .num_clks = ARRAY_SIZE(msm8976_clks),
+ .icc_clks = bimc_pcnoc_snoc_smmnoc_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(bimc_pcnoc_snoc_smmnoc_icc_clks),
};
static struct clk_smd_rpm *msm8992_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_PNOC_CLK] = &clk_smd_rpm_bus_0_pcnoc_clk,
- [RPM_SMD_PNOC_A_CLK] = &clk_smd_rpm_bus_0_pcnoc_a_clk,
- [RPM_SMD_OCMEMGX_CLK] = &clk_smd_rpm_ocmemgx_clk,
- [RPM_SMD_OCMEMGX_A_CLK] = &clk_smd_rpm_ocmemgx_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
- [RPM_SMD_CNOC_CLK] = &clk_smd_rpm_bus_2_cnoc_clk,
- [RPM_SMD_CNOC_A_CLK] = &clk_smd_rpm_bus_2_cnoc_a_clk,
[RPM_SMD_GFX3D_CLK_SRC] = &clk_smd_rpm_gfx3d_clk_src,
[RPM_SMD_GFX3D_A_CLK_SRC] = &clk_smd_rpm_gfx3d_a_clk_src,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_1_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_1_snoc_a_clk,
[RPM_SMD_BB_CLK1] = &clk_smd_rpm_bb_clk1,
[RPM_SMD_BB_CLK1_A] = &clk_smd_rpm_bb_clk1_a,
[RPM_SMD_BB_CLK1_PIN] = &clk_smd_rpm_bb_clk1_pin,
@@ -786,23 +811,15 @@ static struct clk_smd_rpm *msm8992_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_msm8992 = {
.clks = msm8992_clks,
.num_clks = ARRAY_SIZE(msm8992_clks),
+ .icc_clks = bimc_pcnoc_snoc_cnoc_ocmem_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(bimc_pcnoc_snoc_cnoc_ocmem_icc_clks),
};
static struct clk_smd_rpm *msm8994_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_PNOC_CLK] = &clk_smd_rpm_bus_0_pcnoc_clk,
- [RPM_SMD_PNOC_A_CLK] = &clk_smd_rpm_bus_0_pcnoc_a_clk,
- [RPM_SMD_OCMEMGX_CLK] = &clk_smd_rpm_ocmemgx_clk,
- [RPM_SMD_OCMEMGX_A_CLK] = &clk_smd_rpm_ocmemgx_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
- [RPM_SMD_CNOC_CLK] = &clk_smd_rpm_bus_2_cnoc_clk,
- [RPM_SMD_CNOC_A_CLK] = &clk_smd_rpm_bus_2_cnoc_a_clk,
[RPM_SMD_GFX3D_CLK_SRC] = &clk_smd_rpm_gfx3d_clk_src,
[RPM_SMD_GFX3D_A_CLK_SRC] = &clk_smd_rpm_gfx3d_a_clk_src,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_1_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_1_snoc_a_clk,
[RPM_SMD_BB_CLK1] = &clk_smd_rpm_bb_clk1,
[RPM_SMD_BB_CLK1_A] = &clk_smd_rpm_bb_clk1_a,
[RPM_SMD_BB_CLK1_PIN] = &clk_smd_rpm_bb_clk1_pin,
@@ -846,29 +863,17 @@ static struct clk_smd_rpm *msm8994_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_msm8994 = {
.clks = msm8994_clks,
.num_clks = ARRAY_SIZE(msm8994_clks),
+ .icc_clks = bimc_pcnoc_snoc_cnoc_ocmem_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(bimc_pcnoc_snoc_cnoc_ocmem_icc_clks),
};
static struct clk_smd_rpm *msm8996_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_PCNOC_CLK] = &clk_smd_rpm_bus_0_pcnoc_clk,
- [RPM_SMD_PCNOC_A_CLK] = &clk_smd_rpm_bus_0_pcnoc_a_clk,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_1_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_1_snoc_a_clk,
- [RPM_SMD_CNOC_CLK] = &clk_smd_rpm_bus_2_cnoc_clk,
- [RPM_SMD_CNOC_A_CLK] = &clk_smd_rpm_bus_2_cnoc_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
- [RPM_SMD_MMAXI_CLK] = &clk_smd_rpm_mmssnoc_axi_rpm_clk,
- [RPM_SMD_MMAXI_A_CLK] = &clk_smd_rpm_mmssnoc_axi_rpm_a_clk,
[RPM_SMD_IPA_CLK] = &clk_smd_rpm_ipa_clk,
[RPM_SMD_IPA_A_CLK] = &clk_smd_rpm_ipa_a_clk,
[RPM_SMD_CE1_CLK] = &clk_smd_rpm_ce1_clk,
[RPM_SMD_CE1_A_CLK] = &clk_smd_rpm_ce1_a_clk,
- [RPM_SMD_AGGR1_NOC_CLK] = &clk_smd_rpm_branch_aggre1_noc_clk,
- [RPM_SMD_AGGR1_NOC_A_CLK] = &clk_smd_rpm_branch_aggre1_noc_a_clk,
- [RPM_SMD_AGGR2_NOC_CLK] = &clk_smd_rpm_branch_aggre2_noc_clk,
- [RPM_SMD_AGGR2_NOC_A_CLK] = &clk_smd_rpm_branch_aggre2_noc_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_qdss_clk,
[RPM_SMD_QDSS_A_CLK] = &clk_smd_rpm_qdss_a_clk,
[RPM_SMD_BB_CLK1] = &clk_smd_rpm_bb_clk1,
@@ -900,6 +905,8 @@ static struct clk_smd_rpm *msm8996_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_msm8996 = {
.clks = msm8996_clks,
.num_clks = ARRAY_SIZE(msm8996_clks),
+ .icc_clks = msm8996_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(msm8996_icc_clks),
};
static struct clk_smd_rpm *qcs404_clks[] = {
@@ -928,19 +935,15 @@ static struct clk_smd_rpm *qcs404_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_qcs404 = {
.clks = qcs404_clks,
.num_clks = ARRAY_SIZE(qcs404_clks),
+ .icc_clks = bimc_pcnoc_snoc_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(bimc_pcnoc_snoc_icc_clks),
};
static struct clk_smd_rpm *msm8998_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
[RPM_SMD_PCNOC_CLK] = &clk_smd_rpm_bus_0_pcnoc_clk,
[RPM_SMD_PCNOC_A_CLK] = &clk_smd_rpm_bus_0_pcnoc_a_clk,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_1_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_1_snoc_a_clk,
- [RPM_SMD_CNOC_CLK] = &clk_smd_rpm_bus_2_cnoc_clk,
- [RPM_SMD_CNOC_A_CLK] = &clk_smd_rpm_bus_2_cnoc_a_clk,
[RPM_SMD_CE1_CLK] = &clk_smd_rpm_ce1_clk,
[RPM_SMD_CE1_A_CLK] = &clk_smd_rpm_ce1_a_clk,
[RPM_SMD_DIV_CLK1] = &clk_smd_rpm_div_clk1,
@@ -963,12 +966,6 @@ static struct clk_smd_rpm *msm8998_clks[] = {
[RPM_SMD_LN_BB_CLK2_A_PIN] = &clk_smd_rpm_ln_bb_clk2_a_pin,
[RPM_SMD_LN_BB_CLK3_PIN] = &clk_smd_rpm_ln_bb_clk3_pin,
[RPM_SMD_LN_BB_CLK3_A_PIN] = &clk_smd_rpm_ln_bb_clk3_a_pin,
- [RPM_SMD_MMAXI_CLK] = &clk_smd_rpm_mmssnoc_axi_rpm_clk,
- [RPM_SMD_MMAXI_A_CLK] = &clk_smd_rpm_mmssnoc_axi_rpm_a_clk,
- [RPM_SMD_AGGR1_NOC_CLK] = &clk_smd_rpm_aggre1_noc_clk,
- [RPM_SMD_AGGR1_NOC_A_CLK] = &clk_smd_rpm_aggre1_noc_a_clk,
- [RPM_SMD_AGGR2_NOC_CLK] = &clk_smd_rpm_aggre2_noc_clk,
- [RPM_SMD_AGGR2_NOC_A_CLK] = &clk_smd_rpm_aggre2_noc_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_qdss_clk,
[RPM_SMD_QDSS_A_CLK] = &clk_smd_rpm_qdss_a_clk,
[RPM_SMD_RF_CLK1] = &clk_smd_rpm_rf_clk1,
@@ -988,27 +985,19 @@ static struct clk_smd_rpm *msm8998_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_msm8998 = {
.clks = msm8998_clks,
.num_clks = ARRAY_SIZE(msm8998_clks),
+ .icc_clks = msm8998_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(msm8998_icc_clks),
};
static struct clk_smd_rpm *sdm660_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_1_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_1_snoc_a_clk,
- [RPM_SMD_CNOC_CLK] = &clk_smd_rpm_bus_2_cnoc_clk,
- [RPM_SMD_CNOC_A_CLK] = &clk_smd_rpm_bus_2_cnoc_a_clk,
[RPM_SMD_CNOC_PERIPH_CLK] = &clk_smd_rpm_bus_0_pcnoc_clk,
[RPM_SMD_CNOC_PERIPH_A_CLK] = &clk_smd_rpm_bus_0_pcnoc_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
- [RPM_SMD_MMSSNOC_AXI_CLK] = &clk_smd_rpm_mmssnoc_axi_rpm_clk,
- [RPM_SMD_MMSSNOC_AXI_CLK_A] = &clk_smd_rpm_mmssnoc_axi_rpm_a_clk,
[RPM_SMD_IPA_CLK] = &clk_smd_rpm_ipa_clk,
[RPM_SMD_IPA_A_CLK] = &clk_smd_rpm_ipa_a_clk,
[RPM_SMD_CE1_CLK] = &clk_smd_rpm_ce1_clk,
[RPM_SMD_CE1_A_CLK] = &clk_smd_rpm_ce1_a_clk,
- [RPM_SMD_AGGR2_NOC_CLK] = &clk_smd_rpm_aggre2_noc_clk,
- [RPM_SMD_AGGR2_NOC_A_CLK] = &clk_smd_rpm_aggre2_noc_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_qdss_clk,
[RPM_SMD_QDSS_A_CLK] = &clk_smd_rpm_qdss_a_clk,
[RPM_SMD_RF_CLK1] = &clk_smd_rpm_rf_clk1,
@@ -1034,15 +1023,13 @@ static struct clk_smd_rpm *sdm660_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_sdm660 = {
.clks = sdm660_clks,
.num_clks = ARRAY_SIZE(sdm660_clks),
+ .icc_clks = sdm660_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(sdm660_icc_clks),
};
static struct clk_smd_rpm *mdm9607_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_PCNOC_CLK] = &clk_smd_rpm_bus_0_pcnoc_clk,
- [RPM_SMD_PCNOC_A_CLK] = &clk_smd_rpm_bus_0_pcnoc_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
[RPM_SMD_QPIC_CLK] = &clk_smd_rpm_qpic_clk,
[RPM_SMD_QPIC_CLK_A] = &clk_smd_rpm_qpic_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_qdss_clk,
@@ -1056,21 +1043,15 @@ static struct clk_smd_rpm *mdm9607_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_mdm9607 = {
.clks = mdm9607_clks,
.num_clks = ARRAY_SIZE(mdm9607_clks),
+ .icc_clks = bimc_pcnoc_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(bimc_pcnoc_icc_clks),
};
static struct clk_smd_rpm *msm8953_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_PCNOC_CLK] = &clk_smd_rpm_bus_0_pcnoc_clk,
- [RPM_SMD_PCNOC_A_CLK] = &clk_smd_rpm_bus_0_pcnoc_a_clk,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_1_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_1_snoc_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
[RPM_SMD_IPA_CLK] = &clk_smd_rpm_ipa_clk,
[RPM_SMD_IPA_A_CLK] = &clk_smd_rpm_ipa_a_clk,
- [RPM_SMD_SYSMMNOC_CLK] = &clk_smd_rpm_bus_2_sysmmnoc_clk,
- [RPM_SMD_SYSMMNOC_A_CLK] = &clk_smd_rpm_bus_2_sysmmnoc_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_qdss_clk,
[RPM_SMD_QDSS_A_CLK] = &clk_smd_rpm_qdss_a_clk,
[RPM_SMD_BB_CLK1] = &clk_smd_rpm_bb_clk1,
@@ -1092,23 +1073,19 @@ static struct clk_smd_rpm *msm8953_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_msm8953 = {
.clks = msm8953_clks,
.num_clks = ARRAY_SIZE(msm8953_clks),
+ .icc_clks = bimc_pcnoc_snoc_smmnoc_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(bimc_pcnoc_snoc_smmnoc_icc_clks),
};
static struct clk_smd_rpm *sm6125_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_2_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_2_snoc_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_branch_qdss_clk,
[RPM_SMD_QDSS_A_CLK] = &clk_smd_rpm_branch_qdss_a_clk,
[RPM_SMD_RF_CLK1] = &clk_smd_rpm_rf_clk1,
[RPM_SMD_RF_CLK1_A] = &clk_smd_rpm_rf_clk1_a,
[RPM_SMD_RF_CLK2] = &clk_smd_rpm_rf_clk2,
[RPM_SMD_RF_CLK2_A] = &clk_smd_rpm_rf_clk2_a,
- [RPM_SMD_CNOC_CLK] = &clk_smd_rpm_bus_1_cnoc_clk,
- [RPM_SMD_CNOC_A_CLK] = &clk_smd_rpm_bus_1_cnoc_a_clk,
[RPM_SMD_IPA_CLK] = &clk_smd_rpm_ipa_clk,
[RPM_SMD_IPA_A_CLK] = &clk_smd_rpm_ipa_a_clk,
[RPM_SMD_CE1_CLK] = &clk_smd_rpm_ce1_clk,
@@ -1119,12 +1096,6 @@ static struct clk_smd_rpm *sm6125_clks[] = {
[RPM_SMD_LN_BB_CLK2_A] = &clk_smd_rpm_ln_bb_clk2_a,
[RPM_SMD_LN_BB_CLK3] = &clk_smd_rpm_ln_bb_clk3,
[RPM_SMD_LN_BB_CLK3_A] = &clk_smd_rpm_ln_bb_clk3_a,
- [RPM_SMD_QUP_CLK] = &clk_smd_rpm_qup_clk,
- [RPM_SMD_QUP_A_CLK] = &clk_smd_rpm_qup_a_clk,
- [RPM_SMD_MMRT_CLK] = &clk_smd_rpm_mmrt_clk,
- [RPM_SMD_MMRT_A_CLK] = &clk_smd_rpm_mmrt_a_clk,
- [RPM_SMD_MMNRT_CLK] = &clk_smd_rpm_mmnrt_clk,
- [RPM_SMD_MMNRT_A_CLK] = &clk_smd_rpm_mmnrt_a_clk,
[RPM_SMD_SNOC_PERIPH_CLK] = &clk_smd_rpm_bus_0_snoc_periph_clk,
[RPM_SMD_SNOC_PERIPH_A_CLK] = &clk_smd_rpm_bus_0_snoc_periph_a_clk,
[RPM_SMD_SNOC_LPASS_CLK] = &clk_smd_rpm_bus_5_snoc_lpass_clk,
@@ -1134,34 +1105,24 @@ static struct clk_smd_rpm *sm6125_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_sm6125 = {
.clks = sm6125_clks,
.num_clks = ARRAY_SIZE(sm6125_clks),
+ .icc_clks = sm_qnoc_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(sm_qnoc_icc_clks)
};
/* SM6115 */
static struct clk_smd_rpm *sm6115_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_2_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_2_snoc_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_branch_qdss_clk,
[RPM_SMD_QDSS_A_CLK] = &clk_smd_rpm_branch_qdss_a_clk,
[RPM_SMD_RF_CLK1] = &clk_smd_rpm_rf_clk1,
[RPM_SMD_RF_CLK1_A] = &clk_smd_rpm_rf_clk1_a,
[RPM_SMD_RF_CLK2] = &clk_smd_rpm_rf_clk2,
[RPM_SMD_RF_CLK2_A] = &clk_smd_rpm_rf_clk2_a,
- [RPM_SMD_CNOC_CLK] = &clk_smd_rpm_bus_1_cnoc_clk,
- [RPM_SMD_CNOC_A_CLK] = &clk_smd_rpm_bus_1_cnoc_a_clk,
[RPM_SMD_IPA_CLK] = &clk_smd_rpm_ipa_clk,
[RPM_SMD_IPA_A_CLK] = &clk_smd_rpm_ipa_a_clk,
[RPM_SMD_CE1_CLK] = &clk_smd_rpm_ce1_clk,
[RPM_SMD_CE1_A_CLK] = &clk_smd_rpm_ce1_a_clk,
- [RPM_SMD_QUP_CLK] = &clk_smd_rpm_qup_clk,
- [RPM_SMD_QUP_A_CLK] = &clk_smd_rpm_qup_a_clk,
- [RPM_SMD_MMRT_CLK] = &clk_smd_rpm_mmrt_clk,
- [RPM_SMD_MMRT_A_CLK] = &clk_smd_rpm_mmrt_a_clk,
- [RPM_SMD_MMNRT_CLK] = &clk_smd_rpm_mmnrt_clk,
- [RPM_SMD_MMNRT_A_CLK] = &clk_smd_rpm_mmnrt_a_clk,
[RPM_SMD_SNOC_PERIPH_CLK] = &clk_smd_rpm_bus_0_snoc_periph_clk,
[RPM_SMD_SNOC_PERIPH_A_CLK] = &clk_smd_rpm_bus_0_snoc_periph_a_clk,
[RPM_SMD_SNOC_LPASS_CLK] = &clk_smd_rpm_bus_5_snoc_lpass_clk,
@@ -1175,27 +1136,17 @@ static struct clk_smd_rpm *sm6115_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_sm6115 = {
.clks = sm6115_clks,
.num_clks = ARRAY_SIZE(sm6115_clks),
+ .icc_clks = sm_qnoc_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(sm_qnoc_icc_clks)
};
static struct clk_smd_rpm *sm6375_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_2_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_2_snoc_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_branch_qdss_clk,
[RPM_SMD_QDSS_A_CLK] = &clk_smd_rpm_branch_qdss_a_clk,
- [RPM_SMD_CNOC_CLK] = &clk_smd_rpm_bus_1_cnoc_clk,
- [RPM_SMD_CNOC_A_CLK] = &clk_smd_rpm_bus_1_cnoc_a_clk,
[RPM_SMD_IPA_CLK] = &clk_smd_rpm_ipa_clk,
[RPM_SMD_IPA_A_CLK] = &clk_smd_rpm_ipa_a_clk,
- [RPM_SMD_QUP_CLK] = &clk_smd_rpm_qup_clk,
- [RPM_SMD_QUP_A_CLK] = &clk_smd_rpm_qup_a_clk,
- [RPM_SMD_MMRT_CLK] = &clk_smd_rpm_mmrt_clk,
- [RPM_SMD_MMRT_A_CLK] = &clk_smd_rpm_mmrt_a_clk,
- [RPM_SMD_MMNRT_CLK] = &clk_smd_rpm_mmnrt_clk,
- [RPM_SMD_MMNRT_A_CLK] = &clk_smd_rpm_mmnrt_a_clk,
[RPM_SMD_SNOC_PERIPH_CLK] = &clk_smd_rpm_bus_0_snoc_periph_clk,
[RPM_SMD_SNOC_PERIPH_A_CLK] = &clk_smd_rpm_bus_0_snoc_periph_a_clk,
[RPM_SMD_SNOC_LPASS_CLK] = &clk_smd_rpm_bus_5_snoc_lpass_clk,
@@ -1212,31 +1163,21 @@ static struct clk_smd_rpm *sm6375_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_sm6375 = {
.clks = sm6375_clks,
.num_clks = ARRAY_SIZE(sm6375_clks),
+ .icc_clks = sm_qnoc_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(sm_qnoc_icc_clks)
};
static struct clk_smd_rpm *qcm2290_clks[] = {
[RPM_SMD_XO_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo,
[RPM_SMD_XO_A_CLK_SRC] = &clk_smd_rpm_branch_bi_tcxo_a,
- [RPM_SMD_SNOC_CLK] = &clk_smd_rpm_bus_2_snoc_clk,
- [RPM_SMD_SNOC_A_CLK] = &clk_smd_rpm_bus_2_snoc_a_clk,
- [RPM_SMD_BIMC_CLK] = &clk_smd_rpm_bimc_clk,
- [RPM_SMD_BIMC_A_CLK] = &clk_smd_rpm_bimc_a_clk,
[RPM_SMD_QDSS_CLK] = &clk_smd_rpm_branch_qdss_clk,
[RPM_SMD_QDSS_A_CLK] = &clk_smd_rpm_branch_qdss_a_clk,
[RPM_SMD_LN_BB_CLK2] = &clk_smd_rpm_ln_bb_clk2,
[RPM_SMD_LN_BB_CLK2_A] = &clk_smd_rpm_ln_bb_clk2_a,
[RPM_SMD_RF_CLK3] = &clk_smd_rpm_38m4_rf_clk3,
[RPM_SMD_RF_CLK3_A] = &clk_smd_rpm_38m4_rf_clk3_a,
- [RPM_SMD_CNOC_CLK] = &clk_smd_rpm_bus_1_cnoc_clk,
- [RPM_SMD_CNOC_A_CLK] = &clk_smd_rpm_bus_1_cnoc_a_clk,
[RPM_SMD_IPA_CLK] = &clk_smd_rpm_ipa_clk,
[RPM_SMD_IPA_A_CLK] = &clk_smd_rpm_ipa_a_clk,
- [RPM_SMD_QUP_CLK] = &clk_smd_rpm_qup_clk,
- [RPM_SMD_QUP_A_CLK] = &clk_smd_rpm_qup_a_clk,
- [RPM_SMD_MMRT_CLK] = &clk_smd_rpm_mmrt_clk,
- [RPM_SMD_MMRT_A_CLK] = &clk_smd_rpm_mmrt_a_clk,
- [RPM_SMD_MMNRT_CLK] = &clk_smd_rpm_mmnrt_clk,
- [RPM_SMD_MMNRT_A_CLK] = &clk_smd_rpm_mmnrt_a_clk,
[RPM_SMD_SNOC_PERIPH_CLK] = &clk_smd_rpm_bus_0_snoc_periph_clk,
[RPM_SMD_SNOC_PERIPH_A_CLK] = &clk_smd_rpm_bus_0_snoc_periph_a_clk,
[RPM_SMD_SNOC_LPASS_CLK] = &clk_smd_rpm_bus_5_snoc_lpass_clk,
@@ -1258,6 +1199,8 @@ static struct clk_smd_rpm *qcm2290_clks[] = {
static const struct rpm_smd_clk_desc rpm_clk_qcm2290 = {
.clks = qcm2290_clks,
.num_clks = ARRAY_SIZE(qcm2290_clks),
+ .icc_clks = sm_qnoc_icc_clks,
+ .num_icc_clks = ARRAY_SIZE(sm_qnoc_icc_clks)
};
static const struct of_device_id rpm_smd_clk_match_table[] = {
@@ -1333,6 +1276,15 @@ static int rpm_smd_clk_probe(struct platform_device *pdev)
goto err;
}
+ for (i = 0; i < desc->num_icc_clks; i++) {
+ if (!desc->icc_clks[i])
+ continue;
+
+ ret = clk_smd_rpm_handoff(desc->icc_clks[i]);
+ if (ret)
+ goto err;
+ }
+
if (!desc->scaling_before_handover) {
ret = clk_smd_rpm_enable_scaling();
if (ret)
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 19/22] interconnect: qcom: icc-rpm: Fix bucket number
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (17 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 18/22] clk: qcom: smd-rpm: Separate out interconnect bus clocks Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-10 17:46 ` Stephan Gerhold
2023-06-09 20:19 ` [PATCH v2 20/22] interconnect: qcom: icc-rpm: Set bandwidth on both contexts Konrad Dybcio
` (2 subsequent siblings)
21 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Dmitry Baryshkov
SMD RPM only provides two buckets, one each for the active-only and
active-sleep RPM contexts. Use the correct constant to allocate and
operate on them.
Fixes: dcbce7b0a79c ("interconnect: qcom: icc-rpm: Support multiple buckets")
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/icc-rpm.c | 14 +++++++-------
1 file changed, 7 insertions(+), 7 deletions(-)
diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
index 6d40815c5401..3ac47b818afe 100644
--- a/drivers/interconnect/qcom/icc-rpm.c
+++ b/drivers/interconnect/qcom/icc-rpm.c
@@ -249,7 +249,7 @@ static void qcom_icc_pre_bw_aggregate(struct icc_node *node)
size_t i;
qn = node->data;
- for (i = 0; i < QCOM_ICC_NUM_BUCKETS; i++) {
+ for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++) {
qn->sum_avg[i] = 0;
qn->max_peak[i] = 0;
}
@@ -275,7 +275,7 @@ static int qcom_icc_bw_aggregate(struct icc_node *node, u32 tag, u32 avg_bw,
if (!tag)
tag = QCOM_ICC_TAG_ALWAYS;
- for (i = 0; i < QCOM_ICC_NUM_BUCKETS; i++) {
+ for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++) {
if (tag & BIT(i)) {
qn->sum_avg[i] += avg_bw;
qn->max_peak[i] = max_t(u32, qn->max_peak[i], peak_bw);
@@ -300,11 +300,11 @@ static void qcom_icc_bus_aggregate(struct icc_provider *provider,
{
struct icc_node *node;
struct qcom_icc_node *qn;
- u64 sum_avg[QCOM_ICC_NUM_BUCKETS];
+ u64 sum_avg[QCOM_SMD_RPM_STATE_NUM];
int i;
/* Initialise aggregate values */
- for (i = 0; i < QCOM_ICC_NUM_BUCKETS; i++) {
+ for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++) {
agg_avg[i] = 0;
agg_peak[i] = 0;
}
@@ -317,7 +317,7 @@ static void qcom_icc_bus_aggregate(struct icc_provider *provider,
*/
list_for_each_entry(node, &provider->nodes, node_list) {
qn = node->data;
- for (i = 0; i < QCOM_ICC_NUM_BUCKETS; i++) {
+ for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++) {
if (qn->channels)
sum_avg[i] = div_u64(qn->sum_avg[i], qn->channels);
else
@@ -328,7 +328,7 @@ static void qcom_icc_bus_aggregate(struct icc_provider *provider,
}
/* Find maximum values across all buckets */
- for (i = 0; i < QCOM_ICC_NUM_BUCKETS; i++)
+ for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++)
*max_agg_avg = max_t(u64, *max_agg_avg, agg_avg[i]);
}
@@ -339,7 +339,7 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
struct icc_provider *provider;
u64 sum_bw;
u64 active_rate, sleep_rate;
- u64 agg_avg[QCOM_ICC_NUM_BUCKETS], agg_peak[QCOM_ICC_NUM_BUCKETS];
+ u64 agg_avg[QCOM_SMD_RPM_STATE_NUM], agg_peak[QCOM_SMD_RPM_STATE_NUM];
u64 max_agg_avg;
int ret;
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 20/22] interconnect: qcom: icc-rpm: Set bandwidth on both contexts
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (18 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 19/22] interconnect: qcom: icc-rpm: Fix bucket number Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-10 18:00 ` Stephan Gerhold
2023-06-09 20:19 ` [PATCH v2 21/22] interconnect: qcom: icc-rpm: Set correct bandwidth through RPM bw req Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 22/22] interconnect: qcom: icc-rpm: Fix bandwidth calculations Konrad Dybcio
21 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio
Up until now, for some reason we've only been setting bandwidth values
on the active-only context. That pretty much meant that RPM could lift
all votes when entering sleep mode. Or never sleep at all.
That in turn could potentially break things like USB wakeup, as the
connection between APSS and SNoC/PNoC would simply be dead.
Set the values appropriately.
Fixes: 30c8fa3ec61a ("interconnect: qcom: Add MSM8916 interconnect provider driver")
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/icc-rpm.c | 54 +++++++++++++++++++------------------
1 file changed, 28 insertions(+), 26 deletions(-)
diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
index 3ac47b818afe..ac719013077e 100644
--- a/drivers/interconnect/qcom/icc-rpm.c
+++ b/drivers/interconnect/qcom/icc-rpm.c
@@ -205,34 +205,39 @@ static int qcom_icc_qos_set(struct icc_node *node)
}
}
-static int qcom_icc_rpm_set(struct qcom_icc_node *qn, u64 sum_bw)
+static int qcom_icc_rpm_set(struct qcom_icc_node *qn, u64 *bw)
{
- int ret = 0;
+ int ret, rpm_ctx = 0;
+ u64 bw_bps;
if (qn->qos.ap_owned)
return 0;
- if (qn->mas_rpm_id != -1) {
- ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
- RPM_BUS_MASTER_REQ,
- qn->mas_rpm_id,
- sum_bw);
- if (ret) {
- pr_err("qcom_icc_rpm_smd_send mas %d error %d\n",
- qn->mas_rpm_id, ret);
- return ret;
+ for (rpm_ctx = 0; rpm_ctx < QCOM_SMD_RPM_STATE_NUM; rpm_ctx++) {
+ bw_bps = icc_units_to_bps(bw[rpm_ctx]);
+
+ if (qn->mas_rpm_id != -1) {
+ ret = qcom_icc_rpm_smd_send(rpm_ctx,
+ RPM_BUS_MASTER_REQ,
+ qn->mas_rpm_id,
+ bw_bps);
+ if (ret) {
+ pr_err("qcom_icc_rpm_smd_send mas %d error %d\n",
+ qn->mas_rpm_id, ret);
+ return ret;
+ }
}
- }
- if (qn->slv_rpm_id != -1) {
- ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
- RPM_BUS_SLAVE_REQ,
- qn->slv_rpm_id,
- sum_bw);
- if (ret) {
- pr_err("qcom_icc_rpm_smd_send slv %d error %d\n",
- qn->slv_rpm_id, ret);
- return ret;
+ if (qn->slv_rpm_id != -1) {
+ ret = qcom_icc_rpm_smd_send(rpm_ctx,
+ RPM_BUS_SLAVE_REQ,
+ qn->slv_rpm_id,
+ bw_bps);
+ if (ret) {
+ pr_err("qcom_icc_rpm_smd_send slv %d error %d\n",
+ qn->slv_rpm_id, ret);
+ return ret;
+ }
}
}
@@ -337,7 +342,6 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
struct qcom_icc_provider *qp;
struct qcom_icc_node *src_qn = NULL, *dst_qn = NULL;
struct icc_provider *provider;
- u64 sum_bw;
u64 active_rate, sleep_rate;
u64 agg_avg[QCOM_SMD_RPM_STATE_NUM], agg_peak[QCOM_SMD_RPM_STATE_NUM];
u64 max_agg_avg;
@@ -351,14 +355,12 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
qcom_icc_bus_aggregate(provider, agg_avg, agg_peak, &max_agg_avg);
- sum_bw = icc_units_to_bps(max_agg_avg);
-
- ret = qcom_icc_rpm_set(src_qn, sum_bw);
+ ret = qcom_icc_rpm_set(src_qn, agg_avg);
if (ret)
return ret;
if (dst_qn) {
- ret = qcom_icc_rpm_set(dst_qn, sum_bw);
+ ret = qcom_icc_rpm_set(dst_qn, agg_avg);
if (ret)
return ret;
}
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 21/22] interconnect: qcom: icc-rpm: Set correct bandwidth through RPM bw req
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (19 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 20/22] interconnect: qcom: icc-rpm: Set bandwidth on both contexts Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-10 18:46 ` Stephan Gerhold
2023-06-09 20:19 ` [PATCH v2 22/22] interconnect: qcom: icc-rpm: Fix bandwidth calculations Konrad Dybcio
21 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Stephan Gerhold
Currently, we're setting the aggregated-on-provider bandwidth on each
node, individually. That is of course incorrect and results in far too
high votes. Use the correct values to ensure we're not wasting power.
Fixes: 30c8fa3ec61a ("interconnect: qcom: Add MSM8916 interconnect provider driver")
Reported-by: Stephan Gerhold <stephan@gerhold.net>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/icc-rpm.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
index ac719013077e..1508233632f6 100644
--- a/drivers/interconnect/qcom/icc-rpm.c
+++ b/drivers/interconnect/qcom/icc-rpm.c
@@ -355,12 +355,12 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
qcom_icc_bus_aggregate(provider, agg_avg, agg_peak, &max_agg_avg);
- ret = qcom_icc_rpm_set(src_qn, agg_avg);
+ ret = qcom_icc_rpm_set(src_qn, src_qn->sum_avg);
if (ret)
return ret;
if (dst_qn) {
- ret = qcom_icc_rpm_set(dst_qn, agg_avg);
+ ret = qcom_icc_rpm_set(dst_qn, dst_qn->sum_avg);
if (ret)
return ret;
}
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* [PATCH v2 22/22] interconnect: qcom: icc-rpm: Fix bandwidth calculations
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
` (20 preceding siblings ...)
2023-06-09 20:19 ` [PATCH v2 21/22] interconnect: qcom: icc-rpm: Set correct bandwidth through RPM bw req Konrad Dybcio
@ 2023-06-09 20:19 ` Konrad Dybcio
2023-06-10 19:06 ` Stephan Gerhold
21 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-09 20:19 UTC (permalink / raw)
To: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green
Cc: Marijn Suijten, linux-arm-msm, linux-kernel, linux-clk, linux-pm,
Konrad Dybcio, Stephan Gerhold
Up until now, we've been aggregating the bandwidth values and only
dividing them by the bus width of the source node. This was completely
wrong, as different nodes on a given path may (and usually do) have
varying bus widths. That in turn, resulted in the calculated clock rates
being completely bogus - usually they ended up being much higher, as
NoC_A<->NoC_B links are very wide.
Since we're not using the aggregate bandwidth value for anything other
than clock rate calculations, remodel qcom_icc_bus_aggregate() to
calculate the per-context clock rate for a given provider, taking into
account the bus width of every individual node.
Fixes: 30c8fa3ec61a ("interconnect: qcom: Add MSM8916 interconnect provider driver")
Reported-by: Stephan Gerhold <stephan@gerhold.net>
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
---
drivers/interconnect/qcom/icc-rpm.c | 59 ++++++++++++-------------------------
1 file changed, 19 insertions(+), 40 deletions(-)
diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
index 1508233632f6..d177a76abe2a 100644
--- a/drivers/interconnect/qcom/icc-rpm.c
+++ b/drivers/interconnect/qcom/icc-rpm.c
@@ -293,58 +293,44 @@ static int qcom_icc_bw_aggregate(struct icc_node *node, u32 tag, u32 avg_bw,
}
/**
- * qcom_icc_bus_aggregate - aggregate bandwidth by traversing all nodes
+ * qcom_icc_bus_aggregate - calculate bus clock rates by traversing all nodes
* @provider: generic interconnect provider
- * @agg_avg: an array for aggregated average bandwidth of buckets
- * @agg_peak: an array for aggregated peak bandwidth of buckets
- * @max_agg_avg: pointer to max value of aggregated average bandwidth
+ * @agg_clk_rate: array containing the aggregated clock rates in kHz
*/
-static void qcom_icc_bus_aggregate(struct icc_provider *provider,
- u64 *agg_avg, u64 *agg_peak,
- u64 *max_agg_avg)
+static void qcom_icc_bus_aggregate(struct icc_provider *provider, u64 *agg_clk_rate)
{
- struct icc_node *node;
+ u64 agg_avg_rate, agg_rate;
struct qcom_icc_node *qn;
- u64 sum_avg[QCOM_SMD_RPM_STATE_NUM];
+ struct icc_node *node;
int i;
- /* Initialise aggregate values */
- for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++) {
- agg_avg[i] = 0;
- agg_peak[i] = 0;
- }
-
- *max_agg_avg = 0;
-
/*
- * Iterate nodes on the interconnect and aggregate bandwidth
- * requests for every bucket.
+ * Iterate nodes on the provider, aggregate bandwidth requests for
+ * every bucket and convert them into bus clock rates.
*/
list_for_each_entry(node, &provider->nodes, node_list) {
qn = node->data;
for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++) {
if (qn->channels)
- sum_avg[i] = div_u64(qn->sum_avg[i], qn->channels);
+ agg_avg_rate = div_u64(qn->sum_avg[i], qn->channels);
else
- sum_avg[i] = qn->sum_avg[i];
- agg_avg[i] += sum_avg[i];
- agg_peak[i] = max_t(u64, agg_peak[i], qn->max_peak[i]);
+ agg_avg_rate = qn->sum_avg[i];
+
+ agg_rate = max_t(u64, agg_avg_rate, qn->max_peak[i]);
+ do_div(agg_rate, qn->buswidth);
+
+ agg_clk_rate[i] = max_t(u64, agg_clk_rate[i], agg_rate);
}
}
-
- /* Find maximum values across all buckets */
- for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++)
- *max_agg_avg = max_t(u64, *max_agg_avg, agg_avg[i]);
}
static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
{
- struct qcom_icc_provider *qp;
struct qcom_icc_node *src_qn = NULL, *dst_qn = NULL;
+ u64 agg_clk_rate[QCOM_SMD_RPM_STATE_NUM] = { 0 };
struct icc_provider *provider;
+ struct qcom_icc_provider *qp;
u64 active_rate, sleep_rate;
- u64 agg_avg[QCOM_SMD_RPM_STATE_NUM], agg_peak[QCOM_SMD_RPM_STATE_NUM];
- u64 max_agg_avg;
int ret;
src_qn = src->data;
@@ -353,7 +339,9 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
provider = src->provider;
qp = to_qcom_provider(provider);
- qcom_icc_bus_aggregate(provider, agg_avg, agg_peak, &max_agg_avg);
+ qcom_icc_bus_aggregate(provider, agg_clk_rate);
+ active_rate = agg_clk_rate[QCOM_SMD_RPM_ACTIVE_STATE];
+ sleep_rate = agg_clk_rate[QCOM_SMD_RPM_SLEEP_STATE];
ret = qcom_icc_rpm_set(src_qn, src_qn->sum_avg);
if (ret)
@@ -369,15 +357,6 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
if (!qp->bus_clk_desc && !qp->bus_clk)
return 0;
- /* Intentionally keep the rates in kHz as that's what RPM accepts */
- active_rate = max(agg_avg[QCOM_SMD_RPM_ACTIVE_STATE],
- agg_peak[QCOM_SMD_RPM_ACTIVE_STATE]);
- do_div(active_rate, src_qn->buswidth);
-
- sleep_rate = max(agg_avg[QCOM_SMD_RPM_SLEEP_STATE],
- agg_peak[QCOM_SMD_RPM_SLEEP_STATE]);
- do_div(sleep_rate, src_qn->buswidth);
-
/*
* Downstream checks whether the requested rate is zero, but it makes little sense
* to vote for a value that's below the lower threshold, so let's not do so.
--
2.41.0
^ permalink raw reply related [flat|nested] 45+ messages in thread
* Re: [PATCH v2 04/22] clk: qcom: smd-rpm: Export clock scaling availability
2023-06-09 20:19 ` [PATCH v2 04/22] clk: qcom: smd-rpm: Export clock scaling availability Konrad Dybcio
@ 2023-06-10 11:35 ` Stephan Gerhold
2023-06-10 12:15 ` Konrad Dybcio
0 siblings, 1 reply; 45+ messages in thread
From: Stephan Gerhold @ 2023-06-10 11:35 UTC (permalink / raw)
To: Konrad Dybcio
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On Fri, Jun 09, 2023 at 10:19:09PM +0200, Konrad Dybcio wrote:
> Before we issue a call to RPM through clk_smd_rpm_enable_scaling() the
> clock rate requests will not be commited in hardware. This poses a
> race threat since we're accessing the bus clocks directly from within
> the interconnect framework.
>
> Add a marker to indicate that we're good to go with sending new requests
> and export it so that it can be referenced from icc.
>
> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
> ---
> drivers/clk/qcom/clk-smd-rpm.c | 9 +++++++++
> include/linux/soc/qcom/smd-rpm.h | 2 ++
> 2 files changed, 11 insertions(+)
>
> diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
> index 937cb1515968..482fe30ee6f0 100644
> --- a/drivers/clk/qcom/clk-smd-rpm.c
> +++ b/drivers/clk/qcom/clk-smd-rpm.c
> @@ -151,6 +151,7 @@
> #define to_clk_smd_rpm(_hw) container_of(_hw, struct clk_smd_rpm, hw)
>
> static struct qcom_smd_rpm *rpmcc_smd_rpm;
> +static bool smd_rpm_clk_scaling;
>
> struct clk_smd_rpm {
> const int rpm_res_type;
> @@ -385,6 +386,12 @@ static unsigned long clk_smd_rpm_recalc_rate(struct clk_hw *hw,
> return r->rate;
> }
>
> +bool qcom_smd_rpm_scaling_available(void)
> +{
> + return smd_rpm_clk_scaling;
> +}
> +EXPORT_SYMBOL_GPL(qcom_smd_rpm_scaling_available);
> +
> static int clk_smd_rpm_enable_scaling(void)
> {
> int ret;
> @@ -410,6 +417,8 @@ static int clk_smd_rpm_enable_scaling(void)
> return ret;
> }
>
> + smd_rpm_clk_scaling = true;
> +
If you move the platform_device_register_data(&rpdev->dev,
"icc_smd_rpm", ...) from drivers/soc/qcom/smd-rpm.c to here you can
avoid the race completely and drop this API. I think that would be
cleaner. And it will likely probe much faster because probe deferral
is slow. :)
Thanks,
Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 17/22] interconnect: qcom: icc-rpm: Control bus rpmcc from icc
2023-06-09 20:19 ` [PATCH v2 17/22] interconnect: qcom: icc-rpm: Control bus rpmcc from icc Konrad Dybcio
@ 2023-06-10 11:58 ` Stephan Gerhold
2023-06-10 12:14 ` Konrad Dybcio
0 siblings, 1 reply; 45+ messages in thread
From: Stephan Gerhold @ 2023-06-10 11:58 UTC (permalink / raw)
To: Konrad Dybcio
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On Fri, Jun 09, 2023 at 10:19:22PM +0200, Konrad Dybcio wrote:
> The sole purpose of bus clocks that were previously registered with
> rpmcc was to convey the aggregated bandwidth to RPM. There's no good
> reason to keep them outside the interconnect framework, as it only
> adds to the plentiful complexity.
>
> Add the required code to handle these clocks from within SMD RPM ICC.
>
> RPM-owned bus clocks are no longer considered a thing, but sadly we
> have to allow for the existence of HLOS-owned bus clocks, as some
> (mostly older) SoCs (ab)use these for bus scaling (e.g. MSM8998 and
> &mmcc AHB_CLK_SRC).
>
> This in turn is trivially solved with a single *clk, which is filled
> and used iff qp.bus_clk_desc is absent and we have a "bus" clock-names
> entry in the DT node.
>
> This change should(tm) be fully compatible with all sorts of old
> Device Trees as far as the interconnect functionality goes (modulo
> abusing bus clock handles, but that's a mistake in and of itself).
>
> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
> ---
> drivers/interconnect/qcom/icc-rpm.c | 116 +++++++++++++++++++-----------------
> drivers/interconnect/qcom/icc-rpm.h | 13 ++--
> drivers/interconnect/qcom/msm8996.c | 1 -
> drivers/interconnect/qcom/sdm660.c | 1 -
> 4 files changed, 66 insertions(+), 65 deletions(-)
>
> diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
> index b8ecf9538ab9..6d40815c5401 100644
> --- a/drivers/interconnect/qcom/icc-rpm.c
> +++ b/drivers/interconnect/qcom/icc-rpm.c
> [...]
> @@ -364,49 +363,50 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
> return ret;
> }
>
> - for (i = 0; i < qp->num_bus_clks; i++) {
> - /*
> - * Use WAKE bucket for active clock, otherwise, use SLEEP bucket
> - * for other clocks. If a platform doesn't set interconnect
> - * path tags, by default use sleep bucket for all clocks.
> - *
> - * Note, AMC bucket is not supported yet.
> - */
> - if (!strcmp(qp->bus_clks[i].id, "bus_a"))
> - bucket = QCOM_ICC_BUCKET_WAKE;
> - else
> - bucket = QCOM_ICC_BUCKET_SLEEP;
> -
> - rate = icc_units_to_bps(max(agg_avg[bucket], agg_peak[bucket]));
> - do_div(rate, src_qn->buswidth);
> - rate = min_t(u64, rate, LONG_MAX);
^
> -
> - /*
> - * Downstream checks whether the requested rate is zero, but it makes little sense
> - * to vote for a value that's below the lower threshold, so let's not do so.
> - */
> - if (bucket == QCOM_ICC_BUCKET_WAKE && qp->keep_alive)
> - rate = max(ICC_BUS_CLK_MIN_RATE, rate);
> -
> - if (qp->bus_clk_rate[i] == rate)
> - continue;
> -
> - ret = clk_set_rate(qp->bus_clks[i].clk, rate);
> - if (ret) {
> - pr_err("%s clk_set_rate error: %d\n",
> - qp->bus_clks[i].id, ret);
> + /* Some providers don't have a bus clock to scale */
> + if (!qp->bus_clk_desc && !qp->bus_clk)
> + return 0;
> +
> + /* Intentionally keep the rates in kHz as that's what RPM accepts */
> + active_rate = max(agg_avg[QCOM_SMD_RPM_ACTIVE_STATE],
> + agg_peak[QCOM_SMD_RPM_ACTIVE_STATE]);
> + do_div(active_rate, src_qn->buswidth);
> +
> + sleep_rate = max(agg_avg[QCOM_SMD_RPM_SLEEP_STATE],
> + agg_peak[QCOM_SMD_RPM_SLEEP_STATE]);
> + do_div(sleep_rate, src_qn->buswidth);
> +
> + /*
> + * Downstream checks whether the requested rate is zero, but it makes little sense
> + * to vote for a value that's below the lower threshold, so let's not do so.
> + */
> + if (qp->keep_alive)
> + active_rate = max(ICC_BUS_CLK_MIN_RATE, active_rate);
> +
> + /* Some providers have a non-RPM-owned bus clock - convert kHz->Hz for the CCF */
> + if (qp->bus_clk)
> + return clk_set_rate(qp->bus_clk, 1000ULL * max(active_rate, sleep_rate));
Something like the min_t(u64, rate, LONG_MAX)* that was there in the old
code is still needed for the clk_set_rate(). The reason is that the rate
parameter in clk_set_rate() is unsigned long (32-bit on ARM32) while you
do the calculation in fixed u64. This can easily overflow and then the
higher bits will just be cut off.
Consider the following on ARM32:
u64 rate = 1ULL << 32 = 4294967296ULL;
clk_set_rate(clk, rate);
This actually does clk_set_rate(clk, 0) because the upper 32-bit will
just be truncated. So the min() is needed to ensure that we really set
the highest possible.
Also see commit a7d9436a6c85 ("interconnect: qcom: rpm: Prevent integer
overflow in rate") [1].
* I'm not sure why I used LONG_MAX instead of ULONG_MAX back then.
[1]: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=a7d9436a6c85fcb8843c910fd323dcd7f839bf63
> [...]
> int qnoc_probe(struct platform_device *pdev)
> {
> struct device *dev = &pdev->dev;
> @@ -448,6 +448,18 @@ int qnoc_probe(struct platform_device *pdev)
> if (!qp->intf_clks)
> return -ENOMEM;
>
> + if (desc->bus_clk_desc) {
> + qp->bus_clk_desc = devm_kzalloc(dev, sizeof(*qp->bus_clk_desc),
> + GFP_KERNEL);
> + if (!qp->bus_clk_desc)
> + return -ENOMEM;
> +
> + qp->bus_clk_desc = desc->bus_clk_desc;
> + } else if (!IS_ERR(devm_clk_get(dev, "bus"))) {
> + /* Some older SoCs may have a single non-RPM-owned bus clock. */
> + qp->bus_clk = devm_clk_get(dev, "bus");
> + }
Hm, looks like you're requesting the clock twice? devm_clk_get()
allocates memory internally so that's not ideal. It would be better to
call it just once and store the result. Or do you actually want
devm_clk_get_optional(dev, "bus") maybe? The error handling is a bit
weird here.
> [...]
> @@ -490,13 +498,11 @@ int qnoc_probe(struct platform_device *pdev)
> }
>
> regmap_done:
> - ret = devm_clk_bulk_get(dev, qp->num_bus_clks, qp->bus_clks);
> - if (ret)
> - return ret;
> -
> - ret = clk_bulk_prepare_enable(qp->num_bus_clks, qp->bus_clks);
> - if (ret)
> - return ret;
I guess we need dt-binding updates so we can drop the clocks from the
device tree? They're not valid anymore after you remove them from
clk-smd-rpm.c so it would be good to drop them from the DTs to avoid
confusion.
> + if (qp->bus_clk) {
> + ret = clk_prepare_enable(qp->bus_clk);
> + if (ret)
> + return ret;
> + }
The clk_* APIs have internal NULL checks for optional clocks so you
could also omit the if statement and call it unconditionally.
>
> ret = devm_clk_bulk_get(dev, qp->num_intf_clks, qp->intf_clks);
> if (ret)
> @@ -566,7 +572,8 @@ int qnoc_probe(struct platform_device *pdev)
> icc_provider_deregister(provider);
> err_remove_nodes:
> icc_nodes_remove(provider);
> - clk_bulk_disable_unprepare(qp->num_bus_clks, qp->bus_clks);
> + if (qp->bus_clk)
> + clk_disable_unprepare(qp->bus_clk);
>
Same here.
> return ret;
> }
> @@ -578,7 +585,8 @@ int qnoc_remove(struct platform_device *pdev)
>
> icc_provider_deregister(&qp->provider);
> icc_nodes_remove(&qp->provider);
> - clk_bulk_disable_unprepare(qp->num_bus_clks, qp->bus_clks);
> + if (qp->bus_clk)
> + clk_disable_unprepare(qp->bus_clk);
>
And here.
Thanks,
Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 15/22] interconnect: qcom: msm8916: Hook up RPM bus clk definitions
2023-06-09 20:19 ` [PATCH v2 15/22] interconnect: qcom: msm8916: " Konrad Dybcio
@ 2023-06-10 12:02 ` Stephan Gerhold
0 siblings, 0 replies; 45+ messages in thread
From: Stephan Gerhold @ 2023-06-10 12:02 UTC (permalink / raw)
To: Konrad Dybcio
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm, Dmitry Baryshkov
On Fri, Jun 09, 2023 at 10:19:20PM +0200, Konrad Dybcio wrote:
> Assign the necessary definitions to migrate to the new bus clock
> handling mechanism.
>
> Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Reviewed-by: Stephan Gerhold <stephan@gerhold.net>
> ---
> drivers/interconnect/qcom/msm8916.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/interconnect/qcom/msm8916.c b/drivers/interconnect/qcom/msm8916.c
> index 196b05879896..be2a190a8b52 100644
> --- a/drivers/interconnect/qcom/msm8916.c
> +++ b/drivers/interconnect/qcom/msm8916.c
> @@ -1231,6 +1231,7 @@ static const struct qcom_icc_desc msm8916_snoc = {
> .type = QCOM_ICC_NOC,
> .nodes = msm8916_snoc_nodes,
> .num_nodes = ARRAY_SIZE(msm8916_snoc_nodes),
> + .bus_clk_desc = &bus_1_clk,
> .regmap_cfg = &msm8916_snoc_regmap_config,
> .qos_offset = 0x7000,
> };
> @@ -1259,6 +1260,7 @@ static const struct qcom_icc_desc msm8916_bimc = {
> .type = QCOM_ICC_BIMC,
> .nodes = msm8916_bimc_nodes,
> .num_nodes = ARRAY_SIZE(msm8916_bimc_nodes),
> + .bus_clk_desc = &bimc_clk,
> .regmap_cfg = &msm8916_bimc_regmap_config,
> .qos_offset = 0x8000,
> };
> @@ -1328,6 +1330,7 @@ static const struct qcom_icc_desc msm8916_pcnoc = {
> .type = QCOM_ICC_NOC,
> .nodes = msm8916_pcnoc_nodes,
> .num_nodes = ARRAY_SIZE(msm8916_pcnoc_nodes),
> + .bus_clk_desc = &bus_0_clk,
> .regmap_cfg = &msm8916_pcnoc_regmap_config,
> .qos_offset = 0x7000,
> };
>
> --
> 2.41.0
>
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 14/22] interconnect: qcom: msm8939: Hook up RPM bus clk definitions
2023-06-09 20:19 ` [PATCH v2 14/22] interconnect: qcom: msm8939: " Konrad Dybcio
@ 2023-06-10 12:02 ` Stephan Gerhold
0 siblings, 0 replies; 45+ messages in thread
From: Stephan Gerhold @ 2023-06-10 12:02 UTC (permalink / raw)
To: Konrad Dybcio
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm, Dmitry Baryshkov
On Fri, Jun 09, 2023 at 10:19:19PM +0200, Konrad Dybcio wrote:
> Assign the necessary definitions to migrate to the new bus clock
> handling mechanism.
>
> Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Reviewed-by: Stephan Gerhold <stephan@gerhold.net>
> ---
> drivers/interconnect/qcom/msm8939.c | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/drivers/interconnect/qcom/msm8939.c b/drivers/interconnect/qcom/msm8939.c
> index 639566dce45a..94b11b590a8e 100644
> --- a/drivers/interconnect/qcom/msm8939.c
> +++ b/drivers/interconnect/qcom/msm8939.c
> @@ -1284,6 +1284,7 @@ static const struct qcom_icc_desc msm8939_snoc = {
> .type = QCOM_ICC_NOC,
> .nodes = msm8939_snoc_nodes,
> .num_nodes = ARRAY_SIZE(msm8939_snoc_nodes),
> + .bus_clk_desc = &bus_1_clk,
> .regmap_cfg = &msm8939_snoc_regmap_config,
> .qos_offset = 0x7000,
> };
> @@ -1304,6 +1305,7 @@ static const struct qcom_icc_desc msm8939_snoc_mm = {
> .type = QCOM_ICC_NOC,
> .nodes = msm8939_snoc_mm_nodes,
> .num_nodes = ARRAY_SIZE(msm8939_snoc_mm_nodes),
> + .bus_clk_desc = &bus_2_clk,
> .regmap_cfg = &msm8939_snoc_regmap_config,
> .qos_offset = 0x7000,
> };
> @@ -1332,6 +1334,7 @@ static const struct qcom_icc_desc msm8939_bimc = {
> .type = QCOM_ICC_BIMC,
> .nodes = msm8939_bimc_nodes,
> .num_nodes = ARRAY_SIZE(msm8939_bimc_nodes),
> + .bus_clk_desc = &bimc_clk,
> .regmap_cfg = &msm8939_bimc_regmap_config,
> .qos_offset = 0x8000,
> };
> @@ -1403,6 +1406,7 @@ static const struct qcom_icc_desc msm8939_pcnoc = {
> .type = QCOM_ICC_NOC,
> .nodes = msm8939_pcnoc_nodes,
> .num_nodes = ARRAY_SIZE(msm8939_pcnoc_nodes),
> + .bus_clk_desc = &bus_0_clk,
> .regmap_cfg = &msm8939_pcnoc_regmap_config,
> .qos_offset = 0x7000,
> };
>
> --
> 2.41.0
>
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 17/22] interconnect: qcom: icc-rpm: Control bus rpmcc from icc
2023-06-10 11:58 ` Stephan Gerhold
@ 2023-06-10 12:14 ` Konrad Dybcio
2023-06-10 16:20 ` Stephan Gerhold
0 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-10 12:14 UTC (permalink / raw)
To: Stephan Gerhold
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On 10.06.2023 13:58, Stephan Gerhold wrote:
> On Fri, Jun 09, 2023 at 10:19:22PM +0200, Konrad Dybcio wrote:
>> The sole purpose of bus clocks that were previously registered with
>> rpmcc was to convey the aggregated bandwidth to RPM. There's no good
>> reason to keep them outside the interconnect framework, as it only
>> adds to the plentiful complexity.
>>
>> Add the required code to handle these clocks from within SMD RPM ICC.
>>
>> RPM-owned bus clocks are no longer considered a thing, but sadly we
>> have to allow for the existence of HLOS-owned bus clocks, as some
>> (mostly older) SoCs (ab)use these for bus scaling (e.g. MSM8998 and
>> &mmcc AHB_CLK_SRC).
>>
>> This in turn is trivially solved with a single *clk, which is filled
>> and used iff qp.bus_clk_desc is absent and we have a "bus" clock-names
>> entry in the DT node.
>>
>> This change should(tm) be fully compatible with all sorts of old
>> Device Trees as far as the interconnect functionality goes (modulo
>> abusing bus clock handles, but that's a mistake in and of itself).
>>
>> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
>> ---
>> drivers/interconnect/qcom/icc-rpm.c | 116 +++++++++++++++++++-----------------
>> drivers/interconnect/qcom/icc-rpm.h | 13 ++--
>> drivers/interconnect/qcom/msm8996.c | 1 -
>> drivers/interconnect/qcom/sdm660.c | 1 -
>> 4 files changed, 66 insertions(+), 65 deletions(-)
>>
>> diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
>> index b8ecf9538ab9..6d40815c5401 100644
>> --- a/drivers/interconnect/qcom/icc-rpm.c
>> +++ b/drivers/interconnect/qcom/icc-rpm.c
>> [...]
>> @@ -364,49 +363,50 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
>> return ret;
>> }
>>
>> - for (i = 0; i < qp->num_bus_clks; i++) {
>> - /*
>> - * Use WAKE bucket for active clock, otherwise, use SLEEP bucket
>> - * for other clocks. If a platform doesn't set interconnect
>> - * path tags, by default use sleep bucket for all clocks.
>> - *
>> - * Note, AMC bucket is not supported yet.
>> - */
>> - if (!strcmp(qp->bus_clks[i].id, "bus_a"))
>> - bucket = QCOM_ICC_BUCKET_WAKE;
>> - else
>> - bucket = QCOM_ICC_BUCKET_SLEEP;
>> -
>> - rate = icc_units_to_bps(max(agg_avg[bucket], agg_peak[bucket]));
>> - do_div(rate, src_qn->buswidth);
>> - rate = min_t(u64, rate, LONG_MAX);
>
> ^
>
>> -
>> - /*
>> - * Downstream checks whether the requested rate is zero, but it makes little sense
>> - * to vote for a value that's below the lower threshold, so let's not do so.
>> - */
>> - if (bucket == QCOM_ICC_BUCKET_WAKE && qp->keep_alive)
>> - rate = max(ICC_BUS_CLK_MIN_RATE, rate);
>> -
>> - if (qp->bus_clk_rate[i] == rate)
>> - continue;
>> -
>> - ret = clk_set_rate(qp->bus_clks[i].clk, rate);
>> - if (ret) {
>> - pr_err("%s clk_set_rate error: %d\n",
>> - qp->bus_clks[i].id, ret);
>> + /* Some providers don't have a bus clock to scale */
>> + if (!qp->bus_clk_desc && !qp->bus_clk)
>> + return 0;
>> +
>> + /* Intentionally keep the rates in kHz as that's what RPM accepts */
>> + active_rate = max(agg_avg[QCOM_SMD_RPM_ACTIVE_STATE],
>> + agg_peak[QCOM_SMD_RPM_ACTIVE_STATE]);
>> + do_div(active_rate, src_qn->buswidth);
>> +
>> + sleep_rate = max(agg_avg[QCOM_SMD_RPM_SLEEP_STATE],
>> + agg_peak[QCOM_SMD_RPM_SLEEP_STATE]);
>> + do_div(sleep_rate, src_qn->buswidth);
>> +
>> + /*
>> + * Downstream checks whether the requested rate is zero, but it makes little sense
>> + * to vote for a value that's below the lower threshold, so let's not do so.
>> + */
>> + if (qp->keep_alive)
>> + active_rate = max(ICC_BUS_CLK_MIN_RATE, active_rate);
>> +
>> + /* Some providers have a non-RPM-owned bus clock - convert kHz->Hz for the CCF */
>> + if (qp->bus_clk)
>> + return clk_set_rate(qp->bus_clk, 1000ULL * max(active_rate, sleep_rate));
>
> Something like the min_t(u64, rate, LONG_MAX)* that was there in the old
> code is still needed for the clk_set_rate(). The reason is that the rate
> parameter in clk_set_rate() is unsigned long (32-bit on ARM32) while you
> do the calculation in fixed u64. This can easily overflow and then the
> higher bits will just be cut off.
>
> Consider the following on ARM32:
>
> u64 rate = 1ULL << 32 = 4294967296ULL;
> clk_set_rate(clk, rate);
>
> This actually does clk_set_rate(clk, 0) because the upper 32-bit will
> just be truncated. So the min() is needed to ensure that we really set
> the highest possible.
>
> Also see commit a7d9436a6c85 ("interconnect: qcom: rpm: Prevent integer
> overflow in rate") [1].
>
> * I'm not sure why I used LONG_MAX instead of ULONG_MAX back then.
Ughh can we kill arm32 already? It only causes problems :P
Thanks for spotting this..
>
> [1]: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=a7d9436a6c85fcb8843c910fd323dcd7f839bf63
>
>> [...]
>> int qnoc_probe(struct platform_device *pdev)
>> {
>> struct device *dev = &pdev->dev;
>> @@ -448,6 +448,18 @@ int qnoc_probe(struct platform_device *pdev)
>> if (!qp->intf_clks)
>> return -ENOMEM;
>>
>> + if (desc->bus_clk_desc) {
>> + qp->bus_clk_desc = devm_kzalloc(dev, sizeof(*qp->bus_clk_desc),
>> + GFP_KERNEL);
>> + if (!qp->bus_clk_desc)
>> + return -ENOMEM;
>> +
>> + qp->bus_clk_desc = desc->bus_clk_desc;
>> + } else if (!IS_ERR(devm_clk_get(dev, "bus"))) {
>> + /* Some older SoCs may have a single non-RPM-owned bus clock. */
>> + qp->bus_clk = devm_clk_get(dev, "bus");
>> + }
>
> Hm, looks like you're requesting the clock twice? devm_clk_get()
> allocates memory internally so that's not ideal. It would be better to
> call it just once and store the result. Or do you actually want
> devm_clk_get_optional(dev, "bus") maybe? The error handling is a bit
> weird here.
Hmm, right.. I think it should go something like this:
qp->bus_clk = devm_clk_get(dev, "bus");
if (IS_ERR(qp->bus_clk) && PTR_ERR(qp->bus_clk) == -ENOENT)
qp->bus_clk = NULL;
else if (IS_ERR(qp->bus_clk))
return PTR_ERR(qp->bus_clk)
>
>> [...]
>> @@ -490,13 +498,11 @@ int qnoc_probe(struct platform_device *pdev)
>> }
>>
>> regmap_done:
>> - ret = devm_clk_bulk_get(dev, qp->num_bus_clks, qp->bus_clks);
>> - if (ret)
>> - return ret;
>> -
>> - ret = clk_bulk_prepare_enable(qp->num_bus_clks, qp->bus_clks);
>> - if (ret)
>> - return ret;
>
> I guess we need dt-binding updates so we can drop the clocks from the
> device tree? They're not valid anymore after you remove them from
> clk-smd-rpm.c so it would be good to drop them from the DTs to avoid
> confusion.
I wanted to handle that separately as the bindings are technically still
correct.. There'll be a need for some deprecation though.
>
>> + if (qp->bus_clk) {
>> + ret = clk_prepare_enable(qp->bus_clk);
>> + if (ret)
>> + return ret;
>> + }
>
> The clk_* APIs have internal NULL checks for optional clocks so you
> could also omit the if statement and call it unconditionally.
Ack
Konrad
>
>>
>> ret = devm_clk_bulk_get(dev, qp->num_intf_clks, qp->intf_clks);
>> if (ret)
>> @@ -566,7 +572,8 @@ int qnoc_probe(struct platform_device *pdev)
>> icc_provider_deregister(provider);
>> err_remove_nodes:
>> icc_nodes_remove(provider);
>> - clk_bulk_disable_unprepare(qp->num_bus_clks, qp->bus_clks);
>> + if (qp->bus_clk)
>> + clk_disable_unprepare(qp->bus_clk);
>>
>
> Same here.
>
>> return ret;
>> }
>> @@ -578,7 +585,8 @@ int qnoc_remove(struct platform_device *pdev)
>>
>> icc_provider_deregister(&qp->provider);
>> icc_nodes_remove(&qp->provider);
>> - clk_bulk_disable_unprepare(qp->num_bus_clks, qp->bus_clks);
>> + if (qp->bus_clk)
>> + clk_disable_unprepare(qp->bus_clk);
>>
>
> And here.
>
> Thanks,
> Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 04/22] clk: qcom: smd-rpm: Export clock scaling availability
2023-06-10 11:35 ` Stephan Gerhold
@ 2023-06-10 12:15 ` Konrad Dybcio
2023-06-10 18:53 ` Konrad Dybcio
0 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-10 12:15 UTC (permalink / raw)
To: Stephan Gerhold
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On 10.06.2023 13:35, Stephan Gerhold wrote:
> On Fri, Jun 09, 2023 at 10:19:09PM +0200, Konrad Dybcio wrote:
>> Before we issue a call to RPM through clk_smd_rpm_enable_scaling() the
>> clock rate requests will not be commited in hardware. This poses a
>> race threat since we're accessing the bus clocks directly from within
>> the interconnect framework.
>>
>> Add a marker to indicate that we're good to go with sending new requests
>> and export it so that it can be referenced from icc.
>>
>> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
>> ---
>> drivers/clk/qcom/clk-smd-rpm.c | 9 +++++++++
>> include/linux/soc/qcom/smd-rpm.h | 2 ++
>> 2 files changed, 11 insertions(+)
>>
>> diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
>> index 937cb1515968..482fe30ee6f0 100644
>> --- a/drivers/clk/qcom/clk-smd-rpm.c
>> +++ b/drivers/clk/qcom/clk-smd-rpm.c
>> @@ -151,6 +151,7 @@
>> #define to_clk_smd_rpm(_hw) container_of(_hw, struct clk_smd_rpm, hw)
>>
>> static struct qcom_smd_rpm *rpmcc_smd_rpm;
>> +static bool smd_rpm_clk_scaling;
>>
>> struct clk_smd_rpm {
>> const int rpm_res_type;
>> @@ -385,6 +386,12 @@ static unsigned long clk_smd_rpm_recalc_rate(struct clk_hw *hw,
>> return r->rate;
>> }
>>
>> +bool qcom_smd_rpm_scaling_available(void)
>> +{
>> + return smd_rpm_clk_scaling;
>> +}
>> +EXPORT_SYMBOL_GPL(qcom_smd_rpm_scaling_available);
>> +
>> static int clk_smd_rpm_enable_scaling(void)
>> {
>> int ret;
>> @@ -410,6 +417,8 @@ static int clk_smd_rpm_enable_scaling(void)
>> return ret;
>> }
>>
>> + smd_rpm_clk_scaling = true;
>> +
>
> If you move the platform_device_register_data(&rpdev->dev,
> "icc_smd_rpm", ...) from drivers/soc/qcom/smd-rpm.c to here you can
> avoid the race completely and drop this API. I think that would be
> cleaner. And it will likely probe much faster because probe deferral
> is slow. :)
Sounds like an idea.. especially since it's pretty much the only
dependency other than SMDRPM itself!
Konrad
>
> Thanks,
> Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 17/22] interconnect: qcom: icc-rpm: Control bus rpmcc from icc
2023-06-10 12:14 ` Konrad Dybcio
@ 2023-06-10 16:20 ` Stephan Gerhold
2023-06-10 17:54 ` Konrad Dybcio
0 siblings, 1 reply; 45+ messages in thread
From: Stephan Gerhold @ 2023-06-10 16:20 UTC (permalink / raw)
To: Konrad Dybcio
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On Sat, Jun 10, 2023 at 02:14:18PM +0200, Konrad Dybcio wrote:
> On 10.06.2023 13:58, Stephan Gerhold wrote:
> > On Fri, Jun 09, 2023 at 10:19:22PM +0200, Konrad Dybcio wrote:
> >> The sole purpose of bus clocks that were previously registered with
> >> rpmcc was to convey the aggregated bandwidth to RPM. There's no good
> >> reason to keep them outside the interconnect framework, as it only
> >> adds to the plentiful complexity.
> >>
> >> Add the required code to handle these clocks from within SMD RPM ICC.
> >>
> >> RPM-owned bus clocks are no longer considered a thing, but sadly we
> >> have to allow for the existence of HLOS-owned bus clocks, as some
> >> (mostly older) SoCs (ab)use these for bus scaling (e.g. MSM8998 and
> >> &mmcc AHB_CLK_SRC).
> >>
> >> This in turn is trivially solved with a single *clk, which is filled
> >> and used iff qp.bus_clk_desc is absent and we have a "bus" clock-names
> >> entry in the DT node.
> >>
> >> This change should(tm) be fully compatible with all sorts of old
> >> Device Trees as far as the interconnect functionality goes (modulo
> >> abusing bus clock handles, but that's a mistake in and of itself).
> >>
> >> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
> >> ---
> >> drivers/interconnect/qcom/icc-rpm.c | 116 +++++++++++++++++++-----------------
> >> drivers/interconnect/qcom/icc-rpm.h | 13 ++--
> >> drivers/interconnect/qcom/msm8996.c | 1 -
> >> drivers/interconnect/qcom/sdm660.c | 1 -
> >> 4 files changed, 66 insertions(+), 65 deletions(-)
> >>
> >> diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
> >> index b8ecf9538ab9..6d40815c5401 100644
> >> --- a/drivers/interconnect/qcom/icc-rpm.c
> >> +++ b/drivers/interconnect/qcom/icc-rpm.c
> >> [...]
> >> @@ -364,49 +363,50 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
> >> return ret;
> >> }
> >>
> >> - for (i = 0; i < qp->num_bus_clks; i++) {
> >> - /*
> >> - * Use WAKE bucket for active clock, otherwise, use SLEEP bucket
> >> - * for other clocks. If a platform doesn't set interconnect
> >> - * path tags, by default use sleep bucket for all clocks.
> >> - *
> >> - * Note, AMC bucket is not supported yet.
> >> - */
> >> - if (!strcmp(qp->bus_clks[i].id, "bus_a"))
> >> - bucket = QCOM_ICC_BUCKET_WAKE;
> >> - else
> >> - bucket = QCOM_ICC_BUCKET_SLEEP;
> >> -
> >> - rate = icc_units_to_bps(max(agg_avg[bucket], agg_peak[bucket]));
> >> - do_div(rate, src_qn->buswidth);
> >> - rate = min_t(u64, rate, LONG_MAX);
> >
> > ^
> >
> >> -
> >> - /*
> >> - * Downstream checks whether the requested rate is zero, but it makes little sense
> >> - * to vote for a value that's below the lower threshold, so let's not do so.
> >> - */
> >> - if (bucket == QCOM_ICC_BUCKET_WAKE && qp->keep_alive)
> >> - rate = max(ICC_BUS_CLK_MIN_RATE, rate);
> >> -
> >> - if (qp->bus_clk_rate[i] == rate)
> >> - continue;
> >> -
> >> - ret = clk_set_rate(qp->bus_clks[i].clk, rate);
> >> - if (ret) {
> >> - pr_err("%s clk_set_rate error: %d\n",
> >> - qp->bus_clks[i].id, ret);
> >> + /* Some providers don't have a bus clock to scale */
> >> + if (!qp->bus_clk_desc && !qp->bus_clk)
> >> + return 0;
> >> +
> >> + /* Intentionally keep the rates in kHz as that's what RPM accepts */
> >> + active_rate = max(agg_avg[QCOM_SMD_RPM_ACTIVE_STATE],
> >> + agg_peak[QCOM_SMD_RPM_ACTIVE_STATE]);
> >> + do_div(active_rate, src_qn->buswidth);
> >> +
> >> + sleep_rate = max(agg_avg[QCOM_SMD_RPM_SLEEP_STATE],
> >> + agg_peak[QCOM_SMD_RPM_SLEEP_STATE]);
> >> + do_div(sleep_rate, src_qn->buswidth);
> >> +
> >> + /*
> >> + * Downstream checks whether the requested rate is zero, but it makes little sense
> >> + * to vote for a value that's below the lower threshold, so let's not do so.
> >> + */
> >> + if (qp->keep_alive)
> >> + active_rate = max(ICC_BUS_CLK_MIN_RATE, active_rate);
> >> +
> >> + /* Some providers have a non-RPM-owned bus clock - convert kHz->Hz for the CCF */
> >> + if (qp->bus_clk)
> >> + return clk_set_rate(qp->bus_clk, 1000ULL * max(active_rate, sleep_rate));
> >
> > Something like the min_t(u64, rate, LONG_MAX)* that was there in the old
> > code is still needed for the clk_set_rate(). The reason is that the rate
> > parameter in clk_set_rate() is unsigned long (32-bit on ARM32) while you
> > do the calculation in fixed u64. This can easily overflow and then the
> > higher bits will just be cut off.
> >
> > Consider the following on ARM32:
> >
> > u64 rate = 1ULL << 32 = 4294967296ULL;
> > clk_set_rate(clk, rate);
> >
> > This actually does clk_set_rate(clk, 0) because the upper 32-bit will
> > just be truncated. So the min() is needed to ensure that we really set
> > the highest possible.
> >
> > Also see commit a7d9436a6c85 ("interconnect: qcom: rpm: Prevent integer
> > overflow in rate") [1].
> >
> > * I'm not sure why I used LONG_MAX instead of ULONG_MAX back then.
> Ughh can we kill arm32 already? It only causes problems :P
>
Nooooo, "32-bit ought to be enough for anybody"! :P
> >
> > [1]: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=a7d9436a6c85fcb8843c910fd323dcd7f839bf63
> >
> >> [...]
> >> int qnoc_probe(struct platform_device *pdev)
> >> {
> >> struct device *dev = &pdev->dev;
> >> @@ -448,6 +448,18 @@ int qnoc_probe(struct platform_device *pdev)
> >> if (!qp->intf_clks)
> >> return -ENOMEM;
> >>
> >> + if (desc->bus_clk_desc) {
> >> + qp->bus_clk_desc = devm_kzalloc(dev, sizeof(*qp->bus_clk_desc),
> >> + GFP_KERNEL);
> >> + if (!qp->bus_clk_desc)
> >> + return -ENOMEM;
> >> +
> >> + qp->bus_clk_desc = desc->bus_clk_desc;
> >> + } else if (!IS_ERR(devm_clk_get(dev, "bus"))) {
> >> + /* Some older SoCs may have a single non-RPM-owned bus clock. */
> >> + qp->bus_clk = devm_clk_get(dev, "bus");
> >> + }
> >
> > Hm, looks like you're requesting the clock twice? devm_clk_get()
> > allocates memory internally so that's not ideal. It would be better to
> > call it just once and store the result. Or do you actually want
> > devm_clk_get_optional(dev, "bus") maybe? The error handling is a bit
> > weird here.
> Hmm, right.. I think it should go something like this:
>
> qp->bus_clk = devm_clk_get(dev, "bus");
> if (IS_ERR(qp->bus_clk) && PTR_ERR(qp->bus_clk) == -ENOENT)
> qp->bus_clk = NULL;
> else if (IS_ERR(qp->bus_clk))
> return PTR_ERR(qp->bus_clk)
>
This is the same as
qp->bus_clk = devm_clk_get_optional(dev, "bus");
if (IS_ERR(qp->bus_clk))
return PTR_ERR(qp->bus_clk);
clk_get_optional is literally defined as
static inline struct clk *clk_get_optional(struct device *dev, const char *id)
{
struct clk *clk = clk_get(dev, id);
if (clk == ERR_PTR(-ENOENT))
return NULL;
return clk;
}
>
> >
> >> [...]
> >> @@ -490,13 +498,11 @@ int qnoc_probe(struct platform_device *pdev)
> >> }
> >>
> >> regmap_done:
> >> - ret = devm_clk_bulk_get(dev, qp->num_bus_clks, qp->bus_clks);
> >> - if (ret)
> >> - return ret;
> >> -
> >> - ret = clk_bulk_prepare_enable(qp->num_bus_clks, qp->bus_clks);
> >> - if (ret)
> >> - return ret;
> >
> > I guess we need dt-binding updates so we can drop the clocks from the
> > device tree? They're not valid anymore after you remove them from
> > clk-smd-rpm.c so it would be good to drop them from the DTs to avoid
> > confusion.
> I wanted to handle that separately as the bindings are technically still
> correct.. There'll be a need for some deprecation though.
>
OK, that's fine for me (not sure how the DT maintainers would feel about
this though).
Thanks,
Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 19/22] interconnect: qcom: icc-rpm: Fix bucket number
2023-06-09 20:19 ` [PATCH v2 19/22] interconnect: qcom: icc-rpm: Fix bucket number Konrad Dybcio
@ 2023-06-10 17:46 ` Stephan Gerhold
2023-06-10 17:53 ` Konrad Dybcio
0 siblings, 1 reply; 45+ messages in thread
From: Stephan Gerhold @ 2023-06-10 17:46 UTC (permalink / raw)
To: Konrad Dybcio
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm, Dmitry Baryshkov
On Fri, Jun 09, 2023 at 10:19:24PM +0200, Konrad Dybcio wrote:
> SMD RPM only provides two buckets, one each for the active-only and
> active-sleep RPM contexts. Use the correct constant to allocate and
> operate on them.
>
> Fixes: dcbce7b0a79c ("interconnect: qcom: icc-rpm: Support multiple buckets")
> Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
> ---
> drivers/interconnect/qcom/icc-rpm.c | 14 +++++++-------
> 1 file changed, 7 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
> index 6d40815c5401..3ac47b818afe 100644
> --- a/drivers/interconnect/qcom/icc-rpm.c
> +++ b/drivers/interconnect/qcom/icc-rpm.c
> [...]
> @@ -275,7 +275,7 @@ static int qcom_icc_bw_aggregate(struct icc_node *node, u32 tag, u32 avg_bw,
> if (!tag)
> tag = QCOM_ICC_TAG_ALWAYS;
>
> - for (i = 0; i < QCOM_ICC_NUM_BUCKETS; i++) {
> + for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++) {
> if (tag & BIT(i)) {
Hm, I think QCOM_ICC_NUM_BUCKETS is actually intentional here. There is
a hint about this in the description of the commit in your Fixes line:
> This patch studies the implementation from interconnect rpmh driver to
> support multiple buckets. The rpmh driver provides three buckets for
> AMC, WAKE, and SLEEP; this driver only needs to use WAKE and SLEEP
> buckets, but we keep the same way with rpmh driver, this can allow us
> to reuse the DT binding and avoid to define duplicated data structures.
As far as I understand, the idea was to reuse the definitions in
qcom,icc.h and just ignore the AMC bucket for now. AFAIU AMC (or rather
the lack thereof) is basically caching: Sending requests without AMC bit
set is delayed until the next rpmh_flush() call that happens when
entering a deep idle state. It requires some work but I guess
theoretically one could implement exactly the same for RPM.
What you're actually doing here is not fixing the commit but changing
the bindings. On MSM8909 I defined the ICC path for CPU<->RAM like this:
interconnects = <&bimc MAS_APPS_PROC QCOM_ICC_TAG_ACTIVE_ONLY
&bimc SLV_EBI QCOM_ICC_TAG_ACTIVE_ONLY>;
Per definition in qcom,icc.h:
QCOM_ICC_TAG_ACTIVE_ONLY = (AMC | WAKE) = (BIT(0) | BIT(1))
Without your patch series this behaves correctly. It results in an
active-only vote.
The change of behavior is in PATCH 17/22 "interconnect: qcom: icc-rpm:
Control bus rpmcc from icc". It silently switches from
QCOM_ICC_BUCKET_WAKE (1) and QCOM_ICC_BUCKET_SLEEP (2) to
QCOM_SMD_RPM_ACTIVE_STATE (0) and QCOM_SMD_RPM_SLEEP_STATE (1).
In other words, QCOM_ICC_TAG_ACTIVE_ONLY (BIT(0) | BIT(1)) now results
in an active+sleep vote, not an active-only one. :)
There doesn't seem to be an upstream user of the ICC tags/buckets for
icc-rpm yet so personally I would be fine with changing it. However,
then qcom,icc.h should get a clear comment that it's rpmh-only and we
should define a new qcom,icc-rpm.h.
Or perhaps we should just drop this patch and continue using
QCOM_ICC_BUCKET_WAKE and QCOM_ICC_BUCKET_SLEEP as before?
Thanks,
Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 19/22] interconnect: qcom: icc-rpm: Fix bucket number
2023-06-10 17:46 ` Stephan Gerhold
@ 2023-06-10 17:53 ` Konrad Dybcio
0 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-10 17:53 UTC (permalink / raw)
To: Stephan Gerhold
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm, Dmitry Baryshkov
On 10.06.2023 19:46, Stephan Gerhold wrote:
> On Fri, Jun 09, 2023 at 10:19:24PM +0200, Konrad Dybcio wrote:
>> SMD RPM only provides two buckets, one each for the active-only and
>> active-sleep RPM contexts. Use the correct constant to allocate and
>> operate on them.
>>
>> Fixes: dcbce7b0a79c ("interconnect: qcom: icc-rpm: Support multiple buckets")
>> Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
>> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
>> ---
>> drivers/interconnect/qcom/icc-rpm.c | 14 +++++++-------
>> 1 file changed, 7 insertions(+), 7 deletions(-)
>>
>> diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
>> index 6d40815c5401..3ac47b818afe 100644
>> --- a/drivers/interconnect/qcom/icc-rpm.c
>> +++ b/drivers/interconnect/qcom/icc-rpm.c
>> [...]
>> @@ -275,7 +275,7 @@ static int qcom_icc_bw_aggregate(struct icc_node *node, u32 tag, u32 avg_bw,
>> if (!tag)
>> tag = QCOM_ICC_TAG_ALWAYS;
>>
>> - for (i = 0; i < QCOM_ICC_NUM_BUCKETS; i++) {
>> + for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++) {
>> if (tag & BIT(i)) {
>
> Hm, I think QCOM_ICC_NUM_BUCKETS is actually intentional here. There is
> a hint about this in the description of the commit in your Fixes line:
>
>> This patch studies the implementation from interconnect rpmh driver to
>> support multiple buckets. The rpmh driver provides three buckets for
>> AMC, WAKE, and SLEEP; this driver only needs to use WAKE and SLEEP
>> buckets, but we keep the same way with rpmh driver, this can allow us
>> to reuse the DT binding and avoid to define duplicated data structures.
>
> As far as I understand, the idea was to reuse the definitions in
> qcom,icc.h and just ignore the AMC bucket for now. AFAIU AMC (or rather
> the lack thereof) is basically caching: Sending requests without AMC bit
> set is delayed until the next rpmh_flush() call that happens when
> entering a deep idle state. It requires some work but I guess
> theoretically one could implement exactly the same for RPM.
That's trying to shove a cube into a circle-shaped hole.. AMC is a
hardware (well, firmware.. you know what I mean) feature, which SMD
RPM lacks. Plus it'd result in useless allocations.
>
> What you're actually doing here is not fixing the commit but changing
> the bindings. On MSM8909 I defined the ICC path for CPU<->RAM like this:
>
> interconnects = <&bimc MAS_APPS_PROC QCOM_ICC_TAG_ACTIVE_ONLY
> &bimc SLV_EBI QCOM_ICC_TAG_ACTIVE_ONLY>;
>
> Per definition in qcom,icc.h:
>
> QCOM_ICC_TAG_ACTIVE_ONLY = (AMC | WAKE) = (BIT(0) | BIT(1))
>
> Without your patch series this behaves correctly. It results in an
> active-only vote.
>
> The change of behavior is in PATCH 17/22 "interconnect: qcom: icc-rpm:
> Control bus rpmcc from icc". It silently switches from
> QCOM_ICC_BUCKET_WAKE (1) and QCOM_ICC_BUCKET_SLEEP (2) to
> QCOM_SMD_RPM_ACTIVE_STATE (0) and QCOM_SMD_RPM_SLEEP_STATE (1).
>
> In other words, QCOM_ICC_TAG_ACTIVE_ONLY (BIT(0) | BIT(1)) now results
> in an active+sleep vote, not an active-only one. :)
>
> There doesn't seem to be an upstream user of the ICC tags/buckets for
> icc-rpm yet so personally I would be fine with changing it. However,
> then qcom,icc.h should get a clear comment that it's rpmh-only and we
> should define a new qcom,icc-rpm.h.
Right, I'd argue the original commit was in the wrong here. It was trying
to reuse bindings which were intended for a different hw architecture (and
perhaps not described very well - there's no word of RPMh neither in the
name nor in the defines themselves). That's an abuse in my view..
I think introducing RPM-specific bindings and communicating the change
clearly is the way to go. As you've noticed, there are no users so that
should not be problematic at all.
Konrad
>
> Or perhaps we should just drop this patch and continue using
> QCOM_ICC_BUCKET_WAKE and QCOM_ICC_BUCKET_SLEEP as before?
>
> Thanks,
> Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 17/22] interconnect: qcom: icc-rpm: Control bus rpmcc from icc
2023-06-10 16:20 ` Stephan Gerhold
@ 2023-06-10 17:54 ` Konrad Dybcio
0 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-10 17:54 UTC (permalink / raw)
To: Stephan Gerhold
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On 10.06.2023 18:20, Stephan Gerhold wrote:
> On Sat, Jun 10, 2023 at 02:14:18PM +0200, Konrad Dybcio wrote:
>> On 10.06.2023 13:58, Stephan Gerhold wrote:
>>> On Fri, Jun 09, 2023 at 10:19:22PM +0200, Konrad Dybcio wrote:
>>>> The sole purpose of bus clocks that were previously registered with
>>>> rpmcc was to convey the aggregated bandwidth to RPM. There's no good
>>>> reason to keep them outside the interconnect framework, as it only
>>>> adds to the plentiful complexity.
>>>>
>>>> Add the required code to handle these clocks from within SMD RPM ICC.
>>>>
>>>> RPM-owned bus clocks are no longer considered a thing, but sadly we
>>>> have to allow for the existence of HLOS-owned bus clocks, as some
>>>> (mostly older) SoCs (ab)use these for bus scaling (e.g. MSM8998 and
>>>> &mmcc AHB_CLK_SRC).
>>>>
>>>> This in turn is trivially solved with a single *clk, which is filled
>>>> and used iff qp.bus_clk_desc is absent and we have a "bus" clock-names
>>>> entry in the DT node.
>>>>
>>>> This change should(tm) be fully compatible with all sorts of old
>>>> Device Trees as far as the interconnect functionality goes (modulo
>>>> abusing bus clock handles, but that's a mistake in and of itself).
>>>>
>>>> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
>>>> ---
>>>> drivers/interconnect/qcom/icc-rpm.c | 116 +++++++++++++++++++-----------------
>>>> drivers/interconnect/qcom/icc-rpm.h | 13 ++--
>>>> drivers/interconnect/qcom/msm8996.c | 1 -
>>>> drivers/interconnect/qcom/sdm660.c | 1 -
>>>> 4 files changed, 66 insertions(+), 65 deletions(-)
>>>>
>>>> diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
>>>> index b8ecf9538ab9..6d40815c5401 100644
>>>> --- a/drivers/interconnect/qcom/icc-rpm.c
>>>> +++ b/drivers/interconnect/qcom/icc-rpm.c
>>>> [...]
>>>> @@ -364,49 +363,50 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
>>>> return ret;
>>>> }
>>>>
>>>> - for (i = 0; i < qp->num_bus_clks; i++) {
>>>> - /*
>>>> - * Use WAKE bucket for active clock, otherwise, use SLEEP bucket
>>>> - * for other clocks. If a platform doesn't set interconnect
>>>> - * path tags, by default use sleep bucket for all clocks.
>>>> - *
>>>> - * Note, AMC bucket is not supported yet.
>>>> - */
>>>> - if (!strcmp(qp->bus_clks[i].id, "bus_a"))
>>>> - bucket = QCOM_ICC_BUCKET_WAKE;
>>>> - else
>>>> - bucket = QCOM_ICC_BUCKET_SLEEP;
>>>> -
>>>> - rate = icc_units_to_bps(max(agg_avg[bucket], agg_peak[bucket]));
>>>> - do_div(rate, src_qn->buswidth);
>>>> - rate = min_t(u64, rate, LONG_MAX);
>>>
>>> ^
>>>
>>>> -
>>>> - /*
>>>> - * Downstream checks whether the requested rate is zero, but it makes little sense
>>>> - * to vote for a value that's below the lower threshold, so let's not do so.
>>>> - */
>>>> - if (bucket == QCOM_ICC_BUCKET_WAKE && qp->keep_alive)
>>>> - rate = max(ICC_BUS_CLK_MIN_RATE, rate);
>>>> -
>>>> - if (qp->bus_clk_rate[i] == rate)
>>>> - continue;
>>>> -
>>>> - ret = clk_set_rate(qp->bus_clks[i].clk, rate);
>>>> - if (ret) {
>>>> - pr_err("%s clk_set_rate error: %d\n",
>>>> - qp->bus_clks[i].id, ret);
>>>> + /* Some providers don't have a bus clock to scale */
>>>> + if (!qp->bus_clk_desc && !qp->bus_clk)
>>>> + return 0;
>>>> +
>>>> + /* Intentionally keep the rates in kHz as that's what RPM accepts */
>>>> + active_rate = max(agg_avg[QCOM_SMD_RPM_ACTIVE_STATE],
>>>> + agg_peak[QCOM_SMD_RPM_ACTIVE_STATE]);
>>>> + do_div(active_rate, src_qn->buswidth);
>>>> +
>>>> + sleep_rate = max(agg_avg[QCOM_SMD_RPM_SLEEP_STATE],
>>>> + agg_peak[QCOM_SMD_RPM_SLEEP_STATE]);
>>>> + do_div(sleep_rate, src_qn->buswidth);
>>>> +
>>>> + /*
>>>> + * Downstream checks whether the requested rate is zero, but it makes little sense
>>>> + * to vote for a value that's below the lower threshold, so let's not do so.
>>>> + */
>>>> + if (qp->keep_alive)
>>>> + active_rate = max(ICC_BUS_CLK_MIN_RATE, active_rate);
>>>> +
>>>> + /* Some providers have a non-RPM-owned bus clock - convert kHz->Hz for the CCF */
>>>> + if (qp->bus_clk)
>>>> + return clk_set_rate(qp->bus_clk, 1000ULL * max(active_rate, sleep_rate));
>>>
>>> Something like the min_t(u64, rate, LONG_MAX)* that was there in the old
>>> code is still needed for the clk_set_rate(). The reason is that the rate
>>> parameter in clk_set_rate() is unsigned long (32-bit on ARM32) while you
>>> do the calculation in fixed u64. This can easily overflow and then the
>>> higher bits will just be cut off.
>>>
>>> Consider the following on ARM32:
>>>
>>> u64 rate = 1ULL << 32 = 4294967296ULL;
>>> clk_set_rate(clk, rate);
>>>
>>> This actually does clk_set_rate(clk, 0) because the upper 32-bit will
>>> just be truncated. So the min() is needed to ensure that we really set
>>> the highest possible.
>>>
>>> Also see commit a7d9436a6c85 ("interconnect: qcom: rpm: Prevent integer
>>> overflow in rate") [1].
>>>
>>> * I'm not sure why I used LONG_MAX instead of ULONG_MAX back then.
>> Ughh can we kill arm32 already? It only causes problems :P
>>
>
> Nooooo, "32-bit ought to be enough for anybody"! :P
>
>
>>>
>>> [1]: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=a7d9436a6c85fcb8843c910fd323dcd7f839bf63
>>>
>>>> [...]
>>>> int qnoc_probe(struct platform_device *pdev)
>>>> {
>>>> struct device *dev = &pdev->dev;
>>>> @@ -448,6 +448,18 @@ int qnoc_probe(struct platform_device *pdev)
>>>> if (!qp->intf_clks)
>>>> return -ENOMEM;
>>>>
>>>> + if (desc->bus_clk_desc) {
>>>> + qp->bus_clk_desc = devm_kzalloc(dev, sizeof(*qp->bus_clk_desc),
>>>> + GFP_KERNEL);
>>>> + if (!qp->bus_clk_desc)
>>>> + return -ENOMEM;
>>>> +
>>>> + qp->bus_clk_desc = desc->bus_clk_desc;
>>>> + } else if (!IS_ERR(devm_clk_get(dev, "bus"))) {
>>>> + /* Some older SoCs may have a single non-RPM-owned bus clock. */
>>>> + qp->bus_clk = devm_clk_get(dev, "bus");
>>>> + }
>>>
>>> Hm, looks like you're requesting the clock twice? devm_clk_get()
>>> allocates memory internally so that's not ideal. It would be better to
>>> call it just once and store the result. Or do you actually want
>>> devm_clk_get_optional(dev, "bus") maybe? The error handling is a bit
>>> weird here.
>> Hmm, right.. I think it should go something like this:
>>
>> qp->bus_clk = devm_clk_get(dev, "bus");
>> if (IS_ERR(qp->bus_clk) && PTR_ERR(qp->bus_clk) == -ENOENT)
>> qp->bus_clk = NULL;
>> else if (IS_ERR(qp->bus_clk))
>> return PTR_ERR(qp->bus_clk)
>>
>
> This is the same as
>
> qp->bus_clk = devm_clk_get_optional(dev, "bus");
> if (IS_ERR(qp->bus_clk))
> return PTR_ERR(qp->bus_clk);
>
> clk_get_optional is literally defined as
>
> static inline struct clk *clk_get_optional(struct device *dev, const char *id)
> {
> struct clk *clk = clk_get(dev, id);
> if (clk == ERR_PTR(-ENOENT))
> return NULL;
> return clk;
> }
Touche. I'll use this.
Konrad
>
>>
>>>
>>>> [...]
>>>> @@ -490,13 +498,11 @@ int qnoc_probe(struct platform_device *pdev)
>>>> }
>>>>
>>>> regmap_done:
>>>> - ret = devm_clk_bulk_get(dev, qp->num_bus_clks, qp->bus_clks);
>>>> - if (ret)
>>>> - return ret;
>>>> -
>>>> - ret = clk_bulk_prepare_enable(qp->num_bus_clks, qp->bus_clks);
>>>> - if (ret)
>>>> - return ret;
>>>
>>> I guess we need dt-binding updates so we can drop the clocks from the
>>> device tree? They're not valid anymore after you remove them from
>>> clk-smd-rpm.c so it would be good to drop them from the DTs to avoid
>>> confusion.
>> I wanted to handle that separately as the bindings are technically still
>> correct.. There'll be a need for some deprecation though.
>>
>
> OK, that's fine for me (not sure how the DT maintainers would feel about
> this though).
>
> Thanks,
> Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 20/22] interconnect: qcom: icc-rpm: Set bandwidth on both contexts
2023-06-09 20:19 ` [PATCH v2 20/22] interconnect: qcom: icc-rpm: Set bandwidth on both contexts Konrad Dybcio
@ 2023-06-10 18:00 ` Stephan Gerhold
2023-06-10 18:28 ` Konrad Dybcio
0 siblings, 1 reply; 45+ messages in thread
From: Stephan Gerhold @ 2023-06-10 18:00 UTC (permalink / raw)
To: Konrad Dybcio
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On Fri, Jun 09, 2023 at 10:19:25PM +0200, Konrad Dybcio wrote:
> Up until now, for some reason we've only been setting bandwidth values
> on the active-only context. That pretty much meant that RPM could lift
> all votes when entering sleep mode. Or never sleep at all.
>
> That in turn could potentially break things like USB wakeup, as the
> connection between APSS and SNoC/PNoC would simply be dead.
>
Nitpick: Apparently an "active" vote is applied during both active+sleep
until the first "sleep" vote is sent. It's documented only for
regulators [1] but I would expect the same applies to the bandwidths.
This means actual breakage shouldn't have been possible.
The patch itself is still the right thing to do to have the sleep state
correct during deep cpuidle/suspend.
[1]: https://git.codelinaro.org/clo/la/kernel/msm-3.10/-/blob/LA.BR.1.2.9.c26-04700-8x09.0/drivers/regulator/rpm-smd-regulator.c#L199-209
> Set the values appropriately.
>
> Fixes: 30c8fa3ec61a ("interconnect: qcom: Add MSM8916 interconnect provider driver")
> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
> ---
> drivers/interconnect/qcom/icc-rpm.c | 54 +++++++++++++++++++------------------
> 1 file changed, 28 insertions(+), 26 deletions(-)
>
> diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
> index 3ac47b818afe..ac719013077e 100644
> --- a/drivers/interconnect/qcom/icc-rpm.c
> +++ b/drivers/interconnect/qcom/icc-rpm.c
> @@ -205,34 +205,39 @@ static int qcom_icc_qos_set(struct icc_node *node)
> }
> }
>
> -static int qcom_icc_rpm_set(struct qcom_icc_node *qn, u64 sum_bw)
> +static int qcom_icc_rpm_set(struct qcom_icc_node *qn, u64 *bw)
> {
> - int ret = 0;
> + int ret, rpm_ctx = 0;
> + u64 bw_bps;
>
> if (qn->qos.ap_owned)
> return 0;
>
> - if (qn->mas_rpm_id != -1) {
> - ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
> - RPM_BUS_MASTER_REQ,
> - qn->mas_rpm_id,
> - sum_bw);
> - if (ret) {
> - pr_err("qcom_icc_rpm_smd_send mas %d error %d\n",
> - qn->mas_rpm_id, ret);
> - return ret;
> + for (rpm_ctx = 0; rpm_ctx < QCOM_SMD_RPM_STATE_NUM; rpm_ctx++) {
> + bw_bps = icc_units_to_bps(bw[rpm_ctx]);
> +
> + if (qn->mas_rpm_id != -1) {
> + ret = qcom_icc_rpm_smd_send(rpm_ctx,
> + RPM_BUS_MASTER_REQ,
> + qn->mas_rpm_id,
> + bw_bps);
> + if (ret) {
> + pr_err("qcom_icc_rpm_smd_send mas %d error %d\n",
> + qn->mas_rpm_id, ret);
> + return ret;
> + }
> }
> - }
>
> - if (qn->slv_rpm_id != -1) {
> - ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
> - RPM_BUS_SLAVE_REQ,
> - qn->slv_rpm_id,
> - sum_bw);
> - if (ret) {
> - pr_err("qcom_icc_rpm_smd_send slv %d error %d\n",
> - qn->slv_rpm_id, ret);
> - return ret;
> + if (qn->slv_rpm_id != -1) {
> + ret = qcom_icc_rpm_smd_send(rpm_ctx,
> + RPM_BUS_SLAVE_REQ,
> + qn->slv_rpm_id,
> + bw_bps);
> + if (ret) {
> + pr_err("qcom_icc_rpm_smd_send slv %d error %d\n",
> + qn->slv_rpm_id, ret);
> + return ret;
> + }
> }
> }
>
> @@ -337,7 +342,6 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
> struct qcom_icc_provider *qp;
> struct qcom_icc_node *src_qn = NULL, *dst_qn = NULL;
> struct icc_provider *provider;
> - u64 sum_bw;
> u64 active_rate, sleep_rate;
> u64 agg_avg[QCOM_SMD_RPM_STATE_NUM], agg_peak[QCOM_SMD_RPM_STATE_NUM];
> u64 max_agg_avg;
> @@ -351,14 +355,12 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
>
> qcom_icc_bus_aggregate(provider, agg_avg, agg_peak, &max_agg_avg);
>
> - sum_bw = icc_units_to_bps(max_agg_avg);
> -
> - ret = qcom_icc_rpm_set(src_qn, sum_bw);
> + ret = qcom_icc_rpm_set(src_qn, agg_avg);
> if (ret)
> return ret;
>
> if (dst_qn) {
> - ret = qcom_icc_rpm_set(dst_qn, sum_bw);
> + ret = qcom_icc_rpm_set(dst_qn, agg_avg);
> if (ret)
> return ret;
> }
>
> --
> 2.41.0
>
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 20/22] interconnect: qcom: icc-rpm: Set bandwidth on both contexts
2023-06-10 18:00 ` Stephan Gerhold
@ 2023-06-10 18:28 ` Konrad Dybcio
2023-06-10 18:43 ` Stephan Gerhold
0 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-10 18:28 UTC (permalink / raw)
To: Stephan Gerhold
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On 10.06.2023 20:00, Stephan Gerhold wrote:
> On Fri, Jun 09, 2023 at 10:19:25PM +0200, Konrad Dybcio wrote:
>> Up until now, for some reason we've only been setting bandwidth values
>> on the active-only context. That pretty much meant that RPM could lift
>> all votes when entering sleep mode. Or never sleep at all.
>>
>> That in turn could potentially break things like USB wakeup, as the
>> connection between APSS and SNoC/PNoC would simply be dead.
>>
>
> Nitpick: Apparently an "active" vote is applied during both active+sleep
> until the first "sleep" vote is sent. It's documented only for
> regulators [1] but I would expect the same applies to the bandwidths.
> This means actual breakage shouldn't have been possible.
..unless some part of the boot chain voted for the sleep set!
I'm not sure whether the regulator comment also holds for bw, but I
also don't really have a great way to check it.. Would you want me to
alter this commit message somehow?
Konrad
>
> The patch itself is still the right thing to do to have the sleep state
> correct during deep cpuidle/suspend.
>
> [1]: https://git.codelinaro.org/clo/la/kernel/msm-3.10/-/blob/LA.BR.1.2.9.c26-04700-8x09.0/drivers/regulator/rpm-smd-regulator.c#L199-209
>
>> Set the values appropriately.
>>
>> Fixes: 30c8fa3ec61a ("interconnect: qcom: Add MSM8916 interconnect provider driver")
>> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
>> ---
>> drivers/interconnect/qcom/icc-rpm.c | 54 +++++++++++++++++++------------------
>> 1 file changed, 28 insertions(+), 26 deletions(-)
>>
>> diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
>> index 3ac47b818afe..ac719013077e 100644
>> --- a/drivers/interconnect/qcom/icc-rpm.c
>> +++ b/drivers/interconnect/qcom/icc-rpm.c
>> @@ -205,34 +205,39 @@ static int qcom_icc_qos_set(struct icc_node *node)
>> }
>> }
>>
>> -static int qcom_icc_rpm_set(struct qcom_icc_node *qn, u64 sum_bw)
>> +static int qcom_icc_rpm_set(struct qcom_icc_node *qn, u64 *bw)
>> {
>> - int ret = 0;
>> + int ret, rpm_ctx = 0;
>> + u64 bw_bps;
>>
>> if (qn->qos.ap_owned)
>> return 0;
>>
>> - if (qn->mas_rpm_id != -1) {
>> - ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
>> - RPM_BUS_MASTER_REQ,
>> - qn->mas_rpm_id,
>> - sum_bw);
>> - if (ret) {
>> - pr_err("qcom_icc_rpm_smd_send mas %d error %d\n",
>> - qn->mas_rpm_id, ret);
>> - return ret;
>> + for (rpm_ctx = 0; rpm_ctx < QCOM_SMD_RPM_STATE_NUM; rpm_ctx++) {
>> + bw_bps = icc_units_to_bps(bw[rpm_ctx]);
>> +
>> + if (qn->mas_rpm_id != -1) {
>> + ret = qcom_icc_rpm_smd_send(rpm_ctx,
>> + RPM_BUS_MASTER_REQ,
>> + qn->mas_rpm_id,
>> + bw_bps);
>> + if (ret) {
>> + pr_err("qcom_icc_rpm_smd_send mas %d error %d\n",
>> + qn->mas_rpm_id, ret);
>> + return ret;
>> + }
>> }
>> - }
>>
>> - if (qn->slv_rpm_id != -1) {
>> - ret = qcom_icc_rpm_smd_send(QCOM_SMD_RPM_ACTIVE_STATE,
>> - RPM_BUS_SLAVE_REQ,
>> - qn->slv_rpm_id,
>> - sum_bw);
>> - if (ret) {
>> - pr_err("qcom_icc_rpm_smd_send slv %d error %d\n",
>> - qn->slv_rpm_id, ret);
>> - return ret;
>> + if (qn->slv_rpm_id != -1) {
>> + ret = qcom_icc_rpm_smd_send(rpm_ctx,
>> + RPM_BUS_SLAVE_REQ,
>> + qn->slv_rpm_id,
>> + bw_bps);
>> + if (ret) {
>> + pr_err("qcom_icc_rpm_smd_send slv %d error %d\n",
>> + qn->slv_rpm_id, ret);
>> + return ret;
>> + }
>> }
>> }
>>
>> @@ -337,7 +342,6 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
>> struct qcom_icc_provider *qp;
>> struct qcom_icc_node *src_qn = NULL, *dst_qn = NULL;
>> struct icc_provider *provider;
>> - u64 sum_bw;
>> u64 active_rate, sleep_rate;
>> u64 agg_avg[QCOM_SMD_RPM_STATE_NUM], agg_peak[QCOM_SMD_RPM_STATE_NUM];
>> u64 max_agg_avg;
>> @@ -351,14 +355,12 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
>>
>> qcom_icc_bus_aggregate(provider, agg_avg, agg_peak, &max_agg_avg);
>>
>> - sum_bw = icc_units_to_bps(max_agg_avg);
>> -
>> - ret = qcom_icc_rpm_set(src_qn, sum_bw);
>> + ret = qcom_icc_rpm_set(src_qn, agg_avg);
>> if (ret)
>> return ret;
>>
>> if (dst_qn) {
>> - ret = qcom_icc_rpm_set(dst_qn, sum_bw);
>> + ret = qcom_icc_rpm_set(dst_qn, agg_avg);
>> if (ret)
>> return ret;
>> }
>>
>> --
>> 2.41.0
>>
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 20/22] interconnect: qcom: icc-rpm: Set bandwidth on both contexts
2023-06-10 18:28 ` Konrad Dybcio
@ 2023-06-10 18:43 ` Stephan Gerhold
0 siblings, 0 replies; 45+ messages in thread
From: Stephan Gerhold @ 2023-06-10 18:43 UTC (permalink / raw)
To: Konrad Dybcio
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On Sat, Jun 10, 2023 at 08:28:22PM +0200, Konrad Dybcio wrote:
>
>
> On 10.06.2023 20:00, Stephan Gerhold wrote:
> > On Fri, Jun 09, 2023 at 10:19:25PM +0200, Konrad Dybcio wrote:
> >> Up until now, for some reason we've only been setting bandwidth values
> >> on the active-only context. That pretty much meant that RPM could lift
> >> all votes when entering sleep mode. Or never sleep at all.
> >>
> >> That in turn could potentially break things like USB wakeup, as the
> >> connection between APSS and SNoC/PNoC would simply be dead.
> >>
> >
> > Nitpick: Apparently an "active" vote is applied during both active+sleep
> > until the first "sleep" vote is sent. It's documented only for
> > regulators [1] but I would expect the same applies to the bandwidths.
> > This means actual breakage shouldn't have been possible.
> ..unless some part of the boot chain voted for the sleep set!
>
> I'm not sure whether the regulator comment also holds for bw, but I
> also don't really have a great way to check it.. Would you want me to
> alter this commit message somehow?
>
Hm. Well, on a second look you used "could" instead of "definitely does"
everywhere in your commit message. There is a indeed a slight chance
so feel free to just keep it as-is. :D
Thanks,
Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 21/22] interconnect: qcom: icc-rpm: Set correct bandwidth through RPM bw req
2023-06-09 20:19 ` [PATCH v2 21/22] interconnect: qcom: icc-rpm: Set correct bandwidth through RPM bw req Konrad Dybcio
@ 2023-06-10 18:46 ` Stephan Gerhold
0 siblings, 0 replies; 45+ messages in thread
From: Stephan Gerhold @ 2023-06-10 18:46 UTC (permalink / raw)
To: Konrad Dybcio
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On Fri, Jun 09, 2023 at 10:19:26PM +0200, Konrad Dybcio wrote:
> Currently, we're setting the aggregated-on-provider bandwidth on each
> node, individually. That is of course incorrect and results in far too
> high votes. Use the correct values to ensure we're not wasting power.
>
> Fixes: 30c8fa3ec61a ("interconnect: qcom: Add MSM8916 interconnect provider driver")
> Reported-by: Stephan Gerhold <stephan@gerhold.net>
> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Reviewed-by: Stephan Gerhold <stephan@gerhold.net>
> ---
> drivers/interconnect/qcom/icc-rpm.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
> index ac719013077e..1508233632f6 100644
> --- a/drivers/interconnect/qcom/icc-rpm.c
> +++ b/drivers/interconnect/qcom/icc-rpm.c
> @@ -355,12 +355,12 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
>
> qcom_icc_bus_aggregate(provider, agg_avg, agg_peak, &max_agg_avg);
>
> - ret = qcom_icc_rpm_set(src_qn, agg_avg);
> + ret = qcom_icc_rpm_set(src_qn, src_qn->sum_avg);
> if (ret)
> return ret;
>
> if (dst_qn) {
> - ret = qcom_icc_rpm_set(dst_qn, agg_avg);
> + ret = qcom_icc_rpm_set(dst_qn, dst_qn->sum_avg);
> if (ret)
> return ret;
> }
>
> --
> 2.41.0
>
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 04/22] clk: qcom: smd-rpm: Export clock scaling availability
2023-06-10 12:15 ` Konrad Dybcio
@ 2023-06-10 18:53 ` Konrad Dybcio
2023-06-10 19:25 ` Stephan Gerhold
0 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-10 18:53 UTC (permalink / raw)
To: Stephan Gerhold
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On 10.06.2023 14:15, Konrad Dybcio wrote:
>
>
> On 10.06.2023 13:35, Stephan Gerhold wrote:
>> On Fri, Jun 09, 2023 at 10:19:09PM +0200, Konrad Dybcio wrote:
>>> Before we issue a call to RPM through clk_smd_rpm_enable_scaling() the
>>> clock rate requests will not be commited in hardware. This poses a
>>> race threat since we're accessing the bus clocks directly from within
>>> the interconnect framework.
>>>
>>> Add a marker to indicate that we're good to go with sending new requests
>>> and export it so that it can be referenced from icc.
>>>
>>> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
>>> ---
>>> drivers/clk/qcom/clk-smd-rpm.c | 9 +++++++++
>>> include/linux/soc/qcom/smd-rpm.h | 2 ++
>>> 2 files changed, 11 insertions(+)
>>>
>>> diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
>>> index 937cb1515968..482fe30ee6f0 100644
>>> --- a/drivers/clk/qcom/clk-smd-rpm.c
>>> +++ b/drivers/clk/qcom/clk-smd-rpm.c
>>> @@ -151,6 +151,7 @@
>>> #define to_clk_smd_rpm(_hw) container_of(_hw, struct clk_smd_rpm, hw)
>>>
>>> static struct qcom_smd_rpm *rpmcc_smd_rpm;
>>> +static bool smd_rpm_clk_scaling;
>>>
>>> struct clk_smd_rpm {
>>> const int rpm_res_type;
>>> @@ -385,6 +386,12 @@ static unsigned long clk_smd_rpm_recalc_rate(struct clk_hw *hw,
>>> return r->rate;
>>> }
>>>
>>> +bool qcom_smd_rpm_scaling_available(void)
>>> +{
>>> + return smd_rpm_clk_scaling;
>>> +}
>>> +EXPORT_SYMBOL_GPL(qcom_smd_rpm_scaling_available);
>>> +
>>> static int clk_smd_rpm_enable_scaling(void)
>>> {
>>> int ret;
>>> @@ -410,6 +417,8 @@ static int clk_smd_rpm_enable_scaling(void)
>>> return ret;
>>> }
>>>
>>> + smd_rpm_clk_scaling = true;
>>> +
>>
>> If you move the platform_device_register_data(&rpdev->dev,
>> "icc_smd_rpm", ...) from drivers/soc/qcom/smd-rpm.c to here you can
>> avoid the race completely and drop this API. I think that would be
>> cleaner. And it will likely probe much faster because probe deferral
>> is slow. :)
> Sounds like an idea.. especially since it's pretty much the only
> dependency other than SMDRPM itself!
It sounds great, but to not break bisecting one has to:
1. change the registration in soc/smd-rpm to store rpm ptr in driver
data, in addition to parent driver data
2. change icc/smd-rpm to use the device and not parent data
3. add a platform_device_register_data call in clk-smd-rpm that will
always fail because the device is always registered
4. remove the registration from soc/smd-rpm
I know you'd love to see me break my [PATCH xx/42] record, but I'd say
that deserves its own series and *this* patch could be a good transition
middleground :P
Hopefully Stephen, Bjorn and Georgi won't send a hitman after me for
abusing atomic cross-subsystem merges..
Konrad
>
> Konrad
>>
>> Thanks,
>> Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 22/22] interconnect: qcom: icc-rpm: Fix bandwidth calculations
2023-06-09 20:19 ` [PATCH v2 22/22] interconnect: qcom: icc-rpm: Fix bandwidth calculations Konrad Dybcio
@ 2023-06-10 19:06 ` Stephan Gerhold
2023-06-10 19:09 ` Konrad Dybcio
0 siblings, 1 reply; 45+ messages in thread
From: Stephan Gerhold @ 2023-06-10 19:06 UTC (permalink / raw)
To: Konrad Dybcio
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On Fri, Jun 09, 2023 at 10:19:27PM +0200, Konrad Dybcio wrote:
> Up until now, we've been aggregating the bandwidth values and only
> dividing them by the bus width of the source node. This was completely
> wrong, as different nodes on a given path may (and usually do) have
> varying bus widths. That in turn, resulted in the calculated clock rates
> being completely bogus - usually they ended up being much higher, as
> NoC_A<->NoC_B links are very wide.
>
> Since we're not using the aggregate bandwidth value for anything other
> than clock rate calculations, remodel qcom_icc_bus_aggregate() to
> calculate the per-context clock rate for a given provider, taking into
> account the bus width of every individual node.
>
> Fixes: 30c8fa3ec61a ("interconnect: qcom: Add MSM8916 interconnect provider driver")
> Reported-by: Stephan Gerhold <stephan@gerhold.net>
> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
> ---
> drivers/interconnect/qcom/icc-rpm.c | 59 ++++++++++++-------------------------
> 1 file changed, 19 insertions(+), 40 deletions(-)
>
> diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
> index 1508233632f6..d177a76abe2a 100644
> --- a/drivers/interconnect/qcom/icc-rpm.c
> +++ b/drivers/interconnect/qcom/icc-rpm.c
> @@ -293,58 +293,44 @@ static int qcom_icc_bw_aggregate(struct icc_node *node, u32 tag, u32 avg_bw,
> }
>
> /**
> - * qcom_icc_bus_aggregate - aggregate bandwidth by traversing all nodes
> + * qcom_icc_bus_aggregate - calculate bus clock rates by traversing all nodes
> * @provider: generic interconnect provider
> - * @agg_avg: an array for aggregated average bandwidth of buckets
> - * @agg_peak: an array for aggregated peak bandwidth of buckets
> - * @max_agg_avg: pointer to max value of aggregated average bandwidth
> + * @agg_clk_rate: array containing the aggregated clock rates in kHz
> */
> -static void qcom_icc_bus_aggregate(struct icc_provider *provider,
> - u64 *agg_avg, u64 *agg_peak,
> - u64 *max_agg_avg)
> +static void qcom_icc_bus_aggregate(struct icc_provider *provider, u64 *agg_clk_rate)
> {
> - struct icc_node *node;
> + u64 agg_avg_rate, agg_rate;
> struct qcom_icc_node *qn;
> - u64 sum_avg[QCOM_SMD_RPM_STATE_NUM];
> + struct icc_node *node;
> int i;
>
> - /* Initialise aggregate values */
> - for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++) {
> - agg_avg[i] = 0;
> - agg_peak[i] = 0;
> - }
> -
> - *max_agg_avg = 0;
> -
> /*
> - * Iterate nodes on the interconnect and aggregate bandwidth
> - * requests for every bucket.
> + * Iterate nodes on the provider, aggregate bandwidth requests for
> + * every bucket and convert them into bus clock rates.
> */
> list_for_each_entry(node, &provider->nodes, node_list) {
> qn = node->data;
> for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++) {
> if (qn->channels)
> - sum_avg[i] = div_u64(qn->sum_avg[i], qn->channels);
> + agg_avg_rate = div_u64(qn->sum_avg[i], qn->channels);
> else
> - sum_avg[i] = qn->sum_avg[i];
> - agg_avg[i] += sum_avg[i];
> - agg_peak[i] = max_t(u64, agg_peak[i], qn->max_peak[i]);
> + agg_avg_rate = qn->sum_avg[i];
> +
> + agg_rate = max_t(u64, agg_avg_rate, qn->max_peak[i]);
> + do_div(agg_rate, qn->buswidth);
> +
> + agg_clk_rate[i] = max_t(u64, agg_clk_rate[i], agg_rate);
> }
> }
> -
> - /* Find maximum values across all buckets */
> - for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++)
> - *max_agg_avg = max_t(u64, *max_agg_avg, agg_avg[i]);
> }
>
> static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
> {
> - struct qcom_icc_provider *qp;
> struct qcom_icc_node *src_qn = NULL, *dst_qn = NULL;
> + u64 agg_clk_rate[QCOM_SMD_RPM_STATE_NUM] = { 0 };
> struct icc_provider *provider;
> + struct qcom_icc_provider *qp;
> u64 active_rate, sleep_rate;
> - u64 agg_avg[QCOM_SMD_RPM_STATE_NUM], agg_peak[QCOM_SMD_RPM_STATE_NUM];
> - u64 max_agg_avg;
> int ret;
>
> src_qn = src->data;
> @@ -353,7 +339,9 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
> provider = src->provider;
> qp = to_qcom_provider(provider);
>
> - qcom_icc_bus_aggregate(provider, agg_avg, agg_peak, &max_agg_avg);
> + qcom_icc_bus_aggregate(provider, agg_clk_rate);
> + active_rate = agg_clk_rate[QCOM_SMD_RPM_ACTIVE_STATE];
> + sleep_rate = agg_clk_rate[QCOM_SMD_RPM_SLEEP_STATE];
>
> ret = qcom_icc_rpm_set(src_qn, src_qn->sum_avg);
> if (ret)
> @@ -369,15 +357,6 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
> if (!qp->bus_clk_desc && !qp->bus_clk)
> return 0;
>
> - /* Intentionally keep the rates in kHz as that's what RPM accepts */
I kind of liked this comment because otherwise it's not obvious why
you're not converting from "ICC units" anywhere.
Anyway:
Reviewed-by: Stephan Gerhold <stephan@gerhold.net>
Thanks for going through the giant maze to fix this!
Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 22/22] interconnect: qcom: icc-rpm: Fix bandwidth calculations
2023-06-10 19:06 ` Stephan Gerhold
@ 2023-06-10 19:09 ` Konrad Dybcio
0 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-10 19:09 UTC (permalink / raw)
To: Stephan Gerhold
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On 10.06.2023 21:06, Stephan Gerhold wrote:
> On Fri, Jun 09, 2023 at 10:19:27PM +0200, Konrad Dybcio wrote:
>> Up until now, we've been aggregating the bandwidth values and only
>> dividing them by the bus width of the source node. This was completely
>> wrong, as different nodes on a given path may (and usually do) have
>> varying bus widths. That in turn, resulted in the calculated clock rates
>> being completely bogus - usually they ended up being much higher, as
>> NoC_A<->NoC_B links are very wide.
>>
>> Since we're not using the aggregate bandwidth value for anything other
>> than clock rate calculations, remodel qcom_icc_bus_aggregate() to
>> calculate the per-context clock rate for a given provider, taking into
>> account the bus width of every individual node.
>>
>> Fixes: 30c8fa3ec61a ("interconnect: qcom: Add MSM8916 interconnect provider driver")
>> Reported-by: Stephan Gerhold <stephan@gerhold.net>
>> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
>> ---
>> drivers/interconnect/qcom/icc-rpm.c | 59 ++++++++++++-------------------------
>> 1 file changed, 19 insertions(+), 40 deletions(-)
>>
>> diff --git a/drivers/interconnect/qcom/icc-rpm.c b/drivers/interconnect/qcom/icc-rpm.c
>> index 1508233632f6..d177a76abe2a 100644
>> --- a/drivers/interconnect/qcom/icc-rpm.c
>> +++ b/drivers/interconnect/qcom/icc-rpm.c
>> @@ -293,58 +293,44 @@ static int qcom_icc_bw_aggregate(struct icc_node *node, u32 tag, u32 avg_bw,
>> }
>>
>> /**
>> - * qcom_icc_bus_aggregate - aggregate bandwidth by traversing all nodes
>> + * qcom_icc_bus_aggregate - calculate bus clock rates by traversing all nodes
>> * @provider: generic interconnect provider
>> - * @agg_avg: an array for aggregated average bandwidth of buckets
>> - * @agg_peak: an array for aggregated peak bandwidth of buckets
>> - * @max_agg_avg: pointer to max value of aggregated average bandwidth
>> + * @agg_clk_rate: array containing the aggregated clock rates in kHz
>> */
>> -static void qcom_icc_bus_aggregate(struct icc_provider *provider,
>> - u64 *agg_avg, u64 *agg_peak,
>> - u64 *max_agg_avg)
>> +static void qcom_icc_bus_aggregate(struct icc_provider *provider, u64 *agg_clk_rate)
>> {
>> - struct icc_node *node;
>> + u64 agg_avg_rate, agg_rate;
>> struct qcom_icc_node *qn;
>> - u64 sum_avg[QCOM_SMD_RPM_STATE_NUM];
>> + struct icc_node *node;
>> int i;
>>
>> - /* Initialise aggregate values */
>> - for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++) {
>> - agg_avg[i] = 0;
>> - agg_peak[i] = 0;
>> - }
>> -
>> - *max_agg_avg = 0;
>> -
>> /*
>> - * Iterate nodes on the interconnect and aggregate bandwidth
>> - * requests for every bucket.
>> + * Iterate nodes on the provider, aggregate bandwidth requests for
>> + * every bucket and convert them into bus clock rates.
>> */
>> list_for_each_entry(node, &provider->nodes, node_list) {
>> qn = node->data;
>> for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++) {
>> if (qn->channels)
>> - sum_avg[i] = div_u64(qn->sum_avg[i], qn->channels);
>> + agg_avg_rate = div_u64(qn->sum_avg[i], qn->channels);
>> else
>> - sum_avg[i] = qn->sum_avg[i];
>> - agg_avg[i] += sum_avg[i];
>> - agg_peak[i] = max_t(u64, agg_peak[i], qn->max_peak[i]);
>> + agg_avg_rate = qn->sum_avg[i];
>> +
>> + agg_rate = max_t(u64, agg_avg_rate, qn->max_peak[i]);
>> + do_div(agg_rate, qn->buswidth);
>> +
>> + agg_clk_rate[i] = max_t(u64, agg_clk_rate[i], agg_rate);
>> }
>> }
>> -
>> - /* Find maximum values across all buckets */
>> - for (i = 0; i < QCOM_SMD_RPM_STATE_NUM; i++)
>> - *max_agg_avg = max_t(u64, *max_agg_avg, agg_avg[i]);
>> }
>>
>> static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
>> {
>> - struct qcom_icc_provider *qp;
>> struct qcom_icc_node *src_qn = NULL, *dst_qn = NULL;
>> + u64 agg_clk_rate[QCOM_SMD_RPM_STATE_NUM] = { 0 };
>> struct icc_provider *provider;
>> + struct qcom_icc_provider *qp;
>> u64 active_rate, sleep_rate;
>> - u64 agg_avg[QCOM_SMD_RPM_STATE_NUM], agg_peak[QCOM_SMD_RPM_STATE_NUM];
>> - u64 max_agg_avg;
>> int ret;
>>
>> src_qn = src->data;
>> @@ -353,7 +339,9 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
>> provider = src->provider;
>> qp = to_qcom_provider(provider);
>>
>> - qcom_icc_bus_aggregate(provider, agg_avg, agg_peak, &max_agg_avg);
>> + qcom_icc_bus_aggregate(provider, agg_clk_rate);
>> + active_rate = agg_clk_rate[QCOM_SMD_RPM_ACTIVE_STATE];
>> + sleep_rate = agg_clk_rate[QCOM_SMD_RPM_SLEEP_STATE];
>>
>> ret = qcom_icc_rpm_set(src_qn, src_qn->sum_avg);
>> if (ret)
>> @@ -369,15 +357,6 @@ static int qcom_icc_set(struct icc_node *src, struct icc_node *dst)
>> if (!qp->bus_clk_desc && !qp->bus_clk)
>> return 0;
>>
>> - /* Intentionally keep the rates in kHz as that's what RPM accepts */
>
> I kind of liked this comment because otherwise it's not obvious why
> you're not converting from "ICC units" anywhere.
I figured the kerneldoc change would be enough:
@agg_clk_rate: array containing the aggregated clock rates in kHz
>
> Anyway:
>
> Reviewed-by: Stephan Gerhold <stephan@gerhold.net>
>
> Thanks for going through the giant maze to fix this!
Thanks for being there along the way.. This spaghetti is far too much
for a single human..
Konrad
>
> Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 04/22] clk: qcom: smd-rpm: Export clock scaling availability
2023-06-10 18:53 ` Konrad Dybcio
@ 2023-06-10 19:25 ` Stephan Gerhold
2023-06-10 19:39 ` Konrad Dybcio
0 siblings, 1 reply; 45+ messages in thread
From: Stephan Gerhold @ 2023-06-10 19:25 UTC (permalink / raw)
To: Konrad Dybcio
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On Sat, Jun 10, 2023 at 08:53:05PM +0200, Konrad Dybcio wrote:
> On 10.06.2023 14:15, Konrad Dybcio wrote:
> > On 10.06.2023 13:35, Stephan Gerhold wrote:
> >> On Fri, Jun 09, 2023 at 10:19:09PM +0200, Konrad Dybcio wrote:
> >>> Before we issue a call to RPM through clk_smd_rpm_enable_scaling() the
> >>> clock rate requests will not be commited in hardware. This poses a
> >>> race threat since we're accessing the bus clocks directly from within
> >>> the interconnect framework.
> >>>
> >>> Add a marker to indicate that we're good to go with sending new requests
> >>> and export it so that it can be referenced from icc.
> >>>
> >>> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
> >>> ---
> >>> drivers/clk/qcom/clk-smd-rpm.c | 9 +++++++++
> >>> include/linux/soc/qcom/smd-rpm.h | 2 ++
> >>> 2 files changed, 11 insertions(+)
> >>>
> >>> diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
> >>> index 937cb1515968..482fe30ee6f0 100644
> >>> --- a/drivers/clk/qcom/clk-smd-rpm.c
> >>> +++ b/drivers/clk/qcom/clk-smd-rpm.c
> >>> @@ -151,6 +151,7 @@
> >>> #define to_clk_smd_rpm(_hw) container_of(_hw, struct clk_smd_rpm, hw)
> >>>
> >>> static struct qcom_smd_rpm *rpmcc_smd_rpm;
> >>> +static bool smd_rpm_clk_scaling;
> >>>
> >>> struct clk_smd_rpm {
> >>> const int rpm_res_type;
> >>> @@ -385,6 +386,12 @@ static unsigned long clk_smd_rpm_recalc_rate(struct clk_hw *hw,
> >>> return r->rate;
> >>> }
> >>>
> >>> +bool qcom_smd_rpm_scaling_available(void)
> >>> +{
> >>> + return smd_rpm_clk_scaling;
> >>> +}
> >>> +EXPORT_SYMBOL_GPL(qcom_smd_rpm_scaling_available);
> >>> +
> >>> static int clk_smd_rpm_enable_scaling(void)
> >>> {
> >>> int ret;
> >>> @@ -410,6 +417,8 @@ static int clk_smd_rpm_enable_scaling(void)
> >>> return ret;
> >>> }
> >>>
> >>> + smd_rpm_clk_scaling = true;
> >>> +
> >>
> >> If you move the platform_device_register_data(&rpdev->dev,
> >> "icc_smd_rpm", ...) from drivers/soc/qcom/smd-rpm.c to here you can
> >> avoid the race completely and drop this API. I think that would be
> >> cleaner. And it will likely probe much faster because probe deferral
> >> is slow. :)
> > Sounds like an idea.. especially since it's pretty much the only
> > dependency other than SMDRPM itself!
> It sounds great, but to not break bisecting one has to:
>
> 1. change the registration in soc/smd-rpm to store rpm ptr in driver
> data, in addition to parent driver data
>
> 2. change icc/smd-rpm to use the device and not parent data
>
> 3. add a platform_device_register_data call in clk-smd-rpm that will
> always fail because the device is always registered
>
> 4. remove the registration from soc/smd-rpm
>
Logically the icc_smd_rpm device still fits better as child of
smd-rpm and not clk-smd-rpm. So I would probably just continue
registering it on the parent device from clk-smd-rpm.
Then there are no changes necessary in icc_smd_rpm.
You could use this. Both touched files are Bjorn-maintained so should be
manageable to have it in one commit. (note: compile-tested only)
Thanks,
Stephan
From a2610adb2551b01e76b9de8e4cbcc89853814a8f Mon Sep 17 00:00:00 2001
From: Stephan Gerhold <stephan@gerhold.net>
Date: Sat, 10 Jun 2023 21:19:48 +0200
Subject: [PATCH] soc: qcom: smd-rpm: Move icc_smd_rpm registration to
clk-smd-rpm
icc_smd_rpm will do bus clock votes itself rather than taking the
unnecessary detour through the clock subsystem. However, it can only
do that after the clocks have been handed off and scaling has been
enabled in the RPM in clk-smd-rpm.
Move the icc_smd_rpm registration from smd-rpm.c to clk-smd-rpm.c
to avoid any possible races. icc_smd_rpm gets the driver data from
the smd-rpm device, so still register the platform device on the
smd-rpm parent device.
Signed-off-by: Stephan Gerhold <stephan@gerhold.net>
---
drivers/clk/qcom/clk-smd-rpm.c | 21 +++++++++++++++++++++
drivers/soc/qcom/smd-rpm.c | 23 +----------------------
2 files changed, 22 insertions(+), 22 deletions(-)
diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
index e4de74b68797..91adb16889b3 100644
--- a/drivers/clk/qcom/clk-smd-rpm.c
+++ b/drivers/clk/qcom/clk-smd-rpm.c
@@ -1302,12 +1302,20 @@ static struct clk_hw *qcom_smdrpm_clk_hw_get(struct of_phandle_args *clkspec,
return desc->clks[idx] ? &desc->clks[idx]->hw : ERR_PTR(-ENOENT);
}
+static void rpm_smd_unregister_icc(void *data)
+{
+ struct platform_device *icc_pdev = data;
+
+ platform_device_unregister(icc_pdev);
+}
+
static int rpm_smd_clk_probe(struct platform_device *pdev)
{
int ret;
size_t num_clks, i;
struct clk_smd_rpm **rpm_smd_clks;
const struct rpm_smd_clk_desc *desc;
+ struct platform_device *icc_pdev;
rpmcc_smd_rpm = dev_get_drvdata(pdev->dev.parent);
if (!rpmcc_smd_rpm) {
@@ -1357,6 +1365,19 @@ static int rpm_smd_clk_probe(struct platform_device *pdev)
if (ret)
goto err;
+ icc_pdev = platform_device_register_data(pdev->dev.parent,
+ "icc_smd_rpm", -1, NULL, 0);
+ if (IS_ERR(icc_pdev)) {
+ dev_err(&pdev->dev, "Failed to register icc_smd_rpm device: %pE\n",
+ icc_pdev);
+ /* No need to unregister clocks because of this */
+ } else {
+ ret = devm_add_action_or_reset(&pdev->dev, rpm_smd_unregister_icc,
+ icc_pdev);
+ if (ret)
+ goto err;
+ }
+
return 0;
err:
dev_err(&pdev->dev, "Error registering SMD clock driver (%d)\n", ret);
diff --git a/drivers/soc/qcom/smd-rpm.c b/drivers/soc/qcom/smd-rpm.c
index 0c1aa809cc4e..427dd5392b82 100644
--- a/drivers/soc/qcom/smd-rpm.c
+++ b/drivers/soc/qcom/smd-rpm.c
@@ -19,7 +19,6 @@
/**
* struct qcom_smd_rpm - state of the rpm device driver
* @rpm_channel: reference to the smd channel
- * @icc: interconnect proxy device
* @dev: rpm device
* @ack: completion for acks
* @lock: mutual exclusion around the send/complete pair
@@ -27,7 +26,6 @@
*/
struct qcom_smd_rpm {
struct rpmsg_endpoint *rpm_channel;
- struct platform_device *icc;
struct device *dev;
struct completion ack;
@@ -197,7 +195,6 @@ static int qcom_smd_rpm_callback(struct rpmsg_device *rpdev,
static int qcom_smd_rpm_probe(struct rpmsg_device *rpdev)
{
struct qcom_smd_rpm *rpm;
- int ret;
rpm = devm_kzalloc(&rpdev->dev, sizeof(*rpm), GFP_KERNEL);
if (!rpm)
@@ -210,24 +207,7 @@ static int qcom_smd_rpm_probe(struct rpmsg_device *rpdev)
rpm->rpm_channel = rpdev->ept;
dev_set_drvdata(&rpdev->dev, rpm);
- rpm->icc = platform_device_register_data(&rpdev->dev, "icc_smd_rpm", -1,
- NULL, 0);
- if (IS_ERR(rpm->icc))
- return PTR_ERR(rpm->icc);
-
- ret = of_platform_populate(rpdev->dev.of_node, NULL, NULL, &rpdev->dev);
- if (ret)
- platform_device_unregister(rpm->icc);
-
- return ret;
-}
-
-static void qcom_smd_rpm_remove(struct rpmsg_device *rpdev)
-{
- struct qcom_smd_rpm *rpm = dev_get_drvdata(&rpdev->dev);
-
- platform_device_unregister(rpm->icc);
- of_platform_depopulate(&rpdev->dev);
+ return devm_of_platform_populate(&rpdev->dev);
}
static const struct of_device_id qcom_smd_rpm_of_match[] = {
@@ -256,7 +236,6 @@ MODULE_DEVICE_TABLE(of, qcom_smd_rpm_of_match);
static struct rpmsg_driver qcom_smd_rpm_driver = {
.probe = qcom_smd_rpm_probe,
- .remove = qcom_smd_rpm_remove,
.callback = qcom_smd_rpm_callback,
.drv = {
.name = "qcom_smd_rpm",
--
2.40.1
^ permalink raw reply related [flat|nested] 45+ messages in thread
* Re: [PATCH v2 04/22] clk: qcom: smd-rpm: Export clock scaling availability
2023-06-10 19:25 ` Stephan Gerhold
@ 2023-06-10 19:39 ` Konrad Dybcio
2023-06-11 9:20 ` Stephan Gerhold
0 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-10 19:39 UTC (permalink / raw)
To: Stephan Gerhold
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On 10.06.2023 21:25, Stephan Gerhold wrote:
> On Sat, Jun 10, 2023 at 08:53:05PM +0200, Konrad Dybcio wrote:
>> On 10.06.2023 14:15, Konrad Dybcio wrote:
>>> On 10.06.2023 13:35, Stephan Gerhold wrote:
>>>> On Fri, Jun 09, 2023 at 10:19:09PM +0200, Konrad Dybcio wrote:
>>>>> Before we issue a call to RPM through clk_smd_rpm_enable_scaling() the
>>>>> clock rate requests will not be commited in hardware. This poses a
>>>>> race threat since we're accessing the bus clocks directly from within
>>>>> the interconnect framework.
>>>>>
>>>>> Add a marker to indicate that we're good to go with sending new requests
>>>>> and export it so that it can be referenced from icc.
>>>>>
>>>>> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
>>>>> ---
>>>>> drivers/clk/qcom/clk-smd-rpm.c | 9 +++++++++
>>>>> include/linux/soc/qcom/smd-rpm.h | 2 ++
>>>>> 2 files changed, 11 insertions(+)
>>>>>
>>>>> diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
>>>>> index 937cb1515968..482fe30ee6f0 100644
>>>>> --- a/drivers/clk/qcom/clk-smd-rpm.c
>>>>> +++ b/drivers/clk/qcom/clk-smd-rpm.c
>>>>> @@ -151,6 +151,7 @@
>>>>> #define to_clk_smd_rpm(_hw) container_of(_hw, struct clk_smd_rpm, hw)
>>>>>
>>>>> static struct qcom_smd_rpm *rpmcc_smd_rpm;
>>>>> +static bool smd_rpm_clk_scaling;
>>>>>
>>>>> struct clk_smd_rpm {
>>>>> const int rpm_res_type;
>>>>> @@ -385,6 +386,12 @@ static unsigned long clk_smd_rpm_recalc_rate(struct clk_hw *hw,
>>>>> return r->rate;
>>>>> }
>>>>>
>>>>> +bool qcom_smd_rpm_scaling_available(void)
>>>>> +{
>>>>> + return smd_rpm_clk_scaling;
>>>>> +}
>>>>> +EXPORT_SYMBOL_GPL(qcom_smd_rpm_scaling_available);
>>>>> +
>>>>> static int clk_smd_rpm_enable_scaling(void)
>>>>> {
>>>>> int ret;
>>>>> @@ -410,6 +417,8 @@ static int clk_smd_rpm_enable_scaling(void)
>>>>> return ret;
>>>>> }
>>>>>
>>>>> + smd_rpm_clk_scaling = true;
>>>>> +
>>>>
>>>> If you move the platform_device_register_data(&rpdev->dev,
>>>> "icc_smd_rpm", ...) from drivers/soc/qcom/smd-rpm.c to here you can
>>>> avoid the race completely and drop this API. I think that would be
>>>> cleaner. And it will likely probe much faster because probe deferral
>>>> is slow. :)
>>> Sounds like an idea.. especially since it's pretty much the only
>>> dependency other than SMDRPM itself!
>> It sounds great, but to not break bisecting one has to:
>>
>> 1. change the registration in soc/smd-rpm to store rpm ptr in driver
>> data, in addition to parent driver data
>>
>> 2. change icc/smd-rpm to use the device and not parent data
>>
>> 3. add a platform_device_register_data call in clk-smd-rpm that will
>> always fail because the device is always registered
>>
>> 4. remove the registration from soc/smd-rpm
>>
>
> Logically the icc_smd_rpm device still fits better as child of
> smd-rpm and not clk-smd-rpm. So I would probably just continue
> registering it on the parent device from clk-smd-rpm.
> Then there are no changes necessary in icc_smd_rpm.
>
> You could use this. Both touched files are Bjorn-maintained so should be
> manageable to have it in one commit. (note: compile-tested only)
>
> Thanks,
> Stephan
>
> From a2610adb2551b01e76b9de8e4cbcc89853814a8f Mon Sep 17 00:00:00 2001
> From: Stephan Gerhold <stephan@gerhold.net>
> Date: Sat, 10 Jun 2023 21:19:48 +0200
> Subject: [PATCH] soc: qcom: smd-rpm: Move icc_smd_rpm registration to
> clk-smd-rpm
>
> icc_smd_rpm will do bus clock votes itself rather than taking the
> unnecessary detour through the clock subsystem. However, it can only
> do that after the clocks have been handed off and scaling has been
> enabled in the RPM in clk-smd-rpm.
>
> Move the icc_smd_rpm registration from smd-rpm.c to clk-smd-rpm.c
> to avoid any possible races. icc_smd_rpm gets the driver data from
> the smd-rpm device, so still register the platform device on the
> smd-rpm parent device.
>
> Signed-off-by: Stephan Gerhold <stephan@gerhold.net>
> ---
Generally it looks good.. I'll give it a spin next week. One
thing below.
> drivers/clk/qcom/clk-smd-rpm.c | 21 +++++++++++++++++++++
> drivers/soc/qcom/smd-rpm.c | 23 +----------------------
> 2 files changed, 22 insertions(+), 22 deletions(-)
>
> diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
> index e4de74b68797..91adb16889b3 100644
> --- a/drivers/clk/qcom/clk-smd-rpm.c
> +++ b/drivers/clk/qcom/clk-smd-rpm.c
> @@ -1302,12 +1302,20 @@ static struct clk_hw *qcom_smdrpm_clk_hw_get(struct of_phandle_args *clkspec,
> return desc->clks[idx] ? &desc->clks[idx]->hw : ERR_PTR(-ENOENT);
> }
>
> +static void rpm_smd_unregister_icc(void *data)
> +{
> + struct platform_device *icc_pdev = data;
> +
> + platform_device_unregister(icc_pdev);
> +}
> +
> static int rpm_smd_clk_probe(struct platform_device *pdev)
> {
> int ret;
> size_t num_clks, i;
> struct clk_smd_rpm **rpm_smd_clks;
> const struct rpm_smd_clk_desc *desc;
> + struct platform_device *icc_pdev;
>
> rpmcc_smd_rpm = dev_get_drvdata(pdev->dev.parent);
> if (!rpmcc_smd_rpm) {
> @@ -1357,6 +1365,19 @@ static int rpm_smd_clk_probe(struct platform_device *pdev)
> if (ret)
> goto err;
>
> + icc_pdev = platform_device_register_data(pdev->dev.parent,
> + "icc_smd_rpm", -1, NULL, 0);
> + if (IS_ERR(icc_pdev)) {
> + dev_err(&pdev->dev, "Failed to register icc_smd_rpm device: %pE\n",
> + icc_pdev);
> + /* No need to unregister clocks because of this */
> + } else {
> + ret = devm_add_action_or_reset(&pdev->dev, rpm_smd_unregister_icc,
> + icc_pdev);
> + if (ret)
> + goto err;
> + }
> +
> return 0;
> err:
> dev_err(&pdev->dev, "Error registering SMD clock driver (%d)\n", ret);
> diff --git a/drivers/soc/qcom/smd-rpm.c b/drivers/soc/qcom/smd-rpm.c
> index 0c1aa809cc4e..427dd5392b82 100644
> --- a/drivers/soc/qcom/smd-rpm.c
> +++ b/drivers/soc/qcom/smd-rpm.c
> @@ -19,7 +19,6 @@
> /**
> * struct qcom_smd_rpm - state of the rpm device driver
> * @rpm_channel: reference to the smd channel
> - * @icc: interconnect proxy device
> * @dev: rpm device
> * @ack: completion for acks
> * @lock: mutual exclusion around the send/complete pair
> @@ -27,7 +26,6 @@
> */
> struct qcom_smd_rpm {
> struct rpmsg_endpoint *rpm_channel;
> - struct platform_device *icc;
> struct device *dev;
>
> struct completion ack;
> @@ -197,7 +195,6 @@ static int qcom_smd_rpm_callback(struct rpmsg_device *rpdev,
> static int qcom_smd_rpm_probe(struct rpmsg_device *rpdev)
> {
> struct qcom_smd_rpm *rpm;
> - int ret;
>
> rpm = devm_kzalloc(&rpdev->dev, sizeof(*rpm), GFP_KERNEL);
> if (!rpm)
> @@ -210,24 +207,7 @@ static int qcom_smd_rpm_probe(struct rpmsg_device *rpdev)
> rpm->rpm_channel = rpdev->ept;
> dev_set_drvdata(&rpdev->dev, rpm);
>
> - rpm->icc = platform_device_register_data(&rpdev->dev, "icc_smd_rpm", -1,
> - NULL, 0);
> - if (IS_ERR(rpm->icc))
> - return PTR_ERR(rpm->icc);
> -
> - ret = of_platform_populate(rpdev->dev.of_node, NULL, NULL, &rpdev->dev);
> - if (ret)
> - platform_device_unregister(rpm->icc);
> -
> - return ret;
> -}
> -
> -static void qcom_smd_rpm_remove(struct rpmsg_device *rpdev)
> -{
> - struct qcom_smd_rpm *rpm = dev_get_drvdata(&rpdev->dev);
> -
> - platform_device_unregister(rpm->icc);
> - of_platform_depopulate(&rpdev->dev);
> + return devm_of_platform_populate(&rpdev->dev);
> }
>
> static const struct of_device_id qcom_smd_rpm_of_match[] = {
> @@ -256,7 +236,6 @@ MODULE_DEVICE_TABLE(of, qcom_smd_rpm_of_match);
>
> static struct rpmsg_driver qcom_smd_rpm_driver = {
> .probe = qcom_smd_rpm_probe,
> - .remove = qcom_smd_rpm_remove,
This reaches over the removal of the icc registration, the depopulate
call should stay.
Konrad
> .callback = qcom_smd_rpm_callback,
> .drv = {
> .name = "qcom_smd_rpm",
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 04/22] clk: qcom: smd-rpm: Export clock scaling availability
2023-06-10 19:39 ` Konrad Dybcio
@ 2023-06-11 9:20 ` Stephan Gerhold
2023-06-12 12:51 ` Konrad Dybcio
0 siblings, 1 reply; 45+ messages in thread
From: Stephan Gerhold @ 2023-06-11 9:20 UTC (permalink / raw)
To: Konrad Dybcio
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On Sat, Jun 10, 2023 at 09:39:00PM +0200, Konrad Dybcio wrote:
> On 10.06.2023 21:25, Stephan Gerhold wrote:
> > On Sat, Jun 10, 2023 at 08:53:05PM +0200, Konrad Dybcio wrote:
> >> On 10.06.2023 14:15, Konrad Dybcio wrote:
> >>> On 10.06.2023 13:35, Stephan Gerhold wrote:
> >>>> On Fri, Jun 09, 2023 at 10:19:09PM +0200, Konrad Dybcio wrote:
> >>>>> Before we issue a call to RPM through clk_smd_rpm_enable_scaling() the
> >>>>> clock rate requests will not be commited in hardware. This poses a
> >>>>> race threat since we're accessing the bus clocks directly from within
> >>>>> the interconnect framework.
> >>>>>
> >>>>> Add a marker to indicate that we're good to go with sending new requests
> >>>>> and export it so that it can be referenced from icc.
> >>>>>
> >>>>> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
> >>>>> ---
> >>>>> drivers/clk/qcom/clk-smd-rpm.c | 9 +++++++++
> >>>>> include/linux/soc/qcom/smd-rpm.h | 2 ++
> >>>>> 2 files changed, 11 insertions(+)
> >>>>>
> >>>>> diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
> >>>>> index 937cb1515968..482fe30ee6f0 100644
> >>>>> --- a/drivers/clk/qcom/clk-smd-rpm.c
> >>>>> +++ b/drivers/clk/qcom/clk-smd-rpm.c
> >>>>> @@ -151,6 +151,7 @@
> >>>>> #define to_clk_smd_rpm(_hw) container_of(_hw, struct clk_smd_rpm, hw)
> >>>>>
> >>>>> static struct qcom_smd_rpm *rpmcc_smd_rpm;
> >>>>> +static bool smd_rpm_clk_scaling;
> >>>>>
> >>>>> struct clk_smd_rpm {
> >>>>> const int rpm_res_type;
> >>>>> @@ -385,6 +386,12 @@ static unsigned long clk_smd_rpm_recalc_rate(struct clk_hw *hw,
> >>>>> return r->rate;
> >>>>> }
> >>>>>
> >>>>> +bool qcom_smd_rpm_scaling_available(void)
> >>>>> +{
> >>>>> + return smd_rpm_clk_scaling;
> >>>>> +}
> >>>>> +EXPORT_SYMBOL_GPL(qcom_smd_rpm_scaling_available);
> >>>>> +
> >>>>> static int clk_smd_rpm_enable_scaling(void)
> >>>>> {
> >>>>> int ret;
> >>>>> @@ -410,6 +417,8 @@ static int clk_smd_rpm_enable_scaling(void)
> >>>>> return ret;
> >>>>> }
> >>>>>
> >>>>> + smd_rpm_clk_scaling = true;
> >>>>> +
> >>>>
> >>>> If you move the platform_device_register_data(&rpdev->dev,
> >>>> "icc_smd_rpm", ...) from drivers/soc/qcom/smd-rpm.c to here you can
> >>>> avoid the race completely and drop this API. I think that would be
> >>>> cleaner. And it will likely probe much faster because probe deferral
> >>>> is slow. :)
> >>> Sounds like an idea.. especially since it's pretty much the only
> >>> dependency other than SMDRPM itself!
> >> It sounds great, but to not break bisecting one has to:
> >>
> >> 1. change the registration in soc/smd-rpm to store rpm ptr in driver
> >> data, in addition to parent driver data
> >>
> >> 2. change icc/smd-rpm to use the device and not parent data
> >>
> >> 3. add a platform_device_register_data call in clk-smd-rpm that will
> >> always fail because the device is always registered
> >>
> >> 4. remove the registration from soc/smd-rpm
> >>
> >
> > Logically the icc_smd_rpm device still fits better as child of
> > smd-rpm and not clk-smd-rpm. So I would probably just continue
> > registering it on the parent device from clk-smd-rpm.
> > Then there are no changes necessary in icc_smd_rpm.
> >
> > You could use this. Both touched files are Bjorn-maintained so should be
> > manageable to have it in one commit. (note: compile-tested only)
> >
> > Thanks,
> > Stephan
> >
> > From a2610adb2551b01e76b9de8e4cbcc89853814a8f Mon Sep 17 00:00:00 2001
> > From: Stephan Gerhold <stephan@gerhold.net>
> > Date: Sat, 10 Jun 2023 21:19:48 +0200
> > Subject: [PATCH] soc: qcom: smd-rpm: Move icc_smd_rpm registration to
> > clk-smd-rpm
> >
> > icc_smd_rpm will do bus clock votes itself rather than taking the
> > unnecessary detour through the clock subsystem. However, it can only
> > do that after the clocks have been handed off and scaling has been
> > enabled in the RPM in clk-smd-rpm.
> >
> > Move the icc_smd_rpm registration from smd-rpm.c to clk-smd-rpm.c
> > to avoid any possible races. icc_smd_rpm gets the driver data from
> > the smd-rpm device, so still register the platform device on the
> > smd-rpm parent device.
> >
> > Signed-off-by: Stephan Gerhold <stephan@gerhold.net>
> > ---
> Generally it looks good.. I'll give it a spin next week. One
> thing below.
>
> > drivers/clk/qcom/clk-smd-rpm.c | 21 +++++++++++++++++++++
> > drivers/soc/qcom/smd-rpm.c | 23 +----------------------
> > 2 files changed, 22 insertions(+), 22 deletions(-)
> >
> > diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
> > index e4de74b68797..91adb16889b3 100644
> > --- a/drivers/clk/qcom/clk-smd-rpm.c
> > +++ b/drivers/clk/qcom/clk-smd-rpm.c
> > @@ -1302,12 +1302,20 @@ static struct clk_hw *qcom_smdrpm_clk_hw_get(struct of_phandle_args *clkspec,
> > return desc->clks[idx] ? &desc->clks[idx]->hw : ERR_PTR(-ENOENT);
> > }
> >
> > +static void rpm_smd_unregister_icc(void *data)
> > +{
> > + struct platform_device *icc_pdev = data;
> > +
> > + platform_device_unregister(icc_pdev);
> > +}
> > +
> > static int rpm_smd_clk_probe(struct platform_device *pdev)
> > {
> > int ret;
> > size_t num_clks, i;
> > struct clk_smd_rpm **rpm_smd_clks;
> > const struct rpm_smd_clk_desc *desc;
> > + struct platform_device *icc_pdev;
> >
> > rpmcc_smd_rpm = dev_get_drvdata(pdev->dev.parent);
> > if (!rpmcc_smd_rpm) {
> > @@ -1357,6 +1365,19 @@ static int rpm_smd_clk_probe(struct platform_device *pdev)
> > if (ret)
> > goto err;
> >
> > + icc_pdev = platform_device_register_data(pdev->dev.parent,
> > + "icc_smd_rpm", -1, NULL, 0);
> > + if (IS_ERR(icc_pdev)) {
> > + dev_err(&pdev->dev, "Failed to register icc_smd_rpm device: %pE\n",
> > + icc_pdev);
> > + /* No need to unregister clocks because of this */
> > + } else {
> > + ret = devm_add_action_or_reset(&pdev->dev, rpm_smd_unregister_icc,
> > + icc_pdev);
> > + if (ret)
> > + goto err;
> > + }
> > +
> > return 0;
> > err:
> > dev_err(&pdev->dev, "Error registering SMD clock driver (%d)\n", ret);
> > diff --git a/drivers/soc/qcom/smd-rpm.c b/drivers/soc/qcom/smd-rpm.c
> > index 0c1aa809cc4e..427dd5392b82 100644
> > --- a/drivers/soc/qcom/smd-rpm.c
> > +++ b/drivers/soc/qcom/smd-rpm.c
> > @@ -19,7 +19,6 @@
> > /**
> > * struct qcom_smd_rpm - state of the rpm device driver
> > * @rpm_channel: reference to the smd channel
> > - * @icc: interconnect proxy device
> > * @dev: rpm device
> > * @ack: completion for acks
> > * @lock: mutual exclusion around the send/complete pair
> > @@ -27,7 +26,6 @@
> > */
> > struct qcom_smd_rpm {
> > struct rpmsg_endpoint *rpm_channel;
> > - struct platform_device *icc;
> > struct device *dev;
> >
> > struct completion ack;
> > @@ -197,7 +195,6 @@ static int qcom_smd_rpm_callback(struct rpmsg_device *rpdev,
> > static int qcom_smd_rpm_probe(struct rpmsg_device *rpdev)
> > {
> > struct qcom_smd_rpm *rpm;
> > - int ret;
> >
> > rpm = devm_kzalloc(&rpdev->dev, sizeof(*rpm), GFP_KERNEL);
> > if (!rpm)
> > @@ -210,24 +207,7 @@ static int qcom_smd_rpm_probe(struct rpmsg_device *rpdev)
> > rpm->rpm_channel = rpdev->ept;
> > dev_set_drvdata(&rpdev->dev, rpm);
> >
> > - rpm->icc = platform_device_register_data(&rpdev->dev, "icc_smd_rpm", -1,
> > - NULL, 0);
> > - if (IS_ERR(rpm->icc))
> > - return PTR_ERR(rpm->icc);
> > -
> > - ret = of_platform_populate(rpdev->dev.of_node, NULL, NULL, &rpdev->dev);
> > - if (ret)
> > - platform_device_unregister(rpm->icc);
> > -
> > - return ret;
> > -}
> > -
> > -static void qcom_smd_rpm_remove(struct rpmsg_device *rpdev)
> > -{
> > - struct qcom_smd_rpm *rpm = dev_get_drvdata(&rpdev->dev);
> > -
> > - platform_device_unregister(rpm->icc);
> > - of_platform_depopulate(&rpdev->dev);
> > + return devm_of_platform_populate(&rpdev->dev);
> > }
> >
> > static const struct of_device_id qcom_smd_rpm_of_match[] = {
> > @@ -256,7 +236,6 @@ MODULE_DEVICE_TABLE(of, qcom_smd_rpm_of_match);
> >
> > static struct rpmsg_driver qcom_smd_rpm_driver = {
> > .probe = qcom_smd_rpm_probe,
> > - .remove = qcom_smd_rpm_remove,
> This reaches over the removal of the icc registration, the depopulate
> call should stay.
>
I switched the of_platform_populate() to devm_of_platform_populate(),
that's why the remove callback is no longer necessary. It's a bit
hidden, perhaps it would be enough to add to the commit message:
"While at it, switch the remaining of_platform_populate() call to the
devm variant and remove the remove callback."
Or maybe it should be split into two patches.
Thanks,
Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 04/22] clk: qcom: smd-rpm: Export clock scaling availability
2023-06-11 9:20 ` Stephan Gerhold
@ 2023-06-12 12:51 ` Konrad Dybcio
2023-06-12 17:03 ` Konrad Dybcio
0 siblings, 1 reply; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-12 12:51 UTC (permalink / raw)
To: Stephan Gerhold
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On 11.06.2023 11:20, Stephan Gerhold wrote:
> On Sat, Jun 10, 2023 at 09:39:00PM +0200, Konrad Dybcio wrote:
>> On 10.06.2023 21:25, Stephan Gerhold wrote:
>>> On Sat, Jun 10, 2023 at 08:53:05PM +0200, Konrad Dybcio wrote:
>>>> On 10.06.2023 14:15, Konrad Dybcio wrote:
>>>>> On 10.06.2023 13:35, Stephan Gerhold wrote:
>>>>>> On Fri, Jun 09, 2023 at 10:19:09PM +0200, Konrad Dybcio wrote:
>>>>>>> Before we issue a call to RPM through clk_smd_rpm_enable_scaling() the
>>>>>>> clock rate requests will not be commited in hardware. This poses a
>>>>>>> race threat since we're accessing the bus clocks directly from within
>>>>>>> the interconnect framework.
>>>>>>>
>>>>>>> Add a marker to indicate that we're good to go with sending new requests
>>>>>>> and export it so that it can be referenced from icc.
>>>>>>>
>>>>>>> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
>>>>>>> ---
>>>>>>> drivers/clk/qcom/clk-smd-rpm.c | 9 +++++++++
>>>>>>> include/linux/soc/qcom/smd-rpm.h | 2 ++
>>>>>>> 2 files changed, 11 insertions(+)
>>>>>>>
>>>>>>> diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
>>>>>>> index 937cb1515968..482fe30ee6f0 100644
>>>>>>> --- a/drivers/clk/qcom/clk-smd-rpm.c
>>>>>>> +++ b/drivers/clk/qcom/clk-smd-rpm.c
>>>>>>> @@ -151,6 +151,7 @@
>>>>>>> #define to_clk_smd_rpm(_hw) container_of(_hw, struct clk_smd_rpm, hw)
>>>>>>>
>>>>>>> static struct qcom_smd_rpm *rpmcc_smd_rpm;
>>>>>>> +static bool smd_rpm_clk_scaling;
>>>>>>>
>>>>>>> struct clk_smd_rpm {
>>>>>>> const int rpm_res_type;
>>>>>>> @@ -385,6 +386,12 @@ static unsigned long clk_smd_rpm_recalc_rate(struct clk_hw *hw,
>>>>>>> return r->rate;
>>>>>>> }
>>>>>>>
>>>>>>> +bool qcom_smd_rpm_scaling_available(void)
>>>>>>> +{
>>>>>>> + return smd_rpm_clk_scaling;
>>>>>>> +}
>>>>>>> +EXPORT_SYMBOL_GPL(qcom_smd_rpm_scaling_available);
>>>>>>> +
>>>>>>> static int clk_smd_rpm_enable_scaling(void)
>>>>>>> {
>>>>>>> int ret;
>>>>>>> @@ -410,6 +417,8 @@ static int clk_smd_rpm_enable_scaling(void)
>>>>>>> return ret;
>>>>>>> }
>>>>>>>
>>>>>>> + smd_rpm_clk_scaling = true;
>>>>>>> +
>>>>>>
>>>>>> If you move the platform_device_register_data(&rpdev->dev,
>>>>>> "icc_smd_rpm", ...) from drivers/soc/qcom/smd-rpm.c to here you can
>>>>>> avoid the race completely and drop this API. I think that would be
>>>>>> cleaner. And it will likely probe much faster because probe deferral
>>>>>> is slow. :)
>>>>> Sounds like an idea.. especially since it's pretty much the only
>>>>> dependency other than SMDRPM itself!
>>>> It sounds great, but to not break bisecting one has to:
>>>>
>>>> 1. change the registration in soc/smd-rpm to store rpm ptr in driver
>>>> data, in addition to parent driver data
>>>>
>>>> 2. change icc/smd-rpm to use the device and not parent data
>>>>
>>>> 3. add a platform_device_register_data call in clk-smd-rpm that will
>>>> always fail because the device is always registered
>>>>
>>>> 4. remove the registration from soc/smd-rpm
>>>>
>>>
>>> Logically the icc_smd_rpm device still fits better as child of
>>> smd-rpm and not clk-smd-rpm. So I would probably just continue
>>> registering it on the parent device from clk-smd-rpm.
>>> Then there are no changes necessary in icc_smd_rpm.
>>>
>>> You could use this. Both touched files are Bjorn-maintained so should be
>>> manageable to have it in one commit. (note: compile-tested only)
>>>
>>> Thanks,
>>> Stephan
>>>
>>> From a2610adb2551b01e76b9de8e4cbcc89853814a8f Mon Sep 17 00:00:00 2001
>>> From: Stephan Gerhold <stephan@gerhold.net>
>>> Date: Sat, 10 Jun 2023 21:19:48 +0200
>>> Subject: [PATCH] soc: qcom: smd-rpm: Move icc_smd_rpm registration to
>>> clk-smd-rpm
>>>
>>> icc_smd_rpm will do bus clock votes itself rather than taking the
>>> unnecessary detour through the clock subsystem. However, it can only
>>> do that after the clocks have been handed off and scaling has been
>>> enabled in the RPM in clk-smd-rpm.
>>>
>>> Move the icc_smd_rpm registration from smd-rpm.c to clk-smd-rpm.c
>>> to avoid any possible races. icc_smd_rpm gets the driver data from
>>> the smd-rpm device, so still register the platform device on the
>>> smd-rpm parent device.
>>>
>>> Signed-off-by: Stephan Gerhold <stephan@gerhold.net>
>>> ---
>> Generally it looks good.. I'll give it a spin next week. One
>> thing below.
>>
>>> drivers/clk/qcom/clk-smd-rpm.c | 21 +++++++++++++++++++++
>>> drivers/soc/qcom/smd-rpm.c | 23 +----------------------
>>> 2 files changed, 22 insertions(+), 22 deletions(-)
>>>
>>> diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
>>> index e4de74b68797..91adb16889b3 100644
>>> --- a/drivers/clk/qcom/clk-smd-rpm.c
>>> +++ b/drivers/clk/qcom/clk-smd-rpm.c
>>> @@ -1302,12 +1302,20 @@ static struct clk_hw *qcom_smdrpm_clk_hw_get(struct of_phandle_args *clkspec,
>>> return desc->clks[idx] ? &desc->clks[idx]->hw : ERR_PTR(-ENOENT);
>>> }
>>>
>>> +static void rpm_smd_unregister_icc(void *data)
>>> +{
>>> + struct platform_device *icc_pdev = data;
>>> +
>>> + platform_device_unregister(icc_pdev);
>>> +}
>>> +
>>> static int rpm_smd_clk_probe(struct platform_device *pdev)
>>> {
>>> int ret;
>>> size_t num_clks, i;
>>> struct clk_smd_rpm **rpm_smd_clks;
>>> const struct rpm_smd_clk_desc *desc;
>>> + struct platform_device *icc_pdev;
>>>
>>> rpmcc_smd_rpm = dev_get_drvdata(pdev->dev.parent);
>>> if (!rpmcc_smd_rpm) {
>>> @@ -1357,6 +1365,19 @@ static int rpm_smd_clk_probe(struct platform_device *pdev)
>>> if (ret)
>>> goto err;
>>>
>>> + icc_pdev = platform_device_register_data(pdev->dev.parent,
>>> + "icc_smd_rpm", -1, NULL, 0);
>>> + if (IS_ERR(icc_pdev)) {
>>> + dev_err(&pdev->dev, "Failed to register icc_smd_rpm device: %pE\n",
>>> + icc_pdev);
>>> + /* No need to unregister clocks because of this */
>>> + } else {
>>> + ret = devm_add_action_or_reset(&pdev->dev, rpm_smd_unregister_icc,
>>> + icc_pdev);
>>> + if (ret)
>>> + goto err;
>>> + }
>>> +
>>> return 0;
>>> err:
>>> dev_err(&pdev->dev, "Error registering SMD clock driver (%d)\n", ret);
>>> diff --git a/drivers/soc/qcom/smd-rpm.c b/drivers/soc/qcom/smd-rpm.c
>>> index 0c1aa809cc4e..427dd5392b82 100644
>>> --- a/drivers/soc/qcom/smd-rpm.c
>>> +++ b/drivers/soc/qcom/smd-rpm.c
>>> @@ -19,7 +19,6 @@
>>> /**
>>> * struct qcom_smd_rpm - state of the rpm device driver
>>> * @rpm_channel: reference to the smd channel
>>> - * @icc: interconnect proxy device
>>> * @dev: rpm device
>>> * @ack: completion for acks
>>> * @lock: mutual exclusion around the send/complete pair
>>> @@ -27,7 +26,6 @@
>>> */
>>> struct qcom_smd_rpm {
>>> struct rpmsg_endpoint *rpm_channel;
>>> - struct platform_device *icc;
>>> struct device *dev;
>>>
>>> struct completion ack;
>>> @@ -197,7 +195,6 @@ static int qcom_smd_rpm_callback(struct rpmsg_device *rpdev,
>>> static int qcom_smd_rpm_probe(struct rpmsg_device *rpdev)
>>> {
>>> struct qcom_smd_rpm *rpm;
>>> - int ret;
>>>
>>> rpm = devm_kzalloc(&rpdev->dev, sizeof(*rpm), GFP_KERNEL);
>>> if (!rpm)
>>> @@ -210,24 +207,7 @@ static int qcom_smd_rpm_probe(struct rpmsg_device *rpdev)
>>> rpm->rpm_channel = rpdev->ept;
>>> dev_set_drvdata(&rpdev->dev, rpm);
>>>
>>> - rpm->icc = platform_device_register_data(&rpdev->dev, "icc_smd_rpm", -1,
>>> - NULL, 0);
>>> - if (IS_ERR(rpm->icc))
>>> - return PTR_ERR(rpm->icc);
>>> -
>>> - ret = of_platform_populate(rpdev->dev.of_node, NULL, NULL, &rpdev->dev);
>>> - if (ret)
>>> - platform_device_unregister(rpm->icc);
>>> -
>>> - return ret;
>>> -}
>>> -
>>> -static void qcom_smd_rpm_remove(struct rpmsg_device *rpdev)
>>> -{
>>> - struct qcom_smd_rpm *rpm = dev_get_drvdata(&rpdev->dev);
>>> -
>>> - platform_device_unregister(rpm->icc);
>>> - of_platform_depopulate(&rpdev->dev);
>>> + return devm_of_platform_populate(&rpdev->dev);
>>> }
>>>
>>> static const struct of_device_id qcom_smd_rpm_of_match[] = {
>>> @@ -256,7 +236,6 @@ MODULE_DEVICE_TABLE(of, qcom_smd_rpm_of_match);
>>>
>>> static struct rpmsg_driver qcom_smd_rpm_driver = {
>>> .probe = qcom_smd_rpm_probe,
>>> - .remove = qcom_smd_rpm_remove,
>> This reaches over the removal of the icc registration, the depopulate
>> call should stay.
>>
>
> I switched the of_platform_populate() to devm_of_platform_populate(),
> that's why the remove callback is no longer necessary. It's a bit
> hidden, perhaps it would be enough to add to the commit message:
>
> "While at it, switch the remaining of_platform_populate() call to the
> devm variant and remove the remove callback."
>
> Or maybe it should be split into two patches.
Gave it a spin, I think it ends up being worse if an IPA rpm clock is
consumed by one of the icc providers, and that's sadly the case
for almost all platforms (or supposed to be).. :/ Only qcm2290 doesn't
seem to care if we poke the network interface units with half the soc
off :P
Konrad
>
> Thanks,
> Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
* Re: [PATCH v2 04/22] clk: qcom: smd-rpm: Export clock scaling availability
2023-06-12 12:51 ` Konrad Dybcio
@ 2023-06-12 17:03 ` Konrad Dybcio
0 siblings, 0 replies; 45+ messages in thread
From: Konrad Dybcio @ 2023-06-12 17:03 UTC (permalink / raw)
To: Stephan Gerhold
Cc: Andy Gross, Bjorn Andersson, Michael Turquette, Stephen Boyd,
Georgi Djakov, Leo Yan, Evan Green, Marijn Suijten, linux-arm-msm,
linux-kernel, linux-clk, linux-pm
On 12.06.2023 14:51, Konrad Dybcio wrote:
> On 11.06.2023 11:20, Stephan Gerhold wrote:
>> On Sat, Jun 10, 2023 at 09:39:00PM +0200, Konrad Dybcio wrote:
>>> On 10.06.2023 21:25, Stephan Gerhold wrote:
>>>> On Sat, Jun 10, 2023 at 08:53:05PM +0200, Konrad Dybcio wrote:
>>>>> On 10.06.2023 14:15, Konrad Dybcio wrote:
>>>>>> On 10.06.2023 13:35, Stephan Gerhold wrote:
>>>>>>> On Fri, Jun 09, 2023 at 10:19:09PM +0200, Konrad Dybcio wrote:
>>>>>>>> Before we issue a call to RPM through clk_smd_rpm_enable_scaling() the
>>>>>>>> clock rate requests will not be commited in hardware. This poses a
>>>>>>>> race threat since we're accessing the bus clocks directly from within
>>>>>>>> the interconnect framework.
>>>>>>>>
>>>>>>>> Add a marker to indicate that we're good to go with sending new requests
>>>>>>>> and export it so that it can be referenced from icc.
>>>>>>>>
>>>>>>>> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
>>>>>>>> ---
>>>>>>>> drivers/clk/qcom/clk-smd-rpm.c | 9 +++++++++
>>>>>>>> include/linux/soc/qcom/smd-rpm.h | 2 ++
>>>>>>>> 2 files changed, 11 insertions(+)
>>>>>>>>
>>>>>>>> diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
>>>>>>>> index 937cb1515968..482fe30ee6f0 100644
>>>>>>>> --- a/drivers/clk/qcom/clk-smd-rpm.c
>>>>>>>> +++ b/drivers/clk/qcom/clk-smd-rpm.c
>>>>>>>> @@ -151,6 +151,7 @@
>>>>>>>> #define to_clk_smd_rpm(_hw) container_of(_hw, struct clk_smd_rpm, hw)
>>>>>>>>
>>>>>>>> static struct qcom_smd_rpm *rpmcc_smd_rpm;
>>>>>>>> +static bool smd_rpm_clk_scaling;
>>>>>>>>
>>>>>>>> struct clk_smd_rpm {
>>>>>>>> const int rpm_res_type;
>>>>>>>> @@ -385,6 +386,12 @@ static unsigned long clk_smd_rpm_recalc_rate(struct clk_hw *hw,
>>>>>>>> return r->rate;
>>>>>>>> }
>>>>>>>>
>>>>>>>> +bool qcom_smd_rpm_scaling_available(void)
>>>>>>>> +{
>>>>>>>> + return smd_rpm_clk_scaling;
>>>>>>>> +}
>>>>>>>> +EXPORT_SYMBOL_GPL(qcom_smd_rpm_scaling_available);
>>>>>>>> +
>>>>>>>> static int clk_smd_rpm_enable_scaling(void)
>>>>>>>> {
>>>>>>>> int ret;
>>>>>>>> @@ -410,6 +417,8 @@ static int clk_smd_rpm_enable_scaling(void)
>>>>>>>> return ret;
>>>>>>>> }
>>>>>>>>
>>>>>>>> + smd_rpm_clk_scaling = true;
>>>>>>>> +
>>>>>>>
>>>>>>> If you move the platform_device_register_data(&rpdev->dev,
>>>>>>> "icc_smd_rpm", ...) from drivers/soc/qcom/smd-rpm.c to here you can
>>>>>>> avoid the race completely and drop this API. I think that would be
>>>>>>> cleaner. And it will likely probe much faster because probe deferral
>>>>>>> is slow. :)
>>>>>> Sounds like an idea.. especially since it's pretty much the only
>>>>>> dependency other than SMDRPM itself!
>>>>> It sounds great, but to not break bisecting one has to:
>>>>>
>>>>> 1. change the registration in soc/smd-rpm to store rpm ptr in driver
>>>>> data, in addition to parent driver data
>>>>>
>>>>> 2. change icc/smd-rpm to use the device and not parent data
>>>>>
>>>>> 3. add a platform_device_register_data call in clk-smd-rpm that will
>>>>> always fail because the device is always registered
>>>>>
>>>>> 4. remove the registration from soc/smd-rpm
>>>>>
>>>>
>>>> Logically the icc_smd_rpm device still fits better as child of
>>>> smd-rpm and not clk-smd-rpm. So I would probably just continue
>>>> registering it on the parent device from clk-smd-rpm.
>>>> Then there are no changes necessary in icc_smd_rpm.
>>>>
>>>> You could use this. Both touched files are Bjorn-maintained so should be
>>>> manageable to have it in one commit. (note: compile-tested only)
>>>>
>>>> Thanks,
>>>> Stephan
>>>>
>>>> From a2610adb2551b01e76b9de8e4cbcc89853814a8f Mon Sep 17 00:00:00 2001
>>>> From: Stephan Gerhold <stephan@gerhold.net>
>>>> Date: Sat, 10 Jun 2023 21:19:48 +0200
>>>> Subject: [PATCH] soc: qcom: smd-rpm: Move icc_smd_rpm registration to
>>>> clk-smd-rpm
>>>>
>>>> icc_smd_rpm will do bus clock votes itself rather than taking the
>>>> unnecessary detour through the clock subsystem. However, it can only
>>>> do that after the clocks have been handed off and scaling has been
>>>> enabled in the RPM in clk-smd-rpm.
>>>>
>>>> Move the icc_smd_rpm registration from smd-rpm.c to clk-smd-rpm.c
>>>> to avoid any possible races. icc_smd_rpm gets the driver data from
>>>> the smd-rpm device, so still register the platform device on the
>>>> smd-rpm parent device.
>>>>
>>>> Signed-off-by: Stephan Gerhold <stephan@gerhold.net>
>>>> ---
>>> Generally it looks good.. I'll give it a spin next week. One
>>> thing below.
>>>
>>>> drivers/clk/qcom/clk-smd-rpm.c | 21 +++++++++++++++++++++
>>>> drivers/soc/qcom/smd-rpm.c | 23 +----------------------
>>>> 2 files changed, 22 insertions(+), 22 deletions(-)
>>>>
>>>> diff --git a/drivers/clk/qcom/clk-smd-rpm.c b/drivers/clk/qcom/clk-smd-rpm.c
>>>> index e4de74b68797..91adb16889b3 100644
>>>> --- a/drivers/clk/qcom/clk-smd-rpm.c
>>>> +++ b/drivers/clk/qcom/clk-smd-rpm.c
>>>> @@ -1302,12 +1302,20 @@ static struct clk_hw *qcom_smdrpm_clk_hw_get(struct of_phandle_args *clkspec,
>>>> return desc->clks[idx] ? &desc->clks[idx]->hw : ERR_PTR(-ENOENT);
>>>> }
>>>>
>>>> +static void rpm_smd_unregister_icc(void *data)
>>>> +{
>>>> + struct platform_device *icc_pdev = data;
>>>> +
>>>> + platform_device_unregister(icc_pdev);
>>>> +}
>>>> +
>>>> static int rpm_smd_clk_probe(struct platform_device *pdev)
>>>> {
>>>> int ret;
>>>> size_t num_clks, i;
>>>> struct clk_smd_rpm **rpm_smd_clks;
>>>> const struct rpm_smd_clk_desc *desc;
>>>> + struct platform_device *icc_pdev;
>>>>
>>>> rpmcc_smd_rpm = dev_get_drvdata(pdev->dev.parent);
>>>> if (!rpmcc_smd_rpm) {
>>>> @@ -1357,6 +1365,19 @@ static int rpm_smd_clk_probe(struct platform_device *pdev)
>>>> if (ret)
>>>> goto err;
>>>>
>>>> + icc_pdev = platform_device_register_data(pdev->dev.parent,
>>>> + "icc_smd_rpm", -1, NULL, 0);
>>>> + if (IS_ERR(icc_pdev)) {
>>>> + dev_err(&pdev->dev, "Failed to register icc_smd_rpm device: %pE\n",
>>>> + icc_pdev);
>>>> + /* No need to unregister clocks because of this */
>>>> + } else {
>>>> + ret = devm_add_action_or_reset(&pdev->dev, rpm_smd_unregister_icc,
>>>> + icc_pdev);
>>>> + if (ret)
>>>> + goto err;
>>>> + }
>>>> +
>>>> return 0;
>>>> err:
>>>> dev_err(&pdev->dev, "Error registering SMD clock driver (%d)\n", ret);
>>>> diff --git a/drivers/soc/qcom/smd-rpm.c b/drivers/soc/qcom/smd-rpm.c
>>>> index 0c1aa809cc4e..427dd5392b82 100644
>>>> --- a/drivers/soc/qcom/smd-rpm.c
>>>> +++ b/drivers/soc/qcom/smd-rpm.c
>>>> @@ -19,7 +19,6 @@
>>>> /**
>>>> * struct qcom_smd_rpm - state of the rpm device driver
>>>> * @rpm_channel: reference to the smd channel
>>>> - * @icc: interconnect proxy device
>>>> * @dev: rpm device
>>>> * @ack: completion for acks
>>>> * @lock: mutual exclusion around the send/complete pair
>>>> @@ -27,7 +26,6 @@
>>>> */
>>>> struct qcom_smd_rpm {
>>>> struct rpmsg_endpoint *rpm_channel;
>>>> - struct platform_device *icc;
>>>> struct device *dev;
>>>>
>>>> struct completion ack;
>>>> @@ -197,7 +195,6 @@ static int qcom_smd_rpm_callback(struct rpmsg_device *rpdev,
>>>> static int qcom_smd_rpm_probe(struct rpmsg_device *rpdev)
>>>> {
>>>> struct qcom_smd_rpm *rpm;
>>>> - int ret;
>>>>
>>>> rpm = devm_kzalloc(&rpdev->dev, sizeof(*rpm), GFP_KERNEL);
>>>> if (!rpm)
>>>> @@ -210,24 +207,7 @@ static int qcom_smd_rpm_probe(struct rpmsg_device *rpdev)
>>>> rpm->rpm_channel = rpdev->ept;
>>>> dev_set_drvdata(&rpdev->dev, rpm);
>>>>
>>>> - rpm->icc = platform_device_register_data(&rpdev->dev, "icc_smd_rpm", -1,
>>>> - NULL, 0);
>>>> - if (IS_ERR(rpm->icc))
>>>> - return PTR_ERR(rpm->icc);
>>>> -
>>>> - ret = of_platform_populate(rpdev->dev.of_node, NULL, NULL, &rpdev->dev);
>>>> - if (ret)
>>>> - platform_device_unregister(rpm->icc);
>>>> -
>>>> - return ret;
>>>> -}
>>>> -
>>>> -static void qcom_smd_rpm_remove(struct rpmsg_device *rpdev)
>>>> -{
>>>> - struct qcom_smd_rpm *rpm = dev_get_drvdata(&rpdev->dev);
>>>> -
>>>> - platform_device_unregister(rpm->icc);
>>>> - of_platform_depopulate(&rpdev->dev);
>>>> + return devm_of_platform_populate(&rpdev->dev);
>>>> }
>>>>
>>>> static const struct of_device_id qcom_smd_rpm_of_match[] = {
>>>> @@ -256,7 +236,6 @@ MODULE_DEVICE_TABLE(of, qcom_smd_rpm_of_match);
>>>>
>>>> static struct rpmsg_driver qcom_smd_rpm_driver = {
>>>> .probe = qcom_smd_rpm_probe,
>>>> - .remove = qcom_smd_rpm_remove,
>>> This reaches over the removal of the icc registration, the depopulate
>>> call should stay.
>>>
>>
>> I switched the of_platform_populate() to devm_of_platform_populate(),
>> that's why the remove callback is no longer necessary. It's a bit
>> hidden, perhaps it would be enough to add to the commit message:
>>
>> "While at it, switch the remaining of_platform_populate() call to the
>> devm variant and remove the remove callback."
>>
>> Or maybe it should be split into two patches.
> Gave it a spin, I think it ends up being worse if an IPA rpm clock is
> consumed by one of the icc providers, and that's sadly the case
> for almost all platforms (or supposed to be).. :/ Only qcm2290 doesn't
> seem to care if we poke the network interface units with half the soc
> off :P
Actually it's more complex and that rings the same tons-of-probe-deferrals
bell that we have to get rid of.. The icc_smd_rpm driver actually does
probe faster with your patch, but that's not the case with the SoC icc
provider driver which does probe later :/
Don't think either solution is perfect, but I'm willing to take your
solution, as the one I proposed is a bit of a dead end for future
improvements.
Konrad
>
> Konrad
>>
>> Thanks,
>> Stephan
^ permalink raw reply [flat|nested] 45+ messages in thread
end of thread, other threads:[~2023-06-12 17:03 UTC | newest]
Thread overview: 45+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-06-09 20:19 [PATCH v2 00/22] Restructure RPM SMD ICC Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 01/22] soc: qcom: smd-rpm: Add QCOM_SMD_RPM_STATE_NUM Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 02/22] soc: qcom: smd-rpm: Use tabs for defines Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 03/22] clk: qcom: smd-rpm: Move some RPM resources to the common header Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 04/22] clk: qcom: smd-rpm: Export clock scaling availability Konrad Dybcio
2023-06-10 11:35 ` Stephan Gerhold
2023-06-10 12:15 ` Konrad Dybcio
2023-06-10 18:53 ` Konrad Dybcio
2023-06-10 19:25 ` Stephan Gerhold
2023-06-10 19:39 ` Konrad Dybcio
2023-06-11 9:20 ` Stephan Gerhold
2023-06-12 12:51 ` Konrad Dybcio
2023-06-12 17:03 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 05/22] interconnect: qcom: icc-rpm: Introduce keep_alive Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 06/22] interconnect: qcom: icc-rpm: Allow negative QoS offset Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 07/22] interconnect: qcom: Fold smd-rpm.h into icc-rpm.h Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 08/22] interconnect: qcom: smd-rpm: Add rpmcc handling skeleton code Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 09/22] interconnect: qcom: Add missing headers in icc-rpm.h Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 10/22] interconnect: qcom: Define RPM bus clocks Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 11/22] interconnect: qcom: sdm660: Hook up RPM bus clk definitions Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 12/22] interconnect: qcom: msm8996: " Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 13/22] interconnect: qcom: qcs404: " Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 14/22] interconnect: qcom: msm8939: " Konrad Dybcio
2023-06-10 12:02 ` Stephan Gerhold
2023-06-09 20:19 ` [PATCH v2 15/22] interconnect: qcom: msm8916: " Konrad Dybcio
2023-06-10 12:02 ` Stephan Gerhold
2023-06-09 20:19 ` [PATCH v2 16/22] interconnect: qcom: qcm2290: " Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 17/22] interconnect: qcom: icc-rpm: Control bus rpmcc from icc Konrad Dybcio
2023-06-10 11:58 ` Stephan Gerhold
2023-06-10 12:14 ` Konrad Dybcio
2023-06-10 16:20 ` Stephan Gerhold
2023-06-10 17:54 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 18/22] clk: qcom: smd-rpm: Separate out interconnect bus clocks Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 19/22] interconnect: qcom: icc-rpm: Fix bucket number Konrad Dybcio
2023-06-10 17:46 ` Stephan Gerhold
2023-06-10 17:53 ` Konrad Dybcio
2023-06-09 20:19 ` [PATCH v2 20/22] interconnect: qcom: icc-rpm: Set bandwidth on both contexts Konrad Dybcio
2023-06-10 18:00 ` Stephan Gerhold
2023-06-10 18:28 ` Konrad Dybcio
2023-06-10 18:43 ` Stephan Gerhold
2023-06-09 20:19 ` [PATCH v2 21/22] interconnect: qcom: icc-rpm: Set correct bandwidth through RPM bw req Konrad Dybcio
2023-06-10 18:46 ` Stephan Gerhold
2023-06-09 20:19 ` [PATCH v2 22/22] interconnect: qcom: icc-rpm: Fix bandwidth calculations Konrad Dybcio
2023-06-10 19:06 ` Stephan Gerhold
2023-06-10 19:09 ` Konrad Dybcio
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).