Introduce SoC sleep stats driver
Subsystems (4)
v2 cover Introduce SoC sleep stats driver
Changes in v2:
- Convert Documentation to YAML.
- Address stephen's comments from v1.
- Use debugfs instead of sysfs.
- Add sc7180 dts changes for sleep stats
- Add defconfig changes to enable driver
- Include subsystem stats from [1] in this single stats driver.
- Address stephen's comments from [1]
- Update cover letter inline to mention [1]
Qualcomm Technologies, Inc. (QTI)'s chipsets support SoC level low power modes. SoCs Always On Processor/Resource Power Manager produces statistics of the SoC sleep modes involving lowering or powering down of the rails and the oscillator clock.
Additionally multiple subsystems present on SoC like modem, spss, adsp, cdsp maintains their low power mode statistics in shared memory (SMEM).
Statistics includes SoC sleep mode type, number of times LPM entered, time of last entry, exit, and accumulated sleep duration in seconds.
This series adds a driver to read the stats and export to debugfs.
[1] https://lore.kernel.org/patchwork/patch/1149381/
Mahesh Sivasubramanian (2):
dt-bindings: Introduce soc sleep stats bindings for Qualcomm SoCs
soc: qcom: Add SoC sleep stats driver
Maulik Shah (2):
arm64: dts: qcom: sc7180: Enable soc sleep stats
arm64: defconfig: Enable SoC sleep stats driver for Qualcomm
.../bindings/soc/qcom/soc-sleep-stats.yaml | 47 ++++ arch/arm64/boot/dts/qcom/sc7180.dtsi | 5 + arch/arm64/configs/defconfig | 1 + drivers/soc/qcom/Kconfig | 10 + drivers/soc/qcom/Makefile | 1 + drivers/soc/qcom/soc_sleep_stats.c | 279 +++++++++++++++++++++ 6 files changed, 343 insertions(+) create mode 100644 Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml create mode 100644 drivers/soc/qcom/soc_sleep_stats.c
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
PATCH v2 1/4 dt-bindings: Introduce soc sleep stats bindings for Qualcomm SoCs
From: Mahesh Sivasubramanian <[email protected]>
Add device binding documentation for Qualcomm Technology Inc's (QTI) SoC sleep stats driver. The driver is used for displaying SoC sleep statistic maintained by Always On Processor or Resource Power Manager.
--- .../bindings/soc/qcom/soc-sleep-stats.yaml | 47 ++++++++++++++++++++++ 1 file changed, 47 insertions(+) create mode 100644 Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +47 -0
@@ -0,0 +1,47 @@ +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) +%YAML 1.2 +--- +$id: http://devicetree.org/schemas/bindings/soc/qcom/soc-sleep-stats.yaml# +$schema: http://devicetree.org/meta-schemas/core.yaml# + +title: Qualcomm Technologies, Inc. (QTI) SoC sleep stats bindings + +maintainers: + - Maulik Shah <[email protected]> + - Lina Iyer <[email protected]> + +description: | + Always On Processor/Resource Power Manager maintains statistics of the SoC + sleep modes involving powering down of the rails and oscillator clock. + + Statistics includes SoC sleep mode type, number of times low power mode were + entered, time of last entry, time of last exit and accumulated sleep duration. + SoC sleep stats driver provides debugfs interface to show this information. + +properties: + compatible: + enum: + - qcom,rpmh-sleep-stats + - qcom,rpm-sleep-stats + + reg: + maxItems: 1 + +required: + - compatible + - reg + +examples: + # Example of rpmh sleep stats + - | + rpmh_sleep_stats: soc-sleep-stats@c3f0000 { + compatible = "qcom,rpmh-sleep-stats"; + reg = <0 0xc3f0000 0 0x400>; + }; + # Example of rpm sleep stats + - | + rpm_sleep_stats: soc-sleep-stats@4690000 { + compatible = "qcom,rpm-sleep-stats"; + reg = <0 0x04690000 0 0x400>; + }; +...
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
PATCH v2 2/4 soc: qcom: Add SoC sleep stats driver
In reply to Maulik ShahFrom: Mahesh Sivasubramanian <[email protected]>
Lets's add a driver to read the the stats from remote processor and export to debugfs.
The driver creates "soc_sleep_stats" directory in debugfs and adds files for various low power mode available. Below is sample output with command
cat /sys/kernel/debug/soc_sleep_stats/ddr
count = 0
Last Entered At = 0
Last Exited At = 0
Accumulated Duration = 0
--- drivers/soc/qcom/Kconfig | 10 ++ drivers/soc/qcom/Makefile | 1 + drivers/soc/qcom/soc_sleep_stats.c | 279 +++++++++++++++++++++++++++++++++++++ 3 files changed, 290 insertions(+) create mode 100644 drivers/soc/qcom/soc_sleep_stats.c
drivers/soc/qcom/Kconfig +10 -0
@@ -185,6 +185,16 @@ config QCOM_SOCINFO Say yes here to support the Qualcomm socinfo driver, providing information about the SoC to user space. +config QCOM_SOC_SLEEP_STATS + tristate "Qualcomm Technologies, Inc. (QTI) SoC sleep stats driver" + depends on ARCH_QCOM + depends on DEBUG_FS + help + Qualcomm Technologies, Inc. (QTI) SoC sleep stats driver to read + the shared memory exported by the remote processor related to + various SoC level low power modes statistics and export to debugfs + interface. + config QCOM_WCNSS_CTRL tristate "Qualcomm WCNSS control driver" depends on ARCH_QCOM || COMPILE_TEST
drivers/soc/qcom/Makefile +1 -0
@@ -20,6 +20,7 @@ obj-$(CONFIG_QCOM_SMEM_STATE) += smem_state.o obj-$(CONFIG_QCOM_SMP2P) += smp2p.o obj-$(CONFIG_QCOM_SMSM) += smsm.o obj-$(CONFIG_QCOM_SOCINFO) += socinfo.o +obj-$(CONFIG_QCOM_SOC_SLEEP_STATS) += soc_sleep_stats.o obj-$(CONFIG_QCOM_WCNSS_CTRL) += wcnss_ctrl.o obj-$(CONFIG_QCOM_APR) += apr.o obj-$(CONFIG_QCOM_LLCC) += llcc-qcom.o
drivers/soc/qcom/soc_sleep_stats.c new file +279 -0
279 lines not on this page · raw message
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
PATCH v2 3/4 arm64: dts: qcom: sc7180: Enable soc sleep stats
In reply to Maulik ShahSoC sleep stats provides various low power mode stats.
--- arch/arm64/boot/dts/qcom/sc7180.dtsi | 5 +++++ 1 file changed, 5 insertions(+)
arch/arm64/boot/dts/qcom/sc7180.dtsi +5 -0
@@ -745,6 +745,11 @@ }; }; + rpmh_sleep_stats: soc-sleep-stats@c3f0000 { + compatible = "qcom,rpmh-sleep-stats"; + reg = <0 0xc3f0000 0 0x400>; + }; + tcsr_mutex_regs: syscon@1f40000 { compatible = "syscon"; reg = <0 0x01f40000 0 0x40000>;
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
PATCH v2 4/4 arm64: defconfig: Enable SoC sleep stats driver for Qualcomm
In reply to Maulik ShahEnable SoC sleep stats driver. The driver gives statistics for various SoC level low power modes.
--- arch/arm64/configs/defconfig | 1 + 1 file changed, 1 insertion(+)
arch/arm64/configs/defconfig +1 -0
@@ -767,6 +767,7 @@ CONFIG_QCOM_SMD_RPM=y CONFIG_QCOM_SMP2P=y CONFIG_QCOM_SMSM=y CONFIG_QCOM_SOCINFO=m +CONFIG_QCOM_SOC_SLEEP_STATS=y CONFIG_ARCH_R8A774A1=y CONFIG_ARCH_R8A774B1=y CONFIG_ARCH_R8A774C0=y
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
Re: [PATCH v2 1/4] dt-bindings: Introduce soc sleep stats bindings for Qualcomm SoCs
In reply to Maulik ShahQuoted 6 lines · Maulik Shah
From: Mahesh Sivasubramanian <[email protected]>
Add device binding documentation for Qualcomm Technology Inc's (QTI) SoC sleep stats driver. The driver is used for displaying SoC sleep statistic maintained by Always On Processor or Resource Power Manager.
Quoted 9 lines
Signed-off-by Mahesh Sivasubramanian <[email protected]>Signed-off-by Lina Iyer <[email protected]>Signed-off-by Maulik Shah <[email protected]>--- .../bindings/soc/qcom/soc-sleep-stats.yaml | 47 ++++++++++++++++++++++ 1 file changed, 47 insertions(+) create mode 100644 Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml
Re: [PATCH v2 3/4] arm64: dts: qcom: sc7180: Enable soc sleep stats
In reply to Maulik ShahQuoted 13 lines · Maulik Shah
SoC sleep stats provides various low power mode stats.
Signed-off-by Maulik Shah <[email protected]>--- arch/arm64/boot/dts/qcom/sc7180.dtsi | 5 +++++ 1 file changed, 5 insertions(+)arch/arm64/boot/dts/qcom/sc7180.dtsi +0
-0
arch/arm64/boot/dts/qcom/sc7180.dtsi line 748
Quoted 4 lines
arch/arm64/boot/dts/qcom/sc7180.dtsi +1
-0@@ -745,6 +745,11 @@ }; }; + rpmh_sleep_stats: soc-sleep-stats@c3f0000 {
I don't see any reason to reference this node, so you should be able to omit the label(?)
arch/arm64/boot/dts/qcom/sc7180.dtsi line 750
Quoted 2 lines
arch/arm64/boot/dts/qcom/sc7180.dtsi +2
-0+ compatible = "qcom,rpmh-sleep-stats"; + reg = <0 0xc3f0000 0 0x400>;
Please pad the address to 8 digits and sort the nodes by address.
Regards,
Bjorn
Quoted 8 lines
arch/arm64/boot/dts/qcom/sc7180.dtsi +2
-0+ }; + tcsr_mutex_regs: syscon@1f40000 { compatible = "syscon"; reg = <0 0x01f40000 0 0x40000>;Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
Re: [PATCH v2 4/4] arm64: defconfig: Enable SoC sleep stats driver for Qualcomm
In reply to Maulik ShahQuoted 13 lines · Maulik Shah
Enable SoC sleep stats driver. The driver gives statistics for various SoC level low power modes.
Signed-off-by Maulik Shah <[email protected]>--- arch/arm64/configs/defconfig | 1 + 1 file changed, 1 insertion(+)arch/arm64/configs/defconfig +0
-0
arch/arm64/configs/defconfig CONFIG_QCOM_SMD_RPM=y · line 770
Quoted 4 lines
arch/arm64/configs/defconfig +1
-0@@ -767,6 +767,7 @@ CONFIG_QCOM_SMD_RPM=y CONFIG_QCOM_SMP2P=y CONFIG_QCOM_SMSM=y CONFIG_QCOM_SOCINFO=m +CONFIG_QCOM_SOC_SLEEP_STATS=y
Afaict the driver is not crucial to boot the system or to collect the statics, so you should be able to make this =m.
Regards,
Bjorn
Quoted 6 lines
arch/arm64/configs/defconfig +0
-0CONFIG_ARCH_R8A774A1=y CONFIG_ARCH_R8A774B1=y CONFIG_ARCH_R8A774C0=ySignature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
Re: [PATCH v2 1/4] dt-bindings: Introduce soc sleep stats bindings for Qualcomm SoCs
In reply to Maulik ShahQuoted 6 lines · Maulik Shah
From: Mahesh Sivasubramanian <[email protected]>
Add device binding documentation for Qualcomm Technology Inc's (QTI) SoC sleep stats driver. The driver is used for displaying SoC sleep statistic maintained by Always On Processor or Resource Power Manager.
Quoted 65 lines
Signed-off-by Mahesh Sivasubramanian <[email protected]>Signed-off-by Lina Iyer <[email protected]>Signed-off-by Maulik Shah <[email protected]>--- .../bindings/soc/qcom/soc-sleep-stats.yaml | 47 ++++++++++++++++++++++ 1 file changed, 47 insertions(+) create mode 100644 Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yamlDocumentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +47
-0@@ -0,0 +1,47 @@ +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) +%YAML 1.2 +--- +$id: http://devicetree.org/schemas/bindings/soc/qcom/soc-sleep-stats.yaml# +$schema: http://devicetree.org/meta-schemas/core.yaml# + +title: Qualcomm Technologies, Inc. (QTI) SoC sleep stats bindings + +maintainers: + - Maulik Shah <[email protected]> + - Lina Iyer <[email protected]> + +description: | + Always On Processor/Resource Power Manager maintains statistics of the SoC + sleep modes involving powering down of the rails and oscillator clock. + + Statistics includes SoC sleep mode type, number of times low power mode were + entered, time of last entry, time of last exit and accumulated sleep duration. + SoC sleep stats driver provides debugfs interface to show this information. + +properties: + compatible: + enum: + - qcom,rpmh-sleep-stats + - qcom,rpm-sleep-stats + + reg: + maxItems: 1 + +required: + - compatible + - reg + +examples: + # Example of rpmh sleep stats + - | + rpmh_sleep_stats: soc-sleep-stats@c3f0000 { + compatible = "qcom,rpmh-sleep-stats"; + reg = <0 0xc3f0000 0 0x400>; + }; + # Example of rpm sleep stats + - | + rpm_sleep_stats: soc-sleep-stats@4690000 { + compatible = "qcom,rpm-sleep-stats"; + reg = <0 0x04690000 0 0x400>; + }; +...Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
Re: [PATCH v2 1/4] dt-bindings: Introduce soc sleep stats bindings for Qualcomm SoCs
In reply to Maulik ShahQuoting Maulik Shah (2020-02-21 00:49:43)
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml line 4
Quoted 10 lines
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +4
-0@@ -0,0 +1,47 @@ +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) +%YAML 1.2 +--- +$id: http://devicetree.org/schemas/bindings/soc/qcom/soc-sleep-stats.yaml#
Drop 'bindings' from above?
Quoted 5 lines
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +5
-0+$schema: http://devicetree.org/meta-schemas/core.yaml# + +title: Qualcomm Technologies, Inc. (QTI) SoC sleep stats bindings + +maintainers:
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml line 19
Quoted 10 lines
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +10
-0+ - Maulik Shah <[email protected]> + - Lina Iyer <[email protected]> + +description: | + Always On Processor/Resource Power Manager maintains statistics of the SoC + sleep modes involving powering down of the rails and oscillator clock. + + Statistics includes SoC sleep mode type, number of times low power mode were + entered, time of last entry, time of last exit and accumulated sleep duration. + SoC sleep stats driver provides debugfs interface to show this information.
Please remove this last line. It is a Linuxism that doesn't belong in DT bindings. And then make it one paragraph and drop the | because formatting doesn't need to be maintained.
Quoted 11 lines
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +11
-0+ +properties: + compatible: + enum: + - qcom,rpmh-sleep-stats + - qcom,rpm-sleep-stats + + reg: + maxItems: 1 + +required:
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml line 40
Quoted 10 lines
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +10
-0+ - compatible + - reg + +examples: + # Example of rpmh sleep stats + - | + rpmh_sleep_stats: soc-sleep-stats@c3f0000 { + compatible = "qcom,rpmh-sleep-stats"; + reg = <0 0xc3f0000 0 0x400>; + };
I see that on sc7180 aoss-qmp is overlapping with this region.
aoss_qmp: qmp@c300000 {
compatible = "qcom,sc7180-aoss-qmp";
reg = <0 0x0c300000 0 0x100000>;So is this register region really something more like a TCM or RAM area where rpmh combines multiple software concepts into one hardware memory region? The aoss-qmp driver talks about message RAM, so I think this sleep stats stuff is a carveout of the RPMh message RAM. It seems OK if we want to split that message RAM up into multiple DT nodes, but then we'll need to reduce the reg size for the aoss-qmp node.
Finally, the node name 'soc-sleep-stats' is generic, but I wonder why we couldn't name the node 'tcm' or 'memory' or 'msgram'. Similarly for the qmp node it fits better DT style to have the node be something generic.
Quoted 7 lines
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +7
-0+ # Example of rpm sleep stats + - | + rpm_sleep_stats: soc-sleep-stats@4690000 { + compatible = "qcom,rpm-sleep-stats"; + reg = <0 0x04690000 0 0x400>; + }; +...
Re: [PATCH v2 2/4] soc: qcom: Add SoC sleep stats driver
In reply to Maulik ShahQuoting Maulik Shah (2020-02-21 00:49:44)
Quoted 3 lines
From: Mahesh Sivasubramanian <[email protected]>
Lets's add a driver to read the the stats from remote processor
Let's
Quoted 27 lines
and export to debugfs.
The driver creates "
soc_sleep_stats" directory in debugfs and adds files for various low power mode available. Below is sample output with commandcat
/sys/kernel/debug/soc_sleep_stats/ddr
count = 0
Last Entered At = 0
Last Exited At = 0
Accumulated Duration = 0Signed-off-by Mahesh Sivasubramanian <[email protected]>Signed-off-by Lina Iyer <[email protected]>Signed-off-by Maulik Shah <[email protected]>--- drivers/soc/qcom/Kconfig | 10 ++ drivers/soc/qcom/Makefile | 1 + drivers/soc/qcom/soc_sleep_stats.c | 279 +++++++++++++++++++++++++++++++++++++ 3 files changed, 290 insertions(+) create mode 100644 drivers/soc/qcom/soc_sleep_stats.cdrivers/soc/qcom/Kconfig +0
-0
drivers/soc/qcom/Kconfig config QCOM_SOCINFO · line 190
Quoted 6 lines
drivers/soc/qcom/Kconfig +3
-0@@ -185,6 +185,16 @@ config QCOM_SOCINFO Say yes here to support the Qualcomm socinfo driver, providing information about the SoC to user space. +config QCOM_SOC_SLEEP_STATS + tristate "Qualcomm Technologies, Inc. (QTI) SoC sleep stats driver" + depends on ARCH_QCOM
Can you add || COMPILE_TEST?
Quoted 20 lines
drivers/soc/qcom/Kconfig +7
-0+ depends on DEBUG_FS + help + Qualcomm Technologies, Inc. (QTI) SoC sleep stats driver to read + the shared memory exported by the remote processor related to + various SoC level low power modes statistics and export to debugfs + interface. + config QCOM_WCNSS_CTRL tristate "Qualcomm WCNSS control driver" depends on ARCH_QCOM || COMPILE_TESTdrivers/soc/qcom/soc_sleep_stats.c new file +4
-0@@ -0,0 +1,279 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* +* Copyright (c) 2011-2020, The Linux Foundation. All rights reserved. +*/
drivers/soc/qcom/soc_sleep_stats.c line 14
Quoted 10 lines
drivers/soc/qcom/soc_sleep_stats.c new file +10
-0+ +#include <linux/device.h> +#include <linux/io.h> +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/of.h> +#include <linux/of_device.h> +#include <linux/platform_device.h> +#include <linux/slab.h> +#include <linux/uaccess.h>
Why?
drivers/soc/qcom/soc_sleep_stats.c line 17
Quoted 3 lines
drivers/soc/qcom/soc_sleep_stats.c new file +3
-0+#include <linux/debugfs.h> +#include <linux/seq_file.h> +#include <linux/stddef.h>
Why?
Quoted 13 lines
drivers/soc/qcom/soc_sleep_stats.c new file +13
-0+ +#include <linux/soc/qcom/smem.h> +#include <clocksource/arm_arch_timer.h> + +#define NAME_LENGTH 5 + +enum subsystem_item_id { +MODEM =605 , +ADSP , +CDSP , +SLPI , +GPU , +DISPLAY ,
drivers/soc/qcom/soc_sleep_stats.c line 40
Quoted 10 lines
drivers/soc/qcom/soc_sleep_stats.c new file +10
-0+}; + +enum subsystem_pid { +PID_APSS =0 , +PID_MODEM =1 , +PID_ADSP =2 , +PID_SLPI =3 , +PID_CDSP =5 , +PID_GPU =PID_APSS , +PID_DISPLAY =PID_APSS ,
I still don't understand this enums. Why not make them a #define set with known values?
drivers/soc/qcom/soc_sleep_stats.c line 44
Quoted 4 lines
drivers/soc/qcom/soc_sleep_stats.c new file +4
-0+}; + +struct subsystem_data { +char *name;
Can it be const char *?
Quoted 5 lines
drivers/soc/qcom/soc_sleep_stats.c new file +5
-0+enum subsystem_item_id item_id; +enum subsystem_pid pid; +}; + +static struct subsystem_data subsystems[] = {
drivers/soc/qcom/soc_sleep_stats.c line 59
Quoted 10 lines
drivers/soc/qcom/soc_sleep_stats.c new file +10
-0+ { "modem", MODEM, PID_MODEM }, + { "adsp", ADSP, PID_ADSP }, + { "cdsp", CDSP, PID_CDSP }, + { "slpi", SLPI, PID_SLPI }, + { "gpu", GPU, PID_GPU }, + { "display", DISPLAY, PID_DISPLAY }, +}; + +struct stats_config { + u32 offset_addr;
Looks ok but probably can just be unsigned int?
drivers/soc/qcom/soc_sleep_stats.c line 60
Quoted 1 line
drivers/soc/qcom/soc_sleep_stats.c new file +1
-0+u32 num_records;
Why not size_t? Is 32-bit width important?
drivers/soc/qcom/soc_sleep_stats.c line 64
Quoted 4 lines
drivers/soc/qcom/soc_sleep_stats.c new file +4
-0+bool appended_stats_avail; +}; + +static const struct stats_config *config;
This shouldn't be necessary. Can the config be passed into the debugfs file ops private data?
drivers/soc/qcom/soc_sleep_stats.c line 66
Quoted 2 lines
drivers/soc/qcom/soc_sleep_stats.c new file +2
-0+ +struct soc_sleep_stats_data {
This struct is only used in probe. Why not just make probe have more local variables?
drivers/soc/qcom/soc_sleep_stats.c line 78
Quoted 12 lines
drivers/soc/qcom/soc_sleep_stats.c new file +12
-0+phys_addr_t stats_base; +resource_size_t stats_size; +void __iomem *reg; +struct dentry *root; +}; + +struct entry { +__le32 stat_type; +__le32 count; +__le64 last_entered_at; +__le64 last_exited_at; +__le64 accumulated;
These should just be u32 and u64.
drivers/soc/qcom/soc_sleep_stats.c line 84
Quoted 6 lines
drivers/soc/qcom/soc_sleep_stats.c new file +6
-0+}; + +struct appended_entry { +__le32 client_votes; +__le32 reserved[3 ]; +};
Same, just u32.
drivers/soc/qcom/soc_sleep_stats.c line 88
Quoted 4 lines
drivers/soc/qcom/soc_sleep_stats.c new file +4
-0+ +static u64 get_time_in_sec (u64 counter) +{ +do_div (counter,arch_timer_get_rate ());
I don't think arch_timer_get_rate() is exported as a symbol. Why we need to convert this to time? Is the counter not sufficient to understand that so many ticks have passed since it entered and exited?
Quoted 4 lines
drivers/soc/qcom/soc_sleep_stats.c new file +4
-0+ +return counter; +} +
drivers/soc/qcom/soc_sleep_stats.c line 102
Quoted 10 lines
drivers/soc/qcom/soc_sleep_stats.c new file +10
-0+static void print_sleep_stats (struct seq_file *s,struct entry *e) +{ + e->last_entered_at =get_time_in_sec (e->last_entered_at); + e->last_exited_at =get_time_in_sec (e->last_exited_at); + e->accumulated =get_time_in_sec (e->accumulated); + +seq_printf (s,"count = %u\n" , e->count); +seq_printf (s,"Last Entered At = %llu\n" , e->last_entered_at); +seq_printf (s,"Last Exited At = %llu\n" , e->last_exited_at); +seq_printf (s,"Accumulated Duration = %llu\n" , e->accumulated);
Why is count not capitalized but everything else is?
drivers/soc/qcom/soc_sleep_stats.c line 110
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c new file +8
-0+} + +static int subsystem_sleep_stats_show (struct seq_file *s,void *d) +{ +struct subsystem_data *subsystem = s->private; +struct entry *e; + + e =qcom_smem_get (subsystem->pid, subsystem->item_id,NULL );
Do we need to look this up each time we read the stats? Why can't we get this once in probe or when we create the debugfs file?
Quoted 14 lines
drivers/soc/qcom/soc_sleep_stats.c new file +14
-0+if (IS_ERR (e)) +return PTR_ERR (e); + +/* +* If a subsystem is in sleep when reading the sleep stats adjust +* the accumulated sleep duration to show actual sleep time. +*/ +if (e->last_entered_at > e->last_exited_at) + e->accumulated += (arch_timer_read_counter () + - e->last_entered_at); + +print_sleep_stats (s, e); + +return 0 ;
drivers/soc/qcom/soc_sleep_stats.c line 134
Quoted 10 lines
drivers/soc/qcom/soc_sleep_stats.c new file +10
-0+} + +static int soc_sleep_stats_show (struct seq_file *s,void *d) +{ +void __iomem *reg = s->private; +uint32_t offset; +struct entry e; + + offset = offsetof(struct entry , count); + e.count =le32_to_cpu (readl_relaxed (reg + offset));
readl APIs already do endian conversions.
Quoted 9 lines
drivers/soc/qcom/soc_sleep_stats.c new file +9
-0+ + offset = offsetof(struct entry , last_entered_at); + e.last_entered_at =le64_to_cpu (readq_relaxed (reg + offset)); + + offset = offsetof(struct entry , last_exited_at); + e.last_exited_at =le64_to_cpu (readq_relaxed (reg + offset)); + + offset = offsetof(struct entry , accumulated); + e.accumulated =le64_to_cpu (readq_relaxed (reg + offset));
drivers/soc/qcom/soc_sleep_stats.c line 153
Quoted 10 lines
drivers/soc/qcom/soc_sleep_stats.c new file +10
-0+ +print_sleep_stats (s, &e); + +if (config->appended_stats_avail) { +struct appended_entry ae; + + offset = offsetof(struct appended_entry , client_votes) + +sizeof (struct entry ); + ae.client_votes =le32_to_cpu (readl_relaxed (reg + offset)); +seq_printf (s,"Client_votes = 0x%x\n" , ae.client_votes);
Use %#x to avoid having to add 0x prefix. Also, can we replace '=' with ':' so that readability is a bit nicer?
Quoted 14 lines
drivers/soc/qcom/soc_sleep_stats.c new file +14
-0+ } + +return 0 ; +} + +DEFINE_SHOW_ATTRIBUTE (soc_sleep_stats); +DEFINE_SHOW_ATTRIBUTE (subsystem_sleep_stats); + +static const struct stats_config rpm_data = { + .offset_addr =0x14 , + .num_records =2 , + .appended_stats_avail = true, +}; +
drivers/soc/qcom/soc_sleep_stats.c line 177
Quoted 10 lines
drivers/soc/qcom/soc_sleep_stats.c new file +10
-0+static const struct stats_config rpmh_data = { + .offset_addr =0x4 , + .num_records =3 , + .appended_stats_avail = false, +}; + +static const struct of_device_id soc_sleep_stats_table[] = { + { .compatible ="qcom,rpm-sleep-stats" , .data = &rpm_data}, + { .compatible ="qcom,rpmh-sleep-stats" , .data = &rpmh_data}, + { },
Please drop comma after sentinel. It makes a compiler error appear if anything comes after.
drivers/soc/qcom/soc_sleep_stats.c line 178
Quoted 1 line
drivers/soc/qcom/soc_sleep_stats.c new file +1
-0+};
Move this table next to the driver structure please.
drivers/soc/qcom/soc_sleep_stats.c line 183
Quoted 5 lines
drivers/soc/qcom/soc_sleep_stats.c new file +5
-0+ +static int create_debugfs_entries (struct soc_sleep_stats_data *drv) +{ +struct entry *e; +char stat_type[NAME_LENGTH ] = {0 };
Is it a string? Otherwise, seems pretty useless to initialize this to 0 on the stack.
drivers/soc/qcom/soc_sleep_stats.c line 184
Quoted 1 line
drivers/soc/qcom/soc_sleep_stats.c new file +1
-0+uint32_t offset, type;
Just use u32 instead of uint32_t in any kernel code.
drivers/soc/qcom/soc_sleep_stats.c line 189
Quoted 5 lines
drivers/soc/qcom/soc_sleep_stats.c new file +5
-0+int i; + + drv->root =debugfs_create_dir ("soc_sleep_stats" ,NULL ); +if (IS_ERR_OR_NULL (drv->root)) +return PTR_ERR (drv->root);
This is wrong. Debugfs checks have generally been removed because it's not a problem if debugfs fails. When it fails, drv->root will be NULL and any debugfs_create() function will ignore it. Since this driver depends on DEBUGFS being enabled, -ENODEV won't be returned as an error pointer.
drivers/soc/qcom/soc_sleep_stats.c line 193
Quoted 4 lines
drivers/soc/qcom/soc_sleep_stats.c new file +4
-0+ +for (i =0 ; i < config->num_records; i++) { + offset = offsetof(struct entry , stat_type) + + (i *sizeof (struct entry ));
This offset of stuff is sort of complicated. I'd prefer to treat it like how we treat most registers with #defines and base + #define sort of logic. That is easier to read. It also let's us decide to completely reorder struct members and not have to worry that the struct doesn't match how memory is laid out. Instead we have macros that define the offset from some base __iomem pointer.
Quoted 27 lines
drivers/soc/qcom/soc_sleep_stats.c new file +27
-0+ +if (config->appended_stats_avail) + offset += i *sizeof (struct appended_entry ); + + type =le32_to_cpu (readl_relaxed (drv->reg + offset)); +memcpy (stat_type, &type,sizeof (uint32_t )); +debugfs_create_file (stat_type,0444 , drv->root, + drv->reg + offset, + &soc_sleep_stats_fops); + } + +for (i =0 ; i <ARRAY_SIZE (subsystems); i++) { + e =qcom_smem_get (subsystems[i].pid, subsystems[i].item_id, +NULL ); + +if (IS_ERR (e)) +continue ; + +debugfs_create_file (subsystems[i].name,0444 , drv->root, + &subsystems[i], + &subsystem_sleep_stats_fops); + } + +return 0 ; +} + +static int soc_sleep_stats_probe (struct platform_device *pdev)
drivers/soc/qcom/soc_sleep_stats.c line 230
Quoted 10 lines
drivers/soc/qcom/soc_sleep_stats.c new file +10
-0+{ +struct soc_sleep_stats_data *drv; +struct resource *res; +void __iomem *offset_addr; + + drv =devm_kzalloc (&pdev->dev,sizeof (*drv),GFP_KERNEL ); +if (!drv) +return -ENOMEM ; + + config =of_device_get_match_data (&pdev->dev);
Use device_get_match_data() to make this be less DT specific.
drivers/soc/qcom/soc_sleep_stats.c line 238
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c new file +8
-0+if (!config) +return -ENODEV ; + + res =platform_get_resource (pdev,IORESOURCE_MEM ,0 ); +if (!res) +return PTR_ERR (res); + + offset_addr = ioremap_nocache(res->start + config->offset_addr,
Why do we need ioremap_nocache()? Can ioremap() work?
drivers/soc/qcom/soc_sleep_stats.c line 243
Quoted 5 lines
drivers/soc/qcom/soc_sleep_stats.c new file +5
-0+ sizeof(u32)); +if (IS_ERR (offset_addr)) +return PTR_ERR (offset_addr); + + drv->stats_base = res->start |readl_relaxed (offset_addr);
Why | vs. +?
Quoted 14 lines
drivers/soc/qcom/soc_sleep_stats.c new file +14
-0+ drv->stats_size =resource_size (res); +iounmap (offset_addr); + + drv->reg =devm_ioremap (&pdev->dev, drv->stats_base, drv->stats_size); +if (!drv->reg) +return -ENOMEM ; + +if (create_debugfs_entries (drv)) +return -ENODEV ; + +platform_set_drvdata (pdev, drv); + +return 0 ; +}
Re: [PATCH v2 4/4] arm64: defconfig: Enable SoC sleep stats driver for Qualcomm
In reply to Bjorn AnderssonQuoted 22 lines · Bjorn Andersson
Quoted 13 lines · Maulik Shah
Enable SoC sleep stats driver. The driver gives statistics for various SoC level low power modes.
Signed-off-by Maulik Shah <[email protected]>--- arch/arm64/configs/defconfig | 1 + 1 file changed, 1 insertion(+)arch/arm64/configs/defconfig +0
-0arch/arm64/configs/defconfig CONFIG_QCOM_SMD_RPM=y · line 770
Quoted 4 lines
arch/arm64/configs/defconfig +1
-0@@ -767,6 +767,7 @@ CONFIG_QCOM_SMD_RPM=y CONFIG_QCOM_SMP2P=y CONFIG_QCOM_SMSM=y CONFIG_QCOM_SOCINFO=m +CONFIG_QCOM_SOC_SLEEP_STATS=yAfaict the driver is not crucial to boot the system or to collect the statics, so you should be able to make this =m.
Quoted 2 lines
Regards,
Bjorn
Done. Will update in next revision.
Thanks,
Maulik
Quoted 6 lines
Quoted 6 lines
arch/arm64/configs/defconfig +0
-0CONFIG_ARCH_R8A774A1=y CONFIG_ARCH_R8A774B1=y CONFIG_ARCH_R8A774C0=ySignature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
Signature · 1 line
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
Re: [PATCH v2 3/4] arm64: dts: qcom: sc7180: Enable soc sleep stats
In reply to Bjorn AnderssonQuoted 18 lines · Bjorn Andersson
Quoted 16 lines · Maulik Shah
SoC sleep stats provides various low power mode stats.
Signed-off-by Maulik Shah <[email protected]>--- arch/arm64/boot/dts/qcom/sc7180.dtsi | 5 +++++ 1 file changed, 5 insertions(+)arch/arm64/boot/dts/qcom/sc7180.dtsi +0
-0@@ -745,6 +745,11 @@ }; };
Quoted 3 lines
arch/arm64/boot/dts/qcom/sc7180.dtsi line 748
Quoted 1 line
arch/arm64/boot/dts/qcom/sc7180.dtsi +1
-0+ rpmh_sleep_stats: soc-sleep-stats@c3f0000 {I don't see any reason to reference this node, so you should be able to omit the label(?)
Done. Will update in next revision.
Thanks,
Maulik
Quoted 7 lines
arch/arm64/boot/dts/qcom/sc7180.dtsi line 750
Quoted 3 lines
arch/arm64/boot/dts/qcom/sc7180.dtsi +2
-0+ compatible = "qcom,rpmh-sleep-stats"; + reg = <0 0xc3f0000 0 0x400>;Please pad the address to 8 digits and sort the nodes by address.
Regards,
Bjorn
Done. Will update in next revision.
Thanks,
Maulik
Quoted 9 lines
Quoted 9 lines
arch/arm64/boot/dts/qcom/sc7180.dtsi +2
-0+ }; + tcsr_mutex_regs: syscon@1f40000 { compatible = "syscon"; reg = <0 0x01f40000 0 0x40000>;Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
Signature · 1 line
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
Re: [PATCH v2 1/4] dt-bindings: Introduce soc sleep stats bindings for Qualcomm SoCs
In reply to Stephen BoydQuoted 12 lines · Stephen Boyd
Quoting Maulik Shah (2020-02-21 00:49:43)
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml line 4
Quoted 10 lines
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +4
-0@@ -0,0 +1,47 @@ +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) +%YAML 1.2 +--- +$id: http://devicetree.org/schemas/bindings/soc/qcom/soc-sleep-stats.yaml#Drop 'bindings' from above?
Done.
Quoted 10 lines
Quoted 10 lines
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +10
-0+$schema: http://devicetree.org/meta-schemas/core.yaml# + +title: Qualcomm Technologies, Inc. (QTI) SoC sleep stats bindings + +maintainers: + - Maulik Shah <[email protected]> + - Lina Iyer <[email protected]> + +description: | + Always On Processor/Resource Power Manager maintains statistics of the SoC
Quoted 8 lines
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml line 19
Quoted 5 lines
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +5
-0+ sleep modes involving powering down of the rails and oscillator clock. + + Statistics includes SoC sleep mode type, number of times low power mode were + entered, time of last entry, time of last exit and accumulated sleep duration. + SoC sleep stats driver provides debugfs interface to show this information.Please remove this last line. It is a Linuxism that doesn't belong in DT bindings. And then make it one paragraph and drop the | because formatting doesn't need to be maintained.
Done.
Quoted 28 lines
Quoted 11 lines
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +11
-0+ +properties: + compatible: + enum: + - qcom,rpmh-sleep-stats + - qcom,rpm-sleep-stats + + reg: + maxItems: 1 + +required:Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml line 40
Quoted 10 lines
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +10
-0+ - compatible + - reg + +examples: + # Example of rpmh sleep stats + - | + rpmh_sleep_stats: soc-sleep-stats@c3f0000 { + compatible = "qcom,rpmh-sleep-stats"; + reg = <0 0xc3f0000 0 0x400>; + };I see that on sc7180 aoss-qmp is overlapping with this region.
aoss_qmp: qmp@c300000 { compatible = "qcom,sc7180-aoss-qmp"; reg = <0 0x0c300000 0 0x100000>;
Quoted 6 lines
So is this register region really something more like a TCM or RAM area where rpmh combines multiple software concepts into one hardware memory region? The aoss-qmp driver talks about message RAM, so I think this sleep stats stuff is a carveout of the RPMh message RAM. It seems OK if we want to split that message RAM up into multiple DT nodes, but then we'll need to reduce the reg size for the aoss-qmp node.
Yes, i discussed this long back with bjorn, we will limit aoss_qmp size to 0x400. Will include in next revision in DTSI change.
Quoted 3 lines
Finally, the node name 'soc-sleep-stats' is generic, but I wonder why we couldn't name the node 'tcm' or 'memory' or 'msgram'. Similarly for the qmp node it fits better DT style to have the node be something generic.
following DTSI change in this series has comment to drop the label, since no one references this node. if we drop label than just naming it msgram@c3f0000 will also be too generic IMO.
To address both the comments i am planning to drop label and also soc-sleep-stats and just keep rpmh_sleep_stats@c3f0000.
Thanks,
Maulik
Quoted 7 lines
Quoted 7 lines
Documentation/devicetree/bindings/soc/qcom/soc-sleep-stats.yaml new file +7
-0+ # Example of rpm sleep stats + - | + rpm_sleep_stats: soc-sleep-stats@4690000 { + compatible = "qcom,rpm-sleep-stats"; + reg = <0 0x04690000 0 0x400>; + }; +...
Signature · 1 line
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
Re: [PATCH v2 2/4] soc: qcom: Add SoC sleep stats driver
In reply to Stephen BoydQuoted 5 lines · Stephen Boyd
Quoting Maulik Shah (2020-02-21 00:49:44)
Quoted 3 lines
From: Mahesh Sivasubramanian <[email protected]>
Lets's add a driver to read the the stats from remote processor
Let's
Done.
Quoted 30 lines
Quoted 30 lines
and export to debugfs.
The driver creates "
soc_sleep_stats" directory in debugfs and adds files for various low power mode available. Below is sample output with commandcat
/sys/kernel/debug/soc_sleep_stats/ddr
count = 0
Last Entered At = 0
Last Exited At = 0
Accumulated Duration = 0Signed-off-by Mahesh Sivasubramanian <[email protected]>Signed-off-by Lina Iyer <[email protected]>Signed-off-by Maulik Shah <[email protected]>--- drivers/soc/qcom/Kconfig | 10 ++ drivers/soc/qcom/Makefile | 1 + drivers/soc/qcom/soc_sleep_stats.c | 279 +++++++++++++++++++++++++++++++++++++ 3 files changed, 290 insertions(+) create mode 100644 drivers/soc/qcom/soc_sleep_stats.cdrivers/soc/qcom/Kconfig +0
-0@@ -185,6 +185,16 @@ config QCOM_SOCINFO Say yes here to support the Qualcomm socinfo driver, providing information about the SoC to user space.
Quoted 4 lines
drivers/soc/qcom/Kconfig config QCOM_SOCINFO · line 190
Quoted 3 lines
drivers/soc/qcom/Kconfig +3
-0+config QCOM_SOC_SLEEP_STATS + tristate "Qualcomm Technologies, Inc. (QTI) SoC sleep stats driver" + depends on ARCH_QCOMCan you add ||
COMPILE_TEST?
Done.
Quoted 23 lines
Quoted 23 lines
drivers/soc/qcom/Kconfig +7
-0+ depends on DEBUG_FS + help + Qualcomm Technologies, Inc. (QTI) SoC sleep stats driver to read + the shared memory exported by the remote processor related to + various SoC level low power modes statistics and export to debugfs + interface. + config QCOM_WCNSS_CTRL tristate "Qualcomm WCNSS control driver" depends on ARCH_QCOM || COMPILE_TESTdrivers/soc/qcom/soc_sleep_stats.c new file +7
-0@@ -0,0 +1,279 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* +* Copyright (c) 2011-2020, The Linux Foundation. All rights reserved. +*/ + +#include <linux/device.h> +#include <linux/io.h>
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c line 14
Quoted 7 lines
drivers/soc/qcom/soc_sleep_stats.c new file +7
-0+#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/of.h> +#include <linux/of_device.h> +#include <linux/platform_device.h> +#include <linux/slab.h> +#include <linux/uaccess.h> Why?
will remove.
Quoted 4 lines
drivers/soc/qcom/soc_sleep_stats.c line 17
Quoted 3 lines
drivers/soc/qcom/soc_sleep_stats.c new file +3
-0+#include <linux/debugfs.h> +#include <linux/seq_file.h> +#include <linux/stddef.h> Why?
stddef.h was included for offsetof() usage.
from below comments removed offsetof usage in entire driver and also this header.
Quoted 17 lines
Quoted 17 lines
drivers/soc/qcom/soc_sleep_stats.c new file +17
-0+ +#include <linux/soc/qcom/smem.h> +#include <clocksource/arm_arch_timer.h> + +#define NAME_LENGTH 5 + +enum subsystem_item_id { +MODEM =605 , +ADSP , +CDSP , +SLPI , +GPU , +DISPLAY , +}; + +enum subsystem_pid { +PID_APSS =0 ,
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c line 40
Quoted 6 lines
drivers/soc/qcom/soc_sleep_stats.c new file +6
-0+PID_MODEM =1 , +PID_ADSP =2 , +PID_SLPI =3 , +PID_CDSP =5 , +PID_GPU =PID_APSS , + PID_DISPLAY = PID_APSS,I still don't understand this enums. Why not make them a #define set with known values?
Done.
Quoted 5 lines
drivers/soc/qcom/soc_sleep_stats.c line 44
Quoted 4 lines
drivers/soc/qcom/soc_sleep_stats.c new file +4
-0+}; + +struct subsystem_data { +char *name;Can it be const char *?
Done.
Quoted 8 lines
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c new file +8
-0+enum subsystem_item_id item_id; +enum subsystem_pid pid; +}; + +static struct subsystem_data subsystems[] = { + {"modem" ,MODEM ,PID_MODEM }, + {"adsp" ,ADSP ,PID_ADSP }, + {"cdsp" ,CDSP ,PID_CDSP },
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c line 59
Quoted 7 lines
drivers/soc/qcom/soc_sleep_stats.c new file +7
-0+ { "slpi", SLPI, PID_SLPI }, + { "gpu", GPU, PID_GPU }, + { "display", DISPLAY, PID_DISPLAY }, +}; + +struct stats_config { + u32 offset_addr;Looks ok but probably can just be unsigned int?
Done.
Quoted 2 lines
drivers/soc/qcom/soc_sleep_stats.c line 60
Quoted 1 line
drivers/soc/qcom/soc_sleep_stats.c new file +1
-0+u32 num_records;Why not
size_t? Is 32-bit width important?
32-bit width not important.
Quoted 6 lines
drivers/soc/qcom/soc_sleep_stats.c line 64
Quoted 4 lines
drivers/soc/qcom/soc_sleep_stats.c new file +4
-0+bool appended_stats_avail; +}; + +static const struct stats_config *config;This shouldn't be necessary. Can the config be passed into the debugfs file ops private data?
No we are already passing address in debugfs file ops private data from where to start reading stats.
will require to keep this global.
Quoted 4 lines
drivers/soc/qcom/soc_sleep_stats.c line 66
Quoted 2 lines
drivers/soc/qcom/soc_sleep_stats.c new file +2
-0+ +struct soc_sleep_stats_data {This struct is only used in probe. Why not just make probe have more local variables?
This is allocated in probe only
struct soc_sleep_stats_data *drv;
however we pass drv->reg + offset and drv->root in debugfs_create_file, these are used during show to indicate
from where to start reading stats. also during remove,
it currently only uses drv->root member to clean up debugfs, but lets pass entire drv, if there is need to undo something else in future.
Quoted 5 lines
Quoted 5 lines
drivers/soc/qcom/soc_sleep_stats.c new file +5
-0+phys_addr_t stats_base; +resource_size_t stats_size; +void __iomem *reg; +struct dentry *root; +};
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c line 78
Quoted 7 lines
drivers/soc/qcom/soc_sleep_stats.c new file +7
-0+ +struct entry { +__le32 stat_type; +__le32 count; +__le64 last_entered_at; +__le64 last_exited_at; +__le64 accumulated;These should just be u32 and u64.
Done.
Quoted 7 lines
drivers/soc/qcom/soc_sleep_stats.c line 84
Quoted 6 lines
drivers/soc/qcom/soc_sleep_stats.c new file +6
-0+}; + +struct appended_entry { +__le32 client_votes; +__le32 reserved[3 ]; +};Same, just u32.
Done.
Quoted 7 lines
drivers/soc/qcom/soc_sleep_stats.c line 88
Quoted 4 lines
drivers/soc/qcom/soc_sleep_stats.c new file +4
-0+ +static u64 get_time_in_sec (u64 counter) +{ +do_div (counter,arch_timer_get_rate ());I don't think
arch_timer_get_rate()is exported as a symbol. Why we need to convert this to time? Is the counter not sufficient to understand that so many ticks have passed since it entered and exited?
Thanks, yes it is not exported. i think we can display without converting to seconds.
if the need arises to display in seconds, we can add below locally
#define ARCH_TIMER_FREQ 19200000
and then do_div with this value as done in v1.
Quoted 7 lines
Quoted 7 lines
drivers/soc/qcom/soc_sleep_stats.c new file +7
-0+ +return counter; +} + +static void print_sleep_stats (struct seq_file *s,struct entry *e) +{ + e->last_entered_at =get_time_in_sec (e->last_entered_at);
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c line 102
Quoted 7 lines
drivers/soc/qcom/soc_sleep_stats.c new file +7
-0+ e->last_exited_at =get_time_in_sec (e->last_exited_at); + e->accumulated =get_time_in_sec (e->accumulated); + +seq_printf (s,"count = %u\n" , e->count); +seq_printf (s,"Last Entered At = %llu\n" , e->last_entered_at); +seq_printf (s,"Last Exited At = %llu\n" , e->last_exited_at); +seq_printf (s,"Accumulated Duration = %llu\n" , e->accumulated);Why is count not capitalized but everything else is?
Done.
Quoted 10 lines
drivers/soc/qcom/soc_sleep_stats.c line 110
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c new file +8
-0+} + +static int subsystem_sleep_stats_show (struct seq_file *s,void *d) +{ +struct subsystem_data *subsystem = s->private; +struct entry *e; + + e =qcom_smem_get (subsystem->pid, subsystem->item_id,NULL );Do we need to look this up each time we read the stats? Why can't we get this once in probe or when we create the debugfs file?
Yes, we need to look up each time in shared memory for latest values.
Quoted 17 lines
Quoted 17 lines
drivers/soc/qcom/soc_sleep_stats.c new file +17
-0+if (IS_ERR (e)) +return PTR_ERR (e); + +/* +* If a subsystem is in sleep when reading the sleep stats adjust +* the accumulated sleep duration to show actual sleep time. +*/ +if (e->last_entered_at > e->last_exited_at) + e->accumulated += (arch_timer_read_counter () + - e->last_entered_at); + +print_sleep_stats (s, e); + +return 0 ; +} + +static int soc_sleep_stats_show (struct seq_file *s,void *d)
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c line 134
Quoted 7 lines
drivers/soc/qcom/soc_sleep_stats.c new file +7
-0+{ +void __iomem *reg = s->private; +uint32_t offset; +struct entry e; + + offset = offsetof(struct entry , count); + e.count =le32_to_cpu (readl_relaxed (reg + offset));readl APIs already do endian conversions.
Done.
Quoted 13 lines
Quoted 13 lines
drivers/soc/qcom/soc_sleep_stats.c new file +13
-0+ + offset = offsetof(struct entry , last_entered_at); + e.last_entered_at =le64_to_cpu (readq_relaxed (reg + offset)); + + offset = offsetof(struct entry , last_exited_at); + e.last_exited_at =le64_to_cpu (readq_relaxed (reg + offset)); + + offset = offsetof(struct entry , accumulated); + e.accumulated =le64_to_cpu (readq_relaxed (reg + offset)); + +print_sleep_stats (s, &e); + +if (config->appended_stats_avail) {
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c line 153
Quoted 6 lines
drivers/soc/qcom/soc_sleep_stats.c new file +6
-0+struct appended_entry ae; + + offset = offsetof(struct appended_entry , client_votes) + +sizeof (struct entry ); + ae.client_votes =le32_to_cpu (readl_relaxed (reg + offset)); +seq_printf (s,"Client_votes = 0x%x\n" , ae.client_votes);Use %#x to avoid having to add 0x prefix. Also, can we replace '=' with ':' so that readability is a bit nicer?
Done.
Quoted 18 lines
Quoted 18 lines
drivers/soc/qcom/soc_sleep_stats.c new file +18
-0+ } + +return 0 ; +} + +DEFINE_SHOW_ATTRIBUTE (soc_sleep_stats); +DEFINE_SHOW_ATTRIBUTE (subsystem_sleep_stats); + +static const struct stats_config rpm_data = { + .offset_addr =0x14 , + .num_records =2 , + .appended_stats_avail = true, +}; + +static const struct stats_config rpmh_data = { + .offset_addr =0x4 , + .num_records =3 , + .appended_stats_avail = false,
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c line 177
Quoted 6 lines
drivers/soc/qcom/soc_sleep_stats.c new file +6
-0+}; + +static const struct of_device_id soc_sleep_stats_table[] = { + { .compatible ="qcom,rpm-sleep-stats" , .data = &rpm_data}, + { .compatible ="qcom,rpmh-sleep-stats" , .data = &rpmh_data}, + { },Please drop comma after sentinel. It makes a compiler error appear if anything comes after.
Done.
Quoted 2 lines
drivers/soc/qcom/soc_sleep_stats.c line 178
Quoted 1 line
drivers/soc/qcom/soc_sleep_stats.c new file +1
-0+};Move this table next to the driver structure please.
Done.
Quoted 7 lines
drivers/soc/qcom/soc_sleep_stats.c line 183
Quoted 5 lines
drivers/soc/qcom/soc_sleep_stats.c new file +5
-0+ +static int create_debugfs_entries (struct soc_sleep_stats_data *drv) +{ +struct entry *e; +char stat_type[NAME_LENGTH ] = {0 };Is it a string? Otherwise, seems pretty useless to initialize this to 0 on the stack.
yes its string.
Quoted 2 lines
drivers/soc/qcom/soc_sleep_stats.c line 184
Quoted 1 line
drivers/soc/qcom/soc_sleep_stats.c new file +1
-0+uint32_t offset, type;Just use u32 instead of
uint32_tin any kernel code.
Done
Quoted 10 lines
drivers/soc/qcom/soc_sleep_stats.c line 189
Quoted 5 lines
drivers/soc/qcom/soc_sleep_stats.c new file +5
-0+int i; + + drv->root =debugfs_create_dir ("soc_sleep_stats" ,NULL ); +if (IS_ERR_OR_NULL (drv->root)) +return PTR_ERR (drv->root);This is wrong. Debugfs checks have generally been removed because it's not a problem if debugfs fails. When it fails,
drv->rootwill be NULL and anydebugfs_create()function will ignore it. Since this driver depends on DEBUGFS being enabled,-ENODEVwon't be returned as an error pointer.
Done.
Quoted 9 lines
drivers/soc/qcom/soc_sleep_stats.c line 193
Quoted 4 lines
drivers/soc/qcom/soc_sleep_stats.c new file +4
-0+ +for (i =0 ; i < config->num_records; i++) { + offset = offsetof(struct entry , stat_type) + + (i *sizeof (struct entry ));This offset of stuff is sort of complicated. I'd prefer to treat it like how we treat most registers with #defines and base + #define sort of logic. That is easier to read. It also let's us decide to completely reorder struct members and not have to worry that the struct doesn't match how memory is laid out.
Reordering members should not be done. qcom_smem_get() API returns pointer to "struct entry" type and will need keep order of members as it is.
Quoted 2 lines
Instead we have macros that define the
offset from some base__iomempointer.
Okay, i removed offsetof usage from entire driver and using #define now.
Quoted 30 lines
Quoted 30 lines
drivers/soc/qcom/soc_sleep_stats.c new file +30
-0+ +if (config->appended_stats_avail) + offset += i *sizeof (struct appended_entry ); + + type =le32_to_cpu (readl_relaxed (drv->reg + offset)); +memcpy (stat_type, &type,sizeof (uint32_t )); +debugfs_create_file (stat_type,0444 , drv->root, + drv->reg + offset, + &soc_sleep_stats_fops); + } + +for (i =0 ; i <ARRAY_SIZE (subsystems); i++) { + e =qcom_smem_get (subsystems[i].pid, subsystems[i].item_id, +NULL ); + +if (IS_ERR (e)) +continue ; + +debugfs_create_file (subsystems[i].name,0444 , drv->root, + &subsystems[i], + &subsystem_sleep_stats_fops); + } + +return 0 ; +} + +static int soc_sleep_stats_probe (struct platform_device *pdev) +{ +struct soc_sleep_stats_data *drv; +struct resource *res;
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c line 230
Quoted 7 lines
drivers/soc/qcom/soc_sleep_stats.c new file +7
-0+void __iomem *offset_addr; + + drv =devm_kzalloc (&pdev->dev,sizeof (*drv),GFP_KERNEL ); +if (!drv) +return -ENOMEM ; + + config =of_device_get_match_data (&pdev->dev);Use
device_get_match_data()to make this be less DT specific.
Done.
Quoted 9 lines
drivers/soc/qcom/soc_sleep_stats.c line 238
Quoted 8 lines
drivers/soc/qcom/soc_sleep_stats.c new file +8
-0+if (!config) +return -ENODEV ; + + res =platform_get_resource (pdev,IORESOURCE_MEM ,0 ); +if (!res) +return PTR_ERR (res); + + offset_addr = ioremap_nocache(res->start + config->offset_addr,Why do we need
ioremap_nocache()? Canioremap()work?
Will update it to use devm_ioremap().
Quoted 6 lines
drivers/soc/qcom/soc_sleep_stats.c line 243
Quoted 5 lines
drivers/soc/qcom/soc_sleep_stats.c new file +5
-0+ sizeof(u32)); +if (IS_ERR (offset_addr)) +return PTR_ERR (offset_addr); + + drv->stats_base = res->start |readl_relaxed (offset_addr);Why | vs. +?
This is how final address need to be formed.
Thanks,
Maulik
Quoted 14 lines
Quoted 14 lines
drivers/soc/qcom/soc_sleep_stats.c new file +14
-0+ drv->stats_size =resource_size (res); +iounmap (offset_addr); + + drv->reg =devm_ioremap (&pdev->dev, drv->stats_base, drv->stats_size); +if (!drv->reg) +return -ENOMEM ; + +if (create_debugfs_entries (drv)) +return -ENODEV ; + +platform_set_drvdata (pdev, drv); + +return 0 ; +}
Signature · 1 line
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation