| Message ID | 1584505758-21037-3-git-send-email-mkshah@codeaurora.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Introduce SoC sleep stats driver | expand |
Quoting Maulik Shah (2020-03-17 21:29:16) > diff --git a/drivers/soc/qcom/soc_sleep_stats.c b/drivers/soc/qcom/soc_sleep_stats.c > new file mode 100644 > index 0000000..0db7c3d > --- /dev/null > +++ b/drivers/soc/qcom/soc_sleep_stats.c > @@ -0,0 +1,244 @@ > +// SPDX-License-Identifier: GPL-2.0-only > +/* > + * Copyright (c) 2011-2020, The Linux Foundation. All rights reserved. > + */ > + > +#include <linux/debugfs.h> > +#include <linux/device.h> > +#include <linux/io.h> > +#include <linux/module.h> > +#include <linux/of.h> > +#include <linux/platform_device.h> > +#include <linux/seq_file.h> > + > +#include <linux/soc/qcom/smem.h> > +#include <clocksource/arm_arch_timer.h> > + > +#define STAT_TYPE_ADDR 0x0 > +#define COUNT_ADDR 0x4 > +#define LAST_ENTERED_AT_ADDR 0x8 > +#define LAST_EXITED_AT_ADDR 0x10 > +#define ACCUMULATED_ADDR 0x18 > +#define CLIENT_VOTES_ADDR 0x1c > + > +struct subsystem_data { > + const char *name; > + u32 smem_item; > + u32 pid; > +}; > + > +static struct subsystem_data subsystems[] = { This can be const. > + { "modem", 605, 1 }, > + { "adsp", 606, 2 }, > + { "cdsp", 607, 5 }, > + { "slpi", 608, 3 }, > + { "gpu", 609, 0 }, > + { "display", 610, 0 }, > +}; > + > +struct stats_config { > + unsigned int offset_addr; > + unsigned int num_records; > + bool appended_stats_avail; > +}; > + > +struct stats_prv_data { > + const struct stats_config *config; Can we have 'bool has_appended_stats' here instead? > + void __iomem *reg; > +}; > + > +struct sleep_stats { > + u32 stat_type; > + u32 count; > + u64 last_entered_at; > + u64 last_exited_at; > + u64 accumulated; > +}; > + > +struct appended_stats { > + u32 client_votes; > + u32 reserved[3]; > +}; > + > +static void print_sleep_stats(struct seq_file *s, struct sleep_stats *stat) Make stat const please. > +{ > + u64 accumulated = stat->accumulated; > + /* > + * If a subsystem is in sleep when reading the sleep stats adjust > + * the accumulated sleep duration to show actual sleep time. > + */ > + if (stat->last_entered_at > stat->last_exited_at) > + accumulated += arch_timer_read_counter() This assumes that arch_timer_read_counter() is returning the physical counter? Is accumulated duration useful for anything? I don't like that we're relying on the arch timer code to return a certain counter value that may or may not be the same as what is written into smem. > + - stat->last_entered_at; > + > + seq_printf(s, "Count = %u\n", stat->count); > + seq_printf(s, "Last Entered At = %llu\n", stat->last_entered_at); > + seq_printf(s, "Last Exited At = %llu\n", stat->last_exited_at); > + seq_printf(s, "Accumulated Duration = %llu\n", accumulated); > +} > + > +static int subsystem_sleep_stats_show(struct seq_file *s, void *d) > +{ > + struct subsystem_data *subsystem = s->private; > + struct sleep_stats *stat; > + > + stat = qcom_smem_get(subsystem->pid, subsystem->smem_item, NULL); We already get this during probe in create_debugfs_entries() (which is too generic of a name by the way). Why can't we stash that pointer away so that it comes through the 'd' parameter to this function? > + if (IS_ERR(stat)) > + return PTR_ERR(stat); > + > + print_sleep_stats(s, stat); > + > + return 0; > +} > + > +static int soc_sleep_stats_show(struct seq_file *s, void *d) > +{ > + struct stats_prv_data *prv_data = s->private; > + void __iomem *reg = prv_data->reg; > + struct sleep_stats stat; > + > + stat.count = readl(reg + COUNT_ADDR); > + stat.last_entered_at = readq(reg + LAST_ENTERED_AT_ADDR); > + stat.last_exited_at = readq(reg + LAST_EXITED_AT_ADDR); > + stat.accumulated = readq(reg + ACCUMULATED_ADDR); > + > + print_sleep_stats(s, &stat); Ok I see. The same stat info is in SMEM and also in some IO memory location? So we have to read one from an iomem region and one from smem? Please make subsystem_sleep_stats_show() take a 'struct smem_sleep_stats' through 'd' that has the __le32 and __le64 markings and then do the byte swaps into a stack local struct sleep_stats that print_sleep_stats() can use. > + > + if (prv_data->config->appended_stats_avail) { > + struct appended_stats app_stat; Just use a u32 for this. The struct is not useful here. > + > + app_stat.client_votes = readl(reg + CLIENT_VOTES_ADDR); > + seq_printf(s, "Client_votes = %#x\n", app_stat.client_votes); > + } > + > + return 0; > +} > + > +DEFINE_SHOW_ATTRIBUTE(soc_sleep_stats); > +DEFINE_SHOW_ATTRIBUTE(subsystem_sleep_stats); > + > +static struct dentry *create_debugfs_entries(void __iomem *reg, > + struct stats_prv_data *prv_data) I'd prefer this was just inlined into probe. > +{ > + struct dentry *root; > + struct sleep_stats *stat; > + char stat_type[sizeof(u32) + 1] = {0}; > + u32 offset, type; > + int i; > + > + root = debugfs_create_dir("qcom_sleep_stats", NULL); > + > + for (i = 0; i < prv_data[0].config->num_records; i++) { Pass config as another argument instead of attaching it to prv_data please. > + offset = STAT_TYPE_ADDR + (i * sizeof(struct sleep_stats)); > + > + if (prv_data[0].config->appended_stats_avail) > + offset += i * sizeof(struct appended_stats); I was imagining a macro like, SLEEP_STATn(i, has_appended) that did all the math to figure out the offset. > + > + prv_data[i].reg = reg + offset; > + > + type = readl(prv_data[i].reg); > + memcpy(stat_type, &type, sizeof(u32)); Is it a 4-byte ascii register that may or may not be NUL terminated? We should use __raw_readl() then so we don't do an endian swap. Using memcpy_fromio() does this all for you. > + debugfs_create_file(stat_type, 0444, root, Maybe just 0400? Not sure why everyone in the world needs to read this. > + &prv_data[i], > + &soc_sleep_stats_fops); > + } > + > + for (i = 0; i < ARRAY_SIZE(subsystems); i++) { > + stat = qcom_smem_get(subsystems[i].pid, subsystems[i].smem_item, > + NULL); > + if (IS_ERR(stat)) > + continue; > + > + debugfs_create_file(subsystems[i].name, 0444, root, > + &subsystems[i], > + &subsystem_sleep_stats_fops); > + } > + > + return root; > +} > + > +static int soc_sleep_stats_probe(struct platform_device *pdev) > +{ > + struct resource *res; > + void __iomem *reg; > + void __iomem *offset_addr; > + phys_addr_t stats_base; > + resource_size_t stats_size; > + struct dentry *root; > + const struct stats_config *config; > + struct stats_prv_data *prv_data; > + int i; > + > + config = device_get_match_data(&pdev->dev); > + if (!config) > + return -ENODEV; > + > + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); > + if (!res) > + return PTR_ERR(res); > + > + offset_addr = ioremap(res->start + config->offset_addr, sizeof(u32)); > + if (IS_ERR(offset_addr)) > + return PTR_ERR(offset_addr); > + > + stats_base = res->start | readl_relaxed(offset_addr); I still don't get it, but whatever. I highly doubt the lower bits are anything besides 0 so a logical OR vs addition is not different. > + stats_size = resource_size(res); > + iounmap(offset_addr); > + > + reg = devm_ioremap(&pdev->dev, stats_base, stats_size); > + if (!reg) > + return -ENOMEM; > + > + prv_data = devm_kzalloc(&pdev->dev, config->num_records * > + sizeof(struct stats_prv_data), GFP_KERNEL); We have devm_kcalloc() for array allocations. > + if (!prv_data) > + return -ENOMEM; > + > + for (i = 0; i < config->num_records; i++) > + prv_data[i].config = config; > + > + root = create_debugfs_entries(reg, prv_data); > + platform_set_drvdata(pdev, root); > + > + return 0; > +}
Hi, On 3/19/2020 5:55 AM, Stephen Boyd wrote: > Quoting Maulik Shah (2020-03-17 21:29:16) >> diff --git a/drivers/soc/qcom/soc_sleep_stats.c b/drivers/soc/qcom/soc_sleep_stats.c >> new file mode 100644 >> index 0000000..0db7c3d >> --- /dev/null >> +++ b/drivers/soc/qcom/soc_sleep_stats.c >> @@ -0,0 +1,244 @@ >> +// SPDX-License-Identifier: GPL-2.0-only >> +/* >> + * Copyright (c) 2011-2020, The Linux Foundation. All rights reserved. >> + */ >> + >> +#include <linux/debugfs.h> >> +#include <linux/device.h> >> +#include <linux/io.h> >> +#include <linux/module.h> >> +#include <linux/of.h> >> +#include <linux/platform_device.h> >> +#include <linux/seq_file.h> >> + >> +#include <linux/soc/qcom/smem.h> >> +#include <clocksource/arm_arch_timer.h> >> + >> +#define STAT_TYPE_ADDR 0x0 >> +#define COUNT_ADDR 0x4 >> +#define LAST_ENTERED_AT_ADDR 0x8 >> +#define LAST_EXITED_AT_ADDR 0x10 >> +#define ACCUMULATED_ADDR 0x18 >> +#define CLIENT_VOTES_ADDR 0x1c >> + >> +struct subsystem_data { >> + const char *name; >> + u32 smem_item; >> + u32 pid; >> +}; >> + >> +static struct subsystem_data subsystems[] = { > This can be const. we are passing each element of this in debugfs_create_file as (void * data) making this const is saying error passing const as void *. >> + { "modem", 605, 1 }, >> + { "adsp", 606, 2 }, >> + { "cdsp", 607, 5 }, >> + { "slpi", 608, 3 }, >> + { "gpu", 609, 0 }, >> + { "display", 610, 0 }, >> +}; >> + >> +struct stats_config { >> + unsigned int offset_addr; >> + unsigned int num_records; >> + bool appended_stats_avail; >> +}; >> + >> +struct stats_prv_data { >> + const struct stats_config *config; > Can we have 'bool has_appended_stats' here instead? any reason to move? this is per compatible config where rpm based target have appended stats available. > >> + void __iomem *reg; >> +}; >> + >> +struct sleep_stats { >> + u32 stat_type; >> + u32 count; >> + u64 last_entered_at; >> + u64 last_exited_at; >> + u64 accumulated; >> +}; >> + >> +struct appended_stats { >> + u32 client_votes; >> + u32 reserved[3]; >> +}; >> + >> +static void print_sleep_stats(struct seq_file *s, struct sleep_stats *stat) > Make stat const please. Sure, i will address this. > >> +{ >> + u64 accumulated = stat->accumulated; >> + /* >> + * If a subsystem is in sleep when reading the sleep stats adjust >> + * the accumulated sleep duration to show actual sleep time. >> + */ >> + if (stat->last_entered_at > stat->last_exited_at) >> + accumulated += arch_timer_read_counter() > This assumes that arch_timer_read_counter() is returning the physical > counter? Is accumulated duration useful for anything? I don't like that > we're relying on the arch timer code to return a certain counter value > that may or may not be the same as what is written into smem. we are not comparing it with what is written into smem. The idea here is to adjust the accumulated sleep length to reflect close/real value. When a subsystem goes to sleep, it updates entry time and then stays in sleep. Exit time and accumulated duration is updated when subsystem comes out of low power mode. Now if Subsystem updated entry time and went in sleep, while printing accumulated duration will keep giving older value. If read it at interval of say every 10 seconds, if subsystem never comes out during this. Even after 10 seconds, it gives older accumulated duration, while we want to get printed as close to real value. so its updated here. Now when it comes out of sleep, it always prints value given from subsystem. > >> + - stat->last_entered_at; >> + >> + seq_printf(s, "Count = %u\n", stat->count); >> + seq_printf(s, "Last Entered At = %llu\n", stat->last_entered_at); >> + seq_printf(s, "Last Exited At = %llu\n", stat->last_exited_at); >> + seq_printf(s, "Accumulated Duration = %llu\n", accumulated); >> +} >> + >> +static int subsystem_sleep_stats_show(struct seq_file *s, void *d) >> +{ >> + struct subsystem_data *subsystem = s->private; >> + struct sleep_stats *stat; >> + >> + stat = qcom_smem_get(subsystem->pid, subsystem->smem_item, NULL); > We already get this during probe in create_debugfs_entries() (which is > too generic of a name by the way). I will update the name. > Why can't we stash that pointer away > so that it comes through the 'd' parameter to this function? i think you are missing a subsystem restart case, in that pointer may be updated to new value. so we can not just save it and re-use it every time, we don't know when subsystem restart happens and we now need new pointer. > >> + if (IS_ERR(stat)) >> + return PTR_ERR(stat); >> + >> + print_sleep_stats(s, stat); >> + >> + return 0; >> +} >> + >> +static int soc_sleep_stats_show(struct seq_file *s, void *d) >> +{ >> + struct stats_prv_data *prv_data = s->private; >> + void __iomem *reg = prv_data->reg; >> + struct sleep_stats stat; >> + >> + stat.count = readl(reg + COUNT_ADDR); >> + stat.last_entered_at = readq(reg + LAST_ENTERED_AT_ADDR); >> + stat.last_exited_at = readq(reg + LAST_EXITED_AT_ADDR); >> + stat.accumulated = readq(reg + ACCUMULATED_ADDR); >> + >> + print_sleep_stats(s, &stat); > Ok I see. The same stat info is in SMEM and also in some IO memory > location? So we have to read one from an iomem region and one from smem? Yes. > Please make subsystem_sleep_stats_show() take a 'struct > smem_sleep_stats' through 'd' that has the __le32 and __le64 markings > and then do the byte swaps into a stack local struct sleep_stats that > print_sleep_stats() can use. Let me check this, i don't see every user of qcom_smem_get() doing byte swaps. passing this through 'd' may not be required even if we want to do endian conversion. 'd' is alreaedy occupied to pass the subsystem_data structure item. > >> + >> + if (prv_data->config->appended_stats_avail) { >> + struct appended_stats app_stat; > Just use a u32 for this. The struct is not useful here. Okay. i will address this. > >> + >> + app_stat.client_votes = readl(reg + CLIENT_VOTES_ADDR); >> + seq_printf(s, "Client_votes = %#x\n", app_stat.client_votes); >> + } >> + >> + return 0; >> +} >> + >> +DEFINE_SHOW_ATTRIBUTE(soc_sleep_stats); >> +DEFINE_SHOW_ATTRIBUTE(subsystem_sleep_stats); >> + >> +static struct dentry *create_debugfs_entries(void __iomem *reg, >> + struct stats_prv_data *prv_data) > I'd prefer this was just inlined into probe. I would like to keep it same way. > >> +{ >> + struct dentry *root; >> + struct sleep_stats *stat; >> + char stat_type[sizeof(u32) + 1] = {0}; >> + u32 offset, type; >> + int i; >> + >> + root = debugfs_create_dir("qcom_sleep_stats", NULL); >> + >> + for (i = 0; i < prv_data[0].config->num_records; i++) { > Pass config as another argument instead of attaching it to prv_data > please. Okay i will address this. > >> + offset = STAT_TYPE_ADDR + (i * sizeof(struct sleep_stats)); >> + >> + if (prv_data[0].config->appended_stats_avail) >> + offset += i * sizeof(struct appended_stats); > I was imagining a macro like, SLEEP_STATn(i, has_appended) that did all the > math to figure out the offset. Okay i will update driver to use macro. > >> + >> + prv_data[i].reg = reg + offset; >> + >> + type = readl(prv_data[i].reg); >> + memcpy(stat_type, &type, sizeof(u32)); > Is it a 4-byte ascii register that may or may not be NUL terminated? We > should use __raw_readl() then so we don't do an endian swap. Using > memcpy_fromio() does this all for you. memcpy_fromio() seems not copying properly with u32 or u32+1 size. seems it need align to 8 byte. so when i pass u64 size it seems working fine. i will change this to use memcpy_fromio with sizeof(u64). > >> + debugfs_create_file(stat_type, 0444, root, > Maybe just 0400? Not sure why everyone in the world needs to read this. Okay i will address this. > >> + &prv_data[i], >> + &soc_sleep_stats_fops); >> + } >> + >> + for (i = 0; i < ARRAY_SIZE(subsystems); i++) { >> + stat = qcom_smem_get(subsystems[i].pid, subsystems[i].smem_item, >> + NULL); >> + if (IS_ERR(stat)) >> + continue; >> + >> + debugfs_create_file(subsystems[i].name, 0444, root, >> + &subsystems[i], >> + &subsystem_sleep_stats_fops); >> + } >> + >> + return root; >> +} >> + >> +static int soc_sleep_stats_probe(struct platform_device *pdev) >> +{ >> + struct resource *res; >> + void __iomem *reg; >> + void __iomem *offset_addr; >> + phys_addr_t stats_base; >> + resource_size_t stats_size; >> + struct dentry *root; >> + const struct stats_config *config; >> + struct stats_prv_data *prv_data; >> + int i; >> + >> + config = device_get_match_data(&pdev->dev); >> + if (!config) >> + return -ENODEV; >> + >> + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); >> + if (!res) >> + return PTR_ERR(res); >> + >> + offset_addr = ioremap(res->start + config->offset_addr, sizeof(u32)); >> + if (IS_ERR(offset_addr)) >> + return PTR_ERR(offset_addr); >> + >> + stats_base = res->start | readl_relaxed(offset_addr); > I still don't get it, but whatever. I highly doubt the lower bits are > anything besides 0 so a logical OR vs addition is not different. its different and need to do logical OR. > >> + stats_size = resource_size(res); >> + iounmap(offset_addr); >> + >> + reg = devm_ioremap(&pdev->dev, stats_base, stats_size); >> + if (!reg) >> + return -ENOMEM; >> + >> + prv_data = devm_kzalloc(&pdev->dev, config->num_records * >> + sizeof(struct stats_prv_data), GFP_KERNEL); > We have devm_kcalloc() for array allocations. Okay i will address this. > >> + if (!prv_data) >> + return -ENOMEM; >> + >> + for (i = 0; i < config->num_records; i++) >> + prv_data[i].config = config; >> + >> + root = create_debugfs_entries(reg, prv_data); >> + platform_set_drvdata(pdev, root); >> + >> + return 0; >> +} Thanks, Maulik
Quoting Maulik Shah (2020-03-19 08:14:03) > On 3/19/2020 5:55 AM, Stephen Boyd wrote: > > Quoting Maulik Shah (2020-03-17 21:29:16) > >> diff --git a/drivers/soc/qcom/soc_sleep_stats.c b/drivers/soc/qcom/soc_sleep_stats.c > >> new file mode 100644 > >> index 0000000..0db7c3d > >> --- /dev/null > >> +++ b/drivers/soc/qcom/soc_sleep_stats.c > >> @@ -0,0 +1,244 @@ [..] > >> + > >> +static struct subsystem_data subsystems[] = { > > This can be const. > we are passing each element of this in debugfs_create_file as (void * data) > making this const is saying error passing const as void *. I think we can cast it and have a comment. This way the structure can be placed in the RO section of memory. > >> + { "modem", 605, 1 }, > >> + { "adsp", 606, 2 }, > >> + { "cdsp", 607, 5 }, > >> + { "slpi", 608, 3 }, > >> + { "gpu", 609, 0 }, > >> + { "display", 610, 0 }, > >> +}; > >> + > >> +struct stats_config { > >> + unsigned int offset_addr; > >> + unsigned int num_records; > >> + bool appended_stats_avail; > >> +}; > >> + > >> +struct stats_prv_data { > >> + const struct stats_config *config; > > Can we have 'bool has_appended_stats' here instead? > any reason to move? this is per compatible config where rpm based target have appended stats available. It avoids one indirection to figure out something and makes the code clearer. prv_data doesn't have any use for config otherwise. > > > >> +{ > >> + u64 accumulated = stat->accumulated; > >> + /* > >> + * If a subsystem is in sleep when reading the sleep stats adjust > >> + * the accumulated sleep duration to show actual sleep time. > >> + */ > >> + if (stat->last_entered_at > stat->last_exited_at) > >> + accumulated += arch_timer_read_counter() > > This assumes that arch_timer_read_counter() is returning the physical > > counter? Is accumulated duration useful for anything? I don't like that > > we're relying on the arch timer code to return a certain counter value > > that may or may not be the same as what is written into smem. > > we are not comparing it with what is written into smem. The idea here is to adjust the accumulated sleep length to reflect close/real value. > > When a subsystem goes to sleep, it updates entry time and then stays in sleep. > Exit time and accumulated duration is updated when subsystem comes out of low power mode. Who is updating the accumulated time? This debugfs code? Or the subsystems themselves? > > Now if Subsystem updated entry time and went in sleep, while printing accumulated duration > will keep giving older value. > > If read it at interval of say every 10 seconds, if subsystem never comes out during this. > Even after 10 seconds, it gives older accumulated duration, while we want to get > printed as close to real value. so its updated here. > > Now when it comes out of sleep, it always prints value given from subsystem. Thanks, I understand the motivation. > > > > >> + - stat->last_entered_at; > >> + > >> + seq_printf(s, "Count = %u\n", stat->count); > >> + seq_printf(s, "Last Entered At = %llu\n", stat->last_entered_at); > >> + seq_printf(s, "Last Exited At = %llu\n", stat->last_exited_at); > >> + seq_printf(s, "Accumulated Duration = %llu\n", accumulated); > >> +} > >> + > >> +static int subsystem_sleep_stats_show(struct seq_file *s, void *d) > >> +{ > >> + struct subsystem_data *subsystem = s->private; > >> + struct sleep_stats *stat; > >> + > >> + stat = qcom_smem_get(subsystem->pid, subsystem->smem_item, NULL); > > We already get this during probe in create_debugfs_entries() (which is > > too generic of a name by the way). > I will update the name. > > Why can't we stash that pointer away > > so that it comes through the 'd' parameter to this function? > i think you are missing a subsystem restart case, in that pointer may be updated to new value. > so we can not just save it and re-use it every time, we don't know when subsystem restart happens and we now need new pointer. Yes I'm missing subsystem restart. A comment would certainly help. If this is the case, maybe this should be libified and each subsystem driver should call some add/remove function in here so that we can refetch the smem pointer. Doing the lookup each time isn't great. > > > >> + > >> + prv_data[i].reg = reg + offset; > >> + > >> + type = readl(prv_data[i].reg); > >> + memcpy(stat_type, &type, sizeof(u32)); > > Is it a 4-byte ascii register that may or may not be NUL terminated? We > > should use __raw_readl() then so we don't do an endian swap. Using > > memcpy_fromio() does this all for you. > > memcpy_fromio() seems not copying properly with u32 or u32+1 size. seems it need align to 8 byte. > > so when i pass u64 size it seems working fine. i will change this to use memcpy_fromio with sizeof(u64). What is the failure? Do you need to read 4 bytes at a time? On arm64 memcpy_fromio() looks to do byte size reads if the size isn't 8 bytes. Maybe just use __raw_readl() then if it is only 4 bytes and you need a word accessor.
On Wed, Mar 18, 2020 at 09:59:16AM +0530, Maulik Shah wrote: > From: Mahesh Sivasubramanian <msivasub@codeaurora.org> > > Let's add a driver to read the the stats from remote processor > and export to debugfs. > > The driver creates "qcom_sleep_stats" directory in debugfs and adds > files for various low power mode available. Below is sample output > with command > > cat /sys/kernel/debug/qcom_sleep_stats/ddr > count = 0 > Last Entered At = 0 > Last Exited At = 0 > Accumulated Duration = 0 > > Signed-off-by: Mahesh Sivasubramanian <msivasub@codeaurora.org> > Signed-off-by: Lina Iyer <ilina@codeaurora.org> > [mkshah: add subsystem sleep stats, create one file for each stat] > Signed-off-by: Maulik Shah <mkshah@codeaurora.org> > --- > drivers/soc/qcom/Kconfig | 9 ++ > drivers/soc/qcom/Makefile | 1 + > drivers/soc/qcom/soc_sleep_stats.c | 244 +++++++++++++++++++++++++++++++++++++ > 3 files changed, 254 insertions(+) > create mode 100644 drivers/soc/qcom/soc_sleep_stats.c > > diff --git a/drivers/soc/qcom/Kconfig b/drivers/soc/qcom/Kconfig > index d0a73e7..08bc0df3 100644 > --- a/drivers/soc/qcom/Kconfig > +++ b/drivers/soc/qcom/Kconfig > @@ -185,6 +185,15 @@ 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 && DEBUG_FS || COMPILE_TEST depends on QCOM_SMEM Otherwise causes compile error if QCOM_SMEM=n. Guenter > + 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 > diff --git a/drivers/soc/qcom/Makefile b/drivers/soc/qcom/Makefile > index 9fb35c8..e6bd052 100644 > --- a/drivers/soc/qcom/Makefile > +++ b/drivers/soc/qcom/Makefile > @@ -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 > diff --git a/drivers/soc/qcom/soc_sleep_stats.c b/drivers/soc/qcom/soc_sleep_stats.c > new file mode 100644 > index 0000000..0db7c3d > --- /dev/null > +++ b/drivers/soc/qcom/soc_sleep_stats.c > @@ -0,0 +1,244 @@ > +// SPDX-License-Identifier: GPL-2.0-only > +/* > + * Copyright (c) 2011-2020, The Linux Foundation. All rights reserved. > + */ > + > +#include <linux/debugfs.h> > +#include <linux/device.h> > +#include <linux/io.h> > +#include <linux/module.h> > +#include <linux/of.h> > +#include <linux/platform_device.h> > +#include <linux/seq_file.h> > + > +#include <linux/soc/qcom/smem.h> > +#include <clocksource/arm_arch_timer.h> > + > +#define STAT_TYPE_ADDR 0x0 > +#define COUNT_ADDR 0x4 > +#define LAST_ENTERED_AT_ADDR 0x8 > +#define LAST_EXITED_AT_ADDR 0x10 > +#define ACCUMULATED_ADDR 0x18 > +#define CLIENT_VOTES_ADDR 0x1c > + > +struct subsystem_data { > + const char *name; > + u32 smem_item; > + u32 pid; > +}; > + > +static struct subsystem_data subsystems[] = { > + { "modem", 605, 1 }, > + { "adsp", 606, 2 }, > + { "cdsp", 607, 5 }, > + { "slpi", 608, 3 }, > + { "gpu", 609, 0 }, > + { "display", 610, 0 }, > +}; > + > +struct stats_config { > + unsigned int offset_addr; > + unsigned int num_records; > + bool appended_stats_avail; > +}; > + > +struct stats_prv_data { > + const struct stats_config *config; > + void __iomem *reg; > +}; > + > +struct sleep_stats { > + u32 stat_type; > + u32 count; > + u64 last_entered_at; > + u64 last_exited_at; > + u64 accumulated; > +}; > + > +struct appended_stats { > + u32 client_votes; > + u32 reserved[3]; > +}; > + > +static void print_sleep_stats(struct seq_file *s, struct sleep_stats *stat) > +{ > + u64 accumulated = stat->accumulated; > + /* > + * If a subsystem is in sleep when reading the sleep stats adjust > + * the accumulated sleep duration to show actual sleep time. > + */ > + if (stat->last_entered_at > stat->last_exited_at) > + accumulated += arch_timer_read_counter() > + - stat->last_entered_at; > + > + seq_printf(s, "Count = %u\n", stat->count); > + seq_printf(s, "Last Entered At = %llu\n", stat->last_entered_at); > + seq_printf(s, "Last Exited At = %llu\n", stat->last_exited_at); > + seq_printf(s, "Accumulated Duration = %llu\n", accumulated); > +} > + > +static int subsystem_sleep_stats_show(struct seq_file *s, void *d) > +{ > + struct subsystem_data *subsystem = s->private; > + struct sleep_stats *stat; > + > + stat = qcom_smem_get(subsystem->pid, subsystem->smem_item, NULL); > + if (IS_ERR(stat)) > + return PTR_ERR(stat); > + > + print_sleep_stats(s, stat); > + > + return 0; > +} > + > +static int soc_sleep_stats_show(struct seq_file *s, void *d) > +{ > + struct stats_prv_data *prv_data = s->private; > + void __iomem *reg = prv_data->reg; > + struct sleep_stats stat; > + > + stat.count = readl(reg + COUNT_ADDR); > + stat.last_entered_at = readq(reg + LAST_ENTERED_AT_ADDR); > + stat.last_exited_at = readq(reg + LAST_EXITED_AT_ADDR); > + stat.accumulated = readq(reg + ACCUMULATED_ADDR); > + > + print_sleep_stats(s, &stat); > + > + if (prv_data->config->appended_stats_avail) { > + struct appended_stats app_stat; > + > + app_stat.client_votes = readl(reg + CLIENT_VOTES_ADDR); > + seq_printf(s, "Client_votes = %#x\n", app_stat.client_votes); > + } > + > + return 0; > +} > + > +DEFINE_SHOW_ATTRIBUTE(soc_sleep_stats); > +DEFINE_SHOW_ATTRIBUTE(subsystem_sleep_stats); > + > +static struct dentry *create_debugfs_entries(void __iomem *reg, > + struct stats_prv_data *prv_data) > +{ > + struct dentry *root; > + struct sleep_stats *stat; > + char stat_type[sizeof(u32) + 1] = {0}; > + u32 offset, type; > + int i; > + > + root = debugfs_create_dir("qcom_sleep_stats", NULL); > + > + for (i = 0; i < prv_data[0].config->num_records; i++) { > + offset = STAT_TYPE_ADDR + (i * sizeof(struct sleep_stats)); > + > + if (prv_data[0].config->appended_stats_avail) > + offset += i * sizeof(struct appended_stats); > + > + prv_data[i].reg = reg + offset; > + > + type = readl(prv_data[i].reg); > + memcpy(stat_type, &type, sizeof(u32)); > + debugfs_create_file(stat_type, 0444, root, > + &prv_data[i], > + &soc_sleep_stats_fops); > + } > + > + for (i = 0; i < ARRAY_SIZE(subsystems); i++) { > + stat = qcom_smem_get(subsystems[i].pid, subsystems[i].smem_item, > + NULL); > + if (IS_ERR(stat)) > + continue; > + > + debugfs_create_file(subsystems[i].name, 0444, root, > + &subsystems[i], > + &subsystem_sleep_stats_fops); > + } > + > + return root; > +} > + > +static int soc_sleep_stats_probe(struct platform_device *pdev) > +{ > + struct resource *res; > + void __iomem *reg; > + void __iomem *offset_addr; > + phys_addr_t stats_base; > + resource_size_t stats_size; > + struct dentry *root; > + const struct stats_config *config; > + struct stats_prv_data *prv_data; > + int i; > + > + config = device_get_match_data(&pdev->dev); > + if (!config) > + return -ENODEV; > + > + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); > + if (!res) > + return PTR_ERR(res); > + > + offset_addr = ioremap(res->start + config->offset_addr, sizeof(u32)); > + if (IS_ERR(offset_addr)) > + return PTR_ERR(offset_addr); > + > + stats_base = res->start | readl_relaxed(offset_addr); > + stats_size = resource_size(res); > + iounmap(offset_addr); > + > + reg = devm_ioremap(&pdev->dev, stats_base, stats_size); > + if (!reg) > + return -ENOMEM; > + > + prv_data = devm_kzalloc(&pdev->dev, config->num_records * > + sizeof(struct stats_prv_data), GFP_KERNEL); > + if (!prv_data) > + return -ENOMEM; > + > + for (i = 0; i < config->num_records; i++) > + prv_data[i].config = config; > + > + root = create_debugfs_entries(reg, prv_data); > + platform_set_drvdata(pdev, root); > + > + return 0; > +} > + > +static int soc_sleep_stats_remove(struct platform_device *pdev) > +{ > + struct dentry *root = platform_get_drvdata(pdev); > + > + debugfs_remove_recursive(root); > + > + return 0; > +} > + > +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, > +}; > + > +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 }, > + { } > +}; > + > +static struct platform_driver soc_sleep_stats_driver = { > + .probe = soc_sleep_stats_probe, > + .remove = soc_sleep_stats_remove, > + .driver = { > + .name = "soc_sleep_stats", > + .of_match_table = soc_sleep_stats_table, > + }, > +}; > +module_platform_driver(soc_sleep_stats_driver); > + > +MODULE_DESCRIPTION("Qualcomm Technologies, Inc. (QTI) SoC Sleep Stats driver"); > +MODULE_LICENSE("GPL v2");
diff --git a/drivers/soc/qcom/Kconfig b/drivers/soc/qcom/Kconfig index d0a73e7..08bc0df3 100644 --- a/drivers/soc/qcom/Kconfig +++ b/drivers/soc/qcom/Kconfig @@ -185,6 +185,15 @@ 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 && DEBUG_FS || COMPILE_TEST + 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 diff --git a/drivers/soc/qcom/Makefile b/drivers/soc/qcom/Makefile index 9fb35c8..e6bd052 100644 --- a/drivers/soc/qcom/Makefile +++ b/drivers/soc/qcom/Makefile @@ -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 diff --git a/drivers/soc/qcom/soc_sleep_stats.c b/drivers/soc/qcom/soc_sleep_stats.c new file mode 100644 index 0000000..0db7c3d --- /dev/null +++ b/drivers/soc/qcom/soc_sleep_stats.c @@ -0,0 +1,244 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Copyright (c) 2011-2020, The Linux Foundation. All rights reserved. + */ + +#include <linux/debugfs.h> +#include <linux/device.h> +#include <linux/io.h> +#include <linux/module.h> +#include <linux/of.h> +#include <linux/platform_device.h> +#include <linux/seq_file.h> + +#include <linux/soc/qcom/smem.h> +#include <clocksource/arm_arch_timer.h> + +#define STAT_TYPE_ADDR 0x0 +#define COUNT_ADDR 0x4 +#define LAST_ENTERED_AT_ADDR 0x8 +#define LAST_EXITED_AT_ADDR 0x10 +#define ACCUMULATED_ADDR 0x18 +#define CLIENT_VOTES_ADDR 0x1c + +struct subsystem_data { + const char *name; + u32 smem_item; + u32 pid; +}; + +static struct subsystem_data subsystems[] = { + { "modem", 605, 1 }, + { "adsp", 606, 2 }, + { "cdsp", 607, 5 }, + { "slpi", 608, 3 }, + { "gpu", 609, 0 }, + { "display", 610, 0 }, +}; + +struct stats_config { + unsigned int offset_addr; + unsigned int num_records; + bool appended_stats_avail; +}; + +struct stats_prv_data { + const struct stats_config *config; + void __iomem *reg; +}; + +struct sleep_stats { + u32 stat_type; + u32 count; + u64 last_entered_at; + u64 last_exited_at; + u64 accumulated; +}; + +struct appended_stats { + u32 client_votes; + u32 reserved[3]; +}; + +static void print_sleep_stats(struct seq_file *s, struct sleep_stats *stat) +{ + u64 accumulated = stat->accumulated; + /* + * If a subsystem is in sleep when reading the sleep stats adjust + * the accumulated sleep duration to show actual sleep time. + */ + if (stat->last_entered_at > stat->last_exited_at) + accumulated += arch_timer_read_counter() + - stat->last_entered_at; + + seq_printf(s, "Count = %u\n", stat->count); + seq_printf(s, "Last Entered At = %llu\n", stat->last_entered_at); + seq_printf(s, "Last Exited At = %llu\n", stat->last_exited_at); + seq_printf(s, "Accumulated Duration = %llu\n", accumulated); +} + +static int subsystem_sleep_stats_show(struct seq_file *s, void *d) +{ + struct subsystem_data *subsystem = s->private; + struct sleep_stats *stat; + + stat = qcom_smem_get(subsystem->pid, subsystem->smem_item, NULL); + if (IS_ERR(stat)) + return PTR_ERR(stat); + + print_sleep_stats(s, stat); + + return 0; +} + +static int soc_sleep_stats_show(struct seq_file *s, void *d) +{ + struct stats_prv_data *prv_data = s->private; + void __iomem *reg = prv_data->reg; + struct sleep_stats stat; + + stat.count = readl(reg + COUNT_ADDR); + stat.last_entered_at = readq(reg + LAST_ENTERED_AT_ADDR); + stat.last_exited_at = readq(reg + LAST_EXITED_AT_ADDR); + stat.accumulated = readq(reg + ACCUMULATED_ADDR); + + print_sleep_stats(s, &stat); + + if (prv_data->config->appended_stats_avail) { + struct appended_stats app_stat; + + app_stat.client_votes = readl(reg + CLIENT_VOTES_ADDR); + seq_printf(s, "Client_votes = %#x\n", app_stat.client_votes); + } + + return 0; +} + +DEFINE_SHOW_ATTRIBUTE(soc_sleep_stats); +DEFINE_SHOW_ATTRIBUTE(subsystem_sleep_stats); + +static struct dentry *create_debugfs_entries(void __iomem *reg, + struct stats_prv_data *prv_data) +{ + struct dentry *root; + struct sleep_stats *stat; + char stat_type[sizeof(u32) + 1] = {0}; + u32 offset, type; + int i; + + root = debugfs_create_dir("qcom_sleep_stats", NULL); + + for (i = 0; i < prv_data[0].config->num_records; i++) { + offset = STAT_TYPE_ADDR + (i * sizeof(struct sleep_stats)); + + if (prv_data[0].config->appended_stats_avail) + offset += i * sizeof(struct appended_stats); + + prv_data[i].reg = reg + offset; + + type = readl(prv_data[i].reg); + memcpy(stat_type, &type, sizeof(u32)); + debugfs_create_file(stat_type, 0444, root, + &prv_data[i], + &soc_sleep_stats_fops); + } + + for (i = 0; i < ARRAY_SIZE(subsystems); i++) { + stat = qcom_smem_get(subsystems[i].pid, subsystems[i].smem_item, + NULL); + if (IS_ERR(stat)) + continue; + + debugfs_create_file(subsystems[i].name, 0444, root, + &subsystems[i], + &subsystem_sleep_stats_fops); + } + + return root; +} + +static int soc_sleep_stats_probe(struct platform_device *pdev) +{ + struct resource *res; + void __iomem *reg; + void __iomem *offset_addr; + phys_addr_t stats_base; + resource_size_t stats_size; + struct dentry *root; + const struct stats_config *config; + struct stats_prv_data *prv_data; + int i; + + config = device_get_match_data(&pdev->dev); + if (!config) + return -ENODEV; + + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + if (!res) + return PTR_ERR(res); + + offset_addr = ioremap(res->start + config->offset_addr, sizeof(u32)); + if (IS_ERR(offset_addr)) + return PTR_ERR(offset_addr); + + stats_base = res->start | readl_relaxed(offset_addr); + stats_size = resource_size(res); + iounmap(offset_addr); + + reg = devm_ioremap(&pdev->dev, stats_base, stats_size); + if (!reg) + return -ENOMEM; + + prv_data = devm_kzalloc(&pdev->dev, config->num_records * + sizeof(struct stats_prv_data), GFP_KERNEL); + if (!prv_data) + return -ENOMEM; + + for (i = 0; i < config->num_records; i++) + prv_data[i].config = config; + + root = create_debugfs_entries(reg, prv_data); + platform_set_drvdata(pdev, root); + + return 0; +} + +static int soc_sleep_stats_remove(struct platform_device *pdev) +{ + struct dentry *root = platform_get_drvdata(pdev); + + debugfs_remove_recursive(root); + + return 0; +} + +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, +}; + +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 }, + { } +}; + +static struct platform_driver soc_sleep_stats_driver = { + .probe = soc_sleep_stats_probe, + .remove = soc_sleep_stats_remove, + .driver = { + .name = "soc_sleep_stats", + .of_match_table = soc_sleep_stats_table, + }, +}; +module_platform_driver(soc_sleep_stats_driver); + +MODULE_DESCRIPTION("Qualcomm Technologies, Inc. (QTI) SoC Sleep Stats driver"); +MODULE_LICENSE("GPL v2");