Provide USF for the portable equipment.
Subsystems (3)
PATCH v2 1/1 Provide USF for the portable equipment.
From: Dongdong Yang <[email protected]>
This patch provides USF(User Sensitive Feedback factor) auxiliary cpufreq governor to support high level layer sysfs inodes setting for utils adjustment purpose from the identified scenario on portable equipment. Because the power consumption and UI response are more cared for by portable equipment users. And the "screen off" status stands for no request from the user, however, the kernel is still expected to notify the user in time on modem, network or powerkey events occur. USF provides "sched_usf_non_ux_r" sysfs inode to cut down the utils from user space tasks according to high level scenario. In addition, it usually hints more cpufreq demand that the preemptive counts of the tasks on the cpu burst and over the user expecting completed time such as the ratio sysctl_sched_latency to sysctl_sched_min_granularity on "screen on" status, which more likely with more UI. The sysfs inodes "sched_usf_up_l0_r" and "sched_usf_down_r" have been provided to adjust the utils according to high level identified scenario to alloc the cpufreq in time.
Changes in v2
- Add adjust_task_pred_set switch.
- Move adjust_task_pred_demand declaration into sched.h
- Update comments.
Dongdong Yang (1):
sched: Provide USF for the portable equipment.
drivers/staging/Kconfig | 2 + drivers/staging/Makefile | 1 + drivers/staging/fbsched/Kconfig | 10 ++ drivers/staging/fbsched/Makefile | 2 + drivers/staging/fbsched/usf.c | 346 +++++++++++++++++++++++++++++++++++++++ kernel/sched/cpufreq_schedutil.c | 3 + kernel/sched/sched.h | 10 ++ 7 files changed, 374 insertions(+) create mode 100644 drivers/staging/fbsched/Kconfig create mode 100644 drivers/staging/fbsched/Makefile create mode 100644 drivers/staging/fbsched/usf.c
Signature · 1 line
2.7.4
PATCH v2 1/1 sched: Provide USF for the portable equipment.
From: Dongdong Yang <[email protected]>
The power consumption and UI response are more cared for by the portable equipment users. USF(User Sensitive Feedback factor) auxiliary cpufreq governor is providing more utils adjustment settings to the high level by scenario identification.
From the view of portable equipment, screen off status usually stands for no request from the user, however, the kernel is still expected to notify the user in time on modem, network or powerkey events occur. In some scenarios, such as listening to music, low power processors, such as DSP, take more actions and CPU load requirements cut down. It would bring more power consumption benefit if high level have interfaces to adjust utils according to the current scenario and load.
In addition, the portable equipment user usually heavily interact with devices by touch, and other peripherals. The boost preemptive counts are marking the load requirement urgent, vice versa. If such feedback factor could be set to high level according to the scenario, it would contribute to the power consumption and UI response.
If no USF sysfs inode is set, and no screen on or off event, adjust_task_pred_demand shall not be invoked. Once sched_usf_up_l0_r sched_usf_down_r or sched_usf_non_ux_r be set, adjust_task_pred_demand shall be called back to update settings according to high level scenario identification.
We can get about 17% mean power consumption save at listening to music with speaker on "screen off" scenario, as below statistical data from 7766 XiaoMi devices for two weeks with sched_usf_non_ux_r be set:
day1 day2 day3 day4 count 7766.000000 7766.000000 7766.000000 7766.000000 mean 88.035525 85.500282 83.829305 86.054997 std 111.049980 108.258834 107.562583 108.558240 min 0.099000 0.037000 0.067000 0.045000 25% 34.765500 34.021750 34.101500 34.423000 50% 54.950000 55.286500 54.189500 54.248500 75% 95.954000 93.942000 91.738000 94.0592500 80% 114.675000 107.430000 106.378000 108.673000 85% 137.851000 129.511000 127.156500 131.750750 90% 179.669000 170.208500 164.027000 172.348000 95% 272.395000 257.845500 247.750500 263.275750 98% 399.034500 412.170400 391.484000 402.835600
day5 day6 day7 day8 count 7766.000000 7766.00000 7766.000000 7766.000000 mean 82.532677 79.21923 77.611380 81.075081 std 104.870079 101.34819 103.140037 97.506221 min 0.051000 0.02900 0.007000 0.068000 25% 32.873000 33.44400 31.965500 33.863500 50% 52.180500 51.56550 50.806500 53.080000 75% 90.905750 86.82625 83.859250 89.973000 80% 105.455000 99.64700 97.271000 104.225000 85% 128.300000 118.47825 116.570250 126.648250 90% 166.647500 149.18000 150.649500 161.087000 95% 247.208500 224.36050 226.380000 245.291250 98% 393.002000 347.92060 369.791800 378.778600
day9 day10 day11 day12 count 7766.000000 7766.000000 7766.000000 7766.000000 mean 79.989170 83.859417 78.032930 77.060542 std 104.226122 108.893043 102.561715 99.844276 min 0.118000 0.017000 0.028000 0.039000 25% 32.056250 33.454500 31.176250 30.897750 50% 51.506000 54.056000 48.969500 49.069000 75% 88.513500 92.953500 83.506750 84.096000 80% 102.876000 107.845000 97.717000 98.073000 85% 124.363000 128.288000 118.366500 116.869250 90% 160.557000 167.084000 154.342500 148.187500 95% 231.149000 242.925750 236.759000 228.131250 98% 367.206600 388.619100 385.269100 376.541600
day13 day14 count 7766.000000 7766.000000 mean 75.528036 73.702878 std 90.750594 86.796016 min 0.066000 0.054000 25% 31.170500 31.608500 50% 48.758500 49.215000 75% 84.522750 83.053000 80% 97.879000 94.875000 85% 116.680250 113.573750 90% 149.083500 144.089500 95% 226.177750 211.488750 98% 347.011100 331.317100
--- drivers/staging/Kconfig | 2 + drivers/staging/Makefile | 1 + drivers/staging/fbsched/Kconfig | 10 ++ drivers/staging/fbsched/Makefile | 2 + drivers/staging/fbsched/usf.c | 346 +++++++++++++++++++++++++++++++++++++++ kernel/sched/cpufreq_schedutil.c | 3 + kernel/sched/sched.h | 10 ++ 7 files changed, 374 insertions(+) create mode 100644 drivers/staging/fbsched/Kconfig create mode 100644 drivers/staging/fbsched/Makefile create mode 100644 drivers/staging/fbsched/usf.c
drivers/staging/Kconfig +2 -0
@@ -120,4 +120,6 @@ source "drivers/staging/qlge/Kconfig" source "drivers/staging/wfx/Kconfig" +source "drivers/staging/fbsched/Kconfig" + endif # STAGING
drivers/staging/Makefile +1 -0
@@ -50,3 +50,4 @@ obj-$(CONFIG_FIELDBUS_DEV) += fieldbus/ obj-$(CONFIG_KPC2000) += kpc2000/ obj-$(CONFIG_QLGE) += qlge/ obj-$(CONFIG_WFX) += wfx/ +obj-$(CONFIG_SCHED_USF) += fbsched/
drivers/staging/fbsched/Kconfig new file +10 -0
@@ -0,0 +1,10 @@ +config SCHED_USF + bool "User Sensitive Factors for Scheduler" + depends on CPU_FREQ_GOV_SCHEDUTIL && FB + help + Select this option to enable the adjustment on the cpufreq with + the user sensitive factors on schedule. It is special for portable + equipment which more power care and quick response requirement on + screen on. + + If unsure, say N.
drivers/staging/fbsched/Makefile new file +2 -0
@@ -0,0 +1,2 @@ +LINUXINCLUDE += -include $(srctree)/kernel/sched/sched.h +obj-$(CONFIG_SCHED_USF) += usf.o
drivers/staging/fbsched/usf.c new file +346 -0
346 lines not on this page · raw message
kernel/sched/cpufreq_schedutil.c +3 -0
@@ -289,12 +289,15 @@ unsigned long schedutil_cpu_util(int cpu, unsigned long util_cfs,return min (max, util); } +DEFINE_STATIC_KEY_FALSE (adjust_task_pred_set);static unsigned long sugov_get_util (struct sugov_cpu *sg_cpu) {struct rq *rq =cpu_rq (sg_cpu->cpu);unsigned long util =cpu_util_cfs (rq);unsigned long max =arch_scale_cpu_capacity (sg_cpu->cpu); +if (static_branch_unlikely (&adjust_task_pred_set)) +adjust_task_pred_demand (sg_cpu->cpu, &util, rq); sg_cpu->max = max; sg_cpu->bw_dl =cpu_bw_dl (rq);
kernel/sched/sched.h +10 -0
@@ -2540,5 +2540,15 @@ static inline bool is_per_cpu_kthread(struct task_struct *p) }#endif +DECLARE_STATIC_KEY_FALSE (adjust_task_pred_set); +#ifdef CONFIG_SCHED_USF +void adjust_task_pred_demand (int cpuid,unsigned long *util, +struct rq *rq); +#else +static inline void adjust_task_pred_demand (int cpuid, +unsigned long *util,struct rq *rq) +{ } +#endif +void swake_up_all_locked (struct swait_queue_head *q);void __prepare_to_swait (struct swait_queue_head *q,struct swait_queue *wait);
Signature · 1 line
2.7.4
Greg KH
maintainerRe: [PATCH v2] sched: Provide USF for the portable equipment.
Quoted 11 lines · Dongdong Yang
drivers/staging/Kconfig | 2 + drivers/staging/Makefile | 1 + drivers/staging/fbsched/Kconfig | 10 ++ drivers/staging/fbsched/Makefile | 2 + drivers/staging/fbsched/usf.c | 346 +++++++++++++++++++++++++++++++++++++++ kernel/sched/cpufreq_schedutil.c | 3 + kernel/sched/sched.h | 10 ++ 7 files changed, 374 insertions(+) create mode 100644 drivers/staging/fbsched/Kconfig create mode 100644 drivers/staging/fbsched/Makefile create mode 100644 drivers/staging/fbsched/usf.c
I asked for a TODO file, like all other staging drivers have, why did you not include one here?
Greg KH
maintainerRe: [PATCH v2] sched: Provide USF for the portable equipment.
In reply to Dongdong YangQuoted 5 lines · Dongdong Yang
drivers/staging/fbsched/usf.c new file +2
-0@@ -0,0 +1,346 @@ +// SPDX-License-Identifier: GPL-2.0 +/*
drivers/staging/fbsched/usf.c line 12
Quoted 10 lines
drivers/staging/fbsched/usf.c new file +10
-0+* Copyright (C) 2020 XiaoMi Inc. +* Author: Yang Dongdong <[email protected]> +* This program is free software; you can redistribute it and/or modify +* it under the terms of the GNU General Public License version 2 as +* published by the Free Software Foundation. +* +* This program is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. +* See http://www.gnu.org/licenses/gpl-2.0.html for more details.
Please remove the license "boilerplate" text as you have the SPDX line on top.
Quoted 21 lines
drivers/staging/fbsched/usf.c new file +21
-0+*/ + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/platform_device.h> +#include <linux/kthread.h> +#include <linux/cpu.h> +#include <linux/sysfs.h> +#include <linux/kthread.h> +#include <linux/kobject.h> +#include <linux/module.h> +#include <linux/kernel.h> +#include <linux/init.h> +#include <linux/kallsyms.h> +#include <linux/debugfs.h> +#include <linux/fb.h> +#include <linux/notifier.h> + +#define BOOST_MIN_V -100 +#define BOOST_MAX_V 100 +#define LEVEL_TOP 3
drivers/staging/fbsched/usf.c line 43
Quoted 10 lines
drivers/staging/fbsched/usf.c new file +10
-0+ +#define USF_TAG "[usf_sched]" + +DEFINE_PER_CPU(unsigned long[PID_MAX_LIMIT], task_hist_nivcsw); + +static struct { + bool is_sched_usf_enabled; + int enable_debug; + int is_screen_on; + struct kobject *kobj;
A raw kobject? For a driver? are you _SURE_???
drivers/staging/fbsched/usf.c line 44
Quoted 1 line
drivers/staging/fbsched/usf.c new file +1
-0+struct dentry *debugfs_entry;
Why do you need this?
drivers/staging/fbsched/usf.c line 312
Quoted 6 lines
drivers/staging/fbsched/usf.c new file +6
-0+ usf_vdev.enable_debug =0 ; + usf_vdev.debugfs_entry =debugfs_create_file ("usf_dbg" , +0660 ,NULL ,NULL , + &usf_dbg_fops); +if (!usf_vdev.debugfs_entry) +pr_err ("Failed to create usf_dbg!\n" );
How can that value be NULL?
There is no need to check debugfs functions for error values.
But in this case, why are you writing a single file to the root of debugfs? Why not put it in a directory instead? That would be much nicer, don't you think?
thanks,
greg k-h
Greg KH
maintainerRe: [PATCH v2] sched: Provide USF for the portable equipment.
In reply to Dongdong YangQuoted 80 lines · Dongdong Yang
drivers/staging/fbsched/usf.c new file +80
-0+static ssize_t store_sched_usf_up_l0_r (struct kobject *kobj, +struct kobj_attribute *attr, +const char *buf,size_t count) +{ +int val =0 ; +int ret; + + ret =kstrtoint (buf,0 , &val); +if (ret) { +pr_err (USF_TAG "set state fail ret=%d\n" , ret); +return ret; + } + +if ((val >=0 ) && (val <=BOOST_MAX_V )) { + usf_vdev.sysctl_sched_usf_up_l0 = val; + usf_vdev.usf_up_l0 =LEVEL_TOP - +DIV_ROUND_UP (val,BOOST_MAX_V /2 ); + }else { +pr_err (USF_TAG "%d should fall into [%d %d]" , + val,0 ,BOOST_MAX_V ); + } +if ((usf_vdev.sysctl_sched_usf_up_l0 ==0 ) && + (usf_vdev.sysctl_sched_usf_down ==0 )) +static_branch_disable (&adjust_task_pred_set); +else +static_branch_enable (&adjust_task_pred_set); + +return count; +} + +static ssize_t store_sched_usf_down_r (struct kobject *kobj, +struct kobj_attribute *attr, +const char *buf,size_t count) +{ +int val =0 ; +int ret; + + ret =kstrtoint (buf,0 , &val); +if (ret) { +pr_err (USF_TAG "set state fail ret=%d\n" , ret); +return ret; + } + +if ((val >=BOOST_MIN_V ) && (val <=0 )) { + usf_vdev.sysctl_sched_usf_down = val; + usf_vdev.usf_down =DIV_ROUND_UP (-val, -BOOST_MIN_V /2 ); + }else { +pr_err (USF_TAG "%d should fall into [%d %d]" , + val,BOOST_MIN_V ,0 ); + } +if ((usf_vdev.sysctl_sched_usf_up_l0 ==0 ) && + (usf_vdev.sysctl_sched_usf_down ==0 )) +static_branch_disable (&adjust_task_pred_set); +else +static_branch_enable (&adjust_task_pred_set); + +return count; +} + +static ssize_t store_sched_usf_non_ux_r (struct kobject *kobj, +struct kobj_attribute *attr, +const char *buf,size_t count) +{ +int val =0 ; +int ret; + + ret =kstrtoint (buf,0 , &val); +if (ret) { +pr_err (USF_TAG "set state fail ret=%d\n" , ret); +return ret; + } + +if ((val >=BOOST_MIN_V ) && (val <=0 )) { + usf_vdev.sysctl_sched_usf_non_ux = val; + usf_vdev.usf_non_ux =DIV_ROUND_UP (-val, -BOOST_MIN_V /2 ); + }else { +pr_err (USF_TAG "%d should fall into [%d %d]" , + val,BOOST_MIN_V ,0 ); + } +if (usf_vdev.sysctl_sched_usf_non_ux ==0 )
drivers/staging/fbsched/usf.c line 225
Quoted 10 lines
drivers/staging/fbsched/usf.c new file +10
-0+static_branch_disable (&adjust_task_pred_set); + else + static_branch_enable(&adjust_task_pred_set); + +return count; +} + +#define usf_attr_rw (_name) \ +static struct kobj_attribute _name = \ +__ATTR(_name, 0664, show_##_name, store_##_name)
You have a lot of sysfs files here, but no documentation for them at all, why not?
Quoted 57 lines
drivers/staging/fbsched/usf.c new file +57
-0+ +#define usf_show_node (_name, _value) \ +static ssize_t show_##_name \ +(struct kobject *kobj, struct kobj_attribute *attr, char *buf) \ +{ \ + unsigned int len = 0; \ + unsigned int max_len = 8; \ + len += \ + snprintf(buf + len, max_len - len, "%d", \ + usf_vdev.sysctl_##_value); \ + return len; \ +} + +usf_show_node (sched_usf_up_l0_r, sched_usf_up_l0); +usf_show_node (sched_usf_down_r, sched_usf_down); +usf_show_node (sched_usf_non_ux_r, sched_usf_non_ux); + +usf_attr_rw (sched_usf_up_l0_r); +usf_attr_rw (sched_usf_down_r); +usf_attr_rw (sched_usf_non_ux_r); + +static struct attribute *sched_attrs[] = { + &sched_usf_up_l0_r.attr, + &sched_usf_down_r.attr, + &sched_usf_non_ux_r.attr, +NULL , +}; + +static struct attribute_group sched_attr_group = { + .attrs = sched_attrs, +}; + +static int usf_dbg_get (void *data,u64 *val) +{ + *val = (u64 )usf_vdev.enable_debug; + +return 0 ; +} + +static int usf_dbg_set (void *data,u64 val) +{ + usf_vdev.enable_debug = !!val; +return 0 ; +} + +DEFINE_SIMPLE_ATTRIBUTE (usf_dbg_fops, usf_dbg_get, + usf_dbg_set,"%llu\n" ); + +static int __init intera_monitor_init (void ) +{ +int res =-1 ; +struct attribute_group *attr_group; + + res =fb_register_client (&usf_lcd_nb); +if (res <0 ) { +pr_err ("Failed to register usf_lcd_nb!\n" ); +return res;
drivers/staging/fbsched/usf.c line 292
Quoted 10 lines
drivers/staging/fbsched/usf.c new file +10
-0+ } + +/* +* create a sched_usf in cpu_subsys: +* /sys/devices/system/cpu/sched_usf/... +*/ + attr_group = &sched_attr_group; + usf_vdev.kobj =kobject_create_and_add ("sched_usf" , + &cpu_subsys.dev_root->kobj); +
Huh? Why put this there?
And by making a "raw" kobject, you just kept userspace from seeing this information entirely :(
Use a 'struct device' as you are a driver, do not use a raw kobject.
thanks,
greg k-h
Dietmar Eggemann
reviewerRe: [PATCH v2] sched: Provide USF for the portable equipment.
In reply to Dongdong YangQuoted 1 line · Dongdong Yang
From: Dongdong Yang <[email protected]>
[...]
drivers/staging/fbsched/usf.c line 83
Quoted 5 lines
drivers/staging/fbsched/usf.c new file +5
-0+if (unlikely (usf_vdev.enable_debug)) +trace_printk + ("%s: cpu_id=%d non_ux=%d usf_up=%d usf_down=%d util=%lu\n" , +USF_TAG , cpuid, usf_vdev.usf_non_ux, + usf_vdev.usf_up_l0, usf_vdev.usf_down, *util);
trace_printk in code ?
Quoted 39 lines
drivers/staging/fbsched/usf.c new file +39
-0+static int usf_lcd_notifier (struct notifier_block *nb, +unsigned long val,void *data) +{ +struct fb_event *evdata = data; +unsigned int blank; + +if (!evdata) +return 0 ; + +if (val !=FB_EVENT_BLANK ) +return 0 ; + +if (evdata->data && val ==FB_EVENT_BLANK ) { + blank = *(int *)(evdata->data); + +switch (blank) { +case FB_BLANK_POWERDOWN : + usf_vdev.is_screen_on =0 ; +if (usf_vdev.sysctl_sched_usf_non_ux !=0 ) +static_branch_enable (&adjust_task_pred_set); +else +static_branch_disable (&adjust_task_pred_set); + +break ; + +case FB_BLANK_UNBLANK : + usf_vdev.is_screen_on =1 ; +if (usf_vdev.sysctl_sched_usf_up_l0 !=0 || + usf_vdev.sysctl_sched_usf_down !=0 ) +static_branch_enable (&adjust_task_pred_set); +else +static_branch_disable (&adjust_task_pred_set); +break ; +default : +break ; + } + + usf_vdev.is_sched_usf_enabled =1 ; +if (usf_vdev.enable_debug)
drivers/staging/fbsched/usf.c line 134
Quoted 10 lines
drivers/staging/fbsched/usf.c new file +10
-0+trace_printk ("%s : usf_vdev.is_screen_on:%d\n" , + __func__, usf_vdev.is_screen_on); + } +return NOTIFY_OK ; +} + +static struct notifier_block usf_lcd_nb = { + .notifier_call = usf_lcd_notifier, + .priority =INT_MAX , +};
Looks like those notifications should enable/disable the schedutil extension adjust_task_pred_demand(). Who's calling them?
The 3 sched_usf_FOO sys files somehow have an influence here too. How should this work?
I see a fb_register_client() in intera_monitor_init further below.
[...]
drivers/staging/fbsched/usf.c line 245
Quoted 3 lines
drivers/staging/fbsched/usf.c new file +3
-0+usf_attr_rw (sched_usf_up_l0_r); +usf_attr_rw (sched_usf_down_r); +usf_attr_rw (sched_usf_non_ux_r);
What can I do with these three files? What do they stand for?
root@h620:/sys/devices/system/cpu/sched_usf# ls sched_usf_down_r sched_usf_non_ux_r sched_usf_up_l0_r
[...]
Quoted 9 lines
drivers/staging/fbsched/usf.c new file +9
-0+static int __init intera_monitor_init (void ) +{ +int res =-1 ; +struct attribute_group *attr_group; + + res =fb_register_client (&usf_lcd_nb); +if (res <0 ) { +pr_err ("Failed to register usf_lcd_nb!\n" ); +return res;
[...]
kernel test robot
build botRe: [PATCH v2] sched: Provide USF for the portable equipment.
In reply to Dongdong YangThank you for the patch! Yet something to improve:
Hi Dongdong,
Thank you for the patch! Yet something to improve:
[auto build test ERROR on staging/staging-testing]
[also build test ERROR on pm/linux-next tip/sched/core v5.8-rc7 next-20200731] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]
url: https://github.com/0day-ci/linux/commits/Dongdong-Yang/sched-Provide-USF-for-the-portable-equipment/20200731-211347 base: https://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging.git 11c416e3f0ea437661514fa68967a5dba3cdf886 config: ia64-allyesconfig (attached as .config) compiler: ia64-linux-gcc (GCC) 9.3.0
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-9.3.0 make.cross ARCH=ia64
If you fix the issue, kindly add following tag as appropriate
All errors (new ones prefixed by >>):
ia64-linux-ld: section .data..percpu VMA wraps around address space
>> ia64-linux-ld: section .data LMA [0000000014380000,00000000150a8e7f] overlaps section .data..percpu LMA [0000000014340000,000000001634ab47]
arch/ia64/kernel/setup.o: in function `cpu_init':
setup.c:(.text+0x1110): relocation truncated to fit: IMM22 against symbol `cpu_sibling_map' defined in .data..percpu..shared_aligned section in arch/ia64/kernel/smpboot.o
---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org
You can't modify non-staging core code just for staging code.