[V2,3/3] tracing/timerlat: Add user-space interface
Commit Message
Going a step further, we propose a way to use any user-space
workload as the task waiting for the timerlat timer. This is done
via a per-CPU file named osnoise/cpu$id/timerlat_fd file.
The tracef_fd allows a task to open at a time. When a task reads
the file, the timerlat timer is armed for future osnoise/timerlat_period_us
time. When the timer fires, it prints the IRQ latency and
wakes up the user-space thread waiting in the timerlat_fd.
The thread then starts to run, executes the timerlat measurement, prints
the thread scheduling latency and returns to user space.
When the thread rereads the timerlat_fd, the tracer will print the
user-ret(urn) latency, which is an additional metric.
This additional metric is also traced by the tracer and can be used, for
example of measuring the context switch overhead from kernel-to-user and
user-to-kernel, or the response time for an arbitrary execution in
user-space.
The tracer supports one thread per CPU, the thread must be pinned to
the CPU, and it cannot migrate while holding the timerlat_fd. The reason
is that the tracer is per CPU (nothing prohibits the tracer from
allowing migrations in the future). The tracer monitors the migration
of the thread and disables the tracer if detected.
The timerlat_fd is only available for opening/reading when timerlat
tracer is enabled, and NO_OSNOISE_WORKLOAD is set.
The simplest way to activate this feature from user-space is:
-------------------------------- %< -----------------------------------
int main(void)
{
char buffer[1024];
int timerlat_fd;
int retval;
long cpu = 0; /* place in CPU 0 */
cpu_set_t set;
CPU_ZERO(&set);
CPU_SET(cpu, &set);
if (sched_setaffinity(gettid(), sizeof(set), &set) == -1)
return 1;
snprintf(buffer, sizeof(buffer),
"/sys/kernel/tracing/osnoise/per_cpu/cpu%ld/timerlat_fd",
cpu);
timerlat_fd = open(buffer, O_RDONLY);
if (timerlat_fd < 0) {
printf("error opening %s: %s\n", buffer, strerror(errno));
exit(1);
}
for (;;) {
retval = read(timerlat_fd, buffer, 1024);
if (retval < 0)
break;
}
close(timerlat_fd);
exit(0);
}
-------------------------------- >% -----------------------------------
When disabling timerlat, if there is a workload holding the timerlat_fd,
the SIGKILL will be sent to the thread.
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Daniel Bristot de Oliveira <bristot@kernel.org>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
---
Documentation/trace/timerlat-tracer.rst | 78 ++++++
kernel/trace/trace_osnoise.c | 312 +++++++++++++++++++++++-
kernel/trace/trace_output.c | 4 +-
3 files changed, 391 insertions(+), 3 deletions(-)
Comments
Hi Daniel,
kernel test robot noticed the following build warnings:
[auto build test WARNING on linus/master]
[also build test WARNING on v6.4-rc3 next-20230523]
[cannot apply to rostedt-trace/for-next rostedt-trace/for-next-urgent]
[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#_base_tree_information]
url: https://github.com/intel-lab-lkp/linux/commits/Daniel-Bristot-de-Oliveira/tracing-osnoise-Switch-from-PF_NO_SETAFFINITY-to-migrate_disable/20230524-012512
base: linus/master
patch link: https://lore.kernel.org/r/a7b2c215c763e95a56fa1258743332b570c81c9d.1684860626.git.bristot%40kernel.org
patch subject: [PATCH V2 3/3] tracing/timerlat: Add user-space interface
config: loongarch-allyesconfig
compiler: loongarch64-linux-gcc (GCC) 12.1.0
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
# https://github.com/intel-lab-lkp/linux/commit/89216b54eaf490480bc1929f5780f95a688a91bb
git remote add linux-review https://github.com/intel-lab-lkp/linux
git fetch --no-tags linux-review Daniel-Bristot-de-Oliveira/tracing-osnoise-Switch-from-PF_NO_SETAFFINITY-to-migrate_disable/20230524-012512
git checkout 89216b54eaf490480bc1929f5780f95a688a91bb
# save the config file
mkdir build_dir && cp config build_dir/.config
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.1.0 make.cross W=1 O=build_dir ARCH=loongarch olddefconfig
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-12.1.0 make.cross W=1 O=build_dir ARCH=loongarch SHELL=/bin/bash kernel/
If you fix the issue, kindly add following tag where applicable
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202305240459.z43SP4XU-lkp@intel.com/
All warnings (new ones prefixed by >>):
>> kernel/trace/trace_osnoise.c:2645:5: warning: no previous prototype for 'osnoise_create_cpu_timerlat_fd' [-Wmissing-prototypes]
2645 | int osnoise_create_cpu_timerlat_fd(struct dentry *top_dir)
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
vim +/osnoise_create_cpu_timerlat_fd +2645 kernel/trace/trace_osnoise.c
2644
> 2645 int osnoise_create_cpu_timerlat_fd(struct dentry *top_dir)
2646 {
2647 struct dentry *timerlat_fd;
2648 struct dentry *per_cpu;
2649 struct dentry *cpu_dir;
2650 char cpu_str[30]; /* see trace.c: tracing_init_tracefs_percpu() */
2651 long cpu;
2652
2653 /*
2654 * Why not using tracing instance per_cpu/ dir?
2655 *
2656 * Because osnoise/timerlat have a single workload, having
2657 * multiple files like these are wast of memory.
2658 */
2659 per_cpu = tracefs_create_dir("per_cpu", top_dir);
2660 if (!per_cpu)
2661 return -ENOMEM;
2662
2663 for_each_possible_cpu(cpu) {
2664 snprintf(cpu_str, 30, "cpu%ld", cpu);
2665 cpu_dir = tracefs_create_dir(cpu_str, per_cpu);
2666 if (!cpu_dir)
2667 goto out_clean;
2668
2669 timerlat_fd = trace_create_file("timerlat_fd", TRACE_MODE_READ,
2670 cpu_dir, NULL, &timerlat_fd_fops);
2671 if (!timerlat_fd)
2672 goto out_clean;
2673
2674 /* Record the CPU */
2675 d_inode(timerlat_fd)->i_cdev = (void *)(cpu);
2676 }
2677
2678 return 0;
2679
2680 out_clean:
2681 tracefs_remove(per_cpu);
2682 return -ENOMEM;
2683 }
2684
Hi Daniel,
kernel test robot noticed the following build warnings:
[auto build test WARNING on linus/master]
[also build test WARNING on v6.4-rc3 next-20230523]
[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#_base_tree_information]
url: https://github.com/intel-lab-lkp/linux/commits/Daniel-Bristot-de-Oliveira/tracing-osnoise-Switch-from-PF_NO_SETAFFINITY-to-migrate_disable/20230524-012512
base: linus/master
patch link: https://lore.kernel.org/r/a7b2c215c763e95a56fa1258743332b570c81c9d.1684860626.git.bristot%40kernel.org
patch subject: [PATCH V2 3/3] tracing/timerlat: Add user-space interface
config: x86_64-randconfig-s022
compiler: gcc-11 (Debian 11.3.0-12) 11.3.0
reproduce:
# apt-get install sparse
# sparse version: v0.6.4-39-gce1a6720-dirty
# https://github.com/intel-lab-lkp/linux/commit/89216b54eaf490480bc1929f5780f95a688a91bb
git remote add linux-review https://github.com/intel-lab-lkp/linux
git fetch --no-tags linux-review Daniel-Bristot-de-Oliveira/tracing-osnoise-Switch-from-PF_NO_SETAFFINITY-to-migrate_disable/20230524-012512
git checkout 89216b54eaf490480bc1929f5780f95a688a91bb
# save the config file
mkdir build_dir && cp config build_dir/.config
make W=1 C=1 CF='-fdiagnostic-prefix -D__CHECK_ENDIAN__' O=build_dir ARCH=x86_64 olddefconfig
make W=1 C=1 CF='-fdiagnostic-prefix -D__CHECK_ENDIAN__' O=build_dir ARCH=x86_64 SHELL=/bin/bash kernel/trace/
If you fix the issue, kindly add following tag where applicable
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202305241531.5ItU2LLm-lkp@intel.com/
sparse warnings: (new ones prefixed by >>)
>> kernel/trace/trace_osnoise.c:2645:5: sparse: sparse: symbol 'osnoise_create_cpu_timerlat_fd' was not declared. Should it be static?
Hi Daniel,
kernel test robot noticed the following build warnings:
[auto build test WARNING on linus/master]
[also build test WARNING on v6.4-rc3 next-20230524]
[cannot apply to rostedt-trace/for-next rostedt-trace/for-next-urgent]
[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#_base_tree_information]
url: https://github.com/intel-lab-lkp/linux/commits/Daniel-Bristot-de-Oliveira/tracing-osnoise-Switch-from-PF_NO_SETAFFINITY-to-migrate_disable/20230524-012512
base: linus/master
patch link: https://lore.kernel.org/r/a7b2c215c763e95a56fa1258743332b570c81c9d.1684860626.git.bristot%40kernel.org
patch subject: [PATCH V2 3/3] tracing/timerlat: Add user-space interface
config: x86_64-randconfig-x064-20230522
compiler: clang version 14.0.6 (https://github.com/llvm/llvm-project f28c006a5895fc0e329fe15fead81e37457cb1d1)
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
# https://github.com/intel-lab-lkp/linux/commit/89216b54eaf490480bc1929f5780f95a688a91bb
git remote add linux-review https://github.com/intel-lab-lkp/linux
git fetch --no-tags linux-review Daniel-Bristot-de-Oliveira/tracing-osnoise-Switch-from-PF_NO_SETAFFINITY-to-migrate_disable/20230524-012512
git checkout 89216b54eaf490480bc1929f5780f95a688a91bb
# save the config file
mkdir build_dir && cp config build_dir/.config
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross W=1 O=build_dir ARCH=x86_64 olddefconfig
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross W=1 O=build_dir ARCH=x86_64 SHELL=/bin/bash kernel/trace/
If you fix the issue, kindly add following tag where applicable
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202305241530.Pn2xj2vs-lkp@intel.com/
All warnings (new ones prefixed by >>):
kernel/trace/trace_osnoise.c:2364:9: error: implicit declaration of function 'this_cpu_tmr_var' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
tlat = this_cpu_tmr_var();
^
kernel/trace/trace_osnoise.c:2364:9: note: did you mean 'this_cpu_osn_var'?
kernel/trace/trace_osnoise.c:226:41: note: 'this_cpu_osn_var' declared here
static inline struct osnoise_variables *this_cpu_osn_var(void)
^
>> kernel/trace/trace_osnoise.c:2364:7: warning: incompatible integer to pointer conversion assigning to 'struct timerlat_variables *' from 'int' [-Wint-conversion]
tlat = this_cpu_tmr_var();
^ ~~~~~~~~~~~~~~~~~~
kernel/trace/trace_osnoise.c:2365:6: error: incomplete definition of type 'struct timerlat_variables'
tlat->count = 0;
~~~~^
kernel/trace/trace_osnoise.c:2309:9: note: forward declaration of 'struct timerlat_variables'
struct timerlat_variables *tlat;
^
kernel/trace/trace_osnoise.c:2387:25: error: variable has incomplete type 'struct timerlat_sample'
struct timerlat_sample s;
^
kernel/trace/trace_osnoise.c:2387:9: note: forward declaration of 'struct timerlat_sample'
struct timerlat_sample s;
^
kernel/trace/trace_osnoise.c:2393:9: error: implicit declaration of function 'this_cpu_tmr_var' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
tlat = this_cpu_tmr_var();
^
kernel/trace/trace_osnoise.c:2393:7: warning: incompatible integer to pointer conversion assigning to 'struct timerlat_variables *' from 'int' [-Wint-conversion]
tlat = this_cpu_tmr_var();
^ ~~~~~~~~~~~~~~~~~~
kernel/trace/trace_osnoise.c:2401:11: error: incomplete definition of type 'struct timerlat_variables'
if (tlat->uthread_migrate) {
~~~~^
kernel/trace/trace_osnoise.c:2386:9: note: forward declaration of 'struct timerlat_variables'
struct timerlat_variables *tlat;
^
kernel/trace/trace_osnoise.c:2406:16: error: use of undeclared identifier 'per_cpu_timerlat_var'
per_cpu_ptr(&per_cpu_timerlat_var, cpu)->uthread_migrate = 1;
^
kernel/trace/trace_osnoise.c:2406:16: error: use of undeclared identifier 'per_cpu_timerlat_var'
kernel/trace/trace_osnoise.c:2406:16: error: use of undeclared identifier 'per_cpu_timerlat_var'
kernel/trace/trace_osnoise.c:2406:44: error: member reference type 'void' is not a pointer
per_cpu_ptr(&per_cpu_timerlat_var, cpu)->uthread_migrate = 1;
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^
kernel/trace/trace_osnoise.c:2424:46: error: incomplete definition of type 'struct timerlat_variables'
now = ktime_to_ns(hrtimer_cb_get_time(&tlat->timer));
~~~~^
kernel/trace/trace_osnoise.c:2386:9: note: forward declaration of 'struct timerlat_variables'
struct timerlat_variables *tlat;
^
kernel/trace/trace_osnoise.c:2425:20: error: incomplete definition of type 'struct timerlat_variables'
diff = now - tlat->abs_period;
~~~~^
kernel/trace/trace_osnoise.c:2386:9: note: forward declaration of 'struct timerlat_variables'
struct timerlat_variables *tlat;
^
kernel/trace/trace_osnoise.c:2433:18: error: incomplete definition of type 'struct timerlat_variables'
s.seqnum = tlat->count;
~~~~^
kernel/trace/trace_osnoise.c:2386:9: note: forward declaration of 'struct timerlat_variables'
struct timerlat_variables *tlat;
^
kernel/trace/trace_osnoise.c:2437:3: error: implicit declaration of function 'trace_timerlat_sample' is invalid in C99 [-Werror,-Wimplicit-function-declaration]
trace_timerlat_sample(&s);
^
kernel/trace/trace_osnoise.c:2441:7: error: incomplete definition of type 'struct timerlat_variables'
tlat->tracing_thread = false;
~~~~^
kernel/trace/trace_osnoise.c:2386:9: note: forward declaration of 'struct timerlat_variables'
struct timerlat_variables *tlat;
^
kernel/trace/trace_osnoise.c:2446:7: error: incomplete definition of type 'struct timerlat_variables'
tlat->tracing_thread = false;
~~~~^
kernel/trace/trace_osnoise.c:2386:9: note: forward declaration of 'struct timerlat_variables'
struct timerlat_variables *tlat;
^
kernel/trace/trace_osnoise.c:2447:7: error: incomplete definition of type 'struct timerlat_variables'
tlat->kthread = current;
~~~~^
kernel/trace/trace_osnoise.c:2386:9: note: forward declaration of 'struct timerlat_variables'
struct timerlat_variables *tlat;
^
kernel/trace/trace_osnoise.c:2449:21: error: incomplete definition of type 'struct timerlat_variables'
hrtimer_init(&tlat->timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_PINNED_HARD);
~~~~^
kernel/trace/trace_osnoise.c:2386:9: note: forward declaration of 'struct timerlat_variables'
struct timerlat_variables *tlat;
^
kernel/trace/trace_osnoise.c:2450:7: error: incomplete definition of type 'struct timerlat_variables'
tlat->timer.function = timerlat_irq;
~~~~^
kernel/trace/trace_osnoise.c:2386:9: note: forward declaration of 'struct timerlat_variables'
struct timerlat_variables *tlat;
^
kernel/trace/trace_osnoise.c:2450:26: error: use of undeclared identifier 'timerlat_irq'; did you mean 'timerlat_main'?
tlat->timer.function = timerlat_irq;
^~~~~~~~~~~~
timerlat_main
kernel/trace/trace_osnoise.c:1856:12: note: 'timerlat_main' declared here
static int timerlat_main(void *data)
^
fatal error: too many errors emitted, stopping now [-ferror-limit=]
2 warnings and 20 errors generated.
vim +2364 kernel/trace/trace_osnoise.c
2305
2306 static int timerlat_fd_open(struct inode *inode, struct file *file)
2307 {
2308 struct osnoise_variables *osn_var;
2309 struct timerlat_variables *tlat;
2310 long cpu = (long) inode->i_cdev;
2311
2312 mutex_lock(&interface_lock);
2313
2314 /*
2315 * This file is accessible only if timerlat is enabled, and
2316 * NO_OSNOISE_WORKLOAD is set.
2317 */
2318 if (!timerlat_enabled() || test_bit(OSN_WORKLOAD, &osnoise_options)) {
2319 mutex_unlock(&interface_lock);
2320 return -EINVAL;
2321 }
2322
2323 migrate_disable();
2324
2325 osn_var = this_cpu_osn_var();
2326
2327 /*
2328 * The osn_var->pid holds the single access to this file.
2329 */
2330 if (osn_var->pid) {
2331 mutex_unlock(&interface_lock);
2332 migrate_enable();
2333 return -EBUSY;
2334 }
2335
2336 /*
2337 * timerlat tracer is a per-cpu tracer. Check if the user-space too
2338 * is pinned to a single CPU. The tracer laters monitor if the task
2339 * migrates and then disables tracer if it does. However, it is
2340 * worth doing this basic acceptance test to avoid obviusly wrong
2341 * setup.
2342 */
2343 if (current->nr_cpus_allowed > 1 || cpu != smp_processor_id()) {
2344 mutex_unlock(&interface_lock);
2345 migrate_enable();
2346 return -EPERM;
2347 }
2348
2349 /*
2350 * From now on, it is good to go.
2351 */
2352 file->private_data = inode->i_cdev;
2353
2354 get_task_struct(current);
2355
2356 osn_var->kthread = current;
2357 osn_var->pid = current->pid;
2358
2359 /*
2360 * Setup is done.
2361 */
2362 mutex_unlock(&interface_lock);
2363
> 2364 tlat = this_cpu_tmr_var();
2365 tlat->count = 0;
2366
2367 migrate_enable();
2368 return 0;
2369 };
2370
Hi Daniel,
kernel test robot noticed the following build errors:
[auto build test ERROR on linus/master]
[also build test ERROR on v6.4-rc3 next-20230524]
[cannot apply to rostedt-trace/for-next rostedt-trace/for-next-urgent]
[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#_base_tree_information]
url: https://github.com/intel-lab-lkp/linux/commits/Daniel-Bristot-de-Oliveira/tracing-osnoise-Switch-from-PF_NO_SETAFFINITY-to-migrate_disable/20230524-012512
base: linus/master
patch link: https://lore.kernel.org/r/a7b2c215c763e95a56fa1258743332b570c81c9d.1684860626.git.bristot%40kernel.org
patch subject: [PATCH V2 3/3] tracing/timerlat: Add user-space interface
config: i386-randconfig-i014-20230523
compiler: gcc-11 (Debian 11.3.0-12) 11.3.0
reproduce (this is a W=1 build):
# https://github.com/intel-lab-lkp/linux/commit/89216b54eaf490480bc1929f5780f95a688a91bb
git remote add linux-review https://github.com/intel-lab-lkp/linux
git fetch --no-tags linux-review Daniel-Bristot-de-Oliveira/tracing-osnoise-Switch-from-PF_NO_SETAFFINITY-to-migrate_disable/20230524-012512
git checkout 89216b54eaf490480bc1929f5780f95a688a91bb
# save the config file
mkdir build_dir && cp config build_dir/.config
make W=1 O=build_dir ARCH=i386 olddefconfig
make W=1 O=build_dir ARCH=i386 SHELL=/bin/bash kernel/trace/
If you fix the issue, kindly add following tag where applicable
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202305242020.VlsOXEMn-lkp@intel.com/
All error/warnings (new ones prefixed by >>):
kernel/trace/trace_osnoise.c: In function 'timerlat_fd_open':
>> kernel/trace/trace_osnoise.c:2364:16: error: implicit declaration of function 'this_cpu_tmr_var'; did you mean 'this_cpu_osn_var'? [-Werror=implicit-function-declaration]
2364 | tlat = this_cpu_tmr_var();
| ^~~~~~~~~~~~~~~~
| this_cpu_osn_var
>> kernel/trace/trace_osnoise.c:2364:14: warning: assignment to 'struct timerlat_variables *' from 'int' makes pointer from integer without a cast [-Wint-conversion]
2364 | tlat = this_cpu_tmr_var();
| ^
>> kernel/trace/trace_osnoise.c:2365:13: error: invalid use of undefined type 'struct timerlat_variables'
2365 | tlat->count = 0;
| ^~
kernel/trace/trace_osnoise.c: In function 'timerlat_fd_read':
>> kernel/trace/trace_osnoise.c:2387:32: error: storage size of 's' isn't known
2387 | struct timerlat_sample s;
| ^
kernel/trace/trace_osnoise.c:2393:14: warning: assignment to 'struct timerlat_variables *' from 'int' makes pointer from integer without a cast [-Wint-conversion]
2393 | tlat = this_cpu_tmr_var();
| ^
kernel/trace/trace_osnoise.c:2401:25: error: invalid use of undefined type 'struct timerlat_variables'
2401 | if (tlat->uthread_migrate) {
| ^~
In file included from include/asm-generic/percpu.h:7,
from arch/x86/include/asm/percpu.h:390,
from arch/x86/include/asm/current.h:10,
from include/linux/sched.h:12,
from include/linux/kthread.h:6,
from kernel/trace/trace_osnoise.c:19:
>> kernel/trace/trace_osnoise.c:2406:30: error: 'per_cpu_timerlat_var' undeclared (first use in this function)
2406 | per_cpu_ptr(&per_cpu_timerlat_var, cpu)->uthread_migrate = 1;
| ^~~~~~~~~~~~~~~~~~~~
include/linux/percpu-defs.h:219:54: note: in definition of macro '__verify_pcpu_ptr'
219 | const void __percpu *__vpp_verify = (typeof((ptr) + 0))NULL; \
| ^~~
kernel/trace/trace_osnoise.c:2406:17: note: in expansion of macro 'per_cpu_ptr'
2406 | per_cpu_ptr(&per_cpu_timerlat_var, cpu)->uthread_migrate = 1;
| ^~~~~~~~~~~
kernel/trace/trace_osnoise.c:2406:30: note: each undeclared identifier is reported only once for each function it appears in
2406 | per_cpu_ptr(&per_cpu_timerlat_var, cpu)->uthread_migrate = 1;
| ^~~~~~~~~~~~~~~~~~~~
include/linux/percpu-defs.h:219:54: note: in definition of macro '__verify_pcpu_ptr'
219 | const void __percpu *__vpp_verify = (typeof((ptr) + 0))NULL; \
| ^~~
kernel/trace/trace_osnoise.c:2406:17: note: in expansion of macro 'per_cpu_ptr'
2406 | per_cpu_ptr(&per_cpu_timerlat_var, cpu)->uthread_migrate = 1;
| ^~~~~~~~~~~
kernel/trace/trace_osnoise.c:2424:60: error: invalid use of undefined type 'struct timerlat_variables'
2424 | now = ktime_to_ns(hrtimer_cb_get_time(&tlat->timer));
| ^~
kernel/trace/trace_osnoise.c:2425:34: error: invalid use of undefined type 'struct timerlat_variables'
2425 | diff = now - tlat->abs_period;
| ^~
kernel/trace/trace_osnoise.c:2433:32: error: invalid use of undefined type 'struct timerlat_variables'
2433 | s.seqnum = tlat->count;
| ^~
>> kernel/trace/trace_osnoise.c:2437:17: error: implicit declaration of function 'trace_timerlat_sample' [-Werror=implicit-function-declaration]
2437 | trace_timerlat_sample(&s);
| ^~~~~~~~~~~~~~~~~~~~~
kernel/trace/trace_osnoise.c:2441:21: error: invalid use of undefined type 'struct timerlat_variables'
2441 | tlat->tracing_thread = false;
| ^~
kernel/trace/trace_osnoise.c:2446:21: error: invalid use of undefined type 'struct timerlat_variables'
2446 | tlat->tracing_thread = false;
| ^~
kernel/trace/trace_osnoise.c:2447:21: error: invalid use of undefined type 'struct timerlat_variables'
2447 | tlat->kthread = current;
| ^~
kernel/trace/trace_osnoise.c:2449:35: error: invalid use of undefined type 'struct timerlat_variables'
2449 | hrtimer_init(&tlat->timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_PINNED_HARD);
| ^~
kernel/trace/trace_osnoise.c:2450:21: error: invalid use of undefined type 'struct timerlat_variables'
2450 | tlat->timer.function = timerlat_irq;
| ^~
>> kernel/trace/trace_osnoise.c:2450:40: error: 'timerlat_irq' undeclared (first use in this function); did you mean 'timerlat_main'?
2450 | tlat->timer.function = timerlat_irq;
| ^~~~~~~~~~~~
| timerlat_main
kernel/trace/trace_osnoise.c:2453:21: error: invalid use of undefined type 'struct timerlat_variables'
2453 | tlat->abs_period = hrtimer_cb_get_time(&tlat->timer);
| ^~
kernel/trace/trace_osnoise.c:2453:61: error: invalid use of undefined type 'struct timerlat_variables'
2453 | tlat->abs_period = hrtimer_cb_get_time(&tlat->timer);
| ^~
>> kernel/trace/trace_osnoise.c:2459:9: error: implicit declaration of function 'wait_next_period' [-Werror=implicit-function-declaration]
2459 | wait_next_period(tlat);
| ^~~~~~~~~~~~~~~~
kernel/trace/trace_osnoise.c:2462:52: error: invalid use of undefined type 'struct timerlat_variables'
2462 | now = ktime_to_ns(hrtimer_cb_get_time(&tlat->timer));
| ^~
kernel/trace/trace_osnoise.c:2463:26: error: invalid use of undefined type 'struct timerlat_variables'
2463 | diff = now - tlat->abs_period;
| ^~
kernel/trace/trace_osnoise.c:2471:24: error: invalid use of undefined type 'struct timerlat_variables'
2471 | s.seqnum = tlat->count;
| ^~
>> kernel/trace/trace_osnoise.c:2479:25: error: implicit declaration of function 'timerlat_dump_stack'; did you mean 'trace_dump_stack'? [-Werror=implicit-function-declaration]
2479 | timerlat_dump_stack(time_to_us(diff));
| ^~~~~~~~~~~~~~~~~~~
| trace_dump_stack
>> kernel/trace/trace_osnoise.c:2387:32: warning: unused variable 's' [-Wunused-variable]
2387 | struct timerlat_sample s;
| ^
In file included from include/asm-generic/percpu.h:7,
from arch/x86/include/asm/percpu.h:390,
from arch/x86/include/asm/current.h:10,
from include/linux/sched.h:12,
from include/linux/kthread.h:6,
from kernel/trace/trace_osnoise.c:19:
kernel/trace/trace_osnoise.c: In function 'timerlat_fd_release':
kernel/trace/trace_osnoise.c:2500:33: error: 'per_cpu_timerlat_var' undeclared (first use in this function)
2500 | tlat_var = per_cpu_ptr(&per_cpu_timerlat_var, cpu);
| ^~~~~~~~~~~~~~~~~~~~
include/linux/percpu-defs.h:219:54: note: in definition of macro '__verify_pcpu_ptr'
219 | const void __percpu *__vpp_verify = (typeof((ptr) + 0))NULL; \
| ^~~
kernel/trace/trace_osnoise.c:2500:20: note: in expansion of macro 'per_cpu_ptr'
2500 | tlat_var = per_cpu_ptr(&per_cpu_timerlat_var, cpu);
| ^~~~~~~~~~~
kernel/trace/trace_osnoise.c:2502:33: error: invalid use of undefined type 'struct timerlat_variables'
2502 | hrtimer_cancel(&tlat_var->timer);
| ^~
In file included from include/linux/string.h:254,
from arch/x86/include/asm/page_32.h:18,
from arch/x86/include/asm/page.h:14,
from arch/x86/include/asm/thread_info.h:12,
from include/linux/thread_info.h:60,
from arch/x86/include/asm/preempt.h:9,
from include/linux/preempt.h:78,
from include/linux/rcupdate.h:27,
from include/linux/rculist.h:11,
from include/linux/pid.h:5,
from include/linux/sched.h:14,
from include/linux/kthread.h:6,
from kernel/trace/trace_osnoise.c:19:
>> kernel/trace/trace_osnoise.c:2503:35: error: invalid application of 'sizeof' to incomplete type 'struct timerlat_variables'
2503 | memset(tlat_var, 0, sizeof(*tlat_var));
| ^
include/linux/fortify-string.h:451:42: note: in definition of macro '__fortify_memset_chk'
451 | size_t __fortify_size = (size_t)(size); \
| ^~~~
kernel/trace/trace_osnoise.c:2503:9: note: in expansion of macro 'memset'
2503 | memset(tlat_var, 0, sizeof(*tlat_var));
| ^~~~~~
In file included from include/asm-generic/percpu.h:7,
from arch/x86/include/asm/percpu.h:390,
from arch/x86/include/asm/current.h:10,
from include/linux/sched.h:12,
from include/linux/kthread.h:6,
from kernel/trace/trace_osnoise.c:19:
kernel/trace/trace_osnoise.c: In function 'check_timerlat_user_migration':
kernel/trace/trace_osnoise.c:2529:38: error: 'per_cpu_timerlat_var' undeclared (first use in this function)
2529 | per_cpu_ptr(&per_cpu_timerlat_var, cpu)->uthread_migrate = 1;
| ^~~~~~~~~~~~~~~~~~~~
include/linux/percpu-defs.h:219:54: note: in definition of macro '__verify_pcpu_ptr'
219 | const void __percpu *__vpp_verify = (typeof((ptr) + 0))NULL; \
| ^~~
kernel/trace/trace_osnoise.c:2529:25: note: in expansion of macro 'per_cpu_ptr'
2529 | per_cpu_ptr(&per_cpu_timerlat_var, cpu)->uthread_migrate = 1;
| ^~~~~~~~~~~
At top level:
>> kernel/trace/trace_osnoise.c:2618:37: warning: 'timerlat_fd_fops' defined but not used [-Wunused-const-variable=]
2618 | static const struct file_operations timerlat_fd_fops = {
| ^~~~~~~~~~~~~~~~
cc1: some warnings being treated as errors
vim +2364 kernel/trace/trace_osnoise.c
2305
2306 static int timerlat_fd_open(struct inode *inode, struct file *file)
2307 {
2308 struct osnoise_variables *osn_var;
2309 struct timerlat_variables *tlat;
2310 long cpu = (long) inode->i_cdev;
2311
2312 mutex_lock(&interface_lock);
2313
2314 /*
2315 * This file is accessible only if timerlat is enabled, and
2316 * NO_OSNOISE_WORKLOAD is set.
2317 */
2318 if (!timerlat_enabled() || test_bit(OSN_WORKLOAD, &osnoise_options)) {
2319 mutex_unlock(&interface_lock);
2320 return -EINVAL;
2321 }
2322
2323 migrate_disable();
2324
2325 osn_var = this_cpu_osn_var();
2326
2327 /*
2328 * The osn_var->pid holds the single access to this file.
2329 */
2330 if (osn_var->pid) {
2331 mutex_unlock(&interface_lock);
2332 migrate_enable();
2333 return -EBUSY;
2334 }
2335
2336 /*
2337 * timerlat tracer is a per-cpu tracer. Check if the user-space too
2338 * is pinned to a single CPU. The tracer laters monitor if the task
2339 * migrates and then disables tracer if it does. However, it is
2340 * worth doing this basic acceptance test to avoid obviusly wrong
2341 * setup.
2342 */
2343 if (current->nr_cpus_allowed > 1 || cpu != smp_processor_id()) {
2344 mutex_unlock(&interface_lock);
2345 migrate_enable();
2346 return -EPERM;
2347 }
2348
2349 /*
2350 * From now on, it is good to go.
2351 */
2352 file->private_data = inode->i_cdev;
2353
2354 get_task_struct(current);
2355
2356 osn_var->kthread = current;
2357 osn_var->pid = current->pid;
2358
2359 /*
2360 * Setup is done.
2361 */
2362 mutex_unlock(&interface_lock);
2363
> 2364 tlat = this_cpu_tmr_var();
> 2365 tlat->count = 0;
2366
2367 migrate_enable();
2368 return 0;
2369 };
2370
2371 /*
2372 * timerlat_fd_read - Read function for "timerlat_fd" file
2373 * @file: The active open file structure
2374 * @ubuf: The userspace provided buffer to read value into
2375 * @cnt: The maximum number of bytes to read
2376 * @ppos: The current "file" position
2377 *
2378 * Prints 1 on timerlat, the number of interferences on osnoise, -1 on error.
2379 */
2380 static ssize_t
2381 timerlat_fd_read(struct file *file, char __user *ubuf, size_t count,
2382 loff_t *ppos)
2383 {
2384 long cpu = (long) file->private_data;
2385 struct osnoise_variables *osn_var;
2386 struct timerlat_variables *tlat;
> 2387 struct timerlat_sample s;
2388 s64 diff;
2389 u64 now;
2390
2391 migrate_disable();
2392
2393 tlat = this_cpu_tmr_var();
2394
2395 /*
2396 * While in user-space, the thread is migratable. There is nothing
2397 * we can do about it.
2398 * So, if the thread is running on another CPU, stop the machinery.
2399 */
2400 if (cpu == smp_processor_id()) {
2401 if (tlat->uthread_migrate) {
2402 migrate_enable();
2403 return -EINVAL;
2404 }
2405 } else {
> 2406 per_cpu_ptr(&per_cpu_timerlat_var, cpu)->uthread_migrate = 1;
2407 osnoise_taint("timerlat user thread migrate\n");
2408 osnoise_stop_tracing();
2409 migrate_enable();
2410 return -EINVAL;
2411 }
2412
2413 osn_var = this_cpu_osn_var();
2414
2415 /*
2416 * The timerlat in user-space runs in a different order:
2417 * the read() starts from the execution of the previous occurrence,
2418 * sleeping for the next occurrence.
2419 *
2420 * So, skip if we are entering on read() before the first wakeup
2421 * from timerlat IRQ:
2422 */
2423 if (likely(osn_var->sampling)) {
2424 now = ktime_to_ns(hrtimer_cb_get_time(&tlat->timer));
2425 diff = now - tlat->abs_period;
2426
2427 /*
2428 * it was not a timer firing, but some other signal?
2429 */
2430 if (diff < 0)
2431 goto out;
2432
2433 s.seqnum = tlat->count;
2434 s.timer_latency = diff;
2435 s.context = THREAD_URET;
2436
> 2437 trace_timerlat_sample(&s);
2438
2439 notify_new_max_latency(diff);
2440
2441 tlat->tracing_thread = false;
2442 if (osnoise_data.stop_tracing_total)
2443 if (time_to_us(diff) >= osnoise_data.stop_tracing_total)
2444 osnoise_stop_tracing();
2445 } else {
2446 tlat->tracing_thread = false;
2447 tlat->kthread = current;
2448
2449 hrtimer_init(&tlat->timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_PINNED_HARD);
> 2450 tlat->timer.function = timerlat_irq;
2451
2452 /* Annotate now to drift new period */
2453 tlat->abs_period = hrtimer_cb_get_time(&tlat->timer);
2454
2455 osn_var->sampling = 1;
2456 }
2457
2458 /* wait for the next period */
> 2459 wait_next_period(tlat);
2460
2461 /* This is the wakeup from this cycle */
2462 now = ktime_to_ns(hrtimer_cb_get_time(&tlat->timer));
2463 diff = now - tlat->abs_period;
2464
2465 /*
2466 * it was not a timer firing, but some other signal?
2467 */
2468 if (diff < 0)
2469 goto out;
2470
2471 s.seqnum = tlat->count;
2472 s.timer_latency = diff;
2473 s.context = THREAD_CONTEXT;
2474
2475 trace_timerlat_sample(&s);
2476
2477 if (osnoise_data.stop_tracing_total) {
2478 if (time_to_us(diff) >= osnoise_data.stop_tracing_total) {
> 2479 timerlat_dump_stack(time_to_us(diff));
2480 notify_new_max_latency(diff);
2481 osnoise_stop_tracing();
2482 }
2483 }
2484
2485 out:
2486 migrate_enable();
2487 return 0;
2488 }
2489
2490 static int timerlat_fd_release(struct inode *inode, struct file *file)
2491 {
2492 struct osnoise_variables *osn_var;
2493 struct timerlat_variables *tlat_var;
2494 long cpu = (long) file->private_data;
2495
2496 migrate_disable();
2497 mutex_lock(&interface_lock);
2498
2499 osn_var = per_cpu_ptr(&per_cpu_osnoise_var, cpu);
2500 tlat_var = per_cpu_ptr(&per_cpu_timerlat_var, cpu);
2501
2502 hrtimer_cancel(&tlat_var->timer);
> 2503 memset(tlat_var, 0, sizeof(*tlat_var));
2504
2505 osn_var->sampling = 0;
2506 osn_var->pid = 0;
2507
2508 /*
2509 * We are leaving, not being stopped... see stop_kthread();
2510 */
2511 if (osn_var->kthread) {
2512 put_task_struct(osn_var->kthread);
2513 osn_var->kthread = NULL;
2514 }
2515
2516 mutex_unlock(&interface_lock);
2517 migrate_enable();
2518 return 0;
2519 }
2520
On 5/24/23 14:48, kernel test robot wrote:
> Hi Daniel,
>
> kernel test robot noticed the following build errors:
>
> [auto build test ERROR on linus/master]
> [also build test ERROR on v6.4-rc3 next-20230524]
> [cannot apply to rostedt-trace/for-next rostedt-trace/for-next-urgent]
> [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#_base_tree_information]
>
> url: https://github.com/intel-lab-lkp/linux/commits/Daniel-Bristot-de-Oliveira/tracing-osnoise-Switch-from-PF_NO_SETAFFINITY-to-migrate_disable/20230524-012512
> base: linus/master
> patch link: https://lore.kernel.org/r/a7b2c215c763e95a56fa1258743332b570c81c9d.1684860626.git.bristot%40kernel.org
> patch subject: [PATCH V2 3/3] tracing/timerlat: Add user-space interface
> config: i386-randconfig-i014-20230523
> compiler: gcc-11 (Debian 11.3.0-12) 11.3.0
> reproduce (this is a W=1 build):
> # https://github.com/intel-lab-lkp/linux/commit/89216b54eaf490480bc1929f5780f95a688a91bb
> git remote add linux-review https://github.com/intel-lab-lkp/linux
> git fetch --no-tags linux-review Daniel-Bristot-de-Oliveira/tracing-osnoise-Switch-from-PF_NO_SETAFFINITY-to-migrate_disable/20230524-012512
> git checkout 89216b54eaf490480bc1929f5780f95a688a91bb
> # save the config file
> mkdir build_dir && cp config build_dir/.config
> make W=1 O=build_dir ARCH=i386 olddefconfig
> make W=1 O=build_dir ARCH=i386 SHELL=/bin/bash kernel/trace/
>
> If you fix the issue, kindly add following tag where applicable
> | Reported-by: kernel test robot <lkp@intel.com>
> | Closes: https://lore.kernel.org/oe-kbuild-all/202305242020.VlsOXEMn-lkp@intel.com/
>
> All error/warnings (new ones prefixed by >>):
>
> kernel/trace/trace_osnoise.c: In function 'timerlat_fd_open':
>>> kernel/trace/trace_osnoise.c:2364:16: error: implicit declaration of function 'this_cpu_tmr_var'; did you mean 'this_cpu_osn_var'? [-Werror=implicit-function-declaration]
> 2364 | tlat = this_cpu_tmr_var();
> | ^~~~~~~~~~~~~~~~
> | this_cpu_osn_var
Thanks robot, I forgot to test without timerlat enabled. Easy to fix, an ifdef
here and there.
-- Daniel
[...]
>
> +static void check_timerlat_user_migration(pid_t pid, long dest_cpu);
> /*
> * trace_sched_switch - sched:sched_switch trace event handler
> *
> @@ -1196,6 +1199,9 @@ trace_sched_switch_callback(void *data, bool preempt,
> struct osnoise_variables *osn_var = this_cpu_osn_var();
> int workload = test_bit(OSN_WORKLOAD, &osnoise_options);
>
> + if (unlikely(workload && timerlat_enabled()))
> + check_timerlat_user_migration(n->pid, smp_processor_id());
> +
it should be !workload, anyway, I will move this check to a sched_migrate_task
tracepoint because it runs less frequently and...
[...]
the tracepoint also informs the origin CPU, so it can be passed here:
> +
> +static void check_timerlat_user_migration(pid_t pid, long dest_cpu)
> +{
> + struct osnoise_variables *osn_var;
> + long cpu;
> +
and we can avoid this ugly loop.
> + for_each_possible_cpu(cpu) {
> + osn_var = per_cpu_ptr(&per_cpu_osnoise_var, cpu);
> + if (osn_var->pid == pid && dest_cpu != cpu) {
> + per_cpu_ptr(&per_cpu_timerlat_var, cpu)->uthread_migrate = 1;
> + osnoise_taint("timerlat user thread migrate\n");
> + osnoise_stop_tracing();
> + break;
> + }
> + }
Lazy daniel should have had a look first, but at least now I know I also need
to do some ifdeffery.
-- Daniel
@@ -180,3 +180,81 @@ dummy_load_1ms_pd_init, which had the following code (on purpose)::
return 0;
}
+
+User-space interface
+---------------------------
+
+Timerlat allows user-space threads to use timerlat infra-structure to
+measure scheduling latency. This interface is accessible via a per-CPU
+file descriptor inside $tracing_dir/osnoise/per_cpu/cpu$ID/timerlat_fd.
+
+This interface is accessible under the following conditions:
+
+ - timerlat tracer is enable
+ - osnoise workload option is set to NO_OSNOISE_WORKLOAD
+ - The user-space thread is affined to a single processor
+ - The thread opens the file associated with its single processor
+ - Only one thread can access the file at a time
+
+The open() syscall will fail if any of these conditions are not met.
+After opening the file descriptor, the user space can read from it.
+
+The read() system call will run a timerlat code that will arm the
+timer in the future and wait for it as the regular kernel thread does.
+
+When the timer IRQ fires, the timerlat IRQ will execute, report the
+IRQ latency and wake up the thread waiting in the read. The thread will be
+scheduled and report the thread latency via tracer - as for the kernel
+thread.
+
+The difference from the in-kernel timerlat is that, instead of re-arming
+the timer, timerlat will return to the read() system call. At this point,
+the user can run any code.
+
+If the application rereads the file timerlat file descriptor, the tracer
+will report the return from user-space latency, which is the total
+latency. If this is the end of the work, it can be interpreted as the
+response time for the request.
+
+After reporting the total latency, timerlat will restart the cycle, arm
+a timer, and go to sleep for the following activation.
+
+If at any time one of the conditions is broken, e.g., the thread migrates
+while in user space, or the timerlat tracer is disabled, the SIG_KILL
+signal will be sent to the user-space thread.
+
+Here is an basic example of user-space code for timerlat::
+
+ int main(void)
+ {
+ char buffer[1024];
+ int timerlat_fd;
+ int retval;
+ long cpu = 0; /* place in CPU 0 */
+ cpu_set_t set;
+
+ CPU_ZERO(&set);
+ CPU_SET(cpu, &set);
+
+ if (sched_setaffinity(gettid(), sizeof(set), &set) == -1)
+ return 1;
+
+ snprintf(buffer, sizeof(buffer),
+ "/sys/kernel/tracing/osnoise/per_cpu/cpu%ld/timerlat_fd",
+ cpu);
+
+ timerlat_fd = open(buffer, O_RDONLY);
+ if (timerlat_fd < 0) {
+ printf("error opening %s: %s\n", buffer, strerror(errno));
+ exit(1);
+ }
+
+ for (;;) {
+ retval = read(timerlat_fd, buffer, 1024);
+ if (retval < 0)
+ break;
+ }
+
+ close(timerlat_fd);
+ exit(0);
+ }
@@ -181,6 +181,7 @@ struct osn_irq {
#define IRQ_CONTEXT 0
#define THREAD_CONTEXT 1
+#define THREAD_URET 2
/*
* sofirq runtime info.
*/
@@ -238,6 +239,7 @@ struct timerlat_variables {
u64 abs_period;
bool tracing_thread;
u64 count;
+ bool uthread_migrate;
};
static DEFINE_PER_CPU(struct timerlat_variables, per_cpu_timerlat_var);
@@ -1181,6 +1183,7 @@ thread_exit(struct osnoise_variables *osn_var, struct task_struct *t)
osn_var->thread.arrival_time = 0;
}
+static void check_timerlat_user_migration(pid_t pid, long dest_cpu);
/*
* trace_sched_switch - sched:sched_switch trace event handler
*
@@ -1196,6 +1199,9 @@ trace_sched_switch_callback(void *data, bool preempt,
struct osnoise_variables *osn_var = this_cpu_osn_var();
int workload = test_bit(OSN_WORKLOAD, &osnoise_options);
+ if (unlikely(workload && timerlat_enabled()))
+ check_timerlat_user_migration(n->pid, smp_processor_id());
+
if ((p->pid != osn_var->pid) || !workload)
thread_exit(osn_var, p);
@@ -1864,10 +1870,24 @@ static void stop_kthread(unsigned int cpu)
kthread = per_cpu(per_cpu_osnoise_var, cpu).kthread;
if (kthread) {
- kthread_stop(kthread);
+ if (test_bit(OSN_WORKLOAD, &osnoise_options)) {
+ kthread_stop(kthread);
+ } else {
+ /*
+ * This is a user thread waiting on the timerlat_fd. We need
+ * to close all users, and the best way to guarantee this is
+ * by killing the thread. NOTE: this is a purpose specific file.
+ */
+ kill_pid(kthread->thread_pid, SIGKILL, 1);
+ put_task_struct(kthread);
+ }
per_cpu(per_cpu_osnoise_var, cpu).kthread = NULL;
} else {
+ /* if no workload, just return */
if (!test_bit(OSN_WORKLOAD, &osnoise_options)) {
+ /*
+ * This is set in the osnoise tracer case.
+ */
per_cpu(per_cpu_osnoise_var, cpu).sampling = false;
barrier();
return;
@@ -1912,7 +1932,6 @@ static int start_kthread(unsigned int cpu)
barrier();
return 0;
}
-
snprintf(comm, 24, "osnoise/%d", cpu);
}
@@ -1941,6 +1960,11 @@ static int start_per_cpu_kthreads(void)
int retval = 0;
int cpu;
+ if (!test_bit(OSN_WORKLOAD, &osnoise_options)) {
+ if (timerlat_enabled())
+ return 0;
+ }
+
cpus_read_lock();
/*
* Run only on online CPUs in which osnoise is allowed to run.
@@ -2281,6 +2305,238 @@ osnoise_cpus_write(struct file *filp, const char __user *ubuf, size_t count,
return err;
}
+static int timerlat_fd_open(struct inode *inode, struct file *file)
+{
+ struct osnoise_variables *osn_var;
+ struct timerlat_variables *tlat;
+ long cpu = (long) inode->i_cdev;
+
+ mutex_lock(&interface_lock);
+
+ /*
+ * This file is accessible only if timerlat is enabled, and
+ * NO_OSNOISE_WORKLOAD is set.
+ */
+ if (!timerlat_enabled() || test_bit(OSN_WORKLOAD, &osnoise_options)) {
+ mutex_unlock(&interface_lock);
+ return -EINVAL;
+ }
+
+ migrate_disable();
+
+ osn_var = this_cpu_osn_var();
+
+ /*
+ * The osn_var->pid holds the single access to this file.
+ */
+ if (osn_var->pid) {
+ mutex_unlock(&interface_lock);
+ migrate_enable();
+ return -EBUSY;
+ }
+
+ /*
+ * timerlat tracer is a per-cpu tracer. Check if the user-space too
+ * is pinned to a single CPU. The tracer laters monitor if the task
+ * migrates and then disables tracer if it does. However, it is
+ * worth doing this basic acceptance test to avoid obviusly wrong
+ * setup.
+ */
+ if (current->nr_cpus_allowed > 1 || cpu != smp_processor_id()) {
+ mutex_unlock(&interface_lock);
+ migrate_enable();
+ return -EPERM;
+ }
+
+ /*
+ * From now on, it is good to go.
+ */
+ file->private_data = inode->i_cdev;
+
+ get_task_struct(current);
+
+ osn_var->kthread = current;
+ osn_var->pid = current->pid;
+
+ /*
+ * Setup is done.
+ */
+ mutex_unlock(&interface_lock);
+
+ tlat = this_cpu_tmr_var();
+ tlat->count = 0;
+
+ migrate_enable();
+ return 0;
+};
+
+/*
+ * timerlat_fd_read - Read function for "timerlat_fd" file
+ * @file: The active open file structure
+ * @ubuf: The userspace provided buffer to read value into
+ * @cnt: The maximum number of bytes to read
+ * @ppos: The current "file" position
+ *
+ * Prints 1 on timerlat, the number of interferences on osnoise, -1 on error.
+ */
+static ssize_t
+timerlat_fd_read(struct file *file, char __user *ubuf, size_t count,
+ loff_t *ppos)
+{
+ long cpu = (long) file->private_data;
+ struct osnoise_variables *osn_var;
+ struct timerlat_variables *tlat;
+ struct timerlat_sample s;
+ s64 diff;
+ u64 now;
+
+ migrate_disable();
+
+ tlat = this_cpu_tmr_var();
+
+ /*
+ * While in user-space, the thread is migratable. There is nothing
+ * we can do about it.
+ * So, if the thread is running on another CPU, stop the machinery.
+ */
+ if (cpu == smp_processor_id()) {
+ if (tlat->uthread_migrate) {
+ migrate_enable();
+ return -EINVAL;
+ }
+ } else {
+ per_cpu_ptr(&per_cpu_timerlat_var, cpu)->uthread_migrate = 1;
+ osnoise_taint("timerlat user thread migrate\n");
+ osnoise_stop_tracing();
+ migrate_enable();
+ return -EINVAL;
+ }
+
+ osn_var = this_cpu_osn_var();
+
+ /*
+ * The timerlat in user-space runs in a different order:
+ * the read() starts from the execution of the previous occurrence,
+ * sleeping for the next occurrence.
+ *
+ * So, skip if we are entering on read() before the first wakeup
+ * from timerlat IRQ:
+ */
+ if (likely(osn_var->sampling)) {
+ now = ktime_to_ns(hrtimer_cb_get_time(&tlat->timer));
+ diff = now - tlat->abs_period;
+
+ /*
+ * it was not a timer firing, but some other signal?
+ */
+ if (diff < 0)
+ goto out;
+
+ s.seqnum = tlat->count;
+ s.timer_latency = diff;
+ s.context = THREAD_URET;
+
+ trace_timerlat_sample(&s);
+
+ notify_new_max_latency(diff);
+
+ tlat->tracing_thread = false;
+ if (osnoise_data.stop_tracing_total)
+ if (time_to_us(diff) >= osnoise_data.stop_tracing_total)
+ osnoise_stop_tracing();
+ } else {
+ tlat->tracing_thread = false;
+ tlat->kthread = current;
+
+ hrtimer_init(&tlat->timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_PINNED_HARD);
+ tlat->timer.function = timerlat_irq;
+
+ /* Annotate now to drift new period */
+ tlat->abs_period = hrtimer_cb_get_time(&tlat->timer);
+
+ osn_var->sampling = 1;
+ }
+
+ /* wait for the next period */
+ wait_next_period(tlat);
+
+ /* This is the wakeup from this cycle */
+ now = ktime_to_ns(hrtimer_cb_get_time(&tlat->timer));
+ diff = now - tlat->abs_period;
+
+ /*
+ * it was not a timer firing, but some other signal?
+ */
+ if (diff < 0)
+ goto out;
+
+ s.seqnum = tlat->count;
+ s.timer_latency = diff;
+ s.context = THREAD_CONTEXT;
+
+ trace_timerlat_sample(&s);
+
+ if (osnoise_data.stop_tracing_total) {
+ if (time_to_us(diff) >= osnoise_data.stop_tracing_total) {
+ timerlat_dump_stack(time_to_us(diff));
+ notify_new_max_latency(diff);
+ osnoise_stop_tracing();
+ }
+ }
+
+out:
+ migrate_enable();
+ return 0;
+}
+
+static int timerlat_fd_release(struct inode *inode, struct file *file)
+{
+ struct osnoise_variables *osn_var;
+ struct timerlat_variables *tlat_var;
+ long cpu = (long) file->private_data;
+
+ migrate_disable();
+ mutex_lock(&interface_lock);
+
+ osn_var = per_cpu_ptr(&per_cpu_osnoise_var, cpu);
+ tlat_var = per_cpu_ptr(&per_cpu_timerlat_var, cpu);
+
+ hrtimer_cancel(&tlat_var->timer);
+ memset(tlat_var, 0, sizeof(*tlat_var));
+
+ osn_var->sampling = 0;
+ osn_var->pid = 0;
+
+ /*
+ * We are leaving, not being stopped... see stop_kthread();
+ */
+ if (osn_var->kthread) {
+ put_task_struct(osn_var->kthread);
+ osn_var->kthread = NULL;
+ }
+
+ mutex_unlock(&interface_lock);
+ migrate_enable();
+ return 0;
+}
+
+static void check_timerlat_user_migration(pid_t pid, long dest_cpu)
+{
+ struct osnoise_variables *osn_var;
+ long cpu;
+
+ for_each_possible_cpu(cpu) {
+ osn_var = per_cpu_ptr(&per_cpu_osnoise_var, cpu);
+ if (osn_var->pid == pid && dest_cpu != cpu) {
+ per_cpu_ptr(&per_cpu_timerlat_var, cpu)->uthread_migrate = 1;
+ osnoise_taint("timerlat user thread migrate\n");
+ osnoise_stop_tracing();
+ break;
+ }
+ }
+}
+
+
/*
* osnoise/runtime_us: cannot be greater than the period.
*/
@@ -2361,6 +2617,13 @@ static const struct file_operations osnoise_options_fops = {
.write = osnoise_options_write
};
+static const struct file_operations timerlat_fd_fops = {
+ .open = timerlat_fd_open,
+ .read = timerlat_fd_read,
+ .release = timerlat_fd_release,
+ .llseek = generic_file_llseek,
+};
+
#ifdef CONFIG_TIMERLAT_TRACER
#ifdef CONFIG_STACKTRACE
static int init_timerlat_stack_tracefs(struct dentry *top_dir)
@@ -2381,18 +2644,63 @@ static int init_timerlat_stack_tracefs(struct dentry *top_dir)
}
#endif /* CONFIG_STACKTRACE */
+int osnoise_create_cpu_timerlat_fd(struct dentry *top_dir)
+{
+ struct dentry *timerlat_fd;
+ struct dentry *per_cpu;
+ struct dentry *cpu_dir;
+ char cpu_str[30]; /* see trace.c: tracing_init_tracefs_percpu() */
+ long cpu;
+
+ /*
+ * Why not using tracing instance per_cpu/ dir?
+ *
+ * Because osnoise/timerlat have a single workload, having
+ * multiple files like these are wast of memory.
+ */
+ per_cpu = tracefs_create_dir("per_cpu", top_dir);
+ if (!per_cpu)
+ return -ENOMEM;
+
+ for_each_possible_cpu(cpu) {
+ snprintf(cpu_str, 30, "cpu%ld", cpu);
+ cpu_dir = tracefs_create_dir(cpu_str, per_cpu);
+ if (!cpu_dir)
+ goto out_clean;
+
+ timerlat_fd = trace_create_file("timerlat_fd", TRACE_MODE_READ,
+ cpu_dir, NULL, &timerlat_fd_fops);
+ if (!timerlat_fd)
+ goto out_clean;
+
+ /* Record the CPU */
+ d_inode(timerlat_fd)->i_cdev = (void *)(cpu);
+ }
+
+ return 0;
+
+out_clean:
+ tracefs_remove(per_cpu);
+ return -ENOMEM;
+}
+
/*
* init_timerlat_tracefs - A function to initialize the timerlat interface files
*/
static int init_timerlat_tracefs(struct dentry *top_dir)
{
struct dentry *tmp;
+ int retval;
tmp = tracefs_create_file("timerlat_period_us", TRACE_MODE_WRITE, top_dir,
&timerlat_period, &trace_min_max_fops);
if (!tmp)
return -ENOMEM;
+ retval = osnoise_create_cpu_timerlat_fd(top_dir);
+ if (retval)
+ return retval;
+
return init_timerlat_stack_tracefs(top_dir);
}
#else /* CONFIG_TIMERLAT_TRACER */
@@ -1446,6 +1446,8 @@ static struct trace_event trace_osnoise_event = {
};
/* TRACE_TIMERLAT */
+
+static char *timerlat_lat_context[] = {"irq", "thread", "user-ret"};
static enum print_line_t
trace_timerlat_print(struct trace_iterator *iter, int flags,
struct trace_event *event)
@@ -1458,7 +1460,7 @@ trace_timerlat_print(struct trace_iterator *iter, int flags,
trace_seq_printf(s, "#%-5u context %6s timer_latency %9llu ns\n",
field->seqnum,
- field->context ? "thread" : "irq",
+ timerlat_lat_context[field->context],
field->timer_latency);
return trace_handle_return(s);