The -t option has an optional argument.
The usual case is for a short option to be specified without an '='
and for the long version to be specified with an '='
Various forms of this do not work as expected.
For example:
rtla timerlat hist -T50 -tfile.txt
will result in a truncated file name of "ile.txt"
Another example is that the long form without the '=' will result in the
default file name instead of the requested file name.
This patch properly parses the optional argument with and without '='
and with and without spaces for the short form.
This patch was also tested using -t and --trace without providing a file
name both as the last requested option and with a following long and
short option.
For example:
rtla timerlat hist -T50 -t -u
rtla timerlat hist -T50 --trace -u
This fix is applied to both timerlat top and hist
and to osnoise top and hist.
Here is the full testing for rtla timerlat hist.
Before applying the patch
rtla timerlat hist -T50 -t=file.txt
Works as expected, "file.txt"
rtla timerlat hist -T50 -tfile.txt
Truncated file name "ile.txt"
rtla timerlat hist -T50 -t file.txt
Default file name instead of file.txt
rtla timerlat hist -T50 --trace=file.txt
Truncated file name "ile.txt"
rtla timerlat hist -T50 --trace file.txt
Default file name "timerlat_trace.txt" instead of "file.txt"
After applying the patch:
rtla timerlat hist -T50 -t=file.txt
Works as expected, "file.txt"
rtla timerlat hist -T50 -tfile.txt
Works as expected, "file.txt"
rtla timerlat hist -T50 -t file.txt
Works as expected, "file.txt"
rtla timerlat hist -T50 --trace=file.txt
Works as expected, "file.txt"
rtla timerlat hist -T50 --trace file.txt
Works as expected, "file.txt"
In addition the following tests were performed to make sure that
the default file name worked as expected including with trailing
options.
rtla timerlat hist -T50 -t
Works as expected "timerlat_trace.txt"
rtla timerlat hist -T50 --trace
Works as expected "timerlat_trace.txt"
rtla timerlat hist -T50 -t -u
Works as expected "timerlat_trace.txt"
rtla timerlat hist -T50 --trace -u
Works as expected "timerlat_trace.txt"
Link: https://lkml.kernel.org/r/20240515183024.59985-1-jkacur@redhat.com
Cc: Daniel Bristot de Oliveria <bristot@kernel.org>
Signed-off-by: John Kacur <jkacur@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
On short runs it is possible to get no samples on a cpu, like this:
# rtla timerlat hist -u -T50
Index IRQ-001 Thr-001 Usr-001 IRQ-002 Thr-002 Usr-002
2 1 0 0 0 0 0
33 0 1 0 0 0 0
36 0 0 1 0 0 0
49 0 0 0 1 0 0
52 0 0 0 0 1 0
over: 0 0 0 0 0 0
count: 1 1 1 1 1 0
min: 2 33 36 49 52 18446744073709551615
avg: 2 33 36 49 52 -
max: 2 33 36 49 52 0
rtla timerlat hit stop tracing
IRQ handler delay: (exit from idle) 48.21 us (91.09 %)
IRQ latency: 49.11 us
Timerlat IRQ duration: 2.17 us (4.09 %)
Blocking thread: 1.01 us (1.90 %)
swapper/2:0 1.01 us
------------------------------------------------------------------------
Thread latency: 52.93 us (100%)
Max timerlat IRQ latency from idle: 49.11 us in cpu 2
Note, the value 18446744073709551615 is the same as ~0.
Fix this by reporting no results for the min, avg and max if the count
is 0.
Link: https://lkml.kernel.org/r/20240510190318.44295-1-jkacur@redhat.com
Cc: stable@vger.kernel.org
Fixes: 1eeb6328e8 ("rtla/timerlat: Add timerlat hist mode")
Suggested-by: Daniel Bristot de Oliveria <bristot@kernel.org>
Signed-off-by: John Kacur <jkacur@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Add the option allow the users to set a different buffer size for the
trace. For example, in large systems, the user might be interested on
reducing the trace buffer to avoid large tracing files.
The buffer size is specified in kB, and it is only affecting
the tracing instance.
The function trace_set_buffer_size() appears on libtracefs v1.6,
so increase the minimum required version on Makefile.config.
Link: https://lkml.kernel.org/r/e7c9ca5b3865f28e131a49ec3b984fadf2d056c6.1715860611.git.bristot@kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: John Kacur <jkacur@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
After ther -u addition, most of the known users are setting it. And
it makes sense, as it adds more information, and inherits the default
setup for the threads - e.g., cgroups configs.
Thus, if the user-space interface is available, enable -u. Otherwise,
use the in-kernel thread.
Add the -k option to allow the user to request kernel-threads.
Link: https://lkml.kernel.org/r/9241d3089de4091b124f780ed832a0e6646cadaa.1713968967.git.bristot@kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
On many cases, the results right after the startup are different
from the rest of the execution, biasing the results. For example,
on osnoise, the scheduler might take some time to adapt to the new
busy-loop workload.
Add the --warm-up <seconds> option, adding a warm-up phase (in
seconds) where the workload is set, but the results are discarded.
Link: https://lkml.kernel.org/r/e682d5ce5af90f123bd13220f63d5c3d118a92be.1713968967.git.bristot@kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
While the per-cpu values are the results to take into consideration, the
overall system values are also useful.
Add a summary at the bottom of rtla timerlat top showing the overall
results. For instance:
Timer Latency
0 00:00:10 | IRQ Timer Latency (us) | Thread Timer Latency (us)
CPU COUNT | cur min avg max | cur min avg max
0 #10003 | 113 19 150 441 | 134 35 170 459
1 #10003 | 63 8 99 462 | 84 15 119 481
2 #10003 | 3 2 89 396 | 21 8 108 414
3 #10002 | 206 11 210 394 | 223 21 228 415
---------------|----------------------------------------|---------------------------------------
ALL #40011 e0 | 2 137 462 | 8 156 481
Link: https://lkml.kernel.org/r/5eb510d6faeb4ce745e09395196752df75a2dd1a.1713968967.git.bristot@kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Suggested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
timerlat top does some background/font color formatting. While useful
on terminal, it breaks the output on other formats. For example, when
piping the output for pastebin tools, the format strings are printed
as characters. For instance:
[2;37;40m Timer Latency [0;0;0m
0 00:00:01 | IRQ Timer Latency (us) | Thread Timer Latency (us)
[2;30;47mCPU COUNT | cur min avg max | cur min avg max[0;0;0m
0 #1013 | 1 0 1 54 | 5 2 4 57
1 #1013 | 3 0 1 10 | 6 2 4 15
To avoid this problem, do the formatting only if running on a tty,
and in !quiet mode.
Link: https://lkml.kernel.org/r/8288e1544ceab21557d5dda93a0f00339497c649.1713968967.git.bristot@kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
When copying timerlat auto-analysis from a terminal to some web pages or
chats, the \t are being replaced with a single ' ' or ' ', breaking
the output.
For example:
## CPU 3 hit stop tracing, analyzing it ##
IRQ handler delay: 1.30 us (0.11 %)
IRQ latency: 1.90 us
Timerlat IRQ duration: 3.00 us (0.24 %)
Blocking thread: 1223.16 us (99.00 %)
insync:4048 1223.16 us
IRQ interference 4.93 us (0.40 %)
local_timer:236 4.93 us
------------------------------------------------------------------------
Thread latency: 1235.47 us (100%)
Replace \t with spaces to avoid this problem.
Link: https://lkml.kernel.org/r/ec7ed2b2809c22ab0dfc8eb7c805ab9cddc4254a.1713968967.git.bristot@kernel.org
Cc: stable@vger.kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Juri Lelli <juri.lelli@redhat.com>
Fixes: 27e348b221 ("rtla/timerlat: Add auto-analysis core")
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Instead of printing three times the same output, print it only once,
reducing lines and being sure that all no values have the same length.
It also fixes an extra '\n' when running the with kernel threads, like
here:
=============== %< ==============
Timer Latency
0 00:00:01 | IRQ Timer Latency (us) | Thread Timer Latency (us)
CPU COUNT | cur min avg max | cur min avg max
2 #0 | - - - - | 161 161 161 161
3 #0 | - - - - | 161 161 161 161
8 #1 | 54 54 54 54 | - - - -'\n'
---------------|----------------------------------------|---------------------------------------
ALL #1 e0 | 54 54 54 | 161 161 161
=============== %< ==============
This '\n' should have been removed with the user-space support that
added another '\n' if not running with kernel threads.
Link: https://lkml.kernel.org/r/0a4d8085e7cd706733a5dc10a81ca38b82bd4992.1713968967.git.bristot@kernel.org
Cc: stable@vger.kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Juri Lelli <juri.lelli@redhat.com>
Fixes: cdca4f4e5e ("rtla/timerlat_top: Add timerlat user-space support")
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
The timerlat tracer provides an interface for any application to wait
for the timerlat's periodic wakeup. Currently, rtla timerlat uses it
to dispatch its user-space workload (-u option).
But as the tracer interface is generic, rtla timerlat can also be used
to monitor any workload that uses it. For example, a user might
place their own workload to wait on the tracer interface, and
monitor the results with rtla timerlat.
Add the -U option to rtla timerlat top and hist. With this option, rtla
timerlat will not dispatch its workload but only setting up the
system, waiting for a user to dispatch its workload.
The sample code in this patch is an example of python application
that loops in the timerlat tracer fd.
To use it, dispatch:
# rtla timerlat -U
In a terminal, then run the python program on another terminal,
specifying the CPU to run it. For example, setting on CPU 1:
#./timerlat_load.py 1
Then rtla timerlat will start printing the statistics of the
./timerlat_load.py app.
An interesting point is that the "Ret user Timer Latency" value
is the overall response time of the load. The sample load does
a memory copy to exemplify that.
The stop tracing options on rtla timerlat works in this setup
as well, including auto analysis.
Link: https://lkml.kernel.org/r/36e6bcf18fe15c7601048fd4c65aeb193c502cc8.1707229706.git.bristot@kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Use tools/build/ makefiles to build rtla, inheriting the benefits of
it. For example, having a proper way to handle dependencies.
rtla is built using perf infra-structure when building inside the
kernel tree.
At this point, rtla diverges from perf in two points: Documentation
and tarball generation/build.
At the documentation level, rtla is one step ahead, placing the
documentation at Documentation/tools/rtla/, using the same build
tools as kernel documentation. The idea is to move perf
documentation to the same scheme and then share the same makefiles.
rtla has a tarball target that the (old) RHEL8 uses. The tarball was
kept using a simple standalone makefile for compatibility. The
standalone makefile shares most of the code, e.g., flags, with
regular buildings.
The tarball method was set as deprecated. If necessary, we can make
a rtla tarball like perf, which includes the entire tools/build.
But this would also require changes in the user side (the directory
structure changes, and probably the deps to build the package).
Inspired on perf and objtool.
Link: https://lkml.kernel.org/r/57563abf2715d22515c0c54a87cff3849eca5d52.1710519524.git.bristot@kernel.org
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Josh Poimboeuf <jpoimboe@kernel.org>
Cc: John Kacur <jkacur@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
When compiling rtla with clang, I am getting the following warnings:
$ make HOSTCC=clang CC=clang LLVM_IAS=1
[..]
clang -O -g -DVERSION=\"6.8.0-rc3\" -flto=auto -fexceptions
-fstack-protector-strong -fasynchronous-unwind-tables
-fstack-clash-protection -Wall -Werror=format-security
-Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS
$(pkg-config --cflags libtracefs)
-c -o src/osnoise_hist.o src/osnoise_hist.c
src/osnoise_hist.c:138:6: warning: variable 'bucket' is used uninitialized whenever 'if' condition is false [-Wsometimes-uninitialized]
138 | if (data->bucket_size)
| ^~~~~~~~~~~~~~~~~
src/osnoise_hist.c:149:6: note: uninitialized use occurs here
149 | if (bucket < entries)
| ^~~~~~
src/osnoise_hist.c:138:2: note: remove the 'if' if its condition is always true
138 | if (data->bucket_size)
| ^~~~~~~~~~~~~~~~~~~~~~
139 | bucket = duration / data->bucket_size;
src/osnoise_hist.c:132:12: note: initialize the variable 'bucket' to silence this warning
132 | int bucket;
| ^
| = 0
1 warning generated.
[...]
clang -O -g -DVERSION=\"6.8.0-rc3\" -flto=auto -fexceptions
-fstack-protector-strong -fasynchronous-unwind-tables
-fstack-clash-protection -Wall -Werror=format-security
-Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS
$(pkg-config --cflags libtracefs)
-c -o src/timerlat_hist.o src/timerlat_hist.c
src/timerlat_hist.c:181:6: warning: variable 'bucket' is used uninitialized whenever 'if' condition is false [-Wsometimes-uninitialized]
181 | if (data->bucket_size)
| ^~~~~~~~~~~~~~~~~
src/timerlat_hist.c:204:6: note: uninitialized use occurs here
204 | if (bucket < entries)
| ^~~~~~
src/timerlat_hist.c:181:2: note: remove the 'if' if its condition is always true
181 | if (data->bucket_size)
| ^~~~~~~~~~~~~~~~~~~~~~
182 | bucket = latency / data->bucket_size;
src/timerlat_hist.c:175:12: note: initialize the variable 'bucket' to silence this warning
175 | int bucket;
| ^
| = 0
1 warning generated.
This is a legit warning, but data->bucket_size is always > 0 (see
timerlat_hist_parse_args()), so the if is not necessary.
Remove the unneeded if (data->bucket_size) to avoid the warning.
Link: https://lkml.kernel.org/r/6e1b1665cd99042ae705b3e0fc410858c4c42346.1707217097.git.bristot@kernel.org
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Bill Wendling <morbo@google.com>
Cc: Justin Stitt <justinstitt@google.com>
Cc: Donald Zickus <dzickus@redhat.com>
Fixes: 1eeb6328e8 ("rtla/timerlat: Add timerlat hist mode")
Fixes: 829a6c0b56 ("rtla/osnoise: Add the hist mode")
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Variable found is not being initialized, in the case where the desired
mount is not found the variable contains garbage. Fix this by initializing
it to zero.
Link: https://lore.kernel.org/all/20230727150117.627730-1-colin.i.king@gmail.com/
Fixes: a957cbc025 ("rtla: Add -C cgroup support")
Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
If no CPU list is passed, timerlat in user-space will dispatch
one thread per sysconf(_SC_NPROCESSORS_CONF). However, not all
CPU might be available, for instance, if HT is disabled.
Currently, rtla timerlat is stopping the session if an user-space
thread cannot set affinity to a CPU, or if a running user-space
thread is killed. However, this is too restrictive.
So, reduce the error to a debug message, and rtla timerlat run as
long as there is at least one user-space thread alive.
Link: https://lore.kernel.org/lkml/59cf2c882900ab7de91c6ee33b382ac7fa6b4ed0.1694781909.git.bristot@kernel.org
Fixes: cdca4f4e5e ("rtla/timerlat_top: Add timerlat user-space support")
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
timerlat auto-analysis takes note of all IRQs, before or after the
execution of the timerlat thread.
Because we cannot go backward in the trace (we will fix it when
moving to trace-cmd lib?), timerlat aa take note of the last IRQ
execution in the waiting for the IRQ state, and then print it
if it is executed after the expected timer IRQ starting time.
After the thread sample, the timerlat starts recording the next IRQs as
"previous" irq for the next occurrence.
However, if an IRQ happens after the thread measurement but before the
tracing stops, it is classified as a previous IRQ. That is not
wrong, as it can be "previous" for the subsequent activation. What is
wrong is considering it as a potential source for the last activation.
Ignore the IRQ interference that happens after the IRQ starting time for
now. A future improvement for timerlat can be either keeping a list of
previous IRQ execution or using the trace-cmd library. Still, it requires
further investigation - it is a new feature.
Link: https://lore.kernel.org/lkml/a44a3f5c801dcc697bacf7325b65d4a5b0460537.1691162043.git.bristot@kernel.org
Fixes: 27e348b221 ("rtla/timerlat: Add auto-analysis core")
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
When estimating the IRQ timer delay, we are dealing with two different
clock sources: the external clock source that timerlat uses as a reference
and the clock used by the tracer. There are also two moments: the time
reading the clock and the timer in which the event is placed in the
buffer (the trace event timestamp).
If the processor is slow or there is some hardware noise, the difference
between the timestamp and the external clock, read can be longer than the
IRQ handler delay, resulting in a negative time.
If so, set IRQ to start delay as 0. In the end, it is less near-zero and relevant
then the noise.
Link: https://lore.kernel.org/lkml/a066fb667c7136d86dcddb3c7ccd72587db3e7c7.1691162043.git.bristot@kernel.org
Fixes: 27e348b221 ("rtla/timerlat: Add auto-analysis core")
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
The thread thread_thread_sum accounts for thread interference
during a single activation. It was not being zeroed, so it was
accumulating thread interference over all activations.
It was not that visible when timerlat was the highest priority.
Link: https://lore.kernel.org/lkml/97bff55b0141f2d01b47d9450a5672fde147b89a.1691162043.git.bristot@kernel.org
Fixes: 27e348b221 ("rtla/timerlat: Add auto-analysis core")
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
osnoise runs 100% of time by default. It makes sense because osnoise
is preemptive. hwnoise checks preemption once a second, so it
reduces system progress.
Reduce runtime to 75% to avoid problems by default. I added a Fixes
as it might avoid problems for first time users as it lands on distros.
Link: https://lkml.kernel.org/r/af0b7113ffc00031b9af4bb40ef5889a27dadf8c.1686066600.git.bristot@kernel.org
Cc: William White <chwhite@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Fixes: 1f428356c3 ("rtla: Add hwnoise tool")
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Group all start tracing after finishing creating all instances.
The tracing instance starts first for the case of hitting a stop
tracing while enabling other instances. The trace instance is the
one with most valuable information.
Link: https://lkml.kernel.org/r/67da7a703a56f75d7cd46568525145a65501a7e8.1686066600.git.bristot@kernel.org
Cc: William White <chwhite@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Add auto-analysis to timerlat hist, including the --no-aa option to
reduce overhead and --dump-task. --aa-only was not added as it is
already on timerlat top.
Link: https://lkml.kernel.org/r/c2693f47ee83e659a7723fed8035f5d2534f528e.1686066600.git.bristot@kernel.org
Cc: William White <chwhite@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Currently, the auto-analysis is attached to the timerlat top instance.
The idea was to avoid creating another instance just for that, so one
instance could be reused.
The drawback is that, by doing so, the auto-analysis run for the entire
session, consuming CPU time. On my 24 box CPUs for timerlat with a 100
us period consumed 50 % with auto analysis, but only 16 % without.
By creating an instance for auto-analysis, we can keep the processing
stopped until a stop tracing condition is hit. Once it happens,
timerlat auto-analysis can use its own trace instance to parse only
the end of the trace.
By doing so, auto-analysis stop consuming cpu time when it is not
needed.
If the --aa-only is passed, the timerlat top instance is reused for
auto analysis.
Link: https://lkml.kernel.org/r/346b7168c1bae552a415715ec6d23c129a43bdb7.1686066600.git.bristot@kernel.org
Cc: William White <chwhite@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
When the user sets -c <cpu-list> try to move rtla out of the <cpu-list>,
even without an -H option. This is useful to avoid having rtla
interfering with the workload.
This works by removing <cpu-list> from rtla's current affinity.
If rtla fails to move itself away it is not that of a problem as this
is an automatic measure.
Link: https://lkml.kernel.org/r/c54304d90c777310fb85a3e658d1449173759aab.1686066600.git.bristot@kernel.org
Cc: William White <chwhite@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Use a cpumask instead of a char *, reducing memory footprint and code.
No functional change, and in preparation for auto house-keeping.
Link: https://lkml.kernel.org/r/54c46293261d13cb1042d0314486539eeb45fe5d.1686066600.git.bristot@kernel.org
Cc: William White <chwhite@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
To avoid having rtla interfering with the measurement threads, add an
option for the user to set the CPUs in which rtla should run. For
instance:
# rtla timerlat top -H 0 -c 1-7
Will place rtla in the CPU 0, while running the measurement threads in
the CPU 1-7.
Link: https://lkml.kernel.org/r/6a6c78a579a96ba8b02ae67ee1e0ba2cb5e03c4a.1686066600.git.bristot@kernel.org
Cc: William White <chwhite@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Suggested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
The -C option sets a cgroup to the tracer's threads. If the -C option is
passed without arguments, the tracer's thread will inherit rtla's
cgroup. Otherwise, the threads will be placed on the cgroup passed
to the option.
Link: https://lkml.kernel.org/r/cb051477331d292f17c08bf1d66f0e0384bbe5a5.1686066600.git.bristot@kernel.org
Cc: William White <chwhite@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
The "Previous IRQ interference" line is misaligned and without
a \n, breaking the tool's output:
## CPU 12 hit stop tracing, analyzing it ##
Previous IRQ interference: up to 2.22 us IRQ handler delay: 18.06 us (0.00 %)
IRQ latency: 18.52 us
Timerlat IRQ duration: 4.41 us (0.00 %)
Blocking thread: 216.93 us (0.03 %)
Fix the output:
## CPU 7 hit stop tracing, analyzing it ##
Previous IRQ interference: up to 8.93 us
IRQ handler delay: 0.98 us (0.00 %)
IRQ latency: 2.95 us
Timerlat IRQ duration: 11.26 us (0.03 %)
Link: https://lore.kernel.org/linux-trace-devel/8b5819077f15ccf24745c9bf3205451e16ee32d9.1679685525.git.bristot@kernel.org
Fixes: 27e348b221 ("rtla/timerlat: Add auto-analysis core")
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Parsing and formating timerlat data might consume a reasonable
amount of CPU time on very large systems, or when timerlat
has a too short period.
Add an option to run timerlat with auto-analysis enabled while
skipping the statistics parsing. In this mode, rtla timerlat
periodically checks if the tracing is on, going to sleep waiting
for the stop tracing condition to stop tracing, or for the
tracing session to finish.
If the stop tracing condition is hit, the tool prints the auto
analysis. Otherwise, the tool prints the max observed latency and
exit. The max observed latency is captured via tracing_max_latency.
Link: https://lore.kernel.org/linux-trace-devel/4dc514d1d5dc353c537a466a9b5af44c266b6da2.1680106912.git.bristot@kernel.org
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
The hwnoise tool is a special mode for the osnoise top tool.
hwnoise dispatches the osnoise tracer and displays a summary of the noise.
The difference is that it runs the tracer with the OSNOISE_IRQ_DISABLE
option set, thus only allowing only hardware-related noise, resulting in
a simplified output. hwnoise has the same features of osnoise.
An example of the tool's output:
# rtla hwnoise -c 1-11 -T 1 -d 10m -q
Hardware-related Noise
duration: 0 00:10:00 | time is in us
CPU Period Runtime Noise % CPU Aval Max Noise Max Single HW NMI
1 #599 599000000 138 99.99997 3 3 4 74
2 #599 599000000 85 99.99998 3 3 4 75
3 #599 599000000 86 99.99998 4 3 6 75
4 #599 599000000 81 99.99998 4 4 2 75
5 #599 599000000 85 99.99998 2 2 2 75
Link: https://lkml.kernel.org/r/2d6f49a6f3a4f8b51b2c806458b1cff71ad4d014.1675805361.git.bristot@kernel.org
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Cc: Daniel Bristot de Oliveira <bristot@kernel.org>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: Clark Williams <williams@redhat.com>
Cc: Bagas Sanjaya <bagasdotme@gmail.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Currently, timerlat top displays the timerlat tracer latency results, saving
the intuitive timerlat trace for the developer to analyze.
This patch goes a step forward in the automaton of the scheduling latency
analysis by providing a summary of the root cause of a latency higher than
the passed "stop tracing" parameter if the trace stops.
The output is intuitive enough for non-expert users to have a general idea
of the root cause by looking at each factor's contribution percentage while
keeping the technical detail in the output for more expert users to start
an in dept debug or to correlate a root cause with an existing one.
The terminology is in line with recent industry and academic publications
to facilitate the understanding of both audiences.
Here is one example of tool output:
----------------------------------------- %< -----------------------------------------------------
# taskset -c 0 timerlat -a 40 -c 1-23 -q
Timer Latency
0 00:00:12 | IRQ Timer Latency (us) | Thread Timer Latency (us)
CPU COUNT | cur min avg max | cur min avg max
1 #12322 | 0 0 1 15 | 10 3 9 31
2 #12322 | 3 0 1 12 | 10 3 9 23
3 #12322 | 1 0 1 21 | 8 2 8 34
4 #12322 | 1 0 1 17 | 10 2 11 33
5 #12322 | 0 0 1 12 | 8 3 8 25
6 #12322 | 1 0 1 14 | 16 3 11 35
7 #12322 | 0 0 1 14 | 9 2 8 29
8 #12322 | 1 0 1 22 | 9 3 9 34
9 #12322 | 0 0 1 14 | 8 2 8 24
10 #12322 | 1 0 0 12 | 9 3 8 24
11 #12322 | 0 0 0 15 | 6 2 7 29
12 #12321 | 1 0 0 13 | 5 3 8 23
13 #12319 | 0 0 1 14 | 9 3 9 26
14 #12321 | 1 0 0 13 | 6 2 8 24
15 #12321 | 1 0 1 15 | 12 3 11 27
16 #12318 | 0 0 1 13 | 7 3 10 24
17 #12319 | 0 0 1 13 | 11 3 9 25
18 #12318 | 0 0 0 12 | 8 2 8 20
19 #12319 | 0 0 1 18 | 10 2 9 28
20 #12317 | 0 0 0 20 | 9 3 8 34
21 #12318 | 0 0 0 13 | 8 3 8 28
22 #12319 | 0 0 1 11 | 8 3 10 22
23 #12320 | 28 0 1 28 | 41 3 11 41
rtla timerlat hit stop tracing
## CPU 23 hit stop tracing, analyzing it ##
IRQ handler delay: 27.49 us (65.52 %)
IRQ latency: 28.13 us
Timerlat IRQ duration: 9.59 us (22.85 %)
Blocking thread: 3.79 us (9.03 %)
objtool:49256 3.79 us
Blocking thread stacktrace
-> timerlat_irq
-> __hrtimer_run_queues
-> hrtimer_interrupt
-> __sysvec_apic_timer_interrupt
-> sysvec_apic_timer_interrupt
-> asm_sysvec_apic_timer_interrupt
-> _raw_spin_unlock_irqrestore
-> cgroup_rstat_flush_locked
-> cgroup_rstat_flush_irqsafe
-> mem_cgroup_flush_stats
-> mem_cgroup_wb_stats
-> balance_dirty_pages
-> balance_dirty_pages_ratelimited_flags
-> btrfs_buffered_write
-> btrfs_do_write_iter
-> vfs_write
-> __x64_sys_pwrite64
-> do_syscall_64
-> entry_SYSCALL_64_after_hwframe
------------------------------------------------------------------------
Thread latency: 41.96 us (100%)
The system has exit from idle latency!
Max timerlat IRQ latency from idle: 17.48 us in cpu 4
Saving trace to timerlat_trace.txt
----------------------------------------- >% -----------------------------------------------------
In this case, the major factor was the delay suffered by the IRQ handler
that handles timerlat wakeup: 65.52 %. This can be caused by the
current thread masking interrupts, which can be seen in the blocking
thread stacktrace: the current thread (objtool:49256) disabled interrupts
via raw spin lock operations inside mem cgroup, while doing write
syscall in a btrfs file system.
A simple search for the function name on Google shows that this is
a legit case for disabling the interrupts:
cgroup: Use irqsave in cgroup_rstat_flush_locked()
lore.kernel.org/linux-mm/20220301122143.1521823-2-bigeasy@linutronix.de/
The output also prints other reasons for the latency root cause, such as:
- an IRQ that happened before the IRQ handler that caused delays
- The interference from NMI, IRQ, Softirq, and Threads
The details about how these factors affect the scheduling latency
can be found here:
https://bristot.me/demystifying-the-real-time-linux-latency/
Link: https://lkml.kernel.org/r/3d45f40e630317f51ac6d678e2d96d310e495729.1675179318.git.bristot@kernel.org
Cc: Daniel Bristot de Oliveira <bristot@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Currently, timerlat displays a summary of the timerlat tracer results
saving the trace if the system hits a stop condition.
While this represented a huge step forward, the root cause was not
that is accessible to non-expert users.
The auto-analysis fulfill this gap by parsing the trace timerlat runs,
printing an intuitive auto-analysis.
Link: https://lkml.kernel.org/r/1ee073822f6a2cbb33da0c817331d0d4045e837f.1675179318.git.bristot@kernel.org
Cc: Daniel Bristot de Oliveira <bristot@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Calculate average value in osnoise-hist summary with two-digit
precision to avoid displaying too optimitic results.
Link: https://lkml.kernel.org/r/20230103103400.275566-3-br015@umbiko.net
Signed-off-by: Andreas Ziegler <br015@umbiko.net>
Acked-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Sampled durations must be weighted by observed quantity, to arrive at a correct
average duration value.
Perform calculation of total duration by summing (duration * count).
Link: https://lkml.kernel.org/r/20230103103400.275566-2-br015@umbiko.net
Fixes: 829a6c0b56 ("rtla/osnoise: Add the hist mode")
Signed-off-by: Andreas Ziegler <br015@umbiko.net>
Acked-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
rtla_usage(), osnoise_usage() and timerlat_usage() all exit with an
error status.
However when these are called from help, they should exit with a
non-error status.
Fix this by passing the exit status to the functions.
Note, although we remove the subsequent call to exit after calling
usage, we leave it in at the end of a function to suppress the compiler
warning "control reaches end of a non-void function".
Link: https://lkml.kernel.org/r/20221107144313.22470-1-jkacur@redhat.com
Signed-off-by: John Kacur <jkacur@redhat.com>
Acked-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
To allow for distributions and other builders to apply hardening
policy and other customisation, append EXTRA_CFLAGS and EXTRA_LDFLAGS
to the corresponding variables.
Link: https://lore.kernel.org/linux-trace-devel/YtLBshz0nMQ7530H@decadent.org.uk
Signed-off-by: Ben Hutchings <benh@debian.org>
Acked-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
"ln -s" stores the next argument directly as the symlink target, so
it needs to be a relative path. In this case, just "rtla".
Link: https://lore.kernel.org/linux-trace-devel/YtLBXMI6Ui4HLIF1@decadent.org.uk
Fixes: 0605bf009f ("rtla: Add osnoise tool")
Fixes: a828cd18bc ("rtla: Add timerlat tool and timelart top mode")
Signed-off-by: Ben Hutchings <benh@debian.org>
Acked-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
RISC-V uses the same (generic) syscall numbers as ARM64.
Link: https://lkml.kernel.org/r/mvma68wl2ul.fsf@suse.de
Signed-off-by: Andreas Schwab <schwab@suse.de>
Acked-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Avoid double free by making trace_instance_destroy indempotent. When
trace_instance_init fails, it calls trace_instance_destroy, but its only
caller osnoise_destroy_tool calls it again.
Link: https://lkml.kernel.org/r/mvmilnlkyzx.fsf_-_@suse.de
Fixes: 0605bf009f ("rtla: Add osnoise tool")
Signed-off-by: Andreas Schwab <schwab@suse.de>
Acked-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Sedat Dilek reported an error on rtla Makefile when running:
$ make -C tools/ clean
[...]
make[2]: Entering directory
'/home/dileks/src/linux-kernel/git/tools/tracing/rtla'
[...]
'/home/dileks/src/linux-kernel/git/Documentation/tools/rtla'
/bin/sh: 1: test: rtla-make[2]:: unexpected operator <------ The problem
rm: cannot remove '/home/dileks/src/linux-kernel/git': Is a directory
make[2]: *** [Makefile:120: clean] Error 1
make[2]: Leaving directory
This occurred because the rtla calls kernel's Makefile to get the
version in silence mode, e.g.,
$ make -sC ../../.. kernelversion
5.19.0-rc4
But the -s is being ignored when rtla's makefile is called indirectly,
so the output looks like this:
$ make -C ../../.. kernelversion
make: Entering directory '/root/linux'
5.19.0-rc4
make: Leaving directory '/root/linux'
Using 'grep -v make' avoids this problem, e.g.,
$ make -C ../../.. kernelversion | grep -v make
5.19.0-rc4
Thus, add | grep -v make.
Link: https://lkml.kernel.org/r/870c02d4d97a921f02a31fa3b229fc549af61a20.1657747763.git.bristot@kernel.org
Fixes: 8619e32825 ("rtla: Follow kernel version")
Reported-by: Sedat Dilek <sedat.dilek@gmail.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Replace malloc with calloc and add memory allocating check
of mon_cpus before used.
Link: https://lkml.kernel.org/r/20220615073348.6891-1-jianchunfu@cmss.chinamobile.com
Fixes: 7d0dc9576d ("rtla/timerlat: Add --dma-latency option")
Signed-off-by: jianchunfu <jianchunfu@cmss.chinamobile.com>
Acked-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
