aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAge
* Add GTK viewer for reading trace dataSteven Rostedt2009-12-17
| | | | | | This is the start of adding a GTK front end to read the trace data. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add ability to add new comms after first useSteven Rostedt2009-12-17
| | | | | | | | | | | | | To speed up the reading of cmdlines to pid mappings, an unsorted list is used first to register comms (cmdlines). Once the list is used to seach, the list is converted to a sorted array to allow for fast binary searches. But once this was done, there was no way to add more cmdlines mappings. This now adds a slower version to add new cmdlines after the list has been made, but still makes it possible. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add pevent_pid_is_registered()Steven Rostedt2009-12-17
| | | | | | | Add the function pevent_pid_is_registered() that returns 1 if the given pid has a cmdline mapped to it, and 0 otherwise. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Use python-config scriptSteven Rostedt2009-12-17
| | | | | | | | Need to use the python-config script instead of having hard coded includes. Reported-by: Darren Hart <dvhltc@us.ibm.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Start of a tracecmd swig wrapper for pythonDarren Hart2009-12-17
| | | | | | | | | | | | | | | | | | | | | | | | | | | Introduce a python ctracecmd module for use in rapidly prototyping tracing applications. The interface description is provided in ctracecmd.i, it identifies which functions are available from within python. A test python script is provided as tracecmd-test.py. These bindings are expected to change significantly. Eventually I would like to wrap this automated binding with more pythonic objects, most likely including Trace and Event objects which merge the functionality of tracecmd-input, pevent, record, and event structures. This will make development of python apps much more accessible to many application developers. For now, this is mostly a proof of concept and is no where near complete. It can however open a trace file and read all the events from it, displaying them by CPU in chronological order. V2: o Simplified interface file with SWIG ifdefs in the header files V3: o Move attribute removal to interface file o Remove proxy classes and rename module to ctracecmd o Use the Makefile with a phony python target instead of swig.sh Signed-off-by: Darren Hart <dvhltc@us.ibm.com> LKML-Reference: <4B2A7878.6060209@us.ibm.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Change tracecmd_open() to take file nameSteven Rostedt2009-12-17
| | | | | | | | | | | Rename tracecmd_open() to tracecmd_open_fd and add a new tracecmd_open() to take the name of a file. Also added the API for tracecmd_close() but still need to do proper clean up. It just frees the handle, but does nothing with everything else it allocated. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Remove non lib externs from trace-cmd.hSteven Rostedt2009-12-17
| | | | | | | Clean up, to get rid of leftover externs in the trace-cmd.h header file. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Use opaque record type in pevent accessor functionsDarren Hart2009-12-17
| | | | | | | | | | | | | | | | | | The caller of the pevent accessor functions: pevent_data_pid(pevent, data) for example Already have a struct record. Let them send that directly: pevent_data_pid(pevent, record) This decouples the API from the struct implementation and facilitates language bindings which use opaque pointers for passing the C structs around. Signed-off-by: Darren Hart <dvhltc@us.ibm.com> LKML-Reference: <4B297C79.2090004@us.ibm.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add negating of eventsSteven Rostedt2009-12-15
| | | | | | | | | | | Add the '-v' option to disable events. This is advantageous with: trace-cmd record -e all -v -e lock -e syscalls This will enable all events and then disable lock and syscall events before executing the command. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Show ring buffer stastics at the end of tracingSteven Rostedt2009-12-15
| | | | | | | When the tracer is completed, show the ring buffer statistics to give a good idea about overruns and such. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Reset page at ever read by tracecmd_read_at()Steven Rostedt2009-12-15
| | | | | | | | The timestamp was getting messed up by various reads of tracecmd_read_at(). This resets the page everytime that function is used. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add sleep_time between splice readsSteven Rostedt2009-12-15
| | | | | | | | | | | | I found that tracing few events causes trace-cmd to almost spin in a busy loop since it can constantly read a page, and the process of reading that page causes it to run again. We end up just causing events to read. This adds a -s sleep time option that forces trace-cmd to sleep. The default is 1ms. But it can be disabled with 0. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Add stdlib.h to trace-cmd.h for use of free()Steven Rostedt2009-12-15
| | | | | | | The free_record inline function uses free, and requires the stdlib.h header. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Added comments to the parse-event librarySteven Rostedt2009-12-15
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Add free_record() to use to free records readSteven Rostedt2009-12-15
| | | | | | | | | This adds a wrapper for freeing the records. It also fixes a comment in tracecmd_peek_data, where the record returned must not be freed. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Comment the tracecmd_input functionsSteven Rostedt2009-12-15
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Fix invalid write due to cpus and cpu_count confusionDarren Hart2009-12-15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Fix invalid write due to cpus and cpu_count confusion trace-cmd would fail with: # ./trace-cmd record -e sched ls -ltr enable sched cpus: 8 cpu_count: 0 *** glibc detected *** ./trace-cmd: free(): invalid next size (normal): 0x0000000000e760b0 *** ======= Backtrace: ========= /lib64/libc.so.6[0x3f18671ce2] /lib64/libc.so.6(cfree+0x8c)[0x3f1867590c] /lib64/libc.so.6(fclose+0x14b)[0x3f18660d0b] ./trace-cmd[0x40397e] ./trace-cmd(main+0x7df)[0x404777] /lib64/libc.so.6(__libc_start_main+0xf4)[0x3f1861d974] ./trace-cmd[0x4029f9] ======= Memory map: ======== 00400000-00418000 r-xp 00000000 08:04 1922384 /test/dvhart/source/trace-cmd.git/trace-cmd 00617000-00618000 rw-p 00017000 08:04 1922384 /test/dvhart/source/trace-cmd.git/trace-cmd 00e76000-00e97000 rw-p 00000000 00:00 0 [heap] 3f18200000-3f1821c000 r-xp 00000000 08:03 327334 /lib64/ld-2.5.so 3f1841b000-3f1841c000 r--p 0001b000 08:03 327334 /lib64/ld-2.5.so 3f1841c000-3f1841d000 rw-p 0001c000 08:03 327334 /lib64/ld-2.5.so 3f18600000-3f1874c000 r-xp 00000000 08:03 327335 /lib64/libc-2.5.so 3f1874c000-3f1894c000 ---p 0014c000 08:03 327335 /lib64/libc-2.5.so 3f1894c000-3f18950000 r--p 0014c000 08:03 327335 /lib64/libc-2.5.so 3f18950000-3f18951000 rw-p 00150000 08:03 327335 /lib64/libc-2.5.so 3f18951000-3f18956000 rw-p 00000000 00:00 0 3f18a00000-3f18a02000 r-xp 00000000 08:03 327341 /lib64/libdl-2.5.so 3f18a02000-3f18c02000 ---p 00002000 08:03 327341 /lib64/libdl-2.5.so 3f18c02000-3f18c03000 r--p 00002000 08:03 327341 /lib64/libdl-2.5.so 3f18c03000-3f18c04000 rw-p 00003000 08:03 327341 /lib64/libdl-2.5.so 3f19a00000-3f19a0d000 r-xp 00000000 08:03 327350 /lib64/libgcc_s-4.1.2-20080825.so.1 3f19a0d000-3f19c0d000 ---p 0000d000 08:03 327350 /lib64/libgcc_s-4.1.2-20080825.so.1 3f19c0d000-3f19c0e000 rw-p 0000d000 08:03 327350 /lib64/libgcc_s-4.1.2-20080825.so.1 7f4ef8000000-7f4ef8021000 rw-p 00000000 00:00 0 7f4ef8021000-7f4efc000000 ---p 00000000 00:00 0 7f4effbea000-7f4effbec000 rw-p 00000000 00:00 0 7f4effc00000-7f4effc03000 rw-p 00000000 00:00 0 7ffffb0c5000-7ffffb0da000 rw-p 00000000 00:00 0 [stack] 7ffffb1ff000-7ffffb200000 r-xp 00000000 00:00 0 [vdso] ffffffffff600000-ffffffffff601000 r-xp 00000000 00:00 0 [vsyscall] Aborted The cpus and cpu_count line above I added to understand the ambiguity of those variables. The cpus variable appears redundant. This patch uses the global cpu_count directly. If cpu_count should not be updated until later for some reason, then the code could be updated to use cpus instead. The way it was however tries to write to pids[] which has a size of 0. Signed-off-by: Darren Hart <dvhltc@us.ibm.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Set to first record in tracecmd_set_cpu_to_timestamp()Steven Rostedt2009-12-15
| | | | | | | | | The algorithm in tracecmd_set_cpu_to_timestamp() depends on the timestamp of the page being for the page. With reading data off the page, the timestamp gets incremented. This resets the timestamp back to the first record on the current page. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Simplify free_page()Steven Rostedt2009-12-15
| | | | | | | | Both read and mmap do the same thing except the mmap version needs to unmap the page on free. The condition should hold the mmap version. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Have get_page() handle size and indexSteven Rostedt2009-12-15
| | | | | | | get_page() already manages the size field of cpu_data, it should also reset index. This does not need to be done in get_next_page(). Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Update timestamp with offsetSteven Rostedt2009-12-15
| | | | | | | | If the offset of the page is modified, reset the timestamp as well. The tracecmd_set_cpu_to_timestamp depends on the timestamp being in sync with the offset. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Remove duplicate setting of offsetSteven Rostedt2009-12-15
| | | | | | | The offset was set at the beginning and again during updating of the page in get_page. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Disable first / last testingSteven Rostedt2009-12-15
| | | | | | The test for first and last element was left on. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Prevent an infinite loop in searching by timestampSteven Rostedt2009-12-15
| | | | | | | | | | If the search for a record comes down to being on a page where the timestamp is greater than the page but less then the next page to look for, the code will go into an infinite loop. This is because the next page to look for will be the same as the starting page, and it will never continue to move forward. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Add tracecmd_read_cpu_first() and last()Steven Rostedt2009-12-14
| | | | | | Add a interface to get the first and last item per CPU. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* The reading of page info belongs in the mapping of the pageSteven Rostedt2009-12-14
| | | | | | | We need to read the page info when a page is mapped, not when the first element is read. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Move writing to data file to librarySteven Rostedt2009-12-14
| | | | | | | | We need a way to have other applications easily make the output file. This moves the recording of the trace.dat file to the trace-cmd library. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Fix do_read() in case a multiple read is neededSteven Rostedt2009-12-13
| | | | | | | Fix a bug where we tried to read more than needed if a second read is needed. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Add tracecmd_set_cpu_to_timestamp()Steven Rostedt2009-12-12
| | | | | | | | | This adds an API to set the CPU data field to the location specified by a timestamp. It looks for the page that contains or is before the timestamp to allow for walking forward to find the specified event. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* squash meSteven Rostedt2009-12-12
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Move page calculations to get_pageSteven Rostedt2009-12-12
| | | | | | | The get_page code shoud calculate the page calculations of the cpu index, offset and size. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Add test to trace-cmd command line to test tracecmd_read_atSteven Rostedt2009-12-12
| | | | | | | | The trace-cmd command line does not use tracecmd_read_at, but it still needs to be tested. This allows easy access to test the tracecmd_read_at routine. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Use get_page over get_next_page for tracecmd_read_atSteven Rostedt2009-12-12
| | | | | | | | | The capturing of a page with tracecmd_read_at was a hack that used get_next_page by forcing the index to the previous page. Now that we have a more generic get_page, this patch changes tracecmd_read_at to use that. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Add generic mmap funcion to read page from cpu dataSteven Rostedt2009-12-12
| | | | | | | | This patch adds a generic mmap page reading that simply swaps the old mmap with a new mapping. This avoids needing to use get_next_page in an awkward and error prone fashion. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Have trace-cmd record not need a commandSteven Rostedt2009-12-11
| | | | | | | | Currently trace-cmd record requires a command to exec. This patch will just record what is happening if no command is given and will stop when the user hits Ctrl^C. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Have git ignore emacs backup filesSteven Rostedt2009-12-10
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Add sched_switch plugin to make the events look more like the old pluginSteven Rostedt2009-12-10
| | | | | | | | The event output for scheduler context switches and wakeups changed the output format. This adds a plugin to make those changes look more like the sched_switch plugin. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Let trace-cmd start disable function tracingSteven Rostedt2009-12-10
| | | | | | | The disabling of the function tracer was hidden in the trace-cmd record branch. Move it to be used by both record and start. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Add override handler to handle kernel stack trace entrySteven Rostedt2009-12-10
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Do not record if no plugin or event was specifiedSteven Rostedt2009-12-09
| | | | | | | | | An new user may just try doing a record and see what happens, and will be surprised when there is no output to show. This patch adds an error message if no event or plugin is specified. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Add negative and subtraction number evaluationSteven Rostedt2009-12-09
| | | | | | | | If a constant has a negative number in processing symbols, the parser will fail. This patch adds handling of both negative numbers and subtraction. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Add command to show event formats from the data fileSteven Rostedt2009-12-09
| | | | | | | | Sometimes we need to see the event formats that are stored in a data file. This patch adds the options "--events" that prints out these formats. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Added trace-cmd start, stop and resetSteven Rostedt2009-12-09
| | | | | | | | | | | | | | Sometimes it is nice to just kick off the tracer without having to record it into a trace data file. Instead of having to always echo the commands into the tracing debugfs directory, this patch adds the trace-cmd start, which acts like the trace-cmd record, but does not start a command or record the trace via splice. It simply enables the tracer. trace-cmd stop - will stop tracing (echo 0 > tracing_on) trace-cmd reset - will reset the tracer (echo nop > current_tracer) Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Add work around for event output bug in older kernelsJohannes Berg2009-12-07
| | | | | | | | | | | | | | Older kernels have a bug that prints out the macro name instead of what it should represent. This breaks parsing the data from userspace. This especially affects the mac80211 trace events. This adds a work around in the tool to handle the bug when executed on older kernels. See Linux kernel commit 811cb50baf63461ce0bdb234927046131fc7fa8b for details. Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
* Rename tracecmd_handle to tracecmd_inputSteven Rostedt2009-12-04
| | | | | | | | | | | We will soon be adding write functionality to the tracecmd library. This will require a new and separate handle than what is used by the file reader. "handle" is too generic. Rename it to "input" so we can make the output handle called tracecmd_output. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Add handler to pevent functionsSteven Rostedt2009-12-04
| | | | | | | | | Currently pevent uses global variables to store the event types. This will eventually bite us since we may someday want to be able to read multiple files with different formats. This patch adds a handle to pevent, to keep it encapsulated. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Reset cpu data page offset when page is not mappedSteven Rostedt2009-12-02
| | | | | | | | | When we reach the end of the data file and unmap the page, we must also reset the page offset, otherwise the tracecmd_read_at gets confused, and may think that a page can be mapped when it is not. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Space out events more evenlySteven Rostedt2009-12-02
| | | | | | | The data for the events in the output looks better if most of them are lined up in a column. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Added trace_seq_terminateSteven Rostedt2009-12-02
| | | | | | | | Added trace_seq_terminate that zeros out the buffer if the buffer is not full. This is useful when data being returned may be used in a printf directly or more data will be added to the trace_seq. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Add way to retrieve all eventsSteven Rostedt2009-12-01
| | | | | | | This adds the function pevent_list_events that returns a list of event pointers. One can pass in a sort algorithm too. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
generated by cgit v1.2.2 (git 2.25.0) at 2025-04-11 22:13:58 -0400