aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAge
* mac80211 plugin: fix flag parsing errorJohannes Berg2010-04-07
| | | | | | | | | | | This fixes an error in the mac80211 plugin when it parses flags, seems we never noticed because I ordered the flags properly. Also add the new QOS flag. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Fix error path from process_value() failingSteven Rostedt2010-04-06
| | | | | | | | | The parent arg in process_value() may not be changed on error. If we pass in an argument, then it needs to be initialized to NULL if we plan on freeing it on failure. Reported-by: Li Zefan <lizf@cn.fujitsu.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Fix parsing not algorithmSteven Rostedt2010-04-06
| | | | | | | | The process_not() filter parsing did not assign an arg to the parent arg, and the entire expression was ignored. Reported-by: Li Zefan <lizf@cn.fujitsu.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Explain "entries" in the buffer statisticsSteven Rostedt2010-04-06
| | | | | | | | | | | | Seeing in the Buffer statistics: entries: 0 Makes it look like nothing was recorded when the opposite is true. Add a note that explains what this value means. Reported-by: Li Zefan <lizf@cn.fujitsu.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Fix parenthesis parsing of filtersSteven Rostedt2010-04-01
| | | | | | | | | | | | | The parenthesis filter parsing was a bit broken and needed a rewrite. This redesigns the filter parsing to be a bit more sane when it comes to parenthesis. It lets the parenthesis be treated as what they resolve to be. Funny, the example used by kernelshark failed because of this bug: "next_prio < 100 && (prev_prio > 100 && prev_pid != 0)" Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Show lost events if buffer supports itSteven Rostedt2010-04-01
| | | | | | | | | | | If the ring buffer in the kernel sends information via the commit field of the sub buffer header. Than this patch will display where events were lost. If the sub buffer has space and can store the count of dropped events, this patch will show the count as well. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Remove extra page copy in trace-cmd split -pSteven Rostedt2010-03-31
| | | | | | | | | | | | | | | | The while loop that copies pages from one dat file ot the next breaks after it copies the last page that it is expected to. But after the while loop (for other split types) there is also a checkt if a page is not finished and should be copied. But this always copies the page again with the page count option, and causes a duplict page at the end of the trace. The fix is simply to break out of the loop before copying the data when the page count is hit. Then the copy after the while loop will copy the last page needed. Reported-by: Li Zefan <lizf@cn.fujitsu.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Make sure trace-cmd split start and end values are floatsSteven Rostedt2010-03-31
| | | | | | | | | | | | | | | It is a common mistake to do: trace-cmd split my-trace.dat Expecting to be splitting the file my-trace.dat. But this is taken as the starting value and the user will not know that the command ran the split against trace.dat (if it exists). This patch checks to make sure that the start and end values passed in are indeed floating point (or integer) values. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Always append '.1' to output file in trace-cmd splitSteven Rostedt2010-03-31
| | | | | | | | | If a output file is specified, always append a '.1' to it. This removes the confusion of where the output file is written to and it could cause data to be overwritten that was not expected. Reported-by: Li Zefan <lizf@cn.fujitsu.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Fix temp files in trace-cmd split to include directoriesSteven Rostedt2010-03-31
| | | | | | | | | If a directory is specified in the output file for trace-cmd split, then it would fault because of the adding of the prefix ".tmp" to the output name. This patch fixes the issues by parsing out the directory name and basename from the output path. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Fix typo in connect()Steven Rostedt2010-03-31
| | | | | | | The loop after getaddrinfo() needs to test connect() against -1, not 1. Reported-by: Li Zefan <lizf@cn.fujitsu.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Terminate trace-cmd listen nicelySteven Rostedt2010-03-29
| | | | | | | | | | | When killing the trace-cmd listen, it will kill all the listeners and leave the temporary per CPU data files around. This patch adds handlers (also replacing signal() with sigaction()) to allow a Ctrl^C or kill pid, to shutdown the recording nicely and create the final file before exiting. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Remove left over debug in trace-cmd listenSteven Rostedt2010-03-29
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add -l option to trace-cmd listen for log file outputSteven Rostedt2010-03-29
| | | | | | | Add a way to pipe log info to another file, especially useful in daemon mode. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Have listen clean up children with signalsSteven Rostedt2010-03-29
| | | | | | | | Clean up the children that are created by a client with signals instead of leaving them around as zombies until another connection is made. This is a much cleaner solution. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Implement daemon on listenSteven Rostedt2010-03-29
| | | | | | | The daemon was not implement on trace-cmd listen. This implements it by simply calling daemon(). Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add --debug option to trace-cmd listenSteven Rostedt2010-03-29
| | | | | | | Add the option --debug to keep trace-cmd from forking children on connections. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Make listen connection run as a separate processSteven Rostedt2010-03-29
| | | | | | Fork off children when a process is connected, and run continuously. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Move to output dir specified by -d in listen commandSteven Rostedt2010-03-29
| | | | | | | | If -d is specified on the command line for 'trace-cmd listen' then change to that directory to write the data files to. Reported-by: Li Zefan <lizf@cn.fujitsu.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Documentation: Added trace-cmd.dat.5.txtSteven Rostedt2010-03-26
| | | | | | Added a documentation man page on the format of the trace.dat file. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* build: Make the man pages build number dependentSteven Rostedt2010-03-26
| | | | | | | To make way for creating other numbered man pages, modify the make documentation build to be number specific. This required adding the numbers to the text files. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* build: Added installing of documentationSteven Rostedt2010-03-26
| | | | | | Added 'make install_doc' Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* build: Hide the annoying output that xsltproc writes when completedSteven Rostedt2010-03-25
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* build: Prevent rebuild of documentation every timeSteven Rostedt2010-03-25
| | | | | | | | | | Using dependencies to trigger the manpage.xsl waring was causing all man pages to be built every time, event when no files were changed. By converting it to a funciton and calling it first, it will allow the man pages to only be rebuilt if their source files were changed. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* build: Make the building of documentation look cleanerSteven Rostedt2010-03-25
| | | | | | Use the "KBUILD" method to build documentation. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* build: Make doc build handle O=/path/to/output/dirSteven Rostedt2010-03-25
| | | | | | Update the Documentation Makefile to handle building outside the source tree. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Update documentation to match new trace-cmd report optionsSteven Rostedt2010-03-25
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Print raw fields of type long in hexSteven Rostedt2010-03-25
| | | | | | | | When raw fields are printing, check if the field is of type long. Since longs usually store pointers, it makes more sense to print out a hex number for a field declared as long than decimal. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add raw output format to trace-cmd reportSteven Rostedt2010-03-25
| | | | | | | Add '-r' option that prints the events by their field content and ignores the print format of the event. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Add raw format for printing eventsSteven Rostedt2010-03-25
| | | | | | | | | | | | | | | | | Add print_raw field to pevent that will cause the pretty_print function to output the entries by their raw data in the fields, and will ignore the print formats. When set, each field will show its data. So instead of this: sched_switch: task swapper:0 [140] (R) ==> trace-cmd:31807 [140] We have: sched_switch: prev_comm=swapper prev_pid=0 prev_prio=140 prev_state=0 next_comm=trace-cmd next_pid=31807 next_prio=140 Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add disabling of plugins or trace-cmd reportSteven Rostedt2010-03-25
| | | | | | | Add the option -L to disable system plugins and only load local ones. Add the option -N to disable all plugins. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add another backslash in record helpSteven Rostedt2010-03-25
| | | | | | | The help for record has one backslash with the options going to the next line, but it does not have another one for the second line the options run off to. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Clean up trace-cmd usage outputSteven Rostedt2010-03-25
| | | | | | | | | | | | | | | | The usage just printed all commands and their full descriptions. This is too much for all the commands that trace-cmd does. This patch moves the usage out of trace-cmd.c and into a new file called trace-usage.c. It then makes a structure that holds the name of the command, a short description and the full usage (from the original usage). If no command, or the command passed to trace-cmd is not found, then the list of commands and their short usages are shown. If a command is passed in, only the usage for that command is shown. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Print field contents of failed parsed eventSteven Rostedt2010-03-25
| | | | | | | | | | | If an event fails to parse the print string, then it is better to show the contents as raw elements than just to print "FAILED TO PARSE" and drop the event. At least now the events show a bit more information, even if the print string did not parse correctly. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Fix field_is_string()Steven Rostedt2010-03-25
| | | | | | | | | | | | | | The field_is_string() function always returns true because it checks for !strstr(...), which returns 0 if it does not match. Thus the !strstr("char") || !strstr("u8") || !strstr("s8") One would fail and the entire code would end up being true. Also move the field_is_dynamic out of the field_is_string() condition because a field may be dynamic without being a string. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Fix dynamic flag settingSteven Rostedt2010-03-25
| | | | | | | | | | The dynamic flag was not being set because the compare in the field type was looking for an exact match of "__data_loc", when in reality, the field type also contained the array type as well (ie. "__data_loc int"). Just check the first 10 characters of the type. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* build: Have install print nicerSteven Rostedt2010-03-25
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Change the output of installing from: install -d -m 755 '/usr/local/share/trace-cmd/plugins' install plugin_hrtimer.so plugin_kmem.so plugin_sched_switch.so plugin_mac80211.so plugin_jbd2.so '/usr/local/share/trace-cmd/plugins' install -d -m 755 '/usr/local/bin' install trace-cmd '/usr/local/bin' gui build complete install -d -m 755 '/usr/local/bin' install trace-view '/usr/local/bin' install trace-graph '/usr/local/bin' install kernelshark '/usr/local/bin' to: INSTALL plugin_hrtimer.so to /usr/local/share/trace-cmd/plugins INSTALL plugin_kmem.so to /usr/local/share/trace-cmd/plugins INSTALL plugin_sched_switch.so to /usr/local/share/trace-cmd/plugins INSTALL plugin_mac80211.so to /usr/local/share/trace-cmd/plugins INSTALL plugin_jbd2.so to /usr/local/share/trace-cmd/plugins INSTALL trace-cmd to /usr/local/bin INSTALL trace-view to /usr/local/bin INSTALL trace-graph to /usr/local/bin INSTALL kernelshark to /usr/local/bin Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Print port on error using %s not %dSteven Rostedt2010-03-25
| | | | | | | On error to connecting to a machine, the port is printed with a %d. But the variable is of type char * and a %s should be used. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add BUILDING and INSTALL instructions to READMEJohn Kacur2010-03-25
| | | | | | | - Add BUILDING and INSTALL instructions to the README - Change Makefile so that messages don't appear like errors. Signed-off-by: John Kacur <jkacur@redhat.com>
* trace-cmd: Install plugins by defaultJohn Kacur2010-03-24
| | | | | | If you a user does "make install", also install plugins. Signed-off-by: John Kacur <jkacur@redhat.com>
* trace-cmd: Make the default value of prefix /usr/localJohn Kacur2010-03-24
| | | | | | | | Make the default value of prefix /usr/local which makes sense for most end-users Developers who wish to use $HOME can set this on the command line. (use ?= which only sets the default if prefix doesn't already have a value) Signed-off-by: John Kacur <jkacur@redhat.com>
* trace-cmd: Remove DESTDIR_SQ from the makefileJohn Kacur2010-03-24
| | | | | | | DESTDIR_SQ must have been incompletely copied from another project's makefile. It isn't serving a purpose here, so remove it. Signed-off-by: John Kacur <jkacur@redhat.com>
* trace-cmd: Add trace_plugin_dir to .gitignoreJohn Kacur2010-03-24
| | | | | | File trace_plugin_dir is generated during make. Add it to the .gitignore file. Signed-off-by: John Kacur <jkacur@redhat.com>
* parse-events: Add function to handler that is postponedSteven Rostedt2010-03-24
| | | | | | | | This was a silly bug that the function to add to events from plugins never was stored in the temp structure. Reported-by: "Luis Claudio R. Goncalves" <lclaudio@uudg.org> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* build: Fix plugin and install for O=output makesSteven Rostedt2010-03-23
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Let function handlers be overriddenSteven Rostedt2010-03-23
| | | | | | | | | Now that plugins are in both system and user directories, the user directory is processed after the system directory. This makes it highly likely that a function handler may be added again. Instead of warning about it and failing let the new function handler override the previous one. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* build/trace-cmd: Add way to install system default pluginsSteven Rostedt2010-03-23
| | | | | | | | | | If the prefix in the make build is not $(HOME) then a system wide plugin directory is used (in $(prefix)/share/trace-cmd/plugins). This is then compiled into the tracecmd library so that it knows to look for the plugins there. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Handle operations within a conditionSteven Rostedt2010-03-23
| | | | | | | | The conditon ? : does not handle other operations within it (like << which the ext4 tracepoints added). This patch makes the conditions handle operations within them. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add options to trace.dat fileSteven Rostedt2010-03-23
| | | | | | | | Add the start of option to the trace.dat file. This will allow later versions of the file to add data without breaking older versions of trace-cmd. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Added jbd2 pluginSteven Rostedt2010-03-23
| | | | | | | Added the jbd2 plugin that adds the helper functions to handle jbd2_dev_to_name() and jiffies_to_msecs(). Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
generated by cgit v1.2.2 (git 2.25.0) at 2025-04-11 08:04:09 -0400