aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAge
* parse-events: Add strim() to allow whitespace in filter event namesSteven Rostedt2010-02-16
| | | | | | | Adde strim() to remove leading and trailing whitespace from the event and system names being passed into the filter. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Fix comma delimited filtersSteven Rostedt2010-02-16
| | | | | | | | The comma deliminating event names was not being processed correctly. The index pointer was stuck at the comma instead of the next event name and this caused the filtering to fail. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Change report filters to allow regex on event namesSteven Rostedt2010-02-16
| | | | | | | | | | | | | | | | | Use the regex to find events and systems. Since we are now using regular expresions to add events to filters, we can no longer use the '.' to separate the system name from the event name. That is: trace-cmd report -F 'sched.sched_switch' will no longer work. It needs to be: trace-cmd report -F 'sched/sched_switch' Which is a better naming since that is how the files are distinguished anyhow. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Free filters at and of reportSteven Rostedt2010-02-16
| | | | | | Need to call pevent_filter_free() when finished with reporting. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-event: Fix bug where sysname was defined but name was notSteven Rostedt2010-02-15
| | | | | | | If no name is given to the filter, then use the system name as the name and make the system name NULL. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Make filter add just events without filterSteven Rostedt2010-02-15
| | | | | | | Let the filter string contain just events. This will make all records that have this event match. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* build: Add automatically generated warning to generated header filesSteven Rostedt2010-02-15
| | | | | | | | The header files tc_version.h and ks_version.h are created by the build process. Add a comment warning that changes to these files will not be sustained. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Add event filtering on event fieldsSteven Rostedt2010-02-15
| | | | | | | | | | | | | | | Add the ability to filter on event fields. This patch implements the filter on trace-cmd report. Using the following: trace-cmd report -F 'sched.sched_switch:next_prio < 100 || (prev_comm != "swapper" && next_comm != "swapper")' Will return a report where prios are under 100 (Real-time) and does not include any context switch with the swapper task. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Fix reading of !! tokenSteven Rostedt2010-02-15
| | | | | | | | | The !! string needs to be returned as two different tokens. Break out of the token parser when one is encountered. Also break out on '==' and '!=', Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Relicense the library files under LGPLSteven Rostedt2010-02-15
| | | | | | | | | | | | | | | | | | | | All the files that are used to create the libraries: libparsevent and libtracecmd are converted to the LGPL as well as the files to create the plugins. All files now have a boilerplate header that states which license that the file is under. A README file is created as well as the COPYING.LIB which contains the text of the LGPL. All authors of the code that created these files have given their Acks. Acked-by: Darren Hart <dvhltc@us.ibm.com> Acked-by: Frederic Weisbecker <fweisbec@gmail.com> Acked-by: Josh Triplett <josh@joshtriplett.org> Acked-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Replace the event list with a sorted arraySteven Rostedt2010-02-12
| | | | | | | | | | | | | | | | | | | | | | Replace the event list in the pevent structure with a sorted array of event_format pointers. Now the search for an event can be done with a binary search. With the size of events growing, this makes a big difference. Running trace-cmd report on a 300,000 event data file and outputing into /dev/null, went from: real 0m5.639s user 0m5.464s sys 0m0.174s to real 0m1.810s user 0m1.672s sys 0m0.138s Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Free old plot info before initializing new plot infoSteven Rostedt2010-02-12
| | | | | | | | The order of freeing the old plot info and initializing the new plot info was backwards, and caused loading a new graph to output tons of errors about gc being NULL. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Make input buffer const char *Steven Rostedt2010-02-12
| | | | | | | | | | | Make the input buffer that the parse events reads into a const char *. This will be needed by later patches to make sure the input buffer does not get modified. In doing this change, the work around by Johannes Berg needed to be rewritten, since the push_str modified the input buffer. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Move free of record out of show_data in trace-cmd readSteven Rostedt2010-02-12
| | | | | | | There is no reason that the record that is passed to show_data should be freed by show_data. Free it from the location that allocated it. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Move functions into separate routineSteven Rostedt2010-02-12
| | | | | | | | Move the function like macros __print_symbolic() and friends into their own routine. This will allow adding parsing to other types of functions easier. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Clean up function graph pluginSteven Rostedt2010-02-12
| | | | | | | | | | | Because we now have reference counting of records, we do not need to jump through the hoops of allocating previous records' data to use it after reading new records. Remove the hacks from the function graph plugin that does all this. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add locked attribute to record and debug peek freeingSteven Rostedt2010-02-12
| | | | | | | | | | | Add a "locked" attribute to the record structure. This will be set when a peek is returned and cleared on read. This will help debug cases that incorrectly free a record returned by peek. The internal helper function free_next() is added to do the freeing of the next record used by peek. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Fix trace-read to do a read not a peekSteven Rostedt2010-02-12
| | | | | | | A previous change broke the cpu filtering and had the trace-read return a peek and not a read. This would later be freed incorrectly. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add tracecmd_get_cursor()Steven Rostedt2010-02-12
| | | | | | | Add the helper routine tracecmd_get_cursor() that returns the offset of the next tracecmd_read_data() or tracecmd_peek_data(). Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Free record in task plot displaySteven Rostedt2010-02-11
| | | | | | | The record allocating in the displaying of task info must always be freed. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* kernelshark/trace-graph: Free up graph infoSteven Rostedt2010-02-11
| | | | | | | Have kernelshark call the graph info destructor. Have the graph info destructor free the lists. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Initialize time in task plot displaySteven Rostedt2010-02-11
| | | | | | | The task plot display did not always initialize the time that was displayed. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Fix uninitialized variables in CPU plot codeSteven Rostedt2010-02-11
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Make the plots closerSteven Rostedt2010-02-11
| | | | | | | The plots are quite far apart for no apparent reason. Bring them 30 pixels closer. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add target CPU for sched_wakeup in sched pluginSteven Rostedt2010-02-11
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* kernelshark: Add checkbox to make graph follow list cursorSteven Rostedt2010-02-11
| | | | | | | Add a check box, that when set, the graph will follow the cursor of the list. This makes moving the graph with searches nicer. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Limit task plot prev search by count not timeSteven Rostedt2010-02-11
| | | | | | | Since a time slot can hold thousands or just 1 event, search backwards in the task plot and limit it by 20 events not time by pixels. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Display task specific data when zoomed in on task plotSteven Rostedt2010-02-10
| | | | | | | | | When zooming in on a task plot and the events show up in the graph, instead of showing the CPU type data, show information useful for the task. This includes what CPU it is on, and for sched_switch, if a wakeup was detected, it shows the latency from the wakeup. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Limit the search back for task plots display dataSteven Rostedt2010-02-10
| | | | | | | | | | | The task plot was search every CPU for a matching event. If a CPU did not have any events for a given plot, it would search all the way back to the start! This caused a huge delay in processing and would hang trace-graph for seconds while it would search the cpu data backwards. Limit the distance to only 5 pixels in its search. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Remove call to set_cpus_to_time() before find_record()Steven Rostedt2010-02-10
| | | | | | | | The set_cpus_to_time() can be expensive, and find_record() calls it directly. No need to double the time to find the next record for a task plot. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Show wake latency in task plotsSteven Rostedt2010-02-10
| | | | | | | | | Add the ability for a plot to draw a empty box. Then with this ability, change the task plots to show a hollow red box from the time the task was woken, till the time it was scheduled. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Have the graph control the start stop recordsSteven Rostedt2010-02-10
| | | | | | | | | | Move the code from initializing the start cursor locations of the CPUs. Also control when a record is passed to the plot based on its timestamp. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add tracecmd_set_all_cpus_to_timestamp()Steven Rostedt2010-02-10
| | | | | | | Add the helper routine tracecmd_set_all_cpus_to_timestamp() to be used to initialize all CPUs to a given time. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Resize instead of redraw when adding plotsSteven Rostedt2010-02-10
| | | | | | | When adding a new plot, a resize is required since the size of the graph has changed. Not just redrawing the graph. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Insert new task plots below the plot that they were selected onSteven Rostedt2010-02-10
| | | | | | | When selecting to plot a task, insert the plot beneath the plot that it was selected on. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Show comm name for task plotsSteven Rostedt2010-02-10
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Resize graph height when plot is added or removedSteven Rostedt2010-02-10
| | | | | | | | | The graph height was only calculated when the graph was loaded. Now that you can add new plots, we need to recalculate the height anytime a plot is added or removed. Reported-by: Darren Hart <dvhltc@us.ibm.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Fix task plots to print events other than wake ups and schedsSteven Rostedt2010-02-10
| | | | | | | A previous fix to fix sched_switch events broke printing non sched switch events. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Iterate over all records for all plotsSteven Rostedt2010-02-10
| | | | | | | | Change the code to iterate once down all records to create all plots. Use the cpu_hash, task_hash and all_recs to find what plot cares about what data. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Clean up plots when reloading a new trace.datSteven Rostedt2010-02-10
| | | | | | | Need to destroy the the plots with the full destructor when removing them before loading a new trace.dat file. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Add plot_list to hold the plots on a hashSteven Rostedt2010-02-10
| | | | | | | Add an intermediate list item for hashs to let a plot be assigned to more that one hash. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Add hash to lookup pids and cpusSteven Rostedt2010-02-10
| | | | | | | | Add a hash in the graph info to quickly look up to see if plots have been registered to a pid or cpus. This is currently not used but will be in the following patches. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Add destructor routine to plot callbacksSteven Rostedt2010-02-10
| | | | | | | | | | Added a destructor routine to the plot callbacks to be called before the plot is destroyed. This allows the plot to clean up any resources that it allocated before it is freed. Also did some other minor clean ups and freeing. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Pass plot around instead of indexSteven Rostedt2010-02-10
| | | | | | | | | | Use the plot itself to pass to routines instead of the index into the ginfo->plot_array of the plot. This requires that the plot keeps track of its own position in that array, and the trace-plot.c code must update that position when plots are added or removed from the middle of the array. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Remove trace plot listSteven Rostedt2010-02-10
| | | | | | | | Remove the list structure for plots. The plots are stored on an array anyway, and the managing them on a list just adds overhead and maintenance without any added benefit. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Add popup menu to remove plotSteven Rostedt2010-02-10
| | | | | | | | | | | | Add a way to remove a plot. Todo: Add warning to remove plot Add ways to get back CPU plots Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Add way to add task plotsSteven Rostedt2010-02-10
| | | | | | | | | | | | Add a pop up window that lets you trace a given task. TODO: Prevent duplicate plots Add way to remove plots Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Add trace-plot-task.cSteven Rostedt2010-02-10
| | | | | | | Create the code to plot a task. Currently nothing calls it. But that will change in later patches when we collect all task pids. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add tracecmd_read_prev()Steven Rostedt2010-02-10
| | | | | | | Add the tracecmd_read_prev(handle, record) that returns the record before the given record. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add calc_page_offset() helper routineSteven Rostedt2010-02-10
| | | | | | | | | | Use calc_page_offset() to performe the: offset & (handle->page_size - 1) to prevent any typos and bugs. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
generated by cgit v1.2.2 (git 2.25.0) at 2025-04-13 16:39:57 -0400