aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAge
...
* trace-view: Freeze window and wait cursor while doing searchSteven Rostedt2011-02-21
| | | | | | | | When doing a list search, it may take a while if there are a lot of events to search through. Freeze the window and change the cursor to the wait cursor while processing the search. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* kernelshark: Provide a way to freeze the main windowSteven Rostedt2011-02-21
| | | | | | | | | | There may be cases that we need to freeze the main window in order to provide updates to other widgets or the cursor while working to update something. Add trace_freeze_all() and trace_unfreeze_all() Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* kernelshark: Add methods to change the parent window guiSteven Rostedt2011-02-21
| | | | | | | Add trace_set_cursor() and trace_put_cursor() to set and unset the cursor of the parent window. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-view: Add methods to get the store pagesSteven Rostedt2011-02-21
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* kernelshark: Remove comment about function graph tracer in trace-compat.cSteven Rostedt2011-02-21
| | | | | | | | | The trace-compat.c comment header was copied from parse-events.c and included the comment about Frederic Weisbecker writing the code for the function graph tracer output. This has nothing to do with the trace-compat.c file. Remove it. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-view: Use popup for "Not found" display message in searchSteven Rostedt2011-02-21
| | | | | | | When a search is not found, use a pop up to show that answer instead of a printf. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-view: Add way to get any row in the treeSteven Rostedt2011-02-21
| | | | | | | | Currently there's only handlers to get visible rows from the tree. It may be needed to look at other rows from other pages (like in the search function). Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* kernelshark: Have trace_dialog() use parent_window if parent is NULLSteven Rostedt2011-02-21
| | | | | | | Some places in the code does not have access to a window. Default to the parent window if the parent is not supplied. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* kernelshark/trace-view: Fix searching list on later pagesSteven Rostedt2011-02-21
| | | | | | | | | | | Currently the search function only works on the first page of the list. The list is broken up into pages of 1,000,000 items. But if you are on the second or later page (>1,000,000 item) the search will not work. This patch fixes the search feature to work on other pages. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd/report: Add --kallsyms to insert function listSteven Rostedt2011-02-14
| | | | | | | | | If for some reason a trace.dat file is missing the function list (sometimes happens when creating a trace.dat from restore), add away to allow trace-cmd report to pull in a kallsyms file to use instead. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd/report: Fix switch statement for --filter-testSteven Rostedt2011-02-14
| | | | | | | | | | The long opt for the filter test "--filter-test" was not setting the filter-test variable. It was not even tested and would cause the help message to appear. Also add comments around the long op case statements. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Update record filtering to handle childrenSteven Rostedt2011-02-09
| | | | | | | | Now that record -F -c will follow the children, this patch implements the filters to be updated to include the children in the event filters. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Use ptrace to trace children of the processSteven Rostedt2011-02-09
| | | | | | | | | | | | | | | Since there's no good way to know when a child forks, and we want to update the filters when it does. We must resort to ptrace. Yes it sucks, but it works. When performing a trace-cmd record -F -c program The "-c" switch will cause trace-cmd to ptrace the program. If the program forks, trace-cmd will update the filters to follow the children of the program. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Version 1.1Steven Rostedt2011-02-09
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Do not print double \nSteven Rostedt2011-02-07
| | | | | | | | When trace_printk or trace_marker is used, it may end with a '\n'. Do not print out a double '\n'. If the user wants two '\n' then the trace_printk or trace_marker should specify two. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Remove debug printSteven Rostedt2011-02-07
| | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd/doc: Let environment variable override MANPAGE_DOCBOOK_XSLSteven Rostedt2011-02-01
| | | | | | | In case the user wants to use a different manpage/docbook.xsl, let them override the MANPAGE_DOCBOOK_XSL environment variable. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd/doc: Fix installing krenelshark html documentsSteven Rostedt2011-02-01
| | | | | | | Remove space from the do_install script that was making the install of kernelshark html documents fail. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd/doc: Fail if manpage/docbook.xsl is not foundSteven Rostedt2011-02-01
| | | | | | | | Fail if the docbook.xsl is not found. Currently it fails if the file does not exist, but passing in a blank string does not produce an error. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add events to handle sched switch/wakeup filtersSteven Rostedt2011-02-01
| | | | | | | | | | | | The sched_switch and sched_wakeup filters are special for tracing pids. If we are tracing a pid, we want sched_switch and sched_wakeup to be traced if it is waking up or switching to the given pid. Since events are now expanded into a list holding the filter files to be updated. Either reuse or create if necessary the sched_* events to add the extra filters. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Restructure the enable and filtering of eventsSteven Rostedt2011-02-01
| | | | | | | | Use glob to expand the events given on the command line into separate entities on the event_selection list. This will allow easier settings of the filters during execution time. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Null end the filter stringSteven Rostedt2011-02-01
| | | | | | | After reading the filter, make sure to zero out the end of the filter string. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Use arg field size to test for pointerSteven Rostedt2011-02-01
| | | | | | | | | | | | | Using the "len" variable in print_str_arg() to test if the value is an array when the arg field size is zero is incorrect. When the arg field size is zero, the data is an dynamic value regardless if FIELD_IS_ARRAY is set or not. It may not be set for strings at the end of the record. Only assume the data is a pointer if the arg->field.field->size is sizeof long. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Allow update_ftrace_pid() to allow multiple callsSteven Rostedt2011-02-01
| | | | | | | Change the way update_ftrace_pid() to allow multiple calls to add more than one pid. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Use strtoull() in filter conversionSteven Rostedt2011-01-28
| | | | | | | | The value converted by the filters in comparisons uses strtoll() which returns an invalid number when we are dealing with 64bit IP addresses. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd/restore: Allow override where debugfs/tracing and kallsyms come fromSteven Rostedt2011-01-28
| | | | | | | | | | | | Add to trace-cmd restore: -t dir: where dir is the directory to search instead of looking at debugfs/tracing -k file: where to read kallsyms from, instead of reading /proc/kallsyms Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add -T option to report to test filtersSteven Rostedt2011-01-21
| | | | | | | Add the -T option to print the filter strings created by -F and exit. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-event: Add way to just test a filter stringSteven Rostedt2011-01-21
| | | | | | | If pevent->test_filters is set when a filter string is added, then it will print the filters strings that were created and exit. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Rewrite filter algorithm to simplify itSteven Rostedt2011-01-21
| | | | | | | New rewrite of the parse event filter to greatly simplify the logic. Should be much easier to modify it now. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Show where error occurred in filteringSteven Rostedt2011-01-20
| | | | | | | | | | | | | | | Show a nice pointer where an error occurred in filtering: # trace-cmd report -F 'sched_switch:(next_pid==x)||)prev_pid == 1' version = 6 cpus=4 trace-cmd: No such file or directory Error filtering: sched_switch:(next_pid==x)||)prev_pid == 1 (next_pid==x)||)prev_pid == 1 ^ Expected end where ) was found Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Change sched switch/wakeup plugin outputSteven Rostedt2011-01-19
| | | | | | | | | | | | | | | | | | | | | | The current format for sched_switch and sched_wakeup events is: trace-cmd-2271 [003] 6201.438152: sched_switch: 2271:120:S ==> 0:120: kworker/0:1 <idle>-0 [002] 6201.438301: sched_wakeup: 0:?:? + 2270:120:? trace-cmd [002] Success <idle>-0 [002] 6201.438311: sched_switch: 0:120:R ==> 2270:120: trace-cmd trace-cmd-2270 [002] 6201.438393: sched_switch: 2270:120:S ==> 0:120: kworker/0:1 <idle>-0 [000] 6201.438799: sched_wakeup: 0:?:? + 10:120:? kworker/0:1 [000] Success <idle>-0 [000] 6201.438839: sched_switch: 0:120:R ==> 10:120: kworker/0:1 Which kinda sucks. Rewrite it to look like. trace-cmd-2271 [003] 6201.438152: sched_switch: trace-cmd:2271 [120] S ==> kworker/0:1:0 [120] <idle>-0 [002] 6201.438301: sched_wakeup: trace-cmd:2270 [120] success=1 CPU:002 <idle>-0 [002] 6201.438311: sched_switch: kworker/0:1:0 [120] R ==> trace-cmd:2270 [120] trace-cmd-2270 [002] 6201.438393: sched_switch: trace-cmd:2270 [120] S ==> kworker/0:1:0 [120] <idle>-0 [000] 6201.438799: sched_wakeup: kworker/0:1:10 [120] success=1 CPU:000 <idle>-0 [000] 6201.438839: sched_switch: swapper:0 [120] R ==> kworker/0:1:10 [120] Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Always allocate comm for pevent_register_comm()Steven Rostedt2011-01-19
| | | | | | | | | | | The function pevent_register_comm() will allocate its own comm depending on when you execute it. This is stupid and likely to cause bugs. Make it always allocate its own comm and also change the parameter name to accept const char *. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add struct filter_arg_expLai Jiangshan2011-01-19
| | | | | | | | | | | | | | | | | | | | | | | | | | Use "struct filter_arg_exp" instead of "struct filter_arg_op" for FILTER_ARG_EXP and eliminate the warnings: /home/laijs/work/trace-cmd/parse-filter.c: In function ‘get_exp_value’: /home/laijs/work/trace-cmd/parse-filter.c:1461:2: warning: case value ‘4’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c:1464:2: warning: case value ‘5’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c:1467:2: warning: case value ‘6’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c:1470:2: warning: case value ‘7’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c:1473:2: warning: case value ‘8’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c:1476:2: warning: case value ‘9’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c:1479:2: warning: case value ‘10’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c:1482:2: warning: case value ‘11’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c: In function ‘exp_to_str’: /home/laijs/work/trace-cmd/parse-filter.c:1844:2: warning: case value ‘4’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c:1847:2: warning: case value ‘5’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c:1850:2: warning: case value ‘6’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c:1853:2: warning: case value ‘7’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c:1856:2: warning: case value ‘8’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c:1859:2: warning: case value ‘9’ not in enumerated type ‘enum filter_op_type’ /home/laijs/work/trace-cmd/parse-filter.c:1862:2: warning: case value ‘10’ not in enumerated type ‘enum filter_op_type’ Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Add filter on task commSteven Rostedt2011-01-14
| | | | | | | | | | | | | | | | | Add a 'COMM' special field to allow the filter in trace-cmd report to filter on the task's comm. Most records do not have data about what the comm is of the task that recorded the record. But comms are stored in various ways that parse-events can handle. By adding a special field 'COMM' (all caps) the user can now filter out tasks based on their name. This is useful when filtering out (or in) a bunch of threads. -F 'COMM != "trace-cmd"' Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Use conditional assignment of CC and ARDarren Hart2011-01-06
| | | | | | | | | | | Allow the user to specify CC and AR via the environment by only setting them in the Makefile if they don't already exist. This is useful for external build systems for example. Signed-off-by: Darren Hart <dvhart@linux.intel.com> LKML-Reference: <1294333733-7746-1-git-send-email-darren@dvhart.com> CC: Tim Bird <tim.bird@am.sony.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Correct trace._handle usage in event-viewer.pyDarren Hart2011-01-04
| | | | | | | | | | The handle member object was changed to _handle, correct the usage in event-viewer.py. Signed-off-by: Darren Hart <dvhart@linux.intel.com> LKML-Reference: <1294173820-7043-4-git-send-email-dvhart@linux.intel.com> CC: Tom Zanussi <tom.zanussi@intel.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Move trace-usage to lib objectsDarren Hart2011-01-04
| | | | | | | | | | Make the usage function available to shared objects, such as the python ctracecmd.so module built by the SWIG mechanism. Signed-off-by: Darren Hart <dvhart@linux.intel.com> LKML-Reference: <1294173820-7043-3-git-send-email-dvhart@linux.intel.com> CC: Tom Zanussi <tom.zanussi@intel.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Move *_tracing_file into trace-util.cDarren Hart2011-01-04
| | | | | | | | | | Make tracecmd_(get|put)_tracing_file available to shared objects, such the python ctracecmd.so module generated by the SWIG mechanism. Signed-off-by: Darren Hart <dvhart@linux.intel.com> LKML-Reference: <1294173820-7043-2-git-send-email-dvhart@linux.intel.com> CC: Tom Zanussi <tom.zanussi@intel.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add -a option to record to record all eventsSteven Rostedt2011-01-04
| | | | | | | Add the -a option to trace-cmd record that will bring back the old behavior of recording the format of all events. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Only record the formats of the events being recordedSteven Rostedt2011-01-04
| | | | | | No need to record all event formats when only a few are being recorded. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Fix ull warningJohannes Berg2010-12-22
| | | | | | | | Fix this warning: trace-output.c:832: warning: format '%llx' expects type 'long long unsigned int', but argument 3 has type 'off64_t' Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add max and min latencies for report with -wSteven Rostedt2010-12-14
| | | | | | | | | Instead of just showing the average wakeup latency in the -w report, also show the max and min latencies as well as the time stamp that they occurred. Requested-by: Shawn Bohrer <sbohrer@rgmadvisors.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Allow trace-read -w to work without wakeup event success fieldSteven Rostedt2010-12-02
| | | | | | The wakeup event "success" field is going away. Do not depend on it. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-graph: Do not depend on wakeup success fieldSteven Rostedt2010-12-02
| | | | | | | | The "success" field of the wakeup events is going away, don't depend on them there. If they don't exist, then the wakeup was successful, so treat it as such. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* parse-events: Allow pevent_read_number_field() to have NULL fieldSteven Rostedt2010-12-02
| | | | | | | Let pevent_read_number_field() have a NULL field. Just return -1 if one does not exist. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-view: Handle unknown events more gracefullyJan Kiszka2010-12-02
| | | | | | | | | | | | | | Kernel 2.6.35.7 generates invalid trace files due to /sys/kernel/debug/tracing/events/scsi/scsi_dispatch_cmd_timeout/format returning "FORMAT TOO BIG" (seems to be fixed in 2.6.36). This makes pevent_data_event_from_type return NULL, and dereferencing of event crashes the viewer. Catch and report the error properly. Signed-off-by: Jan Kiszka <jan.kiszka@web.de> LKML-Reference: <4CB758C1.8090906@web.de> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* kvm: Do not warn on new fieldsAvi Kivity2010-11-29
| | | | | | | | | The kvm plugin understands a few new fields; don't warn if they are missing, as expected on older kernels. Signed-off-by: Avi Kivity <avi@redhat.com> LKML-Reference: <1291051539-16791-1-git-send-email-avi@redhat.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* kvm: Display the new kvm_exit info1 and info2 fields, if availableAvi Kivity2010-11-23
| | | | | Signed-off-by: Avi Kivity <avi@redhat.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* kvm: Fix typo "UNKOWN"Avi Kivity2010-11-23
| | | | | Signed-off-by: Avi Kivity <avi@redhat.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* kvm: Parse svm exit reasonAvi Kivity2010-11-23
| | | | | | | | svm exit reasons use different code than vmx; use the new "isa" trace field to select the instruction set and display the strings accordingly. Signed-off-by: Avi Kivity <avi@redhat.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
generated by cgit v1.2.2 (git 2.25.0) at 2025-04-08 14:52:28 -0400