aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAge
...
* parse-events: Add PRINT_BPRINTK to separate bprintk strings from dynamicSteven Rostedt2010-11-16
| | | | | | | | | | | When bprintk had a "%s",string format, the args for the bprintk used a dynamic string arg for the string. Unfortunately, this would cause the string to not be found and nothing would be printed. This patch adds a special PRINT_BPRINTK type that lets the processing do something special with the bprintk strings. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Add ability to add options to trace.datSteven Rostedt2010-11-15
| | | | | | | | Add tracecmd_add_option that allows us to add options into the trace.dat file. Currently, options may only be added before the file has been written. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Reset record pointer in set_cpu_to_timestamp()Steven Rostedt2010-11-11
| | | | | | | | | | | | | | The code to reset the cpu to a specific timestamp some time fails if the time stamp does not match the current saved time stamp and we go to another page. Without resetting the timestamp for that page, the algorithm can get confused if the timestamp is currently in the middle of the page and the record we want is on the page but before the timestamp. Simply reset the timestamp to the start of the page when going to another page. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Search $TRACE_CMD_PLUGIN_DIR for plugins.David Sharp2010-10-20
| | | | | | | | | | Allows dynamic specification of the plugin directory so that plugins can be used when the install prefix cannot be predicted, such as when using a pre-compiled version of trace-cmd with Autotest. Signed-off-by: David Sharp <dhsharp@google.com> LKML-Reference: <1285718416-1470-1-git-send-email-dhsharp@google.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Plug startup race between flush_threads and recordersJan Kiszka2010-10-15
| | | | | | | | | | | As the SIGUSR1 handler is so far registered after forking the recorder child, there is a small race window where a flush sent by the main process can actually kill the child. Plug this by registering the signal handlers before the fork. This also allows to unify SIGINT registration. Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> LKML-Reference: <4CB8401B.2020801@siemens.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Do not let tracecmd_peek_data read past # of CPUsSteven Rostedt2010-10-15
| | | | | | | | tracecmd report --cpu # will segfault if # is greater than the last CPU defined in the trace.dat file. Prevent tracecmd_peek_data() from reading past the last CPU. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* plugin_kvm: Fix kvm_exit rip formattingAvi Kivity2010-10-11
| | | | | | | | %0xlx is not a valid printf format. Signed-off-by: Avi Kivity <avi@redhat.com> LKML-Reference: <1286787515-13722-1-git-send-email-avi@redhat.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* plugin_kvm: Remove stray line from kvm_emulate_insn_handler()Avi Kivity2010-10-11
| | | | | | | | 'guest_rip' is not part of kvm_emulate_insn. Signed-off-by: Avi Kivity <avi@redhat.com> LKML-Reference: <1285493922-29797-1-git-send-email-avi@redhat.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd/listen: Force search of unique portSteven Rostedt2010-10-08
| | | | | | | | | | | | | On some systems, the code to find a unique udp port may give a invalid success of the previous port. Which causes two child process to listen to the same port, and makes one of them fail. Force the start of the port search to the port after the last port found, to prevent the same port from being used by two listeners. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd/documentation: Update trace-cmd man page to include stackSteven Rostedt2010-10-06
| | | | | | Add the stack command to the trace-cmd man page. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* trace-cmd: Added trace stack commandSteven Rostedt2010-10-04
| | | | | | | | Added the command "trace-cmd stack" to run the stack tracer. It simply enables the stack tracer and depending on the options it will ouput the current stack, or disable it. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* plugin_kvm: Disassemble instructions for kvm_emulate_insnAvi Kivity2010-09-23
| | | | | | | | | | Override kvm_emulate_insn formatting to use a disassembler to format the emulated instruction. If a disassembler (udis86) is not available, fall back to showing the instruction bytes in hex. Signed-off-by: Avi Kivity <avi@redhat.com> LKML-Reference: <1284906837-2431-1-git-send-email-avi@redhat.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* Merge remote branch 'push/trace-cmd-kvm' into trace-cmdSteven Rostedt2010-09-23
|\ | | | | | | | | | | | | Conflicts: Makefile Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
| * trace-cmd: Add plugin_kvm.soSteven Rostedt2010-06-17
| | | | | | | | | | | | | | | | | | Added code to handle the kvm events that happened whil doing a trace on a 2.6.34 version of the kernel on my Intel box. This adds a kvm plugin to process some of the kvm events. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Use sysconf to figure out # of CPUsSteven Rostedt2010-09-20
| | | | | | | | | | | | | | | | Use the POSIX sysconf instead of the hack to figure out the number of CPUs the current machine has. Suggested-by: Russell King <linux@arm.linux.org.uk> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-graph: Give a better CPU color difference in task plotsSteven Rostedt2010-09-17
| | | | | | | | | | | | | | The CPU color for task plots is a bit close at the lower end. Add some salt to the algorithm to spread it out a bit more. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-graph: Show CPU not task on task plot displaySteven Rostedt2010-09-17
| | | | | | | | | | | | | | The plot of a task, we already know what task it is. It would be nice to know what CPU a task is on at a given location. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-graph: Make zooming in on task plots stay filledSteven Rostedt2010-09-17
| | | | | | | | | | | | | | | | | | Currently when you zoom into a task plot, when there are no events in the viewable area, the line turns blank (as if it were just idle). This patch keeps the plot with a bar even if no events are visible. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-graph: Make zooming in on CPU plots stay filledSteven Rostedt2010-09-17
| | | | | | | | | | | | | | | | | | Currently when you zoom into a CPU plot, when there are no events in the viewable area, the line turns blank (as if it were just idle). This patch keeps the plot with a bar even if no events are visible. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-graph: Fix plots to show start bars in zoomed views.Steven Rostedt2010-09-17
| | | | | | | | | | | | | | | | | | | | | | | | Currently, if a zoomed view has events before the start of the view, those events are not examined to determine if the bar of a graph should be shown at the start. This patch adds code to the cpu and task plots to look for events that happened before the view to determine if a bar should be shown or not. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Add ref counter to recordSteven Rostedt2010-09-17
| | | | | | | | | | | | | | | | Add tracecmd_record_ref() to let an application up the reference to a record to be used at a later moment. This makes sure the record is not freed by the calling function. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Add trace-cmd restoreSteven Rostedt2010-09-13
| | | | | | | | | | | | | | | | The restore feature now allows you to make a trace.dat file from several trace.dat.cpu.X files. This will let you put together files leftover from a crashed trace-cmd record run. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Remove cpus from tracecmd_create_init() parametersSteven Rostedt2010-09-13
| | | | | | | | | | | | | | The tracecmd_create_init() function never used the cpus variable that was passed in. It is not needed, thus it is removed. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Use lseek64 and off64_t in tracecmd_append_cpu_data().David Sharp2010-09-13
| | | | | | | | | | | | | | | | | | | | lseek64 returns 64-bit, but was being truncated to int. Large offsets would cause the subsequent error checking to fail spuriously, leaving behind a bad trace.dat and a collection of trace.dat.cpuX files. Signed-off-by: David Sharp <dhsharp@google.com> LKML-Reference: <1284411266-30442-1-git-send-email-dhsharp@google.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-graph: Fix trace_seq free-before-useStefan Hajnoczi2010-09-07
| | | | | | | | | | | | | | | | | | | | Commit e892161f731e958980fa977e2d7980f3cf0819ce changed trace_seq to dynamically expand its buffer. In two instances the destroy function was called too early, causing a freed s.buffer to be used. Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com> LKML-Reference: <1283684704-5650-1-git-send-email-stefanha@linux.vnet.ibm.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | parse-events: Free token in dynamic_array parsingSteven Rostedt2010-09-03
| | | | | | | | | | | | | | The dynamic_array parsing left out a necessary free_token(). Reported-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | parse-events: Make trace_seq dynamically expand its buffersSteven Rostedt2010-09-03
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The trace_seq code that is used to copy data into a buffer to later be able to print was originally taken from the Linux kernel. But the constraints on the kernel are different than what they are in userspace. Since the kernel uses the trace_seq in critical paths, it can not allocate memory during use, so it used a static page buffer. But in userspace, we do not have this constraint, and are allowed to realloc the buffer when needed. There's no reason to fill the buffer when we hit the buffer size, we can simply allocate more memory. The downside to this is that we must call a destructor on the trace_seq structure. All trace_seq's must be initialized with trace_seq_init(), so this patch matched all trace_seq_init()'s with the new function trace_seq_destroy(). Suggested-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | parse-events: Add pevent_get_field_raw()Steven Rostedt2010-09-02
| | | | | | | | | | | | | | | | | | Add new function pevent_get_field_raw that returns a pointer to the actual data and also returns the size. This is makes handling arrays convenient in plugins. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Add range for cpus to filter on trace-cmd reportSteven Rostedt2010-09-02
| | | | | | | | | | | | | | | | | | | | | | | | Currently, we can only filter the report on one CPU or all CPUS. This patch allows users to add a range: trace-cmd report --cpu 1-5 trace-cmd report --cpu 1,3,6 trace-cmd report --cpu 1,2-4,6,9 trace-cmd report --cpu 1 --cpu 2 --cpu 5,6 --cpu 9-12 Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Don't SIGSEGV when event field format cannot be parsed (v2).David Daney2010-07-23
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | print fmt: "page=%p pfn=%lu order=%d migratetype=%d", REC->page, ({ struct page *__pg = (REC->page); int __sec = page_to_section(__pg); (unsigned long)(__pg - __section_mem_map_addr(__nr_to_section(__sec))); }), REC->order, REC->migratetype This cannot be parsed, leading to a NULL struct event_format* being passed to pevent_get_common_field_val, which produces a SIGSEGV. It would be good to get a parsable format from the kernel, but to remediate the problem for legacy kernels, we can just return an error indicator in this case. This allows some output from trace-cmd report, although perhaps with some missing data. But this is better than crashing. (v2): Do the check in all pevent_get_* Signed-off-by: David Daney <ddaney@caviumnetworks.com> LKML-Reference: <1279832705-14219-1-git-send-email-ddaney@caviumnetworks.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Don't SIGSEGV in trace-graph.c:load_handleDavid Daney2010-07-20
| | | | | | | | | | | | | | | | | | This 'fixes' another crash I encountered. Don't try to dereference NULL records. Signed-off-by: David Daney <ddaney@caviumnetworks.com> LKML-Reference: <1279586810-29859-3-git-send-email-ddaney@caviumnetworks.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Don't try to read unmapped memory (v2).David Daney2010-07-20
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Under some circumstances, I'm not sure exactly which, a trace.dat file may contain a bunch of zeros at the end of one or more of the trace logs. This can lead to tracecmd_peek_data() attempting to read past the end of the mmaped log, causing SIGSEGV. This is a two part fix. 1) For 'new format' data, we always try to read 8 bytes of data. Make sure that they are all on the current page. 2) In pevent_print_event(), if record->size is negative, warn and then skip attempting to print it. Signed-off-by: David Daney <ddaney@caviumnetworks.com> LKML-Reference: <1279586810-29859-2-git-send-email-ddaney@caviumnetworks.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | kernelshark: Add compat gtk_menu_item_get_label() functionSteven Rostedt2010-07-16
| | | | | | | | | | | | | | | | The gtk_menu_item_get_label() function was implemented in gtk 2.16. This adds the compat function for building kernelshark against gtk earlier than 2.16. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Initialize the ftrace info struct to the handlerSteven Rostedt2010-07-16
| | | | | | | | | | | | | | | | The finfo structure to represent ftrace events did not initialize the tracecmd input handler. When tracing the stacktrace event, it would cause a SEGFAULT. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Fix trace-cmd function graph handling of its eventsSteven Rostedt2010-07-14
| | | | | | | | | | | | | | | | | | | | | | | | | | The function graph needs to store the function graph return event for later use. This patch changes the function graph callbacks to use the context to store this information and initialize it when first referenced. The event data is now stored on the tracecmd_input handle, that allows more than one tracecmd_input handle to work with this code in one utility. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | parse-events: Fix bprint %d to be 32 bit even on 64 bit boxesSteven Rostedt2010-07-09
| | | | | | | | | | | | | | | | | | | | The bprint processing of %d was the same as %ld, which would cause it to read 8 bytes instead of 4 bytes on a 64 bit box. This would cause the other parameters after it to be shifted by 4 bytes and produce garbage results. Reported-by: Darren Hart <dvhltc@us.ibm.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Update documentation for added '-i' option to trace-cmd recordSteven Rostedt2010-06-29
| | | | | | | | Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Add option to ignore event not found errorKOSAKI Motohiro2010-06-29
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Currently, if 'trace-cmd start' passes an event that is not found it will cause the command to fail. This can be a problem for an administrator using trace-cmd on several machines that have different modules configured in. One script may be used that has all the events to be traced, but some of the events may not exist on all machines. This means each machine will need its own separate script. Instead of having a separate script, add a '-i' option that lets one ignore failed events. Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> LKML-Reference: <20100629144832.38C6.A69D9226@jp.fujitsu.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | build: Add tags for file names, function prototypesJohn Kacur2010-06-25
| | | | | | | | | | | | | | | | For vi, add tags for file names, function prototypes and external and forward variable declarations. Signed-off-by: John Kacur <jkacur@redhat.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Reset plugin_function variables on unloadSteven Rostedt2010-06-24
| | | | | | | | | | | | | | | | | | | | Seems that unloading the dynamic plugin and loading it again does not reset the static variables back to their original value. On unload, reset these variables back. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Add indentation to function tracerSteven Rostedt2010-06-24
| | | | | | | | | | | | | | | | | | Add a plugin_function.so that will override the function tracer and use the parent functions to determine how to indent a function. This give a nice visual affect similar to what the function graph tracer provides. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-cmd: Load the default ftrace print handlers before pluginsSteven Rostedt2010-06-24
| | | | | | | | | | | | | | | | Load the ftrace function print handlers before loading the plugins. This allows for a loaded plugin to override one of the ftrace default output functions. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | kernelshark: Store the capture output for later displayingSteven Rostedt2010-06-22
| | | | | | | | | | | | | | | | Have the shark_info save the captured output as well as the max buffer size, and have those persistent when closing and opening the capture dialog. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | kernelshark: Add a max display entrySteven Rostedt2010-06-22
| | | | | | | | | | | | | | Let the user change the size of the max output size from a capture. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-graph: Set comm to NULL if field not foundSteven Rostedt2010-06-22
| | | | | | | | | | | | | | | | If the event_comm_field has not been set (does not exist) then we must also set comm to NULL so that later in the function, it does not reference the comm variable. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | kernelshark: Limit the capture output display to 1 MegSteven Rostedt2010-06-22
| | | | | | | | | | | | | | | | Only allow one meg to be displayed at a time in the output display of the capture window. Otherwise, it can fill up too much and waste a lot of memory. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | kernelshark: Change cancel button on capture dialog to closeSteven Rostedt2010-06-22
| | | | | | | | | | | | | | | | | | | | | | "Cancel" is confusing and people may think that the Cancel will just stop the trace and do nothing more. But this is incorrect. The cancel button will stop the trace and update the data for the graph and list. Changing it to "Close" shows that it just closes the window. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | kernelshark: Have capture "Run" button also be "Stop" buttonSteven Rostedt2010-06-22
| | | | | | | | | | | | | | Instead of popping up another dialog to stop the run, have the "Run" button simply change to "Stop". This makes it a lot nicer. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | trace-graph: Test for comm field before using itSteven Rostedt2010-06-22
| | | | | | | | | | | | | | The ftrace_comm_field may not be set. Do not use it if it has not been initialized. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
* | kernelshark: Add more room for capture outputSteven Rostedt2010-06-22
| | | | | | | | | | | | | | Make the capture output display bigger and have only it expand when the dialog expands. Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
generated by cgit v1.2.2 (git 2.25.0) at 2025-04-19 18:12:24 -0400