e4e1bf73b3
line-reader: handle invalid parameters more gracefully
...
This silences coverity which doesn't handle return_if_fail() macros
very well.
2016-08-22 11:02:03 -07:00
46a257da6e
elfparser: suppress warnings about cast alignment
...
These are aligned according to the spec, so we can simply silence them
with the gpointer cast.
2016-08-21 17:32:59 -07:00
e81204dec9
process-model-item: remove duplicate const
2016-08-21 17:32:59 -07:00
a5b08cad94
reader: increment j, not i when swapping counter values
2016-08-21 17:32:59 -07:00
6058c62f44
reader: bswap jitmap item count
2016-08-21 17:32:59 -07:00
94c8551063
kernel-symbol: fix alignment warning
...
Our pointer is always malloc() aligned, so we have a safe cast here.
2016-08-21 17:32:59 -07:00
24716d7c88
elf: add more pango mappings
2016-08-18 14:38:14 +02:00
1ad1b23f57
callgraph: ensure cell renderer is available
2016-08-16 18:03:25 +02:00
c45a55b467
lib: avoid using hexpand in SpProfilerMenuButton
...
This can cause the widget to expand now that GtkHeaderBar supports expand
for start/end packed children.
2016-07-06 23:31:04 -07:00
329743c917
callgraph: Use U64_TO_POINTER
...
This fixes a "cast to pointer from integer of different size" on i586.
https://bugzilla.gnome.org/show_bug.cgi?id=767693
2016-06-15 18:06:25 -04:00
15a2873c25
Move from SpScrolledWindow to GtkScrolledWindow
...
GtkScrolledWindow has max-content-{width,height} properties now. Remove
SpScrolledWindow and bump the gtk+ requirement to 3.21.3 at the same
time.
2016-06-10 21:12:06 +02:00
28d9a7d8b4
callgraph view: Expand cursor row on Right
2016-06-05 21:08:39 +02:00
67efcc462b
Fix handling plural forms
...
https://bugzilla.gnome.org/show_bug.cgi?id=765929
2016-05-03 12:42:20 +03:00
9c6cec9b49
perf: avoid authorization dialog when user owns target pid
...
If we check the UID of the pid from /proc/pid/status and the UID matches
our current user, we should be able to forego the authorization dialog.
This should fix a regression where sysprof worked from jhbuild when
profiling a process you have access to.
2016-04-25 15:11:54 -07:00
afadd90834
perf: ping the sysprofd service before we consider it up
...
This should help catch failures where the peer is not yet up and might
fail to start, before we actually block on communicating with it.
2016-04-24 17:49:03 -07:00
8709eaff14
build: let gcc know our cast is safe
...
Data is already aligned, so casting the ->data field is fine.
2016-04-20 22:35:17 -07:00
e759f2999f
elf: tag GtkSourceView symbols
...
We see this library enough in the platform that it is work tagging.
2016-04-20 20:36:01 -07:00
b97425b300
perf: make sure we emit finished after last process exits
...
This fixes the "auto-stop" feature when the child exits, at least in the
case of /usr/bin/gnome-calculator. However, when testing with gedit I
did not see it accurately clean up, so something still needs to be
troubleshooted.
2016-04-18 22:31:39 -07:00
a594d06b0c
perf: let the compiler know our cast is alignment-safe
...
We know everything is 64-bit aligned, so casting via (void*) squashes
the compiler warning.
2016-04-18 18:52:20 -07:00
7a639e9bf8
menu-button: ensure the scrolled window is always full width
...
When no items were present, this was shrinking. It should always maintain
width of the popover.
2016-04-16 18:58:43 -07:00
57faf5b9e6
menu-button: show process arguments in process list
...
This parses the \0 out of the cmdline proc entry and makes them visible
as both in the list of processes and as tooltips when hovered.
2016-04-16 18:57:50 -07:00
9e082a257d
menu-button: hide the process list when whole-system is enabled
...
No sense in showing a desensitized widget that cannot be toggled except
for discover-ability. But progressive disclosure is probably better.
2016-04-16 18:21:26 -07:00
4fdd7aca71
menu-button: improve error reporting on argv command line
2016-04-16 17:40:07 -07:00
f4aabd8b7c
perf: ignore not-supported failures
...
If the pre-authorize fails due to lacking support for sysprofd, that is
fine, we can fail later on if authorization was required.
2016-04-16 17:16:29 -07:00
f5904bc493
perf: fix typo
...
Typo when polkit/sysprofd is unavailable to pre-authorize the request.
2016-04-16 17:14:42 -07:00
4479c56135
perf: polkit is optional for the library
...
So changing this was a little backwards. The goal here is to allow using
libsysprof-2 without polkit, it just won't have the ability to elevate
privileges.
2016-04-16 17:13:56 -07:00
74d88245c5
perf: use #ifdef not #if
...
This is conditionally added to CFLAGS, so make sure we use #ifdef instead
of #if.
2016-04-16 17:08:38 -07:00
c2eecefe50
build: add polkit cflags/libs to libsysprof-2 target
2016-04-16 17:05:56 -07:00
f307ce50ab
platform: fix missing ;
...
Clearly, we are testing the fallback path well.
2016-04-16 17:00:56 -07:00
f2d40eda89
platform: include gstdio for g_unlink().
2016-04-16 16:57:11 -07:00
9f8f33b210
perf: use #ifdef, not #if
...
config.h does not unconditionally define these, they are not defined when
inactive.
2016-04-16 16:56:41 -07:00
06aaff90e5
perf: optionally enable clockid and use_clockid
...
Some kernels to not have this feature yet, so check for support in
linux/perf_event.h and optionally enable the struct field.
2016-04-16 16:07:11 -07:00
de572f4ed6
memfd: add workaround for systems without memfd
...
Just a quick workaround to use a tmpfile if memfd is not available. We
should try to ensure that we get something on tmpfs, but this is probably
good enough for a fallback.
2016-04-16 15:49:03 -07:00
8291e05da3
menu-button: sync menu button settings when attaching profiler
...
We were failing to launch a program a second or third time, due to the
settings not beeing sync'd to the profiler instance.
2016-04-15 18:19:20 -07:00
ceeb29b0e6
callgraph: add accessor to determine number of functions
...
We can use this to notify the user that not enough samples where collected
to build a callgraph. This can happen when all the samples collected are
ignored (due to being in ignored kernel space address ranges for example).
2016-04-15 18:11:02 -07:00
18d2d81824
perf-source: we need PERF_SAMPLE_TID
...
I mistakingly removed this, which caused us to not get sample delivery.
2016-04-15 05:04:29 -07:00
49e9ad9db8
capture: make counter value a union of int64 and double
...
This allows us to use the type field of the counter to specify if the
counter is a double or a 64-bit integer.
2016-04-15 04:54:55 -07:00
48420b0c87
writer: increment the next counter id, not the return value
2016-04-15 04:51:23 -07:00
253b387323
capture: add sp_capture_reader_peek_frame()
...
It can be useful to peek at the whole frame header (for things like time
or pid or cpu).
2016-04-15 04:51:23 -07:00
ad5f855a2b
capture: add 64-bit time stamp for start time
...
We keep the start time as a iso8601 string, which is useful, but not
enough for relative timestamps to the epoch (start time). Include a
clock time in the same clock domain.
2016-04-15 04:51:23 -07:00
f72306f5a8
writer: use SP_CAPTURE_CURRENT_TIME
...
This should be nsec, not usec based time.
2016-04-15 04:51:23 -07:00
dd821b48e8
perf-source: fix acquisition of time for comm and mmap events
...
We were not getting the proper value for time. We need sample_id_all set
according to my cursory reading of the core.c in the kernel. Also, the
time is aligned to 64-bit (after the comm field).
2016-04-15 04:51:23 -07:00
609ee997da
perf-counter: pass sample_id_all to sysprofd
2016-04-15 04:51:23 -07:00
d7ed2ddf30
writer: allow callers to define a counter range
...
This allows dynamic counter ids that can be allocated in blocks.
2016-04-14 21:36:07 -07:00
1b8cfed5c8
writer: use dup()d fd when creating reader
...
Readers can share the file-offset in the file-descriptor table, but should
really have their own fd that they can close().
2016-04-14 19:25:26 -07:00
cd0a39859e
capture: drop ADD from capture type
...
I opted to not implement this yet, so ensure its not in the capture
format types.
2016-04-14 18:33:23 -07:00
c87a90cd9a
capture: add frame type for defining and setting counters
...
We might want to add a CTRADD type later on, for relative values rather
than absolute. But this should get us started.
Simply define counters upfront, and then set them during the capture
process.
Obviously, we need to come up with a good way to visualize this
information in the UI.
2016-04-14 18:29:39 -07:00
89e75fae2d
profiler: reset timer when starting
...
If we reuse the profiler instance (not convinced this is a good idea yet)
then we could have a nonzero elapsed time. Reset it immediately so that
the UI doesn't jitter between the previous value and 00:00.
2016-04-14 05:39:49 -07:00
a7982ad1e8
perf: delay source start until polkit has authorized
...
Without this, after logging in you are already multiple seconds into your
profiling session recording. Not ideal. So instead, we do the async polkit
auth upfront during SpSource::prepare(), and then toggle ready after we
have received notification.
2016-04-14 04:36:18 -07:00
88d3ae3b74
profiler: extract SpProfiler into an interface
...
It would be nice to be able to support a remote profiler session in the
future, so add SpLocalProfiler implementation of SpProfiler interface.
2016-04-14 02:37:28 -07:00
