mirror/tracy
1
0
Fork 0
mirror of https://github.com/wolfpld/tracy.git synced 2024-11-10 10:41:50 +00:00
Code Issues Packages Projects Releases Wiki Activity
5,030 Commits 2 Branches 28 Tags 44 MiB
85e8a6a81c
Commit Graph

851 Commits

Author SHA1 Message Date
Bartosz Taudul
40281ce2a1 Add default no-op to switch. 2020-03-26 01:07:25 +01:00
Bartosz Taudul
add5b29d03 Report CPU architecture in welcome message. 2020-03-25 21:28:02 +01:00
Bartosz Taudul
ce449ac0e2 Notify server that parameter was handled. 2020-03-25 20:37:26 +01:00
Bartosz Taudul
f114ec3f80 Add code transfer from client to server. 2020-03-25 20:04:55 +01:00
Bartosz Taudul
3e0e120222 Add extra parameter to server queries. 2020-03-25 20:04:01 +01:00
Bartosz Taudul
c999a74d34 Symbol length transfer. 2020-03-25 18:32:36 +01:00
Bartosz Taudul
d47e6819a8 Collect symbol sizes. 2020-03-25 18:28:28 +01:00
Bartosz Taudul
6c0c508280 Ignore kernel-only stacks. 
It is common to receive duplicate stack traces for the same timestamp
(and thread), one containing proper user-space stack, and the second one
containing only kernel frames. Discard the second one, as there's no
documentation how this should be interpreted and the kernel stack is
mostly useless.
 2020-03-21 15:25:30 +01:00
Bartosz Taudul
c7afda2562 Exit processing loops when trace has stopped. 2020-03-10 18:56:49 +01:00
Bartosz Taudul
c6bb08355c Allow specification of port through env variable. 2020-03-08 16:14:36 +01:00
Bartosz Taudul
1da62c2190 Send deferred lock names. 2020-03-08 15:05:35 +01:00
Bartosz Taudul
127224acc6 Send listen port in broadcast message. 2020-03-08 14:37:59 +01:00
Bartosz Taudul
14c896573d Separate config for data and broadcast port. 2020-03-08 14:34:09 +01:00
Bartosz Taudul
2ffaa88c9e Fix typo. 2020-03-08 14:19:08 +01:00
Bartosz Taudul
e7240cb77d Custom lock name transfer. 2020-03-08 13:47:38 +01:00
Bartosz Taudul
f945278959 Fix rpmalloc on android. 2020-03-02 17:10:47 +01:00
Bartosz Taudul
c36ed4b8b8 Boring warning fixes. 2020-03-01 01:48:20 +01:00
Bartosz Taudul
c23984dd6a Fix static assert in rpmalloc. 2020-03-01 01:31:31 +01:00
Bartosz Taudul
e9a32d5dc7 Greatly increase queue block size. 
Previous block size could hold only 256 elements (8KB), which stressed
out the memory allocator. Storing 65536 elements (2MB) per block almost
completely reduces the allocator pressure.
 2020-03-01 01:15:13 +01:00
Bartosz Taudul
82f463724c Update rpmalloc to 1.4.0. 
Notable changes: use C++11 atomics everywhere.
 2020-03-01 01:02:25 +01:00
Bartosz Taudul
710a2a64e4 Fix copy pasta. 2020-02-27 14:08:56 +01:00
Bartosz Taudul
4346620afa No need to copy module name. 2020-02-27 13:45:39 +01:00
Bartosz Taudul
fd8a9465d4 Cosmetics. 2020-02-27 13:40:41 +01:00
Bartosz Taudul
9ae71ac4ee Dl_info doesn't destroy data. 2020-02-27 13:28:45 +01:00
Bartosz Taudul
5f6b3d2cd5 No need for module name intermediate buffer. 2020-02-27 13:24:36 +01:00
Bartosz Taudul
474383b656 Only copy symbol strings, if needed. 2020-02-27 13:17:26 +01:00
Bartosz Taudul
2df6f9068a Don't retrieve symbol name for address. 2020-02-27 12:58:01 +01:00
Bartosz Taudul
be5793987e Don't send symbol name. 2020-02-27 12:49:48 +01:00
Bartosz Taudul
56dce646cc Symbol address decoding on unix. 2020-02-26 23:38:04 +01:00
Bartosz Taudul
4ddafdeeaf Symbol address decoding for old androids. 2020-02-26 23:24:18 +01:00
Bartosz Taudul
7c506d5426 Remove unused variables. 2020-02-26 23:24:11 +01:00
Bartosz Taudul
26cee8acf0 Perform symbol information queries. 2020-02-26 22:35:15 +01:00
Bartosz Taudul
ef05570540 Symbol address decoding (win32 implementation). 2020-02-26 22:32:42 +01:00
Bartosz Taudul
03ff08a934 Increase max name size. 2020-02-26 22:32:09 +01:00
Bartosz Taudul
d1fcf80c2d Move definition of max symbol name size to one place. 2020-02-26 22:30:11 +01:00
Bartosz Taudul
c0f49c648b Validate size. 2020-02-26 22:27:10 +01:00
Bartosz Taudul
890cec9872 Retrieve symbol addresses on unix. 2020-02-26 02:25:45 +01:00
Bartosz Taudul
9231261d73 Retrieve image name on unix. 2020-02-26 02:11:51 +01:00
Bartosz Taudul
fe80a7ed46 Retrieve symbol address on old androids. 2020-02-26 02:06:44 +01:00
Bartosz Taudul
abf8c42a7c Send module name. 2020-02-26 00:33:09 +01:00
Bartosz Taudul
7d0dac9ae2 Store callstack frame module name. 2020-02-26 00:32:47 +01:00
Bartosz Taudul
4cf520db93 Unify copying symbol strings. 2020-02-26 00:02:30 +01:00
Bartosz Taudul
c5b2d14f8c Send sampling period in welcome message. 2020-02-25 23:12:31 +01:00
Bartosz Taudul
2b7f5091f1 Store sampling period. 2020-02-25 23:08:52 +01:00
Bartosz Taudul
3402d16548 Send symbol base address. 2020-02-25 23:03:40 +01:00
Bartosz Taudul
85ffe0ea04 Don't search module list for kernel addresses. 2020-02-24 23:04:53 +01:00
Bartosz Taudul
ece32b47df Zero capacity is invalid. 2020-02-24 23:04:53 +01:00
Bartosz Taudul
9c9e854005 Replace list with vector. 
Maybe next time let's not forget that there's already a custom
allocating vector available.
 2020-02-24 23:04:53 +01:00
Bartosz Taudul
24cd73e366 Fix linux tracing with long pids. 2020-02-23 18:23:53 +01:00
Bartosz Taudul
0fa1d25d98 Disable trace annotations. 2020-02-23 18:20:48 +01:00
Bartosz Taudul
02d200878d Process queue data in-place. 2020-02-23 15:18:24 +01:00
Bartosz Taudul
96034bca3e Force inline AppendData(), NeedDataSize(). 2020-02-23 14:44:19 +01:00
Bartosz Taudul
bd34c24b84 Increase block size. 2020-02-23 12:35:30 +01:00
Bartosz Taudul
26b13abac8 Pre-fill module cache. 2020-02-22 21:32:18 +01:00
Bartosz Taudul
0a02cf32be Add module name cache. 2020-02-22 21:32:10 +01:00
Bartosz Taudul
096e8cd8ae Retrieve module name if symbol name cannot be found. 2020-02-22 21:06:32 +01:00
Bartosz Taudul
d0930e9053 Use maximum possible sampling rate. 2020-02-22 19:08:15 +01:00
Bartosz Taudul
4502858407 Use maximum possible etw buffer size (1MB). 2020-02-22 18:52:38 +01:00
Bartosz Taudul
e270603117 Don't write reference time to memory in each iteration. 2020-02-22 18:52:37 +01:00
Bartosz Taudul
054a6f8563 Send time deltas in callstack sample data packets. 2020-02-22 16:42:47 +01:00
Bartosz Taudul
1ee80e0df5 Send/free callstack sample payloads. 2020-02-22 16:20:43 +01:00
Bartosz Taudul
3b0ed5337b Provide TraceSetInformation() definition for cygwin. 2020-02-22 16:03:07 +01:00
Bartosz Taudul
baf8e6fe80 No support for sampling on 32-bit windows. 
Note that 32-bit applications running on 64-bit windows will perform
sampling.
 2020-02-22 14:16:04 +01:00
Bartosz Taudul
23fe3e623d 64-bit only version of callstack payload sender. 2020-02-22 14:05:01 +01:00
Bartosz Taudul
9e9c7db5b1 Send sampled call stacks. 2020-02-22 13:42:09 +01:00
Bartosz Taudul
f186540c4f Fix callstack pointers in 32-bit builds. 2020-02-22 13:38:09 +01:00
Bartosz Taudul
9b9474ada1 Request stack traces for execution sampling events. 2020-02-22 13:13:49 +01:00
Bartosz Taudul
28d0f387ea Setup execution sampling profiling. 2020-02-22 13:13:32 +01:00
Bartosz Taudul
ad77b4f73b Store current process id. 2020-02-22 13:11:16 +01:00
Bartosz Taudul
1f671fbacc Keep sys trace variables local. 2020-02-22 13:08:35 +01:00
Bartosz Taudul
539ccf5a61 Check provider id in etw callback. 2020-02-22 12:56:33 +01:00
Bartosz Taudul
0b82902618 Optimize scalar DXT1 compression. 2020-02-15 13:43:40 +01:00
Bartosz Taudul
838c0aaaa9 Check if BUS_MCEERR_AR and BUS_MCEERR_AO are defined. 2020-02-12 01:27:03 +01:00
Bartosz Taudul
2c8d519d70 Fix typo. 2020-02-11 15:12:06 +01:00
Bartosz Taudul
abfa4c65df Update fun list of iDevices. 2020-02-10 16:13:32 +01:00
Bartosz Taudul
8d5f4d7363 Always use init once to initialize rpmalloc. 2020-01-30 20:08:34 +01:00
Bartosz Taudul
885fa16373 Don't retrieve connection id, if zone is not active. 2020-01-25 17:21:30 +01:00
Bartosz Taudul
aa94df0845 Replace rpmalloc_thread_initialize with InitRPMallocThread(). 2020-01-25 17:16:08 +01:00
Bartosz Taudul
ab2fbd6164 Move ParamaterSetup() implementation to header. 2020-01-25 16:51:17 +01:00
Bartosz Taudul
13370dc01c Hide RtlWalkFrameChain inside library. 2020-01-25 16:49:29 +01:00
Bartosz Taudul
a90004b983 Move Set/GetThreadName() to Tracy API. 2020-01-25 16:36:58 +01:00
Bartosz Taudul
6f31eb2a9d Disable MSVC idiocy. 2020-01-20 22:49:03 +01:00
Bartosz Taudul
55d03cb03e Hide async queue setup/commit behind macros. 2020-01-19 15:06:11 +01:00
Bartosz Taudul
25082b2bec Don't report CPU topology if delayed init is active. 
Reporting topology requires producer to be available, which creates a
deadlock during delayed init data structures construction.

Calling GetProducer() results in a call to GetProfilerThreadData(),
which in turn calls GetProfilerData() to construct its thread local
variable. However, at this point we already are calling
GetProfilerData() (to construct the profiler itself). This would result
in an incorrect double construction of data, but the code already
prevents this by allowing init code to be entered only once. Hence the
deadlock.

Currently this is a non-issue, as no platform which can report CPU
topology needs to use delayed init.
 2020-01-14 19:41:34 +01:00
Bartosz Taudul
4f8eb53e8b Capture exact tid to pid mapping on windows. 2020-01-14 02:06:22 +01:00
Bartosz Taudul
4ef2ce4622 Fix _mm256_cvtsi256_si32 on gcc. 2019-12-12 02:13:12 +01:00
Bartosz Taudul
129b80ef0f Free source location, if zone is not active. 2019-12-06 00:42:42 +01:00
Bartosz Taudul
b9cdf2cbb7 Expose srcloc allocation in C API. 2019-12-06 00:25:52 +01:00
Bartosz Taudul
399b87fecc Add allocated srcloc zone begin emit functions to C API. 2019-12-06 00:22:49 +01:00
Bartosz Taudul
68ff33d0ba Extract source location allocation functionality. 2019-12-06 00:15:46 +01:00
Bartosz Taudul
e8fcc250a1 Report CPU topology on Linux. 2019-11-30 01:51:29 +01:00
Bartosz Taudul
712403e9fd Transfer, display, save CPU topology data. 2019-11-29 22:41:41 +01:00
Bartosz Taudul
59371eef5a Obtain CPU topology on windows. 2019-11-29 18:29:31 +01:00
thedmd
a1e2c533f6 libbacktrace: Add support for Mach-O (dSYM) 
`macho.cpp` was backported from official libbacktrace repository.
 2019-11-29 12:04:47 +01:00
Bartosz Taudul
a7d2d5f08b Fix DeferItem() call. 2019-11-26 01:10:50 +01:00
Bartosz Taudul
4551553eb4 Implement setting client parameters from server. 2019-11-25 23:59:48 +01:00
Bartosz Taudul
c5c9dfb0c9 Native callstacks are now optional in allocated callstack messages. 2019-11-25 22:54:10 +01:00
Bartosz Taudul
37eef59d54 Implement reading sys time on BSD. 2019-11-21 20:41:57 +01:00
Bartosz Taudul
c7a22cc1ff Use libbacktrace on BSD. 2019-11-21 20:41:57 +01:00
Bartosz Taudul
bd7b0a8197 Support callstack capture on BSD. 2019-11-21 02:34:42 +01:00
Bartosz Taudul
c79449a6a1 Get proper program name on BSD. 2019-11-21 02:16:12 +01:00
Bartosz Taudul
7940977dba Report physical memory size on BSD. 2019-11-21 02:14:08 +01:00
Bartosz Taudul
3854ae11b2 Revert "Remove dead code." 
This reverts commit a36b73f745.
 2019-11-17 17:38:02 +01:00
Bartosz Taudul
a36b73f745 Remove dead code. 2019-11-16 18:34:05 +01:00
Bartosz Taudul
8286b0b72f Plumbing for message call stacks. 2019-11-14 23:40:41 +01:00
Bartosz Taudul
0befc75f83 Fix conflicts with X.h. 2019-11-14 18:24:29 +01:00
Bartosz Taudul
655864eb7c Enable crash handler on cygwin. 
Crash is properly recorded, but the profiler hangs while waiting for
shutdown finish.
 2019-11-07 19:20:13 +01:00
Bartosz Taudul
3fd74a92f9 Native threads are used on mingw. 2019-11-07 19:02:54 +01:00
Bartosz Taudul
351e220d30 Don't calculate queue delay if delayed init is used. 
Queue calibration requires queue access during profiler construction. This
in turn requires construction of profiler data block, *which at this point
is underway*, because the profiler is being constructed.
 2019-06-19 17:29:04 +02:00
Bartosz Taudul
c98f1f0b6b Make sure profiler is initialized only once in delayed init scenario. 2019-06-19 17:28:18 +02:00
Bartosz Taudul
d4f58ddaf3 Use native windows threads on cygwin, mingw. 2019-11-06 01:42:14 +01:00
Bartosz Taudul
ca198e44d3 Remove dead code from concurrentqueue. 2019-11-05 21:40:52 +01:00
Bartosz Taudul
b5590ed197 Include <mutex> for std::once. 2019-11-05 21:40:35 +01:00
Bartosz Taudul
3e9bb80217 More header cleanup. 2019-11-05 20:15:53 +01:00
Bartosz Taudul
6bbf273581 Partial header inclusion cleanup. 2019-11-05 20:09:40 +01:00
Bartosz Taudul
907574e637 Allow remote plot configuration. 2019-11-05 17:45:19 +01:00
Bartosz Taudul
f34609fd9b Set per-cpu kernel buffer size to 512 KB. 
The default setting was causing events to be lost on Android.
 2019-11-03 21:52:20 +01:00
Bartosz Taudul
b8d459d48b Use proper string size (for consistency). 
On Android code path this value is ignored.
 2019-11-03 21:51:49 +01:00
Bartosz Taudul
ca0fae33d1 Remove obsolete assert. 
Before-terminate-events now include events that have time delta
processing, with no memory to free.
 2019-11-01 20:10:24 +01:00
Bartosz Taudul
1f0c18882c Don't collect sys time after application has exited. 2019-10-29 23:05:14 +01:00
Bartosz Taudul
0f2503d334 Send time deltas in GPU time events. 2019-10-25 19:52:01 +02:00
Bartosz Taudul
8fa5188176 Send delta times for context switches. 2019-10-25 19:13:11 +02:00
Bartosz Taudul
25b3cdc1ee Send thread wakeups when handling disconnect request. 2019-10-25 18:22:42 +02:00
Bartosz Taudul
04b132b6e2 Check if requested data size doesn't overflow buffer. 2019-10-24 21:22:22 +02:00
Bartosz Taudul
ba61a9ed84 Transfer time deltas, not absolute times. 
This change significantly reduces network bandwidth requirements.

Implemented for:
- CPU zones,
- GPU zones,
- locks,
- plots,
- memory events.
 2019-10-24 00:06:41 +02:00
Bartosz Taudul
cf88265304 Full 64-bit register is set by rdtsc. 2019-10-21 01:13:55 +02:00
Bartosz Taudul
07b66cd4ab Move fake source location out of loop. 2019-10-20 22:18:05 +02:00
Bartosz Taudul
909503403b Simplify delay calibration. 2019-10-20 22:13:29 +02:00
Bartosz Taudul
c774534b47 Use rdtsc instead of rdtscp. 
But rdtscp is serializing!

No, it's not. Quoting the Intel Instruction Set Reference:

"The RDTSCP instruction is not a serializing instruction, but it does
wait until all previous instructions have executed and all previous
loads are globally visible. But it does not wait for previous stores to
be globally visible, and subsequent instructions may begin execution
before the read operation is performed.",

"The RDTSC instruction is not a serializing instruction. It does not
necessarily wait until all previous instructions have been executed
before reading the counter. Similarly, subsequent instructions may begin
execution before the read operation is performed."

So, the difference is in waiting for prior instructions to finish
executing. Notice that even in the rdtscp case, execution of the
following instructions may commence before time measurement is finished
and data stores may be still pending.

But, you may say, Intel in its "How to Benchmark Code Execution Times"
document shows that using rdtscp is superior to rdstc. Well, not
exactly. What they do show is that when a *single function* is
considered, there are ways to measure its execution time with little to
no error.

This is not what Tracy is doing.

In our case there is no way to determine absolute "this is before" and
"this is after" points of a zone, as we probably already are inside
another zone.  Stopping the CPU execution, so that a deeply nested zone
may be measured with great precision, will skew the measurements of all
parent zones.

And this is not what we want to measure, anyway. We are not interested
in how a *single function* behaves, but how a *whole program* behaves.
The out-of-order CPU behavior may influence the measurements? Good! We
are interested in that. We want to see *how* the code is really
executed. How is *stopping* the CPU to make a timer read an appropriate
thing to do, when we want to see how a program is performing?

At least that's the theory.

And besides all that, the profiling overhead is now reduced.
 2019-10-20 20:52:33 +02:00
Bartosz Taudul
30fc2f02ab Omit calculation of on-stack variable address. 2019-10-20 19:42:29 +02:00
Bartosz Taudul
c3870f8837 Use proper type. 2019-10-10 20:30:08 +02:00
Bartosz Taudul
707f113bda Add missing NOMINMAX definitions. 2019-10-10 20:29:06 +02:00
Bartosz Taudul
7cf3608493 Avoid unused variables. 2019-10-05 02:11:45 +02:00
Bartosz Taudul
e481b5ba22 Add missing thread sent indication. 2019-10-04 19:18:47 +02:00
Bartosz Taudul
9e1935f070 Make C API symbols visible across dlls. 2019-10-03 22:39:26 +02:00
Bartosz Taudul
130365f4ff Inject tracy_systrace into filesystem and use instead of cat. 
Statistics for a one-minute trace:

  Capture tool | Running time | Running regions
---------------+--------------+-----------------
      cat      |    25.11 s   |     392,300
tracy_systrace |    10.41 s   |      12,249
 2019-09-27 15:51:29 +02:00
Bartosz Taudul
3dba4088ee Embed precompiled tracy_systrace for android. 2019-09-27 15:50:58 +02:00
Bartosz Taudul
e13cbf52fd Allow changing tracy port in client. 2019-09-21 15:11:15 +02:00
Bartosz Taudul
a221f121ba Extract lock state handling to a separate context class. 2019-09-21 14:55:14 +02:00
Bartosz Taudul
37661fd2ee Fix 32 bit NEON version of DXT1 compression. 
This reverts commit b32e8fa24e.

Apparently it is possible to receive non-uniform data in alpha channel, which
breaks the original assumption about not needing the mask. This seemed to be a
problem only on 32 bit NEON implementation of DXT1 compression. Other
implementations handle such data without degradation of visual output.
 2019-09-03 21:37:07 +02:00
Bartosz Taudul
7a6564feae Only recycle producers, if there's no data in queue. 
("The queue" is per-thread partial queue here.)

This fixes a problem where one thread writes to the queue, then is
terminated, making the (partially filled) queue available for other
threads to recycle. If another thread re-owns the queue, it will change
the associated thread id, while part of the queue was filled by the
original thread. This obviously created invalid data during dequeue.

The fix makes the recycling process check not only for queue inactivity
(which is marked when the original thread terminates), but also if the
queue is empty, preventing mixing data from different threads.
 2019-08-30 14:28:44 +02:00
Bartosz Taudul
00b26c1acf Fix TRACY_NO_SYSTEM_TRACING. 2019-08-26 18:02:10 +02:00
Bartosz Taudul
fbeee3cf61 Fix (?) invalid function pointer signature. 2019-08-26 17:59:58 +02:00
Bartosz Taudul
78127dc357 System threads only allow limited information queries. 2019-08-25 00:33:22 +02:00
Bartosz Taudul
deb59b4c38 Somehow fix event ordering. 2019-08-24 01:43:55 +02:00
Bartosz Taudul
1e74a89924 Check if there's data to read from kernel. 
Reading from kernel pipe, while being a blocking operation, spin locks the
thread.
 2019-08-24 01:06:21 +02:00
Bartosz Taudul
8f6e94d75c Sleep if sys trace pipe buffer underruns. 2019-08-24 00:42:00 +02:00
Bartosz Taudul
2d50d07438 Allow completely disabling system tracing. 2019-08-21 01:16:25 +02:00
Bartosz Taudul
0cbb853945 Add missing SetThreadName() calls. 2019-08-20 16:23:00 +02:00
Bartosz Taudul
332262dd84 Shorter thread names. 2019-08-20 16:22:54 +02:00