Commit Graph

1085 Commits

Author SHA1 Message Date
Lennox Ho
d3602c0672 Account for Windows while performing implicit TLS detection in CMakeLists.txt 2023-09-17 07:49:38 +08:00
Aliaksey Kandratsenka
dffb4a2f28 bump version to 2.13 2023-09-11 16:23:40 -04:00
Aliaksey Kandratsenka
4ec8c9dbb2 reduce set of nallocx size testing points
Testing every 7th size is a bit slow on slower machines. No need to be
as thorough. We now bump by about 1/128th each step which is still
more steps than size classes we have.
2023-09-10 22:18:51 -04:00
Aliaksey Kandratsenka
e4e7ba93a0 unbreak unnecessary dependency on 64-bit atomics
This unbreaks builds on 32-bit arms and mipsen.
2023-09-10 21:07:28 -04:00
Aliaksey Kandratsenka
523b72f754 make sampling_debug_test actually test debug malloc
We do shell wrapper for actual test run, so we can inspect output of
pprof. But when we set up sampling_debug_test.sh we simply copied
regular sampling_test.sh, which ran same non-debug test binary. Now we
sed-replace contents of shell program when copying, so we test right
binary.

Another thing we fix here is our (still hardcoded) test output path is
now different between sampling{,_debug}_test.sh. So this fixes main
cause of flakiness of our unit tests.
2023-09-10 18:14:57 -04:00
Aliaksey Kandratsenka
2748dd5680 unbreak address access "probing" for generic_fp backtracing
We used msync to verify that address is readable. But msync gives
false positives for PROT_NONE mappings. And we recently got bug report
from user hitting this exact condition.

For correct access check, we steal idea from Abseil and do sigprocmask
with address used as new signal mask and with invalid HOW
argument. This works in today's Linux kernels and is among fastest
methods available. But is brittle w.r.t. possible kernel changes. So
we supply fallback method that does 2 syscalls.

For non-Linux systems we implement usual "write to pipe" trick. Which
also has decent performance, but requires occasional pipe draining and
uses fds which could occasionally be damaged by some forking codes.

We also finally cover all new code with unit test.

Fixes github issue #1426
2023-09-10 17:24:32 -04:00
Ivan Dlugos
7ad1dc7693 fix: cmake config.h defines declaration 2023-09-08 14:38:21 -04:00
Aliaksey Kandratsenka
f7172839a1 turn tcmalloc::TrivialOnce into POD
As we see in github issue #1428, msvc arranges full "init on first
use" initialization for local static usage of TrivialOnce even if that
initialization is completely empty. Fair game, even if stupid.

POD with no initialization should be safely zero-initialized with no
games or tricks from the compilers.

We could have and perhaps at some point should do constexpr for
TrivialOnce and SpinLock (abseil has been liberated from
LinkerInitialized for perphaps well over decade now, including their
fork of SpinLock, of course). But C++ legalese rules are complex
enough and bugs happened in past, so I don't want to be in the tough
business of interpreting standard. So at least for now we keep
things simple.
2023-09-08 14:22:46 -04:00
Aliaksey Kandratsenka
539ed9ca40 bump version 2.12 2023-08-24 15:03:47 -04:00
Aliaksey Kandratsenka
8d634c1f56 don't build mmap_hook when --enable-minimal is given to configure
Refers to github issue #1418
2023-08-24 14:06:25 -04:00
Brett T. Warden
b4ad04982d Set Description field in generated pkg-config files (instead of Summary)
Fixes #1416
2023-08-22 18:09:33 -04:00
Aliaksey Kandratsenka
0a3ca5b43d bump version to 2.11 2023-08-14 22:47:56 -04:00
Aliaksey Kandratsenka
83fccceffa bump README freshness a bit 2023-08-14 22:05:47 -04:00
Ken Raffenetti
c41eb9e8b5 Add MPICH HPC environment detection
Default MPICH builds use the Hydra process manager (mpiexec) which sets
PMI_RANK in the application environment. Update GetUniquePathFromEnv()
test accordingly.

Signed-off-by: Ken Raffenetti <raffenet@mcs.anl.gov>
2023-08-11 15:21:15 -04:00
Aliaksey Kandratsenka
1d2654f3a0 heap-checker: unbreak PTRACE_GETREGS detection on older Linux-es
This unbreaks RHEL6.
2023-08-10 14:30:27 -04:00
Aliaksey Kandratsenka
dbd1071680 link libprofiler with pthread
This unbreaks building on older Linux distros. We missed this at
46d3315ad7 when dropped maybe_thread
stuff, since libprofiler indeed uses pthread, and because on newer
libc-s pthread stuff is now part of regular libc.so.

I am also dropping bogus LIBPROFILER stuff referring to some rpath
badness. Unsure what it was, maybe way back we did libstacktrace as a
proper libtool library, so maybe something was needed. But it is just
a convenience archive this days, so we don't really need to add it
everywhere libprofiler.la is linked.
2023-08-09 23:57:06 -04:00
Aliaksey Kandratsenka
729383b486 make sure that ListerThread runs on properly aligned stack
Without this fix we're failing unit tests on ubuntu 18.04 and centos 7
and 6. It looks like clone() in old glibc-s doesn't align stack, so
lets handle it ourselves. How we didn't hit this much earlier (before
massive thread listing refactoring), I am not sure. Most likely pure
luck(?)
2023-08-09 23:42:56 -04:00
Aliaksey Kandratsenka
51c5e2bec7 massage latest GetUniquePathFromEnv changes
This fixes a number of minor bits (like build details) as well as
making overall code style similar to what we're doing elsewhere.
2023-08-09 16:29:13 -04:00
Artem Polyakov
86450ad99f Add unit test for GetUniquePathFromEnv()
Signed-off-by: Artem Polyakov <artpol84@gmail.com>
2023-08-08 16:44:18 -07:00
Artem Y. Polyakov
881b754da0 Advanced UniquePathFromEnv generation
* Add support for known HPC environments (TODO: needs to be extended
with more nevironments)
* Added the "CPUPROFILE_USE_PID" environment variable to force appending
PID for the non-covered environments
* Preserve the old way of handling the Child-Parent case

Signed-off-by: Artem Polyakov <artpol84@gmail.com>
2023-08-08 16:44:18 -07:00
Aliaksey Kandratsenka
57512e9c3d unbreak -Wthread-safety
It actually found real (but arguably minor) issue with memory region
map locking.

As part of that we're replacing PageHeap::DeleteAndUnlock that had
somewhat ambitious 'move' of SpinLockHolder, with more straightforward
PageHeap::PrepareAndDelete. Doesn't look like we can support move
thingy with thread annotations.
2023-08-06 19:32:32 -04:00
Aliaksey Kandratsenka
862039c185 don't -momit-leaf-frame-pointer when asked for full frame pointers 2023-08-06 14:44:05 -04:00
Aliaksey Kandratsenka
dc25c1fd4c bump version to 2.11rc 2023-07-31 20:11:55 -04:00
Aliaksey Kandratsenka
909fa3e649 unbreak MallocExtension::GetAllocatedSize() for debug allocator
Some years back we fixed memalign vs realloc bug, but we preserved
'wrong' malloc_size/GetAllocatedSize implementation for debug
allocator.

This commit refactors old code making sure we always use right
data_size and it fixes GetAllocatedSize. We update our unittest
accordingly.

Closes #738
2023-07-31 16:28:48 -04:00
Aliaksey Kandratsenka
a51e08b06a drop obsolete TODO file 2023-07-31 15:40:13 -04:00
Aliaksey Kandratsenka
a1c7ce7793 refresh README-s 2023-07-31 15:16:14 -04:00
Aliaksey Kandratsenka
bef6592746 refresh INSTALL 2023-07-31 15:10:56 -04:00
Aliaksey Kandratsenka
e3de2e3242 remove obsolete references to code.google.com
I.e. somehow we managed to still point to (very) old gperftools
hosting location, so lets fix it at last.
2023-07-31 14:28:58 -04:00
Aliaksey Kandratsenka
1ff09a680e drop obsolete deb/rpm packaging stuff 2023-07-31 14:28:40 -04:00
Aliaksey Kandratsenka
8b3f0d6145 undo MarkThreadTemporarilyIdle and make it same as MarkThreadIdle
As noted on github issue #880 'temporarily' thing saves us not just on
freeing thread cache, but also returning thread's share of thread
cache (max_size_) into common pool. And the later has caused trouble
to mongo folk who originally proposed 'temporarily' thing. They claim
they don't use it anymore.

And thus with no users and no clear benefit, it makes no sense for us
to keep this API. For API and ABI compat sake we keep it, but it is
now identical to regular MarkThreadIdle.

Fixes issue #880
2023-07-31 14:11:51 -04:00
Aliaksey Kandratsenka
c3059a56be dont workaround unknown problem in thread_dealloc_unittest
We had some sleep added at the end up thread dealloc unittest claiming
some race trouble with glibc. Which is likely years or even decades
irrelevant.
2023-07-31 12:37:15 -04:00
Aliaksey Kandratsenka
9b91ce917a [win32:patching] define single empty __expand replacement
This unbreaks some cases where patching complains about too short
functions to patch.

What happens is we first locate one of CRT-s (like ucrt or msvcrt) and
patch __expand there, redirecting to our implementation. Then "static"
__expand replacement is patched, but it is usually imported from that
same C runtime DLL. And through several jmp redirections we end up at
our own __expand from libc<1>. Patching that (and other cases) is
wrong, but I am unsure how to fix it properly. So we do most simple
workaround. I found that when it not fails is either in debug builds
where empty expand is not too short or when MSVC deduplicates multiple
identical __expand implementations into single function, or when
64-bit patching has to do extra trampoline thingy. And then our
patching code checks if we're trying to replace some function with
itself. So we "just" take advantage of that and get immediate issue
fixed, while punting on more general "duplicate" patching for later.

Update github issue #667
2023-07-27 19:38:52 -04:00
Aliaksey Kandratsenka
b6cdd8f510 [mingw] dont add libstacktrace to libtcmalloc_minimal
There was this piece of makefile with indention to add stack tracing
functionality (for stuff like growthz, GetCallerStackTrace and
probably heap sampling) to work even in minimal configuration on
mingw.

What is odd is we fail to actually define libstacktrace.la target on
mingw, since libstacktrace.la requires WITH_STACK_TRACE automake
conditional which we don't enable on this platform. And yet somehow it
doesn't fail. It produces empty libstacktrace.la, so build kinda
works. Except at least on my machine it produces racy makefiles. So
lets not pretend and stop breaking our parallel builds.
2023-07-27 19:27:42 -04:00
Aliaksey Kandratsenka
a5cfd38884 [win32] amend and unbreak previous NOMINMAX fix 2023-07-27 19:27:20 -04:00
Aliaksey Kandratsenka
d2c89ba534 don't return raw span when sampling and stacktrace oomed
This is nearly impossible in practice, but still. Somehow we missed
this logic that DoSampledAllocation always returns actual object, but
in that condition where stacktrace_allocator failed to get us
StackTrace object we ended up returning span instead of it's object.
2023-07-24 21:01:35 -04:00
Aliaksey Kandratsenka
59464404d1 capture growthz backtraces outside of pageheap_lock
Actual growthz list is now lockless since we never delete anything
from it. And we now pass special 'locking context' object down page
heap allocation path, both as a documentation that it is under lock
and for tracking whether we needed to grow heap and by how much. Then
whenever lock is released in RAII fashion, we're able to trigger
growthz recording outside of lock.

Closes #1159
2023-07-24 21:01:35 -04:00
Aliaksey Kandratsenka
0d42a48699 move page heap locking under PageHeap
While there is still plenty of code that takes pageheap_lock outside
of page_heap module for all kinds of reasons, at least
bread-and-butter logic of allocating/deallocating larger chunks of
memory is now handling page heap locking inside PageHeap itself. This
gives us flexibility.

Update issue #1159
2023-07-24 21:01:35 -04:00
Aliaksey Kandratsenka
a3e1080c2e handle large alloc reporting locklessly
Which simplifies codes a bit.

Update issue #1159
2023-07-24 21:01:35 -04:00
Aliaksey Kandratsenka
f1eb3c82c6 correctly release memory when system's pagesize is >kPageSize
I.e. this covers case of arms that by default compile tcmalloc for 8k
logical pages (assuming 4k system pages), but can actually run on
systems with 64k pages.

Closes #1135
2023-07-24 21:01:35 -04:00
Aliaksey Kandratsenka
d521e3b30e move page heap allocations with alignment into page heap 2023-07-24 21:01:35 -04:00
Aliaksey Kandratsenka
ad0ca2b83b unbreak large heap fragmentation unittest
Smart compilers again (and lack of -fno-builtin-malloc which we
dropped because of clang).
2023-07-24 20:24:52 -04:00
Aliaksey Kandratsenka
bd3bf9abd7 define NOMINMAX on windows
Otherwise we cannot use some std bits, like numeric_limits.
2023-07-24 20:24:28 -04:00
Aliaksey Kandratsenka
814f340eed add --enable-libgcc-unwinder-by-default for configure
We want distros to ship gperftools with this options, now that they
have _Unwind_Backtrace that is powered by dl_find_object.
2023-07-22 20:45:52 -04:00
Aliaksey Kandratsenka
f06ccc6f79 dont test HAVE_{STDINT,INTTYPES}_H
Those are fairly standard by now. We already require C++11 or later
compiler.
2023-07-22 14:32:40 -04:00
Aliaksey Kandratsenka
86867674f0 don't do syscall on osx
Their compiler barks and we only need this syscall business on FreeBSD.
2023-07-22 14:32:40 -04:00
Aliaksey Kandratsenka
bdae3615ca osx's dyld_present is always true
As reported to us by their compiler's warning.
2023-07-22 14:32:34 -04:00
Aliaksey Kandratsenka
91b8024aa1 drop unused ThreadCache::DropCacheWhichMustBePresent 2023-07-22 14:00:32 -04:00
Aliaksey Kandratsenka
42dde1b883 heap-checker_unittest: run thread-unsafe libc calls less unsafely
Comment there says race is fine, bit it isn't. Crashes under
musl. Lets not crash.
2023-07-22 00:08:38 -04:00
Aliaksey Kandratsenka
8be84e4a5c drop old mmap hooks and introduce internal & simpler mmap_hook.h
Previous implementation wasn't entirely safe w.r.t. 32-bit off_t
systems. Specifically around mmap replacement hook. Also, API was a
lot more general and broad than we actually need.

Sadly, old mmap hooks API was shipped with our public headers. But
thankfully it appears to be unused externally (checked via github
search). So we keep this old API and ABI for the sake of formal API
and ABI compatibility. But this old API is now empty and always
fails (some OS/hardware combinations didn't have functional
implementations of those hooks anyways).

New API is 64-bit clean and only provides us with what we need. Namely
being able to react to virtual address space mapping changes for
logging, heap profiling and heap leak checker. I.e. no pre hooks or
mmap-replacement hooks. We also explicitly not ship this API
externally to give us freedom to change it.

New code is also hopefully tidier and slightly more portable. At least
there are fewer arch-specific ifdef-s.

Another somewhat notable change is, since mmap hook isn't needed in
"minimal" configuration, we now don't override system's
mmap/munmap/etc functions in this configuration. No big deal, but it
reduces risk of damage if we somehow mess those up. I.e. musl's mmap
does few things that our mmap replacement doesn't, such as very fancy
vm_lock thingy. Which doesn't look critical, but is good thing for us
not to interfere with when not necessary.

Fixes issue #1406 and issue #1407. Lets also mention issue #1010 which
is somewhat relevant.
2023-07-21 16:13:19 -04:00
Aliaksey Kandratsenka
6b691cc019 drop patching virtual alloc fns on windows
It wasn't implemented exactly right and, most importantly, there is
nothing on windows that uses those (now dead) mmap hooks.
2023-07-21 14:32:48 -04:00