If lots of threads do lots of malloc/free and they overflow
per-pthread DenseSlabAlloc cache, it causes lots of contention:
31.97% race.old race.old [.] __sanitizer::StaticSpinMutex::LockSlow
17.61% race.old race.old [.] __tsan_read4
10.77% race.old race.old [.] __tsan::SlotLock
Optimize DenseSlabAlloc to use a lock-free stack of batches of nodes.
This way we don't take any locks in steady state at all and do only
1 push/pop per Refill/Drain.
Effect on the added benchmark:
$ TIME="%e %U %S %M" time ./test.old 36 5 2000000
34.51 978.22 175.67 5833592
32.53 891.73 167.03 5790036
36.17 1005.54 201.24 5802828
36.94 1004.76 226.58 5803188
$ TIME="%e %U %S %M" time ./test.new 36 5 2000000
26.44 720.99 13.45 5750704
25.92 721.98 13.58 5767764
26.33 725.15 13.41 5777936
25.93 713.49 13.41 5791796
Reviewed By: melver
Differential Revision: https://reviews.llvm.org/D130002
This change switches tsan to the new runtime which features:
- 2x smaller shadow memory (2x of app memory)
- faster fully vectorized race detection
- small fixed-size vector clocks (512b)
- fast vectorized vector clock operations
- unlimited number of alive threads/goroutimes
Depends on D112602.
Reviewed By: melver
Differential Revision: https://reviews.llvm.org/D112603
This change switches tsan to the new runtime which features:
- 2x smaller shadow memory (2x of app memory)
- faster fully vectorized race detection
- small fixed-size vector clocks (512b)
- fast vectorized vector clock operations
- unlimited number of alive threads/goroutimes
Depends on D112602.
Reviewed By: melver
Differential Revision: https://reviews.llvm.org/D112603
This change switches tsan to the new runtime which features:
- 2x smaller shadow memory (2x of app memory)
- faster fully vectorized race detection
- small fixed-size vector clocks (512b)
- fast vectorized vector clock operations
- unlimited number of alive threads/goroutimes
Depends on D112602.
Reviewed By: melver
Differential Revision: https://reviews.llvm.org/D112603
This change switches tsan to the new runtime which features:
- 2x smaller shadow memory (2x of app memory)
- faster fully vectorized race detection
- small fixed-size vector clocks (512b)
- fast vectorized vector clock operations
- unlimited number of alive threads/goroutimes
Differential Revision: https://reviews.llvm.org/D112603
This change switches tsan to the new runtime which features:
- 2x smaller shadow memory (2x of app memory)
- faster fully vectorized race detection
- small fixed-size vector clocks (512b)
- fast vectorized vector clock operations
- unlimited number of alive threads/goroutimes
Depends on D112602.
Reviewed By: melver
Differential Revision: https://reviews.llvm.org/D112603
This change switches tsan to the new runtime which features:
- 2x smaller shadow memory (2x of app memory)
- faster fully vectorized race detection
- small fixed-size vector clocks (512b)
- fast vectorized vector clock operations
- unlimited number of alive threads/goroutimes
Depends on D112602.
Reviewed By: melver
Differential Revision: https://reviews.llvm.org/D112603
Include info about MBlock/SyncObj memory consumption in the memory profile.
Depends on D110148.
Reviewed By: melver, vitalybuka
Differential Revision: https://reviews.llvm.org/D110149
Now that sanitizer_common mutex has feature-parity with tsan mutex,
switch tsan to the sanitizer_common mutex and remove tsan's custom mutex.
Reviewed By: vitalybuka, melver
Differential Revision: https://reviews.llvm.org/D106379
We've got a user report about heap block allocator overflow.
Bump the L1 capacity of all dense slab allocators to maximum
and be careful to not page the whole L1 array in from .bss.
If OS uses huge pages, this still may cause a limited RSS increase
due to boundary huge pages, but avoiding that looks hard.
Reviewed By: vitalybuka
Differential Revision: https://reviews.llvm.org/D101161
to reflect the new license.
We understand that people may be surprised that we're moving the header
entirely to discuss the new license. We checked this carefully with the
Foundation's lawyer and we believe this is the correct approach.
Essentially, all code in the project is now made available by the LLVM
project under our new license, so you will see that the license headers
include that license only. Some of our contributors have contributed
code under our old license, and accordingly, we have retained a copy of
our old license notice in the top-level files in each project and
repository.
llvm-svn: 351636
check_memcpy test added in r254959 fails on some configurations due to
memset() calls inserted by Clang. Try harder to avoid them:
* Explicitly use internal_memset() instead of empty braced-initializer.
* Replace "new T()" with "new T", as the former generates zero-initialization
for structs in C++11.
llvm-svn: 255136
The new storage (MetaMap) is based on direct shadow (instead of a hashmap + per-block lists).
This solves a number of problems:
- eliminates quadratic behaviour in SyncTab::GetAndLock (https://code.google.com/p/thread-sanitizer/issues/detail?id=26)
- eliminates contention in SyncTab
- eliminates contention in internal allocator during allocation of sync objects
- removes a bunch of ad-hoc code in java interface
- reduces java shadow from 2x to 1/2x
- allows to memorize heap block meta info for Java and Go
- allows to cleanup sync object meta info for Go
- which in turn enabled deadlock detector for Go
llvm-svn: 209810
Dmitry Vyukov