summaryrefslogtreecommitdiffstats
path: root/vespalib
Commit message (Collapse)AuthorAgeFilesLines
* Merge pull request #21469 from ↵Geir Storli2022-03-014-44/+183
|\ | | | | | | | | vespa-engine/toregge/compact-indirect-keys-and-values-in-btree-stress-test Compact indirect keys and values in btree stress test.
| * Compact indirect keys and values in btree stress test.Tor Egge2022-02-284-44/+183
| |
* | Merge pull request #21467 from ↵Henning Baldersheim2022-03-013-0/+27
|\ \ | |/ |/| | | | | vespa-engine/balder/add-and-use-common-adjustTimeoutByDetectedHz Add a simple adjustTimeoutByDetectedHz to compute a suitable timeout/…
| * Add a simple adjustTimeoutByDetectedHz to compute a suitable timeout/naptime ↵Henning Baldersheim2022-02-283-0/+27
| | | | | | | | | | | | based on base VESPA_TIMER_HZ to use for random waiting when doing a poll based appraoch.
* | Restore atomic thread fence in vespalib::GenerationHandler::incGeneration().Tor Egge2022-02-281-0/+5
|/
* Merge pull request #21409 from ↵Geir Storli2022-02-281-19/+29
|\ | | | | | | | | vespa-engine/toregge/use-atomic-instructions-to-get-and-set-btree-internal-child-refs Readers of a btree use atomic read with acquire memory ordering when
| * Readers of a btree use atomic read with acquire memory ordering when gettingTor Egge2022-02-251-19/+29
| | | | | | | | | | | | | | | | btree internal node child ref. Writer of a btree uses atomic write with release memory ordering when setting btree internal node child ref in node that is visible to readers (e.g. when moving child node as part of compacting btree nodes).
* | Remove pointer type mismatch errors (vespalib).Tor Egge2022-02-271-10/+12
| |
* | Merge pull request #21414 from ↵Henning Baldersheim2022-02-262-3/+3
|\ \ | | | | | | | | | | | | vespa-engine/toregge/add-noexcept-specifiers-in-vespalib Add noexcept specifiers (vespalib).
| * | Add noexcept specifiers (vespalib).Tor Egge2022-02-262-3/+3
| | |
* | | Remove inlining warnings (vespalib).Tor Egge2022-02-2614-1/+60
|/ /
* | Merge pull request #21405 from ↵Henning Baldersheim2022-02-251-16/+16
|\ \ | | | | | | | | | | | | vespa-engine/toregge/stop-trying-to-compare-arrays Stop trying to compare arrays (deprecated in C++20)
| * | Stop trying to compare arrays (deprecated in C++20)Tor Egge2022-02-251-16/+16
| | |
* | | Merge pull request #21404 from ↵Henning Baldersheim2022-02-252-3/+3
|\ \ \ | |_|/ |/| | | | | | | | vespa-engine/toregge/stop-using-std-unary-function-and-std-binary-function Stop using std::unary_function and std::binary_function.
| * | Stop using std::unary_function and std::binary_function.Tor Egge2022-02-252-3/+3
| |/
* / Add member functions to get and set btree internal node child refs that willTor Egge2022-02-255-22/+24
|/ | | | use relaxed memory order instead of acquire and release.
* Merge pull request #21397 from vespa-engine/toregge/add-include-for-std-findHenning Baldersheim2022-02-251-0/+1
|\ | | | | Add include to get declaration of std::find
| * Add include to get declaration of std::findTor Egge2022-02-251-0/+1
| |
* | Prepare for different semantics for getData and getChild member functionsTor Egge2022-02-252-3/+3
|/ | | | in internal btree node.
* Extend btree stress test with compaction of tree.Tor Egge2022-02-251-2/+38
|
* Eliminate gcc compiler warnings when using old version of gtest (< 1.11).Tor Egge2022-02-241-0/+8
|
* Extend btree stress test with indirect keys and values.Tor Egge2022-02-241-42/+195
|
* Convert btree stress test to gtest.Tor Egge2022-02-242-41/+46
|
* Merge pull request #21343 from ↵Tor Egge2022-02-244-35/+34
|\ | | | | | | | | vespa-engine/toregge/use-atomic-ops-for-reading-and-writing-data-store-buffer-pointer Use atomic read with acquire memory ordering when reading data store …
| * Remove unused member function.Tor Egge2022-02-231-1/+0
| |
| * Use atomic read with acquire memory ordering when reading data store buffer ↵Tor Egge2022-02-234-35/+35
| | | | | | | | | | | | pointer in reader thread. Use atomic write with release memory ordering when updating data store buffer pointer in writer thread.
* | Move btree scan speed test to vespalib.Tor Egge2022-02-233-0/+191
| |
* | Merge pull request #21355 from vespa-engine/toregge/track-move-to-vespalibHenning Baldersheim2022-02-231-1/+1
|\ \ | | | | | | Track move to vespalib.
| * | Track move to vespalib.Tor Egge2022-02-231-1/+1
| |/
* / Move btree stress test to vespalib.Tor Egge2022-02-233-0/+234
|/
* Use relaxed atomic write when morphing hold entry to new generation.Tor Egge2022-02-221-1/+1
|
* Fix race conditions in vespalib::GenerationHandler discovered by thread ↵Tor Egge2022-02-224-50/+23
| | | | sanitizer.
* Expose current active operation throttler token countTor Brede Vekterli2022-02-223-6/+31
| | | | | | | | | | Allows for inspecting _actual_ operation parallelism vs. the _maximum_ parallelism given by the internal policy window size. Remove the note stating that waiting thread count is for testing only, as we want to be able to expose that as well. Available for both the dynamic and the unlimited implementations, as they both explicitly track active token counts.
* Disable mmap resource limit unit test on Darwin.Tor Egge2022-02-211-3/+3
|
* Make vespalib::Thread state polling thread safeTor Brede Vekterli2022-02-182-6/+9
|
* Form a full release/acquire pair on lock-free hash map shardsTor Brede Vekterli2022-02-181-6/+6
| | | | | | When doing a lock-free read of the underlying `_map` we must form an acquire pair with the release done in `ShardedHashMap::alloc_shard` to ensure we observe all transitive writes within the map shard itself.
* Merge pull request #21266 from ↵Henning Baldersheim2022-02-183-0/+27
|\ | | | | | | | | vespa-engine/balder/add-method-to-get-default-timer-frequency Add a base vespa timer frequency that is used as a base to control po…
| * Add a base vespa timer frequency that is used as a base to control poll ↵Henning Baldersheim2022-02-173-0/+27
| | | | | | | | frequency of various tasks.
* | Handle 64K page size in alloc unit test and rcuvector unit test.Tor Egge2022-02-172-27/+42
|/
* Changes after review feedbackHarald Musum2022-02-171-7/+11
|
* Handle diffrent page sizeHarald Musum2022-02-171-4/+5
|
* Merge pull request #21239 from ↵Geir Storli2022-02-171-2/+14
|\ | | | | | | | | vespa-engine/toregge/detect-dangling-shared-operation-throttler-tokens-when-using-no-limits-operation-throttler Detect dangling shared operation throttler token when using no limits…
| * Detect dangling shared operation throttler token when using no limits ↵Tor Egge2022-02-161-2/+14
| | | | | | | | operation throttler.
* | Use `lock_guard` instead of `unique_lock`Tor Brede Vekterli2022-02-161-1/+1
| |
* | Make CountDownLatch debug function thread safeTor Brede Vekterli2022-02-161-5/+8
| |
* | Fix memory ordering argumentTor Brede Vekterli2022-02-161-1/+1
| |
* | Use standard feature test for atomic_ref support instead of ad-hoc macroTor Brede Vekterli2022-02-161-11/+7
| |
* | Add vespalib utility functions for atomic memory accessTor Brede Vekterli2022-02-162-12/+160
|/ | | | | | | | | Adds the following utilities: * Atomic reference wrapper functions for accessing single memory locations as if they were `std::atomic` instances. * Wrappers for less verbose `std::atomic` loads/stores that also sanity-check that accesses are always lock-free.
* Merge pull request #21209 from ↵Håvard Pettersen2022-02-161-4/+13
|\ | | | | | | | | vespa-engine/vekterli/well-defined-race-cond-checking-in-spinlock-test Use atomic refs to make spin lock race condition testing well-defined
| * Use atomic refs to make spin lock race condition testing well-definedTor Brede Vekterli2022-02-151-4/+13
| | | | | | | | Relaxed loads/stores give the same effect but without the undefined-ness.