Support for user-defined cost in cache classes #982
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
At cache level, "size" is renamed to "cost", representing how much a cache can store (and how much an item cost). "size" is preserved for the count of items. At higher level, we keep the "size" sementics as we are speaking about the size of the cache, whatever this is. This is the first step to a cache limited by memory usage.
... and enhanced two of the lru_cache unittests with log_debug-assisted testing.
The changes in the unit tests are better understood if whitespace changes are ignored (git diff -w).
This test too intermittently fails (likely to the suspected but not yet
tracked-down race condition in logging and/or namedthread code). This
time however it fails either due to a deadlock, or obvious memory corruption
like below:
[ RUN ] ConcurrentCacheMultithreadedTest.accessSameExistingItem
../../../../home/leon/freelancing/kiwix/builddir/SOURCE/libzim/test/concurrentcache.cpp:429: Failure
Expected equality of these values:
zim::Logging::getInMemLogContent()
Which is: "thread#0: Output of interest starts from the next line\na : ConcurrentCache::getOrPut(5) {\n b: ConcurrentCache::getOrPut(5) {\na : ConcurrentCache::getCacheSlot(5) {\na : entered synchronized section\n b: ConcurrentCache::getCacheSlot(5) {\na : lru_cache::getOrPut(5) {\na : already in cache, moved to the beginning of the LRU list.\na : }\na : exiting synchronized section\na : }\n b: entered synchronized section\na : Obtained the cache slot\n b: lru_cache::getOrPut(5) {\na : } (return value: 50)\n\0 b: already in cache, moved to the beginning of the LRU list.\n b: }\n b: exiting synchronized section\n b: }\n b: Obtained the cache slot\n b: } (return value: 50)\n\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0 "
targetOutput
Which is: "thread#0: Output of interest starts from the next line\na : ConcurrentCache::getOrPut(5) {\n b: ConcurrentCache::getOrPut(5) {\na : ConcurrentCache::getCacheSlot(5) {\na : entered synchronized section\n b: ConcurrentCache::getCacheSlot(5) {\na : lru_cache::getOrPut(5) {\na : already in cache, moved to the beginning of the LRU list.\na : }\na : exiting synchronized section\na : }\n b: entered synchronized section\na : Obtained the cache slot\n b: lru_cache::getOrPut(5) {\na : } (return value: 50)\n b: already in cache, moved to the beginning of the LRU list.\n b: }\n b: exiting synchronized section\n b: }\n b: Obtained the cache slot\n b: } (return value: 50)\n"
With diff:
@@ -14,5 +14,5 @@
b: lru_cache::getOrPut(5) {
a : } (return value: 50)
-\0 b: already in cache, moved to the beginning of the LRU list.
+ b: already in cache, moved to the beginning of the LRU list.
b: }
b: exiting synchronized section
@@ -19,4 +19,3 @@
b: }
b: Obtained the cache slot
- b: } (return value: 50)
-\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
+ b: } (return value: 50)\n
[ FAILED ] ConcurrentCacheMultithreadedTest.accessSameExistingItem (2 ms)
If an item that is in the process of being added to the cache is requested from the cache within the time window when it has been (temporarily) evicted from the cache (as a result of the concurrent turmoil), a concurrent evaluation of the same item is started while the previous one is still in progress.
An item being added to the cache is never evicted due to concurrent turmoil. LRU eviction heuristics was changed to only evict items with non-zero cost.
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #982 +/- ##
==========================================
- Coverage 57.97% 57.65% -0.32%
==========================================
Files 103 103
Lines 4850 4922 +72
Branches 2018 2066 +48
==========================================
+ Hits 2812 2838 +26
- Misses 704 706 +2
- Partials 1334 1378 +44 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
Collaborator
Author
|
There seems to be a bug in concurrency orchestration of |
Contributor
|
@veloman-yunkan OK, so I guess this is something to fix before merging? |
Collaborator
Author
@kelson42 Yes, found the bug and filed it as #983. Will fix it tomorrow. |
Contributor
|
@veloman-yunkan Considering you will treat the issue separatly, does that mean this PR is ready? |
Collaborator
Author
@kelson42 Yes, it is. |
Collaborator
Author
Bug-fix is ready (#984). |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR introduces support for user-defined cost to
lru_cacheandConcurrentCacheclasses as an enhancement enabling to fix #947. It was extracted from #960 which in turn was based on #956.