[compiler-rt] [scudo] Update secondary cache time-based release logic. (PR #107507)
via llvm-commits
llvm-commits at lists.llvm.org
Mon Sep 9 11:57:42 PDT 2024
================
@@ -590,35 +596,36 @@ class MapAllocatorCache {
}
}
- void releaseIfOlderThan(CachedBlock &Entry, u64 Time) REQUIRES(Mutex) {
- if (!Entry.isValid() || !Entry.Time)
- return;
- if (Entry.Time > Time) {
- if (OldestTime == 0 || Entry.Time < OldestTime)
- OldestTime = Entry.Time;
- return;
- }
+ inline void release(CachedBlock &Entry) {
+ DCHECK(Entry.Time != 0);
Entry.MemMap.releaseAndZeroPagesToOS(Entry.CommitBase, Entry.CommitSize);
Entry.Time = 0;
}
void releaseOlderThan(u64 Time) EXCLUDES(Mutex) {
ScopedLock L(Mutex);
- if (!EntriesCount || OldestTime == 0 || OldestTime > Time)
+ if (!EntriesCount)
return;
- OldestTime = 0;
- for (uptr I = 0; I < Config::getQuarantineSize(); I++)
- releaseIfOlderThan(Quarantine[I], Time);
- for (uptr I = 0; I < Config::getEntriesArraySize(); I++)
- releaseIfOlderThan(Entries[I], Time);
- }
+ for (uptr I = 0; I < Config::getQuarantineSize(); I++) {
----------------
ChiaHungDuan wrote:
hmmm, I think we should avoid scan the quarantine entries here. I think we can have a function like
```
bool IsQuarantineEnabled() {
return useMemoryTagging<Config>(Options) && QuarantinePos != -1U;
}
```
to bypass scanning the quarantine entries. It doesn't need to be done in this CL by the way.
https://github.com/llvm/llvm-project/pull/107507
More information about the llvm-commits
mailing list