[llvm] PerThreadBumpPtrAllocator: remove dependency on getThreadIndex (PR #209687)
Alexis Engelke via llvm-commits
llvm-commits at lists.llvm.org
Wed Jul 15 00:47:56 PDT 2026
================
@@ -41,25 +53,45 @@ class PerThreadAllocator
/// Allocate \a Size bytes of \a Alignment aligned memory.
void *Allocate(size_t Size, size_t Alignment) {
- assert(getThreadIndex() < NumOfAllocators);
- return Allocators[getThreadIndex()].Allocate(Size, Alignment);
+ return getThreadLocalAllocator().Allocate(Size, Alignment);
}
/// Deallocate \a Ptr to \a Size bytes of memory allocated by this
/// allocator.
void Deallocate(const void *Ptr, size_t Size, size_t Alignment) {
- assert(getThreadIndex() < NumOfAllocators);
- return Allocators[getThreadIndex()].Deallocate(Ptr, Size, Alignment);
+ return getThreadLocalAllocator().Deallocate(Ptr, Size, Alignment);
}
- /// Return allocator corresponding to the current thread.
+ /// Return the calling thread's sub-allocator, creating it on first use.
AllocatorTy &getThreadLocalAllocator() {
- assert(getThreadIndex() < NumOfAllocators);
- return Allocators[getThreadIndex()];
+ // The calling thread's sub-allocator of each instance, indexed by a
+ // process-unique instance id.
+ //
+ // mlir::ThreadLocalCache keys an analogous per-thread map on the instance
+ // pointer and reclaims a thread's slot once the instance dies, but pays a
+ // map lookup and shared_ptr bookkeeping per allocation. Instances here are
+ // few and short-lived, so we prefer the O(1) vector index and accept that a
+ // thread's Cache only grows with the number of instances created.
+ static thread_local std::vector<AllocatorTy *> Cache;
+ if (LLVM_UNLIKELY(Cache.size() <= Id))
+ Cache.resize(Id + 1);
----------------
aengelke wrote:
I'm not at all familiar with this part of LLVMSupport and haven't looked closely at this code, but it does seem odd that creating a lot of PerThreadAllocator causes a permanent memory growth, as ids are never reused.
https://github.com/llvm/llvm-project/pull/209687
More information about the llvm-commits
mailing list