[llvm] 85c2768 - [Support][Parallel] Initialize threadIndex and add assertion checking its usage.
Alexey Lapshin via llvm-commits
llvm-commits at lists.llvm.org
Tue May 2 09:56:49 PDT 2023
Author: Alexey Lapshin
Date: 2023-05-02T18:44:15+02:00
New Revision: 85c2768ce9023ae94d2c70951b3f7e3f33fc2105
URL: https://github.com/llvm/llvm-project/commit/85c2768ce9023ae94d2c70951b3f7e3f33fc2105
DIFF: https://github.com/llvm/llvm-project/commit/85c2768ce9023ae94d2c70951b3f7e3f33fc2105.diff
LOG: [Support][Parallel] Initialize threadIndex and add assertion checking its usage.
That patch adds a check for threadIndex being used with only threads
created by ThreadPoolExecutor. This helps catch two types of errors:
1. If a thread is created not by ThreadPoolExecutor its index may clash
with the index of another thread. Using threadIndex, in that case, may
lead to a data race.
2. Index of the main thread(threadIndex == 0) currently clashes with
the index of thread0 in ThreadPoolExecutor threads. That may lead
to a data race if main thread and thread0 are executed concurrently.
This patch allows execution tasks on the main thread only in case
parallel::strategy.ThreadsRequested == 1. In all other cases,
assertions check that threadIndex != UINT_MAX(i.e. that task
is executed on a thread created by ThreadPoolExecutor).
Differential Revision: https://reviews.llvm.org/D148916
Added:
Modified:
lld/ELF/Relocations.cpp
llvm/include/llvm/Support/Parallel.h
llvm/lib/Support/Parallel.cpp
Removed:
################################################################################
diff --git a/lld/ELF/Relocations.cpp b/lld/ELF/Relocations.cpp
index bda979c3066ac..2f7fcd664c6a5 100644
--- a/lld/ELF/Relocations.cpp
+++ b/lld/ELF/Relocations.cpp
@@ -1537,9 +1537,6 @@ template <class ELFT> void elf::scanRelocations() {
tg.spawn(fn, serial);
}
- // Both the main thread and thread pool index 0 use getThreadIndex()==0. Be
- // careful that they don't concurrently run scanSections. When serial is
- // true, fn() has finished at this point, so running execute is safe.
tg.spawn([] {
RelocationScanner scanner;
for (Partition &part : partitions) {
diff --git a/llvm/include/llvm/Support/Parallel.h b/llvm/include/llvm/Support/Parallel.h
index c9bcad69113d4..2e5476a19d7e8 100644
--- a/llvm/include/llvm/Support/Parallel.h
+++ b/llvm/include/llvm/Support/Parallel.h
@@ -30,6 +30,14 @@ namespace parallel {
extern ThreadPoolStrategy strategy;
#if LLVM_ENABLE_THREADS
+#define GET_THREAD_INDEX_IMPL \
+ if (parallel::strategy.ThreadsRequested == 1) \
+ return 0; \
+ assert((threadIndex != UINT_MAX) && \
+ "getThreadIndex() must be called from a thread created by " \
+ "ThreadPoolExecutor"); \
+ return threadIndex;
+
#ifdef _WIN32
// Direct access to thread_local variables from a
diff erent DLL isn't
// possible with Windows Native TLS.
@@ -38,7 +46,7 @@ unsigned getThreadIndex();
// Don't access this directly, use the getThreadIndex wrapper.
extern thread_local unsigned threadIndex;
-inline unsigned getThreadIndex() { return threadIndex; }
+inline unsigned getThreadIndex() { GET_THREAD_INDEX_IMPL; }
#endif
#else
inline unsigned getThreadIndex() { return 0; }
diff --git a/llvm/lib/Support/Parallel.cpp b/llvm/lib/Support/Parallel.cpp
index df292eba44713..95956bbe7c4de 100644
--- a/llvm/lib/Support/Parallel.cpp
+++ b/llvm/lib/Support/Parallel.cpp
@@ -24,11 +24,11 @@ namespace parallel {
#if LLVM_ENABLE_THREADS
#ifdef _WIN32
-static thread_local unsigned threadIndex;
+static thread_local unsigned threadIndex = UINT_MAX;
-unsigned getThreadIndex() { return threadIndex; }
+unsigned getThreadIndex() { GET_THREAD_INDEX_IMPL; }
#else
-thread_local unsigned threadIndex;
+thread_local unsigned threadIndex = UINT_MAX;
#endif
namespace detail {
@@ -99,10 +99,13 @@ class ThreadPoolExecutor : public Executor {
void add(std::function<void()> F, bool Sequential = false) override {
{
- bool UseSequentialQueue =
- Sequential || parallel::strategy.ThreadsRequested == 1;
+ if (parallel::strategy.ThreadsRequested == 1) {
+ F();
+ return;
+ }
+
std::lock_guard<std::mutex> Lock(Mutex);
- if (UseSequentialQueue)
+ if (Sequential)
WorkQueueSequential.emplace_front(std::move(F));
else
WorkQueue.emplace_back(std::move(F));
@@ -217,13 +220,9 @@ void TaskGroup::spawn(std::function<void()> F, bool Sequential) {
void llvm::parallelFor(size_t Begin, size_t End,
llvm::function_ref<void(size_t)> Fn) {
- // If we have zero or one items, then do not incur the overhead of spinning up
- // a task group. They are surprisingly expensive, and because they do not
- // support nested parallelism, a single entry task group can block parallel
- // execution underneath them.
#if LLVM_ENABLE_THREADS
- auto NumItems = End - Begin;
- if (NumItems > 1 && parallel::strategy.ThreadsRequested != 1) {
+ if (parallel::strategy.ThreadsRequested != 1) {
+ auto NumItems = End - Begin;
// Limit the number of tasks to MaxTasksPerGroup to limit job scheduling
// overhead on large inputs.
auto TaskSize = NumItems / parallel::detail::MaxTasksPerGroup;
More information about the llvm-commits
mailing list