Initialize LLVM time trace profiler on each code generation thread

In https://reviews.llvm.org/D71059 LLVM 11, the time trace profiler was
extended to support multiple threads.

`timeTraceProfilerInitialize` creates a thread local profiler instance.
When a thread finishes `timeTraceProfilerFinishThread` moves a thread
local instance into a global collection of instances. Finally when all
codegen work is complete `timeTraceProfilerWrite` writes data from the
current thread local instance and the instances in global collection
of instances.

Previously, the profiler was intialized on a single thread only. Since
this thread performs no code generation on its own, the resulting
profile was empty.

Update LLVM codegen to initialize & finish time trace profiler on each
code generation thread.

3 files changed, 51 insertions, 5 deletions

diff --git a/compiler/rustc_codegen_llvm/src/lib.rs b/compiler/rustc_codegen_llvm/src/lib.rs
index 8f4d79e7147..64fedb7bc1a 100644
--- a/compiler/rustc_codegen_llvm/src/lib.rs
+++ b/compiler/rustc_codegen_llvm/src/lib.rs
@@ -76,6 +76,27 @@ mod value;
 #[derive(Clone)]
 pub struct LlvmCodegenBackend(());
 
+struct TimeTraceProfiler {
+    enabled: bool,
+}
+
+impl TimeTraceProfiler {
+    fn new(enabled: bool) -> Self {
+        if enabled {
+            unsafe { llvm::LLVMTimeTraceProfilerInitialize() }
+        }
+        TimeTraceProfiler { enabled }
+    }
+}
+
+impl Drop for TimeTraceProfiler {
+    fn drop(&mut self) {
+        if self.enabled {
+            unsafe { llvm::LLVMTimeTraceProfilerFinishThread() }
+        }
+    }
+}
+
 impl ExtraBackendMethods for LlvmCodegenBackend {
     fn new_metadata(&self, tcx: TyCtxt<'_>, mod_name: &str) -> ModuleLlvm {
         ModuleLlvm::new_metadata(tcx, mod_name)
@@ -119,6 +140,34 @@ impl ExtraBackendMethods for LlvmCodegenBackend {
     fn tune_cpu<'b>(&self, sess: &'b Session) -> Option<&'b str> {
         llvm_util::tune_cpu(sess)
     }
+
+    fn spawn_thread<F, T>(time_trace: bool, f: F) -> std::thread::JoinHandle<T>
+    where
+        F: FnOnce() -> T,
+        F: Send + 'static,
+        T: Send + 'static,
+    {
+        std::thread::spawn(move || {
+            let _profiler = TimeTraceProfiler::new(time_trace);
+            f()
+        })
+    }
+
+    fn spawn_named_thread<F, T>(
+        time_trace: bool,
+        name: String,
+        f: F,
+    ) -> std::io::Result<std::thread::JoinHandle<T>>
+    where
+        F: FnOnce() -> T,
+        F: Send + 'static,
+        T: Send + 'static,
+    {
+        std::thread::Builder::new().name(name).spawn(move || {
+            let _profiler = TimeTraceProfiler::new(time_trace);
+            f()
+        })
+    }
 }
 
 impl WriteBackendMethods for LlvmCodegenBackend {
diff --git a/compiler/rustc_codegen_llvm/src/llvm/ffi.rs b/compiler/rustc_codegen_llvm/src/llvm/ffi.rs
index 21d2388fc30..749eec459ac 100644
--- a/compiler/rustc_codegen_llvm/src/llvm/ffi.rs
+++ b/compiler/rustc_codegen_llvm/src/llvm/ffi.rs
@@ -1737,6 +1737,8 @@ extern "C" {
 
     pub fn LLVMTimeTraceProfilerInitialize();
 
+    pub fn LLVMTimeTraceProfilerFinishThread();
+
     pub fn LLVMTimeTraceProfilerFinish(FileName: *const c_char);
 
     pub fn LLVMAddAnalysisPasses(T: &'a TargetMachine, PM: &PassManager<'a>);
diff --git a/compiler/rustc_codegen_llvm/src/llvm_util.rs b/compiler/rustc_codegen_llvm/src/llvm_util.rs
index 246bb88885d..34c28938c85 100644
--- a/compiler/rustc_codegen_llvm/src/llvm_util.rs
+++ b/compiler/rustc_codegen_llvm/src/llvm_util.rs
@@ -113,11 +113,6 @@ unsafe fn configure_llvm(sess: &Session) {
     }
 
     if sess.opts.debugging_opts.llvm_time_trace {
-        // time-trace is not thread safe and running it in parallel will cause seg faults.
-        if !sess.opts.debugging_opts.no_parallel_llvm {
-            bug!("`-Z llvm-time-trace` requires `-Z no-parallel-llvm")
-        }
-
         llvm::LLVMTimeTraceProfilerInitialize();
     }