127 lines
5.0 KiB
Diff
127 lines
5.0 KiB
Diff
From 74daf13fb29f6147971a5e5fc63b3c795c5bffdd Mon Sep 17 00:00:00 2001
|
|
From: Rodrigo Kumpera <kumpera@users.noreply.github.com>
|
|
Date: Fri, 23 Mar 2018 09:43:30 -0700
|
|
Subject: [PATCH] Harden JIT job control against cctors deadlocking. (#7423)
|
|
|
|
* [mini] Don't wait for other threads to finish JITing if the current thread is running a cctor.
|
|
|
|
This eliminates a possible source of deadlocks found during AOT runs of Roslyn.
|
|
|
|
* [mini] Introduce a 1s timeout to JIT job control to ensure we don't simply deadlock in case of execution dependency bugs.
|
|
|
|
* [mini] fix spelling5DD
|
|
---
|
|
mono/metadata/threads-types.h | 3 +++
|
|
mono/metadata/threads.c | 8 ++++++++
|
|
mono/mini/mini-runtime.c | 32 +++++++++++++++++++++++++++-----
|
|
3 files changed, 38 insertions(+), 5 deletions(-)
|
|
|
|
diff --git a/mono/metadata/threads-types.h b/mono/metadata/threads-types.h
|
|
index b55aa6479a1..194d928e4ff 100644
|
|
--- a/mono/metadata/threads-types.h
|
|
+++ b/mono/metadata/threads-types.h
|
|
@@ -292,4 +292,7 @@ mono_thread_internal_describe (MonoInternalThread *internal, GString *str);
|
|
gboolean
|
|
mono_thread_internal_is_current (MonoInternalThread *internal);
|
|
|
|
+gboolean
|
|
+mono_threads_is_current_thread_in_protected_block (void);
|
|
+
|
|
#endif /* _MONO_METADATA_THREADS_TYPES_H_ */
|
|
diff --git a/mono/metadata/threads.c b/mono/metadata/threads.c
|
|
index cf819395e0e..30743de78fa 100644
|
|
--- a/mono/metadata/threads.c
|
|
+++ b/mono/metadata/threads.c
|
|
@@ -257,6 +257,14 @@ mono_thread_get_abort_prot_block_count (MonoInternalThread *thread)
|
|
return (state & ABORT_PROT_BLOCK_MASK) >> ABORT_PROT_BLOCK_SHIFT;
|
|
}
|
|
|
|
+gboolean
|
|
+mono_threads_is_current_thread_in_protected_block (void)
|
|
+{
|
|
+ MonoInternalThread *thread = mono_thread_internal_current ();
|
|
+
|
|
+ return mono_thread_get_abort_prot_block_count (thread) > 0;
|
|
+}
|
|
+
|
|
void
|
|
mono_threads_begin_abort_protected_block (void)
|
|
{
|
|
diff --git a/mono/mini/mini-runtime.c b/mono/mini/mini-runtime.c
|
|
index 22dab7313d6..90fa3a452bf 100644
|
|
--- a/mono/mini/mini-runtime.c
|
|
+++ b/mono/mini/mini-runtime.c
|
|
@@ -1910,8 +1910,15 @@ typedef struct {
|
|
MonoCoopMutex lock;
|
|
} JitCompilationData;
|
|
|
|
+/*
|
|
+Timeout, in millisecounds, that we wait other threads to finish JITing.
|
|
+This value can't be too small or we won't see enough methods being reused and it can't be too big to cause massive stalls due to unforseable circunstances.
|
|
+*/
|
|
+#define MAX_JIT_TIMEOUT_MS 1000
|
|
+
|
|
+
|
|
static JitCompilationData compilation_data;
|
|
-static int jit_methods_waited, jit_methods_multiple, jit_methods_overload, jit_spurious_wakeups;
|
|
+static int jit_methods_waited, jit_methods_multiple, jit_methods_overload, jit_spurious_wakeups_or_timeouts;
|
|
|
|
static void
|
|
mini_jit_init_job_control (void)
|
|
@@ -1976,7 +1983,7 @@ wait_or_register_method_to_compile (MonoMethod *method, MonoDomain *domain)
|
|
mono_counters_register ("JIT compile waited others", MONO_COUNTER_INT|MONO_COUNTER_JIT, &jit_methods_waited);
|
|
mono_counters_register ("JIT compile 1+ jobs", MONO_COUNTER_INT|MONO_COUNTER_JIT, &jit_methods_multiple);
|
|
mono_counters_register ("JIT compile overload wait", MONO_COUNTER_INT|MONO_COUNTER_JIT, &jit_methods_overload);
|
|
- mono_counters_register ("JIT compile spurious wakeups", MONO_COUNTER_INT|MONO_COUNTER_JIT, &jit_spurious_wakeups);
|
|
+ mono_counters_register ("JIT compile spurious wakeups or timeouts", MONO_COUNTER_INT|MONO_COUNTER_JIT, &jit_spurious_wakeups_or_timeouts);
|
|
inited = TRUE;
|
|
}
|
|
|
|
@@ -1993,9 +2000,14 @@ wait_or_register_method_to_compile (MonoMethod *method, MonoDomain *domain)
|
|
|
|
unlock_compilation_data ();
|
|
return FALSE;
|
|
- } else if (jit_tls->active_jit_methods > 0) {
|
|
+ } else if (jit_tls->active_jit_methods > 0 || mono_threads_is_current_thread_in_protected_block ()) {
|
|
//We can't suspend the current thread if it's already JITing a method.
|
|
//Dependency management is too compilated and we want to get rid of this anyways.
|
|
+
|
|
+ //We can't suspend the current thread if it's running a protected block (such as a cctor)
|
|
+ //We can't rely only on JIT nesting as cctor's can be run from outside the JIT.
|
|
+
|
|
+ //Finally, he hit a timeout or spurious wakeup. We're better off just giving up and keep recompiling
|
|
++entry->compilation_count;
|
|
++jit_methods_multiple;
|
|
++jit_tls->active_jit_methods;
|
|
@@ -2015,7 +2027,7 @@ wait_or_register_method_to_compile (MonoMethod *method, MonoDomain *domain)
|
|
++entry->threads_waiting;
|
|
|
|
g_assert (entry->has_cond);
|
|
- mono_coop_cond_wait (&entry->cond, &compilation_data.lock);
|
|
+ mono_coop_cond_timedwait (&entry->cond, &compilation_data.lock, MAX_JIT_TIMEOUT_MS);
|
|
--entry->threads_waiting;
|
|
|
|
if (entry->done) {
|
|
@@ -2023,7 +2035,17 @@ wait_or_register_method_to_compile (MonoMethod *method, MonoDomain *domain)
|
|
unlock_compilation_data ();
|
|
return TRUE;
|
|
} else {
|
|
- ++jit_spurious_wakeups;
|
|
+ //We hit the timeout or a spurious wakeup, fallback to JITing
|
|
+ g_assert (entry->ref_count > 1);
|
|
+ unref_jit_entry (entry);
|
|
+ ++jit_spurious_wakeups_or_timeouts;
|
|
+
|
|
+ ++entry->compilation_count;
|
|
+ ++jit_methods_multiple;
|
|
+ ++jit_tls->active_jit_methods;
|
|
+
|
|
+ unlock_compilation_data ();
|
|
+ return FALSE;
|
|
}
|
|
}
|
|
}
|
|
--
|
|
2.14.1
|
|
|