RegExp: Do on demand JIT compilation

JIT compiling means that we need to obtain executable memory for the
jitted code. That has a non-neglibile cost (in addition to the runtime
cost of compiling the pattern).

Thus, only compile the pattern if we suspect it to be worthwhile:
- when we called match multiple times
- or when the string we match against is quite large

We also assume that the JIT generally works, and thus discard the
bytecode immediately after jitting – if the JIT fails once (either to
compile, or with offset failure), we recompile the bytecode and never
JIT again.

Change-Id: I91e9e10c5ab895af730dbb4a9bc85470dd794d1f
Reviewed-by: Ulf Hermann <ulf.hermann@qt.io>

2 files changed, 52 insertions, 26 deletions

diff --git a/src/qml/jsruntime/qv4regexp.cpp b/src/qml/jsruntime/qv4regexp.cpp
index be7ff77603..1876b29c3e 100644
--- a/src/qml/jsruntime/qv4regexp.cpp
+++ b/src/qml/jsruntime/qv4regexp.cpp
@@ -9,6 +9,10 @@
 
 using namespace QV4;
 
+#if ENABLE(YARR_JIT)
+static constexpr quint8 RegexpJitThreshold = 5;
+#endif
+
 static JSC::RegExpFlags jscFlags(uint flags)
 {
     JSC::RegExpFlags jscFlags = JSC::NoFlags;
@@ -40,12 +44,51 @@ uint RegExp::match(const QString &string, int start, uint *matchOffsets)
     if (!isValid())
         return JSC::Yarr::offsetNoMatch;
 
+#if ENABLE(YARR_JIT)
+    auto *priv = d();
+
+    auto regenerateByteCode = [](Heap::RegExp *regexp) {
+        JSC::Yarr::ErrorCode error = JSC::Yarr::ErrorCode::NoError;
+        JSC::Yarr::YarrPattern yarrPattern(WTF::String(*regexp->pattern), jscFlags(regexp->flags),
+                                           error);
+
+        // As we successfully parsed the pattern before, we should still be able to.
+        Q_ASSERT(error == JSC::Yarr::ErrorCode::NoError);
+
+        regexp->byteCode = JSC::Yarr::byteCompile(
+                                   yarrPattern,
+                                   regexp->internalClass->engine->bumperPointerAllocator).release();
+    };
+    auto removeJitCode = [](Heap::RegExp *regexp) {
+        delete regexp->jitCode;
+        regexp->jitCode = nullptr;
+        regexp->jitFailed = true;
+    };
+    if (!(priv->jitCode || priv->jitFailed)) {
+        if (string.length() > 1024 || priv->matchCount++ == RegexpJitThreshold) {
+            auto *engine = priv->internalClass->engine;
+            if (engine->canJIT()) {
+                JSC::Yarr::ErrorCode error = JSC::Yarr::ErrorCode::NoError;
+                JSC::Yarr::YarrPattern yarrPattern(WTF::String(*priv->pattern), jscFlags(priv->flags), error);
+                if (!yarrPattern.m_containsBackreferences) {
+                    priv->jitCode = new JSC::Yarr::YarrCodeBlock;
+                    JSC::VM *vm = static_cast<JSC::VM *>(engine);
+                    JSC::Yarr::jitCompile(yarrPattern, JSC::Yarr::Char16, vm, *priv->jitCode);
+                }
+            }
+            if (!priv->hasValidJITCode())
+                removeJitCode(priv);
+        }
+        // Note: We can't throw away the byte code as the JIT can't handle everything and
+        // might return offsetJITFail
+    }
+#endif
+
     WTF::String s(string);
 
 #if ENABLE(YARR_JIT)
-    static const uint offsetJITFail = std::numeric_limits<unsigned>::max() - 1;
-    auto *priv = d();
     if (priv->hasValidJITCode()) {
+        static const uint offsetJITFail = std::numeric_limits<unsigned>::max() - 1;
         uint ret = JSC::Yarr::offsetNoMatch;
 #if ENABLE(YARR_JIT_ALL_PARENS_EXPRESSIONS)
         char buffer[8192];
@@ -58,19 +101,10 @@ uint RegExp::match(const QString &string, int start, uint *matchOffsets)
         if (ret != offsetJITFail)
             return ret;
 
+        removeJitCode(priv);
         // JIT failed. We need byteCode to run the interpreter.
-        if (!priv->byteCode) {
-            JSC::Yarr::ErrorCode error = JSC::Yarr::ErrorCode::NoError;
-            JSC::Yarr::YarrPattern yarrPattern(WTF::String(*priv->pattern), jscFlags(priv->flags),
-                                               error);
-
-            // As we successfully parsed the pattern before, we should still be able to.
-            Q_ASSERT(error == JSC::Yarr::ErrorCode::NoError);
-
-            priv->byteCode = JSC::Yarr::byteCompile(
-                                     yarrPattern,
-                                     priv->internalClass->engine->bumperPointerAllocator).release();
-        }
+        Q_ASSERT(!priv->byteCode);
+        regenerateByteCode(priv);
     }
 #endif // ENABLE(YARR_JIT)
 
@@ -175,25 +209,15 @@ void Heap::RegExp::init(ExecutionEngine *engine, const QString &pattern, uint fl
     this->flags = flags;
 
     valid = false;
+    jitFailed = false;
+    matchCount = 0;
 
     JSC::Yarr::ErrorCode error = JSC::Yarr::ErrorCode::NoError;
     JSC::Yarr::YarrPattern yarrPattern(WTF::String(pattern), jscFlags(flags), error);
     if (error != JSC::Yarr::ErrorCode::NoError)
         return;
     subPatternCount = yarrPattern.m_numSubpatterns;
-#if ENABLE(YARR_JIT)
-    if (!yarrPattern.m_containsBackreferences && engine->canJIT()) {
-        jitCode = new JSC::Yarr::YarrCodeBlock;
-        JSC::VM *vm = static_cast<JSC::VM *>(engine);
-        JSC::Yarr::jitCompile(yarrPattern, JSC::Yarr::Char16, vm, *jitCode);
-    }
-#else
     Q_UNUSED(engine);
-#endif
-    if (hasValidJITCode()) {
-        valid = true;
-        return;
-    }
     byteCode = JSC::Yarr::byteCompile(yarrPattern, internalClass->engine->bumperPointerAllocator).release();
     if (byteCode)
         valid = true;
diff --git a/src/qml/jsruntime/qv4regexp_p.h b/src/qml/jsruntime/qv4regexp_p.h
index 1b70354caa..8a178dd2f6 100644
--- a/src/qml/jsruntime/qv4regexp_p.h
+++ b/src/qml/jsruntime/qv4regexp_p.h
@@ -66,6 +66,8 @@ struct RegExp : Base {
     int subPatternCount;
     uint flags;
     bool valid;
+    bool jitFailed;
+    quint8 matchCount;
 
     QString flagsAsString() const;
     int captureCount() const { return subPatternCount + 1; }