Issue #1624 - Add missing error handling and bail on null specifier.

Since this is a http protocol networking feature, it belongs in the networking
branch of our preferences.

README.md

@@ -36,4 +36,4 @@ improvement patches upstream to its origin to mutually improve the platform and
 a future for XUL.
 
 ## Looking for Pale Moon and Basilisk?
-As of [`Issue #969`](https://github.com/MoonchildProductions/UXP/issues/969), release versions `28.9.0` and `2020.03.04` respectively, they have been split off into their own dedicated repositories located at [`MoonchildProductions/Pale-Moon`](https://github.com/MoonchildProductions/Pale-Moon) and [`MoonchildProductions/Basilisk`](https://github.com/MoonchildProductions/Basilisk).
+As of [`Issue #969`](https://repo.palemoon.org/MoonchildProductions/UXP/issues/969), release versions `28.9.0` and `2020.03.04` respectively, they have been split off into their own dedicated repositories located at [`MoonchildProductions/Pale-Moon`](https://repo.palemoon.org/MoonchildProductions/Pale-Moon) and [`MoonchildProductions/Basilisk`](https://repo.palemoon.org/MoonchildProductions/Basilisk).

SECURITY.md

@@ -23,7 +23,7 @@ Generally not security vulnerabilities:
 ## Reporting a Vulnerability
 
 If you find an issue in UXP or the applications it builds on that could impact the security or safety of users please **do not**
-make an issue on GitHub about it. GitHub does not support restricted viewability for security sensitive bugs.
+make an issue on Gitea about it. Gitea does not support restricted viewability for security sensitive bugs.
 
 If you want to report a security-sensitive issue then please go to the [forum](https://forum.palemoon.org) and report the issue
 via a **private message** to **Moonchild** (the founder and prime responsible for security issues).

build/moz.configure/old.configure

@@ -161,6 +161,8 @@ def old_configure_options(*options):
     '--enable-b2g-camera',
     '--enable-b2g-ril',
     '--enable-bundled-fonts',
+    '--enable-chrome-compression',
+    '--enable-chrome-format',
     '--enable-clang-plugin',
     '--enable-content-sandbox',
     '--enable-cookies',
@@ -262,6 +264,7 @@ def old_configure_options(*options):
     '--with-app-name',
     '--with-arch',
     '--with-branding',
+    '--with-chrome-name',
     '--with-crashreporter-enable-percent',
     '--with-cross-lib',
     '--with-debug-label',

config/check_spidermonkey_style.py

@@ -86,6 +86,7 @@ included_inclnames_to_ignore = set([
     'unicode/udat.h',           # ICU
     'unicode/udatpg.h',         # ICU
     'unicode/uenum.h',          # ICU
+    'unicode/uniset.h',         # ICU
     'unicode/unorm.h',          # ICU
     'unicode/unum.h',           # ICU
     'unicode/unumsys.h',        # ICU

config/config.mk

@@ -216,7 +216,7 @@ endif # NO_PROFILE_GUIDED_OPTIMIZE
 
 MAKE_JARS_FLAGS = \
 	-t $(topsrcdir) \
-	-f $(MOZ_JAR_MAKER_FILE_FORMAT) \
+	-f $(MOZ_CHROME_FILE_FORMAT) \
 	$(NULL)
 
 ifdef USE_EXTENSION_MANIFEST

config/milestone.txt

@@ -10,4 +10,4 @@
 # hardcoded milestones in the tree from these two files.
 #--------------------------------------------------------
 
-4.6.0
+4.7.0

config/system-headers

@@ -526,7 +526,6 @@ hardware/lights.h
 hardware/power.h
 hardware_legacy/power.h
 hardware_legacy/uevent.h
-hardware_legacy/vibrator.h
 #endif
 HIToolbox/HIToolbox.h
 hlink.h

config/tests/src-simple/Makefile.in

@@ -22,7 +22,7 @@ REF_MANIFEST = $(if $(USE_EXTENSION_MANIFEST),chrome.manifest,test.manifest)
 
 check-%::
 	if test -d $(FINAL_TARGET); then rm -rf $(FINAL_TARGET); fi;
-	$(MAKE) realchrome MOZ_JAR_MAKER_FILE_FORMAT=$*
+	$(MAKE) realchrome MOZ_CHROME_FILE_FORMAT=$*
 	@echo 'Comparing manifests...'
 	@if ! sort $(MY_MANIFEST) | diff --text -U 0 $(srcdir)/../$(REF_MANIFEST).$* - ; then \
 	  echo 'TEST-UNEXPECTED-FAIL | config/tests/$(REF_MANIFEST).$* | differing content in manifest!' ; \

devtools/client/inspector/rules/test/browser_rules_pseudo-element_01.js

@@ -29,7 +29,7 @@ function* testTopLeft(inspector, view) {
       elementRulesNb: 4,
       firstLineRulesNb: 2,
       firstLetterRulesNb: 1,
-      selectionRulesNb: 0,
+      selectionRulesNb: 1,
       afterRulesNb: 1,
       beforeRulesNb: 2
     }
@@ -166,7 +166,7 @@ function* testParagraph(inspector, view) {
       elementRulesNb: 3,
       firstLineRulesNb: 1,
       firstLetterRulesNb: 1,
-      selectionRulesNb: 1,
+      selectionRulesNb: 2,
       beforeRulesNb: 0,
       afterRulesNb: 0
     });
@@ -216,7 +216,7 @@ function* assertPseudoElementRulesNumbers(selector, inspector, view, ruleNbs) {
     firstLetterRules: elementStyle.rules.filter(rule =>
       rule.pseudoElement === ":first-letter"),
     selectionRules: elementStyle.rules.filter(rule =>
-      rule.pseudoElement === ":-moz-selection"),
+      rule.pseudoElement === ":selection"),
     beforeRules: elementStyle.rules.filter(rule =>
       rule.pseudoElement === ":before"),
     afterRules: elementStyle.rules.filter(rule =>

devtools/client/menus.js

@@ -94,7 +94,7 @@ exports.menuitems = [
     },
     key: {
       id: "browserToolbox",
-      modifiers: "accel,alt,shift",
+      modifiers: isMac ? "accel,alt" : "accel,shift",
       keytext: true
     }
   },

devtools/shared/css/generated/properties-db.js

@@ -1233,7 +1233,7 @@ exports.CSS_PROPERTIES = {
   "-moz-tab-size": {
     "isInherited": true,
     "subproperties": [
-      "-moz-tab-size"
+      "tab-size"
     ],
     "supports": [
       7
@@ -2872,6 +2872,7 @@ exports.CSS_PROPERTIES = {
       "box-shadow",
       "box-sizing",
       "caption-side",
+      "caret-color",
       "clear",
       "clip",
       "clip-path",
@@ -3051,7 +3052,7 @@ exports.CSS_PROPERTIES = {
       "stroke-opacity",
       "stroke-width",
       "-x-system-font",
-      "-moz-tab-size",
+      "tab-size",
       "table-layout",
       "text-align",
       "text-align-last",
@@ -5277,6 +5278,28 @@ exports.CSS_PROPERTIES = {
       "unset"
     ]
   },
+  "caret-color": {
+    "isInherited": true,
+    "subproperties": [
+      "caret-color"
+    ],
+    "supports": [
+      2
+    ],
+    "values": [
+      "COLOR",
+      "auto",
+      "currentColor",
+      "hsl",
+      "hsla",
+      "inherit",
+      "initial",
+      "rgb",
+      "rgba",
+      "transparent",
+      "unset"
+    ]
+  },
   "clear": {
     "isInherited": false,
     "subproperties": [
@@ -7857,6 +7880,7 @@ exports.CSS_PROPERTIES = {
     ],
     "supports": [],
     "values": [
+      "anywhere",
       "break-word",
       "inherit",
       "initial",
@@ -8626,6 +8650,23 @@ exports.CSS_PROPERTIES = {
       "unset"
     ]
   },
+  "tab-size": {
+    "isInherited": true,
+    "subproperties": [
+      "tab-size"
+    ],
+    "supports": [
+      6,
+      7
+    ],
+    "values": [
+      "-moz-calc",
+      "calc",
+      "inherit",
+      "initial",
+      "unset"
+    ]
+  },
   "table-layout": {
     "isInherited": false,
     "subproperties": [
@@ -9317,6 +9358,7 @@ exports.CSS_PROPERTIES = {
     ],
     "supports": [],
     "values": [
+      "anywhere",
       "break-word",
       "inherit",
       "initial",
@@ -9373,6 +9415,7 @@ exports.PSEUDO_ELEMENTS = [
   ":backdrop",
   ":first-letter",
   ":first-line",
+  ":selection",
   ":-moz-selection",
   ":-moz-focus-inner",
   ":-moz-focus-outer",

docs/CONTRIBUTING.md

@@ -93,15 +93,20 @@ With rare exception, it is advisable to use the following style for commit messa
 
 This would directly apply to anyone making a pull request.
 
+Single commits are where issues can be reasonably resolved in a single commit.
 - Single Commit: `Issue #xxx - Cited issue title or appropriate direct description of changes`
   - *`Issue #1083 - Deprecate FUEL extension helper javascript library`*
-- Multi-Part Commits: `Issue #xxx - Part Number: Appropriate direct description of changes`
+
+Multi-part commits would be used for complex issues. However, an exception exists for multi-part commits where the issue is anticipated or ends up being long term such as "Stop using unified compilation of sources". In these instances the multi-part form is not required.
+- Multi-Part Form: `Issue #xxx - Part Number: Appropriate direct description of changes`
   - *`Issue #492 - Part 1: Remove files`*
   - *`Issue #492 - Part 2: Build system, Installer/Packaging`*
   - *`Issue #492 - Part 3: nsUpdateService.js, updater.cpp, nsUpdateDriver.cpp`*
   - *`Issue #492 - Part 4: Remove superfluous brackets in nsUpdateService.js and updater.cpp`*
 
-An exception exists for multi-part commits where the issue is anticipated or ends up being long term such as "Stop using unified compilation of sources". In these instances the multi-part form is not required.
+Occasionally a resolved (and shipped) issue requires further changes to fix bugs. However, If follow-ups are complex enough to be multi-part it should be considered a new issue.
+- Follow-up Form: *`Issue #xxx - Follow-up: Appropriate direct description of changes`*
+  - *`Issue #1643 - Follow-up: Make sure things aren't changed while iterating.`*
 
 ### Commits with no issue
 

dom/animation/test/mozilla/file_discrete-animations.html

@@ -75,11 +75,6 @@ const gMozillaSpecificProperties = {
     from: "ignore",
     to: "stretch-to-fit"
   },
-  "-moz-tab-size": {
-    // https://drafts.csswg.org/css-text-3/#propdef-tab-size
-    from: "1",
-    to: "5"
-  },
   "-moz-text-size-adjust": {
     // https://drafts.csswg.org/css-size-adjust/#propdef-text-size-adjust
     from: "none",

dom/base/Navigator.cpp

@@ -97,11 +97,6 @@
 namespace mozilla {
 namespace dom {
 
-static bool sVibratorEnabled   = false;
-static uint32_t sMaxVibrateMS  = 0;
-static uint32_t sMaxVibrateListLen = 0;
-static const char* kVibrationPermissionType = "vibration";
-
 static void
 AddPermission(nsIPrincipal* aPrincipal, const char* aType, uint32_t aPermission,
               uint32_t aExpireType, int64_t aExpireTime)
@@ -152,12 +147,7 @@ GetPermission(nsIPrincipal* aPrincipal, const char* aType)
 void
 Navigator::Init()
 {
-  Preferences::AddBoolVarCache(&sVibratorEnabled,
-                               "dom.vibrator.enabled", true);
-  Preferences::AddUintVarCache(&sMaxVibrateMS,
-                               "dom.vibrator.max_vibrate_ms", 10000);
-  Preferences::AddUintVarCache(&sMaxVibrateListLen,
-                               "dom.vibrator.max_vibrate_list_len", 128);
+  // Add any Preferences::Add*VarCache(&sPref, "pref", default) here if needed.
 }
 
 Navigator::Navigator(nsPIDOMWindowInner* aWindow)
@@ -705,82 +695,6 @@ Navigator::RefreshMIMEArray()
   }
 }
 
-namespace {
-
-class VibrateWindowListener : public nsIDOMEventListener
-{
-public:
-  VibrateWindowListener(nsPIDOMWindowInner* aWindow, nsIDocument* aDocument)
-  {
-    mWindow = do_GetWeakReference(aWindow);
-    mDocument = do_GetWeakReference(aDocument);
-
-    NS_NAMED_LITERAL_STRING(visibilitychange, "visibilitychange");
-    aDocument->AddSystemEventListener(visibilitychange,
-                                      this, /* listener */
-                                      true, /* use capture */
-                                      false /* wants untrusted */);
-  }
-
-  void RemoveListener();
-
-  NS_DECL_ISUPPORTS
-  NS_DECL_NSIDOMEVENTLISTENER
-
-private:
-  virtual ~VibrateWindowListener()
-  {
-  }
-
-  nsWeakPtr mWindow;
-  nsWeakPtr mDocument;
-};
-
-NS_IMPL_ISUPPORTS(VibrateWindowListener, nsIDOMEventListener)
-
-StaticRefPtr<VibrateWindowListener> gVibrateWindowListener;
-
-static bool
-MayVibrate(nsIDocument* doc) {
-  // Hidden documents cannot start or stop a vibration.
-  return (doc && !doc->Hidden());
-}
-
-NS_IMETHODIMP
-VibrateWindowListener::HandleEvent(nsIDOMEvent* aEvent)
-{
-  nsCOMPtr<nsIDocument> doc =
-    do_QueryInterface(aEvent->InternalDOMEvent()->GetTarget());
-
-  if (!MayVibrate(doc)) {
-    // It's important that we call CancelVibrate(), not Vibrate() with an
-    // empty list, because Vibrate() will fail if we're no longer focused, but
-    // CancelVibrate() will succeed, so long as nobody else has started a new
-    // vibration pattern.
-    nsCOMPtr<nsPIDOMWindowInner> window = do_QueryReferent(mWindow);
-    hal::CancelVibrate(window);
-    RemoveListener();
-    gVibrateWindowListener = nullptr;
-    // Careful: The line above might have deleted |this|!
-  }
-
-  return NS_OK;
-}
-
-void
-VibrateWindowListener::RemoveListener()
-{
-  nsCOMPtr<EventTarget> target = do_QueryReferent(mDocument);
-  if (!target) {
-    return;
-  }
-  NS_NAMED_LITERAL_STRING(visibilitychange, "visibilitychange");
-  target->RemoveSystemEventListener(visibilitychange, this,
-                                    true /* use capture */);
-}
-
-} // namespace
-
 void
 Navigator::AddIdleObserver(MozIdleObserver& aIdleObserver, ErrorResult& aRv)
 {
@@ -809,111 +723,6 @@ Navigator::RemoveIdleObserver(MozIdleObserver& aIdleObserver, ErrorResult& aRv)
   }
 }
 
-void
-Navigator::SetVibrationPermission(bool aPermitted, bool aPersistent)
-{
-  MOZ_ASSERT(NS_IsMainThread());
-
-  nsTArray<uint32_t> pattern;
-  pattern.SwapElements(mRequestedVibrationPattern);
-
-  if (!mWindow) {
-    return;
-  }
-
-  nsCOMPtr<nsIDocument> doc = mWindow->GetExtantDoc();
-
-  if (!MayVibrate(doc)) {
-    return;
-  }
-
-  if (aPermitted) {
-    // Add a listener to cancel the vibration if the document becomes hidden,
-    // and remove the old visibility listener, if there was one.
-    if (!gVibrateWindowListener) {
-      // If gVibrateWindowListener is null, this is the first time we've vibrated,
-      // and we need to register a listener to clear gVibrateWindowListener on
-      // shutdown.
-      ClearOnShutdown(&gVibrateWindowListener);
-    } else {
-      gVibrateWindowListener->RemoveListener();
-    }
-    gVibrateWindowListener = new VibrateWindowListener(mWindow, doc);
-    hal::Vibrate(pattern, mWindow);
-  }
-
-  if (aPersistent) {
-    AddPermission(doc->NodePrincipal(), kVibrationPermissionType,
-                  aPermitted ? nsIPermissionManager::ALLOW_ACTION :
-                               nsIPermissionManager::DENY_ACTION,
-                  nsIPermissionManager::EXPIRE_SESSION, 0);
-  }
-}
-
-bool
-Navigator::Vibrate(uint32_t aDuration)
-{
-  AutoTArray<uint32_t, 1> pattern;
-  pattern.AppendElement(aDuration);
-  return Vibrate(pattern);
-}
-
-bool
-Navigator::Vibrate(const nsTArray<uint32_t>& aPattern)
-{
-  MOZ_ASSERT(NS_IsMainThread());
-
-  if (!mWindow) {
-    return false;
-  }
-
-  nsCOMPtr<nsIDocument> doc = mWindow->GetExtantDoc();
-
-  if (!MayVibrate(doc)) {
-    return false;
-  }
-
-  nsTArray<uint32_t> pattern(aPattern);
-
-  if (pattern.Length() > sMaxVibrateListLen) {
-    pattern.SetLength(sMaxVibrateListLen);
-  }
-
-  for (size_t i = 0; i < pattern.Length(); ++i) {
-    pattern[i] = std::min(sMaxVibrateMS, pattern[i]);
-  }
-
-  // The spec says we check sVibratorEnabled after we've done the sanity
-  // checking on the pattern.
-  if (!sVibratorEnabled) {
-    return true;
-  }
-
-  mRequestedVibrationPattern.SwapElements(pattern);
-  uint32_t permission = GetPermission(mWindow, kVibrationPermissionType);
-
-  if (permission == nsIPermissionManager::ALLOW_ACTION ||
-      mRequestedVibrationPattern.IsEmpty() ||
-      (mRequestedVibrationPattern.Length() == 1 &&
-       mRequestedVibrationPattern[0] == 0)) {
-    // Always allow cancelling vibration and respect session permissions.
-    SetVibrationPermission(true /* permitted */, false /* persistent */);
-    return true;
-  }
-
-  nsCOMPtr<nsIObserverService> obs = services::GetObserverService();
-  if (!obs || permission == nsIPermissionManager::DENY_ACTION) {
-    // Abort without observer service or on denied session permission.
-    SetVibrationPermission(false /* permitted */, false /* persistent */);
-    return true;
-  }
-
-  // Request user permission.
-  obs->NotifyObservers(ToSupports(this), "Vibration:Request", nullptr);
-
-  return true;
-}
-
 //*****************************************************************************
 //  Pointer Events interface
 //*****************************************************************************

dom/base/Navigator.h

@@ -149,9 +149,6 @@ public:
   // NavigatorBinding::ClearCachedUserAgentValue(this);
   void ClearUserAgentCache();
 
-  bool Vibrate(uint32_t aDuration);
-  bool Vibrate(const nsTArray<uint32_t>& aDuration);
-  void SetVibrationPermission(bool aPermitted, bool aPersistent);
   uint32_t MaxTouchPoints();
   void GetAppCodeName(nsString& aAppCodeName, ErrorResult& aRv)
   {

dom/base/ResizeObserverController.cpp

@@ -15,7 +15,7 @@ namespace dom {
 void
 ResizeObserverNotificationHelper::WillRefresh(TimeStamp aTime)
 {
-  MOZ_ASSERT(mOwner, "Why is mOwner already dead when this RefreshObserver is still registered?");
+  MOZ_DIAGNOSTIC_ASSERT(mOwner, "RefreshObserver should have been de-registered on time, but isn't.");
   if (mOwner) {
     mOwner->Notify();
   }
@@ -58,15 +58,19 @@ ResizeObserverNotificationHelper::Register()
 void
 ResizeObserverNotificationHelper::Unregister()
 {
+  if (!mOwner) {
+    // We've outlived our owner, so there's nothing registered anymore.
+    mRegistered = false;
+    return;
+  }
+
   if (!mRegistered) {
     return;
   }
 
   nsRefreshDriver* refreshDriver = GetRefreshDriver();
-  if (!refreshDriver) {
-    // We can't access RefreshDriver now. Just abort the Unregister().
-    return;
-  }
+  MOZ_RELEASE_ASSERT(refreshDriver,
+                     "We should not leave a dangling reference to the observer around");
 
   refreshDriver->RemoveRefreshObserver(this, Flush_Display);
   mRegistered = false;
@@ -75,9 +79,8 @@ ResizeObserverNotificationHelper::Unregister()
 void
 ResizeObserverNotificationHelper::Disconnect()
 {
-  Unregister();
-  // Our owner is dying. Clear our pointer to it, in case we outlive it.
-  mOwner = nullptr;
+  MOZ_RELEASE_ASSERT(!mRegistered, "How can we die when registered?");
+  MOZ_RELEASE_ASSERT(!mOwner, "Forgot to clear weak pointer?");
 }
 
 ResizeObserverNotificationHelper::~ResizeObserverNotificationHelper()
@@ -105,6 +108,10 @@ ResizeObserverController::AddResizeObserver(ResizeObserver* aObserver)
   mResizeObservers.AppendElement(aObserver);
 }
 
+void ResizeObserverController::DetachFromDocument() {
+  mResizeObserverNotificationHelper->Unregister();
+}
+
 void
 ResizeObserverController::Notify()
 {
@@ -112,6 +119,10 @@ ResizeObserverController::Notify()
     return;
   }
 
+  // Hold a strong reference to the document, because otherwise calling
+  // all active observers on it might yank it out from under us.
+  RefPtr<nsIDocument> document(mDocument);
+  
   uint32_t shallowestTargetDepth = 0;
 
   GatherAllActiveObservations(shallowestTargetDepth);
@@ -146,7 +157,7 @@ ResizeObserverController::Notify()
     nsEventStatus status = nsEventStatus_eIgnore;
 
     nsCOMPtr<nsPIDOMWindowInner> window =
-      mDocument->GetWindow()->GetCurrentInnerWindow();
+      document->GetWindow()->GetCurrentInnerWindow();
 
     if (window) {
       nsCOMPtr<nsIScriptGlobalObject> sgo = do_QueryInterface(window);
@@ -178,7 +189,11 @@ ResizeObserverController::BroadcastAllActiveObservations()
 {
   uint32_t shallowestTargetDepth = UINT32_MAX;
 
-  for (auto observer : mResizeObservers) {
+  // Use a copy of the observers as this invokes the callbacks of the observers
+  // which could register/unregister observers at will.
+  nsTArray<RefPtr<ResizeObserver>> tempObservers(mResizeObservers);
+  
+  for (auto observer : tempObservers) {
 
     uint32_t targetDepth = observer->BroadcastActiveObservations();
 
@@ -226,7 +241,10 @@ ResizeObserverController::GetShell() const
 
 ResizeObserverController::~ResizeObserverController()
 {
-  mResizeObserverNotificationHelper->Disconnect();
+  MOZ_RELEASE_ASSERT(
+      !mResizeObserverNotificationHelper->IsRegistered(),
+      "Nothing else should keep a reference to our notification helper when we go away");
+  mResizeObserverNotificationHelper->DetachFromOwner();
 }
 
 } // namespace dom

dom/base/ResizeObserverController.h

@@ -41,6 +41,10 @@ public:
 
   void Disconnect();
 
+  bool IsRegistered() const { return mRegistered; }
+
+  void DetachFromOwner() { mOwner = nullptr; }
+
 protected:
   virtual ~ResizeObserverNotificationHelper();
 
@@ -68,6 +72,7 @@ public:
   void Traverse(nsCycleCollectionTraversalCallback& aCb);
   void Unlink();
 
+  void DetachFromDocument();
   void AddResizeObserver(ResizeObserver* aObserver);
 
   /*

dom/base/nsDocument.cpp

@@ -3644,6 +3644,12 @@ nsDocument::DeleteShell()
   // objects for @font-face rules that came from the style set.
   RebuildUserFontSet();
 
+  if (mResizeObserverController) {
+    // If the shell is going away, we need to remove any links to this document
+    // from the observer.
+    mResizeObserverController->DetachFromDocument();
+  }
+
   nsIPresShell* oldShell = mPresShell;
   mPresShell = nullptr;
   UpdateFrameRequestCallbackSchedulingState(oldShell);

dom/base/nsImageLoadingContent.cpp

@@ -932,8 +932,12 @@ nsImageLoadingContent::LoadImage(nsIURI* aNewURI,
     MOZ_ASSERT(!req, "Shouldn't have non-null request here");
     // If we don't have a current URI, we might as well store this URI so people
     // know what we tried (and failed) to load.
-    if (!mCurrentRequest)
+    if (!mCurrentRequest) {
       mCurrentURI = aNewURI;
+      if (mImageBlockingStatus == nsIContentPolicy::ACCEPT) {
+        mImageBlockingStatus = nsIContentPolicy::REJECT_REQUEST;
+      }
+    }
 
     FireEvent(NS_LITERAL_STRING("error"));
     FireEvent(NS_LITERAL_STRING("loadend"));

dom/base/nsTreeSanitizer.cpp

@@ -1384,6 +1384,8 @@ nsTreeSanitizer::SanitizeChildren(nsINode* aRoot)
         nsAutoString styleText;
         nsContentUtils::GetNodeTextContent(node, false, styleText);
 
+        RemoveAllAttributesFromDescendants(elt);
+        
         nsAutoString sanitizedStyle;
         nsCOMPtr<nsIURI> baseURI = node->GetBaseURI();
         if (SanitizeStyleSheet(styleText,
@@ -1479,6 +1481,17 @@ nsTreeSanitizer::RemoveAllAttributes(nsIContent* aElement)
   }
 }
 
+void nsTreeSanitizer::RemoveAllAttributesFromDescendants(mozilla::dom::Element* aElement) {
+  nsIContent* node = aElement->GetFirstChild();
+  while (node) {
+    if (node->IsElement()) {
+      mozilla::dom::Element* elt = node->AsElement();
+      RemoveAllAttributes(elt);
+    }
+    node = node->GetNextNode(aElement);
+  }
+}
+
 void
 nsTreeSanitizer::InitializeStatics()
 {

dom/base/nsTreeSanitizer.h

@@ -183,6 +183,12 @@ class MOZ_STACK_CLASS nsTreeSanitizer {
      */
     void RemoveAllAttributes(nsIContent* aElement);
 
+    /**
+     * Removes all attributes from the descendants of an element but not from
+     * the element itself.
+     */
+    void RemoveAllAttributesFromDescendants(mozilla::dom::Element* aElement);
+
     /**
      * The whitelist of HTML elements.
      */

dom/fetch/FetchDriver.cpp

@@ -708,10 +708,12 @@ FetchDriver::OnDataAvailable(nsIRequest* aRequest,
   // about races.
 
   if (mObserver) {
+    // Need to keep mObserver alive.
+    RefPtr<FetchDriverObserver> observer = mObserver;
     if (NS_IsMainThread()) {
-      mObserver->OnDataAvailable();
+      observer->OnDataAvailable();
     } else {
-      RefPtr<Runnable> runnable = new DataAvailableRunnable(mObserver);
+      RefPtr<Runnable> runnable = new DataAvailableRunnable(observer);
       nsresult rv = NS_DispatchToMainThread(runnable);
       if (NS_WARN_IF(NS_FAILED(rv))) {
         return rv;
@@ -719,12 +721,15 @@ FetchDriver::OnDataAvailable(nsIRequest* aRequest,
     }
   }
 
-  uint32_t aRead;
+  // Explicitly initialized to 0 because in some cases nsStringInputStream may
+  // not write to aRead.
+  uint32_t aRead = 0;
   MOZ_ASSERT(mResponse);
   MOZ_ASSERT(mPipeOutputStream);
 
   // From "Main Fetch" step 17: SRI-part2.
-  if (mResponse->Type() != ResponseType::Error &&
+  if (mResponse &&
+      mResponse->Type() != ResponseType::Error &&
       !mRequest->GetIntegrity().IsEmpty()) {
     MOZ_ASSERT(mSRIDataVerifier);
 
@@ -766,6 +771,17 @@ FetchDriver::OnDataAvailable(nsIRequest* aRequest,
   nsresult rv = aInputStream->ReadSegments(NS_CopySegmentToStream,
                                            mPipeOutputStream,
                                            aCount, &aRead);
+
+  // If no data was read, it's possible the output stream is closed but the
+  // ReadSegments call followed its contract of returning NS_OK despite write
+  // errors.  Unfortunately, nsIOutputStream has an ill-conceived contract when
+  // taken together with ReadSegments' contract, because the pipe will just
+  // NS_OK if we try and invoke its Write* functions ourselves with a 0 count.
+  // So we must just assume the pipe is broken.
+  if (aRead == 0 && aCount != 0) {
+    return NS_BASE_STREAM_CLOSED;
+  }
+
   return rv;
 }
 

dom/script/ScriptLoader.cpp

@@ -599,38 +599,31 @@ ScriptLoader::CreateModuleScript(ModuleLoadRequest* aRequest)
   return rv;
 }
 
-static bool
-CreateTypeError(JSContext* aCx, ModuleScript* aScript, const nsString& aMessage,
-                JS::MutableHandle<JS::Value> aError)
-{
-  JS::Rooted<JSObject*> module(aCx, aScript->ModuleRecord());
-  JS::Rooted<JSScript*> script(aCx, JS::GetModuleScript(aCx, module));
-  JS::Rooted<JSString*> filename(aCx);
-  filename = JS_NewStringCopyZ(aCx, JS_GetScriptFilename(script));
-  if (!filename) {
-    return false;
-  }
-
-  JS::Rooted<JSString*> message(aCx, JS_NewUCStringCopyZ(aCx, aMessage.get()));
-  if (!message) {
-    return false;
-  }
-
-  return JS::CreateError(aCx, JSEXN_TYPEERR, nullptr, filename, 0, 0, nullptr,
-                         message, aError);
-}
-
 static nsresult
 HandleResolveFailure(JSContext* aCx, ModuleScript* aScript,
-                     const nsAString& aSpecifier)
+                     const nsAString& aSpecifier,
+                     uint32_t aLineNumber, uint32_t aColumnNumber)
 {
-  // TODO: How can we get the line number of the failed import?
+  nsAutoCString url;
+  aScript->BaseURL()->GetAsciiSpec(url);
+
+  JS::Rooted<JSString*> filename(aCx);
+  filename = JS_NewStringCopyZ(aCx, url.get());
+  if (!filename) {
+    return NS_ERROR_OUT_OF_MEMORY;
+  }
 
   nsAutoString message(NS_LITERAL_STRING("Error resolving module specifier: "));
   message.Append(aSpecifier);
 
+  JS::Rooted<JSString*> string(aCx, JS_NewUCStringCopyZ(aCx, message.get()));
+  if (!string) {
+    return NS_ERROR_OUT_OF_MEMORY;
+  }
+
   JS::Rooted<JS::Value> error(aCx);
-  if (!CreateTypeError(aCx, aScript, message, &error)) {
+  if (!JS::CreateError(aCx, JSEXN_TYPEERR, nullptr, filename, aLineNumber,
+                       aColumnNumber, nullptr, string, &error)) {
     return NS_ERROR_OUT_OF_MEMORY;
   }
 
@@ -685,28 +678,37 @@ ResolveRequestedModules(ModuleLoadRequest* aRequest, nsCOMArray<nsIURI>* aUrlsOu
 
   JSContext* cx = jsapi.cx();
   JS::Rooted<JSObject*> moduleRecord(cx, ms->ModuleRecord());
-  JS::Rooted<JSObject*> specifiers(cx, JS::GetRequestedModules(cx, moduleRecord));
+  JS::Rooted<JSObject*> requestedModules(cx);
+  requestedModules = JS::GetRequestedModules(cx, moduleRecord);
+  MOZ_ASSERT(requestedModules);
 
   uint32_t length;
-  if (!JS_GetArrayLength(cx, specifiers, &length)) {
+  if (!JS_GetArrayLength(cx, requestedModules, &length)) {
     return NS_ERROR_FAILURE;
   }
 
-  JS::Rooted<JS::Value> val(cx);
+  JS::Rooted<JS::Value> element(cx);
   for (uint32_t i = 0; i < length; i++) {
-    if (!JS_GetElement(cx, specifiers, i, &val)) {
+    if (!JS_GetElement(cx, requestedModules, i, &element)) {
       return NS_ERROR_FAILURE;
     }
 
+    JS::Rooted<JSString*> str(cx, JS::GetRequestedModuleSpecifier(cx, element));
+    MOZ_ASSERT(str);
+
     nsAutoJSString specifier;
-    if (!specifier.init(cx, val)) {
+    if (!specifier.init(cx, str)) {
       return NS_ERROR_FAILURE;
     }
 
     // Let url be the result of resolving a module specifier given module script and requested.
     nsCOMPtr<nsIURI> uri = ResolveModuleSpecifier(ms, specifier);
     if (!uri) {
-      nsresult rv = HandleResolveFailure(cx, ms, specifier);
+      uint32_t lineNumber = 0;
+      uint32_t columnNumber = 0;
+      JS::GetRequestedModuleSourcePos(cx, element, &lineNumber, &columnNumber);
+
+      nsresult rv = HandleResolveFailure(cx, ms, specifier, lineNumber, columnNumber);
       NS_ENSURE_SUCCESS(rv, rv);
       return NS_ERROR_FAILURE;
     }
@@ -806,6 +808,11 @@ HostResolveImportedModule(JSContext* aCx, JS::Handle<JSObject*> aModule,
   auto script = static_cast<ModuleScript*>(value.toPrivate());
   MOZ_ASSERT(script->ModuleRecord() == aModule);
 
+  // FIXME: in-line module scripts apparently pass a null specifier?
+  // At least make sure that doesn't crash the browser.
+  if (!aSpecifier)
+    return nullptr;
+    
   // Let url be the result of resolving a module specifier given referencing
   // module script and specifier.
   nsAutoJSString string;

dom/smil/nsSMILCSSProperty.cpp

@@ -194,6 +194,7 @@ nsSMILCSSProperty::IsPropertyAnimatable(nsCSSPropertyID aPropID)
   //   writing-mode
 
   switch (aPropID) {
+    case eCSSProperty_caret_color:
     case eCSSProperty_clip:
     case eCSSProperty_clip_rule:
     case eCSSProperty_clip_path:

dom/tests/mochitest/general/mochitest.ini

@@ -122,7 +122,6 @@ run-if = e10s
 [test_storagePermissionsReject.html]
 [test_storagePermissionsRejectForeign.html]
 [test_stylesheetPI.html]
-[test_vibrator.html]
 [test_WebKitCSSMatrix.html]
 [test_windowedhistoryframes.html]
 [test_windowProperties.html]

dom/tests/mochitest/general/test_vibrator.html

@@ -1,93 +0,0 @@
-<!DOCTYPE HTML>
-<html>
-<head>
-  <title>Test for Vibrator</title>
-  <script type="text/javascript" src="/MochiKit/MochiKit.js"></script>
-  <script type="text/javascript" src="/tests/SimpleTest/SimpleTest.js"></script>
-  <link rel="stylesheet" type="text/css" href="/tests/SimpleTest/test.css" />
-</head>
-<body>
-
-<!-- Although we can't test that the vibrator works properly, we can test that
-     navigator.vibrate throws an exception where appropriate. -->
-
-<script class="testbody" type="text/javascript;version=1.7">
-SimpleTest.waitForExplicitFinish();
-var result;
-function expectFailure(param) {
-  result = navigator.vibrate(param);
-  is(result, false, 'vibrate(' + param + ') should have failed.');
-}
-
-function expectSuccess(param) {
-  result = navigator.vibrate(param);
-  is(result, true, 'vibrate(' + param + ') must succeed.');
-}
-
-function tests(aEnabled) {
-  // Some edge cases that the bindings should handle for us.
-  expectSuccess(null);
-  expectSuccess(undefined);
-  // -1 will be converted to the highest unsigned long then clamped.
-  expectSuccess(-1);
-  expectSuccess('a');
-  // -1 will be converted to the highest unsigned long then clamped.
-  expectSuccess([100, -1]);
-  expectSuccess([100, 'a']);
-
-  var maxVibrateMs = SpecialPowers.getIntPref('dom.vibrator.max_vibrate_ms');
-  var maxVibrateListLen = SpecialPowers.getIntPref('dom.vibrator.max_vibrate_list_len');
-
-  // If we pass a vibration pattern with a value higher than max_vibrate_ms or a
-  // pattern longer than max_vibrate_list_len, the call should succeed but the
-  // pattern should be modified to match the restrictions.
-
-  // Values will be clamped to dom.vibrator.max_vibrate_ms.
-  expectSuccess(maxVibrateMs + 1);
-  expectSuccess([maxVibrateMs + 1]);
-
-  var arr = [];
-  for (var i = 0; i < maxVibrateListLen + 1; i++) {
-    arr[i] = 0;
-  }
-  // The array will be truncated to have a length equal to dom.vibrator.max_vibrate_list_len.
-  expectSuccess(arr);
-
-
-  expectSuccess(0);
-  expectSuccess([]);
-  expectSuccess('1000');
-  expectSuccess(1000);
-  expectSuccess(1000.1);
-  expectSuccess([0, 0, 0]);
-  expectSuccess(['1000', 1000]);
-  expectSuccess([1000, 1000]);
-  expectSuccess([1000, 1000.1]);
-
-  // The following loop shouldn't cause us to crash.  See bug 701716.
-  for (var i = 0; i < 10000; i++) {
-    navigator.vibrate([100, 100]);
-  }
-  ok(true, "Didn't crash after issuing a lot of vibrate() calls.");
-  if(!aEnabled)
-    SimpleTest.finish();
-}
-
-SpecialPowers.pushPermissions([
-    {type: 'vibration', allow: true, context: document}
-  ], function() {
-    // Test with the vibrator pref enabled.
-    SpecialPowers.pushPrefEnv({"set": [['dom.vibrator.enabled', true]]}, function() {
-      tests(true);
-      SpecialPowers.pushPrefEnv({"set": [['dom.vibrator.enabled', false]]}, tests(false));
-    });
-    // Everything should be the same when the vibrator is disabled -- in
-    // particular, a disabled vibrator shouldn't eat failures we'd otherwise
-    // observe.
-  }
-);
-
-</script>
-</body>
-
-</html>

dom/webidl/Navigator.webidl

@@ -13,9 +13,9 @@
  * http://www.w3.org/TR/beacon/#sec-beacon-method
  * https://html.spec.whatwg.org/#navigatorconcurrenthardware
  *
- * © Copyright 2004-2011 Apple Computer, Inc., Mozilla Foundation, and
- * Opera Software ASA. You are granted a license to use, reproduce
- * and create derivative works of this document.
+ * © Copyright 2004-2020 Apple Computer, Inc., Mozilla Foundation,
+ * Opera Software ASA and Moonchild Productions. You are granted a license to use,
+ * reproduce and create derivative works of this document.
  */
 
 // http://www.whatwg.org/specs/web-apps/current-work/#the-navigator-object
@@ -124,14 +124,6 @@ interface NavigatorGeolocation {
 };
 Navigator implements NavigatorGeolocation;
 
-// http://www.w3.org/TR/vibration/#vibration-interface
-partial interface Navigator {
-    // We don't support sequences in unions yet
-    //boolean vibrate ((unsigned long or sequence<unsigned long>) pattern);
-    boolean vibrate(unsigned long duration);
-    boolean vibrate(sequence<unsigned long> pattern);
-};
-
 // http://www.w3.org/TR/pointerevents/#extensions-to-the-navigator-interface
 partial interface Navigator {
     [Pref="dom.w3c_pointer_events.enabled"]
@@ -140,17 +132,6 @@ partial interface Navigator {
 
 // Mozilla-specific extensions
 
-// Chrome-only interface for Vibration API permission handling.
-partial interface Navigator {
-    /* Set permission state to device vibration.
-     * @param permitted permission state (true for allowing vibration)
-     * @param persistent make the permission session-persistent
-     */
-    [ChromeOnly]
-    void setVibrationPermission(boolean permitted,
-                                optional boolean persistent = true);
-};
-
 callback interface MozIdleObserver {
   // Time is in seconds and is read only when idle observers are added
   // and removed.

gfx/2d/DrawTargetSkia.cpp

@@ -212,7 +212,7 @@ VerifyRGBXCorners(uint8_t* aData, const IntSize &aSize, const int32_t aStride, S
 #endif
 
 static sk_sp<SkImage>
-GetSkImageForSurface(SourceSurface* aSurface)
+GetSkImageForSurface(SourceSurface* aSurface, Maybe<MutexAutoLock>* aLock)
 {
   if (!aSurface) {
     gfxDebug() << "Creating null Skia image from null SourceSurface";
@@ -220,7 +220,7 @@ GetSkImageForSurface(SourceSurface* aSurface)
   }
 
   if (aSurface->GetType() == SurfaceType::SKIA) {
-    return static_cast<SourceSurfaceSkia*>(aSurface)->GetImage();
+    return static_cast<SourceSurfaceSkia*>(aSurface)->GetImage(aLock);
   }
 
   DataSourceSurface* surf = aSurface->GetDataSurface().take();
@@ -390,9 +390,9 @@ ExtractAlphaBitmap(sk_sp<SkImage> aImage, SkBitmap* aResultBitmap)
 }
 
 static sk_sp<SkImage>
-ExtractAlphaForSurface(SourceSurface* aSurface)
+ExtractAlphaForSurface(SourceSurface* aSurface, Maybe<MutexAutoLock>& aLock)
 {
-  sk_sp<SkImage> image = GetSkImageForSurface(aSurface);
+  sk_sp<SkImage> image = GetSkImageForSurface(aSurface, &aLock);
   if (!image) {
     return nullptr;
   }
@@ -411,7 +411,11 @@ ExtractAlphaForSurface(SourceSurface* aSurface)
 }
 
 static void
-SetPaintPattern(SkPaint& aPaint, const Pattern& aPattern, Float aAlpha = 1.0, Point aOffset = Point(0, 0))
+SetPaintPattern(SkPaint& aPaint,
+                const Pattern& aPattern,
+                Maybe<MutexAutoLock>& aLock,
+                Float aAlpha = 1.0,
+                Point aOffset = Point(0, 0))
 {
   switch (aPattern.GetType()) {
     case PatternType::COLOR: {
@@ -473,7 +477,7 @@ SetPaintPattern(SkPaint& aPaint, const Pattern& aPattern, Float aAlpha = 1.0, Po
     }
     case PatternType::SURFACE: {
       const SurfacePattern& pat = static_cast<const SurfacePattern&>(aPattern);
-      sk_sp<SkImage> image = GetSkImageForSurface(pat.mSurface);
+      sk_sp<SkImage> image = GetSkImageForSurface(pat.mSurface, &aLock);
       if (!image) {
         aPaint.setColor(SK_ColorTRANSPARENT);
         break;
@@ -525,7 +529,7 @@ struct AutoPaintSetup {
     : mNeedsRestore(false), mAlpha(1.0)
   {
     Init(aCanvas, aOptions, aMaskBounds, false);
-    SetPaintPattern(mPaint, aPattern, mAlpha, aOffset);
+    SetPaintPattern(mPaint, aPattern, mLock, mAlpha, aOffset);
   }
 
   AutoPaintSetup(SkCanvas *aCanvas, const DrawOptions& aOptions, const Rect* aMaskBounds = nullptr, bool aForceGroup = false)
@@ -579,6 +583,7 @@ struct AutoPaintSetup {
   SkPaint mPaint;
   bool mNeedsRestore;
   SkCanvas* mCanvas;
+  Maybe<MutexAutoLock> mLock;
   Float mAlpha;
 };
 
@@ -601,7 +606,8 @@ DrawTargetSkia::DrawSurface(SourceSurface *aSurface,
 
   MarkChanged();
 
-  sk_sp<SkImage> image = GetSkImageForSurface(aSurface);
+  Maybe<MutexAutoLock> lock;
+  sk_sp<SkImage> image = GetSkImageForSurface(aSurface, &lock);
   if (!image) {
     return;
   }
@@ -654,7 +660,8 @@ DrawTargetSkia::DrawSurfaceWithShadow(SourceSurface *aSurface,
 
   MarkChanged();
 
-  sk_sp<SkImage> image = GetSkImageForSurface(aSurface);
+  Maybe<MutexAutoLock> lock;
+  sk_sp<SkImage> image = GetSkImageForSurface(aSurface, &lock);
   if (!image) {
     return;
   }
@@ -1407,8 +1414,9 @@ DrawTargetSkia::Mask(const Pattern &aSource,
   MarkChanged();
   AutoPaintSetup paint(mCanvas.get(), aOptions, aSource);
 
+  Maybe<MutexAutoLock> lock;
   SkPaint maskPaint;
-  SetPaintPattern(maskPaint, aMask);
+  SetPaintPattern(maskPaint, aMask, lock);
 
   SkLayerRasterizer::Builder builder;
   builder.addLayer(maskPaint);
@@ -1427,7 +1435,8 @@ DrawTargetSkia::MaskSurface(const Pattern &aSource,
   MarkChanged();
   AutoPaintSetup paint(mCanvas.get(), aOptions, aSource, nullptr, -aOffset);
 
-  sk_sp<SkImage> alphaMask = ExtractAlphaForSurface(aMask);
+  Maybe<MutexAutoLock> lock;
+  sk_sp<SkImage> alphaMask = ExtractAlphaForSurface(aMask, lock);
   if (!alphaMask) {
     gfxDebug() << *this << ": MaskSurface() failed to extract alpha for mask";
     return;
@@ -1456,7 +1465,8 @@ DrawTarget::Draw3DTransformedSurface(SourceSurface* aSurface, const Matrix4x4& a
   fullMat.PostTranslate(-xformBounds.x, -xformBounds.y, 0);
 
   // Read in the source data.
-  sk_sp<SkImage> srcImage = GetSkImageForSurface(aSurface);
+  Maybe<MutexAutoLock> lock;
+  sk_sp<SkImage> srcImage = GetSkImageForSurface(aSurface, &lock);
   if (!srcImage) {
     return true;
   }
@@ -1515,7 +1525,8 @@ DrawTargetSkia::Draw3DTransformedSurface(SourceSurface* aSurface, const Matrix4x
 
   MarkChanged();
 
-  sk_sp<SkImage> image = GetSkImageForSurface(aSurface);
+  Maybe<MutexAutoLock> lock;
+  sk_sp<SkImage> image = GetSkImageForSurface(aSurface, &lock);
   if (!image) {
     return true;
   }
@@ -1599,7 +1610,8 @@ already_AddRefed<SourceSurface>
 DrawTargetSkia::OptimizeGPUSourceSurface(SourceSurface *aSurface) const
 {
   // Check if the underlying SkImage already has an associated GrTexture.
-  sk_sp<SkImage> image = GetSkImageForSurface(aSurface);
+  Maybe<MutexAutoLock> lock;
+  sk_sp<SkImage> image = GetSkImageForSurface(aSurface, &lock);
   if (!image || image->isTextureBacked()) {
     RefPtr<SourceSurface> surface(aSurface);
     return surface.forget();
@@ -1716,7 +1728,8 @@ DrawTargetSkia::CopySurface(SourceSurface *aSurface,
 {
   MarkChanged();
 
-  sk_sp<SkImage> image = GetSkImageForSurface(aSurface);
+  Maybe<MutexAutoLock> lock;
+  sk_sp<SkImage> image = GetSkImageForSurface(aSurface, &lock);
   if (!image) {
     return;
   }
@@ -2061,7 +2074,8 @@ DrawTargetSkia::PopLayer()
         paint.setColor(SK_ColorTRANSPARENT);
       }
 
-      sk_sp<SkImage> alphaMask = ExtractAlphaForSurface(layer.mMask);
+      Maybe<MutexAutoLock> lock;
+      sk_sp<SkImage> alphaMask = ExtractAlphaForSurface(layer.mMask, lock);
       if (!alphaMask) {
         gfxDebug() << *this << ": PopLayer() failed to extract alpha for mask";
       } else {

gfx/2d/SourceSurfaceSkia.cpp

@@ -40,6 +40,25 @@ SourceSurfaceSkia::GetFormat() const
   return mFormat;
 }
 
+sk_sp<SkImage>
+SourceSurfaceSkia::GetImage(Maybe<MutexAutoLock>* aLock) {
+  // If we were provided a lock object, we can let the caller access
+  // a shared SkImage and we know it won't go away while the lock is held.
+  if (aLock) {
+    // We are locked, so now we can check mDrawTarget.
+    // If it's null, then we're not shared and we can unlock eagerly.
+    if (!mDrawTarget) {
+      aLock->reset();
+    }
+  } else {
+    // Otherwise we need to call DrawTargetWillChange to ensure we have our
+    // own copy of the SkImage.
+    DrawTargetWillChange();
+  }
+  sk_sp<SkImage> image = mImage;
+  return image;
+}
+  
 static sk_sp<SkData>
 MakeSkData(unsigned char* aData, const IntSize& aSize, int32_t aStride)
 {

gfx/2d/SourceSurfaceSkia.h

@@ -10,6 +10,10 @@
 #include <vector>
 #include "skia/include/core/SkCanvas.h"
 #include "skia/include/core/SkImage.h"
+#include "mozilla/Maybe.h"
+#include "mozilla/Mutex.h"
+
+typedef mozilla::MutexAutoLock MutexAutoLock;
 
 namespace mozilla {
 
@@ -28,7 +32,7 @@ public:
   virtual IntSize GetSize() const;
   virtual SurfaceFormat GetFormat() const;
 
-  sk_sp<SkImage>& GetImage() { return mImage; }
+  sk_sp<SkImage> GetImage(Maybe<MutexAutoLock>* aLock);
 
   bool InitFromData(unsigned char* aData,
                     const IntSize &aSize,

gfx/thebes/gfxTextRun.cpp

@@ -1071,6 +1071,34 @@ gfxTextRun::GetAdvanceWidth(Range aRange, PropertyProvider *aProvider,
     return result + GetAdvanceForGlyphs(ligatureRange);
 }
 
+gfxFloat
+gfxTextRun::GetMinAdvanceWidth(Range aRange)
+{
+    MOZ_ASSERT(aRange.end <= GetLength(), "Substring out of range");
+
+    Range ligatureRange = aRange;
+    ShrinkToLigatureBoundaries(&ligatureRange);
+
+    gfxFloat result = std::max(
+        ComputePartialLigatureWidth(Range(aRange.start, ligatureRange.start),
+                                    nullptr),
+        ComputePartialLigatureWidth(Range(ligatureRange.end, aRange.end),
+                                    nullptr));
+
+    // XXX Do we need to take spacing into account? When each grapheme cluster
+    // takes its own line, we shouldn't be adding spacings around them.
+    gfxFloat clusterAdvance = 0;
+    for (uint32_t i = ligatureRange.start; i < ligatureRange.end; ++i) {
+        clusterAdvance += GetAdvanceForGlyph(i);
+        if (i + 1 == ligatureRange.end || IsClusterStart(i + 1)) {
+            result = std::max(result, clusterAdvance);
+            clusterAdvance = 0;
+        }
+    }
+
+    return result;
+}
+
 bool
 gfxTextRun::SetLineBreaks(Range aRange,
                           bool aLineBreakBefore, bool aLineBreakAfter,

gfx/thebes/gfxTextRun.h

@@ -297,6 +297,12 @@ public:
         return GetAdvanceWidth(Range(this), nullptr);
     }
 
+    /**
+     * Computes the minimum advance width for a substring assuming line
+     * breaking is allowed everywhere.
+     */
+    gfxFloat GetMinAdvanceWidth(Range aRange);
+
     /**
      * Clear all stored line breaks for the given range (both before and after),
      * and then set the line-break state before aRange.start to aBreakBefore and

hal/Hal.cpp

@@ -94,87 +94,8 @@ WindowIsActive(nsPIDOMWindowInner* aWindow)
   return !document->Hidden();
 }
 
-StaticAutoPtr<WindowIdentifier::IDArrayType> gLastIDToVibrate;
-
-void InitLastIDToVibrate()
-{
-  gLastIDToVibrate = new WindowIdentifier::IDArrayType();
-  ClearOnShutdown(&gLastIDToVibrate);
-}
-
 } // namespace
 
-void
-Vibrate(const nsTArray<uint32_t>& pattern, nsPIDOMWindowInner* window)
-{
-  Vibrate(pattern, WindowIdentifier(window));
-}
-
-void
-Vibrate(const nsTArray<uint32_t>& pattern, const WindowIdentifier &id)
-{
-  AssertMainThread();
-
-  // Only active windows may start vibrations.  If |id| hasn't gone
-  // through the IPC layer -- that is, if our caller is the outside
-  // world, not hal_proxy -- check whether the window is active.  If
-  // |id| has gone through IPC, don't check the window's visibility;
-  // only the window corresponding to the bottommost process has its
-  // visibility state set correctly.
-  if (!id.HasTraveledThroughIPC() && !WindowIsActive(id.GetWindow())) {
-    HAL_LOG("Vibrate: Window is inactive, dropping vibrate.");
-    return;
-  }
-
-  if (!InSandbox()) {
-    if (!gLastIDToVibrate) {
-      InitLastIDToVibrate();
-    }
-    *gLastIDToVibrate = id.AsArray();
-  }
-
-  // Don't forward our ID if we are not in the sandbox, because hal_impl
-  // doesn't need it, and we don't want it to be tempted to read it.  The
-  // empty identifier will assert if it's used.
-  PROXY_IF_SANDBOXED(Vibrate(pattern, InSandbox() ? id : WindowIdentifier()));
-}
-
-void
-CancelVibrate(nsPIDOMWindowInner* window)
-{
-  CancelVibrate(WindowIdentifier(window));
-}
-
-void
-CancelVibrate(const WindowIdentifier &id)
-{
-  AssertMainThread();
-
-  // Although only active windows may start vibrations, a window may
-  // cancel its own vibration even if it's no longer active.
-  //
-  // After a window is marked as inactive, it sends a CancelVibrate
-  // request.  We want this request to cancel a playing vibration
-  // started by that window, so we certainly don't want to reject the
-  // cancellation request because the window is now inactive.
-  //
-  // But it could be the case that, after this window became inactive,
-  // some other window came along and started a vibration.  We don't
-  // want this window's cancellation request to cancel that window's
-  // actively-playing vibration!
-  //
-  // To solve this problem, we keep track of the id of the last window
-  // to start a vibration, and only accepts cancellation requests from
-  // the same window.  All other cancellation requests are ignored.
-
-  if (InSandbox() || (gLastIDToVibrate && *gLastIDToVibrate == id.AsArray())) {
-    // Don't forward our ID if we are not in the sandbox, because hal_impl
-    // doesn't need it, and we don't want it to be tempted to read it.  The
-    // empty identifier will assert if it's used.
-    PROXY_IF_SANDBOXED(CancelVibrate(InSandbox() ? id : WindowIdentifier()));
-  }
-}
-
 template <class InfoType>
 class ObserversManager
 {

hal/Hal.h

@@ -50,41 +50,6 @@ typedef Observer<SystemTimezoneChangeInformation> SystemTimezoneChangeObserver;
 
 namespace MOZ_HAL_NAMESPACE {
 
-/**
- * Turn the default vibrator device on/off per the pattern specified
- * by |pattern|.  Each element in the pattern is the number of
- * milliseconds to turn the vibrator on or off.  The first element in
- * |pattern| is an "on" element, the next is "off", and so on.
- *
- * If |pattern| is empty, any in-progress vibration is canceled.
- *
- * Only an active window within an active tab may call Vibrate; calls
- * from inactive windows and windows on inactive tabs do nothing.
- *
- * If you're calling hal::Vibrate from the outside world, pass an
- * nsIDOMWindow* in place of the WindowIdentifier parameter.
- * The method with WindowIdentifier will be called automatically.
- */
-void Vibrate(const nsTArray<uint32_t>& pattern,
-             nsPIDOMWindowInner* aWindow);
-void Vibrate(const nsTArray<uint32_t>& pattern,
-             const hal::WindowIdentifier &id);
-
-/**
- * Cancel a vibration started by the content window identified by
- * WindowIdentifier.
- *
- * If the window was the last window to start a vibration, the
- * cancellation request will go through even if the window is not
- * active.
- *
- * As with hal::Vibrate(), if you're calling hal::CancelVibrate from the outside
- * world, pass an nsIDOMWindow*. The method with WindowIdentifier will be called
- * automatically.
- */
-void CancelVibrate(nsPIDOMWindowInner* aWindow);
-void CancelVibrate(const hal::WindowIdentifier &id);
-
 /**
  * Determine whether the device's screen is currently enabled.
  */

hal/fallback/FallbackVibration.cpp

@@ -1,22 +0,0 @@
-/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
-/* This Source Code Form is subject to the terms of the Mozilla Public
- * License, v. 2.0. If a copy of the MPL was not distributed with this file,
- * You can obtain one at http://mozilla.org/MPL/2.0/. */
-
-#include "Hal.h"
-
-using mozilla::hal::WindowIdentifier;
-
-namespace mozilla {
-namespace hal_impl {
-
-void
-Vibrate(const nsTArray<uint32_t>& pattern, const hal::WindowIdentifier &)
-{}
-
-void
-CancelVibrate(const hal::WindowIdentifier &)
-{}
-
-} // namespace hal_impl
-} // namespace mozilla

hal/moz.build

@@ -29,7 +29,6 @@ if CONFIG['OS_TARGET'] == 'Linux':
         'fallback/FallbackAlarm.cpp',
         'fallback/FallbackScreenConfiguration.cpp',
         'fallback/FallbackSensor.cpp',
-        'fallback/FallbackVibration.cpp',
         'linux/LinuxMemory.cpp',
         'linux/LinuxPower.cpp',
     ]
@@ -39,7 +38,6 @@ elif CONFIG['OS_TARGET'] == 'WINNT':
         'fallback/FallbackMemory.cpp',
         'fallback/FallbackPower.cpp',
         'fallback/FallbackScreenConfiguration.cpp',
-        'fallback/FallbackVibration.cpp',
         'windows/WindowsSensor.cpp',
     ]
 elif CONFIG['MOZ_WIDGET_TOOLKIT'] == 'cocoa':
@@ -48,7 +46,6 @@ elif CONFIG['MOZ_WIDGET_TOOLKIT'] == 'cocoa':
         'fallback/FallbackMemory.cpp',
         'fallback/FallbackPower.cpp',
         'fallback/FallbackScreenConfiguration.cpp',
-        'fallback/FallbackVibration.cpp',
     ]
 elif CONFIG['OS_TARGET'] in ('OpenBSD', 'NetBSD', 'FreeBSD', 'DragonFly'):
     UNIFIED_SOURCES += [
@@ -57,7 +54,6 @@ elif CONFIG['OS_TARGET'] in ('OpenBSD', 'NetBSD', 'FreeBSD', 'DragonFly'):
         'fallback/FallbackPower.cpp',
         'fallback/FallbackScreenConfiguration.cpp',
         'fallback/FallbackSensor.cpp',
-        'fallback/FallbackVibration.cpp',
     ]
 else:
     UNIFIED_SOURCES += [
@@ -66,7 +62,6 @@ else:
         'fallback/FallbackPower.cpp',
         'fallback/FallbackScreenConfiguration.cpp',
         'fallback/FallbackSensor.cpp',
-        'fallback/FallbackVibration.cpp',
     ]
 
 UNIFIED_SOURCES += [

hal/sandbox/PHal.ipdl

@@ -69,9 +69,6 @@ child:
     async NotifySystemTimezoneChange(SystemTimezoneChangeInformation aSystemTimezoneChangeInfo); 
 
 parent:
-    async Vibrate(uint32_t[] pattern, uint64_t[] id, PBrowser browser);
-    async CancelVibrate(uint64_t[] id, PBrowser browser);
-
     async EnableNetworkNotifications();
     async DisableNetworkNotifications();
     sync GetCurrentNetworkInformation()

hal/sandbox/SandboxHal.cpp

@@ -45,28 +45,6 @@ Hal()
   return sHal;
 }
 
-void
-Vibrate(const nsTArray<uint32_t>& pattern, const WindowIdentifier &id)
-{
-  HAL_LOG("Vibrate: Sending to parent process.");
-
-  AutoTArray<uint32_t, 8> p(pattern);
-
-  WindowIdentifier newID(id);
-  newID.AppendProcessID();
-  Hal()->SendVibrate(p, newID.AsArray(), TabChild::GetFrom(newID.GetWindow()));
-}
-
-void
-CancelVibrate(const WindowIdentifier &id)
-{
-  HAL_LOG("CancelVibrate: Sending to parent process.");
-
-  WindowIdentifier newID(id);
-  newID.AppendProcessID();
-  Hal()->SendCancelVibrate(newID.AsArray(), TabChild::GetFrom(newID.GetWindow()));
-}
-
 void
 EnableNetworkNotifications()
 {
@@ -379,36 +357,6 @@ public:
     hal::UnregisterSystemTimezoneChangeObserver(this);
   }
 
-  virtual bool
-  RecvVibrate(InfallibleTArray<unsigned int>&& pattern,
-              InfallibleTArray<uint64_t>&& id,
-              PBrowserParent *browserParent) override
-  {
-    // We give all content vibration permission.
-    //    TabParent *tabParent = TabParent::GetFrom(browserParent);
-    /* xxxkhuey wtf
-    nsCOMPtr<nsIDOMWindow> window =
-      do_QueryInterface(tabParent->GetBrowserDOMWindow());
-    */
-    WindowIdentifier newID(id, nullptr);
-    hal::Vibrate(pattern, newID);
-    return true;
-  }
-
-  virtual bool
-  RecvCancelVibrate(InfallibleTArray<uint64_t> &&id,
-                    PBrowserParent *browserParent) override
-  {
-    //TabParent *tabParent = TabParent::GetFrom(browserParent);
-    /* XXXkhuey wtf
-    nsCOMPtr<nsIDOMWindow> window =
-      tabParent->GetBrowserDOMWindow();
-    */
-    WindowIdentifier newID(id, nullptr);
-    hal::CancelVibrate(newID);
-    return true;
-  }
-
   virtual bool
   RecvEnableNetworkNotifications() override {
     // We give all content access to this network-status information.

ipc/app/plugin-container.exe.manifest

@@ -29,6 +29,7 @@
   <ms_asmv3:application xmlns:ms_asmv3="urn:schemas-microsoft-com:asm.v3">
     <ms_asmv3:windowsSettings xmlns="http://schemas.microsoft.com/SMI/2005/WindowsSettings">
       <dpiAware>True/PM</dpiAware>
+      <dpiAwareness xmlns="http://schemas.microsoft.com/SMI/2016/WindowsSettings">PerMonitorV2,PerMonitor</dpiAwareness>
     </ms_asmv3:windowsSettings>
   </ms_asmv3:application>
   <compatibility xmlns="urn:schemas-microsoft-com:compatibility.v1">

ipc/testshell/XPCShellEnvironment.cpp

@@ -1,5 +1,5 @@
 /* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
- * This Source Code Form is subject to the terms of the Mozilla Public
+/* This Source Code Form is subject to the terms of the Mozilla Public
  * License, v. 2.0. If a copy of the MPL was not distributed with this
  * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
 

js/moz.configure

@@ -246,3 +246,16 @@ with only_when('--enable-compile-environment'):
 
     set_config('LIBFUZZER', enable_libfuzzer)
     set_define('LIBFUZZER', enable_libfuzzer)
+
+# Initial support for new regexp engine
+# ==================================================
+
+js_option('--enable-new-regexp', default=False, help='Enable new regexp engine')
+
+@depends('--enable-new-regexp')
+def enable_new_regexp(value):
+    if value:
+        return True
+
+set_config('JS_NEW_REGEXP', enable_new_regexp)
+set_define('JS_NEW_REGEXP', enable_new_regexp)

js/public/Class.h

@@ -778,7 +778,7 @@ struct JSClass {
 // application.
 #define JSCLASS_GLOBAL_APPLICATION_SLOTS 5
 #define JSCLASS_GLOBAL_SLOT_COUNT                                             \
-    (JSCLASS_GLOBAL_APPLICATION_SLOTS + JSProto_LIMIT * 2 + 45)
+    (JSCLASS_GLOBAL_APPLICATION_SLOTS + JSProto_LIMIT * 2 + 46)
 #define JSCLASS_GLOBAL_FLAGS_WITH_SLOTS(n)                                    \
     (JSCLASS_IS_GLOBAL | JSCLASS_HAS_RESERVED_SLOTS(JSCLASS_GLOBAL_SLOT_COUNT + (n)))
 #define JSCLASS_GLOBAL_FLAGS                                                  \

js/src/builtin/Module.js

@@ -363,7 +363,7 @@ function InnerModuleInstantiation(module, stack, index)
     // Step 9
     let requestedModules = module.requestedModules;
     for (let i = 0; i < requestedModules.length; i++) {
-        let required = requestedModules[i];
+        let required = requestedModules[i].moduleSpecifier;
         let requiredModule = CallModuleResolveHook(module, required, MODULE_STATUS_UNINSTANTIATED);
 
         index = InnerModuleInstantiation(requiredModule, stack, index);
@@ -591,7 +591,7 @@ function InnerModuleEvaluation(module, stack, index)
     // Step 10
     let requestedModules = module.requestedModules;
     for (let i = 0; i < requestedModules.length; i++) {
-        let required = requestedModules[i];
+        let required = requestedModules[i].moduleSpecifier;
         let requiredModule =
             CallModuleResolveHook(module, required, MODULE_STATUS_INSTANTIATED);
 

js/src/builtin/ModuleObject.cpp

@@ -125,7 +125,7 @@ GlobalObject::initImportEntryProto(JSContext* cx, Handle<GlobalObject*> global)
     if (!DefinePropertiesAndFunctions(cx, proto, protoAccessors, nullptr))
         return false;
 
-    global->setReservedSlot(IMPORT_ENTRY_PROTO, ObjectValue(*proto));
+    global->initReservedSlot(IMPORT_ENTRY_PROTO, ObjectValue(*proto));
     return true;
 }
 
@@ -203,7 +203,7 @@ GlobalObject::initExportEntryProto(JSContext* cx, Handle<GlobalObject*> global)
     if (!DefinePropertiesAndFunctions(cx, proto, protoAccessors, nullptr))
         return false;
 
-    global->setReservedSlot(EXPORT_ENTRY_PROTO, ObjectValue(*proto));
+    global->initReservedSlot(EXPORT_ENTRY_PROTO, ObjectValue(*proto));
     return true;
 }
 
@@ -240,6 +240,71 @@ ExportEntryObject::create(ExclusiveContext* cx,
     return self;
 }
 
+///////////////////////////////////////////////////////////////////////////
+// RequestedModuleObject
+
+/* static */ const Class
+RequestedModuleObject::class_ = {
+    "RequestedModule",
+    JSCLASS_HAS_RESERVED_SLOTS(RequestedModuleObject::SlotCount) |
+    JSCLASS_IS_ANONYMOUS
+};
+
+DEFINE_GETTER_FUNCTIONS(RequestedModuleObject, moduleSpecifier, ModuleSpecifierSlot)
+DEFINE_GETTER_FUNCTIONS(RequestedModuleObject, lineNumber, LineNumberSlot)
+DEFINE_GETTER_FUNCTIONS(RequestedModuleObject, columnNumber, ColumnNumberSlot)
+
+DEFINE_ATOM_ACCESSOR_METHOD(RequestedModuleObject, moduleSpecifier)
+DEFINE_UINT32_ACCESSOR_METHOD(RequestedModuleObject, lineNumber)
+DEFINE_UINT32_ACCESSOR_METHOD(RequestedModuleObject, columnNumber)
+
+/* static */ bool
+RequestedModuleObject::isInstance(HandleValue value)
+{
+    return value.isObject() && value.toObject().is<RequestedModuleObject>();
+}
+
+/* static */ bool
+GlobalObject::initRequestedModuleProto(JSContext* cx, Handle<GlobalObject*> global)
+{
+    static const JSPropertySpec protoAccessors[] = {
+        JS_PSG("moduleSpecifier", RequestedModuleObject_moduleSpecifierGetter, 0),
+        JS_PSG("lineNumber", RequestedModuleObject_lineNumberGetter, 0),
+        JS_PSG("columnNumber", RequestedModuleObject_columnNumberGetter, 0),
+        JS_PS_END
+    };
+
+    RootedObject proto(cx, GlobalObject::createBlankPrototype<PlainObject>(cx, global));
+    if (!proto)
+        return false;
+
+    if (!DefinePropertiesAndFunctions(cx, proto, protoAccessors, nullptr))
+        return false;
+
+    global->initReservedSlot(REQUESTED_MODULE_PROTO, ObjectValue(*proto));
+    return true;
+}
+
+/* static */ RequestedModuleObject*
+RequestedModuleObject::create(JSContext* cx,
+                              HandleAtom moduleSpecifier,
+                              uint32_t lineNumber,
+                              uint32_t columnNumber)
+{
+    MOZ_ASSERT(lineNumber > 0);
+
+    RootedObject proto(cx, cx->global()->getRequestedModulePrototype());
+    RootedObject obj(cx, NewObjectWithGivenProto(cx, &class_, proto));
+    if (!obj)
+        return nullptr;
+
+    RootedRequestedModuleObject self(cx, &obj->as<RequestedModuleObject>());
+    self->initReservedSlot(ModuleSpecifierSlot, StringValue(moduleSpecifier));
+    self->initReservedSlot(LineNumberSlot, Int32Value(lineNumber));
+    self->initReservedSlot(ColumnNumberSlot, Int32Value(columnNumber));
+    return self;
+}
+
 ///////////////////////////////////////////////////////////////////////////
 // IndirectBindingMap
 
@@ -1074,7 +1139,8 @@ ModuleBuilder::ModuleBuilder(ExclusiveContext* cx, HandleModuleObject module,
   : cx_(cx),
     module_(cx, module),
     tokenStream_(tokenStream),
-    requestedModules_(cx, AtomVector(cx)),
+    requestedModuleSpecifiers_(cx, AtomSet(cx)),
+    requestedModules_(cx, RequestedModuleVector(cx)),
     importedBoundNames_(cx, AtomVector(cx)),
     importEntries_(cx, ImportEntryVector(cx)),
     exportEntries_(cx, ExportEntryVector(cx)),
@@ -1083,6 +1149,12 @@ ModuleBuilder::ModuleBuilder(ExclusiveContext* cx, HandleModuleObject module,
     starExportEntries_(cx, ExportEntryVector(cx))
 {}
 
+bool
+ModuleBuilder::init()
+{
+    return requestedModuleSpecifiers_.init();
+}
+
 bool
 ModuleBuilder::buildTables()
 {
@@ -1130,24 +1202,23 @@ ModuleBuilder::buildTables()
 bool
 ModuleBuilder::initModule()
 {
-    RootedArrayObject requestedModules(cx_, createArray<JSAtom*>(requestedModules_));
+    RootedArrayObject requestedModules(cx_, createArray(requestedModules_));
     if (!requestedModules)
         return false;
 
-    RootedArrayObject importEntries(cx_, createArray<ImportEntryObject*>(importEntries_));
+    RootedArrayObject importEntries(cx_, createArray(importEntries_));
     if (!importEntries)
         return false;
 
-    RootedArrayObject localExportEntries(cx_, createArray<ExportEntryObject*>(localExportEntries_));
+    RootedArrayObject localExportEntries(cx_, createArray(localExportEntries_));
     if (!localExportEntries)
         return false;
 
-    RootedArrayObject indirectExportEntries(cx_);
-    indirectExportEntries = createArray<ExportEntryObject*>(indirectExportEntries_);
+    RootedArrayObject indirectExportEntries(cx_, createArray(indirectExportEntries_));
     if (!indirectExportEntries)
         return false;
 
-    RootedArrayObject starExportEntries(cx_, createArray<ExportEntryObject*>(starExportEntries_));
+    RootedArrayObject starExportEntries(cx_, createArray(starExportEntries_));
     if (!starExportEntries)
         return false;
 
@@ -1169,7 +1240,7 @@ ModuleBuilder::processImport(frontend::ParseNode* pn)
     MOZ_ASSERT(pn->pn_right->isKind(PNK_STRING));
 
     RootedAtom module(cx_, pn->pn_right->pn_atom);
-    if (!maybeAppendRequestedModule(module))
+    if (!maybeAppendRequestedModule(module, pn->pn_right))
         return false;
 
     for (ParseNode* spec = pn->pn_left->pn_head; spec; spec = spec->pn_next) {
@@ -1277,7 +1348,7 @@ ModuleBuilder::processExportFrom(frontend::ParseNode* pn)
     MOZ_ASSERT(pn->pn_right->isKind(PNK_STRING));
 
     RootedAtom module(cx_, pn->pn_right->pn_atom);
-    if (!maybeAppendRequestedModule(module))
+    if (!maybeAppendRequestedModule(module, pn->pn_right))
         return false;
 
     for (ParseNode* spec = pn->pn_left->pn_head; spec; spec = spec->pn_next) {
@@ -1346,29 +1417,26 @@ ModuleBuilder::appendExportFromEntry(HandleAtom exportName, HandleAtom moduleReq
 }
 
 bool
-ModuleBuilder::maybeAppendRequestedModule(HandleAtom module)
+ModuleBuilder::maybeAppendRequestedModule(HandleAtom specifier, ParseNode* node)
 {
-    for (auto m : requestedModules_) {
-        if (m == module)
-            return true;
+    if (requestedModuleSpecifiers_.has(specifier)) {
+        return true;
     }
-    return requestedModules_.append(module);
-}
-
-static Value
-MakeElementValue(JSString *string)
-{
-    return StringValue(string);
-}
-
-static Value
-MakeElementValue(JSObject *object)
-{
-    return ObjectValue(*object);
+    uint32_t line;
+    uint32_t column;
+    tokenStream_.srcCoords.lineNumAndColumnIndex(node->pn_pos.begin, &line, &column);
+    JSContext* cx = reinterpret_cast<JSContext*>(cx_);
+    RootedRequestedModuleObject req(cx, RequestedModuleObject::create(cx, specifier, line, column));
+    if (!req) {
+        return false;
+    }
+    return FreezeObject(cx, req) &&
+           requestedModules_.append(req) &&
+           requestedModuleSpecifiers_.put(specifier);
 }
 
 template <typename T>
-ArrayObject* ModuleBuilder::createArray(const GCVector<T>& vector)
+ArrayObject* ModuleBuilder::createArray(const JS::Rooted<GCVector<T>>& vector)
 {
     uint32_t length = vector.length();
     RootedArrayObject array(cx_, NewDenseFullyAllocatedArray(cx_, length));
@@ -1377,7 +1445,7 @@ ArrayObject* ModuleBuilder::createArray(const GCVector<T>& vector)
 
     array->setDenseInitializedLength(length);
     for (uint32_t i = 0; i < length; i++)
-        array->initDenseElement(i, MakeElementValue(vector[i]));
+        array->initDenseElement(i, ObjectValue(*vector[i]));
 
     return array;
 }

js/src/builtin/ModuleObject.h

@@ -100,6 +100,32 @@ class ExportEntryObject : public NativeObject
 typedef Rooted<ExportEntryObject*> RootedExportEntryObject;
 typedef Handle<ExportEntryObject*> HandleExportEntryObject;
 
+class RequestedModuleObject : public NativeObject
+{
+  public:
+    enum
+    {
+        ModuleSpecifierSlot = 0,
+        LineNumberSlot,
+        ColumnNumberSlot,
+        SlotCount
+    };
+
+    static const Class class_;
+    static JSObject* initClass(JSContext* cx, HandleObject obj);
+    static bool isInstance(HandleValue value);
+    static RequestedModuleObject* create(JSContext* cx,
+                                         HandleAtom moduleSpecifier,
+                                         uint32_t lineNumber,
+                                         uint32_t columnNumber);
+    JSAtom* moduleSpecifier() const;
+    uint32_t lineNumber() const;
+    uint32_t columnNumber() const;
+};
+
+typedef Rooted<RequestedModuleObject*> RootedRequestedModuleObject;
+typedef Handle<RequestedModuleObject*> HandleRequestedModuleObject;
+
 class IndirectBindingMap
 {
   public:
@@ -318,6 +344,8 @@ class MOZ_STACK_CLASS ModuleBuilder
     explicit ModuleBuilder(ExclusiveContext* cx, HandleModuleObject module,
                            const frontend::TokenStream& tokenStream);
 
+    bool init();
+
     bool processImport(frontend::ParseNode* pn);
     bool processExport(frontend::ParseNode* pn);
     bool processExportFrom(frontend::ParseNode* pn);
@@ -334,15 +362,20 @@ class MOZ_STACK_CLASS ModuleBuilder
 
   private:
     using AtomVector = GCVector<JSAtom*>;
-    using RootedAtomVector = JS::Rooted<AtomVector>;
     using ImportEntryVector = GCVector<ImportEntryObject*>;
+    using RequestedModuleVector = GCVector<RequestedModuleObject*>;
+    using AtomSet = JS::GCHashSet<JSAtom*>;
+    using RootedAtomVector = JS::Rooted<AtomVector>;
     using RootedImportEntryVector = JS::Rooted<ImportEntryVector>;
     using RootedExportEntryVector = JS::Rooted<ExportEntryVector>;
+    using RootedRequestedModuleVector = JS::Rooted<RequestedModuleVector>;
+    using RootedAtomSet = JS::Rooted<AtomSet>;
 
     ExclusiveContext* cx_;
     RootedModuleObject module_;
     const frontend::TokenStream& tokenStream_;
-    RootedAtomVector requestedModules_;
+    RootedAtomSet requestedModuleSpecifiers_;
+    RootedRequestedModuleVector requestedModules_;
     RootedAtomVector importedBoundNames_;
     RootedImportEntryVector importEntries_;
     RootedExportEntryVector exportEntries_;
@@ -357,10 +390,10 @@ class MOZ_STACK_CLASS ModuleBuilder
     bool appendExportFromEntry(HandleAtom exportName, HandleAtom moduleRequest,
                                HandleAtom importName, frontend::ParseNode* node);
 
-    bool maybeAppendRequestedModule(HandleAtom module);
+    bool maybeAppendRequestedModule(HandleAtom specifier, frontend::ParseNode* node);
 
     template <typename T>
-    ArrayObject* createArray(const GCVector<T>& vector);
+    ArrayObject* createArray(const JS::Rooted<GCVector<T>>& vector);
 };
 
 } // namespace js

js/src/builtin/ReflectParse.cpp

@@ -3741,6 +3741,9 @@ reflect_parse(JSContext* cx, uint32_t argc, Value* vp)
             return false;
 
         ModuleBuilder builder(cx, module, parser.tokenStream);
+        if (!builder.init())
+            return false;
+
         ModuleSharedContext modulesc(cx, module, &cx->global()->emptyGlobalScope(), builder);
         pn = parser.moduleBody(&modulesc);
         if (!pn)

js/src/builtin/moz.build

@@ -0,0 +1,73 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+include('../js-config.mozbuild')
+include('../js-cxxflags.mozbuild')
+
+FINAL_LIBRARY = "js"
+
+# Includes should be relative to parent path
+LOCAL_INCLUDES += ["!..", ".."]
+
+SOURCES += [
+    'AtomicsObject.cpp',
+    'Eval.cpp',
+    'Intl.cpp',
+    'MapObject.cpp',
+    'ModuleObject.cpp',
+    'Object.cpp',
+    'Profilers.cpp',
+    'Promise.cpp',
+    'Reflect.cpp',
+    'ReflectParse.cpp',
+    'RegExp.cpp',
+    'SIMD.cpp',
+    'SymbolObject.cpp',
+    'TestingFunctions.cpp',
+    'TypedObject.cpp',
+    'WeakMapObject.cpp',
+    'WeakSetObject.cpp',
+]
+
+if CONFIG['_MSC_VER']:
+    if CONFIG['CPU_ARCH'] == 'x86':
+        SOURCES['RegExp.cpp'].no_pgo = True # Bug 772303
+
+# Prepare self-hosted JS code for embedding
+GENERATED_FILES += [('selfhosted.out.h', 'selfhosted.js')]
+selfhosted = GENERATED_FILES[('selfhosted.out.h', 'selfhosted.js')]
+selfhosted.script = 'embedjs.py:generate_selfhosted'
+selfhosted.inputs = [
+    '../js.msg',
+    'TypedObjectConstants.h',
+    'SelfHostingDefines.h',
+    'Utilities.js',
+    'Array.js',
+    'AsyncIteration.js',
+    'Classes.js',
+    'Date.js',
+    'Error.js',
+    'Function.js',
+    'Generator.js',
+    'Intl.js',
+    'IntlData.js',
+    'Iterator.js',
+    'Map.js',
+    'Module.js',
+    'Number.js',
+    'Object.js',
+    'Promise.js',
+    'Reflect.js',
+    'RegExp.js',
+    'RegExpGlobalReplaceOpt.h.js',
+    'RegExpLocalReplaceOpt.h.js',
+    'String.js',
+    'Set.js',
+    'Sorting.js',
+    'TypedArray.js',
+    'TypedObject.js',
+    'WeakMap.js',
+    'WeakSet.js',
+]

js/src/ctypes/moz.build

@@ -0,0 +1,26 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+include('../js-config.mozbuild')
+include('../js-cxxflags.mozbuild')
+
+FINAL_LIBRARY = "js"
+
+# Includes should be relative to parent path
+LOCAL_INCLUDES += ["!..", ".."]
+
+SOURCES += [
+    'CTypes.cpp',
+    'Library.cpp',
+]
+
+if not CONFIG['MOZ_SYSTEM_FFI']:
+    LOCAL_INCLUDES += [
+        '!libffi/include',
+        'libffi/src/%s' % CONFIG['FFI_TARGET_DIR'],
+    ]
+
+if CONFIG['_MSC_VER'] and CONFIG['CPU_ARCH'] == 'x86_64':
+    SOURCES['CTypes.cpp'].no_pgo = True # Bug 810661

js/src/devtools/moz.build

@@ -0,0 +1,25 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+include('../js-config.mozbuild')
+include('../js-cxxflags.mozbuild')
+
+FINAL_LIBRARY = "js"
+
+# Includes should be relative to parent path
+LOCAL_INCLUDES += ["!..", ".."]
+
+CONFIGURE_SUBST_FILES += [
+    'rootAnalysis/Makefile',
+]
+
+SOURCES += [
+    'sharkctl.cpp',
+]
+
+if CONFIG['MOZ_INSTRUMENTS']:
+    SOURCES += [
+        'Instruments.cpp',
+    ]

js/src/ds/moz.build

@@ -0,0 +1,17 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+include('../js-config.mozbuild')
+include('../js-cxxflags.mozbuild')
+
+FINAL_LIBRARY = "js"
+
+# Includes should be relative to parent path
+LOCAL_INCLUDES += ["!..", ".."]
+
+SOURCES += [
+    'LifoAlloc.cpp',
+    'MemoryProtectionExceptionHandler.cpp',
+]

js/src/editline/moz.build

@@ -5,7 +5,7 @@
 
 Library('editline')
 
-UNIFIED_SOURCES += [
+SOURCES += [
     'editline.c',
     'sysunix.c',
 ]

js/src/frontend/BytecodeCompiler.cpp

@@ -405,6 +405,9 @@ BytecodeCompiler::compileModule()
     module->init(script);
 
     ModuleBuilder builder(cx, module, parser->tokenStream);
+    if (!builder.init())
+        return nullptr;
+
     ModuleSharedContext modulesc(cx, module, enclosingScope, builder);
     ParseNode* pn = parser->moduleBody(&modulesc);
     if (!pn)

js/src/frontend/moz.build

@@ -0,0 +1,27 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+include('../js-config.mozbuild')
+include('../js-cxxflags.mozbuild')
+
+FINAL_LIBRARY = "js"
+
+# Includes should be relative to parent path
+LOCAL_INCLUDES += ["!..", ".."]
+
+SOURCES += [
+    'BytecodeCompiler.cpp',
+    'BytecodeEmitter.cpp',
+    'FoldConstants.cpp',
+    'NameFunctions.cpp',
+    'ParseNode.cpp',
+    'Parser.cpp',
+    'TokenStream.cpp',
+]
+
+GENERATED_FILES += ['ReservedWordsGenerated.h']
+ReservedWordsGenerated = GENERATED_FILES['ReservedWordsGenerated.h']
+ReservedWordsGenerated.script = 'GenerateReservedWords.py'
+ReservedWordsGenerated.inputs += ['ReservedWords.h']

js/src/gc/Heap.h

@@ -28,6 +28,7 @@
 #include "js/HeapAPI.h"
 #include "js/RootingAPI.h"
 #include "js/TracingAPI.h"
+#include "js/TraceKind.h"
 
 struct JSRuntime;
 
@@ -272,6 +273,23 @@ struct Cell
 
     static MOZ_ALWAYS_INLINE bool needWriteBarrierPre(JS::Zone* zone);
 
+    template <class T>
+    inline bool is() const {
+        return getTraceKind() == JS::MapTypeToTraceKind<T>::kind;
+    }
+
+    template<class T>
+    inline T* as() {
+        MOZ_ASSERT(this->is<T>());
+        return static_cast<T*>(this);
+    }
+
+    template <class T>
+    inline const T* as() const {
+        MOZ_ASSERT(this->is<T>());
+        return static_cast<const T*>(this);
+    }
+
 #ifdef DEBUG
     inline bool isAligned() const;
     void dump(FILE* fp) const;

js/src/gc/Policy.h

@@ -46,6 +46,7 @@ class RegExpObject;
 class SavedFrame;
 class Scope;
 class EnvironmentObject;
+class RequestedModuleObject;
 class ScriptSourceObject;
 class Shape;
 class SharedArrayBufferObject;
@@ -84,6 +85,7 @@ class JitCode;
     D(js::PlainObject*) \
     D(js::PropertyName*) \
     D(js::RegExpObject*) \
+    D(js::RequestedModuleObject*) \
     D(js::SavedFrame*) \
     D(js::Scope*) \
     D(js::ScriptSourceObject*) \

js/src/gc/moz.build

@@ -0,0 +1,31 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+include('../js-config.mozbuild')
+include('../js-cxxflags.mozbuild')
+
+FINAL_LIBRARY = "js"
+
+# Includes should be relative to parent path
+LOCAL_INCLUDES += ["!..", ".."]
+
+SOURCES += [
+    'Allocator.cpp',
+    'Barrier.cpp',
+    'GCTrace.cpp',
+    'Iteration.cpp',
+    'Marking.cpp',
+    'Memory.cpp',
+    'MemoryProfiler.cpp',
+    'Nursery.cpp',
+    'RootMarking.cpp',
+    'Statistics.cpp',
+    'StoreBuffer.cpp',
+    'Tracer.cpp',
+    'Zone.cpp',
+]
+
+if CONFIG['MOZ_DEBUG']:
+    SOURCES += ['Verifier.cpp']

js/src/gdb/moz.build

@@ -5,6 +5,8 @@
 
 GeckoProgram('gdb-tests', linkage=None)
 
+include('../js-cxxflags.mozbuild')
+
 UNIFIED_SOURCES += [
     'gdb-tests.cpp',
     'tests/test-asmjs.cpp',
@@ -40,13 +42,6 @@ if CONFIG['MOZ_ICU_DATA_ARCHIVE']:
 
 OS_LIBS += CONFIG['MOZ_ZLIB_LIBS']
 
-if CONFIG['GNU_CXX']:
-    CXXFLAGS += ['-Wno-shadow', '-fno-strict-aliasing']
-
-# This is intended as a temporary workaround to enable VS2015.
-if CONFIG['_MSC_VER']:
-    CXXFLAGS += ['-wd4312']
-
 DEFINES['topsrcdir'] = '%s/js/src' % TOPSRCDIR
 FINAL_TARGET_PP_FILES += ['gdb-tests-gdb.py.in']
 OBJDIR_FILES.js.src.gdb += ['!/dist/bin/gdb-tests-gdb.py']

js/src/irregexp/moz.build

@@ -0,0 +1,23 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+include('../js-config.mozbuild')
+include('../js-cxxflags.mozbuild')
+
+FINAL_LIBRARY = "js"
+
+# Includes should be relative to parent path
+LOCAL_INCLUDES += ["!..", ".."]
+
+SOURCES += [
+    'NativeRegExpMacroAssembler.cpp',
+    'RegExpAST.cpp',
+    'RegExpCharacters.cpp',
+    'RegExpEngine.cpp',
+    'RegExpInterpreter.cpp',
+    'RegExpMacroAssembler.cpp',
+    'RegExpParser.cpp',
+    'RegExpStack.cpp',
+]

js/src/jit-test/tests/modules/requested-modules.js

@@ -5,7 +5,7 @@ function testRequestedModules(source, expected) {
     var actual = module.requestedModules;
     assertEq(actual.length, expected.length);
     for (var i = 0; i < actual.length; i++) {
-        assertEq(actual[i], expected[i]);
+        assertEq(actual[i].moduleSpecifier, expected[i]);
     }
 }
 

js/src/jit/moz.build

@@ -0,0 +1,233 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+include('../js-config.mozbuild')
+include('../js-cxxflags.mozbuild')
+
+FINAL_LIBRARY = "js"
+
+# Includes should be relative to parent path
+LOCAL_INCLUDES += ["!..", ".."]
+
+UNIFIED_SOURCES += [
+    'AliasAnalysis.cpp',
+    'AliasAnalysisShared.cpp',
+    'AlignmentMaskAnalysis.cpp',
+    'BacktrackingAllocator.cpp',
+    'Bailouts.cpp',
+    'BaselineBailouts.cpp',
+    'BaselineCacheIR.cpp',
+    'BaselineCompiler.cpp',
+    'BaselineDebugModeOSR.cpp',
+    'BaselineFrame.cpp',
+    'BaselineFrameInfo.cpp',
+    'BaselineIC.cpp',
+    'BaselineInspector.cpp',
+    'BaselineJIT.cpp',
+    'BitSet.cpp',
+    'BytecodeAnalysis.cpp',
+    'C1Spewer.cpp',
+    'CacheIR.cpp',
+    'CodeGenerator.cpp',
+    'CompileWrappers.cpp',
+    'EagerSimdUnbox.cpp',
+    'EdgeCaseAnalysis.cpp',
+    'EffectiveAddressAnalysis.cpp',
+    'ExecutableAllocator.cpp',
+    'FlowAliasAnalysis.cpp',
+    'FoldLinearArithConstants.cpp',
+    'InstructionReordering.cpp',
+    'Ion.cpp',
+    'IonAnalysis.cpp',
+    'IonBuilder.cpp',
+    'IonCaches.cpp',
+    'IonOptimizationLevels.cpp',
+    'JitcodeMap.cpp',
+    'JitFrames.cpp',
+    'JitOptions.cpp',
+    'JitSpewer.cpp',
+    'JSONSpewer.cpp',
+    'LICM.cpp',
+    'Linker.cpp',
+    'LIR.cpp',
+    'LoopUnroller.cpp',
+    'Lowering.cpp',
+    'MacroAssembler.cpp',
+    'MCallOptimize.cpp',
+    'MIR.cpp',
+    'MIRGraph.cpp',
+    'MoveResolver.cpp',
+    'OptimizationTracking.cpp',
+    'PerfSpewer.cpp',
+    'ProcessExecutableMemory.cpp',
+    'RangeAnalysis.cpp',
+    'Recover.cpp',
+    'RegisterAllocator.cpp',
+    'RematerializedFrame.cpp',
+    'Safepoints.cpp',
+    'ScalarReplacement.cpp',
+    'shared/BaselineCompiler-shared.cpp',
+    'shared/CodeGenerator-shared.cpp',
+    'shared/Lowering-shared.cpp',
+    'SharedIC.cpp',
+    'Sink.cpp',
+    'Snapshots.cpp',
+    'StupidAllocator.cpp',
+    'TypedObjectPrediction.cpp',
+    'TypePolicy.cpp',
+    'ValueNumbering.cpp',
+    'VMFunctions.cpp',
+    'WasmBCE.cpp',
+]
+
+if CONFIG['MOZ_DEBUG']:
+    UNIFIED_SOURCES += ['Disassembler.cpp']
+
+if not CONFIG['ENABLE_ION']:
+    UNIFIED_SOURCES += [
+        'none/Trampoline-none.cpp'
+    ]
+elif CONFIG['JS_CODEGEN_X86'] or CONFIG['JS_CODEGEN_X64']:
+    UNIFIED_SOURCES += [
+        'x86-shared/Architecture-x86-shared.cpp',
+        'x86-shared/Assembler-x86-shared.cpp',
+        'x86-shared/AssemblerBuffer-x86-shared.cpp',
+        'x86-shared/BaselineCompiler-x86-shared.cpp',
+        'x86-shared/BaselineIC-x86-shared.cpp',
+        'x86-shared/CodeGenerator-x86-shared.cpp',
+        'x86-shared/Lowering-x86-shared.cpp',
+        'x86-shared/MacroAssembler-x86-shared.cpp',
+        'x86-shared/MoveEmitter-x86-shared.cpp',
+    ]
+    UNIFIED_SOURCES += [
+        'x86-shared/Disassembler-x86-shared.cpp',  # using namespace js::jit::X86Encoding;
+    ]
+    if CONFIG['JS_CODEGEN_X64']:
+        UNIFIED_SOURCES += [
+            'x64/Assembler-x64.cpp',
+            'x64/Bailouts-x64.cpp',
+            'x64/BaselineCompiler-x64.cpp',
+            'x64/BaselineIC-x64.cpp',
+            'x64/CodeGenerator-x64.cpp',
+            'x64/Lowering-x64.cpp',
+            'x64/MacroAssembler-x64.cpp',
+            'x64/SharedIC-x64.cpp',
+            'x64/Trampoline-x64.cpp',
+        ]
+    else:
+        UNIFIED_SOURCES += [
+            'x86/Assembler-x86.cpp',
+            'x86/Bailouts-x86.cpp',
+            'x86/BaselineCompiler-x86.cpp',
+            'x86/BaselineIC-x86.cpp',
+            'x86/CodeGenerator-x86.cpp',
+            'x86/Lowering-x86.cpp',
+            'x86/MacroAssembler-x86.cpp',
+            'x86/SharedIC-x86.cpp',
+            'x86/Trampoline-x86.cpp',
+        ]
+elif CONFIG['JS_CODEGEN_ARM']:
+    UNIFIED_SOURCES += [
+        'arm/Architecture-arm.cpp',
+        'arm/Assembler-arm.cpp',
+        'arm/Bailouts-arm.cpp',
+        'arm/BaselineCompiler-arm.cpp',
+        'arm/BaselineIC-arm.cpp',
+        'arm/CodeGenerator-arm.cpp',
+        'arm/disasm/Constants-arm.cpp',
+        'arm/disasm/Disasm-arm.cpp',
+        'arm/Lowering-arm.cpp',
+        'arm/MacroAssembler-arm.cpp',
+        'arm/MoveEmitter-arm.cpp',
+        'arm/SharedIC-arm.cpp',
+        'arm/Trampoline-arm.cpp',
+    ]
+    if CONFIG['JS_SIMULATOR_ARM']:
+        UNIFIED_SOURCES += [
+            'arm/Simulator-arm.cpp'
+        ]
+    elif CONFIG['OS_ARCH'] == 'Darwin':
+        UNIFIED_SOURCES += [
+            'arm/llvm-compiler-rt/arm/aeabi_idivmod.S',
+            'arm/llvm-compiler-rt/arm/aeabi_uidivmod.S',
+        ]
+elif CONFIG['JS_CODEGEN_ARM64']:
+    UNIFIED_SOURCES += [
+        'arm64/Architecture-arm64.cpp',
+        'arm64/Assembler-arm64.cpp',
+        'arm64/Bailouts-arm64.cpp',
+        'arm64/BaselineIC-arm64.cpp',
+        'arm64/CodeGenerator-arm64.cpp',
+        'arm64/Lowering-arm64.cpp',
+        'arm64/MacroAssembler-arm64.cpp',
+        'arm64/MoveEmitter-arm64.cpp',
+        'arm64/SharedIC-arm64.cpp',
+        'arm64/Trampoline-arm64.cpp',
+        'arm64/vixl/Assembler-vixl.cpp',
+        'arm64/vixl/Cpu-vixl.cpp',
+        'arm64/vixl/Decoder-vixl.cpp',
+        'arm64/vixl/Disasm-vixl.cpp',
+        'arm64/vixl/Instructions-vixl.cpp',
+        'arm64/vixl/Instrument-vixl.cpp',
+        'arm64/vixl/MacroAssembler-vixl.cpp',
+        'arm64/vixl/MozAssembler-vixl.cpp',
+        'arm64/vixl/MozInstructions-vixl.cpp',
+        'arm64/vixl/Utils-vixl.cpp'
+    ]
+    if CONFIG['JS_SIMULATOR_ARM64']:
+        UNIFIED_SOURCES += [
+            'arm64/vixl/Debugger-vixl.cpp',
+            'arm64/vixl/Logic-vixl.cpp',
+            'arm64/vixl/MozSimulator-vixl.cpp',
+            'arm64/vixl/Simulator-vixl.cpp'
+        ]
+elif CONFIG['JS_CODEGEN_MIPS32'] or CONFIG['JS_CODEGEN_MIPS64']:
+    UNIFIED_SOURCES += [
+        'mips-shared/Architecture-mips-shared.cpp',
+        'mips-shared/Assembler-mips-shared.cpp',
+        'mips-shared/Bailouts-mips-shared.cpp',
+        'mips-shared/BaselineCompiler-mips-shared.cpp',
+        'mips-shared/BaselineIC-mips-shared.cpp',
+        'mips-shared/CodeGenerator-mips-shared.cpp',
+        'mips-shared/Lowering-mips-shared.cpp',
+        'mips-shared/MacroAssembler-mips-shared.cpp',
+        'mips-shared/MoveEmitter-mips-shared.cpp',
+    ]
+    if CONFIG['JS_CODEGEN_MIPS32']:
+        UNIFIED_SOURCES += [
+            'mips32/Architecture-mips32.cpp',
+            'mips32/Assembler-mips32.cpp',
+            'mips32/Bailouts-mips32.cpp',
+            'mips32/BaselineCompiler-mips32.cpp',
+            'mips32/BaselineIC-mips32.cpp',
+            'mips32/CodeGenerator-mips32.cpp',
+            'mips32/Lowering-mips32.cpp',
+            'mips32/MacroAssembler-mips32.cpp',
+            'mips32/MoveEmitter-mips32.cpp',
+            'mips32/SharedIC-mips32.cpp',
+            'mips32/Trampoline-mips32.cpp',
+        ]
+        if CONFIG['JS_SIMULATOR_MIPS32']:
+            UNIFIED_SOURCES += [
+                'mips32/Simulator-mips32.cpp'
+            ]
+    elif CONFIG['JS_CODEGEN_MIPS64']:
+        UNIFIED_SOURCES += [
+            'mips64/Architecture-mips64.cpp',
+            'mips64/Assembler-mips64.cpp',
+            'mips64/Bailouts-mips64.cpp',
+            'mips64/BaselineCompiler-mips64.cpp',
+            'mips64/BaselineIC-mips64.cpp',
+            'mips64/CodeGenerator-mips64.cpp',
+            'mips64/Lowering-mips64.cpp',
+            'mips64/MacroAssembler-mips64.cpp',
+            'mips64/MoveEmitter-mips64.cpp',
+            'mips64/SharedIC-mips64.cpp',
+            'mips64/Trampoline-mips64.cpp',
+        ]
+        if CONFIG['JS_SIMULATOR_MIPS64']:
+            UNIFIED_SOURCES += [
+                'mips64/Simulator-mips64.cpp'
+            ]

js/src/js-config.mozbuild

@@ -0,0 +1,34 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+FILES_PER_UNIFIED_FILE = 6
+
+# Also set in shell/moz.build
+DEFINES['ENABLE_SHARED_ARRAY_BUFFER'] = True
+
+DEFINES['EXPORT_JS_API'] = True
+
+if CONFIG['NIGHTLY_BUILD']:
+    DEFINES['ENABLE_BINARYDATA'] = True
+
+if CONFIG['NIGHTLY_BUILD']:
+    DEFINES['ENABLE_SIMD'] = True
+
+if CONFIG['JS_HAS_CTYPES']:
+    DEFINES['JS_HAS_CTYPES'] = True
+    for var in ('DLL_PREFIX', 'DLL_SUFFIX'):
+        DEFINES[var] = '"%s"' % CONFIG[var]
+
+if CONFIG['MOZ_LINKER']:
+    DEFINES['MOZ_LINKER'] = True
+
+if CONFIG['JS_HAS_CTYPES']:
+    if not CONFIG['MOZ_SYSTEM_FFI']:
+        # Windows needs this to be linked with a static library.
+        DEFINES['FFI_BUILDING'] = True
+
+if CONFIG['OS_ARCH'] == 'WINNT':
+    DEFINES['_CRT_RAND_S'] = True
+    DEFINES['NOMINMAX'] = True

js/src/js-cxxflags.mozbuild

@@ -0,0 +1,42 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+CFLAGS += CONFIG['MOZ_ICU_CFLAGS']
+CXXFLAGS += CONFIG['MOZ_ICU_CFLAGS']
+
+if CONFIG['_MSC_VER']:
+    # Prevent floating point errors caused by VC++ optimizations
+    CFLAGS += ['-fp:precise']
+    CXXFLAGS += ['-fp:precise']  
+
+    # C4805 warns mixing bool with other integral types in computation.
+    # But given the conversion from bool is specified, and this is a
+    # pattern widely used in code in js/src, suppress this warning here.
+    CXXFLAGS += ['-wd4805']
+
+    # C4661 ("no suitable definition provided for explicit template
+    # instantiation request") is emitted for all Parser methods that
+    # have a Parser<FullParseHandler> definition but no
+    # Parser<SyntaxParseHandler> definition, see bug 1167030.
+    CXXFLAGS += ['-wd4661', '-we4067', '-we4258', '-we4275']
+    CXXFLAGS += ['-wd4146'] # FIXME: unary minus operator applied to unsigned type (bug 1229189)
+
+    # This is intended as a temporary hack to support building with VS2015.
+    # 'noexcept' used with no exception handling mode specified;
+    # termination on exception is not guaranteed. Specify /EHsc
+    CXXFLAGS += ['-wd4577', '-wd4312']
+
+if CONFIG['GNU_CXX']:
+    # Disable strict-aliasing for GCC, which is enabled by default
+    # starting with version 7.1, see Mozilla bug 1363009.
+    CXXFLAGS += ['-Wno-shadow', '-Werror=format', '-fno-strict-aliasing']
+
+if CONFIG['JS_HAS_CTYPES'] and CONFIG['MOZ_SYSTEM_FFI']:
+    CXXFLAGS += CONFIG['MOZ_FFI_CFLAGS']
+
+if CONFIG['JS_CODEGEN_ARM'] and CONFIG['JS_SIMULATOR_ARM']:
+    # Configuration used only for testing.
+    if CONFIG['OS_ARCH'] == 'Linux' or CONFIG['OS_ARCH'] == 'Darwin':
+        CXXFLAGS += [ '-msse2', '-mfpmath=sse' ]

js/src/js-testing.mozbuild

@@ -0,0 +1,48 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+component_engine = ('Core', 'JavaScript Engine')
+component_gc     = ('Core', 'JavaScript: GC')
+component_intl   = ('Core', 'JavaScript: Internationalization API')
+component_jit    = ('Core', 'JavaScript Engine: JIT')
+component_stl    = ('Core', 'JavaScript: Standard Library')
+
+with Files('../public/**'):
+    BUG_COMPONENT = component_engine
+with Files('*'):
+    BUG_COMPONENT = component_engine
+
+with Files('wasm/**'):
+    BUG_COMPONENT = component_jit
+with Files('builtin/**'):
+    BUG_COMPONENT = component_stl
+with Files('ctypes/**'):
+    BUG_COMPONENT = ('Core', 'js-ctypes')
+with Files('gc/**'):
+    BUG_COMPONENT = component_gc
+with Files('jit/**'):
+    BUG_COMPONENT = component_jit
+
+# File-specific metadata
+for gcfile in ['jsgc*', 'devtools/rootAnalysis', 'devtools/gc-ubench', 'devtools/gctrace']:
+    with Files(gcfile):
+        BUG_COMPONENT = component_gc
+for header in ('GCAnnotations.h', 'GCAPI.h', 'HeapAPI.h', 'RootingAPI.h', 'SliceBudget.h', 'SweepingAPI.h', 'TraceKind.h', 'TracingAPI.h', 'WeakMapPtr.h'):
+    with Files('../public/' + header):
+        BUG_COMPONENT = component_gc
+
+for stlfile in ['jsarray.*', 'jsbool*', 'jsdate.*', 'jsnum.*', 'json.*', 'jsstr.*']:
+    with Files(stlfile):
+        BUG_COMPONENT = component_stl
+
+with Files('builtin/Intl*'):
+    BUG_COMPONENT = component_intl
+with Files('builtin/make_intl_data.py'):
+    BUG_COMPONENT = component_intl
+
+with Files('../public/TrackedOptimizationInfo.h'):
+    BUG_COMPONENT = component_jit
+
+TEST_DIRS += ['jsapi-tests', 'tests', 'gdb']

js/src/jsapi-tests/moz.build

@@ -3,6 +3,8 @@
 # License, v. 2.0. If a copy of the MPL was not distributed with this
 # file, You can obtain one at http://mozilla.org/MPL/2.0/.
 
+include(../js-cxxflags.mozbuild)
+
 GeckoProgram('jsapi-tests', linkage=None)
 
 UNIFIED_SOURCES += [
@@ -142,12 +144,5 @@ USE_LIBS += [
 
 OS_LIBS += CONFIG['MOZ_ZLIB_LIBS']
 
-if CONFIG['GNU_CXX']:
-    CXXFLAGS += ['-Wno-shadow', '-Werror=format', '-fno-strict-aliasing']
-
-# This is intended as a temporary workaround to enable VS2015.
-if CONFIG['_MSC_VER']:
-    CXXFLAGS += ['-wd4312']
-
 DEFINES['topsrcdir'] = '%s/js/src' % TOPSRCDIR
 OBJDIR_PP_FILES.js.src['jsapi-tests'] += ['jsapi-tests-gdb.py.in']

js/src/jsapi.cpp

@@ -4729,13 +4729,28 @@ JS::GetRequestedModules(JSContext* cx, JS::HandleObject moduleArg)
     return &moduleArg->as<ModuleObject>().requestedModules();
 }
 
-JS_PUBLIC_API(JSScript*)
-JS::GetModuleScript(JSContext* cx, JS::HandleObject moduleArg)
+JS_PUBLIC_API(JSString*)
+JS::GetRequestedModuleSpecifier(JSContext* cx, JS::HandleValue value)
 {
     AssertHeapIsIdle(cx);
     CHECK_REQUEST(cx);
-    assertSameCompartment(cx, moduleArg);
-    return moduleArg->as<ModuleObject>().script();
+    assertSameCompartment(cx, value);
+    JSObject* obj = &value.toObject();
+    return obj->as<RequestedModuleObject>().moduleSpecifier();
+}
+
+JS_PUBLIC_API(void)
+JS::GetRequestedModuleSourcePos(JSContext* cx, JS::HandleValue value,
+                                uint32_t* lineNumber, uint32_t* columnNumber)
+{
+    AssertHeapIsIdle(cx);
+    CHECK_REQUEST(cx);
+    assertSameCompartment(cx, value);
+    MOZ_ASSERT(lineNumber);
+    MOZ_ASSERT(columnNumber);
+    auto& requested = value.toObject().as<RequestedModuleObject>();
+    *lineNumber = requested.lineNumber();
+    *columnNumber = requested.columnNumber();
 }
 
 JS_PUBLIC_API(JSObject*)
@@ -7039,7 +7054,9 @@ JS::CaptureCurrentStack(JSContext* cx, JS::MutableHandleObject stackp,
 {
     AssertHeapIsIdle(cx);
     CHECK_REQUEST(cx);
-    MOZ_RELEASE_ASSERT(cx->compartment());
+    if (!cx->compartment()) {
+      return false;
+    }
 
     JSCompartment* compartment = cx->compartment();
     Rooted<SavedFrame*> frame(cx);

js/src/jsapi.h

@@ -4386,18 +4386,27 @@ ModuleEvaluate(JSContext* cx, JS::HandleObject moduleRecord);
  * Get a list of the module specifiers used by a source text module
  * record to request importation of modules.
  *
- * The result is a JavaScript array of string values.  To extract the individual
- * values use only JS_GetArrayLength and JS_GetElement with indices 0 to
- * length - 1.
+ * The result is a JavaScript array of object values.  To extract the individual
+ * values use only JS_GetArrayLength and JS_GetElement with indices 0 to length
+ * - 1.
+ *
+ * The element values are objects with the following properties:
+ *  - moduleSpecifier: the module specifier string
+ *  - lineNumber: the line number of the import in the source text
+ *  - columnNumber: the column number of the import in the source text
+ *
+ * These property values can be extracted with GetRequestedModuleSpecifier() and
+ * GetRequestedModuleSourcePos()
  */
 extern JS_PUBLIC_API(JSObject*)
 GetRequestedModules(JSContext* cx, JS::HandleObject moduleRecord);
 
-/*
- * Get the script associated with a module.
- */
-extern JS_PUBLIC_API(JSScript*)
-GetModuleScript(JSContext* cx, JS::HandleObject moduleRecord);
+extern JS_PUBLIC_API(JSString*)
+GetRequestedModuleSpecifier(JSContext* cx, JS::HandleValue requestedModuleObject);
+
+extern JS_PUBLIC_API(void)
+GetRequestedModuleSourcePos(JSContext* cx, JS::HandleValue requestedModuleObject,
+                            uint32_t* lineNumber, uint32_t* columnNumber);
 
 } /* namespace JS */
 

js/src/moz.build

@@ -3,65 +3,15 @@
 # License, v. 2.0. If a copy of the MPL was not distributed with this
 # file, You can obtain one at http://mozilla.org/MPL/2.0/.
 
-# Directory metadata
-component_engine = ('Core', 'JavaScript Engine')
-component_gc     = ('Core', 'JavaScript: GC')
-component_intl   = ('Core', 'JavaScript: Internationalization API')
-component_jit    = ('Core', 'JavaScript Engine: JIT')
-component_stl    = ('Core', 'JavaScript: Standard Library')
-
-FILES_PER_UNIFIED_FILE = 6
-
-with Files('../public/**'):
-    BUG_COMPONENT = component_engine
-with Files('*'):
-    BUG_COMPONENT = component_engine
-
-with Files('wasm/**'):
-    BUG_COMPONENT = component_jit
-with Files('builtin/**'):
-    BUG_COMPONENT = component_stl
-with Files('ctypes/**'):
-    BUG_COMPONENT = ('Core', 'js-ctypes')
-with Files('gc/**'):
-    BUG_COMPONENT = component_gc
-with Files('jit/**'):
-    BUG_COMPONENT = component_jit
-
-# File-specific metadata
-for gcfile in ['jsgc*', 'devtools/rootAnalysis', 'devtools/gc-ubench', 'devtools/gctrace']:
-    with Files(gcfile):
-        BUG_COMPONENT = component_gc
-for header in ('GCAnnotations.h', 'GCAPI.h', 'HeapAPI.h', 'RootingAPI.h', 'SliceBudget.h', 'SweepingAPI.h', 'TraceKind.h', 'TracingAPI.h', 'WeakMapPtr.h'):
-    with Files('../public/' + header):
-        BUG_COMPONENT = component_gc
-
-for stlfile in ['jsarray.*', 'jsbool*', 'jsdate.*', 'jsnum.*', 'json.*', 'jsstr.*']:
-    with Files(stlfile):
-        BUG_COMPONENT = component_stl
-
-with Files('builtin/Intl*'):
-    BUG_COMPONENT = component_intl
-with Files('builtin/make_intl_data.py'):
-    BUG_COMPONENT = component_intl
-
-with Files('../public/TrackedOptimizationInfo.h'):
-    BUG_COMPONENT = component_jit
-
-
-if CONFIG['JS_BUNDLED_EDITLINE']:
-    DIRS += ['editline']
-
-if not CONFIG['JS_DISABLE_SHELL']:
-    DIRS += ['shell']
-
-TEST_DIRS += ['jsapi-tests', 'tests', 'gdb']
+include('js-config.mozbuild')
+include('js-cxxflags.mozbuild')
+include('js-testing.mozbuild')
 
 CONFIGURE_SUBST_FILES += [
-    'devtools/rootAnalysis/Makefile',
     'js-config',
     'js.pc',
 ]
+
 CONFIGURE_DEFINE_FILES += [
     'js-confdefs.h',
 ]
@@ -80,7 +30,6 @@ CONFIGURE_DEFINE_FILES += [
 
 if CONFIG['HAVE_DTRACE']:
     GENERATED_FILES += ['javascript-trace.h']
-
     EXPORTS += ['!javascript-trace.h']
 
 # Changes to internal header files, used externally, massively slow down
@@ -101,7 +50,6 @@ EXPORTS += [
     'jstypes.h',
     'jsversion.h',
     'jswrapper.h',
-    'perf/jsperf.h',
 ]
 
 # If you add a header here, add it to js/src/jsapi-tests/testIntTypesABI.cpp so
@@ -154,122 +102,36 @@ EXPORTS.js += [
     '../public/WeakMapPtr.h',
 ]
 
+DIRS += [
+    'builtin',
+    'devtools',
+    'ds',
+    'frontend',
+    'gc',
+    'irregexp',
+    'jit',
+    'perf',
+    'proxy',
+    'threading',
+    'wasm',
+]
+
+if CONFIG['JS_HAS_CTYPES']:
+    DIRS += ['ctypes']
+
+if CONFIG['JS_BUNDLED_EDITLINE']:
+    DIRS += ['editline']
+
+if CONFIG['JS_NEW_REGEXP']:
+    DIRS += ['regexp']
+
+if not CONFIG['JS_DISABLE_SHELL']:
+    DIRS += ['shell']
+
+if CONFIG['MOZ_VTUNE']:
+    DIRS += ['vtune']
+
 UNIFIED_SOURCES += [
-    'builtin/AtomicsObject.cpp',
-    'builtin/Eval.cpp',
-    'builtin/Intl.cpp',
-    'builtin/MapObject.cpp',
-    'builtin/ModuleObject.cpp',
-    'builtin/Object.cpp',
-    'builtin/Profilers.cpp',
-    'builtin/Promise.cpp',
-    'builtin/Reflect.cpp',
-    'builtin/ReflectParse.cpp',
-    'builtin/SIMD.cpp',
-    'builtin/SymbolObject.cpp',
-    'builtin/TestingFunctions.cpp',
-    'builtin/TypedObject.cpp',
-    'builtin/WeakMapObject.cpp',
-    'builtin/WeakSetObject.cpp',
-    'devtools/sharkctl.cpp',
-    'ds/LifoAlloc.cpp',
-    'ds/MemoryProtectionExceptionHandler.cpp',
-    'frontend/BytecodeCompiler.cpp',
-    'frontend/BytecodeEmitter.cpp',
-    'frontend/FoldConstants.cpp',
-    'frontend/NameFunctions.cpp',
-    'frontend/ParseNode.cpp',
-    'frontend/TokenStream.cpp',
-    'gc/Allocator.cpp',
-    'gc/Barrier.cpp',
-    'gc/GCTrace.cpp',
-    'gc/Iteration.cpp',
-    'gc/Marking.cpp',
-    'gc/Memory.cpp',
-    'gc/MemoryProfiler.cpp',
-    'gc/Nursery.cpp',
-    'gc/RootMarking.cpp',
-    'gc/Statistics.cpp',
-    'gc/Tracer.cpp',
-    'gc/Verifier.cpp',
-    'gc/Zone.cpp',
-    'irregexp/NativeRegExpMacroAssembler.cpp',
-    'irregexp/RegExpAST.cpp',
-    'irregexp/RegExpCharacters.cpp',
-    'irregexp/RegExpEngine.cpp',
-    'irregexp/RegExpInterpreter.cpp',
-    'irregexp/RegExpMacroAssembler.cpp',
-    'irregexp/RegExpParser.cpp',
-    'irregexp/RegExpStack.cpp',
-    'jit/AliasAnalysis.cpp',
-    'jit/AliasAnalysisShared.cpp',
-    'jit/AlignmentMaskAnalysis.cpp',
-    'jit/BacktrackingAllocator.cpp',
-    'jit/Bailouts.cpp',
-    'jit/BaselineBailouts.cpp',
-    'jit/BaselineCacheIR.cpp',
-    'jit/BaselineCompiler.cpp',
-    'jit/BaselineDebugModeOSR.cpp',
-    'jit/BaselineFrame.cpp',
-    'jit/BaselineFrameInfo.cpp',
-    'jit/BaselineIC.cpp',
-    'jit/BaselineInspector.cpp',
-    'jit/BaselineJIT.cpp',
-    'jit/BitSet.cpp',
-    'jit/BytecodeAnalysis.cpp',
-    'jit/C1Spewer.cpp',
-    'jit/CacheIR.cpp',
-    'jit/CodeGenerator.cpp',
-    'jit/CompileWrappers.cpp',
-    'jit/Disassembler.cpp',
-    'jit/EagerSimdUnbox.cpp',
-    'jit/EdgeCaseAnalysis.cpp',
-    'jit/EffectiveAddressAnalysis.cpp',
-    'jit/ExecutableAllocator.cpp',
-    'jit/FlowAliasAnalysis.cpp',
-    'jit/FoldLinearArithConstants.cpp',
-    'jit/InstructionReordering.cpp',
-    'jit/Ion.cpp',
-    'jit/IonAnalysis.cpp',
-    'jit/IonBuilder.cpp',
-    'jit/IonCaches.cpp',
-    'jit/IonOptimizationLevels.cpp',
-    'jit/JitcodeMap.cpp',
-    'jit/JitFrames.cpp',
-    'jit/JitOptions.cpp',
-    'jit/JitSpewer.cpp',
-    'jit/JSONSpewer.cpp',
-    'jit/LICM.cpp',
-    'jit/Linker.cpp',
-    'jit/LIR.cpp',
-    'jit/LoopUnroller.cpp',
-    'jit/Lowering.cpp',
-    'jit/MacroAssembler.cpp',
-    'jit/MCallOptimize.cpp',
-    'jit/MIR.cpp',
-    'jit/MIRGraph.cpp',
-    'jit/MoveResolver.cpp',
-    'jit/OptimizationTracking.cpp',
-    'jit/PerfSpewer.cpp',
-    'jit/ProcessExecutableMemory.cpp',
-    'jit/RangeAnalysis.cpp',
-    'jit/Recover.cpp',
-    'jit/RegisterAllocator.cpp',
-    'jit/RematerializedFrame.cpp',
-    'jit/Safepoints.cpp',
-    'jit/ScalarReplacement.cpp',
-    'jit/shared/BaselineCompiler-shared.cpp',
-    'jit/shared/CodeGenerator-shared.cpp',
-    'jit/shared/Lowering-shared.cpp',
-    'jit/SharedIC.cpp',
-    'jit/Sink.cpp',
-    'jit/Snapshots.cpp',
-    'jit/StupidAllocator.cpp',
-    'jit/TypedObjectPrediction.cpp',
-    'jit/TypePolicy.cpp',
-    'jit/ValueNumbering.cpp',
-    'jit/VMFunctions.cpp',
-    'jit/WasmBCE.cpp',
     'jsalloc.cpp',
     'jsapi.cpp',
     'jsbool.cpp',
@@ -291,16 +153,23 @@ UNIFIED_SOURCES += [
     'jsscript.cpp',
     'jsstr.cpp',
     'jsweakmap.cpp',
-    'perf/jsperf.cpp',
-    'proxy/BaseProxyHandler.cpp',
-    'proxy/CrossCompartmentWrapper.cpp',
-    'proxy/DeadObjectProxy.cpp',
-    'proxy/OpaqueCrossCompartmentWrapper.cpp',
-    'proxy/Proxy.cpp',
-    'proxy/ScriptedProxyHandler.cpp',
-    'proxy/SecurityWrapper.cpp',
-    'proxy/Wrapper.cpp',
-    'threading/Mutex.cpp',
+]
+
+# Can't be built unified
+SOURCES += [
+    'jsarray.cpp',
+    'jsatom.cpp',
+    'jsdtoa.cpp',
+    'jsmath.cpp',
+    'jsutil.cpp',
+]
+
+# Suppress warnings in third-party code.
+# We are keeping this in the main moz.build because it is file specific
+if CONFIG['CLANG_CXX'] or CONFIG['GNU_CXX']:
+    SOURCES['jsdtoa.cpp'].flags += ['-Wno-implicit-fallthrough']
+
+UNIFIED_SOURCES += [
     'vm/ArgumentsObject.cpp',
     'vm/ArrayBufferObject.cpp',
     'vm/AsyncFunction.cpp',
@@ -359,54 +228,10 @@ UNIFIED_SOURCES += [
     'vm/Value.cpp',
     'vm/WeakMapPtr.cpp',
     'vm/Xdr.cpp',
-    'wasm/AsmJS.cpp',
-    'wasm/WasmBaselineCompile.cpp',
-    'wasm/WasmBinaryFormat.cpp',
-    'wasm/WasmBinaryIterator.cpp',
-    'wasm/WasmBinaryToAST.cpp',
-    'wasm/WasmBinaryToExperimentalText.cpp',
-    'wasm/WasmBinaryToText.cpp',
-    'wasm/WasmCode.cpp',
-    'wasm/WasmCompartment.cpp',
-    'wasm/WasmCompile.cpp',
-    'wasm/WasmFrameIterator.cpp',
-    'wasm/WasmGenerator.cpp',
-    'wasm/WasmInstance.cpp',
-    'wasm/WasmIonCompile.cpp',
-    'wasm/WasmJS.cpp',
-    'wasm/WasmModule.cpp',
-    'wasm/WasmSignalHandlers.cpp',
-    'wasm/WasmStubs.cpp',
-    'wasm/WasmTable.cpp',
-    'wasm/WasmTextToBinary.cpp',
-    'wasm/WasmTextUtils.cpp',
-    'wasm/WasmTypes.cpp'
 ]
 
-# jsarray.cpp and jsatom.cpp cannot be built in unified mode because
-# xpcshell is broken during packaging when compiled with gcc-4.8.2
-# builtin/RegExp.cpp cannot be built in unified mode because it is built
-# without PGO
-# frontend/Parser.cpp cannot be built in unified mode because of explicit
-# template instantiations.
-# jsdtoa.cpp cannot be built in unified mode because we want to suppress
-# compiler warnings in third-party dtoa.c.
-# jsmath.cpp cannot be built in unified mode because it needs to pull rand_s
-# from <stdlib.h> on Windows through a preprocessor define.
-# jsutil.cpp cannot be built in unified mode because it is needed for
-# check-vanilla-allocations.
-# StoreBuffer.cpp cannot be built in unified because its template
-# instantiations may or may not be needed depending on what it gets bundled
-# with.
+# Can't be built unified
 SOURCES += [
-    'builtin/RegExp.cpp',
-    'frontend/Parser.cpp',
-    'gc/StoreBuffer.cpp',
-    'jsarray.cpp',
-    'jsatom.cpp',
-    'jsdtoa.cpp',
-    'jsmath.cpp',
-    'jsutil.cpp',
     'vm/Initialization.cpp',
 ]
 
@@ -415,217 +240,14 @@ if CONFIG['JS_POSIX_NSPR']:
         'vm/PosixNSPR.cpp',
     ]
 
-if CONFIG['MOZ_INSTRUMENTS']:
-    SOURCES += [
-        'devtools/Instruments.cpp',
-    ]
-
 if CONFIG['ENABLE_TRACE_LOGGING']:
+    # Can't be built unified
     SOURCES += [
         'vm/TraceLogging.cpp',
         'vm/TraceLoggingGraph.cpp',
         'vm/TraceLoggingTypes.cpp',
     ]
 
-if not CONFIG['ENABLE_ION']:
-    UNIFIED_SOURCES += [
-        'jit/none/Trampoline-none.cpp'
-    ]
-elif CONFIG['JS_CODEGEN_X86'] or CONFIG['JS_CODEGEN_X64']:
-    UNIFIED_SOURCES += [
-        'jit/x86-shared/Architecture-x86-shared.cpp',
-        'jit/x86-shared/Assembler-x86-shared.cpp',
-        'jit/x86-shared/AssemblerBuffer-x86-shared.cpp',
-        'jit/x86-shared/BaselineCompiler-x86-shared.cpp',
-        'jit/x86-shared/BaselineIC-x86-shared.cpp',
-        'jit/x86-shared/CodeGenerator-x86-shared.cpp',
-        'jit/x86-shared/Lowering-x86-shared.cpp',
-        'jit/x86-shared/MacroAssembler-x86-shared.cpp',
-        'jit/x86-shared/MoveEmitter-x86-shared.cpp',
-    ]
-    SOURCES += [
-        'jit/x86-shared/Disassembler-x86-shared.cpp',  # using namespace js::jit::X86Encoding;
-    ]
-    if CONFIG['JS_CODEGEN_X64']:
-        UNIFIED_SOURCES += [
-            'jit/x64/Assembler-x64.cpp',
-            'jit/x64/Bailouts-x64.cpp',
-            'jit/x64/BaselineCompiler-x64.cpp',
-            'jit/x64/BaselineIC-x64.cpp',
-            'jit/x64/CodeGenerator-x64.cpp',
-            'jit/x64/Lowering-x64.cpp',
-            'jit/x64/MacroAssembler-x64.cpp',
-            'jit/x64/SharedIC-x64.cpp',
-            'jit/x64/Trampoline-x64.cpp',
-        ]
-    else:
-        UNIFIED_SOURCES += [
-            'jit/x86/Assembler-x86.cpp',
-            'jit/x86/Bailouts-x86.cpp',
-            'jit/x86/BaselineCompiler-x86.cpp',
-            'jit/x86/BaselineIC-x86.cpp',
-            'jit/x86/CodeGenerator-x86.cpp',
-            'jit/x86/Lowering-x86.cpp',
-            'jit/x86/MacroAssembler-x86.cpp',
-            'jit/x86/SharedIC-x86.cpp',
-            'jit/x86/Trampoline-x86.cpp',
-        ]
-elif CONFIG['JS_CODEGEN_ARM']:
-    UNIFIED_SOURCES += [
-        'jit/arm/Architecture-arm.cpp',
-        'jit/arm/Assembler-arm.cpp',
-        'jit/arm/Bailouts-arm.cpp',
-        'jit/arm/BaselineCompiler-arm.cpp',
-        'jit/arm/BaselineIC-arm.cpp',
-        'jit/arm/CodeGenerator-arm.cpp',
-        'jit/arm/disasm/Constants-arm.cpp',
-        'jit/arm/disasm/Disasm-arm.cpp',
-        'jit/arm/Lowering-arm.cpp',
-        'jit/arm/MacroAssembler-arm.cpp',
-        'jit/arm/MoveEmitter-arm.cpp',
-        'jit/arm/SharedIC-arm.cpp',
-        'jit/arm/Trampoline-arm.cpp',
-    ]
-    if CONFIG['JS_SIMULATOR_ARM']:
-        UNIFIED_SOURCES += [
-            'jit/arm/Simulator-arm.cpp'
-        ]
-        # Configuration used only for testing.
-        if CONFIG['OS_ARCH'] == 'Linux' or CONFIG['OS_ARCH'] == 'Darwin':
-            CXXFLAGS += [ '-msse2', '-mfpmath=sse' ]
-    elif CONFIG['OS_ARCH'] == 'Darwin':
-        SOURCES += [
-            'jit/arm/llvm-compiler-rt/arm/aeabi_idivmod.S',
-            'jit/arm/llvm-compiler-rt/arm/aeabi_uidivmod.S',
-        ]
-elif CONFIG['JS_CODEGEN_ARM64']:
-    UNIFIED_SOURCES += [
-        'jit/arm64/Architecture-arm64.cpp',
-        'jit/arm64/Assembler-arm64.cpp',
-        'jit/arm64/Bailouts-arm64.cpp',
-        'jit/arm64/BaselineIC-arm64.cpp',
-        'jit/arm64/CodeGenerator-arm64.cpp',
-        'jit/arm64/Lowering-arm64.cpp',
-        'jit/arm64/MacroAssembler-arm64.cpp',
-        'jit/arm64/MoveEmitter-arm64.cpp',
-        'jit/arm64/SharedIC-arm64.cpp',
-        'jit/arm64/Trampoline-arm64.cpp',
-        'jit/arm64/vixl/Assembler-vixl.cpp',
-        'jit/arm64/vixl/Cpu-vixl.cpp',
-        'jit/arm64/vixl/Decoder-vixl.cpp',
-        'jit/arm64/vixl/Disasm-vixl.cpp',
-        'jit/arm64/vixl/Instructions-vixl.cpp',
-        'jit/arm64/vixl/Instrument-vixl.cpp',
-        'jit/arm64/vixl/MacroAssembler-vixl.cpp',
-        'jit/arm64/vixl/MozAssembler-vixl.cpp',
-        'jit/arm64/vixl/MozInstructions-vixl.cpp',
-        'jit/arm64/vixl/Utils-vixl.cpp'
-    ]
-    if CONFIG['JS_SIMULATOR_ARM64']:
-        UNIFIED_SOURCES += [
-            'jit/arm64/vixl/Debugger-vixl.cpp',
-            'jit/arm64/vixl/Logic-vixl.cpp',
-            'jit/arm64/vixl/MozSimulator-vixl.cpp',
-            'jit/arm64/vixl/Simulator-vixl.cpp'
-        ]
-elif CONFIG['JS_CODEGEN_MIPS32'] or CONFIG['JS_CODEGEN_MIPS64']:
-    UNIFIED_SOURCES += [
-        'jit/mips-shared/Architecture-mips-shared.cpp',
-        'jit/mips-shared/Assembler-mips-shared.cpp',
-        'jit/mips-shared/Bailouts-mips-shared.cpp',
-        'jit/mips-shared/BaselineCompiler-mips-shared.cpp',
-        'jit/mips-shared/BaselineIC-mips-shared.cpp',
-        'jit/mips-shared/CodeGenerator-mips-shared.cpp',
-        'jit/mips-shared/Lowering-mips-shared.cpp',
-        'jit/mips-shared/MacroAssembler-mips-shared.cpp',
-        'jit/mips-shared/MoveEmitter-mips-shared.cpp',
-    ]
-    if CONFIG['JS_CODEGEN_MIPS32']:
-        UNIFIED_SOURCES += [
-            'jit/mips32/Architecture-mips32.cpp',
-            'jit/mips32/Assembler-mips32.cpp',
-            'jit/mips32/Bailouts-mips32.cpp',
-            'jit/mips32/BaselineCompiler-mips32.cpp',
-            'jit/mips32/BaselineIC-mips32.cpp',
-            'jit/mips32/CodeGenerator-mips32.cpp',
-            'jit/mips32/Lowering-mips32.cpp',
-            'jit/mips32/MacroAssembler-mips32.cpp',
-            'jit/mips32/MoveEmitter-mips32.cpp',
-            'jit/mips32/SharedIC-mips32.cpp',
-            'jit/mips32/Trampoline-mips32.cpp',
-        ]
-        if CONFIG['JS_SIMULATOR_MIPS32']:
-            UNIFIED_SOURCES += [
-                'jit/mips32/Simulator-mips32.cpp'
-            ]
-    elif CONFIG['JS_CODEGEN_MIPS64']:
-        UNIFIED_SOURCES += [
-            'jit/mips64/Architecture-mips64.cpp',
-            'jit/mips64/Assembler-mips64.cpp',
-            'jit/mips64/Bailouts-mips64.cpp',
-            'jit/mips64/BaselineCompiler-mips64.cpp',
-            'jit/mips64/BaselineIC-mips64.cpp',
-            'jit/mips64/CodeGenerator-mips64.cpp',
-            'jit/mips64/Lowering-mips64.cpp',
-            'jit/mips64/MacroAssembler-mips64.cpp',
-            'jit/mips64/MoveEmitter-mips64.cpp',
-            'jit/mips64/SharedIC-mips64.cpp',
-            'jit/mips64/Trampoline-mips64.cpp',
-        ]
-        if CONFIG['JS_SIMULATOR_MIPS64']:
-            UNIFIED_SOURCES += [
-                'jit/mips64/Simulator-mips64.cpp'
-            ]
-
-if CONFIG['OS_ARCH'] == 'WINNT':
-    SOURCES += [
-        'threading/windows/ConditionVariable.cpp',
-        'threading/windows/MutexImpl.cpp',
-        'threading/windows/Thread.cpp',
-    ]
-    # _CRT_RAND_S must be #defined before #including stdlib.h to get rand_s()
-    DEFINES['_CRT_RAND_S'] = True
-else:
-    SOURCES += [
-        'threading/posix/ConditionVariable.cpp',
-        'threading/posix/MutexImpl.cpp',
-        'threading/posix/Thread.cpp',
-    ]
-
-if CONFIG['JS_HAS_CTYPES']:
-    SOURCES += [
-        'ctypes/CTypes.cpp',
-        'ctypes/Library.cpp',
-    ]
-    if not CONFIG['MOZ_SYSTEM_FFI']:
-        LOCAL_INCLUDES += [
-            '!ctypes/libffi/include',
-            'ctypes/libffi/src/%s' % CONFIG['FFI_TARGET_DIR'],
-        ]
-
-if CONFIG['MOZ_VTUNE']:
-    SOURCES += [
-        'vtune/jitprofiling.c'
-    ]
-
-if CONFIG['HAVE_LINUX_PERF_EVENT_H']:
-    SOURCES += [
-        'perf/pm_linux.cpp'
-    ]
-    if CONFIG['LINUX_HEADERS_INCLUDES']:
-        SOURCES['perf/pm_linux.cpp'].flags += [CONFIG['LINUX_HEADERS_INCLUDES']]
-else:
-    SOURCES += [
-        'perf/pm_stub.cpp'
-    ]
-
-GENERATED_FILES += ['frontend/ReservedWordsGenerated.h']
-ReservedWordsGenerated = GENERATED_FILES['frontend/ReservedWordsGenerated.h']
-ReservedWordsGenerated.script = 'frontend/GenerateReservedWords.py'
-ReservedWordsGenerated.inputs += [
-    'frontend/ReservedWords.h'
-]
-
 # JavaScript must be built shared, even for static builds, as it is used by
 # other modules which are always built shared. Failure to do so results in
 # the js code getting copied into xpinstall and jsd as well as mozilla-bin,
@@ -661,65 +283,14 @@ USE_LIBS += [
     'zlib',
 ]
 
-if CONFIG['NIGHTLY_BUILD']:
-    DEFINES['ENABLE_BINARYDATA'] = True
-
-if CONFIG['NIGHTLY_BUILD']:
-    DEFINES['ENABLE_SIMD'] = True
-
-# Also set in shell/moz.build
-DEFINES['ENABLE_SHARED_ARRAY_BUFFER'] = True
-
-DEFINES['EXPORT_JS_API'] = True
-
-if CONFIG['JS_HAS_CTYPES']:
-    DEFINES['JS_HAS_CTYPES'] = True
-    for var in ('DLL_PREFIX', 'DLL_SUFFIX'):
-        DEFINES[var] = '"%s"' % CONFIG[var]
-
-if CONFIG['MOZ_LINKER']:
-    DEFINES['MOZ_LINKER'] = True
-
-if CONFIG['_MSC_VER']:
-    if CONFIG['CPU_ARCH'] == 'x86':
-        SOURCES['builtin/RegExp.cpp'].no_pgo = True # Bug 772303
-    elif CONFIG['CPU_ARCH'] == 'x86_64' and CONFIG['JS_HAS_CTYPES']:
-        SOURCES['ctypes/CTypes.cpp'].no_pgo = True # Bug 810661
-    # Prevent floating point errors caused by VC++ optimizations
-    CFLAGS += ['-fp:precise']
-    CXXFLAGS += ['-fp:precise']    
-    # C4805 warns mixing bool with other integral types in computation.
-    # But given the conversion from bool is specified, and this is a
-    # pattern widely used in code in js/src, suppress this warning here.
-    CXXFLAGS += ['-wd4805']
-    # C4661 ("no suitable definition provided for explicit template
-    # instantiation request") is emitted for all Parser methods that
-    # have a Parser<FullParseHandler> definition but no
-    # Parser<SyntaxParseHandler> definition, see bug 1167030.
-    CXXFLAGS += ['-wd4661']
-    CXXFLAGS += ['-we4067', '-we4258', '-we4275']
-    CXXFLAGS += ['-wd4146'] # FIXME: unary minus operator applied to unsigned type (bug 1229189)
-
-    # This is intended as a temporary hack to support building with VS2015.
-    # 'noexcept' used with no exception handling mode specified;
-    # termination on exception is not guaranteed. Specify /EHsc
-    CXXFLAGS += ['-wd4577']
-    CXXFLAGS += ['-wd4312']
-
 if CONFIG['OS_ARCH'] not in ('WINNT'):
-    OS_LIBS += [
-        'm',
-    ]
+    OS_LIBS += ['m']
 
 if CONFIG['OS_ARCH'] == 'FreeBSD':
-    OS_LIBS += [
-        '-pthread',
-    ]
+    OS_LIBS += ['-pthread']
 
 if CONFIG['OS_ARCH'] == 'Linux':
-    OS_LIBS += [
-        'dl',
-    ]
+    OS_LIBS += ['dl']
 
 if CONFIG['OS_ARCH'] == 'SunOS':
     OS_LIBS += [
@@ -731,65 +302,8 @@ if CONFIG['OS_ARCH'] == 'SunOS':
 
 OS_LIBS += CONFIG['REALTIME_LIBS']
 
-CFLAGS += CONFIG['MOZ_ICU_CFLAGS']
-CXXFLAGS += CONFIG['MOZ_ICU_CFLAGS']
-
 NO_EXPAND_LIBS = True
 
 DIST_INSTALL = True
 
-# Prepare self-hosted JS code for embedding
-GENERATED_FILES += [('selfhosted.out.h', 'selfhosted.js')]
-selfhosted = GENERATED_FILES[('selfhosted.out.h', 'selfhosted.js')]
-selfhosted.script = 'builtin/embedjs.py:generate_selfhosted'
-selfhosted.inputs = [
-    'js.msg',
-    'builtin/TypedObjectConstants.h',
-    'builtin/SelfHostingDefines.h',
-    'builtin/Utilities.js',
-    'builtin/Array.js',
-    'builtin/AsyncIteration.js',
-    'builtin/Classes.js',
-    'builtin/Date.js',
-    'builtin/Error.js',
-    'builtin/Function.js',
-    'builtin/Generator.js',
-    'builtin/Intl.js',
-    'builtin/IntlData.js',
-    'builtin/Iterator.js',
-    'builtin/Map.js',
-    'builtin/Module.js',
-    'builtin/Number.js',
-    'builtin/Object.js',
-    'builtin/Promise.js',
-    'builtin/Reflect.js',
-    'builtin/RegExp.js',
-    'builtin/RegExpGlobalReplaceOpt.h.js',
-    'builtin/RegExpLocalReplaceOpt.h.js',
-    'builtin/String.js',
-    'builtin/Set.js',
-    'builtin/Sorting.js',
-    'builtin/TypedArray.js',
-    'builtin/TypedObject.js',
-    'builtin/WeakMap.js',
-    'builtin/WeakSet.js'
-]
 
-if CONFIG['JS_HAS_CTYPES']:
-    if CONFIG['MOZ_SYSTEM_FFI']:
-        CXXFLAGS += CONFIG['MOZ_FFI_CFLAGS']
-    else:
-        # Windows needs this to be linked with a static library.
-        DEFINES['FFI_BUILDING'] = True
-
-if CONFIG['GNU_CXX']:
-    # Disable strict-aliasing for GCC, which is enabled by default
-    # starting with version 7.1, see Mozilla bug 1363009.
-    CXXFLAGS += ['-Wno-shadow', '-Werror=format', '-fno-strict-aliasing']
-
-# Suppress warnings in third-party code.
-if CONFIG['CLANG_CXX'] or CONFIG['GNU_CXX']:
-    SOURCES['jsdtoa.cpp'].flags += ['-Wno-implicit-fallthrough']
-
-if CONFIG['OS_ARCH'] == 'WINNT':
-    DEFINES['NOMINMAX'] = True

js/src/perf/moz.build

@@ -0,0 +1,29 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+include('../js-config.mozbuild')
+include('../js-cxxflags.mozbuild')
+
+FINAL_LIBRARY = "js"
+
+# Includes should be relative to parent path
+LOCAL_INCLUDES += ["!..", ".."]
+
+EXPORTS += ['jsperf.h']
+
+SOURCES += [
+    'jsperf.cpp',
+]
+
+if CONFIG['HAVE_LINUX_PERF_EVENT_H']:
+    SOURCES += [
+        'pm_linux.cpp'
+    ]
+    if CONFIG['LINUX_HEADERS_INCLUDES']:
+        SOURCES['pm_linux.cpp'].flags += [CONFIG['LINUX_HEADERS_INCLUDES']]
+else:
+    SOURCES += [
+        'pm_stub.cpp'
+    ]

js/src/proxy/moz.build

@@ -0,0 +1,23 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+include('../js-config.mozbuild')
+include('../js-cxxflags.mozbuild')
+
+FINAL_LIBRARY = "js"
+
+# Includes should be relative to parent path
+LOCAL_INCLUDES += ["!..", ".."]
+
+SOURCES += [
+    'BaseProxyHandler.cpp',
+    'CrossCompartmentWrapper.cpp',
+    'DeadObjectProxy.cpp',
+    'OpaqueCrossCompartmentWrapper.cpp',
+    'Proxy.cpp',
+    'ScriptedProxyHandler.cpp',
+    'SecurityWrapper.cpp',
+    'Wrapper.cpp',
+]

js/src/regexp/RegExpTypes.h

@@ -0,0 +1,51 @@
+/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*-
+ * vim: set ts=8 sts=2 et sw=2 tw=80:
+ * This Source Code Form is subject to the terms of the Mozilla Public
+ * License, v. 2.0. If a copy of the MPL was not distributed with this
+ * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
+
+// This file forward-defines Irregexp classes that need to be visible
+// to the rest of Spidermonkey and re-exports them into js::irregexp.
+
+#ifndef regexp_RegExpTypes_h
+#define regexp_RegExpTypes_h
+
+namespace js {
+class MatchPairs;
+}
+
+namespace v8 {
+namespace internal {
+
+struct InputOutputData {
+  const void* inputStart;
+  const void* inputEnd;
+
+  // Index into inputStart (in chars) at which to begin matching.
+  size_t startIndex;
+
+  js::MatchPairs* matches;
+
+  template <typename CharT>
+  InputOutputData(const CharT* inputStart, const CharT* inputEnd,
+                  size_t startIndex, js::MatchPairs* matches)
+    : inputStart(inputStart),
+      inputEnd(inputEnd),
+      startIndex(startIndex),
+      matches(matches)
+  {}
+};
+
+} // namespace internal
+} // namespace v8
+
+
+namespace js {
+namespace irregexp {
+
+using InputOutputData = v8::internal::InputOutputData;
+
+} // namespace irregexp
+} // namespace js
+
+#endif // regexp_RegExpTypes_h

js/src/regexp/VERSION

@@ -0,0 +1,2 @@
+Imported using import-irregexp.py from:
+https://github.com/v8/v8/tree/560f2d8bb3f3a72d78e1a7d7654235d53fdcc83c/src/regexp

js/src/regexp/gen-regexp-special-case.cc

@@ -0,0 +1,165 @@
+// Copyright 2020 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include <fstream>
+#include <iomanip>
+#include <iostream>
+#include <sstream>
+
+#include "regexp/special-case.h"
+
+namespace v8 {
+namespace internal {
+
+static const uc32 kSurrogateStart = 0xd800;
+static const uc32 kSurrogateEnd = 0xdfff;
+static const uc32 kNonBmpStart = 0x10000;
+
+// The following code generates "src/regexp/special-case.cc".
+void PrintSet(std::ofstream& out, const char* name,
+              const icu::UnicodeSet& set) {
+  out << "icu::UnicodeSet Build" << name << "() {\n"
+      << "  icu::UnicodeSet set;\n";
+  for (int32_t i = 0; i < set.getRangeCount(); i++) {
+    if (set.getRangeStart(i) == set.getRangeEnd(i)) {
+      out << "  set.add(0x" << set.getRangeStart(i) << ");\n";
+    } else {
+      out << "  set.add(0x" << set.getRangeStart(i) << ", 0x"
+          << set.getRangeEnd(i) << ");\n";
+    }
+  }
+  out << "  set.freeze();\n"
+      << "  return set;\n"
+      << "}\n\n";
+
+  out << "struct " << name << "Data {\n"
+      << "  " << name << "Data() : set(Build" << name << "()) {}\n"
+      << "  const icu::UnicodeSet set;\n"
+      << "};\n\n";
+
+  out << "//static\n"
+      << "const icu::UnicodeSet& RegExpCaseFolding::" << name << "() {\n"
+      << "  static base::LazyInstance<" << name << "Data>::type set =\n"
+      << "      LAZY_INSTANCE_INITIALIZER;\n"
+      << "  return set.Pointer()->set;\n"
+      << "}\n\n";
+}
+
+void PrintSpecial(std::ofstream& out) {
+  icu::UnicodeSet current;
+  icu::UnicodeSet special_add;
+  icu::UnicodeSet ignore;
+  UErrorCode status = U_ZERO_ERROR;
+  icu::UnicodeSet upper("[\\p{Lu}]", status);
+  CHECK(U_SUCCESS(status));
+
+  // Iterate through all chars in BMP except surrogates.
+  for (UChar32 i = 0; i < kNonBmpStart; i++) {
+    if (i >= kSurrogateStart && i <= kSurrogateEnd) {
+      continue;  // Ignore surrogate range
+    }
+    current.set(i, i);
+    current.closeOver(USET_CASE_INSENSITIVE);
+
+    // Check to see if all characters in the case-folding equivalence
+    // class as defined by UnicodeSet::closeOver all map to the same
+    // canonical value.
+    UChar32 canonical = RegExpCaseFolding::Canonicalize(i);
+    bool class_has_matching_canonical_char = false;
+    bool class_has_non_matching_canonical_char = false;
+    for (int32_t j = 0; j < current.getRangeCount(); j++) {
+      for (UChar32 c = current.getRangeStart(j); c <= current.getRangeEnd(j);
+           c++) {
+        if (c == i) {
+          continue;
+        }
+        UChar32 other_canonical = RegExpCaseFolding::Canonicalize(c);
+        if (canonical == other_canonical) {
+          class_has_matching_canonical_char = true;
+        } else {
+          class_has_non_matching_canonical_char = true;
+        }
+      }
+    }
+    // If any other character in i's equivalence class has a
+    // different canonical value, then i needs special handling.  If
+    // no other character shares a canonical value with i, we can
+    // ignore i when adding alternatives for case-independent
+    // comparison.  If at least one other character shares a
+    // canonical value, then i needs special handling.
+    if (class_has_non_matching_canonical_char) {
+      if (class_has_matching_canonical_char) {
+        special_add.add(i);
+      } else {
+        ignore.add(i);
+      }
+    }
+  }
+
+  // Verify that no Unicode equivalence class contains two non-trivial
+  // JS equivalence classes. Every character in SpecialAddSet has the
+  // same canonical value as every other non-IgnoreSet character in
+  // its Unicode equivalence class. Therefore, if we call closeOver on
+  // a set containing no IgnoreSet characters, the only characters
+  // that must be removed from the result are in IgnoreSet. This fact
+  // is used in CharacterRange::AddCaseEquivalents.
+  for (int32_t i = 0; i < special_add.getRangeCount(); i++) {
+    for (UChar32 c = special_add.getRangeStart(i);
+         c <= special_add.getRangeEnd(i); c++) {
+      UChar32 canonical = RegExpCaseFolding::Canonicalize(c);
+      current.set(c, c);
+      current.closeOver(USET_CASE_INSENSITIVE);
+      current.removeAll(ignore);
+      for (int32_t j = 0; j < current.getRangeCount(); j++) {
+        for (UChar32 c2 = current.getRangeStart(j);
+             c2 <= current.getRangeEnd(j); c2++) {
+          CHECK_EQ(canonical, RegExpCaseFolding::Canonicalize(c2));
+        }
+      }
+    }
+  }
+
+  PrintSet(out, "IgnoreSet", ignore);
+  PrintSet(out, "SpecialAddSet", special_add);
+}
+
+void WriteHeader(const char* header_filename) {
+  std::ofstream out(header_filename);
+  out << std::hex << std::setfill('0') << std::setw(4);
+  out << "// Copyright 2020 the V8 project authors. All rights reserved.\n"
+      << "// Use of this source code is governed by a BSD-style license that\n"
+      << "// can be found in the LICENSE file.\n\n"
+      << "// Automatically generated by regexp/gen-regexp-special-case.cc\n\n"
+      << "// The following functions are used to build UnicodeSets\n"
+      << "// for special cases where the case-folding algorithm used by\n"
+      << "// UnicodeSet::closeOver(USET_CASE_INSENSITIVE) does not match\n"
+      << "// the algorithm defined in ECMAScript 2020 21.2.2.8.2 (Runtime\n"
+      << "// Semantics: Canonicalize) step 3.\n\n"
+      << "#ifdef V8_INTL_SUPPORT\n"
+      << "#include \"src/base/lazy-instance.h\"\n\n"
+      << "#include \"src/regexp/special-case.h\"\n\n"
+      << "#include \"unicode/uniset.h\"\n"
+      << "namespace v8 {\n"
+      << "namespace internal {\n\n";
+
+  PrintSpecial(out);
+
+  out << "\n"
+      << "}  // namespace internal\n"
+      << "}  // namespace v8\n"
+      << "#endif  // V8_INTL_SUPPORT\n";
+}
+
+}  // namespace internal
+}  // namespace v8
+
+int main(int argc, const char** argv) {
+  if (argc != 2) {
+    std::cerr << "Usage: " << argv[0] << " <output filename>\n";
+    std::exit(1);
+  }
+  v8::internal::WriteHeader(argv[1]);
+
+  return 0;
+}

js/src/regexp/import-irregexp.py

@@ -0,0 +1,143 @@
+#!/usr/bin/env python3
+
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this file,
+# You can obtain one at http://mozilla.org/MPL/2.0/.
+
+# This script handles all the mechanical steps of importing irregexp from v8:
+#
+# 1. Acquire the source: either from github, or optionally from a local copy of v8.
+# 2. Copy the contents of v8/src/regexp into js/src/regexp
+#    - Exclude files that we have chosen not to import.
+# 3. While doing so, update #includes:
+#    - Change "src/regexp/*" to "regexp/*".
+#    - Remove other v8-specific headers completely.
+# 4. Add '#include "regexp/regexp-shim.h" in the necessary places.
+# 5. Update the VERSION file to include the correct git hash.
+#
+# Usage:
+#  cd path/to/js/src/regexp
+#  ./import-irregexp.py --path path/to/v8/src/regexp
+#
+# Alternatively, without the --path argument, import-irregexp.py will
+# clone v8 from github into a temporary directory.
+#
+# After running this script, changes to the shim code may be necessary
+# to account for changes in upstream irregexp.
+
+import os
+import re
+import subprocess
+import sys
+from pathlib import Path
+
+
+def get_hash(path):
+    # Get the hash for the current git revision
+    cwd = os.getcwd()
+    os.chdir(path)
+    command = ['git', 'rev-parse', 'HEAD']
+    result = subprocess.check_output(command, encoding='utf-8')
+    os.chdir(cwd)
+    return result.rstrip()
+
+
+def copy_and_update_includes(src_path, dst_path):
+    # List of header files that need to include the shim header
+    need_shim = ['property-sequences.h',
+                 'regexp-ast.h',
+                 'regexp-bytecode-peephole.h',
+                 'regexp-bytecodes.h',
+                 'regexp-dotprinter.h',
+                 'regexp.h',
+                 'regexp-macro-assembler.h',
+                 'regexp-stack.h',
+                 'special-case.h']
+
+    src = open(str(src_path), 'r')
+    dst = open(str(dst_path), 'w')
+
+    # 1. Rewrite includes of V8 regexp headers:
+    regexp_include = re.compile('#include "src/regexp')
+    regexp_include_new = '#include "regexp'
+
+    # 2. Remove includes of other V8 headers
+    other_include = re.compile('#include "src/')
+
+    # 3. If needed, add '#include "regexp/regexp-shim.h"'.
+    #    Note: We get a little fancy to ensure that header files are
+    #    in alphabetic order. `need_to_add_shim` is true if we still
+    #    have to add the shim header in this file. `adding_shim_now`
+    #    is true if we have found a '#include "src/*' and we are just
+    #    waiting to find something alphabetically smaller (or an empty
+    #    line) so that we can insert the shim header in the right place.
+    need_to_add_shim = src_path.name in need_shim
+    adding_shim_now = False
+
+    for line in src:
+        if adding_shim_now:
+            if (line == '\n' or line > '#include "src/regexp/regexp-shim.h"'):
+                dst.write('#include "regexp/regexp-shim.h"\n')
+                need_to_add_shim = False
+                adding_shim_now = False
+
+        if regexp_include.search(line):
+            dst.write(re.sub(regexp_include, regexp_include_new, line))
+        elif other_include.search(line):
+            if need_to_add_shim:
+                adding_shim_now = True
+        else:
+            dst.write(line)
+
+
+def import_from(srcdir, dstdir):
+    excluded = ['OWNERS',
+                'regexp.cc',
+                'regexp-utils.cc',
+                'regexp-utils.h',
+                'regexp-macro-assembler-arch.h']
+
+    for file in srcdir.iterdir():
+        if file.is_dir():
+            continue
+        if str(file.name) in excluded:
+            continue
+        copy_and_update_includes(file, dstdir / file.name)
+
+    # Update VERSION file
+    hash = get_hash(srcdir)
+    version_file = open(str(dstdir / 'VERSION'), 'w')
+    version_file.write('Imported using import-irregexp.py from:\n')
+    version_file.write('https://github.com/v8/v8/tree/%s/src/regexp\n' % hash)
+
+
+if __name__ == '__main__':
+    import argparse
+    import tempfile
+
+    # This script should be run from js/src/regexp to work correctly.
+    current_path = Path(os.getcwd())
+    expected_path = 'js/src/regexp'
+    if not current_path.match(expected_path):
+        raise RuntimeError('%s must be run from %s' % (sys.argv[0],
+                                                       expected_path))
+
+    parser = argparse.ArgumentParser(description='Import irregexp from v8')
+    parser.add_argument('-p', '--path', help='path to v8/src/regexp')
+    args = parser.parse_args()
+
+    if args.path:
+        src_path = Path(args.path)
+
+        if not (src_path / 'regexp.h').exists():
+            print('Usage:\n  import-irregexp.py --path <path/to/v8/src/regexp>')
+            sys.exit(1)
+        import_from(src_path, current_path)
+        sys.exit(0)
+
+    with tempfile.TemporaryDirectory() as tempdir:
+        v8_git = 'https://github.com/v8/v8.git'
+        clone = 'git clone --depth 1 %s %s' % (v8_git, tempdir)
+        os.system(clone)
+        src_path = Path(tempdir) / 'src/regexp'
+        import_from(src_path, current_path)

js/src/regexp/moz.build

@@ -0,0 +1,37 @@
+# -*- Mode: python; indent-tabs-mode: nil; tab-width: 40 -*-
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+include('../js-config.mozbuild')
+include('../js-cxxflags.mozbuild')
+
+FINAL_LIBRARY = "js"
+
+# Includes should be relative to parent path
+LOCAL_INCLUDES += ["!..", ".."]
+
+SOURCES += [
+    'regexp-ast.cc',
+    'regexp-bytecode-generator.cc',
+    'regexp-bytecode-peephole.cc',
+    'regexp-bytecodes.cc',
+    'regexp-compiler-tonode.cc',
+    'regexp-compiler.cc',
+    'regexp-dotprinter.cc',
+    'regexp-interpreter.cc',
+    'regexp-macro-assembler-tracer.cc',
+    'regexp-macro-assembler.cc',
+    'regexp-native-macro-assembler.cc',
+    'regexp-parser.cc',
+    'regexp-shim.cc',
+    'regexp-stack.cc',
+    'util/unicode.cc'
+]
+
+if CONFIG['ENABLE_INTL_API']:
+    CXXFLAGS += ['-DV8_INTL_SUPPORT']
+    SOURCES += [
+        'property-sequences.cc',
+        'special-case.cc'
+    ]

js/src/regexp/property-sequences.h

@@ -0,0 +1,27 @@
+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef V8_REGEXP_PROPERTY_SEQUENCES_H_
+#define V8_REGEXP_PROPERTY_SEQUENCES_H_
+
+#ifdef V8_INTL_SUPPORT
+
+#include "regexp/regexp-shim.h"
+
+namespace v8 {
+namespace internal {
+
+class UnicodePropertySequences : public AllStatic {
+ public:
+  static const uc32 kEmojiFlagSequences[];
+  static const uc32 kEmojiTagSequences[];
+  static const uc32 kEmojiZWJSequences[];
+};
+
+}  // namespace internal
+}  // namespace v8
+
+#endif  // V8_INTL_SUPPORT
+
+#endif  // V8_REGEXP_PROPERTY_SEQUENCES_H_

js/src/regexp/regexp-ast.cc

@@ -0,0 +1,342 @@
+// Copyright 2016 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "regexp/regexp-ast.h"
+
+namespace v8 {
+namespace internal {
+
+#define MAKE_ACCEPT(Name)                                          \
+  void* RegExp##Name::Accept(RegExpVisitor* visitor, void* data) { \
+    return visitor->Visit##Name(this, data);                       \
+  }
+FOR_EACH_REG_EXP_TREE_TYPE(MAKE_ACCEPT)
+#undef MAKE_ACCEPT
+
+#define MAKE_TYPE_CASE(Name)                               \
+  RegExp##Name* RegExpTree::As##Name() { return nullptr; } \
+  bool RegExpTree::Is##Name() { return false; }
+FOR_EACH_REG_EXP_TREE_TYPE(MAKE_TYPE_CASE)
+#undef MAKE_TYPE_CASE
+
+#define MAKE_TYPE_CASE(Name)                              \
+  RegExp##Name* RegExp##Name::As##Name() { return this; } \
+  bool RegExp##Name::Is##Name() { return true; }
+FOR_EACH_REG_EXP_TREE_TYPE(MAKE_TYPE_CASE)
+#undef MAKE_TYPE_CASE
+
+
+static Interval ListCaptureRegisters(ZoneList<RegExpTree*>* children) {
+  Interval result = Interval::Empty();
+  for (int i = 0; i < children->length(); i++)
+    result = result.Union(children->at(i)->CaptureRegisters());
+  return result;
+}
+
+
+Interval RegExpAlternative::CaptureRegisters() {
+  return ListCaptureRegisters(nodes());
+}
+
+
+Interval RegExpDisjunction::CaptureRegisters() {
+  return ListCaptureRegisters(alternatives());
+}
+
+
+Interval RegExpLookaround::CaptureRegisters() {
+  return body()->CaptureRegisters();
+}
+
+
+Interval RegExpCapture::CaptureRegisters() {
+  Interval self(StartRegister(index()), EndRegister(index()));
+  return self.Union(body()->CaptureRegisters());
+}
+
+
+Interval RegExpQuantifier::CaptureRegisters() {
+  return body()->CaptureRegisters();
+}
+
+
+bool RegExpAssertion::IsAnchoredAtStart() {
+  return assertion_type() == RegExpAssertion::START_OF_INPUT;
+}
+
+
+bool RegExpAssertion::IsAnchoredAtEnd() {
+  return assertion_type() == RegExpAssertion::END_OF_INPUT;
+}
+
+
+bool RegExpAlternative::IsAnchoredAtStart() {
+  ZoneList<RegExpTree*>* nodes = this->nodes();
+  for (int i = 0; i < nodes->length(); i++) {
+    RegExpTree* node = nodes->at(i);
+    if (node->IsAnchoredAtStart()) {
+      return true;
+    }
+    if (node->max_match() > 0) {
+      return false;
+    }
+  }
+  return false;
+}
+
+
+bool RegExpAlternative::IsAnchoredAtEnd() {
+  ZoneList<RegExpTree*>* nodes = this->nodes();
+  for (int i = nodes->length() - 1; i >= 0; i--) {
+    RegExpTree* node = nodes->at(i);
+    if (node->IsAnchoredAtEnd()) {
+      return true;
+    }
+    if (node->max_match() > 0) {
+      return false;
+    }
+  }
+  return false;
+}
+
+
+bool RegExpDisjunction::IsAnchoredAtStart() {
+  ZoneList<RegExpTree*>* alternatives = this->alternatives();
+  for (int i = 0; i < alternatives->length(); i++) {
+    if (!alternatives->at(i)->IsAnchoredAtStart()) return false;
+  }
+  return true;
+}
+
+
+bool RegExpDisjunction::IsAnchoredAtEnd() {
+  ZoneList<RegExpTree*>* alternatives = this->alternatives();
+  for (int i = 0; i < alternatives->length(); i++) {
+    if (!alternatives->at(i)->IsAnchoredAtEnd()) return false;
+  }
+  return true;
+}
+
+
+bool RegExpLookaround::IsAnchoredAtStart() {
+  return is_positive() && type() == LOOKAHEAD && body()->IsAnchoredAtStart();
+}
+
+
+bool RegExpCapture::IsAnchoredAtStart() { return body()->IsAnchoredAtStart(); }
+
+
+bool RegExpCapture::IsAnchoredAtEnd() { return body()->IsAnchoredAtEnd(); }
+
+
+// Convert regular expression trees to a simple sexp representation.
+// This representation should be different from the input grammar
+// in as many cases as possible, to make it more difficult for incorrect
+// parses to look as correct ones which is likely if the input and
+// output formats are alike.
+class RegExpUnparser final : public RegExpVisitor {
+ public:
+  RegExpUnparser(std::ostream& os, Zone* zone) : os_(os), zone_(zone) {}
+  void VisitCharacterRange(CharacterRange that);
+#define MAKE_CASE(Name) void* Visit##Name(RegExp##Name*, void* data) override;
+  FOR_EACH_REG_EXP_TREE_TYPE(MAKE_CASE)
+#undef MAKE_CASE
+ private:
+  std::ostream& os_;
+  Zone* zone_;
+};
+
+
+void* RegExpUnparser::VisitDisjunction(RegExpDisjunction* that, void* data) {
+  os_ << "(|";
+  for (int i = 0; i < that->alternatives()->length(); i++) {
+    os_ << " ";
+    that->alternatives()->at(i)->Accept(this, data);
+  }
+  os_ << ")";
+  return nullptr;
+}
+
+
+void* RegExpUnparser::VisitAlternative(RegExpAlternative* that, void* data) {
+  os_ << "(:";
+  for (int i = 0; i < that->nodes()->length(); i++) {
+    os_ << " ";
+    that->nodes()->at(i)->Accept(this, data);
+  }
+  os_ << ")";
+  return nullptr;
+}
+
+
+void RegExpUnparser::VisitCharacterRange(CharacterRange that) {
+  os_ << AsUC32(that.from());
+  if (!that.IsSingleton()) {
+    os_ << "-" << AsUC32(that.to());
+  }
+}
+
+
+void* RegExpUnparser::VisitCharacterClass(RegExpCharacterClass* that,
+                                          void* data) {
+  if (that->is_negated()) os_ << "^";
+  os_ << "[";
+  for (int i = 0; i < that->ranges(zone_)->length(); i++) {
+    if (i > 0) os_ << " ";
+    VisitCharacterRange(that->ranges(zone_)->at(i));
+  }
+  os_ << "]";
+  return nullptr;
+}
+
+
+void* RegExpUnparser::VisitAssertion(RegExpAssertion* that, void* data) {
+  switch (that->assertion_type()) {
+    case RegExpAssertion::START_OF_INPUT:
+      os_ << "@^i";
+      break;
+    case RegExpAssertion::END_OF_INPUT:
+      os_ << "@$i";
+      break;
+    case RegExpAssertion::START_OF_LINE:
+      os_ << "@^l";
+      break;
+    case RegExpAssertion::END_OF_LINE:
+      os_ << "@$l";
+      break;
+    case RegExpAssertion::BOUNDARY:
+      os_ << "@b";
+      break;
+    case RegExpAssertion::NON_BOUNDARY:
+      os_ << "@B";
+      break;
+  }
+  return nullptr;
+}
+
+
+void* RegExpUnparser::VisitAtom(RegExpAtom* that, void* data) {
+  os_ << "'";
+  Vector<const uc16> chardata = that->data();
+  for (int i = 0; i < chardata.length(); i++) {
+    os_ << AsUC16(chardata[i]);
+  }
+  os_ << "'";
+  return nullptr;
+}
+
+
+void* RegExpUnparser::VisitText(RegExpText* that, void* data) {
+  if (that->elements()->length() == 1) {
+    that->elements()->at(0).tree()->Accept(this, data);
+  } else {
+    os_ << "(!";
+    for (int i = 0; i < that->elements()->length(); i++) {
+      os_ << " ";
+      that->elements()->at(i).tree()->Accept(this, data);
+    }
+    os_ << ")";
+  }
+  return nullptr;
+}
+
+
+void* RegExpUnparser::VisitQuantifier(RegExpQuantifier* that, void* data) {
+  os_ << "(# " << that->min() << " ";
+  if (that->max() == RegExpTree::kInfinity) {
+    os_ << "- ";
+  } else {
+    os_ << that->max() << " ";
+  }
+  os_ << (that->is_greedy() ? "g " : that->is_possessive() ? "p " : "n ");
+  that->body()->Accept(this, data);
+  os_ << ")";
+  return nullptr;
+}
+
+
+void* RegExpUnparser::VisitCapture(RegExpCapture* that, void* data) {
+  os_ << "(^ ";
+  that->body()->Accept(this, data);
+  os_ << ")";
+  return nullptr;
+}
+
+void* RegExpUnparser::VisitGroup(RegExpGroup* that, void* data) {
+  os_ << "(?: ";
+  that->body()->Accept(this, data);
+  os_ << ")";
+  return nullptr;
+}
+
+void* RegExpUnparser::VisitLookaround(RegExpLookaround* that, void* data) {
+  os_ << "(";
+  os_ << (that->type() == RegExpLookaround::LOOKAHEAD ? "->" : "<-");
+  os_ << (that->is_positive() ? " + " : " - ");
+  that->body()->Accept(this, data);
+  os_ << ")";
+  return nullptr;
+}
+
+
+void* RegExpUnparser::VisitBackReference(RegExpBackReference* that,
+                                         void* data) {
+  os_ << "(<- " << that->index() << ")";
+  return nullptr;
+}
+
+
+void* RegExpUnparser::VisitEmpty(RegExpEmpty* that, void* data) {
+  os_ << '%';
+  return nullptr;
+}
+
+
+std::ostream& RegExpTree::Print(std::ostream& os, Zone* zone) {  // NOLINT
+  RegExpUnparser unparser(os, zone);
+  Accept(&unparser, nullptr);
+  return os;
+}
+
+
+RegExpDisjunction::RegExpDisjunction(ZoneList<RegExpTree*>* alternatives)
+    : alternatives_(alternatives) {
+  DCHECK_LT(1, alternatives->length());
+  RegExpTree* first_alternative = alternatives->at(0);
+  min_match_ = first_alternative->min_match();
+  max_match_ = first_alternative->max_match();
+  for (int i = 1; i < alternatives->length(); i++) {
+    RegExpTree* alternative = alternatives->at(i);
+    min_match_ = Min(min_match_, alternative->min_match());
+    max_match_ = Max(max_match_, alternative->max_match());
+  }
+}
+
+
+static int IncreaseBy(int previous, int increase) {
+  if (RegExpTree::kInfinity - previous < increase) {
+    return RegExpTree::kInfinity;
+  } else {
+    return previous + increase;
+  }
+}
+
+
+RegExpAlternative::RegExpAlternative(ZoneList<RegExpTree*>* nodes)
+    : nodes_(nodes) {
+  DCHECK_LT(1, nodes->length());
+  min_match_ = 0;
+  max_match_ = 0;
+  for (int i = 0; i < nodes->length(); i++) {
+    RegExpTree* node = nodes->at(i);
+    int node_min_match = node->min_match();
+    min_match_ = IncreaseBy(min_match_, node_min_match);
+    int node_max_match = node->max_match();
+    max_match_ = IncreaseBy(max_match_, node_max_match);
+  }
+}
+
+
+}  // namespace internal
+}  // namespace v8

js/src/regexp/regexp-ast.h

@@ -0,0 +1,615 @@
+// Copyright 2016 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef V8_REGEXP_REGEXP_AST_H_
+#define V8_REGEXP_REGEXP_AST_H_
+
+#include "regexp/regexp-shim.h"
+
+namespace v8 {
+namespace internal {
+
+#define FOR_EACH_REG_EXP_TREE_TYPE(VISIT) \
+  VISIT(Disjunction)                      \
+  VISIT(Alternative)                      \
+  VISIT(Assertion)                        \
+  VISIT(CharacterClass)                   \
+  VISIT(Atom)                             \
+  VISIT(Quantifier)                       \
+  VISIT(Capture)                          \
+  VISIT(Group)                            \
+  VISIT(Lookaround)                       \
+  VISIT(BackReference)                    \
+  VISIT(Empty)                            \
+  VISIT(Text)
+
+#define FORWARD_DECLARE(Name) class RegExp##Name;
+FOR_EACH_REG_EXP_TREE_TYPE(FORWARD_DECLARE)
+#undef FORWARD_DECLARE
+
+class RegExpCompiler;
+class RegExpNode;
+class RegExpTree;
+
+class RegExpVisitor {
+ public:
+  virtual ~RegExpVisitor() = default;
+#define MAKE_CASE(Name) \
+  virtual void* Visit##Name(RegExp##Name*, void* data) = 0;
+  FOR_EACH_REG_EXP_TREE_TYPE(MAKE_CASE)
+#undef MAKE_CASE
+};
+
+
+// A simple closed interval.
+class Interval {
+ public:
+  Interval() : from_(kNone), to_(kNone - 1) {}  // '- 1' for branchless size().
+  Interval(int from, int to) : from_(from), to_(to) {}
+  Interval Union(Interval that) {
+    if (that.from_ == kNone)
+      return *this;
+    else if (from_ == kNone)
+      return that;
+    else
+      return Interval(Min(from_, that.from_), Max(to_, that.to_));
+  }
+
+  bool Contains(int value) { return (from_ <= value) && (value <= to_); }
+  bool is_empty() { return from_ == kNone; }
+  int from() const { return from_; }
+  int to() const { return to_; }
+  int size() const { return to_ - from_ + 1; }
+
+  static Interval Empty() { return Interval(); }
+
+  static constexpr int kNone = -1;
+
+ private:
+  int from_;
+  int to_;
+};
+
+
+// Represents code units in the range from from_ to to_, both ends are
+// inclusive.
+class CharacterRange {
+ public:
+  CharacterRange() : from_(0), to_(0) {}
+  // For compatibility with the CHECK_OK macro
+  CharacterRange(void* null) { DCHECK_NULL(null); }  // NOLINT
+  V8_EXPORT_PRIVATE static void AddClassEscape(char type,
+                                               ZoneList<CharacterRange>* ranges,
+                                               Zone* zone);
+  // Add class escapes. Add case equivalent closure for \w and \W if necessary.
+  V8_EXPORT_PRIVATE static void AddClassEscape(
+      char type, ZoneList<CharacterRange>* ranges,
+      bool add_unicode_case_equivalents, Zone* zone);
+  static Vector<const int> GetWordBounds();
+  static inline CharacterRange Singleton(uc32 value) {
+    return CharacterRange(value, value);
+  }
+  static inline CharacterRange Range(uc32 from, uc32 to) {
+    DCHECK(0 <= from && to <= String::kMaxCodePoint);
+    DCHECK(static_cast<uint32_t>(from) <= static_cast<uint32_t>(to));
+    return CharacterRange(from, to);
+  }
+  static inline CharacterRange Everything() {
+    return CharacterRange(0, String::kMaxCodePoint);
+  }
+  static inline ZoneList<CharacterRange>* List(Zone* zone,
+                                               CharacterRange range) {
+    ZoneList<CharacterRange>* list =
+        new (zone) ZoneList<CharacterRange>(1, zone);
+    list->Add(range, zone);
+    return list;
+  }
+  bool Contains(uc32 i) { return from_ <= i && i <= to_; }
+  uc32 from() const { return from_; }
+  void set_from(uc32 value) { from_ = value; }
+  uc32 to() const { return to_; }
+  void set_to(uc32 value) { to_ = value; }
+  bool is_valid() { return from_ <= to_; }
+  bool IsEverything(uc32 max) { return from_ == 0 && to_ >= max; }
+  bool IsSingleton() { return (from_ == to_); }
+  V8_EXPORT_PRIVATE static void AddCaseEquivalents(
+      Isolate* isolate, Zone* zone, ZoneList<CharacterRange>* ranges,
+      bool is_one_byte);
+  // Whether a range list is in canonical form: Ranges ordered by from value,
+  // and ranges non-overlapping and non-adjacent.
+  V8_EXPORT_PRIVATE static bool IsCanonical(ZoneList<CharacterRange>* ranges);
+  // Convert range list to canonical form. The characters covered by the ranges
+  // will still be the same, but no character is in more than one range, and
+  // adjacent ranges are merged. The resulting list may be shorter than the
+  // original, but cannot be longer.
+  static void Canonicalize(ZoneList<CharacterRange>* ranges);
+  // Negate the contents of a character range in canonical form.
+  static void Negate(ZoneList<CharacterRange>* src,
+                     ZoneList<CharacterRange>* dst, Zone* zone);
+  static const int kStartMarker = (1 << 24);
+  static const int kPayloadMask = (1 << 24) - 1;
+
+ private:
+  CharacterRange(uc32 from, uc32 to) : from_(from), to_(to) {}
+
+  uc32 from_;
+  uc32 to_;
+};
+
+class CharacterSet final {
+ public:
+  explicit CharacterSet(uc16 standard_set_type)
+      : ranges_(nullptr), standard_set_type_(standard_set_type) {}
+  explicit CharacterSet(ZoneList<CharacterRange>* ranges)
+      : ranges_(ranges), standard_set_type_(0) {}
+  ZoneList<CharacterRange>* ranges(Zone* zone);
+  uc16 standard_set_type() const { return standard_set_type_; }
+  void set_standard_set_type(uc16 special_set_type) {
+    standard_set_type_ = special_set_type;
+  }
+  bool is_standard() { return standard_set_type_ != 0; }
+  V8_EXPORT_PRIVATE void Canonicalize();
+
+ private:
+  ZoneList<CharacterRange>* ranges_;
+  // If non-zero, the value represents a standard set (e.g., all whitespace
+  // characters) without having to expand the ranges.
+  uc16 standard_set_type_;
+};
+
+class TextElement final {
+ public:
+  enum TextType { ATOM, CHAR_CLASS };
+
+  static TextElement Atom(RegExpAtom* atom);
+  static TextElement CharClass(RegExpCharacterClass* char_class);
+
+  int cp_offset() const { return cp_offset_; }
+  void set_cp_offset(int cp_offset) { cp_offset_ = cp_offset; }
+  int length() const;
+
+  TextType text_type() const { return text_type_; }
+
+  RegExpTree* tree() const { return tree_; }
+
+  RegExpAtom* atom() const {
+    DCHECK(text_type() == ATOM);
+    return reinterpret_cast<RegExpAtom*>(tree());
+  }
+
+  RegExpCharacterClass* char_class() const {
+    DCHECK(text_type() == CHAR_CLASS);
+    return reinterpret_cast<RegExpCharacterClass*>(tree());
+  }
+
+ private:
+  TextElement(TextType text_type, RegExpTree* tree)
+      : cp_offset_(-1), text_type_(text_type), tree_(tree) {}
+
+  int cp_offset_;
+  TextType text_type_;
+  RegExpTree* tree_;
+};
+
+
+class RegExpTree : public ZoneObject {
+ public:
+  static const int kInfinity = kMaxInt;
+  virtual ~RegExpTree() = default;
+  virtual void* Accept(RegExpVisitor* visitor, void* data) = 0;
+  virtual RegExpNode* ToNode(RegExpCompiler* compiler,
+                             RegExpNode* on_success) = 0;
+  virtual bool IsTextElement() { return false; }
+  virtual bool IsAnchoredAtStart() { return false; }
+  virtual bool IsAnchoredAtEnd() { return false; }
+  virtual int min_match() = 0;
+  virtual int max_match() = 0;
+  // Returns the interval of registers used for captures within this
+  // expression.
+  virtual Interval CaptureRegisters() { return Interval::Empty(); }
+  virtual void AppendToText(RegExpText* text, Zone* zone);
+  V8_EXPORT_PRIVATE std::ostream& Print(std::ostream& os,
+                                        Zone* zone);  // NOLINT
+#define MAKE_ASTYPE(Name)           \
+  virtual RegExp##Name* As##Name(); \
+  virtual bool Is##Name();
+  FOR_EACH_REG_EXP_TREE_TYPE(MAKE_ASTYPE)
+#undef MAKE_ASTYPE
+};
+
+
+class RegExpDisjunction final : public RegExpTree {
+ public:
+  explicit RegExpDisjunction(ZoneList<RegExpTree*>* alternatives);
+  void* Accept(RegExpVisitor* visitor, void* data) override;
+  RegExpNode* ToNode(RegExpCompiler* compiler, RegExpNode* on_success) override;
+  RegExpDisjunction* AsDisjunction() override;
+  Interval CaptureRegisters() override;
+  bool IsDisjunction() override;
+  bool IsAnchoredAtStart() override;
+  bool IsAnchoredAtEnd() override;
+  int min_match() override { return min_match_; }
+  int max_match() override { return max_match_; }
+  ZoneList<RegExpTree*>* alternatives() { return alternatives_; }
+
+ private:
+  bool SortConsecutiveAtoms(RegExpCompiler* compiler);
+  void RationalizeConsecutiveAtoms(RegExpCompiler* compiler);
+  void FixSingleCharacterDisjunctions(RegExpCompiler* compiler);
+  ZoneList<RegExpTree*>* alternatives_;
+  int min_match_;
+  int max_match_;
+};
+
+
+class RegExpAlternative final : public RegExpTree {
+ public:
+  explicit RegExpAlternative(ZoneList<RegExpTree*>* nodes);
+  void* Accept(RegExpVisitor* visitor, void* data) override;
+  RegExpNode* ToNode(RegExpCompiler* compiler, RegExpNode* on_success) override;
+  RegExpAlternative* AsAlternative() override;
+  Interval CaptureRegisters() override;
+  bool IsAlternative() override;
+  bool IsAnchoredAtStart() override;
+  bool IsAnchoredAtEnd() override;
+  int min_match() override { return min_match_; }
+  int max_match() override { return max_match_; }
+  ZoneList<RegExpTree*>* nodes() { return nodes_; }
+
+ private:
+  ZoneList<RegExpTree*>* nodes_;
+  int min_match_;
+  int max_match_;
+};
+
+
+class RegExpAssertion final : public RegExpTree {
+ public:
+  enum AssertionType {
+    START_OF_LINE = 0,
+    START_OF_INPUT = 1,
+    END_OF_LINE = 2,
+    END_OF_INPUT = 3,
+    BOUNDARY = 4,
+    NON_BOUNDARY = 5,
+    LAST_TYPE = NON_BOUNDARY,
+  };
+  RegExpAssertion(AssertionType type, JSRegExp::Flags flags)
+      : assertion_type_(type), flags_(flags) {}
+  void* Accept(RegExpVisitor* visitor, void* data) override;
+  RegExpNode* ToNode(RegExpCompiler* compiler, RegExpNode* on_success) override;
+  RegExpAssertion* AsAssertion() override;
+  bool IsAssertion() override;
+  bool IsAnchoredAtStart() override;
+  bool IsAnchoredAtEnd() override;
+  int min_match() override { return 0; }
+  int max_match() override { return 0; }
+  AssertionType assertion_type() const { return assertion_type_; }
+  JSRegExp::Flags flags() const { return flags_; }
+
+ private:
+  const AssertionType assertion_type_;
+  const JSRegExp::Flags flags_;
+};
+
+
+class RegExpCharacterClass final : public RegExpTree {
+ public:
+  // NEGATED: The character class is negated and should match everything but
+  //     the specified ranges.
+  // CONTAINS_SPLIT_SURROGATE: The character class contains part of a split
+  //     surrogate and should not be unicode-desugared (crbug.com/641091).
+  enum Flag {
+    NEGATED = 1 << 0,
+    CONTAINS_SPLIT_SURROGATE = 1 << 1,
+  };
+  using CharacterClassFlags = base::Flags<Flag>;
+
+  RegExpCharacterClass(
+      Zone* zone, ZoneList<CharacterRange>* ranges, JSRegExp::Flags flags,
+      CharacterClassFlags character_class_flags = CharacterClassFlags())
+      : set_(ranges),
+        flags_(flags),
+        character_class_flags_(character_class_flags) {
+    // Convert the empty set of ranges to the negated Everything() range.
+    if (ranges->is_empty()) {
+      ranges->Add(CharacterRange::Everything(), zone);
+      character_class_flags_ ^= NEGATED;
+    }
+  }
+  RegExpCharacterClass(uc16 type, JSRegExp::Flags flags)
+      : set_(type),
+        flags_(flags),
+        character_class_flags_(CharacterClassFlags()) {}
+  void* Accept(RegExpVisitor* visitor, void* data) override;
+  RegExpNode* ToNode(RegExpCompiler* compiler, RegExpNode* on_success) override;
+  RegExpCharacterClass* AsCharacterClass() override;
+  bool IsCharacterClass() override;
+  bool IsTextElement() override { return true; }
+  int min_match() override { return 1; }
+  // The character class may match two code units for unicode regexps.
+  // TODO(yangguo): we should split this class for usage in TextElement, and
+  //                make max_match() dependent on the character class content.
+  int max_match() override { return 2; }
+  void AppendToText(RegExpText* text, Zone* zone) override;
+  CharacterSet character_set() { return set_; }
+  // TODO(lrn): Remove need for complex version if is_standard that
+  // recognizes a mangled standard set and just do { return set_.is_special(); }
+  bool is_standard(Zone* zone);
+  // Returns a value representing the standard character set if is_standard()
+  // returns true.
+  // Currently used values are:
+  // s : unicode whitespace
+  // S : unicode non-whitespace
+  // w : ASCII word character (digit, letter, underscore)
+  // W : non-ASCII word character
+  // d : ASCII digit
+  // D : non-ASCII digit
+  // . : non-newline
+  // * : All characters, for advancing unanchored regexp
+  uc16 standard_type() const { return set_.standard_set_type(); }
+  ZoneList<CharacterRange>* ranges(Zone* zone) { return set_.ranges(zone); }
+  bool is_negated() const { return (character_class_flags_ & NEGATED) != 0; }
+  JSRegExp::Flags flags() const { return flags_; }
+  bool contains_split_surrogate() const {
+    return (character_class_flags_ & CONTAINS_SPLIT_SURROGATE) != 0;
+  }
+
+ private:
+  CharacterSet set_;
+  const JSRegExp::Flags flags_;
+  CharacterClassFlags character_class_flags_;
+};
+
+
+class RegExpAtom final : public RegExpTree {
+ public:
+  explicit RegExpAtom(Vector<const uc16> data, JSRegExp::Flags flags)
+      : data_(data), flags_(flags) {}
+  void* Accept(RegExpVisitor* visitor, void* data) override;
+  RegExpNode* ToNode(RegExpCompiler* compiler, RegExpNode* on_success) override;
+  RegExpAtom* AsAtom() override;
+  bool IsAtom() override;
+  bool IsTextElement() override { return true; }
+  int min_match() override { return data_.length(); }
+  int max_match() override { return data_.length(); }
+  void AppendToText(RegExpText* text, Zone* zone) override;
+  Vector<const uc16> data() { return data_; }
+  int length() { return data_.length(); }
+  JSRegExp::Flags flags() const { return flags_; }
+  bool ignore_case() const { return (flags_ & JSRegExp::kIgnoreCase) != 0; }
+
+ private:
+  Vector<const uc16> data_;
+  const JSRegExp::Flags flags_;
+};
+
+
+class RegExpText final : public RegExpTree {
+ public:
+  explicit RegExpText(Zone* zone) : elements_(2, zone), length_(0) {}
+  void* Accept(RegExpVisitor* visitor, void* data) override;
+  RegExpNode* ToNode(RegExpCompiler* compiler, RegExpNode* on_success) override;
+  RegExpText* AsText() override;
+  bool IsText() override;
+  bool IsTextElement() override { return true; }
+  int min_match() override { return length_; }
+  int max_match() override { return length_; }
+  void AppendToText(RegExpText* text, Zone* zone) override;
+  void AddElement(TextElement elm, Zone* zone) {
+    elements_.Add(elm, zone);
+    length_ += elm.length();
+  }
+  ZoneList<TextElement>* elements() { return &elements_; }
+
+ private:
+  ZoneList<TextElement> elements_;
+  int length_;
+};
+
+
+class RegExpQuantifier final : public RegExpTree {
+ public:
+  enum QuantifierType { GREEDY, NON_GREEDY, POSSESSIVE };
+  RegExpQuantifier(int min, int max, QuantifierType type, RegExpTree* body)
+      : body_(body),
+        min_(min),
+        max_(max),
+        quantifier_type_(type) {
+    if (min > 0 && body->min_match() > kInfinity / min) {
+      min_match_ = kInfinity;
+    } else {
+      min_match_ = min * body->min_match();
+    }
+    if (max > 0 && body->max_match() > kInfinity / max) {
+      max_match_ = kInfinity;
+    } else {
+      max_match_ = max * body->max_match();
+    }
+  }
+  void* Accept(RegExpVisitor* visitor, void* data) override;
+  RegExpNode* ToNode(RegExpCompiler* compiler, RegExpNode* on_success) override;
+  static RegExpNode* ToNode(int min, int max, bool is_greedy, RegExpTree* body,
+                            RegExpCompiler* compiler, RegExpNode* on_success,
+                            bool not_at_start = false);
+  RegExpQuantifier* AsQuantifier() override;
+  Interval CaptureRegisters() override;
+  bool IsQuantifier() override;
+  int min_match() override { return min_match_; }
+  int max_match() override { return max_match_; }
+  int min() { return min_; }
+  int max() { return max_; }
+  bool is_possessive() { return quantifier_type_ == POSSESSIVE; }
+  bool is_non_greedy() { return quantifier_type_ == NON_GREEDY; }
+  bool is_greedy() { return quantifier_type_ == GREEDY; }
+  RegExpTree* body() { return body_; }
+
+ private:
+  RegExpTree* body_;
+  int min_;
+  int max_;
+  int min_match_;
+  int max_match_;
+  QuantifierType quantifier_type_;
+};
+
+
+class RegExpCapture final : public RegExpTree {
+ public:
+  explicit RegExpCapture(int index)
+      : body_(nullptr),
+        index_(index),
+        min_match_(0),
+        max_match_(0),
+        name_(nullptr) {}
+  void* Accept(RegExpVisitor* visitor, void* data) override;
+  RegExpNode* ToNode(RegExpCompiler* compiler, RegExpNode* on_success) override;
+  static RegExpNode* ToNode(RegExpTree* body, int index,
+                            RegExpCompiler* compiler, RegExpNode* on_success);
+  RegExpCapture* AsCapture() override;
+  bool IsAnchoredAtStart() override;
+  bool IsAnchoredAtEnd() override;
+  Interval CaptureRegisters() override;
+  bool IsCapture() override;
+  int min_match() override { return min_match_; }
+  int max_match() override { return max_match_; }
+  RegExpTree* body() { return body_; }
+  void set_body(RegExpTree* body) {
+    body_ = body;
+    min_match_ = body->min_match();
+    max_match_ = body->max_match();
+  }
+  int index() const { return index_; }
+  const ZoneVector<uc16>* name() const { return name_; }
+  void set_name(const ZoneVector<uc16>* name) { name_ = name; }
+  static int StartRegister(int index) { return index * 2; }
+  static int EndRegister(int index) { return index * 2 + 1; }
+
+ private:
+  RegExpTree* body_;
+  int index_;
+  int min_match_;
+  int max_match_;
+  const ZoneVector<uc16>* name_;
+};
+
+class RegExpGroup final : public RegExpTree {
+ public:
+  explicit RegExpGroup(RegExpTree* body)
+      : body_(body),
+        min_match_(body->min_match()),
+        max_match_(body->max_match()) {}
+  void* Accept(RegExpVisitor* visitor, void* data) override;
+  RegExpNode* ToNode(RegExpCompiler* compiler,
+                     RegExpNode* on_success) override {
+    return body_->ToNode(compiler, on_success);
+  }
+  RegExpGroup* AsGroup() override;
+  bool IsAnchoredAtStart() override { return body_->IsAnchoredAtStart(); }
+  bool IsAnchoredAtEnd() override { return body_->IsAnchoredAtEnd(); }
+  bool IsGroup() override;
+  int min_match() override { return min_match_; }
+  int max_match() override { return max_match_; }
+  Interval CaptureRegisters() override { return body_->CaptureRegisters(); }
+  RegExpTree* body() { return body_; }
+
+ private:
+  RegExpTree* body_;
+  int min_match_;
+  int max_match_;
+};
+
+class RegExpLookaround final : public RegExpTree {
+ public:
+  enum Type { LOOKAHEAD, LOOKBEHIND };
+
+  RegExpLookaround(RegExpTree* body, bool is_positive, int capture_count,
+                   int capture_from, Type type)
+      : body_(body),
+        is_positive_(is_positive),
+        capture_count_(capture_count),
+        capture_from_(capture_from),
+        type_(type) {}
+
+  void* Accept(RegExpVisitor* visitor, void* data) override;
+  RegExpNode* ToNode(RegExpCompiler* compiler, RegExpNode* on_success) override;
+  RegExpLookaround* AsLookaround() override;
+  Interval CaptureRegisters() override;
+  bool IsLookaround() override;
+  bool IsAnchoredAtStart() override;
+  int min_match() override { return 0; }
+  int max_match() override { return 0; }
+  RegExpTree* body() { return body_; }
+  bool is_positive() { return is_positive_; }
+  int capture_count() { return capture_count_; }
+  int capture_from() { return capture_from_; }
+  Type type() { return type_; }
+
+  class Builder {
+   public:
+    Builder(bool is_positive, RegExpNode* on_success,
+            int stack_pointer_register, int position_register,
+            int capture_register_count = 0, int capture_register_start = 0);
+    RegExpNode* on_match_success() { return on_match_success_; }
+    RegExpNode* ForMatch(RegExpNode* match);
+
+   private:
+    bool is_positive_;
+    RegExpNode* on_match_success_;
+    RegExpNode* on_success_;
+    int stack_pointer_register_;
+    int position_register_;
+  };
+
+ private:
+  RegExpTree* body_;
+  bool is_positive_;
+  int capture_count_;
+  int capture_from_;
+  Type type_;
+};
+
+
+class RegExpBackReference final : public RegExpTree {
+ public:
+  explicit RegExpBackReference(JSRegExp::Flags flags)
+      : capture_(nullptr), name_(nullptr), flags_(flags) {}
+  RegExpBackReference(RegExpCapture* capture, JSRegExp::Flags flags)
+      : capture_(capture), name_(nullptr), flags_(flags) {}
+  void* Accept(RegExpVisitor* visitor, void* data) override;
+  RegExpNode* ToNode(RegExpCompiler* compiler, RegExpNode* on_success) override;
+  RegExpBackReference* AsBackReference() override;
+  bool IsBackReference() override;
+  int min_match() override { return 0; }
+  // The back reference may be recursive, e.g. /(\2)(\1)/. To avoid infinite
+  // recursion, we give up. Ignorance is bliss.
+  int max_match() override { return kInfinity; }
+  int index() { return capture_->index(); }
+  RegExpCapture* capture() { return capture_; }
+  void set_capture(RegExpCapture* capture) { capture_ = capture; }
+  const ZoneVector<uc16>* name() const { return name_; }
+  void set_name(const ZoneVector<uc16>* name) { name_ = name; }
+
+ private:
+  RegExpCapture* capture_;
+  const ZoneVector<uc16>* name_;
+  const JSRegExp::Flags flags_;
+};
+
+
+class RegExpEmpty final : public RegExpTree {
+ public:
+  RegExpEmpty() = default;
+  void* Accept(RegExpVisitor* visitor, void* data) override;
+  RegExpNode* ToNode(RegExpCompiler* compiler, RegExpNode* on_success) override;
+  RegExpEmpty* AsEmpty() override;
+  bool IsEmpty() override;
+  int min_match() override { return 0; }
+  int max_match() override { return 0; }
+};
+
+}  // namespace internal
+}  // namespace v8
+
+#endif  // V8_REGEXP_REGEXP_AST_H_

js/src/regexp/regexp-bytecode-generator-inl.h

@@ -0,0 +1,55 @@
+// Copyright 2008-2009 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef V8_REGEXP_REGEXP_BYTECODE_GENERATOR_INL_H_
+#define V8_REGEXP_REGEXP_BYTECODE_GENERATOR_INL_H_
+
+#include "regexp/regexp-bytecode-generator.h"
+
+#include "regexp/regexp-bytecodes.h"
+
+namespace v8 {
+namespace internal {
+
+void RegExpBytecodeGenerator::Emit(uint32_t byte, uint32_t twenty_four_bits) {
+  uint32_t word = ((twenty_four_bits << BYTECODE_SHIFT) | byte);
+  DCHECK(pc_ <= buffer_.length());
+  if (pc_ + 3 >= buffer_.length()) {
+    Expand();
+  }
+  *reinterpret_cast<uint32_t*>(buffer_.begin() + pc_) = word;
+  pc_ += 4;
+}
+
+void RegExpBytecodeGenerator::Emit16(uint32_t word) {
+  DCHECK(pc_ <= buffer_.length());
+  if (pc_ + 1 >= buffer_.length()) {
+    Expand();
+  }
+  *reinterpret_cast<uint16_t*>(buffer_.begin() + pc_) = word;
+  pc_ += 2;
+}
+
+void RegExpBytecodeGenerator::Emit8(uint32_t word) {
+  DCHECK(pc_ <= buffer_.length());
+  if (pc_ == buffer_.length()) {
+    Expand();
+  }
+  *reinterpret_cast<unsigned char*>(buffer_.begin() + pc_) = word;
+  pc_ += 1;
+}
+
+void RegExpBytecodeGenerator::Emit32(uint32_t word) {
+  DCHECK(pc_ <= buffer_.length());
+  if (pc_ + 3 >= buffer_.length()) {
+    Expand();
+  }
+  *reinterpret_cast<uint32_t*>(buffer_.begin() + pc_) = word;
+  pc_ += 4;
+}
+
+}  // namespace internal
+}  // namespace v8
+
+#endif  // V8_REGEXP_REGEXP_BYTECODE_GENERATOR_INL_H_

js/src/regexp/regexp-bytecode-generator.cc

@@ -0,0 +1,395 @@
+// Copyright 2008-2009 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "regexp/regexp-bytecode-generator.h"
+
+#include "regexp/regexp-bytecode-generator-inl.h"
+#include "regexp/regexp-bytecode-peephole.h"
+#include "regexp/regexp-bytecodes.h"
+#include "regexp/regexp-macro-assembler.h"
+
+namespace v8 {
+namespace internal {
+
+RegExpBytecodeGenerator::RegExpBytecodeGenerator(Isolate* isolate, Zone* zone)
+    : RegExpMacroAssembler(isolate, zone),
+      buffer_(Vector<byte>::New(1024)),
+      pc_(0),
+      advance_current_end_(kInvalidPC),
+      jump_edges_(zone),
+      isolate_(isolate) {}
+
+RegExpBytecodeGenerator::~RegExpBytecodeGenerator() {
+  if (backtrack_.is_linked()) backtrack_.Unuse();
+  buffer_.Dispose();
+}
+
+RegExpBytecodeGenerator::IrregexpImplementation
+RegExpBytecodeGenerator::Implementation() {
+  return kBytecodeImplementation;
+}
+
+void RegExpBytecodeGenerator::Bind(Label* l) {
+  advance_current_end_ = kInvalidPC;
+  DCHECK(!l->is_bound());
+  if (l->is_linked()) {
+    int pos = l->pos();
+    while (pos != 0) {
+      int fixup = pos;
+      pos = *reinterpret_cast<int32_t*>(buffer_.begin() + fixup);
+      *reinterpret_cast<uint32_t*>(buffer_.begin() + fixup) = pc_;
+      jump_edges_.emplace(fixup, pc_);
+    }
+  }
+  l->bind_to(pc_);
+}
+
+void RegExpBytecodeGenerator::EmitOrLink(Label* l) {
+  if (l == nullptr) l = &backtrack_;
+  int pos = 0;
+  if (l->is_bound()) {
+    pos = l->pos();
+    jump_edges_.emplace(pc_, pos);
+  } else {
+    if (l->is_linked()) {
+      pos = l->pos();
+    }
+    l->link_to(pc_);
+  }
+  Emit32(pos);
+}
+
+void RegExpBytecodeGenerator::PopRegister(int register_index) {
+  DCHECK_LE(0, register_index);
+  DCHECK_GE(kMaxRegister, register_index);
+  Emit(BC_POP_REGISTER, register_index);
+}
+
+void RegExpBytecodeGenerator::PushRegister(int register_index,
+                                           StackCheckFlag check_stack_limit) {
+  DCHECK_LE(0, register_index);
+  DCHECK_GE(kMaxRegister, register_index);
+  Emit(BC_PUSH_REGISTER, register_index);
+}
+
+void RegExpBytecodeGenerator::WriteCurrentPositionToRegister(int register_index,
+                                                             int cp_offset) {
+  DCHECK_LE(0, register_index);
+  DCHECK_GE(kMaxRegister, register_index);
+  Emit(BC_SET_REGISTER_TO_CP, register_index);
+  Emit32(cp_offset);  // Current position offset.
+}
+
+void RegExpBytecodeGenerator::ClearRegisters(int reg_from, int reg_to) {
+  DCHECK(reg_from <= reg_to);
+  for (int reg = reg_from; reg <= reg_to; reg++) {
+    SetRegister(reg, -1);
+  }
+}
+
+void RegExpBytecodeGenerator::ReadCurrentPositionFromRegister(
+    int register_index) {
+  DCHECK_LE(0, register_index);
+  DCHECK_GE(kMaxRegister, register_index);
+  Emit(BC_SET_CP_TO_REGISTER, register_index);
+}
+
+void RegExpBytecodeGenerator::WriteStackPointerToRegister(int register_index) {
+  DCHECK_LE(0, register_index);
+  DCHECK_GE(kMaxRegister, register_index);
+  Emit(BC_SET_REGISTER_TO_SP, register_index);
+}
+
+void RegExpBytecodeGenerator::ReadStackPointerFromRegister(int register_index) {
+  DCHECK_LE(0, register_index);
+  DCHECK_GE(kMaxRegister, register_index);
+  Emit(BC_SET_SP_TO_REGISTER, register_index);
+}
+
+void RegExpBytecodeGenerator::SetCurrentPositionFromEnd(int by) {
+  DCHECK(is_uint24(by));
+  Emit(BC_SET_CURRENT_POSITION_FROM_END, by);
+}
+
+void RegExpBytecodeGenerator::SetRegister(int register_index, int to) {
+  DCHECK_LE(0, register_index);
+  DCHECK_GE(kMaxRegister, register_index);
+  Emit(BC_SET_REGISTER, register_index);
+  Emit32(to);
+}
+
+void RegExpBytecodeGenerator::AdvanceRegister(int register_index, int by) {
+  DCHECK_LE(0, register_index);
+  DCHECK_GE(kMaxRegister, register_index);
+  Emit(BC_ADVANCE_REGISTER, register_index);
+  Emit32(by);
+}
+
+void RegExpBytecodeGenerator::PopCurrentPosition() { Emit(BC_POP_CP, 0); }
+
+void RegExpBytecodeGenerator::PushCurrentPosition() { Emit(BC_PUSH_CP, 0); }
+
+void RegExpBytecodeGenerator::Backtrack() { Emit(BC_POP_BT, 0); }
+
+void RegExpBytecodeGenerator::GoTo(Label* l) {
+  if (advance_current_end_ == pc_) {
+    // Combine advance current and goto.
+    pc_ = advance_current_start_;
+    Emit(BC_ADVANCE_CP_AND_GOTO, advance_current_offset_);
+    EmitOrLink(l);
+    advance_current_end_ = kInvalidPC;
+  } else {
+    // Regular goto.
+    Emit(BC_GOTO, 0);
+    EmitOrLink(l);
+  }
+}
+
+void RegExpBytecodeGenerator::PushBacktrack(Label* l) {
+  Emit(BC_PUSH_BT, 0);
+  EmitOrLink(l);
+}
+
+bool RegExpBytecodeGenerator::Succeed() {
+  Emit(BC_SUCCEED, 0);
+  return false;  // Restart matching for global regexp not supported.
+}
+
+void RegExpBytecodeGenerator::Fail() { Emit(BC_FAIL, 0); }
+
+void RegExpBytecodeGenerator::AdvanceCurrentPosition(int by) {
+  DCHECK_LE(kMinCPOffset, by);
+  DCHECK_GE(kMaxCPOffset, by);
+  advance_current_start_ = pc_;
+  advance_current_offset_ = by;
+  Emit(BC_ADVANCE_CP, by);
+  advance_current_end_ = pc_;
+}
+
+void RegExpBytecodeGenerator::CheckGreedyLoop(
+    Label* on_tos_equals_current_position) {
+  Emit(BC_CHECK_GREEDY, 0);
+  EmitOrLink(on_tos_equals_current_position);
+}
+
+void RegExpBytecodeGenerator::LoadCurrentCharacterImpl(int cp_offset,
+                                                       Label* on_failure,
+                                                       bool check_bounds,
+                                                       int characters,
+                                                       int eats_at_least) {
+  DCHECK_GE(eats_at_least, characters);
+  if (eats_at_least > characters && check_bounds) {
+    DCHECK(is_uint24(cp_offset + eats_at_least));
+    Emit(BC_CHECK_CURRENT_POSITION, cp_offset + eats_at_least);
+    EmitOrLink(on_failure);
+    check_bounds = false;  // Load below doesn't need to check.
+  }
+
+  DCHECK_LE(kMinCPOffset, cp_offset);
+  DCHECK_GE(kMaxCPOffset, cp_offset);
+  int bytecode;
+  if (check_bounds) {
+    if (characters == 4) {
+      bytecode = BC_LOAD_4_CURRENT_CHARS;
+    } else if (characters == 2) {
+      bytecode = BC_LOAD_2_CURRENT_CHARS;
+    } else {
+      DCHECK_EQ(1, characters);
+      bytecode = BC_LOAD_CURRENT_CHAR;
+    }
+  } else {
+    if (characters == 4) {
+      bytecode = BC_LOAD_4_CURRENT_CHARS_UNCHECKED;
+    } else if (characters == 2) {
+      bytecode = BC_LOAD_2_CURRENT_CHARS_UNCHECKED;
+    } else {
+      DCHECK_EQ(1, characters);
+      bytecode = BC_LOAD_CURRENT_CHAR_UNCHECKED;
+    }
+  }
+  Emit(bytecode, cp_offset);
+  if (check_bounds) EmitOrLink(on_failure);
+}
+
+void RegExpBytecodeGenerator::CheckCharacterLT(uc16 limit, Label* on_less) {
+  Emit(BC_CHECK_LT, limit);
+  EmitOrLink(on_less);
+}
+
+void RegExpBytecodeGenerator::CheckCharacterGT(uc16 limit, Label* on_greater) {
+  Emit(BC_CHECK_GT, limit);
+  EmitOrLink(on_greater);
+}
+
+void RegExpBytecodeGenerator::CheckCharacter(uint32_t c, Label* on_equal) {
+  if (c > MAX_FIRST_ARG) {
+    Emit(BC_CHECK_4_CHARS, 0);
+    Emit32(c);
+  } else {
+    Emit(BC_CHECK_CHAR, c);
+  }
+  EmitOrLink(on_equal);
+}
+
+void RegExpBytecodeGenerator::CheckAtStart(int cp_offset, Label* on_at_start) {
+  Emit(BC_CHECK_AT_START, cp_offset);
+  EmitOrLink(on_at_start);
+}
+
+void RegExpBytecodeGenerator::CheckNotAtStart(int cp_offset,
+                                              Label* on_not_at_start) {
+  Emit(BC_CHECK_NOT_AT_START, cp_offset);
+  EmitOrLink(on_not_at_start);
+}
+
+void RegExpBytecodeGenerator::CheckNotCharacter(uint32_t c,
+                                                Label* on_not_equal) {
+  if (c > MAX_FIRST_ARG) {
+    Emit(BC_CHECK_NOT_4_CHARS, 0);
+    Emit32(c);
+  } else {
+    Emit(BC_CHECK_NOT_CHAR, c);
+  }
+  EmitOrLink(on_not_equal);
+}
+
+void RegExpBytecodeGenerator::CheckCharacterAfterAnd(uint32_t c, uint32_t mask,
+                                                     Label* on_equal) {
+  if (c > MAX_FIRST_ARG) {
+    Emit(BC_AND_CHECK_4_CHARS, 0);
+    Emit32(c);
+  } else {
+    Emit(BC_AND_CHECK_CHAR, c);
+  }
+  Emit32(mask);
+  EmitOrLink(on_equal);
+}
+
+void RegExpBytecodeGenerator::CheckNotCharacterAfterAnd(uint32_t c,
+                                                        uint32_t mask,
+                                                        Label* on_not_equal) {
+  if (c > MAX_FIRST_ARG) {
+    Emit(BC_AND_CHECK_NOT_4_CHARS, 0);
+    Emit32(c);
+  } else {
+    Emit(BC_AND_CHECK_NOT_CHAR, c);
+  }
+  Emit32(mask);
+  EmitOrLink(on_not_equal);
+}
+
+void RegExpBytecodeGenerator::CheckNotCharacterAfterMinusAnd(
+    uc16 c, uc16 minus, uc16 mask, Label* on_not_equal) {
+  Emit(BC_MINUS_AND_CHECK_NOT_CHAR, c);
+  Emit16(minus);
+  Emit16(mask);
+  EmitOrLink(on_not_equal);
+}
+
+void RegExpBytecodeGenerator::CheckCharacterInRange(uc16 from, uc16 to,
+                                                    Label* on_in_range) {
+  Emit(BC_CHECK_CHAR_IN_RANGE, 0);
+  Emit16(from);
+  Emit16(to);
+  EmitOrLink(on_in_range);
+}
+
+void RegExpBytecodeGenerator::CheckCharacterNotInRange(uc16 from, uc16 to,
+                                                       Label* on_not_in_range) {
+  Emit(BC_CHECK_CHAR_NOT_IN_RANGE, 0);
+  Emit16(from);
+  Emit16(to);
+  EmitOrLink(on_not_in_range);
+}
+
+void RegExpBytecodeGenerator::CheckBitInTable(Handle<ByteArray> table,
+                                              Label* on_bit_set) {
+  Emit(BC_CHECK_BIT_IN_TABLE, 0);
+  EmitOrLink(on_bit_set);
+  for (int i = 0; i < kTableSize; i += kBitsPerByte) {
+    int byte = 0;
+    for (int j = 0; j < kBitsPerByte; j++) {
+      if (table->get(i + j) != 0) byte |= 1 << j;
+    }
+    Emit8(byte);
+  }
+}
+
+void RegExpBytecodeGenerator::CheckNotBackReference(int start_reg,
+                                                    bool read_backward,
+                                                    Label* on_not_equal) {
+  DCHECK_LE(0, start_reg);
+  DCHECK_GE(kMaxRegister, start_reg);
+  Emit(read_backward ? BC_CHECK_NOT_BACK_REF_BACKWARD : BC_CHECK_NOT_BACK_REF,
+       start_reg);
+  EmitOrLink(on_not_equal);
+}
+
+void RegExpBytecodeGenerator::CheckNotBackReferenceIgnoreCase(
+    int start_reg, bool read_backward, Label* on_not_equal) {
+  DCHECK_LE(0, start_reg);
+  DCHECK_GE(kMaxRegister, start_reg);
+  Emit(read_backward ? BC_CHECK_NOT_BACK_REF_NO_CASE_BACKWARD
+                     : BC_CHECK_NOT_BACK_REF_NO_CASE,
+       start_reg);
+  EmitOrLink(on_not_equal);
+}
+
+void RegExpBytecodeGenerator::IfRegisterLT(int register_index, int comparand,
+                                           Label* on_less_than) {
+  DCHECK_LE(0, register_index);
+  DCHECK_GE(kMaxRegister, register_index);
+  Emit(BC_CHECK_REGISTER_LT, register_index);
+  Emit32(comparand);
+  EmitOrLink(on_less_than);
+}
+
+void RegExpBytecodeGenerator::IfRegisterGE(int register_index, int comparand,
+                                           Label* on_greater_or_equal) {
+  DCHECK_LE(0, register_index);
+  DCHECK_GE(kMaxRegister, register_index);
+  Emit(BC_CHECK_REGISTER_GE, register_index);
+  Emit32(comparand);
+  EmitOrLink(on_greater_or_equal);
+}
+
+void RegExpBytecodeGenerator::IfRegisterEqPos(int register_index,
+                                              Label* on_eq) {
+  DCHECK_LE(0, register_index);
+  DCHECK_GE(kMaxRegister, register_index);
+  Emit(BC_CHECK_REGISTER_EQ_POS, register_index);
+  EmitOrLink(on_eq);
+}
+
+Handle<HeapObject> RegExpBytecodeGenerator::GetCode(Handle<String> source) {
+  Bind(&backtrack_);
+  Emit(BC_POP_BT, 0);
+
+  Handle<ByteArray> array;
+  if (FLAG_regexp_peephole_optimization) {
+    array = RegExpBytecodePeepholeOptimization::OptimizeBytecode(
+        isolate_, zone(), source, buffer_.begin(), length(), jump_edges_);
+  } else {
+    array = isolate_->factory()->NewByteArray(length());
+    Copy(array->GetDataStartAddress());
+  }
+
+  return array;
+}
+
+int RegExpBytecodeGenerator::length() { return pc_; }
+
+void RegExpBytecodeGenerator::Copy(byte* a) {
+  MemCopy(a, buffer_.begin(), length());
+}
+
+void RegExpBytecodeGenerator::Expand() {
+  Vector<byte> old_buffer = buffer_;
+  buffer_ = Vector<byte>::New(old_buffer.length() * 2);
+  MemCopy(buffer_.begin(), old_buffer.begin(), old_buffer.length());
+  old_buffer.Dispose();
+}
+
+}  // namespace internal
+}  // namespace v8

js/src/regexp/regexp-bytecode-generator.h

@@ -0,0 +1,119 @@
+// Copyright 2012 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef V8_REGEXP_REGEXP_BYTECODE_GENERATOR_H_
+#define V8_REGEXP_REGEXP_BYTECODE_GENERATOR_H_
+
+#include "regexp/regexp-macro-assembler.h"
+
+namespace v8 {
+namespace internal {
+
+// An assembler/generator for the Irregexp byte code.
+class V8_EXPORT_PRIVATE RegExpBytecodeGenerator : public RegExpMacroAssembler {
+ public:
+  // Create an assembler. Instructions and relocation information are emitted
+  // into a buffer, with the instructions starting from the beginning and the
+  // relocation information starting from the end of the buffer. See CodeDesc
+  // for a detailed comment on the layout (globals.h).
+  //
+  // The assembler allocates and grows its own buffer, and buffer_size
+  // determines the initial buffer size. The buffer is owned by the assembler
+  // and deallocated upon destruction of the assembler.
+  RegExpBytecodeGenerator(Isolate* isolate, Zone* zone);
+  virtual ~RegExpBytecodeGenerator();
+  // The byte-code interpreter checks on each push anyway.
+  virtual int stack_limit_slack() { return 1; }
+  virtual bool CanReadUnaligned() { return false; }
+  virtual void Bind(Label* label);
+  virtual void AdvanceCurrentPosition(int by);  // Signed cp change.
+  virtual void PopCurrentPosition();
+  virtual void PushCurrentPosition();
+  virtual void Backtrack();
+  virtual void GoTo(Label* label);
+  virtual void PushBacktrack(Label* label);
+  virtual bool Succeed();
+  virtual void Fail();
+  virtual void PopRegister(int register_index);
+  virtual void PushRegister(int register_index,
+                            StackCheckFlag check_stack_limit);
+  virtual void AdvanceRegister(int reg, int by);  // r[reg] += by.
+  virtual void SetCurrentPositionFromEnd(int by);
+  virtual void SetRegister(int register_index, int to);
+  virtual void WriteCurrentPositionToRegister(int reg, int cp_offset);
+  virtual void ClearRegisters(int reg_from, int reg_to);
+  virtual void ReadCurrentPositionFromRegister(int reg);
+  virtual void WriteStackPointerToRegister(int reg);
+  virtual void ReadStackPointerFromRegister(int reg);
+  virtual void LoadCurrentCharacterImpl(int cp_offset, Label* on_end_of_input,
+                                        bool check_bounds, int characters,
+                                        int eats_at_least);
+  virtual void CheckCharacter(unsigned c, Label* on_equal);
+  virtual void CheckCharacterAfterAnd(unsigned c, unsigned mask,
+                                      Label* on_equal);
+  virtual void CheckCharacterGT(uc16 limit, Label* on_greater);
+  virtual void CheckCharacterLT(uc16 limit, Label* on_less);
+  virtual void CheckGreedyLoop(Label* on_tos_equals_current_position);
+  virtual void CheckAtStart(int cp_offset, Label* on_at_start);
+  virtual void CheckNotAtStart(int cp_offset, Label* on_not_at_start);
+  virtual void CheckNotCharacter(unsigned c, Label* on_not_equal);
+  virtual void CheckNotCharacterAfterAnd(unsigned c, unsigned mask,
+                                         Label* on_not_equal);
+  virtual void CheckNotCharacterAfterMinusAnd(uc16 c, uc16 minus, uc16 mask,
+                                              Label* on_not_equal);
+  virtual void CheckCharacterInRange(uc16 from, uc16 to, Label* on_in_range);
+  virtual void CheckCharacterNotInRange(uc16 from, uc16 to,
+                                        Label* on_not_in_range);
+  virtual void CheckBitInTable(Handle<ByteArray> table, Label* on_bit_set);
+  virtual void CheckNotBackReference(int start_reg, bool read_backward,
+                                     Label* on_no_match);
+  virtual void CheckNotBackReferenceIgnoreCase(int start_reg,
+                                               bool read_backward,
+                                               Label* on_no_match);
+  virtual void IfRegisterLT(int register_index, int comparand, Label* if_lt);
+  virtual void IfRegisterGE(int register_index, int comparand, Label* if_ge);
+  virtual void IfRegisterEqPos(int register_index, Label* if_eq);
+
+  virtual IrregexpImplementation Implementation();
+  virtual Handle<HeapObject> GetCode(Handle<String> source);
+
+ private:
+  void Expand();
+  // Code and bitmap emission.
+  inline void EmitOrLink(Label* label);
+  inline void Emit32(uint32_t x);
+  inline void Emit16(uint32_t x);
+  inline void Emit8(uint32_t x);
+  inline void Emit(uint32_t bc, uint32_t arg);
+  // Bytecode buffer.
+  int length();
+  void Copy(byte* a);
+
+  // The buffer into which code and relocation info are generated.
+  Vector<byte> buffer_;
+  // The program counter.
+  int pc_;
+  Label backtrack_;
+
+  int advance_current_start_;
+  int advance_current_offset_;
+  int advance_current_end_;
+
+  // Stores jump edges emitted for the bytecode (used by
+  // RegExpBytecodePeepholeOptimization).
+  // Key: jump source (offset in buffer_ where jump destination is stored).
+  // Value: jump destination (offset in buffer_ to jump to).
+  ZoneUnorderedMap<int, int> jump_edges_;
+
+  Isolate* isolate_;
+
+  static const int kInvalidPC = -1;
+
+  DISALLOW_IMPLICIT_CONSTRUCTORS(RegExpBytecodeGenerator);
+};
+
+}  // namespace internal
+}  // namespace v8
+
+#endif  // V8_REGEXP_REGEXP_BYTECODE_GENERATOR_H_

js/src/regexp/regexp-bytecode-peephole.h

@@ -0,0 +1,30 @@
+// Copyright 2019 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef V8_REGEXP_REGEXP_BYTECODE_PEEPHOLE_H_
+#define V8_REGEXP_REGEXP_BYTECODE_PEEPHOLE_H_
+
+#include "regexp/regexp-shim.h"
+
+namespace v8 {
+namespace internal {
+
+class ByteArray;
+
+// Peephole optimization for regexp interpreter bytecode.
+// Pre-defined bytecode sequences occuring in the bytecode generated by the
+// RegExpBytecodeGenerator can be optimized into a single bytecode.
+class RegExpBytecodePeepholeOptimization : public AllStatic {
+ public:
+  // Performs peephole optimization on the given bytecode and returns the
+  // optimized bytecode.
+  static Handle<ByteArray> OptimizeBytecode(
+      Isolate* isolate, Zone* zone, Handle<String> source, const byte* bytecode,
+      int length, const ZoneUnorderedMap<int, int>& jump_edges);
+};
+
+}  // namespace internal
+}  // namespace v8
+
+#endif  // V8_REGEXP_REGEXP_BYTECODE_PEEPHOLE_H_

js/src/regexp/regexp-bytecodes.cc

@@ -0,0 +1,45 @@
+// Copyright 2019 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "regexp/regexp-bytecodes.h"
+
+#include <cctype>
+
+
+namespace v8 {
+namespace internal {
+
+void RegExpBytecodeDisassembleSingle(const byte* code_base, const byte* pc) {
+  PrintF("%s", RegExpBytecodeName(*pc));
+
+  // Args and the bytecode as hex.
+  for (int i = 0; i < RegExpBytecodeLength(*pc); i++) {
+    PrintF(", %02x", pc[i]);
+  }
+  PrintF(" ");
+
+  // Args as ascii.
+  for (int i = 1; i < RegExpBytecodeLength(*pc); i++) {
+    unsigned char b = pc[i];
+    PrintF("%c", std::isprint(b) ? b : '.');
+  }
+  PrintF("\n");
+}
+
+void RegExpBytecodeDisassemble(const byte* code_base, int length,
+                               const char* pattern) {
+  PrintF("[generated bytecode for regexp pattern: '%s']\n", pattern);
+
+  ptrdiff_t offset = 0;
+
+  while (offset < length) {
+    const byte* const pc = code_base + offset;
+    PrintF("%p  %4" V8PRIxPTRDIFF "  ", pc, offset);
+    RegExpBytecodeDisassembleSingle(code_base, pc);
+    offset += RegExpBytecodeLength(*pc);
+  }
+}
+
+}  // namespace internal
+}  // namespace v8

js/src/regexp/regexp-bytecodes.h

@@ -0,0 +1,251 @@
+// Copyright 2011 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef V8_REGEXP_REGEXP_BYTECODES_H_
+#define V8_REGEXP_REGEXP_BYTECODES_H_
+
+#include "regexp/regexp-shim.h"
+
+namespace v8 {
+namespace internal {
+
+// Maximum number of bytecodes that will be used (next power of 2 of actually
+// defined bytecodes).
+// All slots between the last actually defined bytecode and maximum id will be
+// filled with BREAKs, indicating an invalid operation. This way using
+// BYTECODE_MASK guarantees no OOB access to the dispatch table.
+constexpr int kRegExpPaddedBytecodeCount = 1 << 6;
+constexpr int BYTECODE_MASK = kRegExpPaddedBytecodeCount - 1;
+// The first argument is packed in with the byte code in one word, but so it
+// has 24 bits, but it can be positive and negative so only use 23 bits for
+// positive values.
+const unsigned int MAX_FIRST_ARG = 0x7fffffu;
+const int BYTECODE_SHIFT = 8;
+STATIC_ASSERT(1 << BYTECODE_SHIFT > BYTECODE_MASK);
+
+// TODO(pthier): Argument offsets of bytecodes should be easily accessible by
+// name or at least by position.
+#define BYTECODE_ITERATOR(V)                                                   \
+  V(BREAK, 0, 4)              /* bc8                                        */ \
+  V(PUSH_CP, 1, 4)            /* bc8 pad24                                  */ \
+  V(PUSH_BT, 2, 8)            /* bc8 pad24 offset32                         */ \
+  V(PUSH_REGISTER, 3, 4)      /* bc8 reg_idx24                              */ \
+  V(SET_REGISTER_TO_CP, 4, 8) /* bc8 reg_idx24 offset32                     */ \
+  V(SET_CP_TO_REGISTER, 5, 4) /* bc8 reg_idx24                              */ \
+  V(SET_REGISTER_TO_SP, 6, 4) /* bc8 reg_idx24                              */ \
+  V(SET_SP_TO_REGISTER, 7, 4) /* bc8 reg_idx24                              */ \
+  V(SET_REGISTER, 8, 8)       /* bc8 reg_idx24 value32                      */ \
+  V(ADVANCE_REGISTER, 9, 8)   /* bc8 reg_idx24 value32                      */ \
+  V(POP_CP, 10, 4)            /* bc8 pad24                                  */ \
+  V(POP_BT, 11, 4)            /* bc8 pad24                                  */ \
+  V(POP_REGISTER, 12, 4)      /* bc8 reg_idx24                              */ \
+  V(FAIL, 13, 4)              /* bc8 pad24                                  */ \
+  V(SUCCEED, 14, 4)           /* bc8 pad24                                  */ \
+  V(ADVANCE_CP, 15, 4)        /* bc8 offset24                               */ \
+  /* Jump to another bytecode given its offset.                             */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07:   0x10 (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x1F:   0x00 (unused) Padding                                   */ \
+  /* 0x20 - 0x3F:   Address of bytecode to jump to                          */ \
+  V(GOTO, 16, 8) /* bc8 pad24 addr32                           */              \
+  /* Check if offset is in range and load character at given offset.        */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07:   0x11 (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x1F:   Offset from current position                            */ \
+  /* 0x20 - 0x3F:   Address of bytecode when load is out of range           */ \
+  V(LOAD_CURRENT_CHAR, 17, 8) /* bc8 offset24 addr32                        */ \
+  /* Load character at given offset without range checks.                   */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07:   0x12 (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x1F:   Offset from current position                            */ \
+  V(LOAD_CURRENT_CHAR_UNCHECKED, 18, 4)    /* bc8 offset24 */                  \
+  V(LOAD_2_CURRENT_CHARS, 19, 8)           /* bc8 offset24 addr32 */           \
+  V(LOAD_2_CURRENT_CHARS_UNCHECKED, 20, 4) /* bc8 offset24 */                  \
+  V(LOAD_4_CURRENT_CHARS, 21, 8)           /* bc8 offset24 addr32 */           \
+  V(LOAD_4_CURRENT_CHARS_UNCHECKED, 22, 4) /* bc8 offset24 */                  \
+  V(CHECK_4_CHARS, 23, 12) /* bc8 pad24 uint32 addr32                    */    \
+  /* Check if current character is equal to a given character               */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07:   0x19 (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x0F:   0x00 (unused) Padding                                   */ \
+  /* 0x10 - 0x1F:   Character to check                                      */ \
+  /* 0x20 - 0x3F:   Address of bytecode when matched                        */ \
+  V(CHECK_CHAR, 24, 8) /* bc8 pad8 uint16 addr32                     */        \
+  V(CHECK_NOT_4_CHARS, 25, 12) /* bc8 pad24 uint32 addr32 */                   \
+  V(CHECK_NOT_CHAR, 26, 8) /* bc8 pad8 uint16 addr32                     */    \
+  V(AND_CHECK_4_CHARS, 27, 16) /* bc8 pad24 uint32 uint32 addr32 */            \
+  /* Checks if the current character combined with mask (bitwise and)       */ \
+  /* matches a character (e.g. used when two characters in a disjunction    */ \
+  /* differ by only a single bit                                            */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07:   0x1c (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x0F:   0x00 (unused) Padding                                   */ \
+  /* 0x10 - 0x1F:   Character to match against (after mask aplied)          */ \
+  /* 0x20 - 0x3F:   Bitmask bitwise and combined with current character     */ \
+  /* 0x40 - 0x5F:   Address of bytecode when matched                        */ \
+  V(AND_CHECK_CHAR, 28, 12)           /* bc8 pad8 uint16 uint32 addr32      */ \
+  V(AND_CHECK_NOT_4_CHARS, 29, 16)    /* bc8 pad24 uint32 uint32 addr32 */     \
+  V(AND_CHECK_NOT_CHAR, 30, 12)       /* bc8 pad8 uint16 uint32 addr32 */      \
+  V(MINUS_AND_CHECK_NOT_CHAR, 31, 12) /* bc8 pad8 uc16 uc16 uc16 addr32 */     \
+  V(CHECK_CHAR_IN_RANGE, 32, 12)      /* bc8 pad24 uc16 uc16 addr32 */         \
+  V(CHECK_CHAR_NOT_IN_RANGE, 33, 12)  /* bc8 pad24 uc16 uc16 addr32 */         \
+  /* Checks if the current character matches any of the characters encoded  */ \
+  /* in a bit table. Similar to/inspired by boyer moore string search       */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07:   0x22 (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x1F:   0x00 (unused) Padding                                   */ \
+  /* 0x20 - 0x3F:   Address of bytecode when bit is set                     */ \
+  /* 0x40 - 0xBF:   Bit table                                               */ \
+  V(CHECK_BIT_IN_TABLE, 34, 24) /* bc8 pad24 addr32 bits128           */       \
+  V(CHECK_LT, 35, 8) /* bc8 pad8 uc16 addr32                       */          \
+  V(CHECK_GT, 36, 8) /* bc8 pad8 uc16 addr32                       */          \
+  V(CHECK_NOT_BACK_REF, 37, 8)                  /* bc8 reg_idx24 addr32 */     \
+  V(CHECK_NOT_BACK_REF_NO_CASE, 38, 8)          /* bc8 reg_idx24 addr32 */     \
+  V(CHECK_NOT_BACK_REF_NO_CASE_UNICODE, 39, 8)  /* UNUSED */                   \
+  V(CHECK_NOT_BACK_REF_BACKWARD, 40, 8)         /* bc8 reg_idx24 addr32 */     \
+  V(CHECK_NOT_BACK_REF_NO_CASE_BACKWARD, 41, 8) /* bc8 reg_idx24 addr32 */     \
+  V(CHECK_NOT_BACK_REF_NO_CASE_UNICODE_BACKWARD, 42, 8) /* UNUSED */           \
+  V(CHECK_NOT_REGS_EQUAL, 43, 12) /* bc8 regidx24 reg_idx32 addr32 */          \
+  V(CHECK_REGISTER_LT, 44, 12)    /* bc8 reg_idx24 value32 addr32 */           \
+  V(CHECK_REGISTER_GE, 45, 12)    /* bc8 reg_idx24 value32 addr32 */           \
+  V(CHECK_REGISTER_EQ_POS, 46, 8) /* bc8 reg_idx24 addr32 */                   \
+  V(CHECK_AT_START, 47, 8) /* bc8 pad24 addr32                           */    \
+  V(CHECK_NOT_AT_START, 48, 8) /* bc8 offset24 addr32 */                       \
+  /* Checks if the current position matches top of backtrack stack          */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07:   0x31 (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x1F:   0x00 (unused) Padding                                   */ \
+  /* 0x20 - 0x3F:   Address of bytecode when current matches tos            */ \
+  V(CHECK_GREEDY, 49, 8) /* bc8 pad24 addr32                           */      \
+  /* Advance character pointer by given offset and jump to another bytecode.*/ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07:   0x32 (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x1F:   Number of characters to advance                         */ \
+  /* 0x20 - 0x3F:   Address of bytecode to jump to                          */ \
+  V(ADVANCE_CP_AND_GOTO, 50, 8) /* bc8 offset24 addr32                    */   \
+  V(SET_CURRENT_POSITION_FROM_END, 51, 4) /* bc8 idx24 */                      \
+  /* Checks if current position + given offset is in range.                 */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07:   0x34 (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x1F:   Offset from current position                            */ \
+  /* 0x20 - 0x3F:   Address of bytecode when position is out of range       */ \
+  V(CHECK_CURRENT_POSITION, 52, 8) /* bc8 idx24 addr32                     */  \
+  /* Combination of:                                                        */ \
+  /* LOAD_CURRENT_CHAR, CHECK_BIT_IN_TABLE and ADVANCE_CP_AND_GOTO          */ \
+  /* Emitted by RegExpBytecodePeepholeOptimization.                         */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07    0x35 (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x1F    Load character offset from current position             */ \
+  /* 0x20 - 0x3F    Number of characters to advance                         */ \
+  /* 0x40 - 0xBF    Bit Table                                               */ \
+  /* 0xC0 - 0xDF    Address of bytecode when character is matched           */ \
+  /* 0xE0 - 0xFF    Address of bytecode when no match                       */ \
+  V(SKIP_UNTIL_BIT_IN_TABLE, 53, 32)                                           \
+  /* Combination of:                                                        */ \
+  /* CHECK_CURRENT_POSITION, LOAD_CURRENT_CHAR_UNCHECKED, AND_CHECK_CHAR    */ \
+  /* and ADVANCE_CP_AND_GOTO                                                */ \
+  /* Emitted by RegExpBytecodePeepholeOptimization.                         */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07    0x36 (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x1F    Load character offset from current position             */ \
+  /* 0x20 - 0x2F    Number of characters to advance                         */ \
+  /* 0x30 - 0x3F    Character to match against (after mask applied)         */ \
+  /* 0x40 - 0x5F:   Bitmask bitwise and combined with current character     */ \
+  /* 0x60 - 0x7F    Minimum number of characters this pattern consumes      */ \
+  /* 0x80 - 0x9F    Address of bytecode when character is matched           */ \
+  /* 0xA0 - 0xBF    Address of bytecode when no match                       */ \
+  V(SKIP_UNTIL_CHAR_AND, 54, 24)                                               \
+  /* Combination of:                                                        */ \
+  /* LOAD_CURRENT_CHAR, CHECK_CHAR and ADVANCE_CP_AND_GOTO                  */ \
+  /* Emitted by RegExpBytecodePeepholeOptimization.                         */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07    0x37 (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x1F    Load character offset from current position             */ \
+  /* 0x20 - 0x2F    Number of characters to advance                         */ \
+  /* 0x30 - 0x3F    Character to match                                      */ \
+  /* 0x40 - 0x5F    Address of bytecode when character is matched           */ \
+  /* 0x60 - 0x7F    Address of bytecode when no match                       */ \
+  V(SKIP_UNTIL_CHAR, 55, 16)                                                   \
+  /* Combination of:                                                        */ \
+  /* CHECK_CURRENT_POSITION, LOAD_CURRENT_CHAR_UNCHECKED, CHECK_CHAR        */ \
+  /* and ADVANCE_CP_AND_GOTO                                                */ \
+  /* Emitted by RegExpBytecodePeepholeOptimization.                         */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07    0x38 (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x1F    Load character offset from current position             */ \
+  /* 0x20 - 0x2F    Number of characters to advance                         */ \
+  /* 0x30 - 0x3F    Character to match                                      */ \
+  /* 0x40 - 0x5F    Minimum number of characters this pattern consumes      */ \
+  /* 0x60 - 0x7F    Address of bytecode when character is matched           */ \
+  /* 0x80 - 0x9F    Address of bytecode when no match                       */ \
+  V(SKIP_UNTIL_CHAR_POS_CHECKED, 56, 20)                                       \
+  /* Combination of:                                                        */ \
+  /* LOAD_CURRENT_CHAR, CHECK_CHAR, CHECK_CHAR and ADVANCE_CP_AND_GOTO      */ \
+  /* Emitted by RegExpBytecodePeepholeOptimization.                         */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07    0x39 (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x1F    Load character offset from current position             */ \
+  /* 0x20 - 0x3F    Number of characters to advance                         */ \
+  /* 0x40 - 0x4F    Character to match                                      */ \
+  /* 0x50 - 0x5F    Other Character to match                                */ \
+  /* 0x60 - 0x7F    Address of bytecode when either character is matched    */ \
+  /* 0x80 - 0x9F    Address of bytecode when no match                       */ \
+  V(SKIP_UNTIL_CHAR_OR_CHAR, 57, 20)                                           \
+  /* Combination of:                                                        */ \
+  /* LOAD_CURRENT_CHAR, CHECK_GT, CHECK_BIT_IN_TABLE, GOTO and              */ \
+  /* and ADVANCE_CP_AND_GOTO                                                */ \
+  /* Emitted by RegExpBytecodePeepholeOptimization.                         */ \
+  /* Bit Layout:                                                            */ \
+  /* 0x00 - 0x07    0x3A (fixed) Bytecode                                   */ \
+  /* 0x08 - 0x1F    Load character offset from current position             */ \
+  /* 0x20 - 0x2F    Number of characters to advance                         */ \
+  /* 0x30 - 0x3F    Character to check if it is less than current char      */ \
+  /* 0x40 - 0xBF    Bit Table                                               */ \
+  /* 0xC0 - 0xDF    Address of bytecode when character is matched           */ \
+  /* 0xE0 - 0xFF    Address of bytecode when no match                       */ \
+  V(SKIP_UNTIL_GT_OR_NOT_BIT_IN_TABLE, 58, 32)
+
+#define COUNT(...) +1
+static constexpr int kRegExpBytecodeCount = BYTECODE_ITERATOR(COUNT);
+#undef COUNT
+
+// Just making sure we assigned values above properly. They should be
+// contiguous, strictly increasing, and start at 0.
+// TODO(jgruber): Do not explicitly assign values, instead generate them
+// implicitly from the list order.
+STATIC_ASSERT(kRegExpBytecodeCount == 59);
+
+#define DECLARE_BYTECODES(name, code, length) \
+  static constexpr int BC_##name = code;
+BYTECODE_ITERATOR(DECLARE_BYTECODES)
+#undef DECLARE_BYTECODES
+
+static constexpr int kRegExpBytecodeLengths[] = {
+#define DECLARE_BYTECODE_LENGTH(name, code, length) length,
+    BYTECODE_ITERATOR(DECLARE_BYTECODE_LENGTH)
+#undef DECLARE_BYTECODE_LENGTH
+};
+
+inline constexpr int RegExpBytecodeLength(int bytecode) {
+  return kRegExpBytecodeLengths[bytecode];
+}
+
+static const char* const kRegExpBytecodeNames[] = {
+#define DECLARE_BYTECODE_NAME(name, ...) #name,
+    BYTECODE_ITERATOR(DECLARE_BYTECODE_NAME)
+#undef DECLARE_BYTECODE_NAME
+};
+
+inline const char* RegExpBytecodeName(int bytecode) {
+  return kRegExpBytecodeNames[bytecode];
+}
+
+void RegExpBytecodeDisassembleSingle(const byte* code_base, const byte* pc);
+void RegExpBytecodeDisassemble(const byte* code_base, int length,
+                               const char* pattern);
+
+}  // namespace internal
+}  // namespace v8
+
+#endif  // V8_REGEXP_REGEXP_BYTECODES_H_

js/src/regexp/regexp-compiler.h

@@ -0,0 +1,621 @@
+// Copyright 2019 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef V8_REGEXP_REGEXP_COMPILER_H_
+#define V8_REGEXP_REGEXP_COMPILER_H_
+
+#include <bitset>
+
+#include "regexp/regexp-nodes.h"
+
+namespace v8 {
+namespace internal {
+
+class DynamicBitSet;
+class Isolate;
+
+namespace regexp_compiler_constants {
+
+// The '2' variant is has inclusive from and exclusive to.
+// This covers \s as defined in ECMA-262 5.1, 15.10.2.12,
+// which include WhiteSpace (7.2) or LineTerminator (7.3) values.
+constexpr uc32 kRangeEndMarker = 0x110000;
+constexpr int kSpaceRanges[] = {
+    '\t',   '\r' + 1, ' ',    ' ' + 1, 0x00A0, 0x00A1, 0x1680,
+    0x1681, 0x2000,   0x200B, 0x2028,  0x202A, 0x202F, 0x2030,
+    0x205F, 0x2060,   0x3000, 0x3001,  0xFEFF, 0xFF00, kRangeEndMarker};
+constexpr int kSpaceRangeCount = arraysize(kSpaceRanges);
+
+constexpr int kWordRanges[] = {'0',     '9' + 1, 'A',     'Z' + 1,        '_',
+                               '_' + 1, 'a',     'z' + 1, kRangeEndMarker};
+constexpr int kWordRangeCount = arraysize(kWordRanges);
+constexpr int kDigitRanges[] = {'0', '9' + 1, kRangeEndMarker};
+constexpr int kDigitRangeCount = arraysize(kDigitRanges);
+constexpr int kSurrogateRanges[] = {kLeadSurrogateStart,
+                                    kLeadSurrogateStart + 1, kRangeEndMarker};
+constexpr int kSurrogateRangeCount = arraysize(kSurrogateRanges);
+constexpr int kLineTerminatorRanges[] = {0x000A, 0x000B, 0x000D,         0x000E,
+                                         0x2028, 0x202A, kRangeEndMarker};
+constexpr int kLineTerminatorRangeCount = arraysize(kLineTerminatorRanges);
+
+// More makes code generation slower, less makes V8 benchmark score lower.
+constexpr int kMaxLookaheadForBoyerMoore = 8;
+// In a 3-character pattern you can maximally step forwards 3 characters
+// at a time, which is not always enough to pay for the extra logic.
+constexpr int kPatternTooShortForBoyerMoore = 2;
+
+}  // namespace regexp_compiler_constants
+
+inline bool IgnoreCase(JSRegExp::Flags flags) {
+  return (flags & JSRegExp::kIgnoreCase) != 0;
+}
+
+inline bool IsUnicode(JSRegExp::Flags flags) {
+  return (flags & JSRegExp::kUnicode) != 0;
+}
+
+inline bool IsSticky(JSRegExp::Flags flags) {
+  return (flags & JSRegExp::kSticky) != 0;
+}
+
+inline bool IsGlobal(JSRegExp::Flags flags) {
+  return (flags & JSRegExp::kGlobal) != 0;
+}
+
+inline bool DotAll(JSRegExp::Flags flags) {
+  return (flags & JSRegExp::kDotAll) != 0;
+}
+
+inline bool Multiline(JSRegExp::Flags flags) {
+  return (flags & JSRegExp::kMultiline) != 0;
+}
+
+inline bool NeedsUnicodeCaseEquivalents(JSRegExp::Flags flags) {
+  // Both unicode and ignore_case flags are set. We need to use ICU to find
+  // the closure over case equivalents.
+  return IsUnicode(flags) && IgnoreCase(flags);
+}
+
+// Details of a quick mask-compare check that can look ahead in the
+// input stream.
+class QuickCheckDetails {
+ public:
+  QuickCheckDetails()
+      : characters_(0), mask_(0), value_(0), cannot_match_(false) {}
+  explicit QuickCheckDetails(int characters)
+      : characters_(characters), mask_(0), value_(0), cannot_match_(false) {}
+  bool Rationalize(bool one_byte);
+  // Merge in the information from another branch of an alternation.
+  void Merge(QuickCheckDetails* other, int from_index);
+  // Advance the current position by some amount.
+  void Advance(int by, bool one_byte);
+  void Clear();
+  bool cannot_match() { return cannot_match_; }
+  void set_cannot_match() { cannot_match_ = true; }
+  struct Position {
+    Position() : mask(0), value(0), determines_perfectly(false) {}
+    uc16 mask;
+    uc16 value;
+    bool determines_perfectly;
+  };
+  int characters() { return characters_; }
+  void set_characters(int characters) { characters_ = characters; }
+  Position* positions(int index) {
+    DCHECK_LE(0, index);
+    DCHECK_GT(characters_, index);
+    return positions_ + index;
+  }
+  uint32_t mask() { return mask_; }
+  uint32_t value() { return value_; }
+
+ private:
+  // How many characters do we have quick check information from.  This is
+  // the same for all branches of a choice node.
+  int characters_;
+  Position positions_[4];
+  // These values are the condensate of the above array after Rationalize().
+  uint32_t mask_;
+  uint32_t value_;
+  // If set to true, there is no way this quick check can match at all.
+  // E.g., if it requires to be at the start of the input, and isn't.
+  bool cannot_match_;
+};
+
+// Improve the speed that we scan for an initial point where a non-anchored
+// regexp can match by using a Boyer-Moore-like table. This is done by
+// identifying non-greedy non-capturing loops in the nodes that eat any
+// character one at a time.  For example in the middle of the regexp
+// /foo[\s\S]*?bar/ we find such a loop.  There is also such a loop implicitly
+// inserted at the start of any non-anchored regexp.
+//
+// When we have found such a loop we look ahead in the nodes to find the set of
+// characters that can come at given distances. For example for the regexp
+// /.?foo/ we know that there are at least 3 characters ahead of us, and the
+// sets of characters that can occur are [any, [f, o], [o]]. We find a range in
+// the lookahead info where the set of characters is reasonably constrained. In
+// our example this is from index 1 to 2 (0 is not constrained). We can now
+// look 3 characters ahead and if we don't find one of [f, o] (the union of
+// [f, o] and [o]) then we can skip forwards by the range size (in this case 2).
+//
+// For Unicode input strings we do the same, but modulo 128.
+//
+// We also look at the first string fed to the regexp and use that to get a hint
+// of the character frequencies in the inputs. This affects the assessment of
+// whether the set of characters is 'reasonably constrained'.
+//
+// We also have another lookahead mechanism (called quick check in the code),
+// which uses a wide load of multiple characters followed by a mask and compare
+// to determine whether a match is possible at this point.
+enum ContainedInLattice {
+  kNotYet = 0,
+  kLatticeIn = 1,
+  kLatticeOut = 2,
+  kLatticeUnknown = 3  // Can also mean both in and out.
+};
+
+inline ContainedInLattice Combine(ContainedInLattice a, ContainedInLattice b) {
+  return static_cast<ContainedInLattice>(a | b);
+}
+
+class BoyerMoorePositionInfo : public ZoneObject {
+ public:
+  bool at(int i) const { return map_[i]; }
+
+  static constexpr int kMapSize = 128;
+  static constexpr int kMask = kMapSize - 1;
+
+  int map_count() const { return map_count_; }
+
+  void Set(int character);
+  void SetInterval(const Interval& interval);
+  void SetAll();
+
+  bool is_non_word() { return w_ == kLatticeOut; }
+  bool is_word() { return w_ == kLatticeIn; }
+
+  using Bitset = std::bitset<kMapSize>;
+  Bitset raw_bitset() const { return map_; }
+
+ private:
+  Bitset map_;
+  int map_count_ = 0;               // Number of set bits in the map.
+  ContainedInLattice w_ = kNotYet;  // The \w character class.
+};
+
+class BoyerMooreLookahead : public ZoneObject {
+ public:
+  BoyerMooreLookahead(int length, RegExpCompiler* compiler, Zone* zone);
+
+  int length() { return length_; }
+  int max_char() { return max_char_; }
+  RegExpCompiler* compiler() { return compiler_; }
+
+  int Count(int map_number) { return bitmaps_->at(map_number)->map_count(); }
+
+  BoyerMoorePositionInfo* at(int i) { return bitmaps_->at(i); }
+
+  void Set(int map_number, int character) {
+    if (character > max_char_) return;
+    BoyerMoorePositionInfo* info = bitmaps_->at(map_number);
+    info->Set(character);
+  }
+
+  void SetInterval(int map_number, const Interval& interval) {
+    if (interval.from() > max_char_) return;
+    BoyerMoorePositionInfo* info = bitmaps_->at(map_number);
+    if (interval.to() > max_char_) {
+      info->SetInterval(Interval(interval.from(), max_char_));
+    } else {
+      info->SetInterval(interval);
+    }
+  }
+
+  void SetAll(int map_number) { bitmaps_->at(map_number)->SetAll(); }
+
+  void SetRest(int from_map) {
+    for (int i = from_map; i < length_; i++) SetAll(i);
+  }
+  void EmitSkipInstructions(RegExpMacroAssembler* masm);
+
+ private:
+  // This is the value obtained by EatsAtLeast.  If we do not have at least this
+  // many characters left in the sample string then the match is bound to fail.
+  // Therefore it is OK to read a character this far ahead of the current match
+  // point.
+  int length_;
+  RegExpCompiler* compiler_;
+  // 0xff for Latin1, 0xffff for UTF-16.
+  int max_char_;
+  ZoneList<BoyerMoorePositionInfo*>* bitmaps_;
+
+  int GetSkipTable(int min_lookahead, int max_lookahead,
+                   Handle<ByteArray> boolean_skip_table);
+  bool FindWorthwhileInterval(int* from, int* to);
+  int FindBestInterval(int max_number_of_chars, int old_biggest_points,
+                       int* from, int* to);
+};
+
+// There are many ways to generate code for a node.  This class encapsulates
+// the current way we should be generating.  In other words it encapsulates
+// the current state of the code generator.  The effect of this is that we
+// generate code for paths that the matcher can take through the regular
+// expression.  A given node in the regexp can be code-generated several times
+// as it can be part of several traces.  For example for the regexp:
+// /foo(bar|ip)baz/ the code to match baz will be generated twice, once as part
+// of the foo-bar-baz trace and once as part of the foo-ip-baz trace.  The code
+// to match foo is generated only once (the traces have a common prefix).  The
+// code to store the capture is deferred and generated (twice) after the places
+// where baz has been matched.
+class Trace {
+ public:
+  // A value for a property that is either known to be true, know to be false,
+  // or not known.
+  enum TriBool { UNKNOWN = -1, FALSE_VALUE = 0, TRUE_VALUE = 1 };
+
+  class DeferredAction {
+   public:
+    DeferredAction(ActionNode::ActionType action_type, int reg)
+        : action_type_(action_type), reg_(reg), next_(nullptr) {}
+    DeferredAction* next() { return next_; }
+    bool Mentions(int reg);
+    int reg() { return reg_; }
+    ActionNode::ActionType action_type() { return action_type_; }
+
+   private:
+    ActionNode::ActionType action_type_;
+    int reg_;
+    DeferredAction* next_;
+    friend class Trace;
+  };
+
+  class DeferredCapture : public DeferredAction {
+   public:
+    DeferredCapture(int reg, bool is_capture, Trace* trace)
+        : DeferredAction(ActionNode::STORE_POSITION, reg),
+          cp_offset_(trace->cp_offset()),
+          is_capture_(is_capture) {}
+    int cp_offset() { return cp_offset_; }
+    bool is_capture() { return is_capture_; }
+
+   private:
+    int cp_offset_;
+    bool is_capture_;
+    void set_cp_offset(int cp_offset) { cp_offset_ = cp_offset; }
+  };
+
+  class DeferredSetRegisterForLoop : public DeferredAction {
+   public:
+    DeferredSetRegisterForLoop(int reg, int value)
+        : DeferredAction(ActionNode::SET_REGISTER_FOR_LOOP, reg),
+          value_(value) {}
+    int value() { return value_; }
+
+   private:
+    int value_;
+  };
+
+  class DeferredClearCaptures : public DeferredAction {
+   public:
+    explicit DeferredClearCaptures(Interval range)
+        : DeferredAction(ActionNode::CLEAR_CAPTURES, -1), range_(range) {}
+    Interval range() { return range_; }
+
+   private:
+    Interval range_;
+  };
+
+  class DeferredIncrementRegister : public DeferredAction {
+   public:
+    explicit DeferredIncrementRegister(int reg)
+        : DeferredAction(ActionNode::INCREMENT_REGISTER, reg) {}
+  };
+
+  Trace()
+      : cp_offset_(0),
+        actions_(nullptr),
+        backtrack_(nullptr),
+        stop_node_(nullptr),
+        loop_label_(nullptr),
+        characters_preloaded_(0),
+        bound_checked_up_to_(0),
+        flush_budget_(100),
+        at_start_(UNKNOWN) {}
+
+  // End the trace.  This involves flushing the deferred actions in the trace
+  // and pushing a backtrack location onto the backtrack stack.  Once this is
+  // done we can start a new trace or go to one that has already been
+  // generated.
+  void Flush(RegExpCompiler* compiler, RegExpNode* successor);
+  int cp_offset() { return cp_offset_; }
+  DeferredAction* actions() { return actions_; }
+  // A trivial trace is one that has no deferred actions or other state that
+  // affects the assumptions used when generating code.  There is no recorded
+  // backtrack location in a trivial trace, so with a trivial trace we will
+  // generate code that, on a failure to match, gets the backtrack location
+  // from the backtrack stack rather than using a direct jump instruction.  We
+  // always start code generation with a trivial trace and non-trivial traces
+  // are created as we emit code for nodes or add to the list of deferred
+  // actions in the trace.  The location of the code generated for a node using
+  // a trivial trace is recorded in a label in the node so that gotos can be
+  // generated to that code.
+  bool is_trivial() {
+    return backtrack_ == nullptr && actions_ == nullptr && cp_offset_ == 0 &&
+           characters_preloaded_ == 0 && bound_checked_up_to_ == 0 &&
+           quick_check_performed_.characters() == 0 && at_start_ == UNKNOWN;
+  }
+  TriBool at_start() { return at_start_; }
+  void set_at_start(TriBool at_start) { at_start_ = at_start; }
+  Label* backtrack() { return backtrack_; }
+  Label* loop_label() { return loop_label_; }
+  RegExpNode* stop_node() { return stop_node_; }
+  int characters_preloaded() { return characters_preloaded_; }
+  int bound_checked_up_to() { return bound_checked_up_to_; }
+  int flush_budget() { return flush_budget_; }
+  QuickCheckDetails* quick_check_performed() { return &quick_check_performed_; }
+  bool mentions_reg(int reg);
+  // Returns true if a deferred position store exists to the specified
+  // register and stores the offset in the out-parameter.  Otherwise
+  // returns false.
+  bool GetStoredPosition(int reg, int* cp_offset);
+  // These set methods and AdvanceCurrentPositionInTrace should be used only on
+  // new traces - the intention is that traces are immutable after creation.
+  void add_action(DeferredAction* new_action) {
+    DCHECK(new_action->next_ == nullptr);
+    new_action->next_ = actions_;
+    actions_ = new_action;
+  }
+  void set_backtrack(Label* backtrack) { backtrack_ = backtrack; }
+  void set_stop_node(RegExpNode* node) { stop_node_ = node; }
+  void set_loop_label(Label* label) { loop_label_ = label; }
+  void set_characters_preloaded(int count) { characters_preloaded_ = count; }
+  void set_bound_checked_up_to(int to) { bound_checked_up_to_ = to; }
+  void set_flush_budget(int to) { flush_budget_ = to; }
+  void set_quick_check_performed(QuickCheckDetails* d) {
+    quick_check_performed_ = *d;
+  }
+  void InvalidateCurrentCharacter();
+  void AdvanceCurrentPositionInTrace(int by, RegExpCompiler* compiler);
+
+ private:
+  int FindAffectedRegisters(DynamicBitSet* affected_registers, Zone* zone);
+  void PerformDeferredActions(RegExpMacroAssembler* macro, int max_register,
+                              const DynamicBitSet& affected_registers,
+                              DynamicBitSet* registers_to_pop,
+                              DynamicBitSet* registers_to_clear, Zone* zone);
+  void RestoreAffectedRegisters(RegExpMacroAssembler* macro, int max_register,
+                                const DynamicBitSet& registers_to_pop,
+                                const DynamicBitSet& registers_to_clear);
+  int cp_offset_;
+  DeferredAction* actions_;
+  Label* backtrack_;
+  RegExpNode* stop_node_;
+  Label* loop_label_;
+  int characters_preloaded_;
+  int bound_checked_up_to_;
+  QuickCheckDetails quick_check_performed_;
+  int flush_budget_;
+  TriBool at_start_;
+};
+
+class GreedyLoopState {
+ public:
+  explicit GreedyLoopState(bool not_at_start);
+
+  Label* label() { return &label_; }
+  Trace* counter_backtrack_trace() { return &counter_backtrack_trace_; }
+
+ private:
+  Label label_;
+  Trace counter_backtrack_trace_;
+};
+
+struct PreloadState {
+  static const int kEatsAtLeastNotYetInitialized = -1;
+  bool preload_is_current_;
+  bool preload_has_checked_bounds_;
+  int preload_characters_;
+  int eats_at_least_;
+  void init() { eats_at_least_ = kEatsAtLeastNotYetInitialized; }
+};
+
+// Analysis performs assertion propagation and computes eats_at_least_ values.
+// See the comments on AssertionPropagator and EatsAtLeastPropagator for more
+// details.
+RegExpError AnalyzeRegExp(Isolate* isolate, bool is_one_byte, RegExpNode* node);
+
+class FrequencyCollator {
+ public:
+  FrequencyCollator() : total_samples_(0) {
+    for (int i = 0; i < RegExpMacroAssembler::kTableSize; i++) {
+      frequencies_[i] = CharacterFrequency(i);
+    }
+  }
+
+  void CountCharacter(int character) {
+    int index = (character & RegExpMacroAssembler::kTableMask);
+    frequencies_[index].Increment();
+    total_samples_++;
+  }
+
+  // Does not measure in percent, but rather per-128 (the table size from the
+  // regexp macro assembler).
+  int Frequency(int in_character) {
+    DCHECK((in_character & RegExpMacroAssembler::kTableMask) == in_character);
+    if (total_samples_ < 1) return 1;  // Division by zero.
+    int freq_in_per128 =
+        (frequencies_[in_character].counter() * 128) / total_samples_;
+    return freq_in_per128;
+  }
+
+ private:
+  class CharacterFrequency {
+   public:
+    CharacterFrequency() : counter_(0), character_(-1) {}
+    explicit CharacterFrequency(int character)
+        : counter_(0), character_(character) {}
+
+    void Increment() { counter_++; }
+    int counter() { return counter_; }
+    int character() { return character_; }
+
+   private:
+    int counter_;
+    int character_;
+  };
+
+ private:
+  CharacterFrequency frequencies_[RegExpMacroAssembler::kTableSize];
+  int total_samples_;
+};
+
+class RegExpCompiler {
+ public:
+  RegExpCompiler(Isolate* isolate, Zone* zone, int capture_count,
+                 bool is_one_byte);
+
+  int AllocateRegister() {
+    if (next_register_ >= RegExpMacroAssembler::kMaxRegister) {
+      reg_exp_too_big_ = true;
+      return next_register_;
+    }
+    return next_register_++;
+  }
+
+  // Lookarounds to match lone surrogates for unicode character class matches
+  // are never nested. We can therefore reuse registers.
+  int UnicodeLookaroundStackRegister() {
+    if (unicode_lookaround_stack_register_ == kNoRegister) {
+      unicode_lookaround_stack_register_ = AllocateRegister();
+    }
+    return unicode_lookaround_stack_register_;
+  }
+
+  int UnicodeLookaroundPositionRegister() {
+    if (unicode_lookaround_position_register_ == kNoRegister) {
+      unicode_lookaround_position_register_ = AllocateRegister();
+    }
+    return unicode_lookaround_position_register_;
+  }
+
+  struct CompilationResult final {
+    explicit CompilationResult(RegExpError err) : error(err) {}
+    CompilationResult(Object code, int registers)
+        : code(code), num_registers(registers) {}
+
+    static CompilationResult RegExpTooBig() {
+      return CompilationResult(RegExpError::kTooLarge);
+    }
+
+    bool Succeeded() const { return error == RegExpError::kNone; }
+
+    const RegExpError error = RegExpError::kNone;
+    Object code;
+    int num_registers = 0;
+  };
+
+  CompilationResult Assemble(Isolate* isolate, RegExpMacroAssembler* assembler,
+                             RegExpNode* start, int capture_count,
+                             Handle<String> pattern);
+
+  // If the regexp matching starts within a surrogate pair, step back to the
+  // lead surrogate and start matching from there.
+  static RegExpNode* OptionallyStepBackToLeadSurrogate(RegExpCompiler* compiler,
+                                                       RegExpNode* on_success,
+                                                       JSRegExp::Flags flags);
+
+  inline void AddWork(RegExpNode* node) {
+    if (!node->on_work_list() && !node->label()->is_bound()) {
+      node->set_on_work_list(true);
+      work_list_->push_back(node);
+    }
+  }
+
+  static const int kImplementationOffset = 0;
+  static const int kNumberOfRegistersOffset = 0;
+  static const int kCodeOffset = 1;
+
+  RegExpMacroAssembler* macro_assembler() { return macro_assembler_; }
+  EndNode* accept() { return accept_; }
+
+  static const int kMaxRecursion = 100;
+  inline int recursion_depth() { return recursion_depth_; }
+  inline void IncrementRecursionDepth() { recursion_depth_++; }
+  inline void DecrementRecursionDepth() { recursion_depth_--; }
+
+  void SetRegExpTooBig() { reg_exp_too_big_ = true; }
+
+  inline bool one_byte() { return one_byte_; }
+  inline bool optimize() { return optimize_; }
+  inline void set_optimize(bool value) { optimize_ = value; }
+  inline bool limiting_recursion() { return limiting_recursion_; }
+  inline void set_limiting_recursion(bool value) {
+    limiting_recursion_ = value;
+  }
+  bool read_backward() { return read_backward_; }
+  void set_read_backward(bool value) { read_backward_ = value; }
+  FrequencyCollator* frequency_collator() { return &frequency_collator_; }
+
+  int current_expansion_factor() { return current_expansion_factor_; }
+  void set_current_expansion_factor(int value) {
+    current_expansion_factor_ = value;
+  }
+
+  Isolate* isolate() const { return isolate_; }
+  Zone* zone() const { return zone_; }
+
+  static const int kNoRegister = -1;
+
+ private:
+  EndNode* accept_;
+  int next_register_;
+  int unicode_lookaround_stack_register_;
+  int unicode_lookaround_position_register_;
+  ZoneVector<RegExpNode*>* work_list_;
+  int recursion_depth_;
+  RegExpMacroAssembler* macro_assembler_;
+  bool one_byte_;
+  bool reg_exp_too_big_;
+  bool limiting_recursion_;
+  bool optimize_;
+  bool read_backward_;
+  int current_expansion_factor_;
+  FrequencyCollator frequency_collator_;
+  Isolate* isolate_;
+  Zone* zone_;
+};
+
+// Categorizes character ranges into BMP, non-BMP, lead, and trail surrogates.
+class UnicodeRangeSplitter {
+ public:
+  V8_EXPORT_PRIVATE UnicodeRangeSplitter(ZoneList<CharacterRange>* base);
+
+  static constexpr int kInitialSize = 8;
+  using CharacterRangeVector = base::SmallVector<CharacterRange, kInitialSize>;
+
+  const CharacterRangeVector* bmp() const { return &bmp_; }
+  const CharacterRangeVector* lead_surrogates() const {
+    return &lead_surrogates_;
+  }
+  const CharacterRangeVector* trail_surrogates() const {
+    return &trail_surrogates_;
+  }
+  const CharacterRangeVector* non_bmp() const { return &non_bmp_; }
+
+ private:
+  void AddRange(CharacterRange range);
+
+  CharacterRangeVector bmp_;
+  CharacterRangeVector lead_surrogates_;
+  CharacterRangeVector trail_surrogates_;
+  CharacterRangeVector non_bmp_;
+};
+
+// We need to check for the following characters: 0x39C 0x3BC 0x178.
+// TODO(jgruber): Move to CharacterRange.
+bool RangeContainsLatin1Equivalents(CharacterRange range);
+
+}  // namespace internal
+}  // namespace v8
+
+#endif  // V8_REGEXP_REGEXP_COMPILER_H_

js/src/regexp/regexp-dotprinter.cc

@@ -0,0 +1,252 @@
+// Copyright 2019 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "regexp/regexp-dotprinter.h"
+
+#include "regexp/regexp-compiler.h"
+
+namespace v8 {
+namespace internal {
+
+// -------------------------------------------------------------------
+// Dot/dotty output
+
+#ifdef DEBUG
+
+class DotPrinterImpl : public NodeVisitor {
+ public:
+  explicit DotPrinterImpl(std::ostream& os) : os_(os) {}
+  void PrintNode(const char* label, RegExpNode* node);
+  void Visit(RegExpNode* node);
+  void PrintAttributes(RegExpNode* from);
+  void PrintOnFailure(RegExpNode* from, RegExpNode* to);
+#define DECLARE_VISIT(Type) virtual void Visit##Type(Type##Node* that);
+  FOR_EACH_NODE_TYPE(DECLARE_VISIT)
+#undef DECLARE_VISIT
+ private:
+  std::ostream& os_;
+};
+
+void DotPrinterImpl::PrintNode(const char* label, RegExpNode* node) {
+  os_ << "digraph G {\n  graph [label=\"";
+  for (int i = 0; label[i]; i++) {
+    switch (label[i]) {
+      case '\\':
+        os_ << "\\\\";
+        break;
+      case '"':
+        os_ << "\"";
+        break;
+      default:
+        os_ << label[i];
+        break;
+    }
+  }
+  os_ << "\"];\n";
+  Visit(node);
+  os_ << "}" << std::endl;
+}
+
+void DotPrinterImpl::Visit(RegExpNode* node) {
+  if (node->info()->visited) return;
+  node->info()->visited = true;
+  node->Accept(this);
+}
+
+void DotPrinterImpl::PrintOnFailure(RegExpNode* from, RegExpNode* on_failure) {
+  os_ << "  n" << from << " -> n" << on_failure << " [style=dotted];\n";
+  Visit(on_failure);
+}
+
+class AttributePrinter {
+ public:
+  explicit AttributePrinter(std::ostream& os)  // NOLINT
+      : os_(os), first_(true) {}
+  void PrintSeparator() {
+    if (first_) {
+      first_ = false;
+    } else {
+      os_ << "|";
+    }
+  }
+  void PrintBit(const char* name, bool value) {
+    if (!value) return;
+    PrintSeparator();
+    os_ << "{" << name << "}";
+  }
+  void PrintPositive(const char* name, int value) {
+    if (value < 0) return;
+    PrintSeparator();
+    os_ << "{" << name << "|" << value << "}";
+  }
+
+ private:
+  std::ostream& os_;
+  bool first_;
+};
+
+void DotPrinterImpl::PrintAttributes(RegExpNode* that) {
+  os_ << "  a" << that << " [shape=Mrecord, color=grey, fontcolor=grey, "
+      << "margin=0.1, fontsize=10, label=\"{";
+  AttributePrinter printer(os_);
+  NodeInfo* info = that->info();
+  printer.PrintBit("NI", info->follows_newline_interest);
+  printer.PrintBit("WI", info->follows_word_interest);
+  printer.PrintBit("SI", info->follows_start_interest);
+  Label* label = that->label();
+  if (label->is_bound()) printer.PrintPositive("@", label->pos());
+  os_ << "}\"];\n"
+      << "  a" << that << " -> n" << that
+      << " [style=dashed, color=grey, arrowhead=none];\n";
+}
+
+void DotPrinterImpl::VisitChoice(ChoiceNode* that) {
+  os_ << "  n" << that << " [shape=Mrecord, label=\"?\"];\n";
+  for (int i = 0; i < that->alternatives()->length(); i++) {
+    GuardedAlternative alt = that->alternatives()->at(i);
+    os_ << "  n" << that << " -> n" << alt.node();
+  }
+  for (int i = 0; i < that->alternatives()->length(); i++) {
+    GuardedAlternative alt = that->alternatives()->at(i);
+    alt.node()->Accept(this);
+  }
+}
+
+void DotPrinterImpl::VisitLoopChoice(LoopChoiceNode* that) {
+  VisitChoice(that);
+}
+
+void DotPrinterImpl::VisitNegativeLookaroundChoice(
+    NegativeLookaroundChoiceNode* that) {
+  VisitChoice(that);
+}
+
+void DotPrinterImpl::VisitText(TextNode* that) {
+  Zone* zone = that->zone();
+  os_ << "  n" << that << " [label=\"";
+  for (int i = 0; i < that->elements()->length(); i++) {
+    if (i > 0) os_ << " ";
+    TextElement elm = that->elements()->at(i);
+    switch (elm.text_type()) {
+      case TextElement::ATOM: {
+        Vector<const uc16> data = elm.atom()->data();
+        for (int i = 0; i < data.length(); i++) {
+          os_ << static_cast<char>(data[i]);
+        }
+        break;
+      }
+      case TextElement::CHAR_CLASS: {
+        RegExpCharacterClass* node = elm.char_class();
+        os_ << "[";
+        if (node->is_negated()) os_ << "^";
+        for (int j = 0; j < node->ranges(zone)->length(); j++) {
+          CharacterRange range = node->ranges(zone)->at(j);
+          os_ << AsUC16(range.from()) << "-" << AsUC16(range.to());
+        }
+        os_ << "]";
+        break;
+      }
+      default:
+        UNREACHABLE();
+    }
+  }
+  os_ << "\", shape=box, peripheries=2];\n";
+  PrintAttributes(that);
+  os_ << "  n" << that << " -> n" << that->on_success() << ";\n";
+  Visit(that->on_success());
+}
+
+void DotPrinterImpl::VisitBackReference(BackReferenceNode* that) {
+  os_ << "  n" << that << " [label=\"$" << that->start_register() << "..$"
+      << that->end_register() << "\", shape=doubleoctagon];\n";
+  PrintAttributes(that);
+  os_ << "  n" << that << " -> n" << that->on_success() << ";\n";
+  Visit(that->on_success());
+}
+
+void DotPrinterImpl::VisitEnd(EndNode* that) {
+  os_ << "  n" << that << " [style=bold, shape=point];\n";
+  PrintAttributes(that);
+}
+
+void DotPrinterImpl::VisitAssertion(AssertionNode* that) {
+  os_ << "  n" << that << " [";
+  switch (that->assertion_type()) {
+    case AssertionNode::AT_END:
+      os_ << "label=\"$\", shape=septagon";
+      break;
+    case AssertionNode::AT_START:
+      os_ << "label=\"^\", shape=septagon";
+      break;
+    case AssertionNode::AT_BOUNDARY:
+      os_ << "label=\"\\b\", shape=septagon";
+      break;
+    case AssertionNode::AT_NON_BOUNDARY:
+      os_ << "label=\"\\B\", shape=septagon";
+      break;
+    case AssertionNode::AFTER_NEWLINE:
+      os_ << "label=\"(?<=\\n)\", shape=septagon";
+      break;
+  }
+  os_ << "];\n";
+  PrintAttributes(that);
+  RegExpNode* successor = that->on_success();
+  os_ << "  n" << that << " -> n" << successor << ";\n";
+  Visit(successor);
+}
+
+void DotPrinterImpl::VisitAction(ActionNode* that) {
+  os_ << "  n" << that << " [";
+  switch (that->action_type_) {
+    case ActionNode::SET_REGISTER_FOR_LOOP:
+      os_ << "label=\"$" << that->data_.u_store_register.reg
+          << ":=" << that->data_.u_store_register.value << "\", shape=octagon";
+      break;
+    case ActionNode::INCREMENT_REGISTER:
+      os_ << "label=\"$" << that->data_.u_increment_register.reg
+          << "++\", shape=octagon";
+      break;
+    case ActionNode::STORE_POSITION:
+      os_ << "label=\"$" << that->data_.u_position_register.reg
+          << ":=$pos\", shape=octagon";
+      break;
+    case ActionNode::BEGIN_SUBMATCH:
+      os_ << "label=\"$" << that->data_.u_submatch.current_position_register
+          << ":=$pos,begin\", shape=septagon";
+      break;
+    case ActionNode::POSITIVE_SUBMATCH_SUCCESS:
+      os_ << "label=\"escape\", shape=septagon";
+      break;
+    case ActionNode::EMPTY_MATCH_CHECK:
+      os_ << "label=\"$" << that->data_.u_empty_match_check.start_register
+          << "=$pos?,$" << that->data_.u_empty_match_check.repetition_register
+          << "<" << that->data_.u_empty_match_check.repetition_limit
+          << "?\", shape=septagon";
+      break;
+    case ActionNode::CLEAR_CAPTURES: {
+      os_ << "label=\"clear $" << that->data_.u_clear_captures.range_from
+          << " to $" << that->data_.u_clear_captures.range_to
+          << "\", shape=septagon";
+      break;
+    }
+  }
+  os_ << "];\n";
+  PrintAttributes(that);
+  RegExpNode* successor = that->on_success();
+  os_ << "  n" << that << " -> n" << successor << ";\n";
+  Visit(successor);
+}
+
+#endif  // DEBUG
+
+void DotPrinter::DotPrint(const char* label, RegExpNode* node) {
+#ifdef DEBUG
+  StdoutStream os;
+  DotPrinterImpl printer(os);
+  printer.PrintNode(label, node);
+#endif  // DEBUG
+}
+
+}  // namespace internal
+}  // namespace v8

js/src/regexp/regexp-dotprinter.h

@@ -0,0 +1,23 @@
+// Copyright 2019 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef V8_REGEXP_REGEXP_DOTPRINTER_H_
+#define V8_REGEXP_REGEXP_DOTPRINTER_H_
+
+#include "regexp/regexp-shim.h"
+
+namespace v8 {
+namespace internal {
+
+class RegExpNode;
+
+class DotPrinter final : public AllStatic {
+ public:
+  static void DotPrint(const char* label, RegExpNode* node);
+};
+
+}  // namespace internal
+}  // namespace v8
+
+#endif  // V8_REGEXP_REGEXP_DOTPRINTER_H_

js/src/regexp/regexp-error.cc

@@ -0,0 +1,22 @@
+// Copyright 2020 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "regexp/regexp-error.h"
+
+namespace v8 {
+namespace internal {
+
+const char* kRegExpErrorStrings[] = {
+#define TEMPLATE(NAME, STRING) STRING,
+    REGEXP_ERROR_MESSAGES(TEMPLATE)
+#undef TEMPLATE
+};
+
+const char* RegExpErrorString(RegExpError error) {
+  DCHECK_LT(error, RegExpError::NumErrors);
+  return kRegExpErrorStrings[static_cast<int>(error)];
+}
+
+}  // namespace internal
+}  // namespace v8

js/src/regexp/regexp-error.h

@@ -0,0 +1,57 @@
+// Copyright 2020 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef V8_REGEXP_REGEXP_ERROR_H_
+#define V8_REGEXP_REGEXP_ERROR_H_
+
+#include "regexp-shim.h"
+
+namespace v8 {
+namespace internal {
+
+#define REGEXP_ERROR_MESSAGES(T)                                          \
+  T(None, "")                                                             \
+  T(StackOverflow, "Maximum call stack size exceeded")                    \
+  T(AnalysisStackOverflow, "Stack overflow")                              \
+  T(TooLarge, "Regular expression too large")                             \
+  T(UnterminatedGroup, "Unterminated group")                              \
+  T(UnmatchedParen, "Unmatched ')'")                                      \
+  T(EscapeAtEndOfPattern, "\\ at end of pattern")                         \
+  T(InvalidPropertyName, "Invalid property name")                         \
+  T(InvalidEscape, "Invalid escape")                                      \
+  T(InvalidDecimalEscape, "Invalid decimal escape")                       \
+  T(InvalidUnicodeEscape, "Invalid Unicode escape")                       \
+  T(NothingToRepeat, "Nothing to repeat")                                 \
+  T(LoneQuantifierBrackets, "Lone quantifier brackets")                   \
+  T(RangeOutOfOrder, "numbers out of order in {} quantifier")             \
+  T(IncompleteQuantifier, "Incomplete quantifier")                        \
+  T(InvalidQuantifier, "Invalid quantifier")                              \
+  T(InvalidGroup, "Invalid group")                                        \
+  T(MultipleFlagDashes, "Multiple dashes in flag group")                  \
+  T(RepeatedFlag, "Repeated flag in flag group")                          \
+  T(InvalidFlagGroup, "Invalid flag group")                               \
+  T(TooManyCaptures, "Too many captures")                                 \
+  T(InvalidCaptureGroupName, "Invalid capture group name")                \
+  T(DuplicateCaptureGroupName, "Duplicate capture group name")            \
+  T(InvalidNamedReference, "Invalid named reference")                     \
+  T(InvalidNamedCaptureReference, "Invalid named capture referenced")     \
+  T(InvalidClassEscape, "Invalid class escape")                           \
+  T(InvalidClassPropertyName, "Invalid property name in character class") \
+  T(InvalidCharacterClass, "Invalid character class")                     \
+  T(UnterminatedCharacterClass, "Unterminated character class")           \
+  T(OutOfOrderCharacterClass, "Range out of order in character class")
+
+enum class RegExpError : uint32_t {
+#define TEMPLATE(NAME, STRING) k##NAME,
+  REGEXP_ERROR_MESSAGES(TEMPLATE)
+#undef TEMPLATE
+      NumErrors
+};
+
+V8_EXPORT_PRIVATE const char* RegExpErrorString(RegExpError error);
+
+}  // namespace internal
+}  // namespace v8
+
+#endif  // V8_REGEXP_REGEXP_ERROR_H_

js/src/regexp/regexp-interpreter.h

@@ -0,0 +1,61 @@
+// Copyright 2011 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// A simple interpreter for the Irregexp byte code.
+
+#ifndef V8_REGEXP_REGEXP_INTERPRETER_H_
+#define V8_REGEXP_REGEXP_INTERPRETER_H_
+
+#include "regexp/regexp.h"
+
+namespace v8 {
+namespace internal {
+
+class V8_EXPORT_PRIVATE IrregexpInterpreter : public AllStatic {
+ public:
+  enum Result {
+    FAILURE = RegExp::kInternalRegExpFailure,
+    SUCCESS = RegExp::kInternalRegExpSuccess,
+    EXCEPTION = RegExp::kInternalRegExpException,
+    RETRY = RegExp::kInternalRegExpRetry,
+  };
+
+  // In case a StackOverflow occurs, a StackOverflowException is created and
+  // EXCEPTION is returned.
+  static Result MatchForCallFromRuntime(Isolate* isolate,
+                                        Handle<JSRegExp> regexp,
+                                        Handle<String> subject_string,
+                                        int* registers, int registers_length,
+                                        int start_position);
+
+  // In case a StackOverflow occurs, EXCEPTION is returned. The caller is
+  // responsible for creating the exception.
+  // RETRY is returned if a retry through the runtime is needed (e.g. when
+  // interrupts have been scheduled or the regexp is marked for tier-up).
+  // Arguments input_start, input_end and backtrack_stack are
+  // unused. They are only passed to match the signature of the native irregex
+  // code.
+  static Result MatchForCallFromJs(Address subject, int32_t start_position,
+                                   Address input_start, Address input_end,
+                                   int* registers, int32_t registers_length,
+                                   Address backtrack_stack,
+                                   RegExp::CallOrigin call_origin,
+                                   Isolate* isolate, Address regexp);
+
+  static Result MatchInternal(Isolate* isolate, ByteArray code_array,
+                              String subject_string, int* registers,
+                              int registers_length, int start_position,
+                              RegExp::CallOrigin call_origin,
+                              uint32_t backtrack_limit);
+
+ private:
+  static Result Match(Isolate* isolate, JSRegExp regexp, String subject_string,
+                      int* registers, int registers_length, int start_position,
+                      RegExp::CallOrigin call_origin);
+};
+
+}  // namespace internal
+}  // namespace v8
+
+#endif  // V8_REGEXP_REGEXP_INTERPRETER_H_

js/src/regexp/regexp-macro-assembler-arch.h

@@ -0,0 +1,291 @@
+/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*-
+ * vim: set ts=8 sts=2 et sw=2 tw=80:
+ * This Source Code Form is subject to the terms of the Mozilla Public
+ * License, v. 2.0. If a copy of the MPL was not distributed with this
+ * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
+
+// Copyright 2020 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// This file implements the NativeRegExpMacroAssembler interface for
+// SpiderMonkey. It provides the same interface as each of V8's
+// architecture-specific implementations.
+
+#ifndef RegexpMacroAssemblerArch_h
+#define RegexpMacroAssemblerArch_h
+
+#include "jit/MacroAssembler.h"
+#include "regexp/regexp-macro-assembler.h"
+
+namespace v8 {
+namespace internal {
+
+struct FrameData {
+  // Character position at the start of the input, stored as a
+  // negative offset from the end of the string (input_end_pointer_).
+  size_t inputStart;
+
+  // The backtrack_stack_pointer_ register points to the top of the stack.
+  // This points to the bottom of the backtrack stack.
+  void* backtrackStackBase;
+
+  // Copy of the input MatchPairs.
+  int32_t* matches;   // pointer to capture array
+  int32_t numMatches; // size of capture array
+};
+
+class SMRegExpMacroAssembler final : public NativeRegExpMacroAssembler {
+ public:
+  SMRegExpMacroAssembler(JSContext* cx, Isolate* isolate,
+                         js::jit::StackMacroAssembler& masm, Zone* zone,
+                         Mode mode, uint32_t num_capture_registers);
+  virtual ~SMRegExpMacroAssembler() {} // Nothing to do here
+
+  virtual int stack_limit_slack();
+  virtual IrregexpImplementation Implementation();
+
+  virtual bool Succeed();
+  virtual void Fail();
+
+  virtual void AdvanceCurrentPosition(int by);
+  virtual void PopCurrentPosition();
+  virtual void PushCurrentPosition();
+  virtual void SetCurrentPositionFromEnd(int by);
+
+  virtual void Backtrack();
+  virtual void Bind(Label* label);
+  virtual void GoTo(Label* label);
+  virtual void PushBacktrack(Label* label);
+
+  virtual void CheckCharacter(uint32_t c, Label* on_equal);
+  virtual void CheckNotCharacter(uint32_t c, Label* on_not_equal);
+  virtual void CheckCharacterGT(uc16 limit, Label* on_greater);
+  virtual void CheckCharacterLT(uc16 limit, Label* on_less);
+  virtual void CheckCharacterAfterAnd(uint32_t c, uint32_t mask,
+                                      Label* on_equal);
+  virtual void CheckNotCharacterAfterAnd(uint32_t c, uint32_t mask,
+                                         Label* on_not_equal);
+  virtual void CheckNotCharacterAfterMinusAnd(uc16 c, uc16 minus, uc16 mask,
+                                              Label* on_not_equal);
+  virtual void CheckGreedyLoop(Label* on_tos_equals_current_position);
+  virtual void CheckCharacterInRange(uc16 from, uc16 to, Label* on_in_range);
+  virtual void CheckCharacterNotInRange(uc16 from, uc16 to,
+                                        Label* on_not_in_range);
+  virtual void CheckAtStart(int cp_offset, Label* on_at_start);
+  virtual void CheckNotAtStart(int cp_offset, Label* on_not_at_start);
+  virtual void CheckPosition(int cp_offset, Label* on_outside_input);
+  virtual void CheckBitInTable(Handle<ByteArray> table, Label* on_bit_set);
+  virtual bool CheckSpecialCharacterClass(uc16 type, Label* on_no_match);
+  virtual void CheckNotBackReference(int start_reg, bool read_backward,
+                                     Label* on_no_match);
+  virtual void CheckNotBackReferenceIgnoreCase(int start_reg,
+                                               bool read_backward,
+                                               Label* on_no_match);
+
+  virtual void LoadCurrentCharacterImpl(int cp_offset, Label* on_end_of_input,
+                                        bool check_bounds, int characters,
+                                        int eats_at_least);
+
+  virtual void AdvanceRegister(int reg, int by);
+  virtual void IfRegisterGE(int reg, int comparand, Label* if_ge);
+  virtual void IfRegisterLT(int reg, int comparand, Label* if_lt);
+  virtual void IfRegisterEqPos(int reg, Label* if_eq);
+  virtual void PopRegister(int register_index);
+  virtual void PushRegister(int register_index,
+                            StackCheckFlag check_stack_limit);
+  virtual void ReadCurrentPositionFromRegister(int reg);
+  virtual void WriteCurrentPositionToRegister(int reg, int cp_offset);
+  virtual void ReadStackPointerFromRegister(int reg);
+  virtual void WriteStackPointerToRegister(int reg);
+  virtual void SetRegister(int register_index, int to);
+  virtual void ClearRegisters(int reg_from, int reg_to);
+
+  virtual Handle<HeapObject> GetCode(Handle<String> source);
+
+ private:
+  size_t frameSize_ = 0;
+
+  void createStackFrame();
+  void initFrameAndRegs();
+  void successHandler();
+  void exitHandler();
+  void backtrackHandler();
+  void stackOverflowHandler();
+
+  // Push a register on the backtrack stack.
+  void Push(js::jit::Register value);
+
+  // Pop a value from the backtrack stack.
+  void Pop(js::jit::Register target);
+
+  void CheckAtStartImpl(int cp_offset, Label* on_cond,
+                        js::jit::Assembler::Condition cond);
+  void CheckCharacterImpl(js::jit::Imm32 c, Label* on_cond,
+                          js::jit::Assembler::Condition cond);
+  void CheckCharacterAfterAndImpl(uint32_t c, uint32_t and_with, Label* on_cond,
+                                  bool negate);
+  void CheckCharacterInRangeImpl(uc16 from, uc16 to, Label* on_cond,
+                                 js::jit::Assembler::Condition cond);
+  void CheckNotBackReferenceImpl(int start_reg, bool read_backward,
+                                 Label* on_no_match, bool ignore_case);
+
+  void LoadCurrentCharacterUnchecked(int cp_offset, int characters);
+
+  void JumpOrBacktrack(Label* to);
+
+  // MacroAssembler methods that take a Label can be called with a
+  // null label, which means that we should backtrack if we would jump
+  // to that label. This is a helper to avoid writing out the same
+  // logic a dozen times.
+  inline js::jit::Label* LabelOrBacktrack(Label* to) {
+    return to ? to->inner() : &backtrack_label_;
+  }
+
+  void CheckBacktrackStackLimit();
+
+  static bool GrowBacktrackStack(RegExpStack* regexp_stack);
+
+  static uint32_t CaseInsensitiveCompareStrings(const char16_t* substring1,
+                                                const char16_t* substring2,
+                                                size_t byteLength);
+  static uint32_t CaseInsensitiveCompareUCStrings(const char16_t* substring1,
+                                                  const char16_t* substring2,
+                                                  size_t byteLength);
+
+  inline int char_size() { return static_cast<int>(mode_); }
+  inline js::jit::Scale factor() {
+    return mode_ == UC16 ? js::jit::TimesTwo : js::jit::TimesOne;
+  }
+
+  js::jit::Address inputStart() {
+    return js::jit::Address(masm_.getStackPointer(),
+                            offsetof(FrameData, inputStart));
+  }
+  js::jit::Address backtrackStackBase() {
+    return js::jit::Address(masm_.getStackPointer(),
+                            offsetof(FrameData, backtrackStackBase));
+  }
+  js::jit::Address matches() {
+    return js::jit::Address(masm_.getStackPointer(),
+                            offsetof(FrameData, matches));
+  }
+  js::jit::Address numMatches() {
+    return js::jit::Address(masm_.getStackPointer(),
+                            offsetof(FrameData, numMatches));
+  }
+
+  // The stack-pointer-relative location of a regexp register.
+  js::jit::Address register_location(int register_index) {
+    return js::jit::Address(masm_.getStackPointer(),
+                            register_offset(register_index));
+  }
+
+  int32_t register_offset(int register_index) {
+    MOZ_ASSERT(register_index >= 0 && register_index <= kMaxRegister);
+    if (num_registers_ <= register_index) {
+      num_registers_ = register_index + 1;
+    }
+    static_assert(alignof(uintptr_t) <= alignof(FrameData));
+    return sizeof(FrameData) + register_index * sizeof(uintptr_t*);
+  }
+
+  JSContext* cx_;
+  js::jit::StackMacroAssembler& masm_;
+
+  /*
+   * This assembler uses the following registers:
+   *
+   * - current_character_:
+   *     Contains the character (or characters) currently being examined.
+   *     Must be loaded using LoadCurrentCharacter before using any of the
+   *     dispatch methods. After a matching pass for a global regexp,
+   *     temporarily stores the index of capture start.
+   * - current_position_:
+   *     Current position in input *as negative byte offset from end of string*.
+   * - input_end_pointer_:
+   *     Points to byte after last character in the input. current_position_ is
+   *     relative to this.
+   * - backtrack_stack_pointer_:
+   *     Points to tip of the (heap-allocated) backtrack stack. The stack grows
+   *     downward (like the native stack).
+   * - temp0_, temp1_, temp2_:
+   *     Scratch registers.
+   *
+   * The native stack pointer is used to access arguments (InputOutputData),
+   * local variables (FrameData), and irregexp's internal virtual registers
+   * (see register_location).
+   */
+
+  js::jit::Register current_character_;
+  js::jit::Register current_position_;
+  js::jit::Register input_end_pointer_;
+  js::jit::Register backtrack_stack_pointer_;
+  js::jit::Register temp0_, temp1_, temp2_;
+
+  js::jit::Label entry_label_;
+  js::jit::Label start_label_;
+  js::jit::Label backtrack_label_;
+  js::jit::Label success_label_;
+  js::jit::Label exit_label_;
+  js::jit::Label stack_overflow_label_;
+  js::jit::Label exit_with_exception_label_;
+
+  // When we generate the code to push a backtrack label's address
+  // onto the backtrack stack, we don't know its final address. We
+  // have to patch it after linking. This is slightly delicate, as the
+  // Label itself (which is allocated on the stack) may not exist by
+  // the time we link. The approach is as follows:
+  //
+  // 1. When we push a label on the backtrack stack (PushBacktrack),
+  //    we bind the label's patchOffset_ field to the offset within
+  //    the code that should be overwritten. This works because each
+  //    label is only pushed by a single instruction.
+  //
+  // 2. When we bind a label (Bind), we check to see if it has a
+  //    bound patchOffset_. If it does, we create a LabelPatch mapping
+  //    its patch offset to the offset of the label itself.
+  //
+  // 3. While linking the code, we walk the list of label patches
+  //    and patch the code accordingly.
+  class LabelPatch {
+   public:
+    LabelPatch(js::jit::CodeOffset patchOffset, size_t labelOffset)
+        : patchOffset_(patchOffset), labelOffset_(labelOffset) {}
+
+    js::jit::CodeOffset patchOffset_;
+    size_t labelOffset_ = 0;
+  };
+
+  js::Vector<LabelPatch, 4, js::SystemAllocPolicy> labelPatches_;
+  void AddLabelPatch(js::jit::CodeOffset patchOffset, size_t labelOffset) {
+    js::AutoEnterOOMUnsafeRegion oomUnsafe;
+    if (!labelPatches_.emplaceBack(patchOffset, labelOffset)) {
+      oomUnsafe.crash("Irregexp label patch");
+    }
+  }
+
+  Mode mode_;
+  int num_registers_;
+  int num_capture_registers_;
+  js::jit::LiveGeneralRegisterSet savedRegisters_;
+
+ public:
+  using TableVector =
+      js::Vector<PseudoHandle<ByteArrayData>, 4, js::SystemAllocPolicy>;
+  TableVector& tables() { return tables_; }
+
+ private:
+  TableVector tables_;
+  void AddTable(PseudoHandle<ByteArrayData> table) {
+    js::AutoEnterOOMUnsafeRegion oomUnsafe;
+    if (!tables_.append(std::move(table))) {
+      oomUnsafe.crash("Irregexp table append");
+    }
+  }
+};
+
+}  // namespace internal
+}  // namespace v8
+
+#endif  // RegexpMacroAssemblerArch_h

js/src/regexp/regexp-macro-assembler-tracer.cc

@@ -0,0 +1,418 @@
+// Copyright 2012 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "regexp/regexp-macro-assembler-tracer.h"
+
+
+namespace v8 {
+namespace internal {
+
+RegExpMacroAssemblerTracer::RegExpMacroAssemblerTracer(
+    Isolate* isolate, RegExpMacroAssembler* assembler)
+    : RegExpMacroAssembler(isolate, assembler->zone()), assembler_(assembler) {
+  IrregexpImplementation type = assembler->Implementation();
+  DCHECK_LT(type, 9);
+  const char* impl_names[] = {"IA32", "ARM", "ARM64", "MIPS",    "S390",
+                              "PPC",  "X64", "X87",   "Bytecode"};
+  PrintF("RegExpMacroAssembler%s();\n", impl_names[type]);
+}
+
+RegExpMacroAssemblerTracer::~RegExpMacroAssemblerTracer() = default;
+
+void RegExpMacroAssemblerTracer::AbortedCodeGeneration() {
+  PrintF(" AbortedCodeGeneration\n");
+  assembler_->AbortedCodeGeneration();
+}
+
+
+// This is used for printing out debugging information.  It makes an integer
+// that is closely related to the address of an object.
+static int LabelToInt(Label* label) {
+  return static_cast<int>(reinterpret_cast<intptr_t>(label));
+}
+
+
+void RegExpMacroAssemblerTracer::Bind(Label* label) {
+  PrintF("label[%08x]: (Bind)\n", LabelToInt(label));
+  assembler_->Bind(label);
+}
+
+
+void RegExpMacroAssemblerTracer::AdvanceCurrentPosition(int by) {
+  PrintF(" AdvanceCurrentPosition(by=%d);\n", by);
+  assembler_->AdvanceCurrentPosition(by);
+}
+
+
+void RegExpMacroAssemblerTracer::CheckGreedyLoop(Label* label) {
+  PrintF(" CheckGreedyLoop(label[%08x]);\n\n", LabelToInt(label));
+  assembler_->CheckGreedyLoop(label);
+}
+
+
+void RegExpMacroAssemblerTracer::PopCurrentPosition() {
+  PrintF(" PopCurrentPosition();\n");
+  assembler_->PopCurrentPosition();
+}
+
+
+void RegExpMacroAssemblerTracer::PushCurrentPosition() {
+  PrintF(" PushCurrentPosition();\n");
+  assembler_->PushCurrentPosition();
+}
+
+
+void RegExpMacroAssemblerTracer::Backtrack() {
+  PrintF(" Backtrack();\n");
+  assembler_->Backtrack();
+}
+
+
+void RegExpMacroAssemblerTracer::GoTo(Label* label) {
+  PrintF(" GoTo(label[%08x]);\n\n", LabelToInt(label));
+  assembler_->GoTo(label);
+}
+
+
+void RegExpMacroAssemblerTracer::PushBacktrack(Label* label) {
+  PrintF(" PushBacktrack(label[%08x]);\n", LabelToInt(label));
+  assembler_->PushBacktrack(label);
+}
+
+
+bool RegExpMacroAssemblerTracer::Succeed() {
+  bool restart = assembler_->Succeed();
+  PrintF(" Succeed();%s\n", restart ? " [restart for global match]" : "");
+  return restart;
+}
+
+
+void RegExpMacroAssemblerTracer::Fail() {
+  PrintF(" Fail();");
+  assembler_->Fail();
+}
+
+
+void RegExpMacroAssemblerTracer::PopRegister(int register_index) {
+  PrintF(" PopRegister(register=%d);\n", register_index);
+  assembler_->PopRegister(register_index);
+}
+
+
+void RegExpMacroAssemblerTracer::PushRegister(
+    int register_index,
+    StackCheckFlag check_stack_limit) {
+  PrintF(" PushRegister(register=%d, %s);\n",
+         register_index,
+         check_stack_limit ? "check stack limit" : "");
+  assembler_->PushRegister(register_index, check_stack_limit);
+}
+
+
+void RegExpMacroAssemblerTracer::AdvanceRegister(int reg, int by) {
+  PrintF(" AdvanceRegister(register=%d, by=%d);\n", reg, by);
+  assembler_->AdvanceRegister(reg, by);
+}
+
+
+void RegExpMacroAssemblerTracer::SetCurrentPositionFromEnd(int by) {
+  PrintF(" SetCurrentPositionFromEnd(by=%d);\n", by);
+  assembler_->SetCurrentPositionFromEnd(by);
+}
+
+
+void RegExpMacroAssemblerTracer::SetRegister(int register_index, int to) {
+  PrintF(" SetRegister(register=%d, to=%d);\n", register_index, to);
+  assembler_->SetRegister(register_index, to);
+}
+
+
+void RegExpMacroAssemblerTracer::WriteCurrentPositionToRegister(int reg,
+                                                                int cp_offset) {
+  PrintF(" WriteCurrentPositionToRegister(register=%d,cp_offset=%d);\n",
+         reg,
+         cp_offset);
+  assembler_->WriteCurrentPositionToRegister(reg, cp_offset);
+}
+
+
+void RegExpMacroAssemblerTracer::ClearRegisters(int reg_from, int reg_to) {
+  PrintF(" ClearRegister(from=%d, to=%d);\n", reg_from, reg_to);
+  assembler_->ClearRegisters(reg_from, reg_to);
+}
+
+
+void RegExpMacroAssemblerTracer::ReadCurrentPositionFromRegister(int reg) {
+  PrintF(" ReadCurrentPositionFromRegister(register=%d);\n", reg);
+  assembler_->ReadCurrentPositionFromRegister(reg);
+}
+
+
+void RegExpMacroAssemblerTracer::WriteStackPointerToRegister(int reg) {
+  PrintF(" WriteStackPointerToRegister(register=%d);\n", reg);
+  assembler_->WriteStackPointerToRegister(reg);
+}
+
+
+void RegExpMacroAssemblerTracer::ReadStackPointerFromRegister(int reg) {
+  PrintF(" ReadStackPointerFromRegister(register=%d);\n", reg);
+  assembler_->ReadStackPointerFromRegister(reg);
+}
+
+void RegExpMacroAssemblerTracer::LoadCurrentCharacterImpl(
+    int cp_offset, Label* on_end_of_input, bool check_bounds, int characters,
+    int eats_at_least) {
+  const char* check_msg = check_bounds ? "" : " (unchecked)";
+  PrintF(
+      " LoadCurrentCharacter(cp_offset=%d, label[%08x]%s (%d chars) (eats at "
+      "least %d));\n",
+      cp_offset, LabelToInt(on_end_of_input), check_msg, characters,
+      eats_at_least);
+  assembler_->LoadCurrentCharacter(cp_offset, on_end_of_input, check_bounds,
+                                   characters, eats_at_least);
+}
+
+class PrintablePrinter {
+ public:
+  explicit PrintablePrinter(uc16 character) : character_(character) { }
+
+  const char* operator*() {
+    if (character_ >= ' ' && character_ <= '~') {
+      buffer_[0] = '(';
+      buffer_[1] = static_cast<char>(character_);
+      buffer_[2] = ')';
+      buffer_[3] = '\0';
+    } else {
+      buffer_[0] = '\0';
+    }
+    return &buffer_[0];
+  }
+
+ private:
+  uc16 character_;
+  char buffer_[4];
+};
+
+
+void RegExpMacroAssemblerTracer::CheckCharacterLT(uc16 limit, Label* on_less) {
+  PrintablePrinter printable(limit);
+  PrintF(" CheckCharacterLT(c=0x%04x%s, label[%08x]);\n",
+         limit,
+         *printable,
+         LabelToInt(on_less));
+  assembler_->CheckCharacterLT(limit, on_less);
+}
+
+
+void RegExpMacroAssemblerTracer::CheckCharacterGT(uc16 limit,
+                                                  Label* on_greater) {
+  PrintablePrinter printable(limit);
+  PrintF(" CheckCharacterGT(c=0x%04x%s, label[%08x]);\n",
+         limit,
+         *printable,
+         LabelToInt(on_greater));
+  assembler_->CheckCharacterGT(limit, on_greater);
+}
+
+
+void RegExpMacroAssemblerTracer::CheckCharacter(unsigned c, Label* on_equal) {
+  PrintablePrinter printable(c);
+  PrintF(" CheckCharacter(c=0x%04x%s, label[%08x]);\n",
+         c,
+         *printable,
+         LabelToInt(on_equal));
+  assembler_->CheckCharacter(c, on_equal);
+}
+
+void RegExpMacroAssemblerTracer::CheckAtStart(int cp_offset,
+                                              Label* on_at_start) {
+  PrintF(" CheckAtStart(cp_offset=%d, label[%08x]);\n", cp_offset,
+         LabelToInt(on_at_start));
+  assembler_->CheckAtStart(cp_offset, on_at_start);
+}
+
+void RegExpMacroAssemblerTracer::CheckNotAtStart(int cp_offset,
+                                                 Label* on_not_at_start) {
+  PrintF(" CheckNotAtStart(cp_offset=%d, label[%08x]);\n", cp_offset,
+         LabelToInt(on_not_at_start));
+  assembler_->CheckNotAtStart(cp_offset, on_not_at_start);
+}
+
+
+void RegExpMacroAssemblerTracer::CheckNotCharacter(unsigned c,
+                                                   Label* on_not_equal) {
+  PrintablePrinter printable(c);
+  PrintF(" CheckNotCharacter(c=0x%04x%s, label[%08x]);\n",
+         c,
+         *printable,
+         LabelToInt(on_not_equal));
+  assembler_->CheckNotCharacter(c, on_not_equal);
+}
+
+
+void RegExpMacroAssemblerTracer::CheckCharacterAfterAnd(
+    unsigned c,
+    unsigned mask,
+    Label* on_equal) {
+  PrintablePrinter printable(c);
+  PrintF(" CheckCharacterAfterAnd(c=0x%04x%s, mask=0x%04x, label[%08x]);\n",
+         c,
+         *printable,
+         mask,
+         LabelToInt(on_equal));
+  assembler_->CheckCharacterAfterAnd(c, mask, on_equal);
+}
+
+
+void RegExpMacroAssemblerTracer::CheckNotCharacterAfterAnd(
+    unsigned c,
+    unsigned mask,
+    Label* on_not_equal) {
+  PrintablePrinter printable(c);
+  PrintF(" CheckNotCharacterAfterAnd(c=0x%04x%s, mask=0x%04x, label[%08x]);\n",
+         c,
+         *printable,
+         mask,
+         LabelToInt(on_not_equal));
+  assembler_->CheckNotCharacterAfterAnd(c, mask, on_not_equal);
+}
+
+
+void RegExpMacroAssemblerTracer::CheckNotCharacterAfterMinusAnd(
+    uc16 c,
+    uc16 minus,
+    uc16 mask,
+    Label* on_not_equal) {
+  PrintF(" CheckNotCharacterAfterMinusAnd(c=0x%04x, minus=%04x, mask=0x%04x, "
+             "label[%08x]);\n",
+         c,
+         minus,
+         mask,
+         LabelToInt(on_not_equal));
+  assembler_->CheckNotCharacterAfterMinusAnd(c, minus, mask, on_not_equal);
+}
+
+
+void RegExpMacroAssemblerTracer::CheckCharacterInRange(
+    uc16 from,
+    uc16 to,
+    Label* on_not_in_range) {
+  PrintablePrinter printable_from(from);
+  PrintablePrinter printable_to(to);
+  PrintF(" CheckCharacterInRange(from=0x%04x%s, to=0x%04x%s, label[%08x]);\n",
+         from,
+         *printable_from,
+         to,
+         *printable_to,
+         LabelToInt(on_not_in_range));
+  assembler_->CheckCharacterInRange(from, to, on_not_in_range);
+}
+
+
+void RegExpMacroAssemblerTracer::CheckCharacterNotInRange(
+    uc16 from,
+    uc16 to,
+    Label* on_in_range) {
+  PrintablePrinter printable_from(from);
+  PrintablePrinter printable_to(to);
+  PrintF(
+      " CheckCharacterNotInRange(from=0x%04x%s," " to=%04x%s, label[%08x]);\n",
+      from,
+      *printable_from,
+      to,
+      *printable_to,
+      LabelToInt(on_in_range));
+  assembler_->CheckCharacterNotInRange(from, to, on_in_range);
+}
+
+
+void RegExpMacroAssemblerTracer::CheckBitInTable(
+    Handle<ByteArray> table, Label* on_bit_set) {
+  PrintF(" CheckBitInTable(label[%08x] ", LabelToInt(on_bit_set));
+  for (int i = 0; i < kTableSize; i++) {
+    PrintF("%c", table->get(i) != 0 ? 'X' : '.');
+    if (i % 32 == 31 && i != kTableMask) {
+      PrintF("\n                                 ");
+    }
+  }
+  PrintF(");\n");
+  assembler_->CheckBitInTable(table, on_bit_set);
+}
+
+
+void RegExpMacroAssemblerTracer::CheckNotBackReference(int start_reg,
+                                                       bool read_backward,
+                                                       Label* on_no_match) {
+  PrintF(" CheckNotBackReference(register=%d, %s, label[%08x]);\n", start_reg,
+         read_backward ? "backward" : "forward", LabelToInt(on_no_match));
+  assembler_->CheckNotBackReference(start_reg, read_backward, on_no_match);
+}
+
+void RegExpMacroAssemblerTracer::CheckNotBackReferenceIgnoreCase(
+    int start_reg, bool read_backward, Label* on_no_match) {
+  PrintF(" CheckNotBackReferenceIgnoreCase(register=%d, %s, label[%08x]);\n",
+         start_reg, read_backward ? "backward" : "forward",
+         LabelToInt(on_no_match));
+  assembler_->CheckNotBackReferenceIgnoreCase(start_reg, read_backward,
+                                              on_no_match);
+}
+
+void RegExpMacroAssemblerTracer::CheckPosition(int cp_offset,
+                                               Label* on_outside_input) {
+  PrintF(" CheckPosition(cp_offset=%d, label[%08x]);\n", cp_offset,
+         LabelToInt(on_outside_input));
+  assembler_->CheckPosition(cp_offset, on_outside_input);
+}
+
+
+bool RegExpMacroAssemblerTracer::CheckSpecialCharacterClass(
+    uc16 type,
+    Label* on_no_match) {
+  bool supported = assembler_->CheckSpecialCharacterClass(type,
+                                                          on_no_match);
+  PrintF(" CheckSpecialCharacterClass(type='%c', label[%08x]): %s;\n",
+         type,
+         LabelToInt(on_no_match),
+         supported ? "true" : "false");
+  return supported;
+}
+
+
+void RegExpMacroAssemblerTracer::IfRegisterLT(int register_index,
+                                              int comparand, Label* if_lt) {
+  PrintF(" IfRegisterLT(register=%d, number=%d, label[%08x]);\n",
+         register_index, comparand, LabelToInt(if_lt));
+  assembler_->IfRegisterLT(register_index, comparand, if_lt);
+}
+
+
+void RegExpMacroAssemblerTracer::IfRegisterEqPos(int register_index,
+                                                 Label* if_eq) {
+  PrintF(" IfRegisterEqPos(register=%d, label[%08x]);\n",
+         register_index, LabelToInt(if_eq));
+  assembler_->IfRegisterEqPos(register_index, if_eq);
+}
+
+
+void RegExpMacroAssemblerTracer::IfRegisterGE(int register_index,
+                                              int comparand, Label* if_ge) {
+  PrintF(" IfRegisterGE(register=%d, number=%d, label[%08x]);\n",
+         register_index, comparand, LabelToInt(if_ge));
+  assembler_->IfRegisterGE(register_index, comparand, if_ge);
+}
+
+
+RegExpMacroAssembler::IrregexpImplementation
+    RegExpMacroAssemblerTracer::Implementation() {
+  return assembler_->Implementation();
+}
+
+
+Handle<HeapObject> RegExpMacroAssemblerTracer::GetCode(Handle<String> source) {
+  PrintF(" GetCode(%s);\n", source->ToCString().get());
+  return assembler_->GetCode(source);
+}
+
+}  // namespace internal
+}  // namespace v8

js/src/regexp/regexp-macro-assembler-tracer.h

@@ -0,0 +1,80 @@
+// Copyright 2008 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef V8_REGEXP_REGEXP_MACRO_ASSEMBLER_TRACER_H_
+#define V8_REGEXP_REGEXP_MACRO_ASSEMBLER_TRACER_H_
+
+#include "regexp/regexp-macro-assembler.h"
+
+namespace v8 {
+namespace internal {
+
+// Decorator on a RegExpMacroAssembler that write all calls.
+class RegExpMacroAssemblerTracer: public RegExpMacroAssembler {
+ public:
+  RegExpMacroAssemblerTracer(Isolate* isolate, RegExpMacroAssembler* assembler);
+  ~RegExpMacroAssemblerTracer() override;
+  void AbortedCodeGeneration() override;
+  int stack_limit_slack() override { return assembler_->stack_limit_slack(); }
+  bool CanReadUnaligned() override { return assembler_->CanReadUnaligned(); }
+  void AdvanceCurrentPosition(int by) override;    // Signed cp change.
+  void AdvanceRegister(int reg, int by) override;  // r[reg] += by.
+  void Backtrack() override;
+  void Bind(Label* label) override;
+  void CheckCharacter(unsigned c, Label* on_equal) override;
+  void CheckCharacterAfterAnd(unsigned c, unsigned and_with,
+                              Label* on_equal) override;
+  void CheckCharacterGT(uc16 limit, Label* on_greater) override;
+  void CheckCharacterLT(uc16 limit, Label* on_less) override;
+  void CheckGreedyLoop(Label* on_tos_equals_current_position) override;
+  void CheckAtStart(int cp_offset, Label* on_at_start) override;
+  void CheckNotAtStart(int cp_offset, Label* on_not_at_start) override;
+  void CheckNotBackReference(int start_reg, bool read_backward,
+                             Label* on_no_match) override;
+  void CheckNotBackReferenceIgnoreCase(int start_reg, bool read_backward,
+                                       Label* on_no_match) override;
+  void CheckNotCharacter(unsigned c, Label* on_not_equal) override;
+  void CheckNotCharacterAfterAnd(unsigned c, unsigned and_with,
+                                 Label* on_not_equal) override;
+  void CheckNotCharacterAfterMinusAnd(uc16 c, uc16 minus, uc16 and_with,
+                                      Label* on_not_equal) override;
+  void CheckCharacterInRange(uc16 from, uc16 to, Label* on_in_range) override;
+  void CheckCharacterNotInRange(uc16 from, uc16 to,
+                                Label* on_not_in_range) override;
+  void CheckBitInTable(Handle<ByteArray> table, Label* on_bit_set) override;
+  void CheckPosition(int cp_offset, Label* on_outside_input) override;
+  bool CheckSpecialCharacterClass(uc16 type, Label* on_no_match) override;
+  void Fail() override;
+  Handle<HeapObject> GetCode(Handle<String> source) override;
+  void GoTo(Label* label) override;
+  void IfRegisterGE(int reg, int comparand, Label* if_ge) override;
+  void IfRegisterLT(int reg, int comparand, Label* if_lt) override;
+  void IfRegisterEqPos(int reg, Label* if_eq) override;
+  IrregexpImplementation Implementation() override;
+  void LoadCurrentCharacterImpl(int cp_offset, Label* on_end_of_input,
+                                bool check_bounds, int characters,
+                                int eats_at_least) override;
+  void PopCurrentPosition() override;
+  void PopRegister(int register_index) override;
+  void PushBacktrack(Label* label) override;
+  void PushCurrentPosition() override;
+  void PushRegister(int register_index,
+                    StackCheckFlag check_stack_limit) override;
+  void ReadCurrentPositionFromRegister(int reg) override;
+  void ReadStackPointerFromRegister(int reg) override;
+  void SetCurrentPositionFromEnd(int by) override;
+  void SetRegister(int register_index, int to) override;
+  bool Succeed() override;
+  void WriteCurrentPositionToRegister(int reg, int cp_offset) override;
+  void ClearRegisters(int reg_from, int reg_to) override;
+  void WriteStackPointerToRegister(int reg) override;
+
+ private:
+  RegExpMacroAssembler* assembler_;
+};
+
+}  // namespace internal
+}  // namespace v8
+
+#endif  // V8_REGEXP_REGEXP_MACRO_ASSEMBLER_TRACER_H_

js/src/regexp/regexp-macro-assembler.cc

@@ -0,0 +1,344 @@
+// Copyright 2012 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "regexp/regexp-macro-assembler.h"
+
+#include "regexp/regexp-stack.h"
+
+#ifdef V8_INTL_SUPPORT
+#include "unicode/uchar.h"
+#include "unicode/unistr.h"
+#endif  // V8_INTL_SUPPORT
+
+namespace v8 {
+namespace internal {
+
+RegExpMacroAssembler::RegExpMacroAssembler(Isolate* isolate, Zone* zone)
+    : slow_safe_compiler_(false),
+      global_mode_(NOT_GLOBAL),
+      isolate_(isolate),
+      zone_(zone) {}
+
+RegExpMacroAssembler::~RegExpMacroAssembler() = default;
+
+int RegExpMacroAssembler::CaseInsensitiveCompareUC16(Address byte_offset1,
+                                                     Address byte_offset2,
+                                                     size_t byte_length,
+                                                     Isolate* isolate) {
+  // This function is not allowed to cause a garbage collection.
+  // A GC might move the calling generated code and invalidate the
+  // return address on the stack.
+  DCHECK_EQ(0, byte_length % 2);
+
+#ifdef V8_INTL_SUPPORT
+  int32_t length = (int32_t)(byte_length >> 1);
+  icu::UnicodeString uni_str_1(reinterpret_cast<const char16_t*>(byte_offset1),
+                               length);
+  return uni_str_1.caseCompare(reinterpret_cast<const char16_t*>(byte_offset2),
+                               length, U_FOLD_CASE_DEFAULT) == 0;
+#else
+  uc16* substring1 = reinterpret_cast<uc16*>(byte_offset1);
+  uc16* substring2 = reinterpret_cast<uc16*>(byte_offset2);
+  size_t length = byte_length >> 1;
+  DCHECK_NOT_NULL(isolate);
+  unibrow::Mapping<unibrow::Ecma262Canonicalize>* canonicalize =
+      isolate->regexp_macro_assembler_canonicalize();
+  for (size_t i = 0; i < length; i++) {
+    unibrow::uchar c1 = substring1[i];
+    unibrow::uchar c2 = substring2[i];
+    if (c1 != c2) {
+      unibrow::uchar s1[1] = {c1};
+      canonicalize->get(c1, '\0', s1);
+      if (s1[0] != c2) {
+        unibrow::uchar s2[1] = {c2};
+        canonicalize->get(c2, '\0', s2);
+        if (s1[0] != s2[0]) {
+          return 0;
+        }
+      }
+    }
+  }
+  return 1;
+#endif  // V8_INTL_SUPPORT
+}
+
+
+void RegExpMacroAssembler::CheckNotInSurrogatePair(int cp_offset,
+                                                   Label* on_failure) {
+  Label ok;
+  // Check that current character is not a trail surrogate.
+  LoadCurrentCharacter(cp_offset, &ok);
+  CheckCharacterNotInRange(kTrailSurrogateStart, kTrailSurrogateEnd, &ok);
+  // Check that previous character is not a lead surrogate.
+  LoadCurrentCharacter(cp_offset - 1, &ok);
+  CheckCharacterInRange(kLeadSurrogateStart, kLeadSurrogateEnd, on_failure);
+  Bind(&ok);
+}
+
+void RegExpMacroAssembler::CheckPosition(int cp_offset,
+                                         Label* on_outside_input) {
+  LoadCurrentCharacter(cp_offset, on_outside_input, true);
+}
+
+void RegExpMacroAssembler::LoadCurrentCharacter(int cp_offset,
+                                                Label* on_end_of_input,
+                                                bool check_bounds,
+                                                int characters,
+                                                int eats_at_least) {
+  // By default, eats_at_least = characters.
+  if (eats_at_least == kUseCharactersValue) {
+    eats_at_least = characters;
+  }
+
+  LoadCurrentCharacterImpl(cp_offset, on_end_of_input, check_bounds, characters,
+                           eats_at_least);
+}
+
+bool RegExpMacroAssembler::CheckSpecialCharacterClass(uc16 type,
+                                                      Label* on_no_match) {
+  return false;
+}
+
+NativeRegExpMacroAssembler::NativeRegExpMacroAssembler(Isolate* isolate,
+                                                       Zone* zone)
+    : RegExpMacroAssembler(isolate, zone) {}
+
+NativeRegExpMacroAssembler::~NativeRegExpMacroAssembler() = default;
+
+bool NativeRegExpMacroAssembler::CanReadUnaligned() {
+  return FLAG_enable_regexp_unaligned_accesses && !slow_safe();
+}
+
+#ifndef COMPILING_IRREGEXP_FOR_EXTERNAL_EMBEDDER
+
+// This method may only be called after an interrupt.
+int NativeRegExpMacroAssembler::CheckStackGuardState(
+    Isolate* isolate, int start_index, RegExp::CallOrigin call_origin,
+    Address* return_address, Code re_code, Address* subject,
+    const byte** input_start, const byte** input_end) {
+  DisallowHeapAllocation no_gc;
+  Address old_pc = PointerAuthentication::AuthenticatePC(return_address, 0);
+  DCHECK_LE(re_code.raw_instruction_start(), old_pc);
+  DCHECK_LE(old_pc, re_code.raw_instruction_end());
+
+  StackLimitCheck check(isolate);
+  bool js_has_overflowed = check.JsHasOverflowed();
+
+  if (call_origin == RegExp::CallOrigin::kFromJs) {
+    // Direct calls from JavaScript can be interrupted in two ways:
+    // 1. A real stack overflow, in which case we let the caller throw the
+    //    exception.
+    // 2. The stack guard was used to interrupt execution for another purpose,
+    //    forcing the call through the runtime system.
+
+    // Bug(v8:9540) Investigate why this method is called from JS although no
+    // stackoverflow or interrupt is pending on ARM64. We return 0 in this case
+    // to continue execution normally.
+    if (js_has_overflowed) {
+      return EXCEPTION;
+    } else if (check.InterruptRequested()) {
+      return RETRY;
+    } else {
+      return 0;
+    }
+  }
+  DCHECK(call_origin == RegExp::CallOrigin::kFromRuntime);
+
+  // Prepare for possible GC.
+  HandleScope handles(isolate);
+  Handle<Code> code_handle(re_code, isolate);
+  Handle<String> subject_handle(String::cast(Object(*subject)), isolate);
+  bool is_one_byte = String::IsOneByteRepresentationUnderneath(*subject_handle);
+  int return_value = 0;
+
+  if (js_has_overflowed) {
+    AllowHeapAllocation yes_gc;
+    isolate->StackOverflow();
+    return_value = EXCEPTION;
+  } else if (check.InterruptRequested()) {
+    AllowHeapAllocation yes_gc;
+    Object result = isolate->stack_guard()->HandleInterrupts();
+    if (result.IsException(isolate)) return_value = EXCEPTION;
+  }
+
+  if (*code_handle != re_code) {  // Return address no longer valid
+    // Overwrite the return address on the stack.
+    intptr_t delta = code_handle->address() - re_code.address();
+    Address new_pc = old_pc + delta;
+    // TODO(v8:10026): avoid replacing a signed pointer.
+    PointerAuthentication::ReplacePC(return_address, new_pc, 0);
+  }
+
+  // If we continue, we need to update the subject string addresses.
+  if (return_value == 0) {
+    // String encoding might have changed.
+    if (String::IsOneByteRepresentationUnderneath(*subject_handle) !=
+        is_one_byte) {
+      // If we changed between an LATIN1 and an UC16 string, the specialized
+      // code cannot be used, and we need to restart regexp matching from
+      // scratch (including, potentially, compiling a new version of the code).
+      return_value = RETRY;
+    } else {
+      *subject = subject_handle->ptr();
+      intptr_t byte_length = *input_end - *input_start;
+      *input_start = subject_handle->AddressOfCharacterAt(start_index, no_gc);
+      *input_end = *input_start + byte_length;
+    }
+  }
+  return return_value;
+}
+
+// Returns a {Result} sentinel, or the number of successful matches.
+int NativeRegExpMacroAssembler::Match(Handle<JSRegExp> regexp,
+                                      Handle<String> subject,
+                                      int* offsets_vector,
+                                      int offsets_vector_length,
+                                      int previous_index, Isolate* isolate) {
+  DCHECK(subject->IsFlat());
+  DCHECK_LE(0, previous_index);
+  DCHECK_LE(previous_index, subject->length());
+
+  // No allocations before calling the regexp, but we can't use
+  // DisallowHeapAllocation, since regexps might be preempted, and another
+  // thread might do allocation anyway.
+
+  String subject_ptr = *subject;
+  // Character offsets into string.
+  int start_offset = previous_index;
+  int char_length = subject_ptr.length() - start_offset;
+  int slice_offset = 0;
+
+  // The string has been flattened, so if it is a cons string it contains the
+  // full string in the first part.
+  if (StringShape(subject_ptr).IsCons()) {
+    DCHECK_EQ(0, ConsString::cast(subject_ptr).second().length());
+    subject_ptr = ConsString::cast(subject_ptr).first();
+  } else if (StringShape(subject_ptr).IsSliced()) {
+    SlicedString slice = SlicedString::cast(subject_ptr);
+    subject_ptr = slice.parent();
+    slice_offset = slice.offset();
+  }
+  if (StringShape(subject_ptr).IsThin()) {
+    subject_ptr = ThinString::cast(subject_ptr).actual();
+  }
+  // Ensure that an underlying string has the same representation.
+  bool is_one_byte = subject_ptr.IsOneByteRepresentation();
+  DCHECK(subject_ptr.IsExternalString() || subject_ptr.IsSeqString());
+  // String is now either Sequential or External
+  int char_size_shift = is_one_byte ? 0 : 1;
+
+  DisallowHeapAllocation no_gc;
+  const byte* input_start =
+      subject_ptr.AddressOfCharacterAt(start_offset + slice_offset, no_gc);
+  int byte_length = char_length << char_size_shift;
+  const byte* input_end = input_start + byte_length;
+  return Execute(*subject, start_offset, input_start, input_end, offsets_vector,
+                 offsets_vector_length, isolate, *regexp);
+}
+
+// Returns a {Result} sentinel, or the number of successful matches.
+// TODO(pthier): The JSRegExp object is passed to native irregexp code to match
+// the signature of the interpreter. We should get rid of JS objects passed to
+// internal methods.
+int NativeRegExpMacroAssembler::Execute(
+    String input,  // This needs to be the unpacked (sliced, cons) string.
+    int start_offset, const byte* input_start, const byte* input_end,
+    int* output, int output_size, Isolate* isolate, JSRegExp regexp) {
+  // Ensure that the minimum stack has been allocated.
+  RegExpStackScope stack_scope(isolate);
+  Address stack_base = stack_scope.stack()->stack_base();
+
+  bool is_one_byte = String::IsOneByteRepresentationUnderneath(input);
+  Code code = Code::cast(regexp.Code(is_one_byte));
+  RegExp::CallOrigin call_origin = RegExp::CallOrigin::kFromRuntime;
+
+  using RegexpMatcherSig = int(
+      Address input_string, int start_offset,  // NOLINT(readability/casting)
+      const byte* input_start, const byte* input_end, int* output,
+      int output_size, Address stack_base, int call_origin, Isolate* isolate,
+      Address regexp);
+
+  auto fn = GeneratedCode<RegexpMatcherSig>::FromCode(code);
+  int result =
+      fn.Call(input.ptr(), start_offset, input_start, input_end, output,
+              output_size, stack_base, call_origin, isolate, regexp.ptr());
+  DCHECK(result >= RETRY);
+
+  if (result == EXCEPTION && !isolate->has_pending_exception()) {
+    // We detected a stack overflow (on the backtrack stack) in RegExp code,
+    // but haven't created the exception yet. Additionally, we allow heap
+    // allocation because even though it invalidates {input_start} and
+    // {input_end}, we are about to return anyway.
+    AllowHeapAllocation allow_allocation;
+    isolate->StackOverflow();
+  }
+  return result;
+}
+
+#endif  // !COMPILING_IRREGEXP_FOR_EXTERNAL_EMBEDDER
+
+// clang-format off
+const byte NativeRegExpMacroAssembler::word_character_map[] = {
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu,  // '0' - '7'
+    0xFFu, 0xFFu, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,  // '8' - '9'
+
+    0x00u, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu,  // 'A' - 'G'
+    0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu,  // 'H' - 'O'
+    0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu,  // 'P' - 'W'
+    0xFFu, 0xFFu, 0xFFu, 0x00u, 0x00u, 0x00u, 0x00u, 0xFFu,  // 'X' - 'Z', '_'
+
+    0x00u, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu,  // 'a' - 'g'
+    0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu,  // 'h' - 'o'
+    0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu, 0xFFu,  // 'p' - 'w'
+    0xFFu, 0xFFu, 0xFFu, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,  // 'x' - 'z'
+    // Latin-1 range
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+    0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u, 0x00u,
+};
+// clang-format on
+
+Address NativeRegExpMacroAssembler::GrowStack(Address stack_pointer,
+                                              Address* stack_base,
+                                              Isolate* isolate) {
+  RegExpStack* regexp_stack = isolate->regexp_stack();
+  size_t size = regexp_stack->stack_capacity();
+  Address old_stack_base = regexp_stack->stack_base();
+  DCHECK(old_stack_base == *stack_base);
+  DCHECK(stack_pointer <= old_stack_base);
+  DCHECK(static_cast<size_t>(old_stack_base - stack_pointer) <= size);
+  Address new_stack_base = regexp_stack->EnsureCapacity(size * 2);
+  if (new_stack_base == kNullAddress) {
+    return kNullAddress;
+  }
+  *stack_base = new_stack_base;
+  intptr_t stack_content_size = old_stack_base - stack_pointer;
+  return new_stack_base - stack_content_size;
+}
+
+}  // namespace internal
+}  // namespace v8

js/src/regexp/regexp-macro-assembler.h

@@ -0,0 +1,280 @@
+// Copyright 2012 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef V8_REGEXP_REGEXP_MACRO_ASSEMBLER_H_
+#define V8_REGEXP_REGEXP_MACRO_ASSEMBLER_H_
+
+#include "regexp/regexp-ast.h"
+#include "regexp/regexp-shim.h"
+#include "regexp/regexp.h"
+
+namespace v8 {
+namespace internal {
+
+static const uc32 kLeadSurrogateStart = 0xd800;
+static const uc32 kLeadSurrogateEnd = 0xdbff;
+static const uc32 kTrailSurrogateStart = 0xdc00;
+static const uc32 kTrailSurrogateEnd = 0xdfff;
+static const uc32 kNonBmpStart = 0x10000;
+static const uc32 kNonBmpEnd = 0x10ffff;
+
+struct DisjunctDecisionRow {
+  RegExpCharacterClass cc;
+  Label* on_match;
+};
+
+
+class RegExpMacroAssembler {
+ public:
+  // The implementation must be able to handle at least:
+  static const int kMaxRegister = (1 << 16) - 1;
+  static const int kMaxCPOffset = (1 << 15) - 1;
+  static const int kMinCPOffset = -(1 << 15);
+
+  static const int kTableSizeBits = 7;
+  static const int kTableSize = 1 << kTableSizeBits;
+  static const int kTableMask = kTableSize - 1;
+
+  static constexpr int kUseCharactersValue = -1;
+
+  enum IrregexpImplementation {
+    kIA32Implementation,
+    kARMImplementation,
+    kARM64Implementation,
+    kMIPSImplementation,
+    kS390Implementation,
+    kPPCImplementation,
+    kX64Implementation,
+    kX87Implementation,
+    kBytecodeImplementation
+  };
+
+  enum StackCheckFlag {
+    kNoStackLimitCheck = false,
+    kCheckStackLimit = true
+  };
+
+  RegExpMacroAssembler(Isolate* isolate, Zone* zone);
+  virtual ~RegExpMacroAssembler();
+  // This function is called when code generation is aborted, so that
+  // the assembler could clean up internal data structures.
+  virtual void AbortedCodeGeneration() {}
+  // The maximal number of pushes between stack checks. Users must supply
+  // kCheckStackLimit flag to push operations (instead of kNoStackLimitCheck)
+  // at least once for every stack_limit() pushes that are executed.
+  virtual int stack_limit_slack() = 0;
+  virtual bool CanReadUnaligned() = 0;
+  virtual void AdvanceCurrentPosition(int by) = 0;  // Signed cp change.
+  virtual void AdvanceRegister(int reg, int by) = 0;  // r[reg] += by.
+  // Continues execution from the position pushed on the top of the backtrack
+  // stack by an earlier PushBacktrack(Label*).
+  virtual void Backtrack() = 0;
+  virtual void Bind(Label* label) = 0;
+  // Dispatch after looking the current character up in a 2-bits-per-entry
+  // map.  The destinations vector has up to 4 labels.
+  virtual void CheckCharacter(unsigned c, Label* on_equal) = 0;
+  // Bitwise and the current character with the given constant and then
+  // check for a match with c.
+  virtual void CheckCharacterAfterAnd(unsigned c,
+                                      unsigned and_with,
+                                      Label* on_equal) = 0;
+  virtual void CheckCharacterGT(uc16 limit, Label* on_greater) = 0;
+  virtual void CheckCharacterLT(uc16 limit, Label* on_less) = 0;
+  virtual void CheckGreedyLoop(Label* on_tos_equals_current_position) = 0;
+  virtual void CheckAtStart(int cp_offset, Label* on_at_start) = 0;
+  virtual void CheckNotAtStart(int cp_offset, Label* on_not_at_start) = 0;
+  virtual void CheckNotBackReference(int start_reg, bool read_backward,
+                                     Label* on_no_match) = 0;
+  virtual void CheckNotBackReferenceIgnoreCase(int start_reg,
+                                               bool read_backward,
+                                               Label* on_no_match) = 0;
+  // Check the current character for a match with a literal character.  If we
+  // fail to match then goto the on_failure label.  End of input always
+  // matches.  If the label is nullptr then we should pop a backtrack address
+  // off the stack and go to that.
+  virtual void CheckNotCharacter(unsigned c, Label* on_not_equal) = 0;
+  virtual void CheckNotCharacterAfterAnd(unsigned c,
+                                         unsigned and_with,
+                                         Label* on_not_equal) = 0;
+  // Subtract a constant from the current character, then and with the given
+  // constant and then check for a match with c.
+  virtual void CheckNotCharacterAfterMinusAnd(uc16 c,
+                                              uc16 minus,
+                                              uc16 and_with,
+                                              Label* on_not_equal) = 0;
+  virtual void CheckCharacterInRange(uc16 from,
+                                     uc16 to,  // Both inclusive.
+                                     Label* on_in_range) = 0;
+  virtual void CheckCharacterNotInRange(uc16 from,
+                                        uc16 to,  // Both inclusive.
+                                        Label* on_not_in_range) = 0;
+
+  // The current character (modulus the kTableSize) is looked up in the byte
+  // array, and if the found byte is non-zero, we jump to the on_bit_set label.
+  virtual void CheckBitInTable(Handle<ByteArray> table, Label* on_bit_set) = 0;
+
+  // Checks whether the given offset from the current position is before
+  // the end of the string.  May overwrite the current character.
+  virtual void CheckPosition(int cp_offset, Label* on_outside_input);
+  // Check whether a standard/default character class matches the current
+  // character. Returns false if the type of special character class does
+  // not have custom support.
+  // May clobber the current loaded character.
+  virtual bool CheckSpecialCharacterClass(uc16 type, Label* on_no_match);
+
+  // Control-flow integrity:
+  // Define a jump target and bind a label.
+  virtual void BindJumpTarget(Label* label) { Bind(label); }
+
+  virtual void Fail() = 0;
+  virtual Handle<HeapObject> GetCode(Handle<String> source) = 0;
+  virtual void GoTo(Label* label) = 0;
+  // Check whether a register is >= a given constant and go to a label if it
+  // is.  Backtracks instead if the label is nullptr.
+  virtual void IfRegisterGE(int reg, int comparand, Label* if_ge) = 0;
+  // Check whether a register is < a given constant and go to a label if it is.
+  // Backtracks instead if the label is nullptr.
+  virtual void IfRegisterLT(int reg, int comparand, Label* if_lt) = 0;
+  // Check whether a register is == to the current position and go to a
+  // label if it is.
+  virtual void IfRegisterEqPos(int reg, Label* if_eq) = 0;
+  virtual IrregexpImplementation Implementation() = 0;
+  V8_EXPORT_PRIVATE void LoadCurrentCharacter(
+      int cp_offset, Label* on_end_of_input, bool check_bounds = true,
+      int characters = 1, int eats_at_least = kUseCharactersValue);
+  virtual void LoadCurrentCharacterImpl(int cp_offset, Label* on_end_of_input,
+                                        bool check_bounds, int characters,
+                                        int eats_at_least) = 0;
+  virtual void PopCurrentPosition() = 0;
+  virtual void PopRegister(int register_index) = 0;
+  // Pushes the label on the backtrack stack, so that a following Backtrack
+  // will go to this label. Always checks the backtrack stack limit.
+  virtual void PushBacktrack(Label* label) = 0;
+  virtual void PushCurrentPosition() = 0;
+  virtual void PushRegister(int register_index,
+                            StackCheckFlag check_stack_limit) = 0;
+  virtual void ReadCurrentPositionFromRegister(int reg) = 0;
+  virtual void ReadStackPointerFromRegister(int reg) = 0;
+  virtual void SetCurrentPositionFromEnd(int by) = 0;
+  virtual void SetRegister(int register_index, int to) = 0;
+  // Return whether the matching (with a global regexp) will be restarted.
+  virtual bool Succeed() = 0;
+  virtual void WriteCurrentPositionToRegister(int reg, int cp_offset) = 0;
+  virtual void ClearRegisters(int reg_from, int reg_to) = 0;
+  virtual void WriteStackPointerToRegister(int reg) = 0;
+
+  // Compares two-byte strings case insensitively.
+  // Called from generated RegExp code.
+  static int CaseInsensitiveCompareUC16(Address byte_offset1,
+                                        Address byte_offset2,
+                                        size_t byte_length, Isolate* isolate);
+
+  // Check that we are not in the middle of a surrogate pair.
+  void CheckNotInSurrogatePair(int cp_offset, Label* on_failure);
+
+  // Controls the generation of large inlined constants in the code.
+  void set_slow_safe(bool ssc) { slow_safe_compiler_ = ssc; }
+  bool slow_safe() { return slow_safe_compiler_; }
+
+  void set_backtrack_limit(uint32_t backtrack_limit) {
+    backtrack_limit_ = backtrack_limit;
+  }
+
+  enum GlobalMode {
+    NOT_GLOBAL,
+    GLOBAL_NO_ZERO_LENGTH_CHECK,
+    GLOBAL,
+    GLOBAL_UNICODE
+  };
+  // Set whether the regular expression has the global flag.  Exiting due to
+  // a failure in a global regexp may still mean success overall.
+  inline void set_global_mode(GlobalMode mode) { global_mode_ = mode; }
+  inline bool global() { return global_mode_ != NOT_GLOBAL; }
+  inline bool global_with_zero_length_check() {
+    return global_mode_ == GLOBAL || global_mode_ == GLOBAL_UNICODE;
+  }
+  inline bool global_unicode() { return global_mode_ == GLOBAL_UNICODE; }
+
+  Isolate* isolate() const { return isolate_; }
+  Zone* zone() const { return zone_; }
+
+ protected:
+  bool has_backtrack_limit() const {
+    return backtrack_limit_ != JSRegExp::kNoBacktrackLimit;
+  }
+  uint32_t backtrack_limit() const { return backtrack_limit_; }
+
+ private:
+  bool slow_safe_compiler_;
+  uint32_t backtrack_limit_ = JSRegExp::kNoBacktrackLimit;
+  GlobalMode global_mode_;
+  Isolate* isolate_;
+  Zone* zone_;
+};
+
+class NativeRegExpMacroAssembler: public RegExpMacroAssembler {
+ public:
+  // Type of input string to generate code for.
+  enum Mode { LATIN1 = 1, UC16 = 2 };
+
+  // Result of calling generated native RegExp code.
+  // RETRY: Something significant changed during execution, and the matching
+  //        should be retried from scratch.
+  // EXCEPTION: Something failed during execution. If no exception has been
+  //        thrown, it's an internal out-of-memory, and the caller should
+  //        throw the exception.
+  // FAILURE: Matching failed.
+  // SUCCESS: Matching succeeded, and the output array has been filled with
+  //        capture positions.
+  enum Result {
+    FAILURE = RegExp::kInternalRegExpFailure,
+    SUCCESS = RegExp::kInternalRegExpSuccess,
+    EXCEPTION = RegExp::kInternalRegExpException,
+    RETRY = RegExp::kInternalRegExpRetry,
+  };
+
+  NativeRegExpMacroAssembler(Isolate* isolate, Zone* zone);
+  ~NativeRegExpMacroAssembler() override;
+  bool CanReadUnaligned() override;
+
+  // Returns a {Result} sentinel, or the number of successful matches.
+  static int Match(Handle<JSRegExp> regexp, Handle<String> subject,
+                   int* offsets_vector, int offsets_vector_length,
+                   int previous_index, Isolate* isolate);
+
+  // Called from RegExp if the backtrack stack limit is hit.
+  // Tries to expand the stack. Returns the new stack-pointer if
+  // successful, and updates the stack_top address, or returns 0 if unable
+  // to grow the stack.
+  // This function must not trigger a garbage collection.
+  static Address GrowStack(Address stack_pointer, Address* stack_top,
+                           Isolate* isolate);
+
+  static int CheckStackGuardState(Isolate* isolate, int start_index,
+                                  RegExp::CallOrigin call_origin,
+                                  Address* return_address, Code re_code,
+                                  Address* subject, const byte** input_start,
+                                  const byte** input_end);
+
+  // Byte map of one byte characters with a 0xff if the character is a word
+  // character (digit, letter or underscore) and 0x00 otherwise.
+  // Used by generated RegExp code.
+  static const byte word_character_map[256];
+
+  static Address word_character_map_address() {
+    return reinterpret_cast<Address>(&word_character_map[0]);
+  }
+
+  // Returns a {Result} sentinel, or the number of successful matches.
+  V8_EXPORT_PRIVATE static int Execute(String input, int start_offset,
+                                       const byte* input_start,
+                                       const byte* input_end, int* output,
+                                       int output_size, Isolate* isolate,
+                                       JSRegExp regexp);
+};
+
+}  // namespace internal
+}  // namespace v8
+
+#endif  // V8_REGEXP_REGEXP_MACRO_ASSEMBLER_H_