![electron-roller[bot]](/assets/img/avatar_default.png)
* chore: bump chromium in DEPS to 140.0.7330.0 * chore: bump chromium in DEPS to 140.0.7331.0 * chore: update patches * fix: gn check failing on crashpad.h Not yet sure what caused this * fix: predictors::PreconnectManager -> content::PreconnectManager CL: https://chromium-review.googlesource.com/c/chromium/src/+/6788473 * chore: bump chromium in DEPS to 140.0.7333.0 * chore: bump chromium in DEPS to 140.0.7335.0 * chore: bump chromium in DEPS to 140.0.7337.0 * chore: update patches * chore: restore some gin utility * 6804057: [Extensions] Validate nodoc is specified as a boolean in schemas https://chromium-review.googlesource.com/c/chromium/src/+/6804057 * fixup! chore: restore some gin utility * fixup! fix: predictors::PreconnectManager -> content::PreconnectManager CL: https://chromium-review.googlesource.com/c/chromium/src/+/6788473 * 6772346: Reset MouseWheelPhaseHandler state when trackpoint scroll is detected https://chromium-review.googlesource.com/c/chromium/src/+/6772346 Not certain about what the "correct" argument to pass here is. A quick dive into the CL suggests that passing `false` is safe to keep things working. The blast radius if this assumption is wrong is that "fling" scroll gestures may not work as expected with the OSR. * 6789383: Uninstall SODA language pack after 30 days of inactivity https://chromium-review.googlesource.com/c/chromium/src/+/6789383 * chore: update libcxx filenames * chore: bump chromium in DEPS to 140.0.7339.0 * chore: update patches * fixup! 6772346: Reset MouseWheelPhaseHandler state when trackpoint scroll is detected https://chromium-review.googlesource.com/c/chromium/src/+/6772346 * chore: bump chromium in DEPS to 140.0.7339.2 --------- Co-authored-by: electron-roller[bot] <84116207+electron-roller[bot]@users.noreply.github.com> Co-authored-by: Samuel Maddock <smaddock@slack-corp.com> Co-authored-by: deepak1556 <hop2deep@gmail.com> Co-authored-by: clavin <clavin@electronjs.org>
77 lines
3.5 KiB
Diff
77 lines
3.5 KiB
Diff
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
|
|
From: Cheng Zhao <zcbenz@gmail.com>
|
|
Date: Thu, 20 Sep 2018 17:45:03 -0700
|
|
Subject: blink_local_frame.patch
|
|
|
|
According to electron/electron#3699, it is unreliable to use |unload|
|
|
event for process.exit('exit'), so we have to do that in
|
|
willReleaseScriptContext.
|
|
|
|
Refs changes in:
|
|
* https://codereview.chromium.org/1657583002
|
|
* https://chromium-review.googlesource.com/c/chromium/src/+/2596880
|
|
* https://chromium-review.googlesource.com/c/chromium/src/+/2597169
|
|
|
|
This patch reverts the changes to fix associated crashes in Electron.
|
|
|
|
diff --git a/third_party/blink/renderer/core/frame/frame.cc b/third_party/blink/renderer/core/frame/frame.cc
|
|
index c9f34fa47702504ccdefb8d61c55f5eaae501085..26df03d777c9ea487cae37f3df91d1df233b75e2 100644
|
|
--- a/third_party/blink/renderer/core/frame/frame.cc
|
|
+++ b/third_party/blink/renderer/core/frame/frame.cc
|
|
@@ -134,14 +134,6 @@ bool Frame::Detach(FrameDetachType type) {
|
|
|
|
DCHECK(!IsDetached());
|
|
|
|
- // TODO(dcheng): FocusController::FrameDetached() *should* fire JS events,
|
|
- // hence the above check for `client_` being null. However, when this was
|
|
- // previously placed before the `FrameDetached()` call, nothing crashes, which
|
|
- // is suspicious. Investigate if we really don't need to fire JS events--and
|
|
- // if we don't, move `forbid_scripts` up to be instantiated sooner and
|
|
- // simplify this code.
|
|
- ScriptForbiddenScope forbid_scripts;
|
|
-
|
|
if (type == FrameDetachType::kRemove) {
|
|
if (provisional_frame_) {
|
|
provisional_frame_->Detach(FrameDetachType::kRemove);
|
|
@@ -166,6 +158,14 @@ bool Frame::Detach(FrameDetachType type) {
|
|
GetWindowProxyManager()->ClearForSwap();
|
|
}
|
|
|
|
+ // TODO(dcheng): FocusController::FrameDetached() *should* fire JS events,
|
|
+ // hence the above check for `client_` being null. However, when this was
|
|
+ // previously placed before the `FrameDetached()` call, nothing crashes, which
|
|
+ // is suspicious. Investigate if we really don't need to fire JS events--and
|
|
+ // if we don't, move `forbid_scripts` up to be instantiated sooner and
|
|
+ // simplify this code.
|
|
+ ScriptForbiddenScope forbid_scripts;
|
|
+
|
|
// After this, we must no longer talk to the client since this clears
|
|
// its owning reference back to our owning LocalFrame.
|
|
client_->Detached(type);
|
|
diff --git a/third_party/blink/renderer/core/frame/local_frame.cc b/third_party/blink/renderer/core/frame/local_frame.cc
|
|
index f4386917072e83f8797d0dbed5f1e2c72659ce80..800a904a8fa47e0a36d8393e11b3e7d7c4a3550e 100644
|
|
--- a/third_party/blink/renderer/core/frame/local_frame.cc
|
|
+++ b/third_party/blink/renderer/core/frame/local_frame.cc
|
|
@@ -738,10 +738,6 @@ bool LocalFrame::DetachImpl(FrameDetachType type) {
|
|
}
|
|
DCHECK(!view_ || !view_->IsAttached());
|
|
|
|
- // This is the earliest that scripting can be disabled:
|
|
- // - FrameLoader::Detach() can fire XHR abort events
|
|
- // - Document::Shutdown() can dispose plugins which can run script.
|
|
- ScriptForbiddenScope forbid_script;
|
|
if (!Client())
|
|
return false;
|
|
|
|
@@ -795,6 +791,11 @@ bool LocalFrame::DetachImpl(FrameDetachType type) {
|
|
DCHECK(!view_->IsAttached());
|
|
Client()->WillBeDetached();
|
|
|
|
+ // This is the earliest that scripting can be disabled:
|
|
+ // - FrameLoader::Detach() can fire XHR abort events
|
|
+ // - Document::Shutdown() can dispose plugins which can run script.
|
|
+ ScriptForbiddenScope forbid_script;
|
|
+
|
|
// TODO(crbug.com/729196): Trace why LocalFrameView::DetachFromLayout crashes.
|
|
CHECK(!view_->IsAttached());
|
|
SetView(nullptr);
|