If you serialize and then deserialize a JSErrorReport with a null filename
field, then we end up passing a null filename to JS::CreateError, which ends
up crashing. Perhaps CreateError should be robust against a null filename,
but the fact that whatever latent null filename issue wasn't causing crashes
until this serialization was added suggests it might be reasonable to deal
with it inside ClonedErrorHolder.
This patch causes a null string to instead become an empty string. We could do
this on the sending side, but by dealing with it on the receiving side we get
the bonus of keeping a malicious sender from crashing us.
Differential Revision: https://phabricator.services.mozilla.com/D60587
--HG--
extra : moz-landing-system : lando
This patch does not change any of the *internals* of process management in
`GeckoProcessManager`, however it upgrades the interface in preparation for
such changes.
* We add a java equivalent to Gecko's `GeckoProcessType` enumeration;
* We add the `GeckoProcessType.Selector` class which will be used for
identifying running content processes;
* We update `ContentParent::MarkAsDead` to use `GeckoProcessType.Selector` for
indicating which process should be marked.
Differential Revision: https://phabricator.services.mozilla.com/D60539
--HG--
extra : moz-landing-system : lando
Also logs JSWindowActor{Child,Parent}::Init and JSWindowActor::{Start,After}Destroy.
Differential Revision: https://phabricator.services.mozilla.com/D45346
--HG--
extra : moz-landing-system : lando
Adds GetManager on JSWindowActor (returns the associated WindowGlobalActor),
and IsInProcess on WindowGlobalActor (returns whether the actor is in process
or not).
Differential Revision: https://phabricator.services.mozilla.com/D57358
--HG--
extra : moz-landing-system : lando
Also logs JSWindowActor{Child,Parent}::Init and JSWindowActor::{Start,After}Destroy.
Differential Revision: https://phabricator.services.mozilla.com/D45346
--HG--
extra : moz-landing-system : lando
Adds GetManager on JSWindowActor (returns the associated WindowGlobalActor),
and IsInProcess on WindowGlobalActor (returns whether the actor is in process
or not).
Differential Revision: https://phabricator.services.mozilla.com/D57358
--HG--
extra : moz-landing-system : lando
This was done by:
This was done by applying:
```
diff --git a/python/mozbuild/mozbuild/code-analysis/mach_commands.py b/python/mozbuild/mozbuild/code-analysis/mach_commands.py
index 789affde7bbf..fe33c4c7d4d1 100644
--- a/python/mozbuild/mozbuild/code-analysis/mach_commands.py
+++ b/python/mozbuild/mozbuild/code-analysis/mach_commands.py
@@ -2007,7 +2007,7 @@ class StaticAnalysis(MachCommandBase):
from subprocess import Popen, PIPE, check_output, CalledProcessError
diff_process = Popen(self._get_clang_format_diff_command(commit), stdout=PIPE)
- args = [sys.executable, clang_format_diff, "-p1", "-binary=%s" % clang_format]
+ args = [sys.executable, clang_format_diff, "-p1", "-binary=%s" % clang_format, '-sort-includes']
if not output_file:
args.append("-i")
```
Then running `./mach clang-format -c <commit-hash>`
Then undoing that patch.
Then running check_spidermonkey_style.py --fixup
Then running `./mach clang-format`
I had to fix four things:
* I needed to move <utility> back down in GuardObjects.h because I was hitting
obscure problems with our system include wrappers like this:
0:03.94 /usr/include/stdlib.h:550:14: error: exception specification in declaration does not match previous declaration
0:03.94 extern void *realloc (void *__ptr, size_t __size)
0:03.94 ^
0:03.94 /home/emilio/src/moz/gecko-2/obj-debug/dist/include/malloc_decls.h:53:1: note: previous declaration is here
0:03.94 MALLOC_DECL(realloc, void*, void*, size_t)
0:03.94 ^
0:03.94 /home/emilio/src/moz/gecko-2/obj-debug/dist/include/mozilla/mozalloc.h:22:32: note: expanded from macro 'MALLOC_DECL'
0:03.94 MOZ_MEMORY_API return_type name##_impl(__VA_ARGS__);
0:03.94 ^
0:03.94 <scratch space>:178:1: note: expanded from here
0:03.94 realloc_impl
0:03.94 ^
0:03.94 /home/emilio/src/moz/gecko-2/obj-debug/dist/include/mozmemory_wrap.h:142:41: note: expanded from macro 'realloc_impl'
0:03.94 #define realloc_impl mozmem_malloc_impl(realloc)
Which I really didn't feel like digging into.
* I had to restore the order of TrustOverrideUtils.h and related files in nss
because the .inc files depend on TrustOverrideUtils.h being included earlier.
* I had to add a missing include to RollingNumber.h
* Also had to partially restore include order in JsepSessionImpl.cpp to avoid
some -WError issues due to some static inline functions being defined in a
header but not used in the rest of the compilation unit.
Differential Revision: https://phabricator.services.mozilla.com/D60327
--HG--
extra : moz-landing-system : lando
rg -l 'mozilla/Move.h' | xargs sed -i 's/#include "mozilla\/Move.h"/#include <utility>/g'
Further manual fixups and cleanups to the include order incoming.
Differential Revision: https://phabricator.services.mozilla.com/D60323
--HG--
extra : moz-landing-system : lando
We need to propagate `ContentParent::MarkAsDead` up to the Android java layer
so that it has a view of the state of content processes that is consistent
with the view of Gecko's content process management.
Differential Revision: https://phabricator.services.mozilla.com/D58564
--HG--
extra : moz-landing-system : lando
As each media controller corresponds to a tab (a browsing context tree), so the controller's life cycle should be equal to the tab. Currently we use the top-level browsing context to represent a tab, so when that browsing context is being destroyed, we should also destroy the corresponding media controller.
Therefore, `MediaControlService` don't need to have methods to access media controller anymore, we should access media controller directly from the top-level canonical browsing context.
Differential Revision: https://phabricator.services.mozilla.com/D58591
--HG--
extra : moz-landing-system : lando
To modify variable and function names to emphasize all media controllers added in the service are active controllers. If the controller doesn't have any controlled media, they won't be added into the service.
Differential Revision: https://phabricator.services.mozilla.com/D58589
--HG--
extra : moz-landing-system : lando
The assertion is added to diagnose whether NotifyContentBlockingEvent
will be called from an in-process document. We have already seen crashes
because of the assertion, so temporarily remove this to avoid people
keep crashing due to this.
Differential Revision: https://phabricator.services.mozilla.com/D60002
--HG--
extra : moz-landing-system : lando
This patch does the following:
1. Add a WindowGlobalChild IPC method - GetContentBlockingEvents. Documents can use the method
retrieve content blocking events stored in the parent process.
2. Update nsIDocShell::GetHasTrackingContentBlocked to return a promise.
3. Replace API in report-site-issue to use the promise-based GetHasTrackingContentBlocked API
Differential Revision: https://phabricator.services.mozilla.com/D56204
--HG--
extra : moz-landing-system : lando
1. Rename mCallerDocumentURI field of PostMessageEvent to mCallerURI as we
might not always have a document, e.g. when we have a sandbox.
2. When we neither have caller's inner window (and thus we have no caller's
window ID) nor caller's URI, use script location as a source name for the error
object when reporting errors to console.
3. When the caller of postMessage has its window in a different process, pass along
the caller's window ID so that the error can be reported correctly.
Differential Revision: https://phabricator.services.mozilla.com/D59526
--HG--
extra : moz-landing-system : lando
This allows us to access contentBlockingLog in the parent process
through a WindowGlobalParent object.
Differential Revision: https://phabricator.services.mozilla.com/D55779
--HG--
extra : moz-landing-system : lando
This allows us to access contentBlockingEvents in the parent process
through a WindowGlobalParent object.
Differential Revision: https://phabricator.services.mozilla.com/D55621
--HG--
extra : moz-landing-system : lando
The OnContentBlockingEvent is no longer needed once we make the
OnContentBlockingEvent parent only.
Differential Revision: https://phabricator.services.mozilla.com/D55788
--HG--
extra : moz-landing-system : lando
This patch adds an IPC message which allows content process to notify
the OnContentBlockingEvent in the parent process. This is needed because
there are some situations that the content blocking happens in content
processes, so we need to tell the parent process to notify it. Such as,
AntiTrackingCommon::AntiTrackingCommon::AddFirstPartyStorageAccessGrantedFor().
Differential Revision: https://phabricator.services.mozilla.com/D55648
--HG--
extra : moz-landing-system : lando
This patch makes the WindowGlobalParent to be able to send the
OnContentBlockingEvent in the parent process.
Differential Revision: https://phabricator.services.mozilla.com/D55645
--HG--
extra : moz-landing-system : lando
This adds a ContentBlockingLog into the WindowGlobalParent. This
ContentBlockingLog is bascially a copy of the log in the content
process. This log in parent is needed for the OnContentBlockingEvent in
parent process since we need an overview of the content blocking events
for OnContentBlockingEvent. And the ContentBlockingLog exits in the
content process in this stage, so we need to maintain a copy of it to
get the overview of the blocking events in parent.
Differential Revision: https://phabricator.services.mozilla.com/D54929
--HG--
extra : moz-landing-system : lando
Right now we do the same thing in two pretty different code paths... That's not
great, so unify them.
Differential Revision: https://phabricator.services.mozilla.com/D59629
--HG--
extra : moz-landing-system : lando
Right now we do the same thing in two pretty different code paths... That's not
great, so unify them.
Differential Revision: https://phabricator.services.mozilla.com/D59629
--HG--
extra : moz-landing-system : lando
This doesn't block the STATE_START notification from the new process, as we currently have a second start notification (when DocumentChannel redirects to the real channel), so this is unchanged.
Differential Revision: https://phabricator.services.mozilla.com/D56818
--HG--
extra : moz-landing-system : lando
When we let the WebIDL callback wrapper report exceptions, it returns an error
result of NS_ERROR_UNEXPECTED, which winds up getting set as a promise
rejection reason or re-reported to the console. This just causes unnecessary
confusion for callers, especially since the error has no location information.
Differential Revision: https://phabricator.services.mozilla.com/D57657
--HG--
extra : moz-landing-system : lando
Splits WebGLContext into ClientWebGLContext and HostWebGLContext. The Client enables the JS-control of a WebGL context in a content procecss while the Host executes the WebGL graphics operations (via a WebGLContext that maintains much of the existing code) in the compositor process. At this point, the cross-process behavior is disabled -- this series of patches is an incremental step toward that final goal.
Differential Revision: https://phabricator.services.mozilla.com/D54018
--HG--
extra : moz-landing-system : lando
Under some (yet) unclear conditions it's possible for a content
process to send session storage data for a browsing context that's
already been discarded. This leads to an assertion failure when
deserializing the IPC message. This commit works around this issue
by sending the browsing context's ID over IPC and issuing a warning
instead of asserting that the browsing context still exists.
Differential Revision: https://phabricator.services.mozilla.com/D59096
--HG--
extra : moz-landing-system : lando
* Makes it possible to selectively enable TRR for pbmode/container/window/etc
Differential Revision: https://phabricator.services.mozilla.com/D48363
--HG--
extra : moz-landing-system : lando
Note that this also implicitly adds support for the view-source+srcdoc configuration, and setting of the BaseURI, which were both in the nsDocShell version but not the others.
Differential Revision: https://phabricator.services.mozilla.com/D57887
--HG--
extra : moz-landing-system : lando
Now that we have UTF8String in the WebIDL, we can remove quite a few of the
conversions. Do that, and lift the remaining string conversions up as needed.
Also deindent Servo_ComputeColor while touching it.
Most of the remaining copies are because either bug 1606994, or because they're
WebIDL attributes that we still need to serialize back as UTF-16 (bug 1606995).
Differential Revision: https://phabricator.services.mozilla.com/D58687
--HG--
extra : moz-landing-system : lando
Note that this also implicitly adds support for the view-source+srcdoc configuration, and setting of the BaseURI, which were both in the nsDocShell version but not the others.
Differential Revision: https://phabricator.services.mozilla.com/D57887
--HG--
extra : moz-landing-system : lando
This makes clear where the information comes from, and also that there are some
bits of information that we should pass down from the child that we don't, like
allowfullscreen and the frame name.
Differential Revision: https://phabricator.services.mozilla.com/D58535
--HG--
extra : moz-landing-system : lando
Introduce `ControlledMediaState` to media controller, so now we can now how many controlled media are playing, which allow us to update `mIsPlaying` correctly.
Differential Revision: https://phabricator.services.mozilla.com/D57574
--HG--
extra : moz-landing-system : lando
Modify IPC method to support notify `ControlledMediaState` to the media controller, and remove the previous implementation.
Differential Revision: https://phabricator.services.mozilla.com/D57572
--HG--
extra : moz-landing-system : lando
When a content or plug-in process crashes too early we haven't initialized the
CrashReporterHost for that process. This will cause the crash to be orphaned,
i.e. to miss most of its crash annotations. We added code to finalize those
crashes in bug 1282776 so that we wouldn't miss them entirely. This ensured
that crash reports would have both their .dmp and .extra files but the patch
failed to modify the code that notified various listeners about the crash
report's presence.
This changes always send the crash ID alongside the crash notifications, even
for orphaned crashes, so that listeners such as the content crash handler or
the test harnesses can always find the minidump and .extra file. Additionally
orphaned crashes are recorded in the CrashManager and in telemetry just like
normal crashes.
This also re-enables dom/ipc/tests/process_error.xul which failed frequently
because of this bug.
Differential Revision: https://phabricator.services.mozilla.com/D57634
--HG--
extra : moz-landing-system : lando
When we open a new window from a content process, we create a nested event
loop to wait for it to be initialized by the parent. The problem with this is
that the OpenWindow code which calls the window provider expects the window to
be in-process and uninitialized, so that it can load its own initial URI into
it, and correctly fulfil the spec-codified contract of window.open(). If
another caller initiates a load in the new window during the nested event
loop, those invariants are broken, and any manner of undefined behavior can
occur.
This patch adds a new flag to the BrowsingContext, marking it as uninitialized
until the end of the nested event loop, and blocking any attempts to load a
new URI into it in the meantime.
Differential Revision: https://phabricator.services.mozilla.com/D57667
--HG--
extra : moz-landing-system : lando
This doesn't solve all problems with potential reentrancy during window.open
nested event loops, but it does improve the situation somewhat. Since any
window in the same BrowsingContextGroup can target any window in the same
group, we need to suspend all windows in the group, not just the root of the
new window's parent. We also need to make sure we suspend all in-process
windows, even if we have out-of-process frames somewhere in the parent chain.
This patch takes care of suspending timeouts and input event handling in all
of these cases. It doesn't block all potential paths for running code in the
suspended windows, though, so the next patch explicitly prevents the
problematic reentrancy.
Differential Revision: https://phabricator.services.mozilla.com/D57666
--HG--
extra : moz-landing-system : lando
IPC crashes if you try to send large messages. The message manager
already drops messages that are too big (and used to collect telemetry
for which messages were being dropped).
I think that some message manager messages have been ported over to JS
window actors and started crashing on beta, so let's just give the
same behavior for JS window actors as we do for the message manager.
Differential Revision: https://phabricator.services.mozilla.com/D57583
--HG--
extra : moz-landing-system : lando
When we open a new window from a content process, we create a nested event
loop to wait for it to be initialized by the parent. The problem with this is
that the OpenWindow code which calls the window provider expects the window to
be in-process and uninitialized, so that it can load its own initial URI into
it, and correctly fulfil the spec-codified contract of window.open(). If
another caller initiates a load in the new window during the nested event
loop, those invariants are broken, and any manner of undefined behavior can
occur.
This patch adds a new flag to the BrowsingContext, marking it as uninitialized
until the end of the nested event loop, and blocking any attempts to load a
new URI into it in the meantime.
Differential Revision: https://phabricator.services.mozilla.com/D57667
--HG--
extra : moz-landing-system : lando
This doesn't solve all problems with potential reentrancy during window.open
nested event loops, but it does improve the situation somewhat. Since any
window in the same BrowsingContextGroup can target any window in the same
group, we need to suspend all windows in the group, not just the root of the
new window's parent. We also need to make sure we suspend all in-process
windows, even if we have out-of-process frames somewhere in the parent chain.
This patch takes care of suspending timeouts and input event handling in all
of these cases. It doesn't block all potential paths for running code in the
suspended windows, though, so the next patch explicitly prevents the
problematic reentrancy.
Differential Revision: https://phabricator.services.mozilla.com/D57666
--HG--
extra : moz-landing-system : lando
There is no functional change with this commit. The default implementation for GetCanceled() is still to check if the status code is a failure.
However, it can be argued that as you had to call Cancel() on the nsIChannel, having to check the nsIHttpChannelInternal interface to determine if you had been canceled in the past was rather a non obvious path.
It makes more sense to check the nsIChannel interface to determine if it's been canceled already and this allows for finer granularity if needed in the future.
Differential Revision: https://phabricator.services.mozilla.com/D55268
--HG--
extra : moz-landing-system : lando
Granted origins cause a third-party tracker browsing context to not get
full first-party storage access after successfully calling the storage
access API or a heuristic granting ephemeral access.
For example, after https://tracker.example calls the storage access API
successfully in the third-party context, they embed
https://other-tracker.example, and that load fails because of ETP
restrictions. Here what happens is that https://other-tracker.example
is mistakenly considered the granted origin, and because such a
permission doesn't exist, access is denied.
Differential Revision: https://phabricator.services.mozilla.com/D57493
--HG--
extra : moz-landing-system : lando
Add a document.addCertException function to about:certerror pages, and use it on the desktop certerror page.
Also, as the CallerIsTrusted* functions expect URLs like about:certerror, but GeckoView error pages are data URLs, and so need to be handled differently for these special error-page methods to be exposed on their documents.
Example usage of document.addCertException:
document.addCertException(
true|false /* true == temporary, false == permanent */
).then(
() => {
location.reload();
},
err => {
console.error(err);
}
);
Differential Revision: https://phabricator.services.mozilla.com/D56974
--HG--
extra : moz-landing-system : lando
Granted origins cause a third-party tracker browsing context to not get
full first-party storage access after successfully calling the storage
access API or a heuristic granting ephemeral access.
For example, after https://tracker.example calls the storage access API
successfully in the third-party context, they embed
https://other-tracker.example, and that load fails because of ETP
restrictions. Here what happens is that https://other-tracker.example
is mistakenly considered the granted origin, and because such a
permission doesn't exist, access is denied.
Differential Revision: https://phabricator.services.mozilla.com/D57493
--HG--
extra : moz-landing-system : lando
Granted origins cause a third-party tracker browsing context to not get
full first-party storage access after successfully calling the storage
access API or a heuristic granting ephemeral access.
For example, after https://tracker.example calls the storage access API
successfully in the third-party context, they embed
https://other-tracker.example, and that load fails because of ETP
restrictions. Here what happens is that https://other-tracker.example
is mistakenly considered the granted origin, and because such a
permission doesn't exist, access is denied.
Differential Revision: https://phabricator.services.mozilla.com/D57493
--HG--
extra : moz-landing-system : lando
This makes LoginManagerPrompter.jsm and LoginManagerAuthPrompter.jsm easier to maintain and switches
LoginManagerPrompter.jsm to be a service with static methods which is easier for GeckoView to implement and is cleaner.
Differential Revision: https://phabricator.services.mozilla.com/D57400
--HG--
rename : toolkit/components/passwordmgr/LoginManagerPrompter.jsm => toolkit/components/passwordmgr/LoginManagerAuthPrompter.jsm
rename : toolkit/components/passwordmgr/nsILoginManagerPrompter.idl => toolkit/components/passwordmgr/nsILoginManagerAuthPrompter.idl
extra : moz-landing-system : lando
Add a document.addCertException function to about:certerror pages, and use it on the desktop certerror page.
Also, as the CallerIsTrusted* functions expect URLs like about:certerror, but GeckoView error pages are data URLs, and so need to be handled differently for these special error-page methods to be exposed on their documents.
Example usage of document.addCertException:
document.addCertException(
true|false /* true == temporary, false == permanent */
).then(
() => {
location.reload();
},
err => {
console.error(err);
}
);
Differential Revision: https://phabricator.services.mozilla.com/D56974
--HG--
extra : moz-landing-system : lando
There is no functional change with this commit. The default implementation for GetCanceled() is still to check if the status code is a failure.
However, it can be argued that as you had to call Cancel() on the nsIChannel, having to check the nsIHttpChannelInternal interface to determine if you had been canceled in the past was rather a non obvious path.
It makes more sense to check the nsIChannel interface to determine if it's been canceled already and this allows for finer granularity if needed in the future.
Differential Revision: https://phabricator.services.mozilla.com/D55268
--HG--
extra : moz-landing-system : lando
foo ? true : false
or:
foo ? false : true
Have always a nicer way to be written.
Differential Revision: https://phabricator.services.mozilla.com/D56842
--HG--
extra : moz-landing-system : lando
This is generally pretty straightforward, and rewrites nearly all calls. It
skips the ones that it can detect using frame script globals like
`sendAsyncMessage`, though.
Differential Revision: https://phabricator.services.mozilla.com/D53740
--HG--
extra : moz-landing-system : lando
This commit implements a simple data replication scheme, where
each content process receives a copy of the relevant session
storage data when navigating, and sending all its session storage
data to the parent process before shutting down.
Differential Revision: https://phabricator.services.mozilla.com/D55662
--HG--
extra : moz-landing-system : lando
The OnContentBlockingEvent is no longer needed once we make the
OnContentBlockingEvent parent only.
Differential Revision: https://phabricator.services.mozilla.com/D55788
--HG--
extra : moz-landing-system : lando
This patch adds an IPC message which allows content process to notify
the OnContentBlockingEvent in the parent process. This is needed because
there are some situations that the content blocking happens in content
processes, so we need to tell the parent process to notify it. Such as,
AntiTrackingCommon::AntiTrackingCommon::AddFirstPartyStorageAccessGrantedFor().
Differential Revision: https://phabricator.services.mozilla.com/D55648
--HG--
extra : moz-landing-system : lando
This patch makes the WindowGlobalParent to be able to send the
OnContentBlockingEvent in the parent process.
Differential Revision: https://phabricator.services.mozilla.com/D55645
--HG--
extra : moz-landing-system : lando
This adds a ContentBlockingLog into the WindowGlobalParent. This
ContentBlockingLog is bascially a copy of the log in the content
process. This log in parent is needed for the OnContentBlockingEvent in
parent process since we need an overview of the content blocking events
for OnContentBlockingEvent. And the ContentBlockingLog exits in the
content process in this stage, so we need to maintain a copy of it to
get the overview of the blocking events in parent.
Differential Revision: https://phabricator.services.mozilla.com/D54929
--HG--
extra : moz-landing-system : lando
This is generally pretty straightforward, and rewrites nearly all calls. It
skips the ones that it can detect using frame script globals like
`sendAsyncMessage`, though.
Differential Revision: https://phabricator.services.mozilla.com/D53740
--HG--
extra : moz-landing-system : lando
If `nsGlobalWindowOuter::ResizeTo()` etc is called before receiving
`BrowserChild::RecvUIResolutionChanged()`, the `aX`, `aY`, `aCx` and `aCy`
values are computed with old scale.
This patch makes `BrowserChild::SendSetDimensions()` set scale in the
remote process and `BrowserParent::RecvSetDimensions()` recompute each
value with current scale.
Differential Revision: https://phabricator.services.mozilla.com/D54150
--HG--
extra : moz-landing-system : lando
This intended to fix some windows builds, but that didn't end up working.
This removes some unused members and such, and fixes some missing includes
that they uncover (whoops).
This was needed because some windows headers used in the sandbox redefine STRICT
(which is used by `StyleContain`) and `TRANSPARENT`, which is used by some WR
stuff.
Differential Revision: https://phabricator.services.mozilla.com/D56317
--HG--
extra : moz-landing-system : lando
This is generally pretty straightforward, and rewrites nearly all calls. It
skips the ones that it can detect using frame script globals like
`sendAsyncMessage`, though.
Differential Revision: https://phabricator.services.mozilla.com/D53740
--HG--
extra : moz-landing-system : lando
This makes it much simpler to track down the source of message handler errors
by linking the cross-process caller chains which sent the problematic message.
Differential Revision: https://phabricator.services.mozilla.com/D50883
--HG--
extra : moz-landing-system : lando
In order to be able to reasonably debug error results from things like
JSWindowActor.sendQuery, we need to be able to clone errors across process
boundaries, so that they can be propagated to the caller that initiated a
query. The standard for the structured clone algorithm does not allow cloning
errors directly, so this patch instead adds a chrome-only wrapper object which
supports structured clone writing, and on reading, automatically decodes to
the error object it wraps. Callers who wish to clone an Error or Exception
object simply need to wrap it in a ClonedErrorHolder before sending.
Differential Revision: https://phabricator.services.mozilla.com/D50881
--HG--
extra : moz-landing-system : lando
This makes it much simpler to track down the source of message handler errors
by linking the cross-process caller chains which sent the problematic message.
Differential Revision: https://phabricator.services.mozilla.com/D50883
--HG--
extra : moz-landing-system : lando
In order to be able to reasonably debug error results from things like
JSWindowActor.sendQuery, we need to be able to clone errors across process
boundaries, so that they can be propagated to the caller that initiated a
query. The standard for the structured clone algorithm does not allow cloning
errors directly, so this patch instead adds a chrome-only wrapper object which
supports structured clone writing, and on reading, automatically decodes to
the error object it wraps. Callers who wish to clone an Error or Exception
object simply need to wrap it in a ClonedErrorHolder before sending.
Differential Revision: https://phabricator.services.mozilla.com/D50881
--HG--
extra : moz-landing-system : lando
* The parent needs to be able to request the child to provide its untrusted
modules telemetry. This is done via `GetUntrustedModulesData`.
* The child needs to be able to determine which of its module loads are trusted,
and which are not. Since the child is sandboxed, it must delegate that work
to the parent process. This is done via `GetModulesTrust`.
Differential Revision: https://phabricator.services.mozilla.com/D53681
--HG--
extra : moz-landing-system : lando
This is generally pretty straightforward, and rewrites nearly all calls. It
skips the ones that it can detect using frame script globals like
`sendAsyncMessage`, though.
Differential Revision: https://phabricator.services.mozilla.com/D53740
--HG--
extra : moz-landing-system : lando
The inclusions were removed with the following very crude script and the
resulting breakage was fixed up by hand. The manual fixups did either
revert the changes done by the script, replace a generic header with a more
specific one or replace a header with a forward declaration.
find . -name "*.idl" | grep -v web-platform | grep -v third_party | while read path; do
interfaces=$(grep "^\(class\|interface\).*:.*" "$path" | cut -d' ' -f2)
if [ -n "$interfaces" ]; then
if [[ "$interfaces" == *$'\n'* ]]; then
regexp="\("
for i in $interfaces; do regexp="$regexp$i\|"; done
regexp="${regexp%%\\\|}\)"
else
regexp="$interfaces"
fi
interface=$(basename "$path")
rg -l "#include.*${interface%%.idl}.h" . | while read path2; do
hits=$(grep -v "#include.*${interface%%.idl}.h" "$path2" | grep -c "$regexp" )
if [ $hits -eq 0 ]; then
echo "Removing ${interface} from ${path2}"
grep -v "#include.*${interface%%.idl}.h" "$path2" > "$path2".tmp
mv -f "$path2".tmp "$path2"
fi
done
fi
done
Differential Revision: https://phabricator.services.mozilla.com/D55442
--HG--
extra : moz-landing-system : lando
* The parent needs to be able to request the child to provide its untrusted
modules telemetry. This is done via `GetUntrustedModulesData`.
* The child needs to be able to determine which of its module loads are trusted,
and which are not. Since the child is sandboxed, it must delegate that work
to the parent process. This is done via `GetModulesTrust`.
Differential Revision: https://phabricator.services.mozilla.com/D53681
--HG--
extra : moz-landing-system : lando
The inclusions were removed with the following very crude script and the
resulting breakage was fixed up by hand. The manual fixups did either
revert the changes done by the script, replace a generic header with a more
specific one or replace a header with a forward declaration.
find . -name "*.idl" | grep -v web-platform | grep -v third_party | while read path; do
interfaces=$(grep "^\(class\|interface\).*:.*" "$path" | cut -d' ' -f2)
if [ -n "$interfaces" ]; then
if [[ "$interfaces" == *$'\n'* ]]; then
regexp="\("
for i in $interfaces; do regexp="$regexp$i\|"; done
regexp="${regexp%%\\\|}\)"
else
regexp="$interfaces"
fi
interface=$(basename "$path")
rg -l "#include.*${interface%%.idl}.h" . | while read path2; do
hits=$(grep -v "#include.*${interface%%.idl}.h" "$path2" | grep -c "$regexp" )
if [ $hits -eq 0 ]; then
echo "Removing ${interface} from ${path2}"
grep -v "#include.*${interface%%.idl}.h" "$path2" > "$path2".tmp
mv -f "$path2".tmp "$path2"
fi
done
fi
done
Differential Revision: https://phabricator.services.mozilla.com/D55442
--HG--
extra : moz-landing-system : lando
* Removed the STATUS_LINK constant.
* Removed the statusType parameter from nsIWebBrowserChrome::setStatus.
* Removed the setStatusWithContext method. No one uses the information from
statusContext.
* Removed the nsIWebBrowserChrome2 interface as setStatusWithContext was the
only method.
Differential Revision: https://phabricator.services.mozilla.com/D55195
--HG--
extra : moz-landing-system : lando
Looks like this can sometimes fail with moz-extension URIs, so we shouldn't crash
Differential Revision: https://phabricator.services.mozilla.com/D54249
--HG--
extra : moz-landing-system : lando
Without DocumentChannel, nsExtProtocolChannel::OpenURL calls into nsExternalHelperAppService::LoadURI in the content process.
We then manually forward this to the parent process over PContent, create a RemoteWindowContext around the browser parent, and then call LoadURI again.
With DocumemntChannel, the nsExtProtocolChannel already lives in the parent, so we just need to provide a RemoteWindowContext directly (that the code accesses via GetInterface on the callbacks).
Differential Revision: https://phabricator.services.mozilla.com/D54247
--HG--
extra : moz-landing-system : lando
Looks like this can sometimes fail with moz-extension URIs, so we shouldn't crash
Differential Revision: https://phabricator.services.mozilla.com/D54249
--HG--
extra : moz-landing-system : lando
Without DocumentChannel, nsExtProtocolChannel::OpenURL calls into nsExternalHelperAppService::LoadURI in the content process.
We then manually forward this to the parent process over PContent, create a RemoteWindowContext around the browser parent, and then call LoadURI again.
With DocumemntChannel, the nsExtProtocolChannel already lives in the parent, so we just need to provide a RemoteWindowContext directly (that the code accesses via GetInterface on the callbacks).
Differential Revision: https://phabricator.services.mozilla.com/D54247
--HG--
extra : moz-landing-system : lando
Looks like this can sometimes fail with moz-extension URIs, so we shouldn't crash
Differential Revision: https://phabricator.services.mozilla.com/D54249
--HG--
extra : moz-landing-system : lando
Without DocumentChannel, nsExtProtocolChannel::OpenURL calls into nsExternalHelperAppService::LoadURI in the content process.
We then manually forward this to the parent process over PContent, create a RemoteWindowContext around the browser parent, and then call LoadURI again.
With DocumemntChannel, the nsExtProtocolChannel already lives in the parent, so we just need to provide a RemoteWindowContext directly (that the code accesses via GetInterface on the callbacks).
Differential Revision: https://phabricator.services.mozilla.com/D54247
--HG--
extra : moz-landing-system : lando
This change saves a step in several places around the tree and is
epsilon more efficient.
Differential Revision: https://phabricator.services.mozilla.com/D54545
--HG--
extra : moz-landing-system : lando
I also removed the unused forward declaration of GMPLoader. It used to
be needed for the declaration of XRE_InitChildProcess.
Differential Revision: https://phabricator.services.mozilla.com/D54819
--HG--
extra : moz-landing-system : lando
This was added for XUL fennec, which was dropped at the time of Firefox 7.
But we're still reading this pref and env var and system info data, so let's
stop doing that (also because 'cpucount' is unlikely to be 1 on modern
desktop machines, we're dropping arbitrary user pref sets for geckoview,
and I don't think it's even possible to set an env var for app execution
on android as a user - certainly not easy).
Depends on D55265
Differential Revision: https://phabricator.services.mozilla.com/D55266
--HG--
extra : moz-landing-system : lando
The change to moz.build unveiled a couple unified build issues that I also had
to fix.
Depends on D55365
Differential Revision: https://phabricator.services.mozilla.com/D55366
--HG--
extra : moz-landing-system : lando
There's just no use for it, it was an unnecessary virtual interface, with a single concreate class and only ever used in C++.
Differential Revision: https://phabricator.services.mozilla.com/D55069
--HG--
extra : moz-landing-system : lando
And clean-up surrounding code by removing useless namespace qualification.
It seems like the first call to Broadcast is useless? Or am I missing something?
Differential Revision: https://phabricator.services.mozilla.com/D55000
--HG--
extra : moz-landing-system : lando
This allows test toolkit/components/places/tests/browser/browser_multi_redirect_frecency.js and others to pass when fission is enabled.
The content process expects to know the chain of redirects encountered while opening a URI. The DocumentChannelParent gather that information and sends it to the new ContentChild which can then propagate the information to the new nsDocShell.
The data used to only be passed around during same-origin redirects when fission mode was enabled.
In order to allow for move semantics and preventing unnecessary copy of the DocumentChannelRedirect array, we make the nsIChildProcessChannelListener::onChannelReady property C++ only (noscript).
As we have only one concrete nsIChildProcessChannelListener class (ChildProcessListener) and that the unique OnChannelReady implementation is infallible; we really don't need for the method to return nsresult (so we declare it nostdcall). This helps simplify that call.
Differential Revision: https://phabricator.services.mozilla.com/D54909
--HG--
extra : moz-landing-system : lando
A typo prevented for moves to actually occur. Fix other IPDL entry points while at it that could benefit from move semantics.
Depends on D54907
Differential Revision: https://phabricator.services.mozilla.com/D54908
--HG--
extra : moz-landing-system : lando
Rather that setting the property bag on both the child and parent from the docshell; we first set it on the parent instead, and once the redirect (or process switch) has completed we carry that bag across.
Differential Revision: https://phabricator.services.mozilla.com/D53924
--HG--
extra : moz-landing-system : lando
Looks like this can sometimes fail with moz-extension URIs, so we shouldn't crash
Differential Revision: https://phabricator.services.mozilla.com/D54249
--HG--
extra : moz-landing-system : lando
Without DocumentChannel, nsExtProtocolChannel::OpenURL calls into nsExternalHelperAppService::LoadURI in the content process.
We then manually forward this to the parent process over PContent, create a RemoteWindowContext around the browser parent, and then call LoadURI again.
With DocumemntChannel, the nsExtProtocolChannel already lives in the parent, so we just need to provide a RemoteWindowContext directly (that the code accesses via GetInterface on the callbacks).
Differential Revision: https://phabricator.services.mozilla.com/D54247
--HG--
extra : moz-landing-system : lando
It's a better name, and will avoid confusion when I add other stylesheet caches
outside of the CSS loader.
Depends on D54556
Differential Revision: https://phabricator.services.mozilla.com/D54557
--HG--
rename : layout/style/nsLayoutStylesheetCache.cpp => layout/style/GlobalStyleSheetCache.cpp
rename : layout/style/nsLayoutStylesheetCache.h => layout/style/GlobalStyleSheetCache.h
extra : moz-landing-system : lando
This function works on all GeckoProcessTypes, not just those for child
processes.
Differential Revision: https://phabricator.services.mozilla.com/D54375
--HG--
extra : moz-landing-system : lando
This function works on all GeckoProcessTypes, not just those for child
processes.
Differential Revision: https://phabricator.services.mozilla.com/D54375
--HG--
extra : moz-landing-system : lando
In short - if a user forcibly terminates the browser because it seems
to be permanently hung, we currently do not get a change to record the
hang. This is unfortunate, because these likely represent the most
egregious hangs in terms of user frustration. This patch seeks to
address that.
If a hang exceeds 8192ms (the current definition of a "permahang" in
existing BHR terms), then we decide to immediately persist it to disk,
in case we never get a chance to return to the main thread and
submit it. On the next start of the browser, we read the file from
disk on a background thread, and just submit it using the normal
mechanism.
Regarding the handling of the file itself, I tried to do the simplest
thing I could - as far as I can tell there is no standard simple
serialization mechanism available directly to C++ in Gecko, so I just
serialized it by hand. I didn't take any special care with endianness
or anything as I can't think of a situation in which we really care
at all about these files being transferable between architectures. I
directly used PR_Write / PR_Read instead of doing something fancy
like memory mapping the file, because I don't think performance is a
critical concern here and it offers a simple protection against
reading out of bounds.
Differential Revision: https://phabricator.services.mozilla.com/D52566
--HG--
extra : moz-landing-system : lando
The dynamic toolbar transition doesn't affect on background tabs since to
switch tabs the dynamic toolbar should be restored to its original state (i.e.,
completely visible state).
Differential Revision: https://phabricator.services.mozilla.com/D52336
--HG--
extra : moz-landing-system : lando
In short - if a user forcibly terminates the browser because it seems
to be permanently hung, we currently do not get a change to record the
hang. This is unfortunate, because these likely represent the most
egregious hangs in terms of user frustration. This patch seeks to
address that.
If a hang exceeds 8192ms (the current definition of a "permahang" in
existing BHR terms), then we decide to immediately persist it to disk,
in case we never get a chance to return to the main thread and
submit it. On the next start of the browser, we read the file from
disk on a background thread, and just submit it using the normal
mechanism.
Regarding the handling of the file itself, I tried to do the simplest
thing I could - as far as I can tell there is no standard simple
serialization mechanism available directly to C++ in Gecko, so I just
serialized it by hand. I didn't take any special care with endianness
or anything as I can't think of a situation in which we really care
at all about these files being transferable between architectures. I
directly used PR_Write / PR_Read instead of doing something fancy
like memory mapping the file, because I don't think performance is a
critical concern here and it offers a simple protection against
reading out of bounds.
Differential Revision: https://phabricator.services.mozilla.com/D52566
--HG--
extra : moz-landing-system : lando
These two functions both have a large number of arguments that are largely identical (and ones missing from the PContent one are mostly oversights).
Unifying both to take a single argument and return the same result means we can unify the serialization and return value (promise) handling.
Differential Revision: https://phabricator.services.mozilla.com/D53382
--HG--
extra : moz-landing-system : lando
This involves two new IPC messages (both async) to propagate the change in
compositor options (of which APZ enablement is one) from the GPU process to
the parent process (via PCompositorBridge) and on to the content process
(via PBrowser).
The support is only partial, in that going from non-APZ to APZ is only
supported if APZ was enabled at the time the window was created.
Depends on D51467
Differential Revision: https://phabricator.services.mozilla.com/D51468
--HG--
extra : moz-landing-system : lando
In short - if a user forcibly terminates the browser because it seems
to be permanently hung, we currently do not get a change to record the
hang. This is unfortunate, because these likely represent the most
egregious hangs in terms of user frustration. This patch seeks to
address that.
If a hang exceeds 8192ms (the current definition of a "permahang" in
existing BHR terms), then we decide to immediately persist it to disk,
in case we never get a chance to return to the main thread and
submit it. On the next start of the browser, we read the file from
disk on a background thread, and just submit it using the normal
mechanism.
Regarding the handling of the file itself, I tried to do the simplest
thing I could - as far as I can tell there is no standard simple
serialization mechanism available directly to C++ in Gecko, so I just
serialized it by hand. I didn't take any special care with endianness
or anything as I can't think of a situation in which we really care
at all about these files being transferable between architectures. I
directly used PR_Write / PR_Read instead of doing something fancy
like memory mapping the file, because I don't think performance is a
critical concern here and it offers a simple protection against
reading out of bounds.
Differential Revision: https://phabricator.services.mozilla.com/D52566
--HG--
extra : moz-landing-system : lando
Preventing RemoteWorkerService from existing in "web COOP+COEP" content
processes prevents Shared/ServiceWorkers from being spawned in them because there
won't be an associated RemoteWorkerServiceParent that registers with the parent
process RemoteWorkerManager.
Differential Revision: https://phabricator.services.mozilla.com/D50815
--HG--
extra : moz-landing-system : lando
Andrea Marchesini
Andrew McCreight
Aaron Klotz
Ciure Andrei
Arnaud Renevier
Nika Layzell
Bogdan Tara
Frederik Braun
Kashav Madan
Narcis Beleuzu
Kris Maglione
Mihai Alexandru Michis
Csoregi Natalia
Dimi Lee
Sylvestre Ledru
Razvan Maries
Emilio Cobos Álvarez
John Dai
Andreas Farre
Valentin Gosu
Cosmin Sabou
Luca Greco
Zibi Braniecki
Jean-Yves Avenard
alwu
pbz
Anny Gakhokidze
David Teller
Eric Rahm
Tim Huang
Matt Woodrow
David Parks
Yaron Tausky
Brian Hackett
Dorel Luca
Boris Zbarsky
shindli
shravanrn@gmail.com
Gabriele Svelto
Hiroyuki Ikezoe
Ehsan Akhgari
Gurzau Raul
Thomas Wisniewski
Matthew Noorenberghe
Noemi Erli
Oana Pop Rus
Johann Hofmann
Masayuki Nakano
Kagami Sascha Rosylight
Daniel Varga
Jim Porter
Masatoshi Kimura
Thomas Nguyen
Anny G
Dave Townsend
Kershaw Chang
Nathan Froyd
Gijs Kruitbosch
Olli Pettay
Coroiu Cristina
Emma Malysz
Brindusan Cristian
Andreea Pavel
Doug Thayer
Perry Jiang
Botond Ballo
Micah Tigley
Jonathan Kew