| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
| |
We need to export dummy symbols to keep C-ABI
Unfortuinately symbol version assembly does not work
if these symbols get redefined here.
These extra allocations are checking and logging "Out of memory"
Task-number: QTBUG-61521
Change-Id: Iea8c7cbf873cee40921000720451653842b03c07
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This makes Chromium Content API to be able to notify about text
selection changes triggered by non-user events (eg. JavaScript, IME,
autofill).
Based on:
https://codereview.chromium.org/2903833002
Corresponding Chromium bug report:
https://bugs.chromium.org/p/chromium/issues/detail?id=671986
Change-Id: I86a7f203d789853199469b301facd06fe5ba54bd
Task-number: QTBUG-53134
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
| |
Task-number: QTBUG-61954
Change-Id: I61bb677293fb2769b4a4499b7127939b6cf18d70
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
http://crrev.com/2384163002 skipped initializing RenderSandboxHostLinux
If the sandbox and zygote are both disabled. But the RendererSocket is
used even in this case by the Linux ChildProcessLauncherHelper, so it
lead to a crash.
Instead initialize the RenderSandboxHostLinux unconditionally. This
appears to be what some code expects in multi-process code.
Tested:
cr run chrome --no-zygote --no-sandbox http://www.google.com
BUG=712779
Change-Id: I391205092c13c6c2a4e7c59c74729d2b068dcb9f
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Remove logic to include the scanning in the .gn files.
This depends on Android code currently stripped from qtwebengine.git.
Also, we don't need it, since we scan as part of the documentation
generation.
Task-number: QTBUG-61693
Change-Id: I868fd35ce48169da625f16b903134a4e62be5a31
Reviewed-by: Viktor Engelmann <viktor.engelmann@qt.io>
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The problem was that sometimes a find reply would come in from a frame that
has already been removed. This was assumed not to be possible (which is the
reason for the original DCHECK present). The right thing to do is just
ignore these replies, so the DCHECK has been turned into an early exit.
BUG=624738
Review-Url: https://codereview.chromium.org/2399923002
Cr-Commit-Position: refs/heads/master@{#458551}
Change-Id: I94f27a4f17a5a27b56a78c09e1b9b5e8f059f2a2
Task-number: QTBUG-60472
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
|
|
|
|
|
|
|
| |
These parts have been moved to components/spellcheck
Change-Id: I73f16bf7b490a32ba29de173ade9f1a082e60d01
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
|
|
|
|
|
|
|
|
| |
The windows fonts warmup was using a PPAPI locks.
Task-number: QTBUG-61170
Change-Id: If8370667091ead3c5b96f2661796892395fdd76b
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
|
|
|
| |
Return MEDIA_DEVICE_NOT_SUPPORTED to JavaScript instead of triggering DCHECK in
MediaStreamManager::SetupDeviceCaptureRequest.
Task-number: QTBUG-60832
Change-Id: Iadb99244fc4345580faa1a3a1af3264d91c3fa8f
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Chromium asserts while logging the GpuHostMsg_ChannelEstablished
message. No other message is known to cause this malfunction. Replace
assert with an error message as a workaround.
Most probably this workaround is not needed in newer Chromium versions
than 56-based since the corresponding message has been removed in
upstream while switching to mojom API.
Change-Id: I0671bef088db410a8fa2aaeca4df16b37fb462f8
Task-number: QTBUG-60924
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Service Workers may be loaded over different schemes than HTTP(S)
(chromium itself provides interfaces to register schemes for this!).
Despite this, chromium assumes in many places, that Service Workers
were always loaded over HTTP(S). The code for registering Service
Workers often blindly dereferences the requests "header" pointer
without checking it for NULL beforehand and assumes the status code
to be an HTTP status (implying an error if it is not between
200 and 299).
Task-number: QTBUG-56169
Change-Id: Id63789bbe974e1958981b651ccc15cfbb067b4af
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When v8_use_external_startup_data is disabled natives_blob.bin and
snapshot_blob.bin external files are not generated. Skip try to open
them for their file descriptors to suppress "Ignore invalid file" error
messages due to non-existent files.
This change adds back guards removed in
https://codereview.chromium.org/2684433003
Change-Id: I0bb93c0e05df18905083423a84d1e56c22232e4d
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
| |
GCC 4.8 doesn't ship with the stdatomic.h header.
Change-Id: I702194330478b56dbbbb1720e725669543a001ba
Reviewed-by: Peter Varga <pvarga@inf.u-szeged.hu>
|
|
|
|
|
| |
Change-Id: If646130fb9327d8b82aa9d8c25d7639fc89d5bbf
Reviewed-by: Peter Varga <pvarga@inf.u-szeged.hu>
|
|
|
|
|
|
|
|
| |
An operator<= must return true on identity comparisons. This is asserted
by libstdc++ in debug mode.
Change-Id: I1fbc98f2742a960e0fa8d5873ab4afd7cb73cc2b
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
|
|
|
|
|
|
|
|
| |
It causes inconsistency in the database of quota permissions.
Chromium registers origins without paths and later tries to query
them including paths.
Change-Id: I4f905a52afb5ceba8bbf854c2e23d4b070526fec
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
On some platforms we get error 35 (deadlock avoided) on sandboxed
machines when extra checking is enabled in the pthread locks. These
appears to be false positives as no deadlocks occur when checking is
disabled.
For now we disable the checking to fix debug builds on Linux.
Task-number: QTBUG-60889
Change-Id: Ic5bcde246597e06fc920a496f2ad9a1ae1298808
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
|
|
|
|
|
|
|
|
|
|
|
| |
Fix compilation with older Linux kernel's (e.g. Red Hat 6.6).
The enum was introduced in Linux kernel version 3.2 (kernel commit
d26a6635b2421).
Task-number: QTBUG-48298
Task-number: QTBUG-57014
Change-Id: I05b86520fc162d55d043d70d1b92aea5f7a61bc5
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Service Workers may be loaded over different schemes than HTTP(S)
(chromium itself provides interfaces to register schemes for this!).
Despite this, chromium assumes here, that Service Workers
were always loaded over HTTP(S). The code for registering Service
Workers blindly dereferences the requests "header" pointer
without checking it for NULL beforehand.
Task-number: QTBUG-56169
Change-Id: If7ba87fbf0cc032cc3b3c34d9d4e0d1041c93178
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
|
|
|
|
|
|
|
| |
Fix a few issues triggered while trying to build on mac after the
rebase on Chromium 58
Change-Id: I998d104862d807caa6b176c62bebc9e57461968b
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
|
|
|
|
|
|
|
| |
This file doesn't exist in sources, and the change was lost in rebasing
Change-Id: I33d604ecdac86db97bf7f29f9d759f34bf6ef17c
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
|
|
|
|
|
|
|
| |
The compression uses android sdk files we don't have.
Change-Id: Iafb3753796122d8304a8c873e724986cd24c298d
Reviewed-by: Peter Varga <pvarga@inf.u-szeged.hu>
|
|
|
|
|
|
|
| |
Part lost in rebase.
Change-Id: Id6851ea0678ec8e85ea75913977776e4d014c36c
Reviewed-by: Peter Varga <pvarga@inf.u-szeged.hu>
|
|\
| |
| |
| | |
Change-Id: I8c280ca9068fdda9cf6276725bfb8608eccb497a
|
| |
| |
| |
| |
| | |
Change-Id: Ib913750d687f29e036959c62a0cabf2412a8faa7
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
|
| |
| |
| |
| |
| |
| |
| | |
Upstream broke building against newer libpng versions.
Change-Id: Ice1979e28b1bcd68751fba23c2321cf3c4a34120
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
platforms
BUG=618739
Review-Url: https://codereview.chromium.org/2748993004
Change-Id: Iefc50ebc75c200c304c0060ac1dbe91aec31f674
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The deployment target will always be higher than 10.9, because
Qt 5.9 requires it to be at least 10.10.
Task-number: QTBUG-60438
Change-Id: I5a97ba828f7267731e50010c4345caa19bb8939a
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Do not create references within returned temporary objects.
OutputDeviceInfo is returned by value so creating reference to
AudioParameters will end up in best case in corrupted
parameters and pipeline initialization failed.
Change-Id: I754642a4bb051a2556648d4b2f0133ab0ffdf669
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
If bootstrap.py is called with a python2 executable but the first python
executable in PATH is a python3, then the gn-build of gn will fail.
Fix this by passing the exact python location to gn it was bootstrapped
with.
Change-Id: I17d9cd3614c3a3bee5d54a08a27b9e08d29ada4d
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
When a touchpad is used on macOS, Qt receives both scroll wheel
events and touch events, which it forwards to Chromium.
The gesture detector in Chromium transforms the touch events into
scroll / fling gestures, and these interfere with the original scroll
wheel events from Qt, leading to broken scrolling.
This used to work in Qt 5.7 and lower because the scroll wheel events
had a higher precedence than gestures, but this is not the case
with newer versions of Chromium.
Furthermore the gesture detector in Chromium has the assumption that
it receives touch events from a touchscreen device, and not from a
touchpad, which implies different gesture behaviors (moving one finger
on a touchpad moves the mouse, whereas on a touchscreen it causes the
screen to move, aka scrolling). Thus it is not correct to use the
gesture detector for handling macOS touchpad events.
The fix is a temporary hack which discards scroll and fling gestures
on macOS, thus processing only wheel events received from Qt.
The proper fix will be to remove usage of the gesture detector on
macOS, and rather transform and forward QNativeGestures which
are received directly from the operating system, similarly to how
the Chrome browser handles it.
Task-number: QTBUG-58779
Change-Id: I1b1fc950d033c175aebd617d38591bf5d94b5adc
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| |
| |
| |
| |
| | |
Change-Id: I34d99968d134bfca48b0ac89ff1fa171084561df
Reviewed-by: Kai Koehne <kai.koehne@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Fixes a few problems with using undeclared functions and ambigious
code.
Task-number: QTBUG-59776
Change-Id: Ifcaecda4d210c52362616d984bd1259fc763874c
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The DCHECK in quota_manager.cc ensures that only the standard origins
can request persistent storage. This patch removes the DCHECK and allows
qrc:/ and custom URL scheme handlers to use quotas.
Change-Id: I68a5915d66599fbb3ea844fe1ec3f0b646cbc7b8
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| | |
As we don't build with the exact same clang version as Google, clang
doesn't necessarily support all the C++ attributes used.
Change-Id: I8992fb255c803e2c03ab859cffa050499034d20e
Reviewed-by: Alexandru Croitor <alexandru.croitor@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Make sure that the target sysroot is not used with host toolchain,
even when os and cpu match.
Task-number: QTBUG-59768
Change-Id: Ib52ca2c3d0f61aaed4567d15185af923b43be9fc
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Due to change 8ae5d7f0 in chromium DWriteFontProxy is now
also initialized when not using the sandbox. This triggers
on win10 DCHECK which fails if font fallback is supported,
but reported windows version (due to missing correct
manifest file) is 8.0. Skip the DCHECK for qtwebengine
builds.
Change-Id: I6fcef46b1e0f0dc8fec8fec62cc46322efcaaad9
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
There have been issues with too long file paths in grit.py
when doing shadow builds on Windows.
Change-Id: Ia0ff5e214bcd079853db32432b58da046698e73e
Task-number: QTBUG-59544
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| | |
Older versions of the eventlog message compiler can not handle UTF-8
encoded messages, so convert to UTF-16.
Change-Id: I63cb4745d2f77fa269269807e0c99f7c2a57b1c8
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| | |
Treat qrc origins like file origins to work-around GURL being terrible
at handling non-http urls.
Change-Id: Ic41a620fb2616c0272c104e707a188b75f10ac0b
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
|
| |
| |
| |
| |
| | |
Change-Id: Ic6d1c1d1d10853b357f07bbb5408afd4ed03d84b
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
|
| |
| |
| |
| |
| |
| |
| | |
We do not need these just because we use most of Aura.
Change-Id: I49f7894bcc6b18b8719ebf59927154730f9e54d2
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
| |
| |
| |
| |
| | |
Change-Id: I724bc1e58c8e988fa5e3afbcc6075ae7eb234843
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
|
| |
| |
| |
| |
| | |
Change-Id: If62418e29741cef6790690eba2b4c890cd8fcaee
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| | |
Add option to add external ozone platform. Assumes the external ozone
platform is also set as the default ozone platform.
Change-Id: I683b56828f9adbc920fcfc3f63aa3ab8b78a6084
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
|
| |
| |
| |
| |
| |
| |
| | |
Introduce use_libcxx gn arg to explicitly ask for -stdlib=libc++
Change-Id: I5b8449d937c8af848b9f89c6a9879742b83f46d6
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
|
| |
| |
| |
| |
| | |
Change-Id: I406d1c85de39cbef013f651eb1beadc619e60154
Reviewed-by: Joerg Bornemann <joerg.bornemann@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| | |
Merge unbundle version of ffmpeg.gn and disable code that accesses the
avformat internal struct.
Change-Id: I0a8923e9ba72f2da6d5e2d1834f63c2a7c740b37
Reviewed-by: Michal Klocek <michal.klocek@qt.io>
|
| |
| |
| |
| |
| |
| |
| | |
Some headers have moved from i18n to common.
Change-Id: I1aa6775dd293fa21fbef02eeb521d553a804498b
Reviewed-by: Michael Brüning <michael.bruning@qt.io>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Use just header names wihtout directory prefix as a source
file for shim headers generation. Add 'prefix' instead so
'process_file_template' can workout correctly 'output'
and 'input' dependencies.
Change-Id: Id71a55edb68fc1b30e756e0713681fe150d84055
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>
|