Firefox uses multiple processes. It has intentional leaks, and when
running with ASAN, we have suppressions to eliminate those. When running
ASAN builds through CI tests, when Firefox exits, each of the processes
(parent and child) exits and goes through its leaks and when there are
(which is a given), the ASAN runtime runs llvm-symbolizer to symbolicate
and match against suppressions. So each process runs llvm-symbolizer. At
the same time.
Some of the addresses to symbolicate are in libxul. Which contains all
DWARF info, making it a ~1GB monster. Oh, and because you're lucky,
things align perfectly such that libxul size is a multiple of the page
size. That makes llvm-symbolizer pread() the file instead of mmap()ing
it. Did I say there are multiple processes? So suddenly you have n
processes simultaneously allocating and filling 1GB of memory each, on
CI machines that have enough memory for the job they usually run, but
not enough for a sudden rush of n GB.
And things go awry. When you're lucky and the OOM killer didn't take
care of killing the CI entirely, symbolication couldn't happen and the
suppressions are not matched, and leaks are reported.
This all turns out it originates in how llvm-symbolicate chooses between
pread() and mmap(), which turns out is just defaults not being made for
binary files.
Differential Revision: https://phabricator.services.mozilla.com/D16010
--HG--
extra : moz-landing-system : lando
With all the previous efforts, the scrollbox binding now does nothing but to create an inner box element holding the scrolling content.
It turned out that inner box can be easily removed. The padding set by the document sheets can be moved to the srollbox element directly.
The only gotcha is XULScrollElement::ScrollByIndex() -- it can now reach the child item frames directly from the scrolled frame.
Depends on D15169
Differential Revision: https://phabricator.services.mozilla.com/D15170
--HG--
extra : moz-landing-system : lando
Add support for enumerations to boolean types, and use it to only show
a deprecation message when background.persistent is false.
Differential Revision: https://phabricator.services.mozilla.com/D15507
--HG--
extra : moz-landing-system : lando
This can legitimately happen while paused since the watchmanager calling this
is dispatching the calls. As such they're out of sync with the paused state,
and we need to allow updating the time while paused.
FireTimeUpdate does ignore the call if the time hasn't actually been updated,
so the only impact from this is that we could do a lot of unnecessary
dispatching while paused without noticing.
Differential Revision: https://phabricator.services.mozilla.com/D15731
--HG--
extra : moz-landing-system : lando
In order to make the history easier to navigate, this changeset includes the
modifications required to make <xul:browser> actually work as a Custom Element,
and switches the app to use it instead of the XBL browser.
Differential Revision: https://phabricator.services.mozilla.com/D14911
--HG--
extra : moz-landing-system : lando
Launching processes takes enough time that we should avoid blocking the
parent process's IPC I/O thread for it; it's less bad for responsiveness
than blocking the main thread, but it's not good.
On Windows we need to use a dedicated thread, because the sandbox isn't
thread-safe and it asserts that the same thread is used for every
launch. Otherwise, a thread pool is used. (Or, in the Web Replay
middleman process, where there isn't enough of XPCOM for any of this,
launching the actual content processes remains on the I/O thread.)
Depends on D15886
Differential Revision: https://phabricator.services.mozilla.com/D8946
--HG--
extra : moz-landing-system : lando
We can directly set environment variables for the child process on
all platforms now, instead of changing the parent's environment and
inheriting the changes. This simplifies memory management, but more
importantly it's necessary for thread safety to allow launching
processes from a thread pool.
Depends on D8944
Differential Revision: https://phabricator.services.mozilla.com/D8945
--HG--
extra : moz-landing-system : lando
The test is actually redundant with the TEXTREL check in
check_binary.py (R_386_PC32 relocations will only happen as text
relocations)
Differential Revision: https://phabricator.services.mozilla.com/D16002
--HG--
extra : moz-landing-system : lando
Adds some includes that are needed for functions used in nsUpdateDriver.cpp.
Differential Revision: https://phabricator.services.mozilla.com/D16063
--HG--
extra : moz-landing-system : lando
This change updates the structs VRDisplayState, VRLayer_2D_Content, and VRLayer_Stereo_Immersive to not have 'm' prefix for its members.
Differential Revision: https://phabricator.services.mozilla.com/D16000
--HG--
extra : moz-landing-system : lando
I needed to scale the layout device pixels coming from Windows before passing them to ScreenForRect(). Also, I'm using GetRect() directly instead of GetRectDisplayPix() * scale now, to avoid an unnecessary double scale & round which was making fullscreen windows off by one pixel in many cases.
Differential Revision: https://phabricator.services.mozilla.com/D15604
--HG--
extra : moz-landing-system : lando
This introduces a DISABLE_USER_IMPERSONATION define when EARLY_BETA_OR_EARLIER is set, if that is present the maintenance service will not attempt to get an impersonation token for the user's updater process, and the updater will not attempt to use any token it is given. The bulk of the changes are restoring the old failure status codes and the tests that expect them, sorry for the noise that causes.
Differential Revision: https://phabricator.services.mozilla.com/D15781
--HG--
extra : moz-landing-system : lando
gfxUtils::EncodeSourceSurface no longer uses a stringly-typed API to create a
`imgIEncoder` for the relevant MIME type. Instead, we now use an enum class and
switch on it to create the encoder.
Depends on D14816
Differential Revision: https://phabricator.services.mozilla.com/D14817
--HG--
extra : moz-landing-system : lando
mozilla::PrintfTarget::cvt_f release asserts that the desired printf
fit into a statically-sized buffer. However, this may not be the case
if the user requested a larger width or precision. Handle this
unusual case by allocating a temporary buffer.
MozReview-Commit-ID: 2WicecHDzDR
Differential Revision: https://phabricator.services.mozilla.com/D15989
--HG--
extra : moz-landing-system : lando
