Because the .xdata format on ARM64 can only encode sizes up to 1M (much too small for our JIT code regions), we register a function table callback to provide RUNTIME_FUNCTIONs at runtime. Windows doesn't seem to care about the size fields on RUNTIME_FUNCTIONs that are created in this way, so the same RUNTIME_FUNCTION can work for any address in the region. We'll set up a generic one in RegisterExecutableMemory and the callback can just return a pointer to it.
Differential Revision: https://phabricator.services.mozilla.com/D16261
--HG--
extra : moz-landing-system : lando
This patch removes the StopWatch code that was used in the first version of
about:performance, and not being used anymore.
Differential Revision: https://phabricator.services.mozilla.com/D7453
--HG--
extra : moz-landing-system : lando
These tests mostly use either the debugger (requires separate compartemnts for
debugger/debuggee) or require a new compartment for things like nukeAllCCWs().
Differential Revision: https://phabricator.services.mozilla.com/D16172
--HG--
extra : moz-landing-system : lando
I added this optimization in bug 1299107 to share more shapes across
compartments. Unfortunately this doesn't play well with same-compartment
realms (ICs can misbehave) because it relies on compartments being isolated
from each other.
I think we should remove this optimization:
* Fixing the IC issue is impossible without deoptimizing everything.
* I added it mainly for chrome globals. The shared-JSM-global work has eliminated
the need for this there.
* Same-compartment realms win memory back by eliminating CCWs etc.
* It's quite a lot of complicated code.
Differential Revision: https://phabricator.services.mozilla.com/D16170
--HG--
extra : moz-landing-system : lando
We want to use this shell flag in automation. Some globals really need their
own compartment so tests can use newGlobal({newCompartment: true}) to opt-out.
Differential Revision: https://phabricator.services.mozilla.com/D16166
--HG--
extra : moz-landing-system : lando
Bindgen is only used when building js or toolkit, so we only need to
include the configure part in js/moz.configure, which is included in
both cases.
Depends on D16293
Differential Revision: https://phabricator.services.mozilla.com/D16294
--HG--
extra : moz-landing-system : lando
This patch was intended to be a pure refactoring of existing code with
no side-effects, moving the logic for handling for-of/for-of-iterclose
trynotes inside TryNoteIter to avoid duplicating logic in all users of
TryNoteIter. However, it turns out that there was a subtle preexisting
bug in TryNoteIter that is fixed by the refactoring. Specifically, the
logic to skip from a for-of-iterclose to its enclosing for-of must run
before the logic to skip trynotes based on stack depth. Otherwise, the
stack depth code may filter out the enclosing for-of (see the attached
test case for an example) and we will skip too many try-notes.
Differential Revision: https://phabricator.services.mozilla.com/D14783
--HG--
extra : moz-landing-system : lando
This also adds a GetVariablesObject helper so we don't have to duplicate the
logic there.
Differential Revision: https://phabricator.services.mozilla.com/D13698
--HG--
extra : moz-landing-system : lando
Only attempt to add breakpoints for the current source in Thread._addSource(),
reducing the number of costly Debugger::findScripts() calls made when loading a
page.
In addition, speed up findScripts() itself by moving the cheaper URL/source
checks in commonFilter() ahead of the line number checks.
Differential Revision: https://phabricator.services.mozilla.com/D15534
--HG--
extra : moz-landing-system : lando
Csoregi Natalia
Jon Coppeard
David Major
Tarek Ziadé
Jan de Mooij
Narcis Beleuzu
Mike Hommey
Noemi Erli
Tooru Fujisawa
Iain Ireland
Razvan Maries
Daniel Varga
Sean Stangl
Ashley Hauck
Coroiu Cristina
André Bargull
wartmanm