VAAPI HW surfaces are released at ReleaseUnusedVAAPIFrames() and we use DMABufSurface::IsUsed() flag
to detect unused surfaces. Then we call GetUnusedDMABufSurfaceWrapper() to get unused surfaces to
re-use them for decoding.
Because DMABufSurface::IsUsed() flag is updated asynchronously it may change
between ReleaseUnusedVAAPIFrames() and GetUnusedDMABufSurfaceWrapper() calls.
In that case GetUnusedDMABufSurfaceWrapper() may return unused but also unreleased surface
as it was marked as used in time of ReleaseUnusedVAAPIFrames() call.
In this patch explicitly release VAAPI data of any surface before we re-use it.
Also fail when we try to upload data to already used DMABufSurfaceYUV surface.
Differential Revision: https://phabricator.services.mozilla.com/D85842
A late getEventTimings RDP request is being cancelled while we are exiting the test case and we have already closed the devtools toolbox.
The related network request is very likely unrelated to the scenario being tested (I suspect it may be an HTTP request triggered by some
privileged code and being automatically added to the webconsole panel, it is not unlikely that we are not explicitly checking if the http
requests logged in the webconsole are actually related to the extension that the devtools toolbox is targeting).
In the meantime, the attached patch is adding to the PromiseTestUtils's "global allow list" the rejections for any late RDP request
that may still be pending while the test is exiting.
Differential Revision: https://phabricator.services.mozilla.com/D86000
The assertion crash found by fuzzing is fixed after moving useraction state from
BrowsingContext to WindowContext, see bug 1611961.
Differential Revision: https://phabricator.services.mozilla.com/D85752
Previously we didn't compact ObjectGroups, but doing this would hopfully unlock more memory savings in bug 1654853.
Most of the trickiness here comes from accessing objects whose object group is forwarded, so require special-case handling to get their class, compartment etc.
Differential Revision: https://phabricator.services.mozilla.com/D86033
Add dump method to stencil-related structs, that dumps the struct as JSON,
and `dumpStencil()` shell-builtin as a consumer.
Some structs have dumpFields method, given that JSONPrinter only provide
beginObjectProperty/endObject to print a property with an object value.
There's some design decision about when to use string-representation and when
to use object.
Currently simple single-field struct like ScopeIndex uses string-representation
like "ScopeIndex(0)", and ScriptAtom also uses string-representation.
There's an exception in ScriptAtom case in DumpScriptThing,
that we need to differentiate string and string-representation of other
structs, and for simplicity, ScriptAtom case uses object with "type" property,
and other single-field structs use string-representation.
The current dumpStencil does not dump CompilationInfo.asmJS (FIXME) and
ScriptStencil.immutableScriptData.
Most of ScriptStencil.immutableScriptData field can be dumped by `dis()`
function.
Differential Revision: https://phabricator.services.mozilla.com/D86042
This is the third part of series of patches to Frame without tls pointer.
Here we preserve initial tls in all entry stubs and then use it to find a proper tls instance for a given frame.
To find the TlsData* for specific frame we start from a entry stub's tls
and then track tls through all possible cross-instance calls. This logic
is implemented in GetNearestEffectiveTls procedure.
Then, we use this new procedure to make singal handling free from Frame::tls.
Differential Revision: https://phabricator.services.mozilla.com/D83044
This is a followup patch for removing Frame::tls.
Now, we are preserving caller's and callee's tls'es for all possible cross-instance calls in the previously allocated abi slots.
We also use preserved tls values to restore the caller's tls in Ion. Baseline doesn't need this because it restores the caller tls from its private stack slot after the call.
Differential Revision: https://phabricator.services.mozilla.com/D82888
We are going to remove Frame::tls and support trampolines for indirect calls, so we need to get rid of using Frame::tls.
In this and the followup patches I will iteratively remove all dependencies of Frame::tls and remove it eventually.
In this patch I changed wasm ABI to allocate two stack slots after stack args to preserve caller's and callee's tls'es in the near future.
Differential Revision: https://phabricator.services.mozilla.com/D82881
On Linux PID_MAX_LIMIT is 2^22. By default pid_max was set to 2^15, but
systemd since version 243 sets it to 2^22.
nsHttpHandler::NewChannelId uses PID shifted by 32 bits to generate
channelId, and that results in values greater than 2^53.
Such values are fine for uint64_t, but when we pass channelId to
devtools it gets converted to JavaScript floats, and values greater
than 2^53 get errors in lower two bits. This in turn leads to much
sorrow, like request id collisions or "no headers for this request".
This patch shifts the PID by one bit less, resulting in channelIds
that always fit into safe JavaScript integer range for all PIDs.
Differential Revision: https://phabricator.services.mozilla.com/D85990
Add a modular approach for the integration of `static-analysis` module in order
to be able to share components of it with other modules, like the integration of
`clangd` in `vscode` where we need to have access to the configuration of `clang-tidy`
in order to have `in-ide` `static-analysis` messages.
In this initial step we make a separate module for the clang-tidy configuration.
Differential Revision: https://phabricator.services.mozilla.com/D85979
In order to have a cross platform ide for C++ language support we've added `clangd`
extenssion and artifact part of `vscode` suite.
To generate the configuration you simply run:
`./mach ide vscode `.
Differential Revision: https://phabricator.services.mozilla.com/D85416
Use assertions for consistency with other intrinsics which are already asserting
that the correct number of arguments with the correct types are used.
Differential Revision: https://phabricator.services.mozilla.com/D85854
The first intrinsics optimised in CacheIR were still using emitNativeCalleeGuard.
Later one we've removed this guard.
Differential Revision: https://phabricator.services.mozilla.com/D85853
Similar to Ion only optimise the case when the object isn't actually a wrapper.
We don't have a separate MArrayBufferByteLength MIR node, so use the same
approach as used in IonBuilder and use a fixed slot load to optimise the
call.
Differential Revision: https://phabricator.services.mozilla.com/D85851
Add dump method to stencil-related structs, that dumps the struct as JSON,
and `dumpStencil()` shell-builtin as a consumer.
Some structs have dumpFields method, given that JSONPrinter only provide
beginObjectProperty/endObject to print a property with an object value.
There's some design decision about when to use string-representation and when
to use object.
Currently simple single-field struct like ScopeIndex uses string-representation
like "ScopeIndex(0)", and ScriptAtom also uses string-representation.
There's an exception in ScriptAtom case in DumpScriptThing,
that we need to differentiate string and string-representation of other
structs, and for simplicity, ScriptAtom case uses object with "type" property,
and other single-field structs use string-representation.
The current dumpStencil does not dump CompilationInfo.asmJS (FIXME) and
ScriptStencil.immutableScriptData.
Most of ScriptStencil.immutableScriptData field can be dumped by `dis()`
function.
Differential Revision: https://phabricator.services.mozilla.com/D86042
This is a followup patch for removing Frame::tls.
Now, we are preserving caller's and callee's tls'es for all possible cross-instance calls in the previously allocated abi slots.
We also use preserved tls values to restore the caller's tls in Ion. Baseline doesn't need this because it restores the caller tls from its private stack slot after the call.
Differential Revision: https://phabricator.services.mozilla.com/D82888
We are going to remove Frame::tls and support trampolines for indirect calls, so we need to get rid of using Frame::tls.
In this and the followup patches I will iteratively remove all dependencies of Frame::tls and remove it eventually.
In this patch I changed wasm ABI to allocate two stack slots after stack args to preserve caller's and callee's tls'es in the near future.
Differential Revision: https://phabricator.services.mozilla.com/D82881
Contrary to python2, python3 considers the values in cl_lnotab to be signed
integers, so with python3, offsets larger than 127 would be encoded in a way
that would make them wrong, or worse, negative.
Differential Revision: https://phabricator.services.mozilla.com/D86122
Edgar Chen
Tom Schuster
Martin Stransky
Gregory Mierzwinski
Csoregi Natalia
Luca Greco
Mark Banner
Geoff Lankow
Jon Coppeard
Tooru Fujisawa
Dmitry Bezhetskov
Hiroyuki Ikezoe
Andi-Bogdan Postelnicu
Matt Woodrow
Cosmin Sabou
Platon Pronko
André Bargull
Erik Nordin
Butkovits Atila
Mike Hommey
Botond Ballo
Itiel
Karl Tomlinson
Kagami Sascha Rosylight