Splits PromiseHandle into two TF builtins to account for catch
prediction. An exception in PromiseHandleReject builtin results in a
"caught" prediction whereas an expception in PromiseHandle results in a
"promise rejection" prediction.
An extra is_exception_caught bit is added to Code to mark this catch
prediction behavior.
BUG=v8:5343
Review-Url: https://codereview.chromium.org/2572623002
Cr-Commit-Position: refs/heads/master@{#41683}
The fast-path for dynamic local variables was previously checking the
lookup variable rather than the shadowed variable when deciding whether
to add a hole check.
BUG=669540
Review-Url: https://codereview.chromium.org/2551023004
Cr-Commit-Position: refs/heads/master@{#41677}
It reduces the add operation for certain checked load/stores by pre-do the add in compiling phase.
BUG=
Review-Url: https://codereview.chromium.org/2563483005
Cr-Commit-Position: refs/heads/master@{#41671}
The assignment in default parameters (e.g., function foo(a = somedefault) { }
doesn't flow through PreParserFactory::NewAssignment even if the comment claims so.
Thus, piping through the variables is not needed.
Kudos to neis@ for pointing this out.
BUG=v8:5501
R=neis@chromium.org
Review-Url: https://codereview.chromium.org/2569983003
Cr-Commit-Position: refs/heads/master@{#41670}
There was previously a race between different phases recording their
first entry -- and thus, their insert order -- on the main and
concurrent-compilation thread. This would later manifest as a segfault
when creating the sorted array of phases for --turbo-stats (as two
phases would have the same insert order and so there would be a gap in
the array).
Review-Url: https://codereview.chromium.org/2572713003
Cr-Commit-Position: refs/heads/master@{#41669}
Reason for revert:
Still flakes on ARM.
Original issue's description:
> [deoptimizer] Enable test that should no longer fail.
>
> R=jarin@chromium.org
> TEST=cctest/test-cpu-profiler/CollectDeoptEvents
>
> Committed: https://crrev.com/3f9c2c590d687562d6adb531d1159738c07013f4
> Cr-Commit-Position: refs/heads/master@{#41664}
TBR=jarin@chromium.org,machenbach@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
Review-Url: https://codereview.chromium.org/2568403003
Cr-Commit-Position: refs/heads/master@{#41667}
https://crrev.com/c3a6ca68d0646b10885ef7017557eaf463db2e4a changed what kind of information is dumped which broke some properties of the output that IRHydra relies on:
1. Because IRHydra does not have access to the whole scripts it relies on function sources dumps to map IR and deoptimization to the actual source. This means that source positions must be dumped in a way that allows for this mapping and previously we used positions relative to the beginning of the function to allow that. Now we are dumping absolute script positions which means that function sources must be annotated with the start position so that IRHydra (or any other similar tools) could translate absolute script positions into function-relative ones.
2. When dumping inlined functions we should dump which function are we inlining into because position itself is not enough to recover the inlining stack from the output.
BUG=
Review-Url: https://codereview.chromium.org/2573653002
Cr-Commit-Position: refs/heads/master@{#41663}
Currently the jump label's distance value in LCodeGen::DoWrapReceiver() of X86 is unconsistent.
For avoiding potential bugs caused by this unconsistent usage for the same label when DeoptEveryNTimes()
return true, This CL unify the label's distance value usage in LCodeGen::DoWrapReceiver().
For more informations, please refer CL #41419 (https://codereview.chromium.org/2539403002/ ).
BUG=
Review-Url: https://codereview.chromium.org/2554493002
Cr-Commit-Position: refs/heads/master@{#41659}
When ParseOnBackground is done, we should construct a trace event and dump all
background parsing runtime statistics as argument. In tracing we don't want to
merge it back to to main thread, instead we show the trace event together with
background parsing trace event on script streamer thread track in Trace Viewer.
To perf sheriff: A series of counters ParseBackgroundxxx and
PreparseBackgroundxxx will be taken into account in this patch, thus runtime
statistics graph will increase after this patch gets landed.
BUG=v8:5089
Review-Url: https://codereview.chromium.org/2559403002
Cr-Commit-Position: refs/heads/master@{#41658}
This is to improve the performance of common operation of
extracting contextId or contextGroupId out of Context.
BUG=none
Review-Url: https://codereview.chromium.org/2558913004
Cr-Commit-Position: refs/heads/master@{#41657}
Generalize Messages to include an error level.
Add a parameter to AddMessageHandler to select which error levels to receive, using a mask (default being just errors, i.e. the current behavior).
BUG=v8:4203
R=dgozman@chromium.org,machenbach@chromium.org,danno@chromium.org,bmeurer@chromium.org,jochen@chromium.org
Review-Url: https://codereview.chromium.org/2526703002
Cr-Commit-Position: refs/heads/master@{#41648}
Reason for revert:
Tree closer: https://build.chromium.org/p/client.v8/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20MSAN/builds/12409
Original issue's description:
> [heap] Initialize the owner on each page after lospace allocation
>
> The least two bits of the owner field of a Page are used to determine
> whether the Page is part of a large object. If these bits are not equal
> to 0x11, the page is part of a large object and needs special handling
> e.g. in MemoryChunk::FromAnyPointerAddress to determine which chunk it
> belongs to.
>
> This CL fixes an issue in which the store buffer overflows after
> a large object space allocation but before the object has been fully
> initialized. Store buffer overflow handling attempts to look up the
> chunk of a page, but fails to do so correctly since the page's owner
> field has not yet been initialized.
>
> This CL ensures that the owner field of all pages belonging to a large
> object allocation are initialized to a value that is interpreted
> correctly.
>
> BUG=chromium:672041
>
> Committed: https://crrev.com/9b6808bfb5366beebe3af30a06f9851edb2039d4
> Cr-Commit-Position: refs/heads/master@{#41641}
TBR=mlippautz@chromium.org,jgruber@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:672041
Review-Url: https://codereview.chromium.org/2562273004
Cr-Commit-Position: refs/heads/master@{#41644}
This CL moves all methods related to scripts and eval origin (HasScript,
GetScript, IsEval, GetEvalOrigin) from JSStackFrame to StackFrameBase,
because it also applies to WasmFrames.
This makes the AppendFileLocation method append the same information to
WasmStackFrames and AsmJsWasmStackFrames than to JSStackFrames.
R=titzer@chromium.org, mstarzinger@chromium.org
BUG=v8:4203
Review-Url: https://codereview.chromium.org/2557923005
Cr-Commit-Position: refs/heads/master@{#41642}
The least two bits of the owner field of a Page are used to determine
whether the Page is part of a large object. If these bits are not equal
to 0x11, the page is part of a large object and needs special handling
e.g. in MemoryChunk::FromAnyPointerAddress to determine which chunk it
belongs to.
This CL fixes an issue in which the store buffer overflows after
a large object space allocation but before the object has been fully
initialized. Store buffer overflow handling attempts to look up the
chunk of a page, but fails to do so correctly since the page's owner
field has not yet been initialized.
This CL ensures that the owner field of all pages belonging to a large
object allocation are initialized to a value that is interpreted
correctly.
BUG=chromium:672041
Review-Url: https://codereview.chromium.org/2565713002
Cr-Commit-Position: refs/heads/master@{#41641}
The current logic in Isolate::GetLocationFromStackTrace just ignores
wasm frames, making the computed location point to the first javascript
frame, like this:
test.js:17: RuntimeError: divide by zero
module.exports.main();
^
RuntimeError: divide by zero
at main (<WASM>[1]+5)
at test.js:17:16
This CL not only fixes the location to point to the top-most wasm
frame, but also exposes to the embedder that the script of that location
is a wasm script, allowing for custom printing of wasm locations.
The Shell::ReportException method now checks for this flag, and prints
wasm locations like this:
<WASM>[0]+5: RuntimeError: divide by zero
RuntimeError: divide by zero
at main (<WASM>[0]+5)
at test/message/wasm-trap.js:15:16
R=titzer@chromium.org, yangguo@chromium.org
BUG=chromium:613110
Review-Url: https://codereview.chromium.org/2563673002
Cr-Commit-Position: refs/heads/master@{#41640}
Add the operator in preparation for actual perf work. The operator is replaced
by the same runtime call as before, during lowering.
The CallConstructWithSpreadParameters is a bit silly at the moment, but will
hold more once we add feedback.
BUG=v8:5659
Review-Url: https://codereview.chromium.org/2561103003
Cr-Commit-Position: refs/heads/master@{#41636}
For generator-based functions (e.g. async functions) we force variables to be
context-allocated. Due to a bug in the parser, this didn't always work
correctly. For instance, in "async function foo([a]) { ... }" the variable "a"
could become stack-allocated due to context allocation being forced on the wrong
scope.
Besides fixing this, I'm also cleaning up some related code in the async parsing
setup and adding some guards.
R=adamk@chromium.org, littledan@chromium.org
BUG=
Review-Url: https://codereview.chromium.org/2561093002
Cr-Commit-Position: refs/heads/master@{#41635}
