When finding conflicts, there's no reason to keep looking for registers that are clearly
not going to be available to a candidate live range.
BUG=v8:5644
Review-Url: https://codereview.chromium.org/2559733002
Cr-Commit-Position: refs/heads/master@{#41602}
Reason for revert:
gc-stress failures
Original issue's description:
> [wasm] Fix location for error in asm.js ToNumber conversion
>
> In the asm.js code translated to wasm, we call imported functions via a
> WASM_TO_JS stub, which first calls the function and then calls ToNumber
> on the return value. Exceptions can happen in both calls.
> We were only ever reporting the location of the function call, whereas
> asm.js code executed via turbofan reported the location of the type
> coercion operator ("+" on "+foo()" or "|" on "foo()|0").
>
> This CL implements the same behaviour for asm.js code translated to
> wasm. The following is changed:
> - the AsmWasmBuilder records the parent node when descending on a binary
> operator (also "+foo()" is represented by a binary operation).
> - it stores not one location per call in the source position side
> table, but two (one for the call, one for the parent which does the
> type coercion).
> - the wasm compiler annotates the source positions "0" and "1" to the
> two calls in the WASM_TO_JS wrapper (only if the module origin is
> asm.js).
> - during stack trace generation (in the StackTraceIterator), when we
> move from the WASM_TO_JS frame to the WASM frame, we remember at which
> call inside the WASM_TO_JS wrapper we are, and encode this information
> in the generated caller state, used for the WASM frame.
> - the same information is also stored in the FrameArray which is used
> to reconstruct the stack trace later.
>
> R=titzer@chromium.org, bradnelson@chromium.org
> CC=jgruber@chromium.org
> BUG=v8:4203,v8:5724
>
> Committed: https://crrev.com/94cd46b55e24fa2bb7b06b3da4d5ba7f029bc262
> Cr-Commit-Position: refs/heads/master@{#41599}
TBR=bradnelson@chromium.org,mstarzinger@chromium.org,titzer@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=v8:4203,v8:5724
Review-Url: https://codereview.chromium.org/2563613003
Cr-Commit-Position: refs/heads/master@{#41601}
In the asm.js code translated to wasm, we call imported functions via a
WASM_TO_JS stub, which first calls the function and then calls ToNumber
on the return value. Exceptions can happen in both calls.
We were only ever reporting the location of the function call, whereas
asm.js code executed via turbofan reported the location of the type
coercion operator ("+" on "+foo()" or "|" on "foo()|0").
This CL implements the same behaviour for asm.js code translated to
wasm. The following is changed:
- the AsmWasmBuilder records the parent node when descending on a binary
operator (also "+foo()" is represented by a binary operation).
- it stores not one location per call in the source position side
table, but two (one for the call, one for the parent which does the
type coercion).
- the wasm compiler annotates the source positions "0" and "1" to the
two calls in the WASM_TO_JS wrapper (only if the module origin is
asm.js).
- during stack trace generation (in the StackTraceIterator), when we
move from the WASM_TO_JS frame to the WASM frame, we remember at which
call inside the WASM_TO_JS wrapper we are, and encode this information
in the generated caller state, used for the WASM frame.
- the same information is also stored in the FrameArray which is used
to reconstruct the stack trace later.
R=titzer@chromium.org, bradnelson@chromium.orgCC=jgruber@chromium.org
BUG=v8:4203,v8:5724
Review-Url: https://codereview.chromium.org/2555243002
Cr-Commit-Position: refs/heads/master@{#41599}
Lowercase 1 character strings occur frequently in minified code. Add a
cache for them, so that we don't need to compute the hash + do the hash
table lookup for each occurrence.
BUG=
Review-Url: https://codereview.chromium.org/2541353002
Cr-Commit-Position: refs/heads/master@{#41597}
Fix 7a6f294ffe.
The first correction enables correct execution DoMathMinMax when two
input registers are the same register.
The second correction adds NOP instructions after branch instructions
in tests macro_float_minmaxf(32|64).
TEST=cctest/test-macro-assembler-mips[64]/macro_float_minmax_f32
cctest/test-macro-assembler-mips[64]/macro_float_minmax_f64
mjsunit/regress/math-min
BUG=
Review-Url: https://codereview.chromium.org/2556793003
Cr-Commit-Position: refs/heads/master@{#41596}
We have been assuming in several places that ContainsDot or ToInt32 is
sufficient to check a value is a valid double or int.
Refactoring all the checks to one place and making them cope with booleans
or other unexpected types being present.
BUG=672044
R=titzer@chromium.org
Review-Url: https://codereview.chromium.org/2555323003
Cr-Commit-Position: refs/heads/master@{#41595}
Use of eval in a function wraps it in a context.
This throws off assumptions not checked until later,
which is at odds with incremental validation and conversion.
Check that module parameters are PARAMETER location early.
BUG=672045
R=titzer@chromium.org
Review-Url: https://codereview.chromium.org/2558813004
Cr-Commit-Position: refs/heads/master@{#41594}
Wrap the liveness bitvectors from the bytecode liveness analysis with a
helper class, which makes the register/accumulator bits explicit.
Review-Url: https://codereview.chromium.org/2552723004
Cr-Commit-Position: refs/heads/master@{#41589}
Aside from the default snapshot, there is no need for additional context
snapshots to have the ability to replace the global proxy and global object
after deserialization. Changes include:
- Changes to the API to better distinguish default context snapshot from
additional context snapshots.
- Disallow global handles when creating snapshots.
- Allow extensions when creating snapshots.
This solves the issue of not being able to having accessors and interceptors on
the global object of contexts to be serialized.
R=jochen@chromium.org, peria@chromium.org
BUG=chromium:617892
Review-Url: https://codereview.chromium.org/2557743003
Cr-Commit-Position: refs/heads/master@{#41588}
The patch was reverted due to a bug - we failed to evict OSR-optimized
code in the case where the SharedFunctionInfo OptimizedCodeMap was
empty/cleared.
Since we OSR code rarely, it makes sense to store it and look for it on the native context rather than the SharedFunctionInfo. This makes the OptimizedCodeMap data structure more space efficient, as it doesn't have to store an ast ID for the OSR entry point.
Review-Url: https://codereview.chromium.org/2561083002
Cr-Commit-Position: refs/heads/master@{#41584}
This CL attempts to set the maybe-assigned flag for variables that are written
to as part of a destructuring or loop header.
For instance, in the following two cases we now mark x as maybe-assigned.
a) [x] = [1];
b) for (x of [1,2,3]) {};
There's more work to do here, this is just a first step.
R=adamk@chromium.org, mstarzinger@chromium.org
BUG=v8:5636
Review-Url: https://codereview.chromium.org/2562443003
Cr-Commit-Position: refs/heads/master@{#41582}
Using x&(x-1) to check for power of two masks usable at runtime
speeds up the life benchmark.
Borrowing this from SimplifiedLowering for the AsmJsRemS internal
wasm opcode.
Leaving this out for general wasm as we should be doing this optimization
in LLVM.
BUG=v8:4203
TEST=None
R=bmeurer@chromium.org
Review-Url: https://codereview.chromium.org/2556963005
Cr-Commit-Position: refs/heads/master@{#41581}
Speeds up some benchmarks that make heavy use of derived constructors.
BUG=chromium:672075
Review-Url: https://codereview.chromium.org/2557963004
Cr-Commit-Position: refs/heads/master@{#41580}
Currently when the number passed to TryNumberToSize is 1 << 64,
it gets away with a bug caused by rounding of mantissa.
Then the number will be casted to 0 and TryNumberToSize
will return true. This patch fix this by making the range check
more accurate.
BUG=v8:5712
Review-Url: https://codereview.chromium.org/2548243004
Cr-Commit-Position: refs/heads/master@{#41578}
We were losing the stats for the first instance. Recording them
as soon as code objects are produced. This way, we have them available
for compile-only benchmarks.
Review-Url: https://codereview.chromium.org/2556963003
Cr-Commit-Position: refs/heads/master@{#41574}
First step towards making arguments and rest parameters optimizable by
splitting the allocations for the actual object and the elements. The
object allocations can already be escape analyzed this way, the elements
would need special support in the deoptimizer and the escape analysis,
but that can be done as a second separate step.
R=jarin@chromium.org
BUG=v8:5726
Review-Url: https://codereview.chromium.org/2557283002
Cr-Commit-Position: refs/heads/master@{#41573}
-- Moves promiseHasHandlerSymbol to inobject property
-- Ports PromiseResolveClosure to TF
-- Fix a non spec async-await test which fails now because we do a map
check for native promise check (instead of IsPromise). Changing the
constructor (in the test) invalidates the map check.
This patch results in a 7.1% performance improvement in the bluebird
benchmark (over 5 runs).
BUG=v8:5343
Review-Url: https://codereview.chromium.org/2541283002
Cr-Commit-Position: refs/heads/master@{#41569}
jasongin@ created this patch.
dcc50445a3
This patch adds the support to emit a trace event by using a comma-separated
list of categories, so that the trace event will be emitted if there is at least
one category is enabled in the categories list.
TBR=jochen@chromium.org
Review-Url: https://codereview.chromium.org/2558193002
Cr-Commit-Position: refs/heads/master@{#41567}
Due to the isOwn check, functions inherited through prototype will not be
included in a preview.
BUG=645053
Review-Url: https://codereview.chromium.org/2554623003
Cr-Commit-Position: refs/heads/master@{#41566}
Getter properties are not currently included in the protocol's
Runtime.ObjectPreview. DevTools currently shows getter properties
when evaluating arrays in the console, and this CL brings them into
the preview generated for RemoteObjects.
Corresponding DevTools CL: https://codereview.chromium.org/2521513006/
BUG=666882
Review-Url: https://codereview.chromium.org/2508423002
Cr-Commit-Position: refs/heads/master@{#41565}
Previously we created 3 FixedArrays and then filled them up with
values. This meant that during the creation of the second and third
FixedArray, there were one and two FixedArrays respectively, without
any values in it which broke the FixedArrayVerify.
This patch fills each FixedArray with the correct values before
creating new ones.
BUG=chromium:672051
Review-Url: https://codereview.chromium.org/2554323003
Cr-Commit-Position: refs/heads/master@{#41564}
We're still collecting use counter data for this situation.
BUG=v8:4973
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
Review-Url: https://codereview.chromium.org/2510873005
Cr-Commit-Position: refs/heads/master@{#41563}
When an octal escape sequence is in a string in strict mode:
- Octal literals are not allowed in strict mode.
+ Octal escape sequences are not allowed in strict mode.
When an octal escape sequence is in a template string:
- Octal literals are not allowed in template strings.
+ Octal escape sequences are not allowed in template strings.
BUG=v8:4973
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
Review-Url: https://codereview.chromium.org/2551633002
Cr-Commit-Position: refs/heads/master@{#41560}
This patch introduces:
1. ICStats class to store ic statistics items produced by V8,
2. A disabled by default tracing category v8.ic_stats,
3. An trace event V8.ICStats that contains ic statistics items in args,
We store ic statistics items in an array until the array is full to reduce
the number of trace events.
TBR=jkummerow@chromium.org,ishell@chromium.org
Review-Url: https://codereview.chromium.org/2503183002
Cr-Commit-Position: refs/heads/master@{#41559}
When v8 fails an access check, it invokes a helper to try to see if it
can service the request via an access check interceptor. Invoking the
access check interceptor can throw an exception (e.g. a SecurityError).
Unfortunately, the failed access check property helpers and the
interceptor helpers don't agree on how to propagate the exception: if
the interceptor helper detects a scheduled exception, it promotes the
exception to a pending exception and returns to the failed access check
property helper.
The failed access check property helper also has an early return in
case of a scheduled exception. However, this doesn't work, as the
previously thrown exception is no longer scheduled, as it's been
promoted to a pending exception. Thus, the failed access check property
helper always end up calling the failed access check callback as well.
Since Blink's implementation of the failed access check callback also
throws an exception, this conflicts with the previously-thrown,
already-pending exception.
With this patch, the failed access check property helpers check for a
pending exception rather than a scheduled exception after invoking the
interceptor, so the exception can be propagated correctly.
BUG=v8:5715
R=yangguo@chromium.org,jochen@chromium.org
Review-Url: https://codereview.chromium.org/2550423002
Cr-Commit-Position: refs/heads/master@{#41556}
Introduces:
- a new AST node representing the GetIterator() algorithm in the specification, to be used by ForOfStatement, YieldExpression (in the case of delegating yield*), and the future `for-await-of` loop proposed in http://tc39.github.io/proposal-async-iteration/#sec-async-iterator-value-unwrap-functions.
- a new opcode (JumpIfJSReceiver), which is useful for `if Type(object) is not Object` checks which are common throughout the specification. This node is easily eliminated by TurboFan.
The AST node is desugared specially in bytecode, rather than manually when building the AST. The benefit of this is that desugaring in the BytecodeGenerator is much simpler and easier to understand than desugaring the AST.
This also reduces parse time very slightly, and allows us to use LoadIC rather than KeyedLoadIC, which seems to have better baseline performance. This results in a ~20% improvement in test/js-perf-test/Iterators micro-benchmarks, which I believe owes to the use of the slightly faster LoadIC as opposed to the KeyedLoadIC in the baseline case. Both produce identical optimized code via TurboFan when the type check can be eliminated, and the load can be replaced with a constant value.
BUG=v8:4280
R=bmeurer@chromium.org, rmcilroy@chromium.org, adamk@chromium.org, neis@chromium.org, jarin@chromium.orgTBR=rossberg@chromium.org
Review-Url: https://codereview.chromium.org/2557593004
Cr-Commit-Position: refs/heads/master@{#41555}
Since we OSR code rarely, it makes sense to store it and look for it on the native context rather than the SharedFunctionInfo. This makes the OptimizedCodeMap data structure more space efficient, as it doesn't have to store an ast ID for the OSR entry point.
BUG=
Review-Url: https://codereview.chromium.org/2549753002
Cr-Commit-Position: refs/heads/master@{#41554}
There were two bugs, one partly hiding the other one:
1) We generate the ToNumber conversion for each WASM_TO_JS wrapper,
even if the expected return type is void.
2) The return node in the WASM_TO_JS wrapper did not use the effect of
the ToNumber conversion.
This CL fixes both, and adds test cases to check that we do throw an
error trying to convert (e.g.) Symbol to a number, but only if the
return type is not void.
Additional test check that a user-provided valueOf method is actually
called the correct number of times.
R=titzer@chromium.org, bradnelson@chromium.org
BUG=v8:4203
Review-Url: https://codereview.chromium.org/2552123004
Cr-Commit-Position: refs/heads/master@{#41552}
Both @@match and @@split internally use dynamically growing fixed
arrays. Shrink to fit when wrapping these in a JSArray to avoid
excessive memory usage.
BUG=chromium:670205,chromium:670708
Review-Url: https://codereview.chromium.org/2556773002
Cr-Commit-Position: refs/heads/master@{#41550}
