With this patch, we can generate simple immediate-shift instructions for
immediates outside the range "0 <= imm < width". Several related
instruction selectors have also been updated accordingly.
Example of generated code:
---- Before --- ---- After ----
movz w0, #33 lsr w0, w1, #1
lsr w0, w1, w0
BUG=
Review URL: https://codereview.chromium.org/1179893003
Cr-Commit-Position: refs/heads/master@{#28977}
Before selecting multiply-accumulate for a multiplication with add operation,
check that the multiply can't be reduced to add-with-shift. This prevents
simple multiplications by 3, 5, etc turning into register moves and madd
instructions.
Review URL: https://codereview.chromium.org/1180863002
Cr-Commit-Position: refs/heads/master@{#28976}
This fixes the graph wiring of implicit JSToNumber nodes inserted by
JSTypedLowering, to be correctly hooked into a surrounding exceptional
continuation.
R=bmeurer@chromium.org
TEST=mjsunit/compiler/try-binop,test262
Review URL: https://codereview.chromium.org/1178153004
Cr-Commit-Position: refs/heads/master@{#28975}
The spec says ToObject is called on the receiver, and this is
observable if you call sort on a primitive. This patch trivially
adds the call and a test.
BUG=v8:4125
R=adamk
LOG=Y
Review URL: https://codereview.chromium.org/1178193004
Cr-Commit-Position: refs/heads/master@{#28972}
%AddElement is not intended for objects which are not arrays, and
its behavior may go away with future refactorings. This patch gets
rid of it if the receiver of from or of is not the intrinsic Array
object.
Array.of and Array.from previously papered over failures in calling
[[DefineOwnProperty]] when setting array elements. This patch
makes them lead to exceptions, and adds tests to assert that
the appropriate exceptions are thrown.
BUG=v8:4168
R=adamk
CC=rossberg,verwaest
LOG=Y
Review URL: https://codereview.chromium.org/1181623003
Cr-Commit-Position: refs/heads/master@{#28969}
Reason for revert:
Blocks revert of https://codereview.chromium.org/1175973002
Original issue's description:
> Replace SetObjectProperty / DefineObjectProperty with less powerful alternatives where relevant.
>
> @yangguo: please look at the debugger part of the CL.
> @ishell: please look at the rest.
>
> Additionally:
> - Ensure the LookupIterator for named properties does not accidentally get indexes in.
> - Fix the return value for typed array assignments to be the incoming value.
>
> BUG=v8:4137
> LOG=n
>
> Committed: https://crrev.com/15aa811f8fe2708a757c3b53ca89db736aa8b222
> Cr-Commit-Position: refs/heads/master@{#28954}
TBR=yangguo@chromium.org,verwaest@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=v8:4137
Review URL: https://codereview.chromium.org/1181733002
Cr-Commit-Position: refs/heads/master@{#28957}
@yangguo: please look at the debugger part of the CL.
@ishell: please look at the rest.
Additionally:
- Ensure the LookupIterator for named properties does not accidentally get indexes in.
- Fix the return value for typed array assignments to be the incoming value.
BUG=v8:4137
LOG=n
Review URL: https://codereview.chromium.org/1178503004
Cr-Commit-Position: refs/heads/master@{#28954}
Merge a following arithmetic or logical right shift into the existing shift
of ARM64's Int32MulHigh or Uint32MulHigh code.
BUG=
Review URL: https://codereview.chromium.org/1179503003
Cr-Commit-Position: refs/heads/master@{#28945}
Reason for revert:
Breaks InstructionSelectorTest.Word64ShrWithWord64AndWithImmediate on debug builds (but not optdebug builds). I'll investigate.
Original issue's description:
> [arm64][turbofan]: Handle any immediate shift.
>
> With this patch, we can generate simple immediate-shift instructions for
> immediates outside the range "0 <= imm < width". Several related
> instruction selectors have also been updated accordingly.
>
> Example of generated code:
>
> ---- Before --- ---- After ----
> movz w0, #33 lsr w0, w1, #1
> lsr w0, w1, w0
>
> BUG=
>
> Committed: https://crrev.com/36d771bbfa4af5efcc1c1dcf5b234445cb7ee722
> Cr-Commit-Position: refs/heads/master@{#28943}
TBR=bmeurer@chromium.org,ulan@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=
Review URL: https://codereview.chromium.org/1176393002
Cr-Commit-Position: refs/heads/master@{#28944}
With this patch, we can generate simple immediate-shift instructions for
immediates outside the range "0 <= imm < width". Several related
instruction selectors have also been updated accordingly.
Example of generated code:
---- Before --- ---- After ----
movz w0, #33 lsr w0, w1, #1
lsr w0, w1, w0
BUG=
Review URL: https://codereview.chromium.org/1179733004
Cr-Commit-Position: refs/heads/master@{#28943}
The alternative of passing an object template at context creation is
unfeasible because we need a context to instantiate the template.
At the time we create the context from snapshot or bootstrap from
scratch, we would already need that template instance, leading to a
chicken-and-egg problem.
This is an alternative that is simpler and less intrusive.
R=domenic@chromium.org, jochen@chromium.org
Review URL: https://codereview.chromium.org/1182443002
Cr-Commit-Position: refs/heads/master@{#28938}
Reason for revert:
Test failures are bogus. Snapshot blob and natives blob are out of sync due to build being weird.
Original issue's description:
> Revert of Promise assimilation fix. (patchset #8 id:160001 of https://codereview.chromium.org/1098663002/)
>
> Reason for revert:
> Test failures: https://chromegw.corp.google.com/i/client.v8/builders/V8%20Linux64/builds/3829
>
> Original issue's description:
> > Promise assimilation fix.
> >
> > Let x be a fulfilled promise and y be another promise. |x.then(() => y)|
> > should call |y.then|, but the current implementation calls PromiseChain.
> > We can see the difference when we set a custom function to |y.then|.
> >
> > This CL fixes the spec violation, but as a result |then| is no longer
> > a wrapper of |chain| and in some cases it does not work well with
> > |accept| or |chain|. That is not a problem for ES6 promise users because
> > ES6 promise doesn't have them.
> >
> > LOG=N
> > BUG=477921
> >
> > Committed: https://crrev.com/2f57dff3ea0c45e1a61b334fda962460f89d71bc
> > Cr-Commit-Position: refs/heads/master@{#28926}
>
> TBR=arv@chromium.org,caitpotter88@gmail.com,rossberg@chromium.org,yhirano@chromium.org
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=477921
>
> Committed: https://crrev.com/5bb75f514027f79303396dba823c2d78c6add83b
> Cr-Commit-Position: refs/heads/master@{#28927}
TBR=arv@chromium.org,caitpotter88@gmail.com,rossberg@chromium.org,yhirano@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=477921
Review URL: https://codereview.chromium.org/1181533006
Cr-Commit-Position: refs/heads/master@{#28928}
Reason for revert:
Test failures: https://chromegw.corp.google.com/i/client.v8/builders/V8%20Linux64/builds/3829
Original issue's description:
> Promise assimilation fix.
>
> Let x be a fulfilled promise and y be another promise. |x.then(() => y)|
> should call |y.then|, but the current implementation calls PromiseChain.
> We can see the difference when we set a custom function to |y.then|.
>
> This CL fixes the spec violation, but as a result |then| is no longer
> a wrapper of |chain| and in some cases it does not work well with
> |accept| or |chain|. That is not a problem for ES6 promise users because
> ES6 promise doesn't have them.
>
> LOG=N
> BUG=477921
>
> Committed: https://crrev.com/2f57dff3ea0c45e1a61b334fda962460f89d71bc
> Cr-Commit-Position: refs/heads/master@{#28926}
TBR=arv@chromium.org,caitpotter88@gmail.com,rossberg@chromium.org,yhirano@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=477921
Review URL: https://codereview.chromium.org/1176163004
Cr-Commit-Position: refs/heads/master@{#28927}
Let x be a fulfilled promise and y be another promise. |x.then(() => y)|
should call |y.then|, but the current implementation calls PromiseChain.
We can see the difference when we set a custom function to |y.then|.
This CL fixes the spec violation, but as a result |then| is no longer
a wrapper of |chain| and in some cases it does not work well with
|accept| or |chain|. That is not a problem for ES6 promise users because
ES6 promise doesn't have them.
LOG=N
BUG=477921
Review URL: https://codereview.chromium.org/1098663002
Cr-Commit-Position: refs/heads/master@{#28926}
This is needed in order to allow expansion of a throwing node into a
set of nodes that produce different effects for the successful and the
exceptional continuation.
R=bmeurer@chromium.org
Review URL: https://codereview.chromium.org/1179543002
Cr-Commit-Position: refs/heads/master@{#28918}
After https://codereview.chromium.org/1180433003 the code-size exceeeds
the limit for MIPS. We have some optimizations in place for MIPS64 that
will be ported back to MIPS, and will investigate other code-size
improvements so we can re-enable this test.
BUG=
Review URL: https://codereview.chromium.org/1166353005
Cr-Commit-Position: refs/heads/master@{#28913}
Frame Elider requires a sane CFG which should have such dummy end block.
BUG=
Review URL: https://codereview.chromium.org/1166293004
Cr-Commit-Position: refs/heads/master@{#28911}
On Mac for the requested minimal version (10.5) only old stl (from gcc 4.2.1) can be used so unfortunately we can't use numeric_limits<>::lowest() which were added in 874c54e05e
BUG=
TEST=cctest/test-assembler-mips64, cctest/test-assembler-mips
Review URL: https://codereview.chromium.org/1175463003
Cr-Commit-Position: refs/heads/master@{#28907}
Up until now we can only inline based on JSFunction, because of the way
the deoptimization works. With this change we will be able to inline
based on the SharedFunctionInfo and materialize the JSFunction from a
literal or a stack slot when necessary.
R=jarin@chromium.org
Review URL: https://codereview.chromium.org/1169103004
Cr-Commit-Position: refs/heads/master@{#28906}
This change rewrites regexps like (ab|ac|z|ad|ae|af) into (a[b-f]|z). We can only reorder disjunctions like this for case-dependent regexps. For case-independent regexps, the disjunctions should be pre-sorted for best results.
R=yangguo@chromium.org
BUG=chromium:482998
LOG=n
Review URL: https://codereview.chromium.org/1180433003
Cr-Commit-Position: refs/heads/master@{#28902}
This deprecates the aforementioned mutator in favor of a simpler
NodeProperties::ReplaceUses that doesn't perform any relaxation.
Preparation for enabling support for try-catch statements.
R=bmeurer@chromium.org
TEST=unittests/NodePropertiesTest
Review URL: https://codereview.chromium.org/1172773003
Cr-Commit-Position: refs/heads/master@{#28897}
The old version would always clobber.
NOTRY=true
NOTREECHECKS=true
Review URL: https://codereview.chromium.org/1177503005
Cr-Commit-Position: refs/heads/master@{#28895}
Don't DCHECK when neutering that the buffer is not a SharedArrayBuffer;
instead, just return early.
BUG=chromium:498142,chromium:497295
R=jarin@chromium.org
LOG=n
Review URL: https://codereview.chromium.org/1174753002
Cr-Commit-Position: refs/heads/master@{#28892}
Reason for revert:
[Sheriff] Still lots of breakages on some builders. E.g. http://build.chromium.org/p/client.v8/builders/V8%20Linux%20-%20nosnap/builds/3313
Original issue's description:
> Update Test262 to 5/30
>
> This updates the test runner to run all tests in both sloppy
> and strict mode.
>
> It also marks the test in the status file as failing in
> sloppy mode even if all it does is marking the test as
> [PASS, FAIL]. A future CL will have to validate that the
> tests pass and fail in the correct mode.
>
> Committed: https://crrev.com/d869f4a4801d4ef6868c266c07f9e29d2e29cba5
> Cr-Commit-Position: refs/heads/master@{#28879}
TBR=rossberg@chromium.org,arv@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
Review URL: https://codereview.chromium.org/1173933002
Cr-Commit-Position: refs/heads/master@{#28890}
For top-level code the closure passed into context allocation methods
needs to be replaced with a sentinel to canonicalize is to the empty
function object.
R=bmeurer@chromium.org
TEST=mjsunit/regress/regress-4169
BUG=v8:4169
LOG=N
Review URL: https://codereview.chromium.org/1172013002
Cr-Commit-Position: refs/heads/master@{#28888}
The initial implementation of this method takes a couple shortcuts:
- At some points in the spec, we should be making a reference to
"Table 49" and instead do property lookup on the constructor.
This is an issue with some other TypedArray methods too.
- The spec indicates that, if constructor and @@species haven't been
messed with in particular ways, then there should be a memcpy
from the old to the new. That fast path is not implemented here
in this patch, but because V8 doesn't canonicalize NaN, the result
isn't observably different.
BUG=v8:3578
LOG=Y
R=arv
Review URL: https://codereview.chromium.org/1170023002
Cr-Commit-Position: refs/heads/master@{#28882}
This updates the test runner to run all tests in both sloppy
and strict mode.
It also marks the test in the status file as failing in
sloppy mode even if all it does is marking the test as
[PASS, FAIL]. A future CL will have to validate that the
tests pass and fail in the correct mode.
Review URL: https://codereview.chromium.org/1136553008
Cr-Commit-Position: refs/heads/master@{#28879}
