These also lower to subgraphs that have to be connected to the effect
and control chains, otherwise removing the atomic regions around heap
allocations would still be unsound.
R=jarin@chromium.org
Review URL: https://codereview.chromium.org/1916763003
Cr-Commit-Position: refs/heads/master@{#35762}
This will allow to pull in gyp as a deps to the same location
as chromium (tools/gyp not build/gyp), needed for gn switch.
This is the first step of a 3-way move.
1) Copy v8.gyp in v8
2) Update references in embedders (follow up)
3) Remove old v8.gyp (follow up)
BUG=chromium:474921
LOG=n
NOTRY=true
Review URL: https://codereview.chromium.org/1920793002
Cr-Commit-Position: refs/heads/master@{#35760}
This adds a baseline tier to the compilation pipeline. Currently this
tier is used to model a path from the interpreter to optimized code via
full-codegen code (to ensure sufficient type feedback). Switching from
the unoptimized tier to the baseline tier is limited to happen only when
there are no activations of the given function on the stack.
R=rmcilroy@chromium.org,bmeurer@chromium.org
Review URL: https://codereview.chromium.org/1903273004
Cr-Commit-Position: refs/heads/master@{#35757}
Use the FastNewSloppyArgumentsStub in the interpreter when function doesn't have
duplicate parameters.
BUG=v8:4280
LOG=N
Review URL: https://codereview.chromium.org/1909903003
Cr-Commit-Position: refs/heads/master@{#35754}
The Oddball::to_number_raw field contains the actual double value of the
Oddball converted to a number, and is located at the same offset as the
HeapNumber::value field, so for lowering changes we don't need to check
for undefined (or any other oddball explicitly).
R=jarin@chromium.org
Review URL: https://codereview.chromium.org/1922443002
Cr-Commit-Position: refs/heads/master@{#35753}
This patch make sure that the nop instructions used to mark floating-point
arguments live range begin will not be moved by the scheduler.
R=jarin@chromium.org
Review URL: https://codereview.chromium.org/1896933004
Cr-Commit-Position: refs/heads/master@{#35751}
The ObjectIs<Type> predicates compile down to diamonds (in the general
case), and those should be connected properly to the control and effect
chain in the linearization pass.
R=jarin@chromium.org
Review URL: https://codereview.chromium.org/1912743002
Cr-Commit-Position: refs/heads/master@{#35748}
port 0231a7efa2 (r35743)
original commit message:
This allows us to get rid of the "push TruncateFloat64ToInt32 into Phi"
trick that was used in the MachineOperatorReducer to combine the
ChangeTaggedToFloat64 and TruncateFloat64ToInt32 operations. Instead of
doing that later, we can just introduce the proper operator during the
representation selection directly.
Also separate the TruncateFloat64ToInt32 machine operator, which had two
different meanings depending on a flag (either JavaScript truncation or
C++ style round to zero). Now there's a TruncateFloat64ToWord32 which
represents the JavaScript truncation (implemented via TruncateDoubleToI
macro + code stub) and the RoundFloat64ToInt32, which implements the C++
round towards zero operation (in the same style as the other WebAssembly
driven Round* machine operators).
BUG=
Review URL: https://codereview.chromium.org/1912403002
Cr-Commit-Position: refs/heads/master@{#35745}
port 5c8609de9d (r35724)
original commit message:
This ensures the InterpreterEntryTrampoline heals code entry fields
inside closures when being called without a valid bytecode array. This
is preparatory work to allow removal of bytecode when switching some
functions to other types of code.
BUG=
Review URL: https://codereview.chromium.org/1920713002
Cr-Commit-Position: refs/heads/master@{#35744}
This allows us to get rid of the "push TruncateFloat64ToInt32 into Phi"
trick that was used in the MachineOperatorReducer to combine the
ChangeTaggedToFloat64 and TruncateFloat64ToInt32 operations. Instead of
doing that later, we can just introduce the proper operator during the
representation selection directly.
Also separate the TruncateFloat64ToInt32 machine operator, which had two
different meanings depending on a flag (either JavaScript truncation or
C++ style round to zero). Now there's a TruncateFloat64ToWord32 which
represents the JavaScript truncation (implemented via TruncateDoubleToI
macro + code stub) and the RoundFloat64ToInt32, which implements the C++
round towards zero operation (in the same style as the other WebAssembly
driven Round* machine operators).
R=jarin@chromium.org
Review URL: https://codereview.chromium.org/1919513002
Cr-Commit-Position: refs/heads/master@{#35743}
If a deferred block has multiple predecessors, they have to be
all deferred. Otherwise, we can run into a situation where if a range
that spills only in deferred blocks inserts its spill in the block, and
other ranges need moves inserted by ResolveControlFlow in the predecessors,
the register of the range spilled in the deferred block may be clobbered.
To avoid that, when a deferred block has multiple predecessors, and some
are not deferred, we add a non-deferred block to collect all such edges.
This CL addresses the validator assertion failure the referenced issue, as well
as the greedy allocator failure - which was caused by the situation described
above.
BUG=v8:4940
LOG=n
Review URL: https://codereview.chromium.org/1912093005
Cr-Commit-Position: refs/heads/master@{#35742}
Need to use the kHashFieldSlot rather than kHashFieldOffset for
pointer-sized memory accesses.
(Fix for "[builtins] Migrate String.prototype.charCodeAt and String.prototype.charAt to TurboFan.")
R=bmeurer@chromium.org, epertoso@chromium.org
BUG=
Review URL: https://codereview.chromium.org/1907393002
Cr-Commit-Position: refs/heads/master@{#35741}
Reason for revert:
Need to fix the #undef logic.
Original issue's description:
> Fix interpreter unittest for embedded constant pools.
>
> The offset from fp to the register file is based on the frame size
> -- which is one slot larger when embedded constant pools are enabled.
>
> TEST=unittests/DecodeBytecodeAndOperands
> R=rmcilroy@chromium.org, bmeurer@chromium.org, oth@chromium.org, mstarzinger@chromium.org
> BUG=
TBR=bmeurer@chromium.org,mstarzinger@chromium.org,oth@chromium.org,rmcilroy@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=
Review URL: https://codereview.chromium.org/1909323003
Cr-Commit-Position: refs/heads/master@{#35736}
This also removes the destructor of the class in question, which removed
any added decorator from the graph. However the adding of the decorator
happens explicitly, so symmetry suggests that removal should also happen
explicitly instead of implicitly in the destructor.
R=bmeurer@chromium.org
Review URL: https://codereview.chromium.org/1914473002
Cr-Commit-Position: refs/heads/master@{#35732}
Removes some control edges added from the RawMachineAssembler to the end of the graph.
Adds a parameter that tells the Verifier to ignore effect and control inputs.
Review URL: https://codereview.chromium.org/1912853003
Cr-Commit-Position: refs/heads/master@{#35731}
Refactor the Scope object to automatically enable strict mode when
initialized as a "module" scope, relieving the caller of this
responsibility.
BUG=v8:4941
LOG=N
R=adamk@chromium.org
Review URL: https://codereview.chromium.org/1906923002
Cr-Commit-Position: refs/heads/master@{#35730}
The approximate mode enables taking an approximate stack trace from GC,
where the top frames might be missing if inlined. Note that in that case,
the frame summary will refer to optimized code, so it will not be possible
to take source position. (The user of the summary will have to handle the
case frame_summary.abstract_code()->kind() == AbstractCode::OPTIMIZED_CODE
specially.)
Review URL: https://codereview.chromium.org/1907443002
Cr-Commit-Position: refs/heads/master@{#35728}
There's no point in running the SimplifiedOperatorReducer also during
the late optimization pass, as it will not do any useful work at that
point.
R=jarin@chromium.org
Review URL: https://codereview.chromium.org/1909363002
Cr-Commit-Position: refs/heads/master@{#35727}
Reason for revert:
Appears to break Android crbug.com/604422
Original issue's description:
> [GN] Define USE_EABI_HARDFLOAT=1 when arm_float_abi=="hard".
>
> Add this define to the config used for mksnapshot. This fixes a bug
> where certain applications would fail at runtime on Chromecast.
>
> BUG=592660
> LOG=Y
> Bug: internal b/27495984
>
> Test: Formerly broken Cast apps load and run as expected.
>
> Committed: https://crrev.com/86357d5235ceba61c151f0b6e509bcb365860454
> Cr-Commit-Position: refs/heads/master@{#35183}
TBR=dpranke@chromium.org,alokp@chromium.org,titzer@chromium.org,slan@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=592660,604422
LOG=n
Review URL: https://codereview.chromium.org/1906373002
Cr-Commit-Position: refs/heads/master@{#35725}
This ensures the InterpreterEntryTrampoline heals code entry fields
inside closures when being called without a valid bytecode array. This
is preparatory work to allow removal of bytecode when switching some
functions to other types of code.
R=rmcilroy@chromium.org
BUG=v8:4280
LOG=n
Review URL: https://codereview.chromium.org/1904093002
Cr-Commit-Position: refs/heads/master@{#35724}
Get rid of further typing checks from ChangeLowering and put them into
the representation selection pass instead (encoding the information in
the operator instead).
Drive-by-change: Rename ChangeSmiToInt32 to ChangeTaggedSignedToInt32
for consistency about naming Tagged, TaggedSigned and TaggedPointer.
R=jarin@chromium.org
Review URL: https://codereview.chromium.org/1909343002
Cr-Commit-Position: refs/heads/master@{#35723}
Currently we are using UnsafeCurrent in async signal handler to acquire the
isolate of VM thread, but we want to get rid of that since it prevents V8 from
being thread agnostic.
This patch replaces UnsafeCurrent with a static map, where we store a map of
samplers for threads, and makes it accessible by signal handler.
BUG=v8:4889
LOG=n
Review URL: https://codereview.chromium.org/1900473002
Cr-Commit-Position: refs/heads/master@{#35722}
Adds IncStub and DecStub TurboFan code stubs and hooks them up to the
interpreter's Inc and Dec bytecodes (which are used for count
operations, e.g. i++).
BUG=v8:4280
LOG=N
Review URL: https://codereview.chromium.org/1901083002
Cr-Commit-Position: refs/heads/master@{#35720}
Fixes a mistake made in r35618 for register OperandSize calculations.
BUG=605470
LOG=N
Review URL: https://codereview.chromium.org/1908033002
Cr-Commit-Position: refs/heads/master@{#35719}
The new bytecodes replace two runtime functions. They are still unsupported by the bytecode graphbuilder, though.
BUG=v8:4907
LOG=n
Review URL: https://codereview.chromium.org/1904933002
Cr-Commit-Position: refs/heads/master@{#35716}
This is a follow-up to 58429beb7b
"Fix KeyedStore stub selection for STRING_WRAPPER_ELEMENTS".
BUG=chromium:602184
LOG=n
Review URL: https://codereview.chromium.org/1912443004
Cr-Commit-Position: refs/heads/master@{#35715}
The feature was deprecated in M49 and flagged off in M50.
This patch removes it entirely from the codebase.
Review URL: https://codereview.chromium.org/1909433003
Cr-Commit-Position: refs/heads/master@{#35714}
If we have to convert a float64 value to tagged representation and we
already know that the value is either in Signed31/Signed32 or
Unsigned32 range, then we can just convert the float64 to word32 and
use the fast word32 to tagged conversion. Doing this in
ChangeLowering (or the effect linearization pass) would be unsound, as
the types on the nodes are no longer usable.
This removes all Type uses from effect linearization. There's still some
work to be done for ChangeLowering tho.
R=jarin@chromium.org
Review URL: https://codereview.chromium.org/1908093002
Cr-Commit-Position: refs/heads/master@{#35713}
