This CL introduces TurboAssembler, a super-class of Assembler and sub-class
of MacroAssembler. TurboAssembler contains all the functionality that is used
by Turbofan and previously was part of MacroAssembler. TurboAssembler has
access to the isolate but, in contrast to MacroAssembler, does not expect to
be running on the main thread.
Bug: v8:6048
Change-Id: If5693f56a45fe057e5011168e830d01a3f2f772d
Reviewed-on: https://chromium-review.googlesource.com/559674
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46477}
This adds support for exception control flow by adding a counter behind throw
statements (never incremented), as well as a counter for catch and finally
blocks.
Bug: v8:6000
Change-Id: I3959772c889b543ab5e186ad7cd710e55a8aec23
Reviewed-on: https://chromium-review.googlesource.com/558993
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46476}
This CL refactors the internal representation of JavaScript-exposed
WebAssembly objects to be more like other such objects in V8. By introducing
a new instance type for each of the JS-exposed types, we get more robust
typechecking without using embedder fields (which were previously used
when these objects where instance type JS_API_OBJECT).
In addition to the new instance types, the subclasses X of JSObject
(WasmInstanceObject, WasmMemoryObject, WasmModuleObject, WasmTableObject)
now have appropriate Is##X() methods on Object and are now robust.
BUG=v8:6547
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_chromium_rel_ng
Review-Url: https://codereview.chromium.org/2964943002
Cr-Commit-Position: refs/heads/master@{#46475}
This CL changes for floats what https://chromium-review.googlesource.com/c/558964/
changed for doubles.
Original message:
On x86, signalling NaNs get converted to quiet NaNs when they get push
on the stack and popped again. This happens in the code generation for
arm, specifically for the vmov instruction with the immediate parameter.
This CL replaces the vmov function in assembler-arm to take the
immediate as a uint64_t instead of a double, to guarantee that the bit
pattern does not change even if the parameter is a signalling NaN.
New in this CL:
Although src/double.h existed already, src/float.h did not exist yet.
I created the file in this CL, and moved the classes Float32 and
Float64 there, which already existed in src/deoptimizer.h.
R=titzer@chromium.org, martyn.capewell@arm.com, v8-arm-ports@googlegroups.com
BUG=v8:6564
Change-Id: I6a3f1f154af9c8cd4bb8e7e856235d3eee5e9edd
Reviewed-on: https://chromium-review.googlesource.com/561009
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Martyn Capewell <martyn.capewell@arm.com>
Reviewed-by: Ben Titzer <titzer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46473}
This CL contains new created tests for instructions INS and EXT.
TEST=cctest/test-assembler-mips[64]/Ins,
cctest/test-assembler-mips[64]/Ext
BUG=
Change-Id: Ie62037ee425a5cba3700c5025608fdada144add5
Reviewed-on: https://chromium-review.googlesource.com/561319
Commit-Queue: Miran Karić <Miran.Karic@imgtec.com>
Reviewed-by: Miran Karić <Miran.Karic@imgtec.com>
Cr-Commit-Position: refs/heads/master@{#46468}
On MIPS R6 we have BC instruction that has 26bit offset. We exploit this
in order to generate deoptimizer table with two instructions per entry
instead of earlier three instructions per entry.
Bug:
Change-Id: If6638a9d4aa1515bfaa61530767d26ce5a78b0b0
Reviewed-on: https://chromium-review.googlesource.com/536972
Reviewed-by: Miran Karić <Miran.Karic@imgtec.com>
Reviewed-by: Ivica Bogosavljevic <ivica.bogosavljevic@imgtec.com>
Commit-Queue: Ivica Bogosavljevic <ivica.bogosavljevic@imgtec.com>
Cr-Commit-Position: refs/heads/master@{#46466}
The PrepareCallCFunction methods would take a scratch register as arguments when
they can simply use ip. The reason to do this, apart from simplifying the code,
is that TurboFan's code generator uses r9 as a scratch for this when it could
just as well use ip.
Bug: v8:6553
Change-Id: I0ed762d201f7ff8b8ed074da758227a8db95fca7
Reviewed-on: https://chromium-review.googlesource.com/558934
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#46465}
Blink needs %ErrorPrototype% in order to properly set up the inheritance
chain from DOMException, as specified in WebIDL:
https://heycam.github.io/webidl/#es-DOMException-specialness
This patch is similar to commit 5ec1cddcd ("Expose %IteratorPrototype% as an
intrinsic in the public API"), with the difference that there was no entry
for %ErrorPrototype% in any of the mappings in contexts.h.
Bug: chromium:556950, chromium:737497
Cq-Include-Trybots: master.tryserver.chromium.linux:linux_chromium_rel_ng
Change-Id: Iadc5b2b844f29f6c9640b6a89769d233931366e9
Reviewed-on: https://chromium-review.googlesource.com/559058
Reviewed-by: Adam Klein <adamk@chromium.org>
Commit-Queue: Raphael Kubo da Costa (rakuco) <raphael.kubo.da.costa@intel.com>
Cr-Commit-Position: refs/heads/master@{#46464}
This adds a new binary block coverage mode (in addition to the existing count
block coverage), as well as a few transformation passes to reduce the number of
uselessly reported ranges.
Bug: v8:6000
Change-Id: I4fb234ca015990d00aa2f1dccb87f76ba4748994
Reviewed-on: https://chromium-review.googlesource.com/552642
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Aleksey Kozyatinskiy <kozyatinskiy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46463}
Adds constant folding for the StringConcat bytecode to
NativeContextSpecialization. Can reduce operator to either a fully folded
constant string, or a JSAdd or a StringConcat with a reduced number of
operators.
BUG=v8:6243, chromium:738312
Change-Id: I6b2be6a3d95230a23f3c7390a4f7be5181c49a2a
Reviewed-on: https://chromium-review.googlesource.com/559146
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46461}
Adds a CheckElimination reducer to eliminate checks which have become
unecessary due to inlining of heap constants.
BUG=v8:6243, chromium:738312
Change-Id: Ie50b274bd07c86466eead08b2f21d2b63dd9e01c
Reviewed-on: https://chromium-review.googlesource.com/559129
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46460}
If a thin string can be dereferenced for StringConcat we still need to check
whether the dereferenced string is a sequential string itself (it could be
an external string).
BUG=v8:6243
Change-Id: I146541512525726f092580512c0b5f02d33685a7
Reviewed-on: https://chromium-review.googlesource.com/558994
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46459}
Port 9b3174b295
Original Commit Message:
Remove all IsHeapObject/IsSmi checks from assembler and also from
the macro-assembler functions that Turbofan code generation uses.
Note for porters: In case it's unclear which macro-assembler
functions need to be modified, it may be best to wait until I
split MacroAssembler in a followup-CL, which will make that clear.
R=neis@chromium.org, joransiu@ca.ibm.com, jyan@ca.ibm.com, michael_dawson@ca.ibm.com
BUG=v8:6048
LOG=N
Change-Id: Iae7d74709930a25246ab919a65dfecf1175945f7
Reviewed-on: https://chromium-review.googlesource.com/562656
Reviewed-by: Junliang Yan <jyan@ca.ibm.com>
Commit-Queue: Jaideep Bajwa <bjaideep@ca.ibm.com>
Cr-Commit-Position: refs/heads/master@{#46458}
Print the object that is being destructured and update the error
message.
Previously,
d8> var [a] = {}
(d8):1: TypeError: [Symbol.iterator] is not a function
Now,
d8> var [a] = {}
(d8):1: TypeError: {} is not iterable
Bug: v8:6513, v8:5532
Change-Id: I5cbfe7c7e20632bce1a48bd38a1b0c98d0ff0660
Reviewed-on: https://chromium-review.googlesource.com/557370
Commit-Queue: Sathya Gunasekaran <gsathya@chromium.org>
Reviewed-by: Adam Klein <adamk@chromium.org>
Reviewed-by: Caitlin Potter <caitp@igalia.com>
Cr-Commit-Position: refs/heads/master@{#46457}
The ')'-in-parameter checking is not necessary for
CompileFunctionInContext. The arguments array is expected to be an
array of identifiers, not an array of arbitrary strings that get
concatenated.
Furthermore, there's no reason to have the .toString() representation
look like it came from CreateDynamicFunction(), and in fact inserting
line breaks makes it more complicated to map line and column numbers
correctly.
Overall, the --harmony-function-tostring behavior only makes
CompileFunctionInContext worse, so this CL removes it.
R=littledan@chromium.org, adamk@chromium.org, caitp@igalia.com
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_trusty_blink_rel
Bug: v8:4958
Change-Id: Ifbc8a83216ca6a6979da1199972aa65f4bee36c3
Reviewed-on: https://chromium-review.googlesource.com/558220
Reviewed-by: Daniel Ehrenberg <littledan@chromium.org>
Commit-Queue: Josh Wolfe <jwolfe@igalia.com>
Cr-Commit-Position: refs/heads/master@{#46456}
... so that it can be used during off-thread code assembly.
R=bmeurer@chromium.org
Bug: v8:6048
Change-Id: Iaa9b534b23d02da69c2b2395c1eacfdcffb3ac2f
Reviewed-on: https://chromium-review.googlesource.com/561677
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46455}
Tsan figured out that I called compilation_units_.empty() outside a
lock.
Original message:
This CL adds a new function {InitializeCompilationUnit} to initialize
a single compilation unit and not just all compilation units at once.
This is necessary for streaming compilation eventually. This also
required some refactoring on how the working queue for compilation units
works. Previously the synchronization was done with an atomic counter,
now it is done with a lock. Note that the code to finish compilation
of a module still only works if the working queue gets only empty when
all work is done. I plan to change this in a different CL.
Since the code would not be tested without streaming compilation, I added
an experimental flag and a test to test the new code.
R=clemensh@chromium.org, mtrofin@chromium.org
Change-Id: Ia83560d1c70f0802271a88c514e0d1cb0458f6c4
Reviewed-on: https://chromium-review.googlesource.com/561458
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46454}
In the original CL I moved an error check backwards, unfortunately
behind a vector lookup which should not happen when there is an error.
Now I also move the vector lookup backwards.
Original message:
We stored the size of a function body before we check that
these values are valid. This caused a failing DCHECK in the constructor
of WireBytesRef which checked for integer overflows. With this CL we
check the size of the function body before we create the WireBytesRef.
R=clemensh@chromium.org
Bug: chromium:738097
Change-Id: Ie65b3cfcbcd6bdb3f04b0760673d9c7b7a0d1057
Reviewed-on: https://chromium-review.googlesource.com/561519
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46453}
This CL adds a few transformations that clean up the set of reported
source ranges. Duplicates, empty, and uncovered ranges are removed, and
nested/consecutive ranges are merged if possible.
BUG=v8:6000
Change-Id: I421ee35ce8292cfe84c1eea4f653762cea5d909d
Reviewed-on: https://chromium-review.googlesource.com/558411
Reviewed-by: Aleksey Kozyatinskiy <kozyatinskiy@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46450}
Due to the dictionary refactoring we ended up calling the superclass' helper
which in turn made us decode the values wrongly for the GlobalDictionary.
Change-Id: I4298b6a437ef2d84b69b7e980470c3cf5af79944
Reviewed-on: https://chromium-review.googlesource.com/561701
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Commit-Queue: Camillo Bruni <cbruni@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46449}
Replaces the old PromotionQueue that was kept at the end of to space.
This change allows a future implementation to use
(a) a thread-local promotion list and
(b) enables work stealing for concurrent scenarios.
Bug: chromium:738865
Change-Id: I4c983a36e69ad4a9462eb9a59f586a85e51acbde
Reviewed-on: https://chromium-review.googlesource.com/561141
Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46446}
This reverts commit 6c8aed76ff.
Reason for revert: Breaks some debug bots:
https://build.chromium.org/p/client.v8/builders/V8%20Linux64%20-%20debug/builds/16754https://build.chromium.org/p/client.v8/builders/V8%20Win64%20-%20debug/builds/17654
Original change's description:
> [wasm] Check the size of a function body before storing it
>
> We stored the size of a function body before we check that
> these values are valid. This caused a failing DCHECK in the constructor
> of WireBytesRef which checked for integer overflows. With this CL we
> check the size of the function body before we create the WireBytesRef.
>
> R=clemensh@chromium.org
>
> Bug: chromium:738097
> Change-Id: I18f8b628c1499aae9c8e9340ea73c87f19e6f1d7
> Reviewed-on: https://chromium-review.googlesource.com/561000
> Commit-Queue: Andreas Haas <ahaas@chromium.org>
> Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#46442}
TBR=ahaas@chromium.org,clemensh@chromium.org
Change-Id: Ifd533c0dee369c746bc97fea13275ebc09ed5eff
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: chromium:738097
Reviewed-on: https://chromium-review.googlesource.com/561517
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46445}
- Check for value map in favour of value instance type
Bug: v8:6557
Change-Id: I820267a87c169143d7a3c567d6c9508bae492350
Reviewed-on: https://chromium-review.googlesource.com/561358
Commit-Queue: Maya Lekova <mslekova@google.com>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46444}
We stored the size of a function body before we check that
these values are valid. This caused a failing DCHECK in the constructor
of WireBytesRef which checked for integer overflows. With this CL we
check the size of the function body before we create the WireBytesRef.
R=clemensh@chromium.org
Bug: chromium:738097
Change-Id: I18f8b628c1499aae9c8e9340ea73c87f19e6f1d7
Reviewed-on: https://chromium-review.googlesource.com/561000
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46442}
This is the first step in optimizing Map and Set iterators. This ports
all the base functionality including
- Set.prototype.entries
- Set.prototype.values
- %SetPrototypeIterator%.next
- Map.prototype.entries
- Map.prototype.keys
- Map.prototype.values
- %MapPrototypeIterator%.next
to C++ and removes all the dead code and the previous half JavaScript
implementation. The next step is to port core parts to CodeStubAssembler
and finally inline the fast-paths into TurboFan directly. The relevant
design document is at:
https://docs.google.com/document/d/13z1fvRVpe_oEroplXEEX0a3WK94fhXorHjcOMsDmR-8
Most of this work is very similar to how the Array iterator works and we
mostly follow the same process for the implementation.
R=jgruber@chromium.org
Bug: v8:6571
Change-Id: Ieb253d6705ba4077c697a5ff0cb6f87f9c4056ff
Reviewed-on: https://chromium-review.googlesource.com/561138
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46441}
TruncateDoubleToI generated a 32-bit result but did not clear the upper
32 bits. This violated the invariant that the upper 32 bits should be
cleared when the result is 32 bits. This change fixes the bug mentioned
below. Clearing the upper 32 bits is also done on x64.
R=v8-arm-ports@googlegroups.com, titzer@chromium.org, martyn.capewell@arm.com
Bug: chromium:738952
Change-Id: I7e23e03fbed380ff08803db41fbae6382957ba08
Reviewed-on: https://chromium-review.googlesource.com/559671
Reviewed-by: Martyn Capewell <martyn.capewell@arm.com>
Reviewed-by: Ben Titzer <titzer@chromium.org>
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46440}
https://chromium-review.googlesource.com/561001 was not sufficient,
because we don't use the --gerrit flag in V8 anymore.
Now we explicitly pass an option for private upload.
TBR=hablich@chromium.org,tandrii@chromium.org
NOTRY=true
Bug: v8:6574
Change-Id: Ib8e2883795f26f22ac47e648b37c47f3a0a60a40
Reviewed-on: https://chromium-review.googlesource.com/561378
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46439}
jtt will now print the complete transition tree of a given Map in gdb.
Change-Id: I07031dd3d463bec0072e8ac696406279ff057489
Reviewed-on: https://chromium-review.googlesource.com/558257
Commit-Queue: Camillo Bruni <cbruni@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46437}
Remove all IsHeapObject/IsSmi checks from assembler and also from
the macro-assembler functions that Turbofan code generation uses.
Note for porters: In case it's unclear which macro-assembler
functions need to be modified, it may be best to wait until I
split MacroAssembler in a followup-CL, which will make that clear.
Bug: v8:6048
Change-Id: Ife0735cc6f48713c9ec493faf2dac5e553d1c06b
Reviewed-on: https://chromium-review.googlesource.com/561015
Commit-Queue: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46436}
Make use of UseScratchRegisterScope instead of using the ip register directly in
code stubs, builtin and the deoptimizer. In a lot of cases, we can
simply use a different register rather than using the new scope.
Bug: v8:6553
Change-Id: Ibc8a9a78bb88f3850c6e8b45871cc3a5b3971b3b
Reviewed-on: https://chromium-review.googlesource.com/544837
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#46429}
