Design doc:
https://docs.google.com/document/d/1szInbXZfaErWW70d30hJsOLL0Es-l5_g8d2rXm1ZBqI/edit?usp=sharing
V8 can already collect data about how many times each basic block in the
builtins is run. This change enables using that data for profile-guided
optimization. New comments in BUILD.gn describe how to use this feature.
A few implementation details worth mentioning, which aren't covered in
the design doc:
- BasicBlockProfilerData currently contains an array of RPO numbers.
However, this array is always just [0, 1, 2, 3, ...], so this change
removes that array. A new DCHECK in BasicBlockInstrumentor::Instrument
ensures that the removal is valid.
- RPO numbers, while useful for printing data that matches with the
stringified schedule, are not useful for matching profiling data with
blocks that haven't been scheduled yet. This change adds a new array
of block IDs in BasicBlockProfilerData, so that block counters can be
used for PGO.
- Basic block counters need to be written to a file so that they can be
provided to a subsequent run of mksnapshot, but the design doc doesn't
specify the transfer format or what file is used. In this change, I
propose using the existing v8.log file for that purpose. Block count
records look like this:
block,TestLessThanHandler,37,29405
This line indicates that block ID 37 in TestLessThanHandler was run
29405 times. If multiple lines refer to the same block, the reader
adds them all together. I like this format because it's easy to use:
- V8 already has robust logic for creating the log file, naming it to
avoid conflicts in multi-process situations, etc.
- Line order doesn't matter, and interleaved writes from various
logging sources are fine, given that V8 writes each line atomically.
- Combining multiple sources of profiling data is as simple as
concatenating their v8.log files together.
- It is a good idea to avoid making any changes based on profiling data
if the function being compiled doesn't match the one that was
profiled, since it is common to use profiling data downloaded from a
central lab which is updated only periodically. To check whether a
function matches, I propose using a hash of the Graph state right
before scheduling. This might be stricter than necessary, as some
changes to the function might be small enough that the profile data is
still relevant, but I'd rather err on the side of not making incorrect
changes. This hash is also written to the v8.log file, in a line that
looks like this:
builtin_hash,LdaZeroHandler,3387822046
Bug: v8:10470
Change-Id: I429e5ce5efa94e01e7489deb3996012cf860cf13
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2220765
Commit-Queue: Seth Brenith <seth.brenith@microsoft.com>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69008}
DateTimeFormat's fractionalSecondDigits shipped in M84, so we
can drop the --harmony_intl_dateformat_fractional_second_digits
flag now.
Bug: v8:10485
Change-Id: If7a1a8599722073d7382635219a6fb46cdc47474
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2311410
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Frank Tang <ftang@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69007}
This CL fixes passing the receiver of a fast call as an effect input
only in cases where the opcode supports it.
It also introduces a test for callbacks without fallback support and a
test where ConvertReceiver is not introduced.
Bug: chromium:1052746
Change-Id: I6f396f4c9cbaab7ae915c908a9f805d9770f8946
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2312777
Commit-Queue: Maya Lekova <mslekova@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69005}
Bug: v8:10720
Change-Id: I7a05bfb3c87c4f0f5516608da5d42fdaff466536
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2304572
Commit-Queue: Richard Stotz <rstz@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69001}
This reverts commit d8f8a7e210.
Reason for revert: Breaks code_serializer variant - https://cr-buildbucket.appspot.com/build/8874070652992164976
Original change's description:
> Reland "[flags] warn about contradictory flags"
>
> This is a reland of b8f9166664
> Difference to previous CL: Additional functionality to specify
> incompatible flags based on GN variables and extra-flags, used
> to fix the issues that came up on the waterfall.
>
> This also changes the rules regarding repeated flags: While
> explicitly repeated flags are allowed for boolean values as long
> as they are identical, repeated flags or explicit flags in the
> presence of an active implication are disallowed for non-boolean
> flags. The latter simplifies specifying conflict rules in
> variants.py. Otherwise a rule like
>
> INCOMPATIBLE_FLAGS_PER_EXTRA_FLAG = {
> "--gc-interval=*": ["--gc-interval=*"],
> }
>
> wouldn't work because specifying the same GC interval twice
> wouldn't actually count as a conflict. This was an issue with
> test/mjsunit/wasm/gc-buffer.js, which specifies
> --gc-interval=500 exactly like the extra flag by the stress bot.
>
> Also, this now expands contradictory flags checking to d8 flags
> for consistency.
>
> Original change's description:
> > [flags] warn about contradictory flags
> >
> > Design Doc: https://docs.google.com/document/d/1lkvu8crkK7Ei39qjkPCFijpNyxWXsOktG9GB-7K34jM/
> >
> > Bug: v8:10577
> > Change-Id: Ib9cfdffa401c48c895bf31caed5ee03545beddab
> > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2154792
> > Reviewed-by: Clemens Backes <clemensb@chromium.org>
> > Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> > Reviewed-by: Georg Neis <neis@chromium.org>
> > Reviewed-by: Tamer Tas <tmrts@chromium.org>
> > Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
> > Cr-Commit-Position: refs/heads/master@{#68168}
>
> Bug: v8:10577
> Change-Id: I268e590ee18a535b13dee14eeb15ddd0a9ee8341
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2235115
> Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
> Reviewed-by: Tamer Tas <tmrts@chromium.org>
> Reviewed-by: Clemens Backes <clemensb@chromium.org>
> Reviewed-by: Georg Neis <neis@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#68989}
TBR=machenbach@chromium.org,neis@chromium.org,clemensb@chromium.org,tebbi@chromium.org,tmrts@chromium.org
Change-Id: I7969065b0edbc463a94e530485bc2ab623d77b62
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:10577
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2312782
Reviewed-by: Maya Lekova <mslekova@chromium.org>
Commit-Queue: Maya Lekova <mslekova@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68992}
The object is passed now as an v8::ApiObject instead of unwrapped
C++ pointer and the embedder should do the unwrapping.
Bug: chromium:1052746
Change-Id: If5671c5fdbbe8d58435c7bd9aceccf5e17f8ea21
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2304571
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Commit-Queue: Maya Lekova <mslekova@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68991}
This is a reland of b8f9166664
Difference to previous CL: Additional functionality to specify
incompatible flags based on GN variables and extra-flags, used
to fix the issues that came up on the waterfall.
This also changes the rules regarding repeated flags: While
explicitly repeated flags are allowed for boolean values as long
as they are identical, repeated flags or explicit flags in the
presence of an active implication are disallowed for non-boolean
flags. The latter simplifies specifying conflict rules in
variants.py. Otherwise a rule like
INCOMPATIBLE_FLAGS_PER_EXTRA_FLAG = {
"--gc-interval=*": ["--gc-interval=*"],
}
wouldn't work because specifying the same GC interval twice
wouldn't actually count as a conflict. This was an issue with
test/mjsunit/wasm/gc-buffer.js, which specifies
--gc-interval=500 exactly like the extra flag by the stress bot.
Also, this now expands contradictory flags checking to d8 flags
for consistency.
Original change's description:
> [flags] warn about contradictory flags
>
> Design Doc: https://docs.google.com/document/d/1lkvu8crkK7Ei39qjkPCFijpNyxWXsOktG9GB-7K34jM/
>
> Bug: v8:10577
> Change-Id: Ib9cfdffa401c48c895bf31caed5ee03545beddab
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2154792
> Reviewed-by: Clemens Backes <clemensb@chromium.org>
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Reviewed-by: Georg Neis <neis@chromium.org>
> Reviewed-by: Tamer Tas <tmrts@chromium.org>
> Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#68168}
Bug: v8:10577
Change-Id: I268e590ee18a535b13dee14eeb15ddd0a9ee8341
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2235115
Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
Reviewed-by: Tamer Tas <tmrts@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68989}
Some platforms disable reading of bytes in the .text section,
so move the metadata into a separate .rodata section.
Bug: v8:10707
Change-Id: I30ef7a180f489f175c31f9d4dcd02115c9f516c2
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2301113
Commit-Queue: Toan Pham <toanpham@google.com>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68984}
This CL introduces a new pipeline mode in which each optimization
triggers both a Turbofan and an NCI compilation job. The TF code is
installed, the NCI code is inserted into the code cache for future
consumption by other contexts.
--turbo-nci enables this mode.
The old configuration (with NCI replacing TF) is still available under
the --turbo-nci-as-highest-tier flag. This flag remains useful for
testing purposes.
Drive-by: Refactor tracing in compiler.cc.
Bug: v8:8888
Change-Id: I62522e61788762250ff717eef84eae914e266f3b
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2299360
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Mythri Alle <mythria@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68983}
If we insert a second SIMPLE_PROPERTY_ARRAY we move from kWeakRef
encoding to kFullTransitionArray encoding. As always, we are searching
in the meantime in a background thread.
Bug: v8:7790
Change-Id: I5e3d85657dda4c199e50b9c35e7f617f1821a8f0
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2243218
Reviewed-by: Nico Hartmann <nicohartmann@chromium.org>
Commit-Queue: Santiago Aboy Solanes <solanes@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68980}
This CL makes the ScriptContextTable concurrently accessible from the
background thread (in particular ScriptContextTable::get_context).
A cctest is added to check synchronization with tsan.
Bug: v8:7790
Change-Id: I2e2dc8c6a7cfa369787959c4d5ed5f357f4720fa
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2260567
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Commit-Queue: Nico Hartmann <nicohartmann@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68979}
The load splat, load extend, load zero macros are essentially the same,
consolidate them into a single macro.
Change-Id: Ic812043b37524deb3a9e6ddc223bb95ae77e1d4d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2304715
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Reviewed-by: Bill Budge <bbudge@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68978}
v128.const (kExprS128) is the only constant expression supported
(similar to the other value types).
Bug: v8:10731
Change-Id: I9b11b47a851903dfd79d3590eff67b615057f81c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2308389
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Reviewed-by: Bill Budge <bbudge@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68975}
Prototype in TurboFan x64 and interpreter, bailout in Liftoff.
Suggested in https://github.com/WebAssembly/simd/pull/237.
Bug: v8:10713
Change-Id: I5346c351fb2ec5240b74013e62aef07c46d5d9b6
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2300924
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Bill Budge <bbudge@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68973}
On desktop systems, we use a very basic tiering strategy: Everything is
initially compiled with Liftoff, and once that is done, the module can
start being used. Concurrently to the execution, we re-compile all code
with TurboFan, and hot-swap each function once TurboFan finishes.
We should start using a more dynamic strategy where each function is
tiered-up when judged necessary. This change will then tier-up each
liftoff function once it has been called 5 times.
I then added a counter in the native module, that is updated directly
from Liftoff code, and a runtime call is then made when the counter
reaches the goal.
R=clemensb@chromium.org
CC=thibaudm@chromium.org
Bug: v8:10728
Change-Id: I8dc2b02fdff8d97781bb1cf496886594b3d7f644
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2306803
Commit-Queue: Arnaud Robin <arobin@google.com>
Reviewed-by: Thibaud Michaud <thibaudm@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68971}
V8 already depends on cppgc_base which means that unittests does not
need to depend on cppgc_for_testing any longer.
Move the cppgc_for_testing dependency to the stand-alone
cppgc_unittests binary
Bug: v8:10674
Change-Id: I07bfe30901eb1683d2e0ee0189f73b5244884f27
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2310249
Auto-Submit: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Anton Bikineev <bikineev@chromium.org>
Reviewed-by: Anton Bikineev <bikineev@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68970}
With work on NCI proceeding, it makes sense to test multiple
pipeline configurations.
The nci variant (passes --turbo-nci) now spawns dedicated NCI
compilation jobs and inserts generated code into the code cache.
The nci_as_highest_tier variant (passes --turbo-nci-as-highest-tier)
simply replaces TF with NCI code (no extra jobs, no extra caching).
This mode stresses NCI generated code more than the nci variant, in
which NCI code only runs on cache hits.
Bug: v8:8888
Change-Id: I4c2a43cce5271a6c288e7aba195dcc9daed6af9d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2299361
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68964}
They have been replaced with {array,struct}.new_with_rtt.
Also, rework tests that used those instructions.
Bug: v8:7748
Change-Id: I2aaccb1958bf2b8d6cad4969abc612216856393d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2307318
Commit-Queue: Manos Koukoutos <manoskouk@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68961}
For debugging code, disable opcode merging. Otherwise, the effect of the
first merged opcode would not be observable when stepping.
R=thibaudm@chromium.org
Bug: v8:10350
Change-Id: Id656c9dee8f9676bf3d7881f3782e5ead76b5e71
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2306802
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Thibaud Michaud <thibaudm@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68960}
We currently still merge opcodes (i.e. i32 comparisons plus a br_if).
This CL adds a test for this, which checks for the current behaviour.
A follow-up CL will fix this and update the expected output accordingly.
R=thibaudm@chromium.org
Bug: v8:10350
Change-Id: I846aa931a3ec1a27043f04e830503d5732ae473e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2307232
Reviewed-by: Thibaud Michaud <thibaudm@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68957}
I was looking around and couldn't find any tests for select with v128
params. Adding it for completion.
Change-Id: I9e3770e37eaddb23fc99649579b175536c1df844
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2308093
Reviewed-by: Bill Budge <bbudge@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68951}
The CanTailCall check only passes if the return locations are the
same in the caller and the callee. However, stack returns are expected
to be at a different offset depending on the stack space reserved for
parameters.
R=clemensb@chromium.org
Bug: v8:7431
Change-Id: Iaac15fce889d6cd7d1ac88f320a872202281fb5a
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2289789
Reviewed-by: Georg Neis <neis@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Commit-Queue: Thibaud Michaud <thibaudm@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68933}
Exposes initial_async_iterator_prototype the same way that
initial_iterator_prototype is exposed, to let blink implement
async iterables better.
Bug: chromium:1087157
Change-Id: I94f65eedb3aad0bf62cb4808935f320fe4bf4e9c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2300779
Commit-Queue: Marijn Kruisselbrink <mek@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68921}
The {operator==} on {VarState} did not check the spill offset, so when
merging stack states, we forgot to move stack values if both source and
destination were stack slots, but at different offsets.
This CL fixes this by removing the {operator==}, because the semantics
(and use) are not clear, and it's only used in one place anyway.
The equality check was mostly redundant, so inlining it also makes the
code smaller and faster.
R=ahaas@chromium.org
Bug: v8:10702
Change-Id: I6c8b2cfd1002274175c9a17d305692e4631fd7dc
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2304574
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68916}
... by migrating old-style code
MyObject* obj = new (zone) MyObject(...)
to the new style
MyObject* obj = zone->New<MyObject>(...)
... and prohibiting accidental use of the old-style.
Bug: v8:10689
Change-Id: Id75774ac12e3d0f95cb3a538066dffbf7815e438
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2300490
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68905}
This is exactly the same issue as in https://crrev.com/c/2299364 for test-torque.cc
Change-Id: I066d93918c94d0c68278c72d9b60ec92a1c5f68b
Bug: v8:10201
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2300546
Commit-Queue: Toon Verwaest <verwaest@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Auto-Submit: Victor Gomes <victorgomes@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68897}
We must perform "smi-checks" before loading an object's map
whenever the object might be an i31ref.
Bug: v8:7748
Change-Id: I2d9839ddcb0c2e8c35b9bea38afe50d55dd084cb
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2299370
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68893}
This CL ports MarkingVerifier from blink.
The existing verifier checks only references on heap.
This new verifier checks references both on heap and on stack.
Bug: chromium:1056170
Change-Id: I083dcb0087125312cca34a2201015a9aecfe6ea4
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2300484
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Reviewed-by: Anton Bikineev <bikineev@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68891}
We cannot allocate large arrays exceeding the size of
kMaxRegularHeapObjectSize in young space. Bailout of optimization in
such cases.
Bug: chromium:1105746
Change-Id: I4f7357c2dd7b3e70d747f9067660725ecf6ae768
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2300481
Reviewed-by: Nico Hartmann <nicohartmann@chromium.org>
Commit-Queue: Mythri Alle <mythria@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68889}
