SourcePosition::InliningId() refers to a the new table DeoptimizationInputData::InliningPositions(), which provides the following data for every inlining id:
- The inlined SharedFunctionInfo as an offset into DeoptimizationInfo::LiteralArray
- The SourcePosition of the inlining. Recursively, this yields the full inlining stack.
Before the Code object is created, the same information can be found in CompilationInfo::inlined_functions().
If SourcePosition::InliningId() is SourcePosition::kNotInlined, it refers to the outer (non-inlined) function.
So every SourcePosition has full information about its inlining stack, as long as the corresponding Code object is known. The internal represenation of a source position is a positive 64bit integer.
All compilers create now appropriate source positions for inlined functions. In the case of Turbofan, this required using AstGraphBuilderWithPositions for inlined functions too. So this class is now moved to a header file.
At the moment, the additional information in source positions is only used in --trace-deopt and --code-comments. The profiler needs to be updated, at the moment it gets the correct script offsets from the deopt info, but the wrong script id from the reconstructed deopt stack, which can lead to wrong outputs. This should be resolved by making the profiler use the new inlining information for deopts.
I activated the inlined deoptimization tests in test-cpu-profiler.cc for Turbofan, changing them to a case where the deopt stack and the inlining position agree. It is currently still broken for other cases.
The following additional changes were necessary:
- The source position table (internal::SourcePositionTableBuilder etc.) supports now 64bit source positions. Encoding source positions in a single 64bit int together with the difference encoding in the source position table results in very little overhead for the inlining id, since only 12% of the source positions in Octane have a changed inlining id.
- The class HPositionInfo was effectively dead code and is now removed.
- SourcePosition has new printing and information facilities, including computing a full inlining stack.
- I had to rename compiler/source-position.{h,cc} to compiler/compiler-source-position-table.{h,cc} to avoid clashes with the new src/source-position.cc file.
- I wrote the new wrapper PodArray for ByteArray. It is a template working with any POD-type. This is used in DeoptimizationInputData::InliningPositions().
- I removed HInlinedFunctionInfo and HGraph::inlined_function_infos, because they were only used for the now obsolete Crankshaft inlining ids.
- Crankshaft managed a list of inlined functions in Lithium: LChunk::inlined_functions. This is an analog structure to CompilationInfo::inlined_functions. So I removed LChunk::inlined_functions and made Crankshaft use CompilationInfo::inlined_functions instead, because this was necessary to register the offsets into the literal array in a uniform way. This is a safe change because LChunk::inlined_functions has no other uses and the functions in CompilationInfo::inlined_functions have a strictly longer lifespan, being created earlier (in Hydrogen already).
BUG=v8:5432
Review-Url: https://codereview.chromium.org/2451853002
Cr-Commit-Position: refs/heads/master@{#40975}
Adds an IsInterpreted() function to both SharedFunctionInfo and JSFunction.
This is used to fix the test-heap code-aging tests since Ignition doesn't
age code.
BUG=v8:4680
Review-Url: https://codereview.chromium.org/2481433002
Cr-Commit-Position: refs/heads/master@{#40868}
This enables Ignition unconditionally for all code that is destined for
optimization with TurboFan. This ensures all optimization attempts will
go through the BytecodeGraphBuilder and that the AstGraphBuilder pipe is
dried out in practice.
patch from issue 2427953002 at patchset 120001 (http://crrev.com/2427953002#ps120001)
R=mvstanton@chromium.org,rmcilroy@chromium.org
Review-Url: https://codereview.chromium.org/2453973004
Cr-Commit-Position: refs/heads/master@{#40663}
This enables Ignition unconditionally for all code that is destined for
optimization with TurboFan. This ensures all optimization attempts will
go through the BytecodeGraphBuilder and that the AstGraphBuilder pipe is
dried out in practice.
R=mvstanton@chromium.org
Review-Url: https://chromiumcodereview.appspot.com/2427953002
Cr-Commit-Position: refs/heads/master@{#40462}
The CL #39795 (https://codereview.chromium.org/2345593003 ) added MultiReturnSelect_f32/MultiReturnSelect_f64 tests.
Because those tests need 2 float point return registers and x87 only has 1 FP return register, Those tests failed at x87 port.
This CL skips MultiReturnSelect_f32/MultiReturnSelect_f64 tests for x87.
BUG=
Review-Url: https://codereview.chromium.org/2405743002
Cr-Commit-Position: refs/heads/master@{#40113}
This makes sure we run the module tests against all variants using
Ignition as the first compilation tier. It will henceforth extend the
test coverage to the BytecodeGraphBuilder as well.
R=neis@chromium.org
Review-Url: https://codereview.chromium.org/2397733002
Cr-Commit-Position: refs/heads/master@{#39992}
Ignition collects type feedback for binary and compare operations in type
feedback vector and FCG uses Binary/CompareOpICs to collect type feedback.
The feedback collected by ignition is not used by crankshaft. This hits the
performance, when trying to optimize functions that did not tier upto FCG.
This cl merges the feedback collected by ignition and FCG when passing to
crankshaft.
BUG=v8:4280
Review-Url: https://codereview.chromium.org/2361043002
Cr-Commit-Position: refs/heads/master@{#39753}
This patch gives the ability for the embedder to ask for the
module requests of a module, and to pass a ResolveCallback
into Module::Instantiate().
In d8, I've implemented a simple module_map that's used
along with this API to allow loading, compiling, instantiating,
and evaluating a whole tree of modules.
No path resolution is yet implemented, meaning that all
import paths are relative to whatever directory d8 runs
in. And no imports are linked to the exports of the
requested module.
BUG=v8:1569
Review-Url: https://codereview.chromium.org/2351113004
Cr-Commit-Position: refs/heads/master@{#39569}
This is a first implementation of inlining into graphs that have been
created using the {BytecodeGraphBuilder}. Note that inlining sticks to
graphs of the same kind, we only ever inline AstGraph into AstGraph or
BytecodeGraph into BytecodeGraph, no mixed inlining.
R=bmeurer@chromium.org,rmcilroy@chromium.org
TEST=cctest/test-run-inlining
BUG=v8:5251
Review-Url: https://codereview.chromium.org/2262033003
Cr-Commit-Position: refs/heads/master@{#39439}
Adds support to collect allocation site feedback for Array function calls
to the call bytecode handler.
BUG=v8:4280, v8:4780
LOG=N
Review-Url: https://codereview.chromium.org/2307903002
Cr-Commit-Position: refs/heads/master@{#39283}
This is analogous to the variable liveness analysis we do in the AstGraphBuilder, but on the bytecode registers.
BUG=
Review-Url: https://codereview.chromium.org/2307863002
Cr-Commit-Position: refs/heads/master@{#39248}
In ignition, allocation site mementos were disabled when creating array
literals. Enabled them in this cl.
BUG=v8:4280
LOG=N
Review-Url: https://codereview.chromium.org/2294913006
Cr-Commit-Position: refs/heads/master@{#39234}
Also disable four cpu-profiler tests under asan that fail
with the GN setup.
BUG=chromium:474921,v8:5243
Review-Url: https://codereview.chromium.org/2309263002
Cr-Commit-Position: refs/heads/master@{#39171}
Our Type class has a semantic and representational dimension.
Much code in src/ast, Crankshaft and Turbofan is based on it.
Going forward in Turbofan we'd like to remove representational information
entirely. To that end, new type AstType has been created to preserve
existing behavior for the benefit of Crankshaft and the AST.
BUG=
Review-Url: https://codereview.chromium.org/2302283002
Cr-Commit-Position: refs/heads/master@{#39135}
Collect type feedback in the bytecode handler for 'new' bytecode. The
earlier cl (https://codereview.chromium.org/2153433002/) was reverted
because that implementation did not collect allocation site feedback.
This regressed delta blue by an order of magnitude. This implementation
includes collection of allocation site feedback.
Reland of https://codereview.chromium.org/2190293003/ with a bug fix.
BUG=v8:4280, v8:4780
LOG=N
Review-Url: https://codereview.chromium.org/2225923003
Cr-Commit-Position: refs/heads/master@{#39120}
There's no need to preserve the exact callee for lazy bailouts
from JSCallFunction in the AstGraphBuilder, as fullcodegen code
will never look at that value after the callee returns. So we
just push optimized_out instead.
BUG=v8:5267
R=jarin@chromium.org
Review-Url: https://codereview.chromium.org/2285183002
Cr-Commit-Position: refs/heads/master@{#38963}
Reason for revert:
[Sheriff] Fails on nosnap debug:
https://build.chromium.org/p/client.v8/builders/V8%20Linux%20-%20nosnap%20-%20debug/builds/8403
Original issue's description:
> [Interpreter] Collect type feedback for 'new' in the bytecode handler
>
> Collect type feedback in the bytecode handler for 'new' bytecode. The
> earlier cl (https://codereview.chromium.org/2153433002/) was reverted
> because that implementation did not collect allocation site feedback.
> This regressed delta blue by an order of magnitude. This implementation
> includes collection of allocation site feedback.
>
> BUG=v8:4280, v8:4780
> LOG=N
>
> Committed: https://crrev.com/9d5e6129c4c7f9cbfe81a5fad2a470f219fe137c
> Cr-Commit-Position: refs/heads/master@{#38364}
TBR=bmeurer@chromium.org,rmcilroy@chromium.org,balazs.kilvady@imgtec.com,mythria@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=v8:4280, v8:4780
Review-Url: https://codereview.chromium.org/2212343002
Cr-Commit-Position: refs/heads/master@{#38368}
Collect type feedback in the bytecode handler for 'new' bytecode. The
earlier cl (https://codereview.chromium.org/2153433002/) was reverted
because that implementation did not collect allocation site feedback.
This regressed delta blue by an order of magnitude. This implementation
includes collection of allocation site feedback.
BUG=v8:4280, v8:4780
LOG=N
Review-Url: https://codereview.chromium.org/2190293003
Cr-Commit-Position: refs/heads/master@{#38364}
This adds the possibility to address test cases in the
status file with the variant under which the test is running.
This is only allowed in top-level sections.
Example:
[{
'test-case': [PASS, SLOW],
}]
['variant == foo', {
'test-case': [FAIL],
}]
The test case "test-case" is marked as slow in all variants.
Additionally, in variant foo, it'll be expected to fail.
This CL also exemplifies the new feature with test cases
running under the ignition_turbofan variant. The
corresponding legacy flag is deprecated.
BUG=v8:5238
Review-Url: https://codereview.chromium.org/2203013002
Cr-Commit-Position: refs/heads/master@{#38342}
So far we treated SignedSmall and Signed32 feedback the same for number
operations. However it would be beneficial to generate (a lot) less code
if we only do a Smi check on the inputs instead of doing the full Smi +
HeapNumber + conversion check that we need to do for Signed32 feedback.
R=epertoso@chromium.org
BUG=v8:4583
Review-Url: https://codereview.chromium.org/2207893002
Cr-Commit-Position: refs/heads/master@{#38290}
Without this fix, two rules [PASS, MODIFIER] and [FAIL]
would be joined as [PASS, MODIFIER, FAIL], undermining the
intention of the second rule to mark a test as failing.
BUG=v8:4680,v8:4900
Review-Url: https://codereview.chromium.org/2201033002
Cr-Commit-Position: refs/heads/master@{#38238}
This switches our inlining tests (i.e. cctest/test-run-inlining) to rely
on global object instead of function context specialization, which is
more in sync with what we are actually shipping. It will also allow us
to test inlining with the BytecodeGraphBuilder without having to add
support for function context specialization just for testing purposes.
R=bmeurer@chromium.org
TEST=cctest/test-run-inlining
BUG=v8:5251
Review-Url: https://codereview.chromium.org/2200673002
Cr-Commit-Position: refs/heads/master@{#38209}
The flag was introduced for ignition development. It can only
be used when running ignition tests in isolation on the bots.
The bots only use ignition_turbo in isolation since a while
and don't pass the --ignition flag anymore.
BUG=v8:5238
Review-Url: https://codereview.chromium.org/2197123002
Cr-Commit-Position: refs/heads/master@{#38206}
The test was calling OptimizeFunctionOnNextCall on a function before
ever executing it - crankshaft therefore didn't have any type info and
was generating a soft deoptimization bailout. Make sure we execute the
function before calling OptimizeFunctionOnNextCall to avoid this issue.
BUG=
Review-Url: https://codereview.chromium.org/2168603003
Cr-Commit-Position: refs/heads/master@{#38171}
Port 580fdf3c05
This also reverses the MachineType stored for partial unaligned access support
such that it records the unsupported types, rather than supported types.
BUG=
Review-Url: https://codereview.chromium.org/2182493003
Cr-Commit-Position: refs/heads/master@{#38065}
Always use the BytecodeGraphBuilder when the --turbo-from-bytecode
is enabled, assuming the function should be compiled for Ignition.
Adds a new MaybeOptimizeIgnition function to runtime-profiler
which is called if the function should be optimized from bytecode
rather than going via full-codegen.
BUG=v8:4280
Committed: https://crrev.com/9ca7db914be88e6792a88eab4a1988ee031d70c4
Review-Url: https://codereview.chromium.org/2156753002
Cr-Original-Commit-Position: refs/heads/master@{#37921}
Cr-Commit-Position: refs/heads/master@{#38002}
Skip test-run-load-store/RunUnalignedLoadStoreUnalignedAccess on ARM until
UnalignedLoad and UnalignedStore operators get implemented
NOTRY=true
Review-Url: https://codereview.chromium.org/2182433002
Cr-Commit-Position: refs/heads/master@{#38001}
Reason for revert:
Breaks tsan:
https://build.chromium.org/p/client.v8/builders/V8%20Linux64%20TSAN/builds/10758
Original issue's description:
> [Intepreter] Always use BytecodeGraphBuilder when --turbo-from-bytecode
>
> Always use the BytecodeGraphBuilder when the --turbo-from-bytecode
> is enabled, assuming the function should be compiled for Ignition.
> Adds a new MaybeOptimizeIgnition function to runtime-profiler
> which is called if the function should be optimized from bytecode
> rather than going via full-codegen.
>
> BUG=v8:4280
>
> Committed: https://crrev.com/9ca7db914be88e6792a88eab4a1988ee031d70c4
> Cr-Commit-Position: refs/heads/master@{#37921}
TBR=mstarzinger@chromium.org,rmcilroy@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=v8:4280
Review-Url: https://codereview.chromium.org/2165223002
Cr-Commit-Position: refs/heads/master@{#37925}
Always use the BytecodeGraphBuilder when the --turbo-from-bytecode
is enabled, assuming the function should be compiled for Ignition.
Adds a new MaybeOptimizeIgnition function to runtime-profiler
which is called if the function should be optimized from bytecode
rather than going via full-codegen.
BUG=v8:4280
Review-Url: https://codereview.chromium.org/2156753002
Cr-Commit-Position: refs/heads/master@{#37921}
Reason for revert:
This cl causes a large regression in octane (https://chromeperf.appspot.com/group_report?bug_id=629503). I have to investigate the reason before I can reland this.
Original issue's description:
> [Interpreter] Collect type feedback for 'new' in the bytecode handler
>
> Collect type feedback in the bytecode handler for 'new' bytecode. The
> current implementation does not collect allocation site feedback.
>
> BUG=v8:4280, v8:4780
> LOG=N
>
> Committed: https://crrev.com/1eadc76419b323fb2e55ae9953142f801704aa59
> Cr-Commit-Position: refs/heads/master@{#37862}
TBR=rmcilroy@chromium.org,bmeurer@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=v8:4280, v8:4780
Review-Url: https://codereview.chromium.org/2165633003
Cr-Commit-Position: refs/heads/master@{#37872}
Collect type feedback in the bytecode handler for 'new' bytecode. The
current implementation does not collect allocation site feedback.
BUG=v8:4280, v8:4780
LOG=N
Review-Url: https://codereview.chromium.org/2153433002
Cr-Commit-Position: refs/heads/master@{#37862}
This allows to pass deoptimization reasons to the profiler without the
requirement of always providing a source position. The absence of deopt
reasons is now communicated via a sentinel as the deopt id value. The
deoptimization reasons recently added to TurboFan are now passed to the
profiler.
R=bmeurer@chromium.org
TEST=cctest/test-cpu-profiler
Review-Url: https://codereview.chromium.org/2159793002
Cr-Commit-Position: refs/heads/master@{#37852}
Collect type feedback in the call bytecode handler. The current
implementation only collects feedback for JS function objects. The other
objects and Array functions do not collect any feedback. They will be
marked Megamorphic.
BUG=v8:4280, v8:4780
LOG=N
Review-Url: https://codereview.chromium.org/2122183002
Cr-Commit-Position: refs/heads/master@{#37700}
