With this CL the out-of-line code of TrapIf will call a builtin instead
of doing a direct runtime call, which is cheaper. In the best case, the
out-of-line code now consists of a single call instruction. The builtin
will load the trapID and then call the runtime to throw a trap.
R=titzer@chromium.org, clemensh@chromium.org
Change-Id: I8a75ba09f40236a3d5cbdc17c1317baf0a3c915a
Reviewed-on: https://chromium-review.googlesource.com/445937
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Ben Titzer <titzer@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#43366}
Test the wasm interpreter entry stub by creating two wasm functions A
and B, make A pass arguments to B, then redirect B to be executed in the
interpreter.
Test different number and types or arguments.
BUG=v8:5822
R=titzer@chromium.org
Review-Url: https://codereview.chromium.org/2651793003
Cr-Commit-Position: refs/heads/master@{#43353}
Native resources allocated by v8, as internal implementation detail,
and held by a Foreign object, must be released when the Isolate is
torn down. Example: wasm::WasmModule allocated by wasm compile, and
held throughout the lifetime of the WebAssembly.Module object.
This change:
- Extends Managed<CppType> with a mechanism for doing just that
- Separates the role of Managed<CppType> to be strictly an owner of
the lifetime of the native resource. For cases where that's not
desirable, we can polymorphically use Foregin.
- moves managed.h out of wasm, since it's not wasm-specific.
BUG=680065
Review-Url: https://codereview.chromium.org/2676513008
Cr-Commit-Position: refs/heads/master@{#43350}
- Adds new machine types SimdBool4/8/16 for the different boolean vector types.
- Adds a kSimdMaskRegisters flag for each platform. These are all false for now.
- Removes Create, ExtractLane, ReplaceLane, Equal, NotEqual, Swizzle and Shuffle
opcodes from the Boolean types. These are unlikely to be well supported natively,
and can be synthesized using Select.
- Changes the signature of Relational opcodes to return boolean vectors.
- Changes the signature of Select opcodes to take boolean vectors.
- Updates the ARM implementation of Relational and Select opcodes.
LOG=N
BUG=v8:4124
Review-Url: https://codereview.chromium.org/2700813002
Cr-Commit-Position: refs/heads/master@{#43348}
This makes it easier to implement asynchronous compilation by hiding all the implementation details of both synchronous and asynchronous compilation within wasm-module.cc, whereas before the code in wasm-js.cc actually implemented asynchronous compilation in terms of synchronous.
BUG=
Review-Url: https://codereview.chromium.org/2695813005
Cr-Commit-Position: refs/heads/master@{#43310}
For functions declared in source code, the .toString() representation
will be an excerpt of the source code.
* For functions declared with the "function" keyword, the excerpt
starts at the "function" or "async" keyword and ends at the final "}".
The previous behavior would start the excerpt at the "(" of the
parameter list, and prepend a canonical `"function " + name` or
similar, which would discard comments and formatting surrounding the
function's name. Anonymous functions declared as function expressions
no longer get the name "anonymous" in their toString representation.
* For methods, the excerpt starts at the "get", "set", "*" (for
generator methods), or property name, whichever comes first.
Previously, the toString representation for methods would use a
canonical prefix before the "(" of the parameter list. Note that any
"static" keyword is omitted.
* For arrow functions and class declarations, the excerpt is unchanged.
For functions created with the Function, GeneratorFunction, or
AsyncFunction constructors:
* The string separating the parameter text and body text is now
"\n) {\n", where previously it was "\n/*``*/) {\n" or ") {\n".
* At one point, newline normalization was required by the spec here,
but that was removed from the spec, and so this CL does not do it.
Included in this CL is a fix for CreateDynamicFunction parsing. ')'
and '`' characters in the parameter string are no longer disallowed,
and Function("a=function(", "}){") is no longer allowed.
BUG=v8:4958, v8:4230
Review-Url: https://codereview.chromium.org/2156303002
Cr-Commit-Position: refs/heads/master@{#43262}
- Renames select, swizzle, and shuffle to be consistent with the S128 and
existing S32x4 ops, and reflect that these aren't arithmetic.
e.g. I16x8Swizzle -> S16x8Swizzle.
- Implements S16x8 and S8x16 Select operations and tests.
- Implements S128And, Or, Xor, Not operations and tests.
- Implements Swizzle for 32x4 formats.
- Refactors test macros that generate SIMD code.
TEST=cctest/test-run-wasm-simd/*
LOG=N
BUG=v8:4124
Review-Url: https://codereview.chromium.org/2683713003
Cr-Commit-Position: refs/heads/master@{#43168}
The use of setjmp/longjmp makes the cctests in test-run-wasm and
test-run-wasm-64 flaky on Windows, and I think that it is better not
to use it. With this CL I replace it as follows:
Similar to the setjmp/longjmp implementation we still call a C
function when a trap happens. However, instead of calling longjmp in
this C function we just set a flag which indicates that a trap
happened and then return. After we return from the C function we leave
the frame of the current wasm function and return with a RET
instruction. At the end of a test the wasm test runner checks the flag
to see if a trap happened.
Please take a special look at the LeaveFrame function on arm64.
R=titzer@chromium.org, clemensh@chromium.org, v8-arm-ports@googlegroups.comCC=jarin@chromium.org
Review-Url: https://codereview.chromium.org/2685583003
Cr-Commit-Position: refs/heads/master@{#43095}
- Adds logical and arithmetic shifts for all integer types.
- Adds min and max for all integer types.
- Adds saturating add and subtract for small integer types.
- Removes lane operations from the MachineOperatorCache.
LOG=N
BUG=v8:4124
Review-Url: https://codereview.chromium.org/2668013003
Cr-Commit-Position: refs/heads/master@{#43005}
Apparently it happens quite easily that different NaNs are produced in
the interpreter than in the execution of the compiled code. This
non-determinism caused problems for the fuzzer which compares the
equality of the results of the interpreter and the compiled code.
I decided therefore to refactor the detection of non-determinism in the
interpreter. Instead of tracking whether potentially non-deterministic
NaNs were produced, I track now whether potentially non-deterministic
NaNs could have been observed. The only way the NaN non-determinism can
be observed is by observing the non-deterministic bit pattern of the
NaN. AFAICT the only way to observe the bit pattern is with a
I(32|64)_REINTERPRET_F(32|64) instruction or with a F(32|64)_STORE
followed by a load. Therefore I flag an execution as potentially
non-deterministic when either a NaN is reinterpreted to an int, or when
a NaN is stored to memory.
R=titzer@chromium.org, eholk@chromium.org
BUG=682180
Review-Url: https://codereview.chromium.org/2671803002
Cr-Commit-Position: refs/heads/master@{#42917}
I removed some constant folding optimizations for float instruction in
https://codereview.chromium.org/2647353007 because they were incorrect
if the input was a signalling NaN. Removing these optimizations, however
had an unexpectedly big impact on asm.js performance. With this CL I
restore the optimizations again when the source origin is not wasm. In
JavaScript signalling NaNs are not observable and therefore the
optimizations are correct.
R=titzer@chromium.org
BUG=chromium:686654
Review-Url: https://codereview.chromium.org/2666903002
Cr-Commit-Position: refs/heads/master@{#42850}
Always execute the implicit return if we fell off the function bytes.
This is not considered an additional "step" as it is not executing a
wasm instruction.
Otherwise, we might pause at an invalid position (one after the
function bytes).
R=titzer@chromium.org
BUG=v8:5822
Review-Url: https://codereview.chromium.org/2650293003
Cr-Commit-Position: refs/heads/master@{#42730}
Implement stepping by remembering the current step action in the wasm
interpreter handle in WasmDebugInfo, and using it when continuing
execution in the interpreter.
The control flow is as follows: After module compilation, the user sets
a breakpoint in wasm. The respective function is redirected to the
interpreter and the breakpoint is set on the interpreter. When it is
hit, we notify all debug event listeners, which might prepare stepping.
When returning from these listeners, before continuing execution, we
check whether stepping was requested and continue execution in the
interpreter accordingly.
Stepping from Wasm to JS and vice versa will be implemented and tested
in a follow-up CL. Testing this requires breakpoints and stepping in
Wasm to be exposed via the inspector interface, such that we can write
an inspector test. This mixed JS-Wasm-execution is hard to set up in a
cctest.
R=titzer@chromium.org, yangguo@chromium.org
BUG=
Review-Url: https://codereview.chromium.org/2649533002
Cr-Commit-Position: refs/heads/master@{#42624}
Also fixes check for table segments to be performed against actual size not declared one.
Makes us pass memory.wast and linking.wast tests (modulo issue 5860).
R=titzer@chromium.org
BUG=
Review-Url: https://codereview.chromium.org/2649553002
Cr-Commit-Position: refs/heads/master@{#42607}
Test that setting breakpoints works for wasm, and that they are hit
correctly.
This basically tests all the layers involved: Compiling and running
wasm interpreter entries, passing arguments to the interpreter, storing
break point infos in wasm objects, getting the right BreakLocation from
wasm frames, and getting stack information from interpreted frames.
BUG=v8:5822
R=titzer@chromium.org, yangguo@chromium.org
Review-Url: https://codereview.chromium.org/2629883002
Cr-Commit-Position: refs/heads/master@{#42560}
The index input of load and store nodes has to be of type int32. However
it is still possible that this index input was lowered by the
Int64Lowering, for example if the index was a kTruncateInt64ToInt32
node. With this CL we check if there exists a replacement of the index
and if so, use the replacement instead of the original index as the
input for the lowered load and store nodes.
BUG=v8:5874
R=titzer@chromium.org, eholk@chromium.org
Review-Url: https://codereview.chromium.org/2644603004
Cr-Commit-Position: refs/heads/master@{#42559}
Make wasm code generation (including deserialization) aware of
allow_codegen_callback - if one were set by the host - akin to what we
do for `eval`.
This allows web pages that opt out of unsafe-eval to also opt out of
wasm scenarios.
BUG=v8:5869
Review-Url: https://codereview.chromium.org/2646713002
Cr-Commit-Position: refs/heads/master@{#42519}
Reason for revert:
compilation problems on mips
Original issue's description:
> [wasm] Fix I32ReinterpretF32 and I64ReinterpretF64 on ia32.
>
> On ia32 return statements in C++ automatically convert signalling NaNs
> to quiet NaNs, even when bit_cast is used. This CL removes all uses of
> bit_cast<float> and bit_cast<double> in the wasm compiler and wasm
> interpreter.
>
> R=titzer@chromium.org, clemensh@chromium.org
>
> Review-Url: https://codereview.chromium.org/2639353002
> Cr-Commit-Position: refs/heads/master@{#42512}
> Committed: 7739affa5bTBR=clemensh@chromium.org,titzer@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
Review-Url: https://codereview.chromium.org/2645693003
Cr-Commit-Position: refs/heads/master@{#42514}
On ia32 return statements in C++ automatically convert signalling NaNs
to quiet NaNs, even when bit_cast is used. This CL removes all uses of
bit_cast<float> and bit_cast<double> in the wasm compiler and wasm
interpreter.
R=titzer@chromium.org, clemensh@chromium.org
Review-Url: https://codereview.chromium.org/2639353002
Cr-Commit-Position: refs/heads/master@{#42512}
The IA32AddPair and IA32SubPair instructions were using an input register as a
temporary value, which led to registers sometimes being clobbered when they
shouldn't have been. This led to problems, for example, in calling printf to
format doubles:
printf("%f", 1.2345) => 0.61725 (on x86)
BUG= https://bugs.chromium.org/p/v8/issues/detail?id=5800
Review-Url: https://codereview.chromium.org/2637583002
Cr-Commit-Position: refs/heads/master@{#42486}
Instead, it is supposed to just return an empty context if it failed.
Also don't invoke interceptors (we don't for the parts that deserialize
from the snapshot anyways).
BUG=v8:5830
R=yangguo@chromium.org
Review-Url: https://codereview.chromium.org/2636903002
Cr-Commit-Position: refs/heads/master@{#42404}
Reason for revert:
OK, the failure really does seem to be due to this patch: It triggers Clang to crash
FAILED: obj/test/unittests/unittests/function-body-decoder-unittest.obj
E:\b\build\slave\cache\cipd\goma/gomacc.exe ../../third_party/llvm-build/Release+Asserts/bin/clang-cl.exe /nologo /showIncludes /FC @obj/test/unittests/unittests/function-body-decoder-unittest.obj.rsp /c ../../test/unittests/wasm/function-body-decoder-unittest.cc /Foobj/test/unittests/unittests/function-body-decoder-unittest.obj /Fd"obj/test/unittests/unittests_cc.pdb"
Assertion failed: (NumGaps == 0 || Bias < MaxDefRange) && "large ranges should not have gaps", file E:\b\build\slave\win_upload_clang\build\src\third_party\llvm\lib\MC\MCCodeView.cpp, line 531
Wrote crash dump file "C:\Users\CHROME~2\AppData\Local\Temp\goma_temp.5068\clang-cl.exe-563144.dmp"
Let's leave it out for now.
Original issue's description:
> Reland of [wasm] Enforce that function bodies end with the \"end\" opcode. (patchset #1 id:1 of https://codereview.chromium.org/2628883006/ )
>
> Reason for revert:
> Try a reland; this might not have been the source of tree-closing.
>
> Original issue's description:
> > Revert of [wasm] Enforce that function bodies end with the \"end\" opcode. (patchset #3 id:40001 of https://codereview.chromium.org/2630553002/ )
> >
> > Reason for revert:
> > Caused tree to close by failing compilation:
> >
> > https://build.chromium.org/p/client.v8/builders/V8%20Win64%20-%20clang/builds/4451
> >
> > Original issue's description:
> > > [wasm] Enforce that function bodies end with the \"end\" opcode.
> > >
> > > R=rossberg@chromium.org
> > > BUG=chromium:575167
> > >
> > > Review-Url: https://codereview.chromium.org/2630553002
> > > Cr-Commit-Position: refs/heads/master@{#42286}
> > > Committed: fcc6e85ec6
> >
> > TBR=mtrofin@chromium.org,rossberg@chromium.org,jbroman@chromium.org,titzer@chromium.org
> > # Skipping CQ checks because original CL landed less than 1 days ago.
> > NOPRESUBMIT=true
> > NOTREECHECKS=true
> > NOTRY=true
> > BUG=chromium:575167
> >
> > Review-Url: https://codereview.chromium.org/2628883006
> > Cr-Commit-Position: refs/heads/master@{#42287}
> > Committed: 1d32a3989b
>
> TBR=mtrofin@chromium.org,rossberg@chromium.org,jbroman@chromium.org,titzer@chromium.org
> # Skipping CQ checks because original CL landed less than 1 days ago.
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=chromium:575167
>
> Review-Url: https://codereview.chromium.org/2628203003
> Cr-Commit-Position: refs/heads/master@{#42296}
> Committed: e539bd8e0eTBR=mtrofin@chromium.org,rossberg@chromium.org,jbroman@chromium.org,titzer@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:575167
Review-Url: https://codereview.chromium.org/2633583002
Cr-Commit-Position: refs/heads/master@{#42298}
Reason for revert:
Try a reland; this might not have been the source of tree-closing.
Original issue's description:
> Revert of [wasm] Enforce that function bodies end with the \"end\" opcode. (patchset #3 id:40001 of https://codereview.chromium.org/2630553002/ )
>
> Reason for revert:
> Caused tree to close by failing compilation:
>
> https://build.chromium.org/p/client.v8/builders/V8%20Win64%20-%20clang/builds/4451
>
> Original issue's description:
> > [wasm] Enforce that function bodies end with the \"end\" opcode.
> >
> > R=rossberg@chromium.org
> > BUG=chromium:575167
> >
> > Review-Url: https://codereview.chromium.org/2630553002
> > Cr-Commit-Position: refs/heads/master@{#42286}
> > Committed: fcc6e85ec6
>
> TBR=mtrofin@chromium.org,rossberg@chromium.org,jbroman@chromium.org,titzer@chromium.org
> # Skipping CQ checks because original CL landed less than 1 days ago.
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=chromium:575167
>
> Review-Url: https://codereview.chromium.org/2628883006
> Cr-Commit-Position: refs/heads/master@{#42287}
> Committed: 1d32a3989bTBR=mtrofin@chromium.org,rossberg@chromium.org,jbroman@chromium.org,titzer@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:575167
Review-Url: https://codereview.chromium.org/2628203003
Cr-Commit-Position: refs/heads/master@{#42296}
- Refactor Dispatch tables to have separate function, signature tables
- New Relocation type for WasmFunctionTableReference, assembler, compiler support.
- RelocInfo helper functions for Wasm references
Review-Url: https://codereview.chromium.org/2627543003
Cr-Commit-Position: refs/heads/master@{#42192}
Reason for revert:
F32x4Add / Sub are still failing. I'll have to investigate on ARM hardware when I get back.
Original issue's description:
> [WASM] Fix failing Wasm SIMD F32x4 tests.
> - Perform lane checks using FP compare instead of reinterpret casts. 0 and -0
> will be different under I32 compare.
> - Some arithmetic operations can generate NaN results, such as adding -Inf
> and +Inf. Skip these tests until we have a way to do more sophisticated
> FP comparisons in the SIMD tests.
> - Eliminate a redundant F32x4 parameter for FP SIMD vector checking. We will only have this one FP type.
>
> LOG=N
> BUG=v8:4124
>
> Review-Url: https://codereview.chromium.org/2594043002
> Cr-Commit-Position: refs/heads/master@{#42154}
> Committed: 5560bbb498TBR=titzer@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=v8:4124
Review-Url: https://codereview.chromium.org/2624713002
Cr-Commit-Position: refs/heads/master@{#42162}
- Perform lane checks using FP compare instead of reinterpret casts. 0 and -0
will be different under I32 compare.
- Some arithmetic operations can generate NaN results, such as adding -Inf
and +Inf. Skip these tests until we have a way to do more sophisticated
FP comparisons in the SIMD tests.
- Eliminate a redundant F32x4 parameter for FP SIMD vector checking. We will only have this one FP type.
LOG=N
BUG=v8:4124
Review-Url: https://codereview.chromium.org/2594043002
Cr-Commit-Position: refs/heads/master@{#42154}
This CL simplifies the relation between the wasm graph builder, the
wasm decoder, and the wasm module they work on.
BUG=
Review-Url: https://codereview.chromium.org/2612643002
Cr-Commit-Position: refs/heads/master@{#42056}
On ARM Neon at least, denormals flush to zero, which may not match
regular FP behavior.
LOG=N
BUG=v8:4124
Review-Url: https://codereview.chromium.org/2598583002
Cr-Commit-Position: refs/heads/master@{#41895}
