- 15 Oct, 2020 19 commits
-
-
Ross McIlroy authored
In certain situations a phi might not be used by later code, and so is neither spilled nor has a register allocated to it. Handle this by removing the incorrect DCHECK. BUG=chromium:1137979,v8:9684 Change-Id: I702dc05dba22e23dac5c1a366a770f18bac45c52 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2471998 Commit-Queue: Tobias Tebbi <tebbi@chromium.org> Reviewed-by:
Tobias Tebbi <tebbi@chromium.org> Auto-Submit: Ross McIlroy <rmcilroy@chromium.org> Cr-Commit-Position: refs/heads/master@{#70536}
-
Manos Koukoutos authored
Changes: - Introduce turbofan builtin WasmAllocatePair. - Implement call_ref for WasmJSFunction in wasm-compiler.cc. - Remove WasmJSFunction trap. - Improve and extend call-ref.js test. Bug: v8:9495 Change-Id: I8b4d1ab70cbbe9ae37887a6241d409eec638fd28 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2463226 Commit-Queue: Manos Koukoutos <manoskouk@chromium.org> Reviewed-by:
Clemens Backes <clemensb@chromium.org> Cr-Commit-Position: refs/heads/master@{#70535}
-
Ross McIlroy authored
When merging blocks in the mid-tier register allocator, try to avoid spilling registers if we can instead introduce a gap move to move the register choosen to hold a virtual register to the register that is holding that virtual register in the successor block. This gives around 1.5% speedup on Octane. BUG=v8:9684 Change-Id: I352d1e0e2c3b89ab6bd203d76339b93ef474028f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2466117 Commit-Queue: Ross McIlroy <rmcilroy@chromium.org> Reviewed-by:
-
Martin Bidlingmaier authored
Bug: v8:11023 Change-Id: I0722338c6d82b665120f9ef8c7076e0c4983b4c2 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2474773Reviewed-by:
Jakob Gruber <jgruber@chromium.org> Commit-Queue: Martin Bidlingmaier <mbid@google.com> Cr-Commit-Position: refs/heads/master@{#70533}
-
Victor Gomes authored
- Add a frame inheritance hierarchy comment. - Rename StandardFrame from frame.h to CommonFrame: StandardFrame usually means a JavaScript frame in other files. - Create a TypedFrame and adjust every frame that depends on it. Change-Id: I105a4ba95954c02e43bcbe2b677e554b9e9af092 Bug: v8:10933 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2431568 Commit-Queue: Victor Gomes <victorgomes@chromium.org> Reviewed-by:
Igor Sheludko <ishell@chromium.org> Reviewed-by:
Toon Verwaest <verwaest@chromium.org> Reviewed-by:
Jakob Kummerow <jkummerow@chromium.org> Cr-Commit-Position: refs/heads/master@{#70532}
-
Victor Gomes authored
Change-Id: Idd0443968cc097a4e7339d7f26ca049909a8eddc Bug: chromium:1138776, v8:10201 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2474791Reviewed-by:
Andreas Haas <ahaas@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Auto-Submit: Victor Gomes <victorgomes@chromium.org> Cr-Commit-Position: refs/heads/master@{#70531}
-
Zhao Jiazhong authored
The RoundFloat/RoundDouble functions should return the Canonical NaN if the input is a NaN. Change-Id: I19928a8a3d78770757c6fe2e240254efe9944bdc Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2475493 Commit-Queue: Zhao Jiazhong <zhaojiazhong-hf@loongson.cn> Reviewed-by:
-
Clemens Backes authored
... at function start. Otherwise we run into a position mismatch: In a non-flooded function, we add the function-entry breakpoint (for "hook on function call") with the position of the first opcode. In the flooded function though, we skip that special breakpoint because we will stop at the first instruction anyway. But then the first instruction is non-breakable, so we don't actually emit a breakpoint for it. Hence during OSR we do not find a corresponding position in the new code. This CL fixes this by postponing the function-entry breakpoint until the first breakable opcode is found, and only emits it if that position does not have a breakpoint anyway. This way, we can also move the handling for function-entry breakpoints from {StartFunctionBody} to {EmitDebuggingInfo}, where it fits much better. R=thibaudm@chromium.org Bug: chromium:1137710 Change-Id: Idfa658fa0897cca89ba5ee3066cd414f68864d06 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2474774Reviewed-by:Thibaud Michaud <thibaudm@chromium.org> Commit-Queue: Clemens Backes <clemensb@chromium.org> Cr-Commit-Position: refs/heads/master@{#70529}
-
Clemens Backes authored
Stepping always happens in Liftoff now, and always by byte offset. Thus remove the redundant "wasm-stepping-byte-offset" test, which was fully subsumed by "wasm-stepping-liftoff". Also, rename "wasm-stepping-liftoff" to "wasm-stepping". R=thibaudm@chromium.org Bug: chromium:1137710 Change-Id: Ifb68ce795ecdcbb1f85500dc4be4c2e64d15a9c6 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2474116Reviewed-by:
-
Michael Lippautz authored
Template specializations must use exact types to match. Bug: chromium:1056170 Change-Id: I2278e9b40712fc209044fe565023029cc5ae3ff3 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2474776 Commit-Queue: Michael Lippautz <mlippautz@chromium.org> Reviewed-by:
Omer Katz <omerkatz@chromium.org> Cr-Commit-Position: refs/heads/master@{#70527}
-
Georg Neis authored
In particular: initial values of local ArchOpcode variables that get overwritten anyways. Creating these variables uninitialized makes that obvious. Change-Id: Ia205b5397c60769a46bf28ed60b299ac652f4b28 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2470557 Auto-Submit: Georg Neis <neis@chromium.org> Reviewed-by:
Nico Hartmann <nicohartmann@chromium.org> Commit-Queue: Nico Hartmann <nicohartmann@chromium.org> Cr-Commit-Position: refs/heads/master@{#70526}
-
Omer Katz authored
Cppgc exposes EphemeronPair that contains a WeakMember key and a Member value and can be used to denote ephemeron semantics in the standalone library. Tracing EphemeronPairs goes through TraceEphemeron that is exposed on the api for the blink usecase. Bug: chromium:1056170 Change-Id: I9fbaa284fa2034248cdf36ea8b0cd5be6a55f676 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2467842 Commit-Queue: Omer Katz <omerkatz@chromium.org> Reviewed-by:
Michael Lippautz <mlippautz@chromium.org> Cr-Commit-Position: refs/heads/master@{#70525}
-
Santiago Aboy Solanes authored
This gives Arm32/64 test parity with x64. Bug: v8:10833 Change-Id: I51c3a61c1529dd17782c60ca5aa6508c6e57ce1a Fixed: v8:10833 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2467850 Commit-Queue: Santiago Aboy Solanes <solanes@chromium.org> Reviewed-by:
Ross McIlroy <rmcilroy@chromium.org> Cr-Commit-Position: refs/heads/master@{#70524}
-
Javad Amiri authored
Bug: v8:9533 Change-Id: I912bd5acd2cdb4c9d111711d17a01ba635b76660 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2463006 Commit-Queue: Ulan Degenbaev <ulan@chromium.org> Reviewed-by:
Ulan Degenbaev <ulan@chromium.org> Reviewed-by:
-
Georg Neis authored
This reverts commit 44708a5b. Reason for revert: https://ci.chromium.org/p/v8/builders/ci/V8%20Linux64%20TSAN/33692 Original change's description: > [compiler, heap] Create LocalHeap outside of ExecuteJob > > Create LocalHeap directly in the Task or in GetOptimizedCodeNow and > pass its reference as argument to ExecuteJob. This allows us to create > LocalHeap differently for the main and background thread, e.g. by > passing an additional argument to the constructor in the future. > It will be required in the future anyways when the main thread will > have its own LocalHeap/LocalIsolate. > > Extending the scope of LocalHeap, also made > HandleBase::IsDereferenceAllowed more precise and uncovered two > potential issues: heap accesses in > OptimizingCompileDispatcher::CompileNext and PipelineImpl::AssembleCode > with --code-comments. > > LocalHeap can now be created in the parked state. Also fixed a data > race with LocalHeap's destructor publishing write barrier entries > without holding the lock. > > Bug: v8:10315 > Change-Id: I9226972601a07b87108cd66efbbb6a0d118af58d > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2460818 > Commit-Queue: Georg Neis <neis@chromium.org> > Reviewed-by: Leszek Swirski <leszeks@chromium.org> > Reviewed-by: Santiago Aboy Solanes <solanes@chromium.org> > Reviewed-by: Georg Neis <neis@chromium.org> > Reviewed-by: Ulan Degenbaev <ulan@chromium.org> > Cr-Commit-Position: refs/heads/master@{#70521} TBR=ulan@chromium.org,neis@chromium.org,leszeks@chromium.org,solanes@chromium.org,dinfuehr@chromium.org Change-Id: I9dd1f8ca6237d5716b6d8938cef0ee3f642f3166 No-Presubmit: true No-Tree-Checks: true No-Try: true Bug: v8:10315 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2474118Reviewed-by:
Georg Neis <neis@chromium.org> Commit-Queue: Georg Neis <neis@chromium.org> Cr-Commit-Position: refs/heads/master@{#70522}
-
Dominik Inführ authored
Create LocalHeap directly in the Task or in GetOptimizedCodeNow and pass its reference as argument to ExecuteJob. This allows us to create LocalHeap differently for the main and background thread, e.g. by passing an additional argument to the constructor in the future. It will be required in the future anyways when the main thread will have its own LocalHeap/LocalIsolate. Extending the scope of LocalHeap, also made HandleBase::IsDereferenceAllowed more precise and uncovered two potential issues: heap accesses in OptimizingCompileDispatcher::CompileNext and PipelineImpl::AssembleCode with --code-comments. LocalHeap can now be created in the parked state. Also fixed a data race with LocalHeap's destructor publishing write barrier entries without holding the lock. Bug: v8:10315 Change-Id: I9226972601a07b87108cd66efbbb6a0d118af58d Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2460818 Commit-Queue: Georg Neis <neis@chromium.org> Reviewed-by:
Leszek Swirski <leszeks@chromium.org> Reviewed-by:
Santiago Aboy Solanes <solanes@chromium.org> Reviewed-by:
-
Ng Zhi An authored
These instructions will be used for prototyping Wasm SIMD's store lane later on, separated the implementation for assembler and disassembler into this patch to make things smaller. Curiously, movhps and movlhps seems to have the same encoding, 0f 16, so I'm not sure not sure how to differentiate them in the disassembler besides using the mod field, since movlhps only takes xmm registers, whereas movhps always take 1 operand. Bug: v8:10975 Change-Id: I8be9a31b1c9a5515038f9c8c55ef30d1ba063ea7 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2471977Reviewed-by:
Bill Budge <bbudge@chromium.org> Commit-Queue: Zhi An Ng <zhin@chromium.org> Cr-Commit-Position: refs/heads/master@{#70520}
-
Zhao Jiazhong authored
dst may be the same register as src0/src1, so it shouldn't be overwritten if we still need src0 and src1's values. And the NaN was not properly canonicalized, this CL adds fmin/fmax instructions to canonicalize the result. Change-Id: Ia65829015eb6c4de160298719d694ca9490883b7 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2465775Reviewed-by:
Zhi An Ng <zhin@chromium.org> Commit-Queue: Zhao Jiazhong <zhaojiazhong-hf@loongson.cn> Cr-Commit-Position: refs/heads/master@{#70519}
-
Ng Zhi An authored
LoadTransform is always a SIMD operation (it always results in a v128), so it should unconditionally set has_simd_. Bug: chromium:1137583 Change-Id: I8496e787c89edec734f4bbfd16dd8b5995fab98a Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2472638Reviewed-by:
-
- 14 Oct, 2020 21 commits
-
-
Ng Zhi An authored
Make everything consistent, pinsr family was converted in https://crrev.com/c/2443494. Bug: v8:10933 Change-Id: I9d09bd477520ce71fccdcf4336135b54c058185c Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2470203Reviewed-by:
-
Ng Zhi An authored
We were already using it to define the SSE instructions. Bug: v8:10933 Change-Id: I8c70c027449ee8b0d00a06298087310ced11cafc Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2470200Reviewed-by:
-
Michael Lippautz authored
Bug: chromium:1056170 Change-Id: Ib03a09db010f3ad06701520fc39e7e83055dbb9e Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2467855 Commit-Queue: Michael Lippautz <mlippautz@chromium.org> Reviewed-by:
-
Z Nguyen-Huu authored
It is related to Reduce consecutive overflow addition with constants. Turned out that we needs to consider also effect use before relaxing it. This fixed the issue that fuzzer found in e93a369f. Bug: chromium:1137586 Change-Id: I32fee5ecc7a6ce40d6f739f9c6e2440a647a2222 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2469597 Commit-Queue: Z Nguyen-Huu <duongn@microsoft.com> Reviewed-by:
-
Milad Fa authored
kModeMask needs to match the previous declaration of mode_mask which was removed in this CL: https://crrev.com/c/2452689 Bug: v8:11007 Change-Id: I2435309b0147b05438902eef440816e3f82aff9b Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2466053 Commit-Queue: Milad Fa <mfarazma@redhat.com> Reviewed-by:
-
Milad Fa authored
Due to a bug on AIX, some of the glibc FP functions do not preserve the sign bit when a negative input is passed by value and the output is rounded to 0: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97086 This CL forces the use of "-0.0" in such cases. Change-Id: If9935596e32e97720f3cb22f27975267ac1124d7 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2468618Reviewed-by:
-
Ng Zhi An authored
The todo is "fixed", in that we found the root cause, and recent refactorings have given us more breathing space in the number of opcodes, and also a static_assert was added to give a clearer error message. Bug: v8:10930 Change-Id: Ied47bf6a61a2bc70949c45f9d00d714b313a5192 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2469157Reviewed-by:
-
Martin Bidlingmaier authored
This CL enables the functionality that was added in d4febb6b by flipping the corresponding feature flag. Cq-Include-Trybots: luci.v8.try:v8_linux64_fyi_rel_ng Bug: v8:10765 Bug: v8:11021 Change-Id: Id061a274b016c71e6a4f7d7934a9c287d3124228 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2470568 Commit-Queue: Martin Bidlingmaier <mbid@google.com> Reviewed-by:
-
Zentaro Kavanagh authored
- -Wfinal-dtor-non-final-class warns on classes with final dtors but not final classes. - Error messages are better when the class is marked final. - Fix existing issues in code base and remove warning exemption Bug: chromium:999886 Test: no errors building Change-Id: Ied2a7a2ff890ecbaf0a4c84f5323f0c9d32def58 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2467000Reviewed-by:
-
Clemens Backes authored
Instead of querying the platform for the number of available threads, and allocating exactly N+1 queues, do grow the number of queues dynamically. This allows for more than N+1 concurrent threads, which then allows us to contribute to compilation instead of waiting doing nothing. This will be added in a follow-up CL. Special care is being taken to not synchronize too much between threads. We take a shared mutex whenever stealing tasks, but not on the default path where we pick a unit from the task's own queue. R=thibaudm@chromium.org CC=etiennep@chromium.org Bug: v8:11005 Change-Id: I1f67f15fb22b95ef246c37eb80c03132d8a1d149 Cq-Include-Trybots: luci.v8.try:v8_linux64_tsan_rel_ng Cq-Include-Trybots: luci.v8.try:v8_linux64_tsan_isolates_rel_ng Cq-Include-Trybots: luci.v8.try:v8_linux_gc_stress_dbg_ng Cq-Include-Trybots: luci.v8.try:v8_mac64_gc_stress_dbg_ng Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2467844 Commit-Queue: Clemens Backes <clemensb@chromium.org> Reviewed-by:
-
Dominik Inführ authored
Scope might still be in progress and needs to be closed when starting tear down. There will be no GC after starting tear down anymore. Bug: v8:11022, v8:10315 Change-Id: I50ea02b13b84ef4fbbc08985ca9e25e0b0ec856d Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2470572Reviewed-by:
-
Seth Brenith authored
This change adds test functions to check that the Torque-generated verifiers can catch a few basic kinds of errors and crash the process. Bug: v8:7793 Change-Id: If0d2b1e8834c3e602c2677253ad3a920566414bb Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2469039Reviewed-by:
-
Victor Gomes authored
Change-Id: I468d64df5d1a06a395249d16c8974d3dec85fe7b Bug: chromium:1138197, v8:10201 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2470570 Commit-Queue: Victor Gomes <victorgomes@chromium.org> Commit-Queue: Leszek Swirski <leszeks@chromium.org> Auto-Submit: Victor Gomes <victorgomes@chromium.org> Reviewed-by:
-
Almothana Athamneh authored
Bug: v8:11018 Change-Id: I36fc948ebea7c5a648307f82e1940678068d2990 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2470559 Commit-Queue: Almothana Athamneh <almuthanna@chromium.org> Reviewed-by:
Liviu Rau <liviurau@chromium.org> Reviewed-by:
Michael Achenbach <machenbach@chromium.org> Cr-Commit-Position: refs/heads/master@{#70504}
-
Vicky Kontoura authored
This CL adds a basic tiering strategy for the js-to-wasm wrappers. When applicable, calls to exported WebAssembly functions are initially handled through the generic js-to-wasm wrapper. If these calls through the generic wrapper reach a constant threshold, the specific (per-signature) wrapper is compiled synchronously for the function and the generic wrapper is replaced. Bug: v8:10982 Change-Id: I65e706daffb5cb6e723ce2f7b785f7ecb7b2fa7b Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2461243 Commit-Queue: Vicky Kontoura <vkont@google.com> Reviewed-by:
-
Victor Gomes authored
Change-Id: I2f262f4545de9e421310094d0dfab2f6147869b5 Bug: v8:10201 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2466116Reviewed-by:
Junliang Yan <junyan@redhat.com> Reviewed-by:
-
Clemens Backes authored
With boolean validation, we don't keep the PC for stack values any more. This CL fixes the --trace-wasm-decoding logic to just not print the opcode which produced a value. The producer can also be found by looking back in the trace. This also makes the tracing output a lot more concise, hence easier to read. Also fix the TraceFailed method to not try to print buffer relative offsets if no PC is there. R=zhin@chromium.org Bug: v8:10969 Change-Id: I5a7a69ea5aa461a277401d87ee24635266517d3c Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2465837Reviewed-by:
-
Martin Bidlingmaier authored
We fall back from irregexp to the experimental engine if a backtrack limit is exceeded and the experimental engine can handle the regexp. The feature can be turned on with a boolean flag, and an uint-valued flag controls the default backtrack limit. For regexps that are constructed with an explicit backtrack limit (API, %NewRegExpWithBacktrackLimit), we choose the lower of the explicit and default backtrack limits. The default backtrack limit does not apply to regexps that can't be handled by the experimental engine, and for such regexps an explicitly specified backtrack limit is handled as before by returning null if we exceed it. Cq-Include-Trybots: luci.v8.try:v8_linux64_fyi_rel_ng Bug: v8:10765 Change-Id: I580df79bd847520985b6c2c2159bc427315c89d1 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2436341 Commit-Queue: Martin Bidlingmaier <mbid@google.com> Reviewed-by:
-
Victor Gomes authored
Change-Id: Icd094eaa12b957bc7a658807aaa565665a184c81 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2470561 Commit-Queue: Victor Gomes <victorgomes@chromium.org> Commit-Queue: Igor Sheludko <ishell@chromium.org> Reviewed-by:
-
Michael Lippautz authored
Bug: chromium:1056170 Change-Id: I65a2b38c85a93ac2822cb7d2b7ac4bd66540348a Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2468996 Commit-Queue: Michael Lippautz <mlippautz@chromium.org> Reviewed-by:
-
Victor Gomes authored
Change-Id: Ie8e2a87fa079b602f895c3c98053b7e7dfc61f45 Bug: v8:10201 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2440098Reviewed-by:
-
