- 05 Jan, 2018 3 commits
-
-
Benedikt Meurer authored
Previously the Promise builtins would always use a runtime function to schedule a new microtask, which is unnecessarily expensive. Since the runtime function only adds the microtask to a FixedArray (potentially growing that array) and increments the number of pending microtasks, it is fairly straight-forward to do this in CSA land instead. This change improves the Bluebird benchmarks by 2-4% on average. Bug: v8:7253 Change-Id: I77e96b9e5afbb4bdbe129b6bb289d9905ed581bf Reviewed-on: https://chromium-review.googlesource.com/851972 Commit-Queue: Benedikt Meurer <bmeurer@chromium.org> Reviewed-by:
Yang Guo <yangguo@chromium.org> Cr-Commit-Position: refs/heads/master@{#50372}
-
Clemens Hammacher authored
WasmGraphBuilder contains fields which store state which is modified during graph construction, other fields store information which is never modified. Make the latter fields const. R=mtrofin@chromium.org Change-Id: I3ea57fbca6b24247989f5ae7260ffed2013ad82a Reviewed-on: https://chromium-review.googlesource.com/850396 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by:
Mircea Trofin <mtrofin@chromium.org> Cr-Commit-Position: refs/heads/master@{#50371}
-
Clemens Hammacher authored
Add a flag which controls our mitigations for executing untrusted code, in particular mitigations against speculative side channel attacks. The new flag (--untrusted-code-mitigations) is enabled by default, and currently only controls the mitigations we do in wasm. In the future, it should also replace the --extra-masking flag controlling our javascript mitigations. R=bradnelson@chromium.org, bmeurer@chromium.org CC=hablich@chromium.org Bug: chromium:798964 Change-Id: I94a008fdb95b6bce07d547a8b81ec9f23f392496 Reviewed-on: https://chromium-review.googlesource.com/850212Reviewed-by:
Brad Nelson <bradnelson@chromium.org> Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Cr-Commit-Position: refs/heads/master@{#50370}
-
- 04 Jan, 2018 21 commits
-
-
Sathya Gunasekaran authored
Create a new function kind for initializer functions and ban arguments if used in such a function. Bug: v8:5367, v8:7183 Cq-Include-Trybots: luci.v8.try:v8_linux_noi18n_rel_ng Change-Id: Id3089e587b3d6a25f27224045f250e032b831818 Reviewed-on: https://chromium-review.googlesource.com/850547 Commit-Queue: Sathya Gunasekaran <gsathya@chromium.org> Reviewed-by:
Adam Klein <adamk@chromium.org> Cr-Commit-Position: refs/heads/master@{#50369}
-
Yang Guo authored
These APIs have been marked as deprecated for more than 6 months. Cq-Include-Trybots: master.tryserver.chromium.linux:linux_chromium_rel_ng Change-Id: I4f05a8904c367f335be7e159709f564362c8bbd9 Reviewed-on: https://chromium-review.googlesource.com/848782Reviewed-by:
-
Benedikt Meurer authored
This changes the names of the builtins on the Promise.prototype to match the general naming convention and make it easier to tell what the builtin is about just by looking at its name. Bug: v8:7253 Change-Id: I906cc6fc14a6b9d8ffe9c0dcfd94b50439c4cfcf Reviewed-on: https://chromium-review.googlesource.com/850698Reviewed-by:
Sathya Gunasekaran <gsathya@chromium.org> Commit-Queue: Benedikt Meurer <bmeurer@chromium.org> Cr-Commit-Position: refs/heads/master@{#50367}
-
Benedikt Meurer authored
This is a simple refactoring to make the code dealing with the marker symbols for the default resolve/reject handlers more readable. Bug: v8:7253 Change-Id: Ife279e77007d52ad11fed3726009d60d7c352c33 Reviewed-on: https://chromium-review.googlesource.com/850696Reviewed-by:
-
Michal Majewski authored
Bug: v8:6917 Change-Id: I7bc5d09ff19a9cf7cb4bee68ca6cf34f5cfa8b19 Reviewed-on: https://chromium-review.googlesource.com/850874 Commit-Queue: Michael Achenbach <machenbach@chromium.org> Reviewed-by:
Michael Achenbach <machenbach@chromium.org> Cr-Commit-Position: refs/heads/master@{#50365}
-
Sathya Gunasekaran authored
This patch breaks out bailout reasons into two enum classes. This helps save 3 bits on the SharedFunctionInfo as we don't have to track the abort reasons. Change-Id: Ic2e7e7e32b0fa31491f1c6f0003a61390d68fd97 Reviewed-on: https://chromium-review.googlesource.com/848244Reviewed-by:
Ben Titzer <titzer@chromium.org> Reviewed-by:
Benedikt Meurer <bmeurer@chromium.org> Reviewed-by:
Ross McIlroy <rmcilroy@chromium.org> Commit-Queue: Sathya Gunasekaran <gsathya@chromium.org> Cr-Commit-Position: refs/heads/master@{#50364}
-
Mike Stanton authored
The initial fast array may change, invalidating assumptions. Bug: chromium:798026 Change-Id: Iddcc40867221a2a58aef33b64e7399e0f2784e89 Reviewed-on: https://chromium-review.googlesource.com/850356 Commit-Queue: Michael Stanton <mvstanton@chromium.org> Reviewed-by:
Jaroslav Sevcik <jarin@chromium.org> Cr-Commit-Position: refs/heads/master@{#50363}
-
Franziska Hinkelmann authored
This reverts commit d5fbf7c5. Reason for revert: Performance regression, see https://bugs.chromium.org/p/chromium/issues/detail?id=798279 Original change's description: > [api] Prefer Descriptor interceptor over Getter in GetPropertyAttributes > > Also fix GetPropertyDescriptorWithInterceptor so that it only calls the > interceptor once. > > R=ahaas@chromium.org, franzih@chromium.org > > Bug: node:17480, node:17481 > Change-Id: I2c3813f80df2962ec909bae7267884ce0b8ccbef > Reviewed-on: https://chromium-review.googlesource.com/816515 > Commit-Queue: Franziska Hinkelmann <franzih@chromium.org> > Reviewed-by: Franziska Hinkelmann <franzih@chromium.org> > Cr-Commit-Position: refs/heads/master@{#50260} TBR=timothygu99@gmail.com,ahaas@chromium.org,franzih@chromium.org,sergiyb@chromium.org # Not skipping CQ checks because original CL landed > 1 day ago. Bug: node:17480, node:17481 Change-Id: I4997e0f3a330d719026e56dd83c1bb999b986bcf Reviewed-on: https://chromium-review.googlesource.com/850355Reviewed-by:
Franziska Hinkelmann <franzih@chromium.org> Commit-Queue: Franziska Hinkelmann <franzih@chromium.org> Cr-Commit-Position: refs/heads/master@{#50362}
-
Michal Majewski authored
I added additional exception logging in the execution.py since errors in processing results were really difficult to debug. There is a problem on Windows with class serialization when it comes from dynamically loaded module. To fix it I moved all output processors to the tools/testrunner/outproc/ and import them in test/*/testcfg.py. Bug: v8:6917 Change-Id: Ida604641d659b006e91faf1d56a37769ec47f5f3 Reviewed-on: https://chromium-review.googlesource.com/842784 Commit-Queue: Michał Majewski <majeski@google.com> Reviewed-by:
-
Tobias Tebbi authored
DeadValue was a constant node of type None. This is unsound in the presence of re-scheduling. This CL adds a value input to DeadValue, which preserves the dependency on the original node of type None. This reland addresses the bug that the EffectControlLinearizer could destroy dependencies of DeadValue by attaching DeadValue nodes to the effect chain in the EffectControlLinearizer. Bug: chromium:796041 chromium:798938 Change-Id: If47b54a7986d257eb63b437f855769b503679ff5 Reviewed-on: https://chromium-review.googlesource.com/850392Reviewed-by:
-
Benedikt Meurer authored
The Promise builtins call out to arbitrary JavaScript a lot, but always use the generic Call builtin with the "any" mode, which means that the builtin has to dispatch based on the type of receiver. Yet these builtins always know for sure that the receiver is either undefined or definitely not null or undefined. So this changes the builtins to pass the correct ConvertReceiverMode and thus save a couple of checks. Bug: v8:7253 Change-Id: I4efbc1e379169cbeefc40e5455713cd7fbe947ed Reviewed-on: https://chromium-review.googlesource.com/850000 Commit-Queue: Benedikt Meurer <bmeurer@chromium.org> Reviewed-by:
-
Mike Stanton authored
Array.prototype.{forEach, filter, map, every} get this support with the help of a new opcode NumberIsFloat64Hole. Bug: v8:1956 Change-Id: Ic6a785590cec66bae4c1462c19d6843c0aa5473b Reviewed-on: https://chromium-review.googlesource.com/847435Reviewed-by:Daniel Clifford <danno@chromium.org> Commit-Queue: Michael Stanton <mvstanton@chromium.org> Cr-Commit-Position: refs/heads/master@{#50358}
-
Mike Stanton authored
Bug: v8:1956 Change-Id: Ie941811110b3c106e252a2621544864673074da5 Reviewed-on: https://chromium-review.googlesource.com/846759Reviewed-by:
-
Tobias Tebbi authored
Revert "[turbofan] add value input to DeadValue" and "[turbofan] add regression test for chromium:796041" This reverts https://chromium-review.googlesource.com/c/v8/v8/+/848995 and https://chromium-review.googlesource.com/c/v8/v8/+/847011 Bug: chromium:798938 Change-Id: I4be8e5bca77037a278fd9882f0d76de1ae12c23f TBR: jarin@chromium.org Reviewed-on: https://chromium-review.googlesource.com/849995Reviewed-by:
Tobias Tebbi <tebbi@chromium.org> Commit-Queue: Tobias Tebbi <tebbi@chromium.org> Cr-Commit-Position: refs/heads/master@{#50356}
-
Clemens Hammacher authored
This adds support for get_local and set_local to the wasm compile fuzzer. Each function will have between 0 and 32 locals of random type. For set_local, we generate a value of the respective type and store it in the local. For get_local, we load any local and convert it to the wanted type. Note that with get_local, we now also check that parameters are passed correctly between functions. Drive-by: Fix parameters passed to the main function (was [1,2,3] for the interpreter, but [1,1,1] for compiled code). R=ahaas@chromium.org, eholk@chromium.org Change-Id: I38e85fe25b1fb4ac298fa81ec8e33711294e78bb Reviewed-on: https://chromium-review.googlesource.com/847535Reviewed-by:
Eric Holk <eholk@chromium.org> Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Cr-Commit-Position: refs/heads/master@{#50355}
-
Sigurd Schneider authored
These checks seem to be redundant, since the relevant information of both Array.prototype and Object.prototype is already guarded by the NoElements protector, which guarantees no indexed data and accessor properties. Bug: v8:7127, v8:7205 Change-Id: I0785c31670c04a3db29eefdc2f75cc63971fed3b Reviewed-on: https://chromium-review.googlesource.com/848780 Commit-Queue: Sigurd Schneider <sigurds@chromium.org> Reviewed-by:
-
Sigurd Schneider authored
This removes a stability check from CanInlineArrayResizeOperation. This is safe, since all prototype maps are either stable or in dictionary mode, where the latter is ruled out separately. Bug: v8:7127, v8:7205 Change-Id: I3a07805cb89f41adf031c6c2a6dd5b2defcb87a8 Reviewed-on: https://chromium-review.googlesource.com/848778 Commit-Queue: Sigurd Schneider <sigurds@chromium.org> Reviewed-by:
-
Sigurd Schneider authored
Bug: v8:7127, v8:7205 Change-Id: I9c4af8b604845861f4e29b872cc516fdeb89b519 Reviewed-on: https://chromium-review.googlesource.com/848999Reviewed-by:
-
Mostyn Bramley-Moore authored
Bug: v8:7127, v8:7204, v8:7205 Change-Id: Iedea388590991e4d0edcf59e02c767b6c04460e7 Reviewed-on: https://chromium-review.googlesource.com/849216Reviewed-by:
-
Michael Achenbach authored
This reverts commit 1685b5d2. Reason for revert: Was probably caused by infra change: https://crrev.com/c/845781 Original change's description: > Revert "[test] Move has unexpected output to outproc." > > This reverts commit 71605b3e. > > Reason for revert: Seems to break static-initializers step: > https://build.chromium.org/p/client.v8/builders/V8%20Linux64/builds/22156 > > Original change's description: > > [test] Move has unexpected output to outproc. > > > > Expected outcomes optimized to serialize [PASS] as None. > > > > Keeping expected outcomes inside output processors should be > > optimized in the future. Few possible optimizations: > > - separate classes for tests that are expected to PASS - done as > > an example in mozilla test suite. > > - cache output processors inside testcase. > > - share output processors between copies of the same test - needs > > some updates to the create_variant to update outproc only if > > expected outcomes changed. > > > > Bug: v8:6917 > > Change-Id: Ie73f1dcdf17fdfc65bce27228f818b1dd1e420c9 > > Reviewed-on: https://chromium-review.googlesource.com/843025 > > Commit-Queue: Michael Achenbach <machenbach@chromium.org> > > Reviewed-by: Michael Achenbach <machenbach@chromium.org> > > Cr-Commit-Position: refs/heads/master@{#50347} > > TBR=machenbach@chromium.org,sergiyb@chromium.org,majeski@google.com > > Change-Id: Ice1f3aee0a26f7f38996459d38fd6e0bd964113d > No-Presubmit: true > No-Tree-Checks: true > No-Try: true > Bug: v8:6917 > Reviewed-on: https://chromium-review.googlesource.com/849572 > Reviewed-by: Bill Budge <bbudge@chromium.org> > Commit-Queue: Bill Budge <bbudge@chromium.org> > Cr-Commit-Position: refs/heads/master@{#50348} TBR=bbudge@chromium.org,machenbach@chromium.org,sergiyb@chromium.org,majeski@google.com Change-Id: I7a522b6487de6e96985d223524533493eb9171f0 No-Presubmit: true No-Tree-Checks: true No-Try: true Bug: v8:6917 Reviewed-on: https://chromium-review.googlesource.com/848975Reviewed-by:
-
Tobias Tebbi authored
The missing regression test for https://chromium-review.googlesource.com/c/v8/v8/+/847011 Bug: chromium:796041 Change-Id: I3d791d6485221d2fa68def2c7be96c48822aa651 Reviewed-on: https://chromium-review.googlesource.com/848995Reviewed-by:
-
- 03 Jan, 2018 16 commits
-
-
Bill Budge authored
This reverts commit 71605b3e. Reason for revert: Seems to break static-initializers step: https://build.chromium.org/p/client.v8/builders/V8%20Linux64/builds/22156 Original change's description: > [test] Move has unexpected output to outproc. > > Expected outcomes optimized to serialize [PASS] as None. > > Keeping expected outcomes inside output processors should be > optimized in the future. Few possible optimizations: > - separate classes for tests that are expected to PASS - done as > an example in mozilla test suite. > - cache output processors inside testcase. > - share output processors between copies of the same test - needs > some updates to the create_variant to update outproc only if > expected outcomes changed. > > Bug: v8:6917 > Change-Id: Ie73f1dcdf17fdfc65bce27228f818b1dd1e420c9 > Reviewed-on: https://chromium-review.googlesource.com/843025 > Commit-Queue: Michael Achenbach <machenbach@chromium.org> > Reviewed-by: Michael Achenbach <machenbach@chromium.org> > Cr-Commit-Position: refs/heads/master@{#50347} TBR=machenbach@chromium.org,sergiyb@chromium.org,majeski@google.com Change-Id: Ice1f3aee0a26f7f38996459d38fd6e0bd964113d No-Presubmit: true No-Tree-Checks: true No-Try: true Bug: v8:6917 Reviewed-on: https://chromium-review.googlesource.com/849572Reviewed-by:
Bill Budge <bbudge@chromium.org> Commit-Queue: Bill Budge <bbudge@chromium.org> Cr-Commit-Position: refs/heads/master@{#50348}
-
Michal Majewski authored
Expected outcomes optimized to serialize [PASS] as None. Keeping expected outcomes inside output processors should be optimized in the future. Few possible optimizations: - separate classes for tests that are expected to PASS - done as an example in mozilla test suite. - cache output processors inside testcase. - share output processors between copies of the same test - needs some updates to the create_variant to update outproc only if expected outcomes changed. Bug: v8:6917 Change-Id: Ie73f1dcdf17fdfc65bce27228f818b1dd1e420c9 Reviewed-on: https://chromium-review.googlesource.com/843025 Commit-Queue: Michael Achenbach <machenbach@chromium.org> Reviewed-by:
-
Michael Achenbach authored
This seperates OS-specific code of the Command class into a Posix and a WindowsCommand to simplify scattered OS checks. This also removes some temporary mac debug output that's obsolete after https://crbug.com/v8/6927 got resolved. TBR=sergiyb@chromium.org Bug: v8:6917 Change-Id: Iaab3b527ce556dfba797a164ae58e8dd358eb56f Reviewed-on: https://chromium-review.googlesource.com/847000Reviewed-by:
-
Michael Achenbach authored
This reverts commit ce8f5141. Reason for revert: Changes a layout test: https://build.chromium.org/p/client.v8.fyi/builders/V8-Blink%20Linux%2064/builds/20693 Original change's description: > Skip Execution::Call in CompileFunctionInContext. > > We execute the top-level function only to get to the wrapped function. > We could do the same by simply instantiating it. > > Other approaches would change the parser so that the top-level function > is the wrapped function. However, that change violates existing scoping > invariants and fixing it would add a lot more complexity to the parser. > > R=adamk@chromium.org, marja@chromium.org > > Bug: v8:7172 > Change-Id: I7272b8b58fc739d47a44da3a9d4a914af3e8cf3d > Reviewed-on: https://chromium-review.googlesource.com/836367 > Reviewed-by: Adam Klein <adamk@chromium.org> > Commit-Queue: Yang Guo <yangguo@chromium.org> > Cr-Commit-Position: refs/heads/master@{#50333} TBR=adamk@chromium.org,marja@chromium.org,yangguo@chromium.org Change-Id: I922fdd2410512d837705263e092a09e041b9d28f No-Presubmit: true No-Tree-Checks: true No-Try: true Bug: v8:7172 Reviewed-on: https://chromium-review.googlesource.com/849215Reviewed-by:
-
Michael Achenbach authored
This replaces multiprocessing with threading in tests to make python coverage work. TBR=sergiyb@chromium.org Bug: v8:6917 Change-Id: Idff763dfefa4a7fc782133d94089b3a5b00a194d Reviewed-on: https://chromium-review.googlesource.com/844735Reviewed-by:
-
Michael Achenbach authored
The tests already run on upload and commit. This makes them run on the corresponding continuous tester. NOTRY=true TBR=sergiyb@chromium.org Bug: v8:6917 Change-Id: I7dedd482ea54f0879854055c97b27f825f06d6e2 Reviewed-on: https://chromium-review.googlesource.com/846807Reviewed-by:
-
Sigurd Schneider authored
This should improve performance in cases where receiver or argument types are unknown. Bug: v8:7127, v8:7092 Change-Id: I72f1fcdc088bc817c1cc42bf27ecee91965b7680 Reviewed-on: https://chromium-review.googlesource.com/846761Reviewed-by:
-
Sigurd Schneider authored
Bug: v8:7127, v8:7204, v8:7205 Change-Id: I05d6bc2e20e29eaa683ad3aa94af24a4309bcdc7 Reviewed-on: https://chromium-review.googlesource.com/847484Reviewed-by:
-
Tobias Tebbi authored
DeadValue was a constant node of type None. This is unsound in the presence of re-scheduling. This CL adds a value input to DeadValue, which preserves the dependency on the original node of type None. Bug: chromium:796041 Change-Id: I3ac459bf661fb78c56552e8201aa18a7dbc4d182 Reviewed-on: https://chromium-review.googlesource.com/847011 Commit-Queue: Tobias Tebbi <tebbi@chromium.org> Reviewed-by:
-
Franziska Hinkelmann authored
Use the script name from the shared function info to create an inline entry. Otherwise functions are attributed to the wrong file in the CpuProfileNode. See https://github.com/GoogleCloudPlatform/cloud-profiler-nodejs/issues/89 Bug: v8:7203, v8:7241 Change-Id: I8ea31943741770e6611275a9c93375922b934547 Reviewed-on: https://chromium-review.googlesource.com/848093Reviewed-by:
-
Mike Stanton authored
This code is highly repetitive, and while we've held off on large refactorings (using "Builder" classes, templates or macros), it's useful to find a few common points that don't change very much. Bug: Change-Id: Ib0880558cf667f5d07c8343328461609ccb1e6b1 Reviewed-on: https://chromium-review.googlesource.com/848913Reviewed-by:
-
Sigurd Schneider authored
Bug: v8:7127, v8:7204, v8:7205 Change-Id: Idd3c11dc92fbd2c7ccab3aece7b2ad189619fc2c Reviewed-on: https://chromium-review.googlesource.com/833916Reviewed-by:
-
Mythri authored
Now that we have an API to request code cache, we want to decouple compilation from serialization. As a first step, we will add CompileEager option (used when we want to produce full code cache) and DeferredProduceCodeOption to NoCacheReason. This is so that we can properly bucket the compilation time and collect statistics about the cache behaviour. Once, blink and node start using the new API, we can remove the code to produce code cache from the compilation. Bug: chromium:783124 Cq-Include-Trybots: master.tryserver.chromium.linux:linux_chromium_rel_ng Change-Id: I35dbb6b0af39940450d412ff75b769603398b2f6 Reviewed-on: https://chromium-review.googlesource.com/828977 Commit-Queue: Mythri Alle <mythria@chromium.org> Reviewed-by:
-
Benedikt Meurer authored
This allows TurboFan to inline API function calls with more than 7 parameters. It's now possible to inline API function calls with up to 127 parameters, which should be sufficiently large. Bug: chromium:794865 Change-Id: I974fed740cd3506d0b8d75ed922227364c51a1fb Reviewed-on: https://chromium-review.googlesource.com/848712Reviewed-by:
-
Sigurd Schneider authored
Rename FastArray{Push,Pop,Shift,Slice} to ArrayPrototype{Push,Pop,Shift,Slice}. Bug: v8:7127, v8:7204, v8:7205 Change-Id: I409fcb4b088f1acece7c6520690f419cad36e78c Reviewed-on: https://chromium-review.googlesource.com/846982Reviewed-by: -
Yang Guo authored
We execute the top-level function only to get to the wrapped function. We could do the same by simply instantiating it. Other approaches would change the parser so that the top-level function is the wrapped function. However, that change violates existing scoping invariants and fixing it would add a lot more complexity to the parser. R=adamk@chromium.org, marja@chromium.org Bug: v8:7172 Change-Id: I7272b8b58fc739d47a44da3a9d4a914af3e8cf3d Reviewed-on: https://chromium-review.googlesource.com/836367Reviewed-by:
-
