- 05 Jan, 2022 16 commits
-
-
Shu-yu Guo authored
This reverts commit be6bd4f4. Reason for revert: Consistent timeouts on Linux and Mac, e.g. https://ci.chromium.org/ui/p/v8/builders/ci/V8%20Linux%20-%20debug/37973/overview https://ci.chromium.org/ui/p/v8/builders/ci/V8%20Mac64%20-%20debug/37346/overview Original change's description: > [wasm] Fast paths in EvaluateInitExpression > > We add fast paths for the most common types of expressions in > {EvaluateInitExpression} to improve instantiation time. We fall back to > full expression decoding for less common operators, or for expressions > with operands. > > Bug: chromium:1284557 > Change-Id: I39a1816176974058b801cdad6eaaa6da156cea04 > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3367627 > Reviewed-by: Clemens Backes <clemensb@chromium.org> > Commit-Queue: Manos Koukoutos <manoskouk@chromium.org> > Cr-Commit-Position: refs/heads/main@{#78497} Bug: chromium:1284557 Change-Id: If09468eb1e790d4359573ddff8b653fe84b0e11e No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3368602 Auto-Submit: Shu-yu Guo <syg@chromium.org> Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Commit-Queue: Shu-yu Guo <syg@chromium.org> Owners-Override: Shu-yu Guo <syg@chromium.org> Cr-Commit-Position: refs/heads/main@{#78502}
-
Piotr Sikora authored
Signed-off-by:
Piotr Sikora <piotrsikora@google.com> Change-Id: I35415a80ded1a90007c70347e9fffd97f47243a9 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3346681Reviewed-by:
Victor Gomes <victorgomes@chromium.org> Cr-Commit-Position: refs/heads/main@{#78501}
-
Piotr Sikora authored
Signed-off-by:
-
Piotr Sikora authored
Clang 12 doesn't support -Wno-bitwise-instead-of-logical, so silence it with -Wno-unknown-warning-option. GCC requires using GNU dialect of the C++ standard, using optimizations (otherwise "always_inline" fails to inline), and produces a lot of warnings that had to be silenced. Signed-off-by:
-
Omer Katz authored
New text is moved over from BlinkGCDesign.md Bug: chromium:1283934 Change-Id: I10a84c91a642e96c494d6e523d6d89059afaa1ca Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3366658Reviewed-by:
Michael Lippautz <mlippautz@chromium.org> Commit-Queue: Omer Katz <omerkatz@chromium.org> Cr-Commit-Position: refs/heads/main@{#78498}
-
Manos Koukoutos authored
We add fast paths for the most common types of expressions in {EvaluateInitExpression} to improve instantiation time. We fall back to full expression decoding for less common operators, or for expressions with operands. Bug: chromium:1284557 Change-Id: I39a1816176974058b801cdad6eaaa6da156cea04 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3367627Reviewed-by:Clemens Backes <clemensb@chromium.org> Commit-Queue: Manos Koukoutos <manoskouk@chromium.org> Cr-Commit-Position: refs/heads/main@{#78497}
-
Manos Koukoutos authored
We implement loop peeling for wasm, currently available behind a flag. Loops are peeled regardless of size. Bug: v8:11510 Change-Id: Ia4c883abdee83df632b2611584d608c44e3295c8 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3367615Reviewed-by:
Andreas Haas <ahaas@chromium.org> Reviewed-by:
Tobias Tebbi <tebbi@chromium.org> Commit-Queue: Manos Koukoutos <manoskouk@chromium.org> Cr-Commit-Position: refs/heads/main@{#78496}
-
Leszek Swirski authored
This reverts commit c7c5b492. Reason for revert: Looks like test needs to be disabled for noi18n: https://ci.chromium.org/ui/p/v8/builders/ci/V8%20Linux%20-%20noi18n%20-%20debug/40832/overview Original change's description: > [scanner] Combine surrogate pairs at start when scanning private names > > Bug: v8:12523 > Change-Id: Ic3779fe6f20965d177d99d0a570a735df72e4fde > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3366994 > Reviewed-by: Leszek Swirski <leszeks@chromium.org> > Commit-Queue: Shu-yu Guo <syg@chromium.org> > Cr-Commit-Position: refs/heads/main@{#78493} Bug: v8:12523 Change-Id: I678d69a7acb793ed03ce049a05c37685d0cdee1a No-Presubmit: true No-Tree-Checks: true No-Try: true Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3368106 Auto-Submit: Leszek Swirski <leszeks@chromium.org> Commit-Queue: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com> Cr-Commit-Position: refs/heads/main@{#78495}
-
Clemens Backes authored
Introduce a build-time flag to disable all CET shadow-stack manipulation. This will allow us to develop the feature without breaking production code, and enable it all at once once the feature is ready. R=mlippautz@chromium.org Bug: v8:12522, v8:11246, chromium:1284445, chromium:1284599 Change-Id: Iedc1b9a0c0c74f484bb76d86c84809798c0931b9 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3368101Reviewed-by:
-
Shu-yu Guo authored
Bug: v8:12523 Change-Id: Ic3779fe6f20965d177d99d0a570a735df72e4fde Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3366994Reviewed-by:
Leszek Swirski <leszeks@chromium.org> Commit-Queue: Shu-yu Guo <syg@chromium.org> Cr-Commit-Position: refs/heads/main@{#78493}
-
Clemens Backes authored
When computing the code size estimate for {PrepareAndStartCompile}, we did not consider Liftoff code in the async path. Other invocations checked {FLAG_liftoff} to decide whether Liftoff code will be generated. This CL fixes the async path to do the same, and renames {uses_liftoff} to {include_liftoff} to match the name of the parameter in {EstimateNativeModuleCodeSize}. R=ahaas@chromium.org Bug: v8:12520 Change-Id: Ic92237dc05ac96ddd88c3e8788cd443c83bd446f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3367624Reviewed-by: -
Clemens Backes authored
The jump table sizes were added to the estimated code size, and then again added for computing the reservation size for the code. This CL moves the jump table size from {EstimateNativeModuleCodeSize} to {EstimateNativeModuleMetaDataSize} so it is still considered for the total memory associated with the {NativeModule}, but only added once for the code space reservation. R=ahaas@chromium.org Bug: v8:12520 Change-Id: I871e54833659a0d466f3e8359bb3b515c85dd3cb Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3367622Reviewed-by: -
Benedikt Meurer authored
The V8InspectorSessionImpl constructor accepts a state, as either text or CBOR encoded, and generally ignores all invalid inputs, except for the case where it's a valid value, but not a dictionary value, in which case it'll leak the value and crash upon casting to a `DictionaryValue`. This is purely an issue with the test driver, so no security impact on Chromium in the wild. Fixed: chromium:1281031 Change-Id: I7b4d0aea83370499b1274d3fa214a14dc098d2f2 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3361838 Commit-Queue: Benedikt Meurer <bmeurer@chromium.org> Auto-Submit: Benedikt Meurer <bmeurer@chromium.org> Reviewed-by:
Yang Guo <yangguo@chromium.org> Cr-Commit-Position: refs/heads/main@{#78490}
-
Benedikt Meurer authored
This method performs exactly the same operation as the official `v8::Exception::GetStackTrace()`, which is already used in other places, so there's no point to have a duplicate of that in the debug interface. Bug: chromium:1283162 Change-Id: I09dd07f678165e1565bd77173e8ce64636ef649b Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3366659 Auto-Submit: Benedikt Meurer <bmeurer@chromium.org> Reviewed-by:
-
v8-ci-autoroll-builder authored
Rolling v8/build: https://chromium.googlesource.com/chromium/src/build/+log/44c14db..ccc9811 Rolling v8/third_party/catapult: https://chromium.googlesource.com/catapult/+log/1227b26..ec88714 Rolling v8/third_party/depot_tools: https://chromium.googlesource.com/chromium/tools/depot_tools/+log/54c265e..02d65ea Rolling v8/tools/clang: https://chromium.googlesource.com/chromium/src/tools/clang/+log/8b73305..2d10229 R=v8-waterfall-sheriff@grotations.appspotmail.com,mtv-sf-v8-sheriff@grotations.appspotmail.com Change-Id: Iade1fe67ff6f3dea3eacc7b614150da806e3ed20 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3365993 Commit-Queue: v8-ci-autoroll-builder <v8-ci-autoroll-builder@chops-service-accounts.iam.gserviceaccount.com> Bot-Commit: v8-ci-autoroll-builder <v8-ci-autoroll-builder@chops-service-accounts.iam.gserviceaccount.com> Cr-Commit-Position: refs/heads/main@{#78488}
-
Hao Xu authored
Change-Id: I7b20a32973c7592c6e47477b1d98bb0d72b27e33 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3347571Reviewed-by:
-
- 04 Jan, 2022 19 commits
-
-
Piotr Sikora authored
There is no reason for bazel/config to be used as an external dependency (we can replace "@v8//bazel/config" as easily as "@config") and it makes integration with other Bazel workspaces much harded than it needs to be. Signed-off-by:
-
Piotr Sikora authored
Signed-off-by:
-
Milad Fa authored
Naming a class member function the same as a class name could cause compilation issues with gcc: ``` error: changes meaning of 'StackFrameInfo' from 'class v8::internal::StackFrameInfo' ``` This CL changes the function name to fix the problem. Change-Id: I085018504deefefa99dbf2ff8638bc0e872fdbc8 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3366703Reviewed-by:
Benedikt Meurer <bmeurer@chromium.org> Commit-Queue: Milad Farazmand <mfarazma@redhat.com> Cr-Commit-Position: refs/heads/main@{#78484}
-
Clemens Backes authored
If a fixed register is defined for an input, we did only spill the sibling SIMD register if the other sibling was allocated. This is not correct. If only the sibling is in use (e.g. s1 colliding with q0) we also have to spill that sibling. R=mslekova@chromium.org Bug: chromium:1283042, v8:12330 Change-Id: I6a22eaf461774a0b4603ec3ff17062134a528161 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3359615Reviewed-by:
Maya Lekova <mslekova@chromium.org> Commit-Queue: Clemens Backes <clemensb@chromium.org> Cr-Commit-Position: refs/heads/main@{#78483}
-
Clemens Backes authored
The mid-tier register allocator did not handle block merges correctly where a SIMD register was partially overlapping with a non-SIMD register. This CL fixes that, and reorders the code to allow for early exits. R=mslekova@chromium.org Bug: chromium:1282224, v8:12330 Change-Id: I2e9275d5c1aaa764ecb63fbf8fa197b68d6b6c3c Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3358294Reviewed-by:
-
Clemens Backes authored
If such a frame is near the top of the stack frame, move to the frame below instead, which is the caller of OS::DebugBreak. Also, rename dcheck_stop_handler to v8_stop_handler since we handle more than DCHECKs there. R=leszeks@chromium.org No-Try: true Change-Id: Ib31c2dc8278ec779a00babfdc952453e66e5f110 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3366238Reviewed-by:
-
Samuel Groß authored
Previously, guard regions were created by allocating pages with PROT_NONE and relying on an allocation hint. This could fail however, for example on Fuchsia (where it would allocate a VMO to back the guard region) and possibly on Windows (where a placeholder mapping was replaced by a "real" mapping). Introducing an explicit VirtualAddressSpace::AllocateGuardRegion routine now makes this operation more efficient and effectively guarantees that it cannot fail if used correctly: in a regular subspace, there is no need to allocate anything when creating guard regions since the address space reservation backing the subspace is guaranteed to be inaccessible when no pages are allocated in it. Bug: chromium:1218005 Change-Id: I6945f17616b6b8dad47241af96d4cb1f660e8858 Cq-Include-Trybots: luci.v8.try:v8_linux64_heap_sandbox_dbg_ng,v8_linux_arm64_sim_heap_sandbox_dbg_ng Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3366237Reviewed-by:
-
Benedikt Meurer authored
This change fixes the implementation of the previously introduced API `Runtime.setMaxCallStackSizeToCapture` to work correctly and also apply (consistently) to stack traces captured by V8 when exceptions are thrown. It does so in a fully backwards compatible manner. This change thus makes the previous fix for catapult (which landed in http://crrev.com/c/3347789) effective, and therefore ensures that real world performance benchmarks aren't affected by the use of the `Runtime` domain in the catapult test framework. Note this is basically a reland of crrev.com/c/3361839, but without touching the stack traces for console messages (which led to the regressions in crbug/1283516, crbug/1283523, etc.). Fixed: chromium:1280831 Bug: chromium:1283162, chromium:1278650, chromium:1258599 Bug: chromium:1280803, chromium:1280832, chromium:1280818 Doc: https://bit.ly/v8-cheaper-inspector-stack-traces Change-Id: I3dcec7b75d76ca267fac8bd6fcb2cda60d5e60dd Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3364086Reviewed-by:
-
Piotr Sikora authored
Signed-off-by:
-
Benedikt Meurer authored
This sprinkles some more trace events in the disabled by default "v8.inspector" category, to help with understanding performance impact of stack trace capturing better. Bug: chromium:1283162 Change-Id: I6085d587f241635fbb6934bef3adc95f58c5d2aa Doc: https://bit.ly/v8-cheaper-inspector-stack-traces Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3364085Reviewed-by:
-
Manos Koukoutos authored
We unify the implementation of element segment expression entries with other initializer expressions: we represent them with a {WireBytesRef} and decode them with {InitExprInterface}. Except for reducing code duplication, this also fixes a bug where {global.get} entries in element segments could reference invalid globals. Changes: - Change {WasmElemSegment::Entry} to a union of a {WireBytesRef} initializer expression and a {uint32_t} function index. - In module-decoder, change parsing of expression entries to use {consume_init_expr}. Add type checking to {consume_element_func_index}, to complement type checking happening in {consume_init_expr}. - In module-instantiate.cc: - Move instantiation of indirect tables before loading of element segments. This way, when we call {UpdateDispatchTables} in {SetTableEntry}, the indirect table for the current table will also be updated. - Consolidate table entry instantiation into {SetTableEntry}, which handles lazily instantiated functions, or dispatches to {WasmTableObject::Set}. - Rename {InitializeIndirectFunctionTables} to {InitializeNonDefaultableTables}. - Change {InitializeNonDefaultableTables} and {LoadElemSegmentImpl} to use {EvaluateInitExpression}. - Add a test to exclude mutable/non-imported globals from the element section. - Update tests as needed. - Update .js module emission in wasm-fuzzer-common. Change-Id: I29c541bbca8531e8d0312ed95869c8e78a5a0c57 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3364082Reviewed-by: -
Omer Katz authored
Chromium builds indicate that moving an optional doesn't reset the source, and the source still indicates it has a value. That may be a bug in base::optional, but we should fix it here first to resolve current crashes. Bug: chromium:1154636 Cq-Include-Trybots: luci.v8.try:v8_linux_blink_rel Change-Id: Ibfb53b6d06d5f0310e68b200cc27ca318a5a57e5 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3366235Reviewed-by:
-
Clemens Backes authored
The virtual register should be prefixed with a 'v' to match the printing of virtual registers in other places. R=mslekova@chromium.org Bug: v8:12330 Change-Id: Ib79ace97b1c497efa3de85e1e48f5b07bb76d6cb Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3358293Reviewed-by:
-
Clemens Backes authored
The mid-tier register allocator already did some consistency checks; this CL extends them, and removes a redundant check. The added check ensures that no two virtual registers are assigned to the same register. A separate check for the correctness of the {allocated_registers_bits_} bitset is folded into {CheckConsistency}. A second check that an allocated register is contained in {allocated_registers_bits_} is removed. R=mslekova@chromium.org Bug: v8:12330 Change-Id: I6420eede145f88006c49e6ab16fdbeabffb8c9c7 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3358291Reviewed-by: -
Clemens Backes authored
This fixes an unbalanced return stack that was caused by popping the return address and jumping to it, instead of pushing it back and returning properly. R=leszeks@chromium.org Bug: v8:11246 Change-Id: I5c58c587cc0f5433c0a3595f5ed4c765e90d1a30 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3365267Reviewed-by:
-
Maya Lekova authored
Change-Id: I6893f37b7b56759341a1d43c21fa52b3836fde27 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3353368 Auto-Submit: Maya Lekova <mslekova@chromium.org> Reviewed-by:
-
Manos Koukoutos authored
See related CL for context. Changes: - In InitExprInterface, add the ability to evaluate function references as index only. Remove the global buffers and use the ones passed with the instance object instead. - In WasmElemSegment, add a field indicating if elements should be parsed as expressions or indices. Change module-decoder.cc to reflect this change. - In module-instantiate, change the signatures of LoadElemSegment, LoadElemSegmentImpl, and EvaluateInitExpr. Move the latter out of InstanceBuilder. Change-Id: I1df54393b2005fba49380654bdd40429bd4869dd Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3364081Reviewed-by:
-
Clemens Backes authored
For low-cost exception handling, it's important to be able to quickly drop frames from the stack until reaching the exception handler. The Intel shadow stack offers an instruction to avoid offending stack discipline, incsspq, which drops N values from the stack. This CL integrates that instruction for v8 exception handling. Bug: v8:11246 Change-Id: I908f0ab8bb3de6c36e6078e27b65132287328f2d Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3289637Reviewed-by:
Jakob Gruber <jgruber@chromium.org> Reviewed-by:
-
v8-ci-autoroll-builder authored
Rolling v8/build: https://chromium.googlesource.com/chromium/src/build/+log/555c8b4..44c14db Rolling v8/buildtools/third_party/libc++abi/trunk: https://chromium.googlesource.com/external/github.com/llvm/llvm-project/libcxxabi/+log/2715a6c..c884e7a Rolling v8/buildtools/third_party/libunwind/trunk: https://chromium.googlesource.com/external/github.com/llvm/llvm-project/libunwind/+log/4ead610..6a10e3e R=v8-waterfall-sheriff@grotations.appspotmail.com,mtv-sf-v8-sheriff@grotations.appspotmail.com Change-Id: I8cc3640b698cba2d84b0e1c11d97ec1eedbb743e Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3364392 Commit-Queue: v8-ci-autoroll-builder <v8-ci-autoroll-builder@chops-service-accounts.iam.gserviceaccount.com> Bot-Commit: v8-ci-autoroll-builder <v8-ci-autoroll-builder@chops-service-accounts.iam.gserviceaccount.com> Cr-Commit-Position: refs/heads/main@{#78468}
-
- 03 Jan, 2022 5 commits
-
-
Anton Bershanskiy authored
Commit 84f3877c moved IsInRange to base::IsInRange and updated src/parsing/keywords-gen.h, but did not update tools/gen-keywords-gen-h.py. Bug: v8:12507 Change-Id: I914ba73feac3bac6fd5d08d14d17149faf6c5c76 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3356200Reviewed-by:
-
Benedikt Meurer authored
This changes the StackFrameInfo to either hold on to a pair of (Script,source position) or a pair of (SharedFunctioInfo,bytecode offset) similar to what we do for MessageLocation. The idea here is to defer the costly bytecode offset to source position lookup until really needed, and in particular, avoid the costly lookup during stack trace capturing. On the `standalone.js` benchmark in crbug.com/1283162#c1, this reduces overall average execution time by roughly 25%, and the performance is almost back to where it was before crrev.com/c/3302794 (being only 12% slower than before on the `standalone.js` test case). Note that due to unrelated limitations we cannot encode -1 as bytecode offset in the flags field of the StackFrameInfo, and so we treat this case specially (happens when stack trace capturing is triggered in the function entry sequence) and just eagerly resolve it to the source position. Bug: chromium:1278650, chromium:1283162, chromium:1280803 Bug: chromium:1280818, chromium:1280831, chromium:1280832 Doc: https://bit.ly/v8-cheaper-inspector-stack-traces Change-Id: If7cf62fce48d32c0f188895d1f8c9eee51b9e70d Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3359633Reviewed-by:
-
Samuel Groß authored
This is in line with PartitionAlloc's DiscardSystemPagesInternal. When the sandbox is enabled, OS::DiscardSystemPages is used instead of PA's version. As such, these two implementations should ideally be mostly identical. Using MADV_FREE instead of MADV_DONTNEED as was previously done appears to cause some memory regressions. Bug: chromium:1276887 Change-Id: Ied92b106e9894d428e599801d753ab4c8cffd874 Cq-Include-Trybots: luci.v8.try:v8_linux64_heap_sandbox_dbg_ng,v8_linux_arm64_sim_heap_sandbox_dbg_ng Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3364090Reviewed-by:
-
Omer Katz authored
Clear cached events if there is no embedder recorder. Bug: chromium:1154636 Change-Id: I9ad3b752ea242d07b417ce3022936789c47afc6a Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3358292Reviewed-by:
-
Manos Koukoutos authored
Int64Lowering may produce projection nodes with floating control input. When inlining, we need to connect such nodes to the caller's start node instead of the control dependency of the call node. Bug: v8:12506, v8:12166 Change-Id: I1a726dc7b0ad40e98f3b745298062c2f7194288a Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3352221Reviewed-by:
-
