- 02 Nov, 2020 22 commits
-
-
Victor Gomes authored
In case of no arguments adaptor frame, we massage the arguments in InvokePrologue pushing undefined objects if the actual argument count is below the parameter count. This CL adds a stack overflow check before pushing these undefined objects to the stack. Change-Id: I2a88bf6fdfd17958f6f6884143a67d50ea842fd2 Bug: v8:10201 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2491039Reviewed-by:
Igor Sheludko <ishell@chromium.org> Reviewed-by:
Georg Neis <neis@chromium.org> Commit-Queue: Victor Gomes <victorgomes@chromium.org> Cr-Commit-Position: refs/heads/master@{#70927}
-
Victor Gomes authored
Change-Id: I7df25ca2c7caabed429cfdc0b4aab0aeb5e03fcd Bug: v8:10201 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2463222Reviewed-by:
-
Milad Fa authored
Port 808a5d76 Original Commit Message: InterpreterEntryTrampoline checks 1. If there is any optimization marker, and process the marker by calling the correct runtime function 2. If there is any optimized code, it installs optimized code and starts executing optimized code. Earlier it was OK to do these two checks in any order, since it wasn't possible that we have both a marker and optimized code. Turboprop code calls out to IET to process the optimization marker / install next tier optimized code. So we have to check for optimization marker before we check for optimized code. R=mythria@chromium.org, joransiu@ca.ibm.com, junyan@redhat.com, midawson@redhat.com BUG= LOG=N Change-Id: I37f470809e244f675862e109ae609170a2b019f0 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2513792Reviewed-by:
Junliang Yan <junyan@redhat.com> Commit-Queue: Milad Fa <mfarazma@redhat.com> Cr-Commit-Position: refs/heads/master@{#70925}
-
Vasili Skurydzin authored
Change-Id: I14941fcc34773791a86c6fb250237279641fd690 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2510070Reviewed-by:
Milad Fa <mfarazma@redhat.com> Reviewed-by:
Michael Dawson <midawson@redhat.com> Reviewed-by:
Joran Siu <joransiu@ca.ibm.com> Reviewed-by:
Ross McIlroy <rmcilroy@chromium.org> Commit-Queue: Milad Fa <mfarazma@redhat.com> Cr-Commit-Position: refs/heads/master@{#70924}
-
Andreas Haas authored
With the recent switch from the tasks API to the jobs API for thread management within WebAssembly compilation we got the problem that TurboFan compilation of a previous compilation can block the Liftoff compilation of a new compilation. With synchronous compilation, this can cause significant delays for the user. With this CL we increase the priority of synchronous compilation so that at least synchronous compilation can only be blocked by other synchronous compilation. This fixes issues that we saw on autocad. Why is it okay to increase the priority of synchronous compilation? * On the main thread, synchronous compilation is only allowed for small modules which should not take long to compile. Also, the compilation blocks the main thread and should finish as fast as possible. * On worker threads, delaying other background work is not such a big issue. Downsides: * This does not only increase the priority of the initial compilation, but also for the TurboFan optimization. Similar to above, for small modules on the main thread this should not be a big deal because compilation is fast. Big modules can only be compiled on worker threads. Note that this is supposed to be a fix for the problems we see at the moment with autocad on stable and beta. Eventually compilation job management should be done by the wasm engine for all WebAssembly compilation, so that not each WebAssemly module has to do its own compilation job management. R=clemensb@chromium.org Bug: chromium:1142686, v8:11088 Change-Id: Iee4948b2fcad944f587918e9452e6888258857f9 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2512911 Commit-Queue: Andreas Haas <ahaas@chromium.org> Reviewed-by:
Clemens Backes <clemensb@chromium.org> Cr-Commit-Position: refs/heads/master@{#70923}
-
Clemens Backes authored
This adds a first simple version of the inspector fuzzer, which is a stripped-down version of the inspector-test executable. The fuzzer generates inputs which are compatible with inspector-test. There are still memory leaks, and the fuzzer will probably run into timeouts most of the time. Both of this will be addressed in follow-ups. R=szuend@chromium.org, machenbach@chromium.org Bug: chromium:1142437 Change-Id: I4d13da460f571d791a3642b0705a1f07b442c11b Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2505722 Commit-Queue: Clemens Backes <clemensb@chromium.org> Reviewed-by:
Simon Zünd <szuend@chromium.org> Reviewed-by:
Michael Achenbach <machenbach@chromium.org> Cr-Commit-Position: refs/heads/master@{#70922}
-
Tobias Tebbi authored
Drive-by fixes: - Use constexpr types to determine C++ type names. - Fix factory constructors to not skip write barriers in old generation. Change-Id: I0ebbfd56c06ad41d02836fb48531ae7eded166bf Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2400994Reviewed-by:
Ulan Degenbaev <ulan@chromium.org> Reviewed-by:
Nico Hartmann <nicohartmann@chromium.org> Commit-Queue: Tobias Tebbi <tebbi@chromium.org> Cr-Commit-Position: refs/heads/master@{#70921}
-
Michael Achenbach authored
Change-Id: Ic72d37bd7707633569a01a6639a91f2aef6f03e8 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2512915Reviewed-by:
-
Camillo Bruni authored
- Move all createElement helpers onto a separate DOM class - Make ICPanel.update async - Show number of selected IC events in the ICPanel header - Use shared bound functions for event listeners in the ICPanel groups - Use triangle to mark opened and closed ICPanel groups - Use global --border-color CSS variable Bug: v8:10644 Change-Id: Ib35d94db1019d5cdcee057f0f047472f478ab3be Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2507718Reviewed-by:
Sathya Gunasekaran <gsathya@chromium.org> Commit-Queue: Camillo Bruni <cbruni@chromium.org> Cr-Commit-Position: refs/heads/master@{#70919}
-
Frank Emrich authored
The comment at the beginning of ordered-hash-table.h desribing the layout of ordered hash tables seems to have become outdated. 1. There was a "kPrefixSize + " missing in the description of the left bound of the first non-fixed-sized part in normal and obsoleted tables. The following only affects obsoleted tables: 2. The bucket count doesn't reside at [kPrefixSize + 0], instead it is unchanged w.r.t normal tables, and resides at [kPrefixSize + 2] 3. The next newer table takes place of the element count, residing at [kPrefixSize + 0]. This can be seen by NextTableIndex() being defined as NumberOfElementsIndex() 4. The number of removed holes is never written anywhere when obsoleting a table. Instead, the number of deleted elements count is used for this. This can be see by DCHECK_EQ(table->NumberOfDeletedElements(), removed_holes_index); in method Rehash 5. kClearedTableSentinel is written to the number of deleted element slot when clearing a table, as seen by table->SetNumberOfDeletedElements(kClearedTableSentinel); in method Clear Bug: v8:7569, v8:11074 Change-Id: I0e6750fc57c90b6c09fc41f001cd555f7c455fd3 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2505763 Commit-Queue: Frank Emrich <emrich@google.com> Reviewed-by:
-
Zhi An Ng authored
Have a new macro-assembler function Pshufb that does not require dst and src to be equal, since the vpshufb can have 3 operands. On SEE, if dst and src are not equal, emit a movapd(dst, src). This saves a move on AVX. Small cleanup to use kScratchDoubleReg2 instead of requesting for a tmp (x64). Bug: v8:9561 Change-Id: I131ad0456b272da857350762582cac1fb240ae40 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2513868 Commit-Queue: Zhi An Ng <zhin@chromium.org> Reviewed-by:
-
Ulan Degenbaev authored
On newer compilers the {operator delete} with explicit {size_t} argument would be instantiated for {v8::BackingStore} and used in the destructor of {std::unique_ptr<v8::BackingStore>}. The {size_t} argument is wrong though, since the pointer actually points to a {v8::internal::BackingStore} object. The solution is to explicitly provide a {operator delete}, preventing an implicitly generated {size_t} operator. Bug:v8:11081 Change-Id: Iee0aa47a67f0e41000bea628942f7e3d70198b83 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2506712 Commit-Queue: Ulan Degenbaev <ulan@chromium.org> Reviewed-by:Camillo Bruni <cbruni@chromium.org> Cr-Commit-Position: refs/heads/master@{#70916}
-
Mythri A authored
This cl adds support for tiering up in TurboProp. This cl makes necessary changes to support tier up but doesn't tier up yet. More specifically this cl: 1. Introduces a new flag for interrupt_budget_for_midtier and updates code to use the correct interrupt_budget. 2. Introduces a flag turboprop_as_midtier and necessary support to tier up. When this flag is enabled, we introduce checks for tierup and updating interrupt budget. Bug: v8:9684 Change-Id: I58785ce4b9de46488a22d3b4d0cebedac460a773 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2460822 Commit-Queue: Mythri Alle <mythria@chromium.org> Reviewed-by:
Jakob Gruber <jgruber@chromium.org> Reviewed-by:
-
Mythri A authored
InterpreterEntryTrampoline checks 1. If there is any optimization marker, and process the marker by calling the correct runtime function 2. If there is any optimized code, it installs optimized code and starts executing optimized code. Earlier it was OK to do these two checks in any order, since it wasn't possible that we have both a marker and optimized code. Turboprop code calls out to IET to process the optimization marker / install next tier optimized code. So we have to check for optimization marker before we check for optimized code. Bug: v8:9684 Change-Id: Iaefb51aec9c2de0bcbacbdbd8a5dff513e573036 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2505721Reviewed-by:
-
Hans Wennborg authored
Use pointer arithmetic to go back to the first nodes_ element instead of computing the address manually. Bug: none Change-Id: I887c41908dc6891ed15aa03698ca2305924dcab2 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2512904Reviewed-by:
Michael Lippautz <mlippautz@chromium.org> Commit-Queue: Hans Wennborg <hans@chromium.org> Cr-Commit-Position: refs/heads/master@{#70913}
-
Frank Emrich authored
This turns the member functions CollectKeysTo and CopyEnumKeysTo of BaseNameDictionary into helper function in keys.cc; they are only used there. Further, CollectKeysTo is renamed to CollectKeysFromDictionary. EnumIndexComparator is moved from keys.cc to dictionary.h. All moves are motivated by https://chromium-review.googlesource.com/c/v8/v8/+/2489692 needing these function in other places. Bug: v8:7569 Change-Id: Ia8039e98fd00cef45dec376f3c401635b2321761 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2509597Reviewed-by:
Marja Hölttä <marja@chromium.org> Reviewed-by:
-
Zhi An Ng authored
Extract Shufps to handle both AVX and SSE cases, in the SSE case it will copy src to dst if they are not the same. This allows us to use it in Liftoff as well, without the extra copy when AVX is supported. In other places, the usage of Shufps is unnecessary, since they are within a clause checking for non-AVX support, so we can simply use the shufps (non-macro-assembler). Bug: v8:9561 Change-Id: Icb043d7a43397c1b0810ece2666be567f0f5986c Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2513866Reviewed-by:
-
Zhi An Ng authored
These instructions require SSE4_1, we need to enable SSE4_1 before calling them. These will only fail when there is no AVX and no SSE4_2, and we unfortunately is not a configuration we test (we test for no-AVX, and no-AVX && no-SSE4_1 (which also means no-SSE4_2) && no-SSSE3 and no-SSE3). Change-Id: I3d49b9c68385c4b5b631c4a82e5e46d388500c06 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2513869Reviewed-by:
-
Clemens Backes authored
This also extracts the {FrontendChannelImpl}, the {SetTimeoutTask}, and the {SetTimeoutExtension} for reuse by the inspector fuzzer. R=szuend@chromium.org Bug: chromium:1142437 Change-Id: I75e49d6fbb4b801ace1ffc9cac963d6b83c52f36 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2505717Reviewed-by: -
v8-ci-autoroll-builder authored
Rolling v8/build: https://chromium.googlesource.com/chromium/src/build/+log/0dfb09a..eb675d7 TBR=machenbach@chromium.org,tmrts@chromium.org,v8-waterfall-sheriff@grotations.appspotmail.com Change-Id: Ic43d9ba2687399b41e04672356163ed075ec6f4d Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2512779Reviewed-by:
v8-ci-autoroll-builder <v8-ci-autoroll-builder@chops-service-accounts.iam.gserviceaccount.com> Commit-Queue: v8-ci-autoroll-builder <v8-ci-autoroll-builder@chops-service-accounts.iam.gserviceaccount.com> Cr-Commit-Position: refs/heads/master@{#70908}
-
Zhi An Ng authored
Bug: v8:11038 Change-Id: Ifb50d8667b03afa045a957ffc8f3df799252e561 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2500939Reviewed-by:
Jakob Kummerow <jkummerow@chromium.org> Commit-Queue: Zhi An Ng <zhin@chromium.org> Cr-Commit-Position: refs/heads/master@{#70907}
-
Zhi An Ng authored
Similar to the changes in the disassembly, break up simulator changes into smaller functions. This change only copy pastes implementation into various helper function, and does not follow the decoding rules in the manual yet, it is still using SpecialValue. Subsequent changes will clean this up. Bug: v8:11074 Change-Id: Ic2905406e905fc3df01825047763d083aa0fbff4 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2505248Reviewed-by:
-
- 01 Nov, 2020 1 commit
-
-
v8-ci-autoroll-builder authored
Rolling v8/build: https://chromium.googlesource.com/chromium/src/build/+log/079c81c..0dfb09a Rolling v8/third_party/catapult: https://chromium.googlesource.com/catapult/+log/7030291..dcd4fb3 TBR=machenbach@chromium.org,tmrts@chromium.org,v8-waterfall-sheriff@grotations.appspotmail.com Change-Id: I4dbfc06c044626d24160d42d5adf08e9ae968960 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2512546Reviewed-by:
-
- 31 Oct, 2020 1 commit
-
-
v8-ci-autoroll-builder authored
Rolling v8/build: https://chromium.googlesource.com/chromium/src/build/+log/b8645e7..079c81c Rolling v8/third_party/catapult: https://chromium.googlesource.com/catapult/+log/24bd418..7030291 Rolling v8/third_party/depot_tools: https://chromium.googlesource.com/chromium/tools/depot_tools/+log/e8dff37..9396c2b Rolling v8/tools/clang: https://chromium.googlesource.com/chromium/src/tools/clang/+log/12b3cd3..e5230a2 Rolling v8/tools/luci-go: git_revision:576741d3eed0fa33971fb34cd823650e6f5b47fb..git_revision:1a022d3a4c50be4207ee93451255d71896416596 Rolling v8/tools/luci-go: git_revision:576741d3eed0fa33971fb34cd823650e6f5b47fb..git_revision:1a022d3a4c50be4207ee93451255d71896416596 Rolling v8/tools/luci-go: git_revision:576741d3eed0fa33971fb34cd823650e6f5b47fb..git_revision:1a022d3a4c50be4207ee93451255d71896416596 TBR=machenbach@chromium.org,tmrts@chromium.org,v8-waterfall-sheriff@grotations.appspotmail.com Change-Id: I52b4a621b1b1b74d3f4e7071cf438154f30ef5bd Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2512544Reviewed-by:
-
- 30 Oct, 2020 14 commits
-
-
Camillo Bruni authored
- Introduce LazyTable that only shows the first 100 entries and can lazily expand - Make StatsPanel.update method async to not block the UI - Show "Total" entry as last item in StatsPanel tables - Fix StatsPanel table entries text alignment - Remove unused getters from StatsPanel class Drive-by-fix: - MapPanel._displayedMapsInTree is undefined when expanding to a sub-transition tree Bug: v8:10644 Change-Id: I5ce7c8b1ee825515cf790a3e52534c3069d8be89 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2507716 Commit-Queue: Camillo Bruni <cbruni@chromium.org> Reviewed-by:
Shu-yu Guo <syg@chromium.org> Cr-Commit-Position: refs/heads/master@{#70903}
-
Milad Fa authored
Change-Id: Icd452df211c85d83c82812124d14fe6a31e733b7 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2509553Reviewed-by:
-
Etienne Pierre-doray authored
IsCompleted is backwards. For a more consistent api, the function is renamed IsActive and logic is flipped. Following up on https://chromium-review.googlesource.com/c/v8/v8/+/2461840 The intend is to make the distinction between IsActive and IsValid obvious. Change-Id: Iaf00b9f6ffa8f1efe93ae29f09899737ef20f04d Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2510969Reviewed-by:
-
Georg Neis authored
After c90ff8bd this is purely a cosmetic change. Change-Id: If804f1b7558ab2f8096e576570f68b848df7423c Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2509592Reviewed-by:
Zhi An Ng <zhin@chromium.org> Commit-Queue: Zhi An Ng <zhin@chromium.org> Auto-Submit: Georg Neis <neis@chromium.org> Cr-Commit-Position: refs/heads/master@{#70900}
-
Michael Lippautz authored
- Provide GetRealStackAddressForSlot that deals with ASAN fake stacks properly, also accounting for the fact that ASAN gets its real stack address in a nested call. - Fix cppgc on-stack getter. - Reuse platform routines in global handles. Bug: chromium:1139914, chromium:1056170 Change-Id: If11a40d543b33edcea220bb70f170ac018e15053 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2509594 Commit-Queue: Michael Lippautz <mlippautz@chromium.org> Reviewed-by:
-
Clemens Backes authored
The idea was to distribute the publish limits to reduce the chance that different tasks are publishing at the same time. We always used the same task_id to compute the limits though, so all queues had the same limit. This CL fixes that by using the task id of the respective queue. R=thibaudm@chromium.org Change-Id: I5231aa35b2b53d625fafbb2130b886390fb7d38a Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2502292Reviewed-by:
Thibaud Michaud <thibaudm@chromium.org> Commit-Queue: Clemens Backes <clemensb@chromium.org> Cr-Commit-Position: refs/heads/master@{#70898}
-
Michael Lippautz authored
We previously assumed that a fake stack should be mapped back to a real stack based on fake-stack offsets. This is not correct: Fake and real stack are disjoint and both contain the corresponding slot values. For global handles this means that on-stack handles must be registered using their real stack frame base to be able to purge them occasionally based on the current stack address. When dealing with a slot though, the GC can just dereference the slot for a value, indeppendent of whether the slot is in a fake or real frame. Drive-by: Fix tests that do not want stack handles by creating handles on heap. Change-Id: I2c86c8e047bd0d48c24c2642b2b4dba284a93909 Bug: chromium:1139914 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2507720 Commit-Queue: Michael Lippautz <mlippautz@chromium.org> Reviewed-by:
-
Andreas Haas authored
This function is equivalent to CheckHasMemory by now. R=clemensb@chromium.org Bug: v8:11074 Change-Id: I05fbd160d6e5b8d7433133052a8390521564632b Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2509590Reviewed-by:
-
Almothana Athamneh authored
Relanding after failing test on ia32 was fixed. Bug: v8:11072 Change-Id: I939ed92f8fb7ea84314a79eb62f0d4aed272bea3 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2509589Reviewed-by:
Liviu Rau <liviurau@chromium.org> Commit-Queue: Almothana Athamneh <almuthanna@chromium.org> Cr-Commit-Position: refs/heads/master@{#70895}
-
Benedikt Meurer authored
Building these objects takes a lot of time and memory for realistic applications and exposing them via the Scope view in DevTools isn't practical either. We have a replacement in the Console now, and if this needs more exposure we can think about other, more scalable ways with better UX. Fixed: v8:10986 Bug: chromium:1141781 Change-Id: I6177d63a987749889a9880cf0738031191eb5705 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2507696 Commit-Queue: Clemens Backes <clemensb@chromium.org> Auto-Submit: Benedikt Meurer <bmeurer@chromium.org> Reviewed-by:
-
Zhi An Ng authored
These operations can be moved into an existing macro list, since they are simple operations that generate only 1 instruction. The benefit is that they have support for AVX 3-operand instruction, and does not have to force dst to be equals to src. Bug: v8:9561 Change-Id: I9ec1d2496d14cb9f0fb3b4854ca39887eb5bf49b Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2505240Reviewed-by:
-
Martin Bidlingmaier authored
This commit adds the 'l' (linear) RegExp flag (as in e.g. /asdf|123/l) that forces execution in linear time. These regexps are handled by the experimental engine. If the experimental engine cannot handle the pattern, an exception is thrown on creation of the regexp. The commit also adds a new global V8 flag and changes an existing one: * --enable-experimental-engine, which turns on recognition of the RegExp 'l' flag. Previously this flag also caused all supported regexps to be executed by the experimental engine; this is not the case anymore. * --default-to-experimental-regexp-engine takes over the previous semantics of --enable-experimental-regexp-engine: We execute all supported regexps with the experimental engine. Cq-Include-Trybots: luci.v8.try:v8_linux64_fyi_rel_ng Bug: v8:10765 Change-Id: I5622a89b19404105e8be280d454e9fdd63c003b3 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2461244Reviewed-by:
-
Leszek Swirski authored
GeneralizeField already calculates the field owner. Change-Id: I0730471479d4289d3d027ea63178ddae659ee2b2 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2507717 Commit-Queue: Leszek Swirski <leszeks@chromium.org> Reviewed-by:
-
v8-ci-autoroll-builder authored
Rolling v8/build: https://chromium.googlesource.com/chromium/src/build/+log/acfd692..b8645e7 Rolling v8/third_party/aemu-linux-x64: zy6Zme4YpIqml3lhXc3o7a0r_gVd0_gcggle2n2j330C..uqYfh08DTmvjUgXl-HxNY1ptIb0KpyuY0I9MFaci9-AC Rolling v8/third_party/android_sdk/public: ijpIFSitwBfaEdO9VXBGPqDHUVzPimXy_whw3aHTN9oC..V__2Ycej-H2-6AcXX5A3gi7sIk74SuN44PBm2uC_N1sC Rolling v8/third_party/catapult: https://chromium.googlesource.com/catapult/+log/15be7eb..24bd418 Rolling v8/tools/clang: https://chromium.googlesource.com/chromium/src/tools/clang/+log/f9b4a22..12b3cd3 TBR=machenbach@chromium.org,tmrts@chromium.org,v8-waterfall-sheriff@grotations.appspotmail.com Change-Id: I8250aad9d82eb7ec2f5da2aa60b2dee040df6612 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2507341Reviewed-by:
-
- 29 Oct, 2020 2 commits
-
-
Zhi An Ng authored
The main fix here is on non-AVX builds, the shifts require dst == src. Calling macro-assembler functions like Psrlw assumes that we already met that requirement (which we usually specify in TurboFan's instruction-selector). On Liftoff, we need to do that check manually. This is done by using the helper EmitSimdShiftOpImm which will do this check, and dispatches to AVX if supported, or mov if dst != src. Also fix a couple of places where we forgot to annotate the required SSE extension for the instruction. Bug: v8:11078 Change-Id: Icc31df9ab80c041f4bb3bd48444658368eaa2c91 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2505251Reviewed-by:
-
Zhi An Ng authored
On AVX, many instructions can have 3 operands, unlike SSE which only has 2. So on SSE we use DefineSameAsFirst on the dst. But on AVX, using that will cause some unnecessary moves. This change moves a bunch of instructions that have single instruction codegen into a macro list which supports the this non-restricted AVX codegen. Bug: v8:9561 Change-Id: I348a8396e8a1129daf2e1ed08ae8526e1bc3a73b Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2505254Reviewed-by:
-
