1. 20 Oct, 2021 1 commit
  2. 06 Sep, 2021 1 commit
  3. 02 Mar, 2021 1 commit
  4. 06 Aug, 2020 1 commit
    • Leszek Swirski's avatar
      [runtime] Move string table off-heap · 1546be9c
      Leszek Swirski authored
      Changes the isolate's string table into an off-heap structure. This
      allows the string table to be resized without allocating on the V8 heap,
      and potentially triggering a GC. This allows existing strings to be
      inserted into the string table without requiring allocation.
      
      This has two important benefits:
      
        1) It allows the deserializer to insert strings directly into the
           string table, rather than having to defer string insertion until
           deserialization completes.
      
        2) It simplifies the concurrent string table lookup to allow resizing
           the table inside the write lock, therefore eliminating the race
           where two concurrent lookups could both resize the table.
      
      The off-heap string table has the following properties:
      
        1) The general hashmap behaviour matches the HashTable, i.e. open
           addressing, power-of-two sized, quadratic probing. This could, of
           course, now be changed.
      
        2) The empty and deleted sentinels are changed to Smi 0 and 1,
           respectively, to make those comparisons a bit cheaper and not
           require roots access.
      
        3) When the HashTable is resized, the old elements array is kept
           alive in a linked list of previous arrays, so that concurrent
           lookups don't lose the data they're accessing. This linked list
           is cleared by the GC, as then we know that all threads are in
           a safepoint.
      
        4) The GC treats the hash table entries as weak roots, and only walks
           them for non-live reference clearing and for evacuation.
      
        5) Since there is no longer a FixedArray to serialize for the startup
           snapshot, there is now a custom serialization of the string table,
           and the string table root is considered unserializable during weak
           root iteration. As a bonus, the custom serialization is more
           efficient, as it skips non-string entries.
      
      As a drive-by, rename LookupStringExists_NoAllocate to
      TryStringToIndexOrLookupExisting, to make it clearer that it returns
      a non-string for the case when the string is an array index. As another
      drive-by, extract StringSet into a separate header.
      
      Bug: v8:10729
      Change-Id: I9c990fb2d74d1fe222920408670974a70e969bca
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2339104
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Reviewed-by: 's avatarJakob Gruber <jgruber@chromium.org>
      Reviewed-by: 's avatarUlan Degenbaev <ulan@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#69270}
      1546be9c
  5. 06 Feb, 2020 1 commit
  6. 04 Feb, 2020 1 commit
    • Leszek Swirski's avatar
      [offthread] Add an OffThreadIsolate · 7a20b6b9
      Leszek Swirski authored
      The Factory/OffThreadFactory allows us to cleanly separate object
      construction behaviour between main-thread and off-thread in a
      syntactically consistent way (so that methods templated on the factory
      type can be made to work on both).
      
      However, there are cases where we also have to access the Isolate, for
      handle creation or exception throwing. So far we have been pushing more
      and more "customization points" into the factories to allow these
      factory-templated methods to dispatch on this isolate behaviour via
      these factory methods. Unfortunately, this is an increasing layering
      violation between Factory and Isolate, particularly around exception
      handling.
      
      Now, we introduce an OffThreadIsolate, analogous to Isolate in the same
      way as OffThreadFactory is analogous to Factory. All methods which were
      templated on Factory are now templated on Isolate, and methods which
      used to take an Isolate, and which were recently changed to take a
      templated Factory, are changed/reverted to take a templated Isolate.
      OffThreadFactory gets an isolate() method to match Factory's.
      
      Notably, FactoryHandle is changed to "HandleFor", where the template
      argument can be either of the Isolate type or the Factory type (allowing
      us to dispatch on both depending on what is available).
      
      Bug: chromium:1011762
      Change-Id: Id144176f7da534dd76f3d535ab2ade008b6845e3
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2030909
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Reviewed-by: 's avatarToon Verwaest <verwaest@chromium.org>
      Reviewed-by: 's avatarUlan Degenbaev <ulan@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#66101}
      7a20b6b9
  7. 15 Jan, 2020 1 commit
    • Leszek Swirski's avatar
      [offthread] Add OffThreadFactory · e659917a
      Leszek Swirski authored
      Introduce OffThreadFactory with initial string construction support.
      
      The OffThreadFactory shares with Factory a new CRTP base class, called
      FactoryBase. Methods in FactoryBase return a FactoryHandle<Factory, T>
      alias, which is Handle<T> for normal Factory and a new OffThreadHandle<T>
      for OffThreadFactory. OffThreadHandle<T> behaves like Handle<T>, except
      it stores the object in-line rather than needing external storage.
      
      Any shared factory methods are moved into FactoryBase, which uses CRTP
      to call the sub-class's AllocateRaw method (plus a few more customization
      points which need Isolate access on the main thread).
      
      Methods that used to take an Isolate or Factory, and are needed off the
      main thread, are now expected to be templated on the factory type and
      to use the appropriate handle.
      
      Once an OffThreadFactory has finished being used (e.g. off-thread
      compilation completed) its pages are "Published" into the main-thread
      Heap. To deal with string internalization without creating a bunch of
      ThinStrings, this is done in two stages:
      
        1. 'FinishOffThread': The off-thread pages are walked to
           collect all slots pointing to "internalized" strings. After this is
           called it is invalid to allocate any more objects with the factory.
        2. 'Publish': On the main thread, we transform these slots into
           <Handle to holder, offset> pairs, then for each saved slot
           re-internalize its string and update the slot to point to the
           internalized string.
      
      Bug: chromium:1011762
      Change-Id: I008a694da3c357de34362bd86fe7e1f46b535d5e
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1992434
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Reviewed-by: 's avatarUlan Degenbaev <ulan@chromium.org>
      Reviewed-by: 's avatarToon Verwaest <verwaest@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#65787}
      e659917a
  8. 20 Nov, 2019 1 commit
  9. 16 Sep, 2019 1 commit
  10. 12 Sep, 2019 1 commit
  11. 20 Aug, 2019 1 commit
  12. 12 Jun, 2019 1 commit
    • Seth Brenith's avatar
      [torque] Stricter object field verification, part 1 · 3834c637
      Seth Brenith authored
      This change adjusts object initialization order for a few classes so
      that the GC can never see those objects in an invalid, partially-
      initialized state.
      
      AccessorInfo: Just zeros out a few fields upon construction. This is the
      simplest case.
      
      FunctionTemplateInfo: Slightly changes the order in which fields are
      set, so that the Smi field is set ahead of the call to SetCallHandler,
      which can GC. Also a pretty simple case.
      
      JSListFormat, JSPluralRules, JSRelativeTimeFormat, JSSegmenter: The spec
      requires that we start with OrdinaryCreateFromConstructor, which has
      observable side effects (it fetches the prototype from the new.target).
      So we split JSObject::New in half: the first half does all of the user-
      visible things and returns a Map, which we can pass to the second half
      when we're ready to actually allocate the object.
      
      JSTypedArray: Extends the pattern from JSListFormat into Torque code:
      start with a Map and don't allocate the object until we're ready to set
      all of its properties.
      
      Bug: v8:9311
      Change-Id: Id7703e8a0727ec756c774cfbb56af787658a111a
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1646844
      Commit-Queue: Seth Brenith <seth.brenith@microsoft.com>
      Reviewed-by: 's avatarJakob Gruber <jgruber@chromium.org>
      Reviewed-by: 's avatarSigurd Schneider <sigurds@chromium.org>
      Reviewed-by: 's avatarUlan Degenbaev <ulan@chromium.org>
      Reviewed-by: 's avatarBenedikt Meurer <bmeurer@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#62123}
      3834c637
  13. 23 May, 2019 2 commits
  14. 22 May, 2019 1 commit
  15. 21 May, 2019 1 commit
  16. 12 Mar, 2019 1 commit
  17. 15 Feb, 2019 1 commit
  18. 09 Feb, 2019 1 commit
  19. 26 Dec, 2018 1 commit
  20. 18 Dec, 2018 1 commit
  21. 17 Dec, 2018 1 commit
  22. 25 Nov, 2018 1 commit
  23. 24 Nov, 2018 1 commit
  24. 19 Oct, 2018 1 commit
  25. 11 Oct, 2018 1 commit
  26. 24 Sep, 2018 3 commits
  27. 21 Sep, 2018 2 commits
  28. 20 Sep, 2018 1 commit
  29. 04 Sep, 2018 1 commit
  30. 22 Aug, 2018 1 commit
    • Camillo Bruni's avatar
      Reland "[runtime] Improve for-in performance" · 827e6d12
      Camillo Bruni authored
      This is a reland of 8fa7f9ed
      
      Original change's description:
      > [runtime] Improve for-in performance
      > 
      > - Add fast-path String conversion for Smi (which is the most common case)
      >   This improves for-in by ~10% on non-initialized enum-caches
      > - Don't use the NumberStringCache for large indices to not overflow the cache
      >   during key collection. This improves worst-case performance by ~2.5x
      > - Drop number_to_string_native and number_to_string_runtime counters
      > 
      > Bug: v8:7717
      > Change-Id: Ic1ff385e3374e6a7e7e7bdb9ae75fb8c238105d1
      > Reviewed-on: https://chromium-review.googlesource.com/1167049
      > Reviewed-by: Toon Verwaest <verwaest@chromium.org>
      > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > Commit-Queue: Camillo Bruni <cbruni@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#55233}
      
      Bug: v8:7717
      Change-Id: Ie29041b20ca4a06b8e74a4378e7c1118373072ae
      Reviewed-on: https://chromium-review.googlesource.com/1183721
      Commit-Queue: Camillo Bruni <cbruni@chromium.org>
      Reviewed-by: 's avatarUlan Degenbaev <ulan@chromium.org>
      Reviewed-by: 's avatarToon Verwaest <verwaest@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#55316}
      827e6d12
  31. 21 Aug, 2018 1 commit
    • Aseem Garg's avatar
      Revert "[runtime] Improve for-in performance" · a4235f00
      Aseem Garg authored
      This reverts commit 8fa7f9ed.
      
      Reason for revert: Speculating that this breaks GC stress
      
      Original change's description:
      > [runtime] Improve for-in performance
      > 
      > - Add fast-path String conversion for Smi (which is the most common case)
      >   This improves for-in by ~10% on non-initialized enum-caches
      > - Don't use the NumberStringCache for large indices to not overflow the cache
      >   during key collection. This improves worst-case performance by ~2.5x
      > - Drop number_to_string_native and number_to_string_runtime counters
      > 
      > Bug: v8:7717
      > Change-Id: Ic1ff385e3374e6a7e7e7bdb9ae75fb8c238105d1
      > Reviewed-on: https://chromium-review.googlesource.com/1167049
      > Reviewed-by: Toon Verwaest <verwaest@chromium.org>
      > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > Commit-Queue: Camillo Bruni <cbruni@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#55233}
      
      TBR=ulan@chromium.org,cbruni@chromium.org,verwaest@chromium.org
      
      Change-Id: I8d0332478afcd7c6a3f8fbf1f044b9aa870b6b13
      No-Presubmit: true
      No-Tree-Checks: true
      No-Try: true
      Bug: v8:7717
      Reviewed-on: https://chromium-review.googlesource.com/1182676Reviewed-by: 's avatarAseem Garg <aseemgarg@chromium.org>
      Commit-Queue: Aseem Garg <aseemgarg@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#55241}
      a4235f00
  32. 20 Aug, 2018 2 commits
  33. 26 Jun, 2018 3 commits
    • Georg Neis's avatar
      Reland "Reland "Introduce MutableHeapNumber class."" · f1c79e02
      Georg Neis authored
      This is a reland of f0bcbc90.
      A few casts were still wrong.
      
      Original change's description:
      > Reland "Introduce MutableHeapNumber class."
      >
      > This is a reland of 40ac6b18, which
      > was incorrect due to a bad merge.
      >
      > Original change's description:
      > > Introduce MutableHeapNumber class.
      > >
      > > V8 knows heap numbers and mutable heap numbers. They have
      > > difference instance types, but in C++ code we've used the
      > > same class for both (HeapNumber). Confusingly, however,
      > > IsHeapNumber would return false for mutable heap numbers,
      > > while HeapNumber::cast would succeed.
      > >
      > > This CL adds a separate class MutableHeapNumber and
      > > eliminates the confusing behavior.
      > >
      [...]
      > TBR=bmeurer@chromium.org
      > TBR=ulan@chromium.org
      >
      > Change-Id: I3af1014c949821dfac0754a3e48c65ce1bad1ad1
      > Reviewed-on: https://chromium-review.googlesource.com/1114539
      > Reviewed-by: Georg Neis <neis@chromium.org>
      > Commit-Queue: Georg Neis <neis@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#54022}
      
      Change-Id: I19a33da4b6abcd445b528a84d4f56ba1964d337b
      Reviewed-on: https://chromium-review.googlesource.com/1114100
      Commit-Queue: Georg Neis <neis@chromium.org>
      Reviewed-by: 's avatarBenedikt Meurer <bmeurer@chromium.org>
      Reviewed-by: 's avatarUlan Degenbaev <ulan@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#54027}
      f1c79e02
    • Georg Neis's avatar
      Revert "Reland "Introduce MutableHeapNumber class."" · 722dfb70
      Georg Neis authored
      This reverts commit f0bcbc90.
      
      Reason for revert: Still failing bots.
      
      Original change's description:
      > Reland "Introduce MutableHeapNumber class."
      > 
      > This is a reland of 40ac6b18, which
      > was incorrect due to a bad merge.
      > 
      > Original change's description:
      > > Introduce MutableHeapNumber class.
      > >
      > > V8 knows heap numbers and mutable heap numbers. They have
      > > difference instance types, but in C++ code we've used the
      > > same class for both (HeapNumber). Confusingly, however,
      > > IsHeapNumber would return false for mutable heap numbers,
      > > while HeapNumber::cast would succeed.
      > >
      > > This CL adds a separate class MutableHeapNumber and
      > > eliminates the confusing behavior.
      > >
      > > TBR=bmeurer@chromium.org
      > >
      > > Change-Id: Id894d177c7fe8cc3f451be80c273b50daee91378
      > > Reviewed-on: https://chromium-review.googlesource.com/1113544
      > > Commit-Queue: Georg Neis <neis@chromium.org>
      > > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > > Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
      > > Cr-Commit-Position: refs/heads/master@{#54012}
      > 
      > TBR=bmeurer@chromium.org
      > TBR=ulanchromium.org
      > 
      > Change-Id: I3af1014c949821dfac0754a3e48c65ce1bad1ad1
      > Reviewed-on: https://chromium-review.googlesource.com/1114539
      > Reviewed-by: Georg Neis <neis@chromium.org>
      > Commit-Queue: Georg Neis <neis@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#54022}
      
      TBR=ulan@chromium.org,jarin@chromium.org,neis@chromium.org,bmeurer@chromium.org
      
      Change-Id: I99c226e95dfb0b913903cc83193f6e51de8c1b47
      No-Presubmit: true
      No-Tree-Checks: true
      No-Try: true
      Reviewed-on: https://chromium-review.googlesource.com/1114099Reviewed-by: 's avatarGeorg Neis <neis@chromium.org>
      Commit-Queue: Georg Neis <neis@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#54024}
      722dfb70
    • Georg Neis's avatar
      Reland "Introduce MutableHeapNumber class." · f0bcbc90
      Georg Neis authored
      This is a reland of 40ac6b18, which
      was incorrect due to a bad merge.
      
      Original change's description:
      > Introduce MutableHeapNumber class.
      >
      > V8 knows heap numbers and mutable heap numbers. They have
      > difference instance types, but in C++ code we've used the
      > same class for both (HeapNumber). Confusingly, however,
      > IsHeapNumber would return false for mutable heap numbers,
      > while HeapNumber::cast would succeed.
      >
      > This CL adds a separate class MutableHeapNumber and
      > eliminates the confusing behavior.
      >
      > TBR=bmeurer@chromium.org
      >
      > Change-Id: Id894d177c7fe8cc3f451be80c273b50daee91378
      > Reviewed-on: https://chromium-review.googlesource.com/1113544
      > Commit-Queue: Georg Neis <neis@chromium.org>
      > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#54012}
      
      TBR=bmeurer@chromium.org
      TBR=ulanchromium.org
      
      Change-Id: I3af1014c949821dfac0754a3e48c65ce1bad1ad1
      Reviewed-on: https://chromium-review.googlesource.com/1114539Reviewed-by: 's avatarGeorg Neis <neis@chromium.org>
      Commit-Queue: Georg Neis <neis@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#54022}
      f0bcbc90