Skip to content

V
V8
Switch branch/tag
  1. 27 Jan, 2022 1 commit
    • Manos Koukoutos's avatar
      [wasm-gc] Implement iso-recursive types · fe531e1f
      Manos Koukoutos authored 
      This CL replaces the equirecursive type system for wasm-gc with the
isorecursive hybrid type system presented here:
https://github.com/WebAssembly/gc/issues/257.

In broad strokes, this includes the following changes:
- In the module decoder, remove equirecursive types. Implement recursive
  type groups, subtype definitions, and function/struct/array
  definitions. Treat nominal modules as syntactic sugar of an
  isorecursive module, where all types belong in the same recursive
  group.
- Remove rtt.sub and all related infrastructure.
- Change subtyping to work with explicit supertypes only.
- Add ValidSubtypeDefinition in subtyping, to check that subtype
  declarations are valid during decoding.
- Remove the subtyping cache.
- Add support for functions to have specific signature index in
  WasmModuleBuilder and in test-gc.cc.
- Adapt tests.

Current restrictions:
- Recursive groups are not stored beyond decoding.
- Type canonicalization is not implemented. No tests relying on types
  being considered identical post-canonicalization.
- No cross-module subtyping is possible. Tests relying on cross-module
  subtyping have been commented out.

Bug: v8:7748
Change-Id: I69fd04ecc5611f6230c95d5c89d1c520163fffae
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3416239Reviewed-by: 's avatarTobias Tebbi <tebbi@chromium.org>
Reviewed-by: 's avatarJakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Manos Koukoutos <manoskouk@chromium.org>
Cr-Commit-Position: refs/heads/main@{#78816}
      fe531e1f
  3. 10 Nov, 2021 1 commit
  5. 02 Nov, 2021 1 commit
    • Manos Koukoutos's avatar
      [wasm] Simplify 'ref' field for API-defined functions · 48b2b891
      Manos Koukoutos authored 
      Design doc:  https://bit.ly/3jEVgzz

Summary:
We change the context for WasmJSFunction and WasmCapiFunction from a
tuple containing the instance to a triple WasmAPIFunctionRef =
{isolate root, context, callable}. This way we do not have to maintain
the correct instance at runtime. Also, a few places in the code get
simplified.

Changes:
- In WasmGraphBuilder, support having a WasmAPIFunctionRef at parameter
  0.
- Remove unpacking of (instance, callable) tuple from code generators.
- Remove the part in WasmGraphBuilder and LiftoffCompiler that used to
  set the instance field of the function reference.
- Modify code that handles the 'ref' field in wasm-objects.*, factory.*
  and c-api.cc.
- Fix the recorded safepoint address for arm when calling a C function
  from wasm.
- (Drive-by) Remove WasmAllocatePair.

Bug: v8:11510
Change-Id: I2a4ef3abaf9da36c4a2d85b434965a40a289b9ec
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3236719
Commit-Queue: Manos Koukoutos <manoskouk@chromium.org>
Reviewed-by: 's avatarJakob Kummerow <jkummerow@chromium.org>
Reviewed-by: 's avatarMichael Lippautz <mlippautz@chromium.org>
Reviewed-by: 's avatarMaya Lekova <mslekova@chromium.org>
Cr-Commit-Position: refs/heads/main@{#77639}
      48b2b891
  7. 04 Aug, 2021 1 commit
  9. 19 Jul, 2021 1 commit
    • Jakob Gruber's avatar
      [compiler] Make JSFunction bg-serialized · 0dba97f8
      Jakob Gruber authored 
      This wraps up the transition away from kSerialized ref kinds.

Since JSFunctionRef is a complex type, we don't attempt full
consistency on the background thread. Instead, we serialize functions
on the background in a partially-racy manner, in which consistency
between different JSFunction fields is *not* guaranteed. Consistency
is later verified through a new compilation dependency kind during
finalization.

Bug: v8:7790, v8:12004
Change-Id: Ic2b78af9c9fe183c8769d323132bb304b151dc75
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2968404
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: 's avatarIgor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#75789}
      0dba97f8
  11. 24 Jun, 2021 1 commit
  13. 07 Jun, 2021 1 commit
  15. 23 Mar, 2021 1 commit
  17. 26 Jan, 2021 1 commit
    • Paolo Severini's avatar
      [test][turbofan] Add %ObserveNode intrinsic for node-specific tests · 4e9f6513
      Paolo Severini authored 
      (Initially copied from nicohartmann@ CL
https://chromium-review.googlesource.com/c/v8/v8/+/2135631)

This CL adds a new intrinsic %ObserveNode(expr) which has noop semantics
but triggers the new NodeObserver set on the OptimizedCompilationInfo
when the node generated for expr is created or changed in any phase
(until EffectControlLinearization).

This provides the infrastructure to write reasonable unit tests that
check for the construction of or lowering to specific nodes (e.g.
depending on feedback).

When %ObserveNode(expr) is used an object of class ObserveNodeManager is
registered to every Reducer/GraphReducer and is notified by the Reducer
with all node changes. The same logic is added to classes
SimplifiedLowering/RepresentationSelector, which do not inherit from
class Reducer.

Observed Node modifications currently are:
 * The Node Operator
 * The Node type
 * Node replacements

A first use case (cctest/test-sloppy-equality.cc) is included in this CL.

Change-Id: Idc5a5e38af8b1d9a2ec5021bf821c4e4e1406220
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2555219
Commit-Queue: Paolo Severini <paolosev@microsoft.com>
Reviewed-by: 's avatarGeorg Neis <neis@chromium.org>
Reviewed-by: 's avatarNico Hartmann <nicohartmann@chromium.org>
Cr-Commit-Position: refs/heads/master@{#72331}
      4e9f6513
  19. 10 Nov, 2020 1 commit
  21. 09 Nov, 2020 2 commits
  23. 05 Nov, 2020 2 commits
  25. 28 Oct, 2020 1 commit
  27. 03 Apr, 2020 1 commit
    • Ross McIlroy's avatar
      [TurboProp] Remove unreachable successor basic blocks from schedule. · 4a2ef63c
      Ross McIlroy authored 
      Effect-control-linearizer will update a basic block to connect it
directly to the end node if it has an Unreachable node. Usually the
block would already have been connected directly to end (via a Throw
node) already, however in some cases it can be connected indirectly
(via a branch, where both end in a throw node).

If this happens, and the Effect-control-linearizer is maintaining the
schedule (e.g., for TurboProp), it will cause the end block to have
unreachable predecessor blocks, which can cause issues with the
register allocator.

To fix this, have the BasicBlockUpdater remove all successor blocks
from the schedule, when they become Unreachable. Also add some tests
to cover this in effect-control-linearizer-unittests.

BUG=v8:10332,v8:9684

Change-Id: Ibce140e6d1f61751a86247e6f8c36075723a1e55
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2120537
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: 's avatarTobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#66994}
      4a2ef63c
  29. 19 Feb, 2020 1 commit
  31. 10 Dec, 2019 1 commit
  33. 05 Dec, 2019 1 commit
  35. 30 Aug, 2019 1 commit
  37. 28 Aug, 2019 1 commit
  39. 05 Jul, 2019 1 commit
  41. 01 Jul, 2019 1 commit
  43. 23 May, 2019 2 commits
  45. 03 May, 2019 2 commits
  47. 02 May, 2019 1 commit
  49. 12 Apr, 2019 1 commit
  51. 18 Mar, 2019 2 commits
  53. 25 Feb, 2019 1 commit
  55. 08 Feb, 2019 1 commit
  57. 02 Nov, 2018 1 commit
  59. 22 Oct, 2018 1 commit
  61. 18 Oct, 2018 2 commits
  63. 16 Oct, 2018 1 commit
  65. 12 Oct, 2018 1 commit
  67. 21 Sep, 2018 1 commit