1. 18 Jun, 2021 1 commit
  2. 16 Jun, 2021 1 commit
  3. 14 Jun, 2021 1 commit
  4. 10 Jun, 2021 1 commit
  5. 09 Jun, 2021 1 commit
  6. 10 May, 2021 1 commit
  7. 17 Mar, 2021 1 commit
  8. 20 Jan, 2021 1 commit
  9. 25 Nov, 2020 1 commit
  10. 20 Nov, 2020 1 commit
  11. 17 Nov, 2020 1 commit
  12. 16 Nov, 2020 1 commit
  13. 11 Nov, 2020 1 commit
  14. 28 Oct, 2020 2 commits
    • Leszek Swirski's avatar
      [serializer] Add write barrier for forward ref writes · 1301daeb
      Leszek Swirski authored
      Forward reference resolution writes didn't have a write barrier,
      which means the slot wouldn't be recorded if there was an active
      slot recording marker running.
      
      Now use the same SlotAccessor interface as the other deserializer
      writes, to make sure that the correct write barrier is called.
      
      As a drive-by, clean up SlotAccessorForHeapObject into two static
      constructors, to differentiate between access by slot index and
      offset.
      
      Fixed: v8:11065
      Bug: v8:10460
      Change-Id: I5b3a3d94057763324d6e1727d96b65c73ba5d7b4
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2504263
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Commit-Queue: Jakob Gruber <jgruber@chromium.org>
      Auto-Submit: Leszek Swirski <leszeks@chromium.org>
      Reviewed-by: 's avatarJakob Gruber <jgruber@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70839}
      1301daeb
    • Jakob Gruber's avatar
      [code] Move embedded metadata to the .rodata section · 779b0edd
      Jakob Gruber authored
      The embedded metadata section is the off-heap equivalent to an on-heap
      Code object's metadata section. It contains no executable data, thus
      .rodata is the natural home for it. Another motivation is that some
      platforms do not grant read permissions on the .text section.
      
      Embedded blob stats before:
      
        EmbeddedData:
          Total size: 1322944
          Data size:  25952
          Code size:  1296992
      
      And after:
      
        EmbeddedData:
          Total size: 1323372
          Data size:  121452
          Code size:  1201920
      
      (Slight size increase due to additional padding.)
      
      Bug: v8:11036,v8:10707
      Change-Id: Ib6b54a7e947966c7bd2fcc1e7e44c85e352f0063
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2502334Reviewed-by: 's avatarClemens Backes <clemensb@chromium.org>
      Reviewed-by: 's avatarUlan Degenbaev <ulan@chromium.org>
      Reviewed-by: 's avatarLeszek Swirski <leszeks@chromium.org>
      Auto-Submit: Jakob Gruber <jgruber@chromium.org>
      Commit-Queue: Ulan Degenbaev <ulan@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70822}
      779b0edd
  15. 22 Oct, 2020 1 commit
    • Jakob Gruber's avatar
      [code] Prepare to move metadata out of the instructions area · 0d0a3416
      Jakob Gruber authored
      No major functional changes in this CL, mostly it moves code around
      to make follow-up CLs less messy.
      
       - Document Code layout.
       - New concepts: 'body' and 'metadata' areas of Code objects. The
         metadata area contains metadata tables, the body area includes
         both instructions and metadata (this is currently the 'instructions'
         area). Add accessors for these new areas.
       - An interesting detail: embedded builtins will have non-adjacent
         instruction and metadata areas, thus a concept of 'body' doesn't
         make sense there.
       - Also add raw_instruction_X_future accessors; these are used where
         we are actually interested in the instructions range, not the entire
         body. In a follow-up, current raw_instruction_X accessors will be
         replaced by raw_body_X, and raw_instruction_X_future by
         raw_instruction_X.
      
      Bug: v8:11036
      Change-Id: I1d85146b652e0c097c3602d4db1862d5d3898a7e
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2491023
      Commit-Queue: Jakob Gruber <jgruber@chromium.org>
      Reviewed-by: 's avatarClemens Backes <clemensb@chromium.org>
      Reviewed-by: 's avatarDominik Inführ <dinfuehr@chromium.org>
      Reviewed-by: 's avatarLeszek Swirski <leszeks@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70701}
      0d0a3416
  16. 09 Oct, 2020 1 commit
  17. 08 Oct, 2020 1 commit
    • Leszek Swirski's avatar
      [serializer] Introduce HandleScopes for object contents · c212d10d
      Leszek Swirski authored
      Create a HandleScope when serializing an object's contents, to reduce
      the number of live handles during serialization. There's only a couple
      of cases where these handles have to outlive the serialized contents,
      and for these cases we introduce GlobalHandleVector or similar manual
      strong root mechanisms.
      
      In particular, backrefs don't actually need to exist as a handle vector
      (the object addresses are already referred to by the reference map's
      IdentityMap), except for DCHECKs, so this becomes a DEBUG-only global
      handle vector.
      
      To support this manual strong-rooting, the HotObjectList is split up
      into a strong-rooted find-only class in Serializer, and a Handle
      vector in Deserializer.
      
      Bug: chromium:1075999
      Change-Id: I586eeeb543e3f6c934c168961b068f2c34e72456
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2449980Reviewed-by: 's avatarUlan Degenbaev <ulan@chromium.org>
      Reviewed-by: 's avatarJakob Gruber <jgruber@chromium.org>
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70411}
      c212d10d
  18. 07 Oct, 2020 1 commit
    • Leszek Swirski's avatar
      Reland^4 "[serializer] Allocate during deserialization" · 3c508b38
      Leszek Swirski authored
      This relands commit 3f4e9bbe.
      which was a reland of c4a062a9
      which was a reland of 28a30c57
      which was a reland of 5d7a29c9
      
      The change had an issue that embedders implementing heap tracing (e.g.
      Unified Heap with Blink) could be passed an uninitialized pointer if
      marking happened during deserialization of an object containing such a
      pointer. Because of the 0xdeadbed0 uninitialized filler value, these
      embedders would then receive the value 0xdeadbed0deadbed0 as the
      'pointer', and crash on dereference.
      
      There is, however, special handling already for null pointers in heap
      tracing, also for dealing with not-yet initialized values. So, we can
      make the uninitialized Smi filler be 0x00000000, and that will make such
      embedded fields have a nullptr representation, making them follow the
      normal uninitialized value bailouts.
      
      In addition, it relands the following dependent changes, which are
      relanding unchanged and are followup performance improvements.
      Relanding them in the same change should allow for cleaner reverts
      should they be needed.
      
      This relands commit 76ad3ab5
      [identity-map] Change resize heuristic
      
      This relands commit 77cc96aa
      [identity-map] Cache the calculated Hash
      
      This relands commit bee5b996
      [serializer] Remove Deserializer::Initialize
      
      This relands commit c8f73f22
      [serializer] Cache instance type in PostProcessNewObject
      
      This relands commit 4e7c99ab
      [identity-map] Remove double-lookups in IdentityMap
      
      Original change's description:
      > Reland^3 "[serializer] Allocate during deserialization"
      >
      > This is a reland of c4a062a9
      > which was a reland of 28a30c57
      > which was a reland of 5d7a29c9
      >
      > Fixes TSAN errors from non-atomic writes in the deserializer. Now all
      > writes are (relaxed) atomic.
      >
      > Original change's description:
      > > Reland^2 "[serializer] Allocate during deserialization"
      > >
      > > This is a reland of 28a30c57
      > > which was a reland of 5d7a29c9
      > >
      > > The crashes were from calling RegisterDeserializerFinished on a null
      > > Isolate pointer, for a deserializer that was never initialised
      > > (specifically, ReadOnlyDeserializer when ROHeap is shared).
      > >
      > > Original change's description:
      > > > Reland "[serializer] Allocate during deserialization"
      > > >
      > > > This is a reland of 5d7a29c9
      > > >
      > > > This reland shuffles around the order of checks in Heap::AllocateRawWith
      > > > to not check the new space addresses until it's known that this is a new
      > > > space allocation. This fixes an UBSan failure during read-only space
      > > > deserialization, which happens before the new space is initialized.
      > > >
      > > > It also fixes some issues discovered by --stress-snapshot, around
      > > > serializing ThinStrings (which are now elided as part of serialization),
      > > > handle counts (I bumped the maximum handle count in that check), and
      > > > clearing map transitions (the map backpointer field needed a Smi
      > > > uninitialized value check).
      > > >
      > > > Original change's description:
      > > > > [serializer] Allocate during deserialization
      > > > >
      > > > > This patch removes the concept of reservations and a specialized
      > > > > deserializer allocator, and instead makes the deserializer allocate
      > > > > directly with the Heap's Allocate method.
      > > > >
      > > > > The major consequence of this is that the GC can now run during
      > > > > deserialization, which means that:
      > > > >
      > > > >   a) Deserialized objects are visible to the GC, and
      > > > >   b) Objects that the deserializer/deserialized objects point to can
      > > > >      move.
      > > > >
      > > > > Point a) is mostly not a problem due to previous work in making
      > > > > deserialized objects "GC valid", i.e. making sure that they have a valid
      > > > > size before any subsequent allocation/safepoint. We now additionally
      > > > > have to initialize the allocated space with a valid tagged value -- this
      > > > > is a magic Smi value to keep "uninitialized" checks simple.
      > > > >
      > > > > Point b) is solved by Handlifying the deserializer. This involves
      > > > > changing any vectors of objects into vectors of Handles, and any object
      > > > > keyed map into an IdentityMap (we can't use Handles as keys because
      > > > > the object's address is no longer a stable hash).
      > > > >
      > > > > Back-references can no longer be direct chunk offsets, so instead the
      > > > > deserializer stores a Handle to each deserialized object, and the
      > > > > backreference is an index into this handle array. This encoding could
      > > > > be optimized in the future with e.g. a second pass over the serialized
      > > > > array which emits a different bytecode for objects that are and aren't
      > > > > back-referenced.
      > > > >
      > > > > Additionally, the slot-walk over objects to initialize them can no
      > > > > longer use absolute slot offsets, as again an object may move and its
      > > > > slot address would become invalid. Now, slots are walked as relative
      > > > > offsets to a Handle to the object, or as absolute slots for the case of
      > > > > root pointers. A concept of "slot accessor" is introduced to share the
      > > > > code between these two modes, and writing the slot (including write
      > > > > barriers) is abstracted into this accessor.
      > > > >
      > > > > Finally, the Code body walk is modified to deserialize all objects
      > > > > referred to by RelocInfos before doing the RelocInfo walk itself. This
      > > > > is because RelocInfoIterator uses raw pointers, so we cannot allocate
      > > > > during a RelocInfo walk.
      > > > >
      > > > > As a drive-by, the VariableRawData bytecode is tweaked to use tagged
      > > > > size rather than byte size -- the size is expected to be tagged-aligned
      > > > > anyway, so now we get an extra few bits in the size encoding.
      > > > >
      > > > > Bug: chromium:1075999
      > > > > Change-Id: I672c42f553f2669888cc5e35d692c1b8ece1845e
      > > > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2404451
      > > > > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > > > > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > > > > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > > > > Cr-Commit-Position: refs/heads/master@{#70229}
      
      Bug: chromium:1075999
      Change-Id: Ib514a4ef16bd02bfb60d046ecbf8fae1ead64a98
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2452689
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Reviewed-by: 's avatarUlan Degenbaev <ulan@chromium.org>
      Reviewed-by: 's avatarJakob Gruber <jgruber@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70366}
      3c508b38
  19. 05 Oct, 2020 3 commits
    • Adam Klein's avatar
      Revert "Reland^3 "[serializer] Allocate during deserialization"" · a10ec2be
      Adam Klein authored
      This reverts commit 3f4e9bbe, along
      with the following dependent changes (reverted to make this a clean revert):
      76ad3ab5 [identity-map] Change resize heuristic
      77cc96aa [identity-map] Cache the calculated Hash
      bee5b996 [serializer] Remove Deserializer::Initialize
      c8f73f22 [serializer] Cache instance type in PostProcessNewObject
      4e7c99ab [identity-map] Remove double-lookups in IdentityMap
      
      Reason for revert: major crash spike on Canary (https://crbug.com/1135027)
      
      Original change's description:
      > Reland^3 "[serializer] Allocate during deserialization"
      >
      > This is a reland of c4a062a9
      > which was a reland of 28a30c57
      > which was a reland of 5d7a29c9
      >
      > Fixes TSAN errors from non-atomic writes in the deserializer. Now all
      > writes are (relaxed) atomic.
      >
      > Original change's description:
      > > Reland^2 "[serializer] Allocate during deserialization"
      > >
      > > This is a reland of 28a30c57
      > > which was a reland of 5d7a29c9
      > >
      > > The crashes were from calling RegisterDeserializerFinished on a null
      > > Isolate pointer, for a deserializer that was never initialised
      > > (specifically, ReadOnlyDeserializer when ROHeap is shared).
      > >
      > > Original change's description:
      > > > Reland "[serializer] Allocate during deserialization"
      > > >
      > > > This is a reland of 5d7a29c9
      > > >
      > > > This reland shuffles around the order of checks in Heap::AllocateRawWith
      > > > to not check the new space addresses until it's known that this is a new
      > > > space allocation. This fixes an UBSan failure during read-only space
      > > > deserialization, which happens before the new space is initialized.
      > > >
      > > > It also fixes some issues discovered by --stress-snapshot, around
      > > > serializing ThinStrings (which are now elided as part of serialization),
      > > > handle counts (I bumped the maximum handle count in that check), and
      > > > clearing map transitions (the map backpointer field needed a Smi
      > > > uninitialized value check).
      > > >
      > > > Original change's description:
      > > > > [serializer] Allocate during deserialization
      > > > >
      > > > > This patch removes the concept of reservations and a specialized
      > > > > deserializer allocator, and instead makes the deserializer allocate
      > > > > directly with the Heap's Allocate method.
      > > > >
      > > > > The major consequence of this is that the GC can now run during
      > > > > deserialization, which means that:
      > > > >
      > > > >   a) Deserialized objects are visible to the GC, and
      > > > >   b) Objects that the deserializer/deserialized objects point to can
      > > > >      move.
      > > > >
      > > > > Point a) is mostly not a problem due to previous work in making
      > > > > deserialized objects "GC valid", i.e. making sure that they have a valid
      > > > > size before any subsequent allocation/safepoint. We now additionally
      > > > > have to initialize the allocated space with a valid tagged value -- this
      > > > > is a magic Smi value to keep "uninitialized" checks simple.
      > > > >
      > > > > Point b) is solved by Handlifying the deserializer. This involves
      > > > > changing any vectors of objects into vectors of Handles, and any object
      > > > > keyed map into an IdentityMap (we can't use Handles as keys because
      > > > > the object's address is no longer a stable hash).
      > > > >
      > > > > Back-references can no longer be direct chunk offsets, so instead the
      > > > > deserializer stores a Handle to each deserialized object, and the
      > > > > backreference is an index into this handle array. This encoding could
      > > > > be optimized in the future with e.g. a second pass over the serialized
      > > > > array which emits a different bytecode for objects that are and aren't
      > > > > back-referenced.
      > > > >
      > > > > Additionally, the slot-walk over objects to initialize them can no
      > > > > longer use absolute slot offsets, as again an object may move and its
      > > > > slot address would become invalid. Now, slots are walked as relative
      > > > > offsets to a Handle to the object, or as absolute slots for the case of
      > > > > root pointers. A concept of "slot accessor" is introduced to share the
      > > > > code between these two modes, and writing the slot (including write
      > > > > barriers) is abstracted into this accessor.
      > > > >
      > > > > Finally, the Code body walk is modified to deserialize all objects
      > > > > referred to by RelocInfos before doing the RelocInfo walk itself. This
      > > > > is because RelocInfoIterator uses raw pointers, so we cannot allocate
      > > > > during a RelocInfo walk.
      > > > >
      > > > > As a drive-by, the VariableRawData bytecode is tweaked to use tagged
      > > > > size rather than byte size -- the size is expected to be tagged-aligned
      > > > > anyway, so now we get an extra few bits in the size encoding.
      > > > >
      > > > > Bug: chromium:1075999
      > > > > Change-Id: I672c42f553f2669888cc5e35d692c1b8ece1845e
      > > > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2404451
      > > > > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > > > > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > > > > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > > > > Cr-Commit-Position: refs/heads/master@{#70229}
      > > >
      > > > Bug: chromium:1075999
      > > > Change-Id: Ibc77cc48b3440b4a28b09746cfc47e50c340ce54
      > > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2440828
      > > > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > > > Auto-Submit: Leszek Swirski <leszeks@chromium.org>
      > > > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > > > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > > > Cr-Commit-Position: refs/heads/master@{#70267}
      > >
      > > Tbr: jgruber@chromium.org,ulan@chromium.org
      > > Bug: chromium:1075999
      > > Change-Id: Iaa8dc54895866ada0e34a7c9e8fff9ae1cb13f2d
      > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2444991
      > > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > > Cr-Commit-Position: refs/heads/master@{#70279}
      >
      > Tbr: jgruber@chromium.org,ulan@chromium.org
      > Cq-Include-Trybots: luci.v8.try:v8_linux64_tsan_rel_ng,v8_linux64_tsan_no_cm_rel_ng,v8_linux64_tsan_isolates_rel_ng
      > Bug: chromium:1075999
      > Change-Id: I0b9b11644aebc4cc8b07c62a0f765b24e4d73d89
      > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2445872
      > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > Auto-Submit: Leszek Swirski <leszeks@chromium.org>
      > Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#70288}
      
      TBR=ulan@chromium.org,jgruber@chromium.org,leszeks@chromium.org,dinfuehr@chromium.org
      
      Bug: chromium:1075999, chromium:1135027
      Change-Id: I5d0d9e49c0302d94ff7291834f5f18e7a0839eb7
      Cq-Include-Trybots: luci.v8.try:v8_linux64_tsan_rel_ng,v8_linux64_tsan_no_cm_rel_ng,v8_linux64_tsan_isolates_rel_ng
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2451030Reviewed-by: 's avatarAdam Klein <adamk@chromium.org>
      Commit-Queue: Adam Klein <adamk@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70328}
      a10ec2be
    • Leszek Swirski's avatar
      [serializer] Remove Deserializer::Initialize · bee5b996
      Leszek Swirski authored
      Remove the separate Initialize method from Deserializer, opting instead
      to pass around SnapshotData where appropriate and pass the isolate
      directly into the Deserializer's constructor.
      
      Change-Id: I0092fadd9c81f14b2ce75145fd81af37c3947c65
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2448466
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Commit-Queue: Dan Elphick <delphick@chromium.org>
      Auto-Submit: Leszek Swirski <leszeks@chromium.org>
      Reviewed-by: 's avatarDan Elphick <delphick@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70310}
      bee5b996
    • Leszek Swirski's avatar
      [serializer] Cache instance type in PostProcessNewObject · c8f73f22
      Leszek Swirski authored
      Rather than having repeated IsFoo checks in PostProcessNewObject, which
      means repeated handle accesses, map word accesses, and map pointer
      decompressions, cache the instance type once and check it with
      InstanceTypeChecker.
      
      This gives a measurable 2-3% improvement in deserialization time (in my
      informal local measurements).
      
      Bug: chromium:1075999
      Change-Id: I3e11588ad5d1c6ee2bbf93b82fa52c66496a325c
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2440578
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Reviewed-by: 's avatarJakob Gruber <jgruber@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70301}
      c8f73f22
  20. 02 Oct, 2020 3 commits
    • Leszek Swirski's avatar
      Reland^3 "[serializer] Allocate during deserialization" · 3f4e9bbe
      Leszek Swirski authored
      This is a reland of c4a062a9
      which was a reland of 28a30c57
      which was a reland of 5d7a29c9
      
      Fixes TSAN errors from non-atomic writes in the deserializer. Now all
      writes are (relaxed) atomic.
      
      Original change's description:
      > Reland^2 "[serializer] Allocate during deserialization"
      >
      > This is a reland of 28a30c57
      > which was a reland of 5d7a29c9
      >
      > The crashes were from calling RegisterDeserializerFinished on a null
      > Isolate pointer, for a deserializer that was never initialised
      > (specifically, ReadOnlyDeserializer when ROHeap is shared).
      >
      > Original change's description:
      > > Reland "[serializer] Allocate during deserialization"
      > >
      > > This is a reland of 5d7a29c9
      > >
      > > This reland shuffles around the order of checks in Heap::AllocateRawWith
      > > to not check the new space addresses until it's known that this is a new
      > > space allocation. This fixes an UBSan failure during read-only space
      > > deserialization, which happens before the new space is initialized.
      > >
      > > It also fixes some issues discovered by --stress-snapshot, around
      > > serializing ThinStrings (which are now elided as part of serialization),
      > > handle counts (I bumped the maximum handle count in that check), and
      > > clearing map transitions (the map backpointer field needed a Smi
      > > uninitialized value check).
      > >
      > > Original change's description:
      > > > [serializer] Allocate during deserialization
      > > >
      > > > This patch removes the concept of reservations and a specialized
      > > > deserializer allocator, and instead makes the deserializer allocate
      > > > directly with the Heap's Allocate method.
      > > >
      > > > The major consequence of this is that the GC can now run during
      > > > deserialization, which means that:
      > > >
      > > >   a) Deserialized objects are visible to the GC, and
      > > >   b) Objects that the deserializer/deserialized objects point to can
      > > >      move.
      > > >
      > > > Point a) is mostly not a problem due to previous work in making
      > > > deserialized objects "GC valid", i.e. making sure that they have a valid
      > > > size before any subsequent allocation/safepoint. We now additionally
      > > > have to initialize the allocated space with a valid tagged value -- this
      > > > is a magic Smi value to keep "uninitialized" checks simple.
      > > >
      > > > Point b) is solved by Handlifying the deserializer. This involves
      > > > changing any vectors of objects into vectors of Handles, and any object
      > > > keyed map into an IdentityMap (we can't use Handles as keys because
      > > > the object's address is no longer a stable hash).
      > > >
      > > > Back-references can no longer be direct chunk offsets, so instead the
      > > > deserializer stores a Handle to each deserialized object, and the
      > > > backreference is an index into this handle array. This encoding could
      > > > be optimized in the future with e.g. a second pass over the serialized
      > > > array which emits a different bytecode for objects that are and aren't
      > > > back-referenced.
      > > >
      > > > Additionally, the slot-walk over objects to initialize them can no
      > > > longer use absolute slot offsets, as again an object may move and its
      > > > slot address would become invalid. Now, slots are walked as relative
      > > > offsets to a Handle to the object, or as absolute slots for the case of
      > > > root pointers. A concept of "slot accessor" is introduced to share the
      > > > code between these two modes, and writing the slot (including write
      > > > barriers) is abstracted into this accessor.
      > > >
      > > > Finally, the Code body walk is modified to deserialize all objects
      > > > referred to by RelocInfos before doing the RelocInfo walk itself. This
      > > > is because RelocInfoIterator uses raw pointers, so we cannot allocate
      > > > during a RelocInfo walk.
      > > >
      > > > As a drive-by, the VariableRawData bytecode is tweaked to use tagged
      > > > size rather than byte size -- the size is expected to be tagged-aligned
      > > > anyway, so now we get an extra few bits in the size encoding.
      > > >
      > > > Bug: chromium:1075999
      > > > Change-Id: I672c42f553f2669888cc5e35d692c1b8ece1845e
      > > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2404451
      > > > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > > > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > > > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > > > Cr-Commit-Position: refs/heads/master@{#70229}
      > >
      > > Bug: chromium:1075999
      > > Change-Id: Ibc77cc48b3440b4a28b09746cfc47e50c340ce54
      > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2440828
      > > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > > Auto-Submit: Leszek Swirski <leszeks@chromium.org>
      > > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > > Cr-Commit-Position: refs/heads/master@{#70267}
      >
      > Tbr: jgruber@chromium.org,ulan@chromium.org
      > Bug: chromium:1075999
      > Change-Id: Iaa8dc54895866ada0e34a7c9e8fff9ae1cb13f2d
      > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2444991
      > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#70279}
      
      Tbr: jgruber@chromium.org,ulan@chromium.org
      Cq-Include-Trybots: luci.v8.try:v8_linux64_tsan_rel_ng,v8_linux64_tsan_no_cm_rel_ng,v8_linux64_tsan_isolates_rel_ng
      Bug: chromium:1075999
      Change-Id: I0b9b11644aebc4cc8b07c62a0f765b24e4d73d89
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2445872
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Auto-Submit: Leszek Swirski <leszeks@chromium.org>
      Reviewed-by: 's avatarDominik Inführ <dinfuehr@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70288}
      3f4e9bbe
    • Clemens Backes's avatar
      Revert "Reland^2 "[serializer] Allocate during deserialization"" · a81da102
      Clemens Backes authored
      This reverts commit c4a062a9.
      
      Reason for revert: TSan issues: https://ci.chromium.org/p/v8/builders/ci/V8%20Linux64%20TSAN/33504
      
      Original change's description:
      > Reland^2 "[serializer] Allocate during deserialization"
      >
      > This is a reland of 28a30c57
      > which was a reland of 5d7a29c9
      >
      > The crashes were from calling RegisterDeserializerFinished on a null
      > Isolate pointer, for a deserializer that was never initialised
      > (specifically, ReadOnlyDeserializer when ROHeap is shared).
      >
      > Original change's description:
      > > Reland "[serializer] Allocate during deserialization"
      > >
      > > This is a reland of 5d7a29c9
      > >
      > > This reland shuffles around the order of checks in Heap::AllocateRawWith
      > > to not check the new space addresses until it's known that this is a new
      > > space allocation. This fixes an UBSan failure during read-only space
      > > deserialization, which happens before the new space is initialized.
      > >
      > > It also fixes some issues discovered by --stress-snapshot, around
      > > serializing ThinStrings (which are now elided as part of serialization),
      > > handle counts (I bumped the maximum handle count in that check), and
      > > clearing map transitions (the map backpointer field needed a Smi
      > > uninitialized value check).
      > >
      > > Original change's description:
      > > > [serializer] Allocate during deserialization
      > > >
      > > > This patch removes the concept of reservations and a specialized
      > > > deserializer allocator, and instead makes the deserializer allocate
      > > > directly with the Heap's Allocate method.
      > > >
      > > > The major consequence of this is that the GC can now run during
      > > > deserialization, which means that:
      > > >
      > > >   a) Deserialized objects are visible to the GC, and
      > > >   b) Objects that the deserializer/deserialized objects point to can
      > > >      move.
      > > >
      > > > Point a) is mostly not a problem due to previous work in making
      > > > deserialized objects "GC valid", i.e. making sure that they have a valid
      > > > size before any subsequent allocation/safepoint. We now additionally
      > > > have to initialize the allocated space with a valid tagged value -- this
      > > > is a magic Smi value to keep "uninitialized" checks simple.
      > > >
      > > > Point b) is solved by Handlifying the deserializer. This involves
      > > > changing any vectors of objects into vectors of Handles, and any object
      > > > keyed map into an IdentityMap (we can't use Handles as keys because
      > > > the object's address is no longer a stable hash).
      > > >
      > > > Back-references can no longer be direct chunk offsets, so instead the
      > > > deserializer stores a Handle to each deserialized object, and the
      > > > backreference is an index into this handle array. This encoding could
      > > > be optimized in the future with e.g. a second pass over the serialized
      > > > array which emits a different bytecode for objects that are and aren't
      > > > back-referenced.
      > > >
      > > > Additionally, the slot-walk over objects to initialize them can no
      > > > longer use absolute slot offsets, as again an object may move and its
      > > > slot address would become invalid. Now, slots are walked as relative
      > > > offsets to a Handle to the object, or as absolute slots for the case of
      > > > root pointers. A concept of "slot accessor" is introduced to share the
      > > > code between these two modes, and writing the slot (including write
      > > > barriers) is abstracted into this accessor.
      > > >
      > > > Finally, the Code body walk is modified to deserialize all objects
      > > > referred to by RelocInfos before doing the RelocInfo walk itself. This
      > > > is because RelocInfoIterator uses raw pointers, so we cannot allocate
      > > > during a RelocInfo walk.
      > > >
      > > > As a drive-by, the VariableRawData bytecode is tweaked to use tagged
      > > > size rather than byte size -- the size is expected to be tagged-aligned
      > > > anyway, so now we get an extra few bits in the size encoding.
      > > >
      > > > Bug: chromium:1075999
      > > > Change-Id: I672c42f553f2669888cc5e35d692c1b8ece1845e
      > > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2404451
      > > > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > > > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > > > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > > > Cr-Commit-Position: refs/heads/master@{#70229}
      > >
      > > Bug: chromium:1075999
      > > Change-Id: Ibc77cc48b3440b4a28b09746cfc47e50c340ce54
      > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2440828
      > > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > > Auto-Submit: Leszek Swirski <leszeks@chromium.org>
      > > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > > Cr-Commit-Position: refs/heads/master@{#70267}
      >
      > Tbr: jgruber@chromium.org,ulan@chromium.org
      > Bug: chromium:1075999
      > Change-Id: Iaa8dc54895866ada0e34a7c9e8fff9ae1cb13f2d
      > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2444991
      > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#70279}
      
      TBR=ulan@chromium.org,jgruber@chromium.org,leszeks@chromium.org
      
      Change-Id: Ib2f01db4cd9b55639d6a4af971bda865edb45e84
      No-Presubmit: true
      No-Tree-Checks: true
      No-Try: true
      Bug: chromium:1075999
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2445250Reviewed-by: 's avatarClemens Backes <clemensb@chromium.org>
      Commit-Queue: Clemens Backes <clemensb@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70280}
      a81da102
    • Leszek Swirski's avatar
      Reland^2 "[serializer] Allocate during deserialization" · c4a062a9
      Leszek Swirski authored
      This is a reland of 28a30c57
      which was a reland of 5d7a29c9
      
      The crashes were from calling RegisterDeserializerFinished on a null
      Isolate pointer, for a deserializer that was never initialised
      (specifically, ReadOnlyDeserializer when ROHeap is shared).
      
      Original change's description:
      > Reland "[serializer] Allocate during deserialization"
      >
      > This is a reland of 5d7a29c9
      >
      > This reland shuffles around the order of checks in Heap::AllocateRawWith
      > to not check the new space addresses until it's known that this is a new
      > space allocation. This fixes an UBSan failure during read-only space
      > deserialization, which happens before the new space is initialized.
      >
      > It also fixes some issues discovered by --stress-snapshot, around
      > serializing ThinStrings (which are now elided as part of serialization),
      > handle counts (I bumped the maximum handle count in that check), and
      > clearing map transitions (the map backpointer field needed a Smi
      > uninitialized value check).
      >
      > Original change's description:
      > > [serializer] Allocate during deserialization
      > >
      > > This patch removes the concept of reservations and a specialized
      > > deserializer allocator, and instead makes the deserializer allocate
      > > directly with the Heap's Allocate method.
      > >
      > > The major consequence of this is that the GC can now run during
      > > deserialization, which means that:
      > >
      > >   a) Deserialized objects are visible to the GC, and
      > >   b) Objects that the deserializer/deserialized objects point to can
      > >      move.
      > >
      > > Point a) is mostly not a problem due to previous work in making
      > > deserialized objects "GC valid", i.e. making sure that they have a valid
      > > size before any subsequent allocation/safepoint. We now additionally
      > > have to initialize the allocated space with a valid tagged value -- this
      > > is a magic Smi value to keep "uninitialized" checks simple.
      > >
      > > Point b) is solved by Handlifying the deserializer. This involves
      > > changing any vectors of objects into vectors of Handles, and any object
      > > keyed map into an IdentityMap (we can't use Handles as keys because
      > > the object's address is no longer a stable hash).
      > >
      > > Back-references can no longer be direct chunk offsets, so instead the
      > > deserializer stores a Handle to each deserialized object, and the
      > > backreference is an index into this handle array. This encoding could
      > > be optimized in the future with e.g. a second pass over the serialized
      > > array which emits a different bytecode for objects that are and aren't
      > > back-referenced.
      > >
      > > Additionally, the slot-walk over objects to initialize them can no
      > > longer use absolute slot offsets, as again an object may move and its
      > > slot address would become invalid. Now, slots are walked as relative
      > > offsets to a Handle to the object, or as absolute slots for the case of
      > > root pointers. A concept of "slot accessor" is introduced to share the
      > > code between these two modes, and writing the slot (including write
      > > barriers) is abstracted into this accessor.
      > >
      > > Finally, the Code body walk is modified to deserialize all objects
      > > referred to by RelocInfos before doing the RelocInfo walk itself. This
      > > is because RelocInfoIterator uses raw pointers, so we cannot allocate
      > > during a RelocInfo walk.
      > >
      > > As a drive-by, the VariableRawData bytecode is tweaked to use tagged
      > > size rather than byte size -- the size is expected to be tagged-aligned
      > > anyway, so now we get an extra few bits in the size encoding.
      > >
      > > Bug: chromium:1075999
      > > Change-Id: I672c42f553f2669888cc5e35d692c1b8ece1845e
      > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2404451
      > > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > > Cr-Commit-Position: refs/heads/master@{#70229}
      >
      > Bug: chromium:1075999
      > Change-Id: Ibc77cc48b3440b4a28b09746cfc47e50c340ce54
      > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2440828
      > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > Auto-Submit: Leszek Swirski <leszeks@chromium.org>
      > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#70267}
      
      Tbr: jgruber@chromium.org,ulan@chromium.org
      Bug: chromium:1075999
      Change-Id: Iaa8dc54895866ada0e34a7c9e8fff9ae1cb13f2d
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2444991Reviewed-by: 's avatarUlan Degenbaev <ulan@chromium.org>
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70279}
      c4a062a9
  21. 01 Oct, 2020 2 commits
    • Zhi An Ng's avatar
      Revert "Reland "[serializer] Allocate during deserialization"" · c7c0e790
      Zhi An Ng authored
      This reverts commit 28a30c57.
      
      Reason for revert: Broke Test262 https://ci.chromium.org/p/v8/builders/ci/V8%20Linux%20-%20shared/38638?
      
      Original change's description:
      > Reland "[serializer] Allocate during deserialization"
      >
      > This is a reland of 5d7a29c9
      >
      > This reland shuffles around the order of checks in Heap::AllocateRawWith
      > to not check the new space addresses until it's known that this is a new
      > space allocation. This fixes an UBSan failure during read-only space
      > deserialization, which happens before the new space is initialized.
      >
      > It also fixes some issues discovered by --stress-snapshot, around
      > serializing ThinStrings (which are now elided as part of serialization),
      > handle counts (I bumped the maximum handle count in that check), and
      > clearing map transitions (the map backpointer field needed a Smi
      > uninitialized value check).
      >
      > Original change's description:
      > > [serializer] Allocate during deserialization
      > >
      > > This patch removes the concept of reservations and a specialized
      > > deserializer allocator, and instead makes the deserializer allocate
      > > directly with the Heap's Allocate method.
      > >
      > > The major consequence of this is that the GC can now run during
      > > deserialization, which means that:
      > >
      > >   a) Deserialized objects are visible to the GC, and
      > >   b) Objects that the deserializer/deserialized objects point to can
      > >      move.
      > >
      > > Point a) is mostly not a problem due to previous work in making
      > > deserialized objects "GC valid", i.e. making sure that they have a valid
      > > size before any subsequent allocation/safepoint. We now additionally
      > > have to initialize the allocated space with a valid tagged value -- this
      > > is a magic Smi value to keep "uninitialized" checks simple.
      > >
      > > Point b) is solved by Handlifying the deserializer. This involves
      > > changing any vectors of objects into vectors of Handles, and any object
      > > keyed map into an IdentityMap (we can't use Handles as keys because
      > > the object's address is no longer a stable hash).
      > >
      > > Back-references can no longer be direct chunk offsets, so instead the
      > > deserializer stores a Handle to each deserialized object, and the
      > > backreference is an index into this handle array. This encoding could
      > > be optimized in the future with e.g. a second pass over the serialized
      > > array which emits a different bytecode for objects that are and aren't
      > > back-referenced.
      > >
      > > Additionally, the slot-walk over objects to initialize them can no
      > > longer use absolute slot offsets, as again an object may move and its
      > > slot address would become invalid. Now, slots are walked as relative
      > > offsets to a Handle to the object, or as absolute slots for the case of
      > > root pointers. A concept of "slot accessor" is introduced to share the
      > > code between these two modes, and writing the slot (including write
      > > barriers) is abstracted into this accessor.
      > >
      > > Finally, the Code body walk is modified to deserialize all objects
      > > referred to by RelocInfos before doing the RelocInfo walk itself. This
      > > is because RelocInfoIterator uses raw pointers, so we cannot allocate
      > > during a RelocInfo walk.
      > >
      > > As a drive-by, the VariableRawData bytecode is tweaked to use tagged
      > > size rather than byte size -- the size is expected to be tagged-aligned
      > > anyway, so now we get an extra few bits in the size encoding.
      > >
      > > Bug: chromium:1075999
      > > Change-Id: I672c42f553f2669888cc5e35d692c1b8ece1845e
      > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2404451
      > > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > > Cr-Commit-Position: refs/heads/master@{#70229}
      >
      > Bug: chromium:1075999
      > Change-Id: Ibc77cc48b3440b4a28b09746cfc47e50c340ce54
      > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2440828
      > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > Auto-Submit: Leszek Swirski <leszeks@chromium.org>
      > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#70267}
      
      TBR=ulan@chromium.org,jgruber@chromium.org,leszeks@chromium.org
      
      Change-Id: Ieed68332ef6a7ad36db061e3f48be0f28673d7a2
      No-Presubmit: true
      No-Tree-Checks: true
      No-Try: true
      Bug: chromium:1075999
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2441608Reviewed-by: 's avatarZhi An Ng <zhin@chromium.org>
      Commit-Queue: Zhi An Ng <zhin@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70268}
      c7c0e790
    • Leszek Swirski's avatar
      Reland "[serializer] Allocate during deserialization" · 28a30c57
      Leszek Swirski authored
      This is a reland of 5d7a29c9
      
      This reland shuffles around the order of checks in Heap::AllocateRawWith
      to not check the new space addresses until it's known that this is a new
      space allocation. This fixes an UBSan failure during read-only space
      deserialization, which happens before the new space is initialized.
      
      It also fixes some issues discovered by --stress-snapshot, around
      serializing ThinStrings (which are now elided as part of serialization),
      handle counts (I bumped the maximum handle count in that check), and
      clearing map transitions (the map backpointer field needed a Smi
      uninitialized value check).
      
      Original change's description:
      > [serializer] Allocate during deserialization
      >
      > This patch removes the concept of reservations and a specialized
      > deserializer allocator, and instead makes the deserializer allocate
      > directly with the Heap's Allocate method.
      >
      > The major consequence of this is that the GC can now run during
      > deserialization, which means that:
      >
      >   a) Deserialized objects are visible to the GC, and
      >   b) Objects that the deserializer/deserialized objects point to can
      >      move.
      >
      > Point a) is mostly not a problem due to previous work in making
      > deserialized objects "GC valid", i.e. making sure that they have a valid
      > size before any subsequent allocation/safepoint. We now additionally
      > have to initialize the allocated space with a valid tagged value -- this
      > is a magic Smi value to keep "uninitialized" checks simple.
      >
      > Point b) is solved by Handlifying the deserializer. This involves
      > changing any vectors of objects into vectors of Handles, and any object
      > keyed map into an IdentityMap (we can't use Handles as keys because
      > the object's address is no longer a stable hash).
      >
      > Back-references can no longer be direct chunk offsets, so instead the
      > deserializer stores a Handle to each deserialized object, and the
      > backreference is an index into this handle array. This encoding could
      > be optimized in the future with e.g. a second pass over the serialized
      > array which emits a different bytecode for objects that are and aren't
      > back-referenced.
      >
      > Additionally, the slot-walk over objects to initialize them can no
      > longer use absolute slot offsets, as again an object may move and its
      > slot address would become invalid. Now, slots are walked as relative
      > offsets to a Handle to the object, or as absolute slots for the case of
      > root pointers. A concept of "slot accessor" is introduced to share the
      > code between these two modes, and writing the slot (including write
      > barriers) is abstracted into this accessor.
      >
      > Finally, the Code body walk is modified to deserialize all objects
      > referred to by RelocInfos before doing the RelocInfo walk itself. This
      > is because RelocInfoIterator uses raw pointers, so we cannot allocate
      > during a RelocInfo walk.
      >
      > As a drive-by, the VariableRawData bytecode is tweaked to use tagged
      > size rather than byte size -- the size is expected to be tagged-aligned
      > anyway, so now we get an extra few bits in the size encoding.
      >
      > Bug: chromium:1075999
      > Change-Id: I672c42f553f2669888cc5e35d692c1b8ece1845e
      > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2404451
      > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#70229}
      
      Bug: chromium:1075999
      Change-Id: Ibc77cc48b3440b4a28b09746cfc47e50c340ce54
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2440828
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Auto-Submit: Leszek Swirski <leszeks@chromium.org>
      Reviewed-by: 's avatarUlan Degenbaev <ulan@chromium.org>
      Reviewed-by: 's avatarJakob Gruber <jgruber@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70267}
      28a30c57
  22. 30 Sep, 2020 2 commits
    • Leszek Swirski's avatar
      Revert "[serializer] Allocate during deserialization" · 74f3665c
      Leszek Swirski authored
      This reverts commit 5d7a29c9.
      
      Reason for revert: UBSan -- https://ci.chromium.org/p/v8/builders/ci/V8%20Linux64%20UBSan/13100
      
      Original change's description:
      > [serializer] Allocate during deserialization
      >
      > This patch removes the concept of reservations and a specialized
      > deserializer allocator, and instead makes the deserializer allocate
      > directly with the Heap's Allocate method.
      >
      > The major consequence of this is that the GC can now run during
      > deserialization, which means that:
      >
      >   a) Deserialized objects are visible to the GC, and
      >   b) Objects that the deserializer/deserialized objects point to can
      >      move.
      >
      > Point a) is mostly not a problem due to previous work in making
      > deserialized objects "GC valid", i.e. making sure that they have a valid
      > size before any subsequent allocation/safepoint. We now additionally
      > have to initialize the allocated space with a valid tagged value -- this
      > is a magic Smi value to keep "uninitialized" checks simple.
      >
      > Point b) is solved by Handlifying the deserializer. This involves
      > changing any vectors of objects into vectors of Handles, and any object
      > keyed map into an IdentityMap (we can't use Handles as keys because
      > the object's address is no longer a stable hash).
      >
      > Back-references can no longer be direct chunk offsets, so instead the
      > deserializer stores a Handle to each deserialized object, and the
      > backreference is an index into this handle array. This encoding could
      > be optimized in the future with e.g. a second pass over the serialized
      > array which emits a different bytecode for objects that are and aren't
      > back-referenced.
      >
      > Additionally, the slot-walk over objects to initialize them can no
      > longer use absolute slot offsets, as again an object may move and its
      > slot address would become invalid. Now, slots are walked as relative
      > offsets to a Handle to the object, or as absolute slots for the case of
      > root pointers. A concept of "slot accessor" is introduced to share the
      > code between these two modes, and writing the slot (including write
      > barriers) is abstracted into this accessor.
      >
      > Finally, the Code body walk is modified to deserialize all objects
      > referred to by RelocInfos before doing the RelocInfo walk itself. This
      > is because RelocInfoIterator uses raw pointers, so we cannot allocate
      > during a RelocInfo walk.
      >
      > As a drive-by, the VariableRawData bytecode is tweaked to use tagged
      > size rather than byte size -- the size is expected to be tagged-aligned
      > anyway, so now we get an extra few bits in the size encoding.
      >
      > Bug: chromium:1075999
      > Change-Id: I672c42f553f2669888cc5e35d692c1b8ece1845e
      > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2404451
      > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#70229}
      
      TBR=ulan@chromium.org,jgruber@chromium.org,leszeks@chromium.org
      
      Change-Id: I2bd792a24861e8f54897e51522769b50f8f814e2
      No-Presubmit: true
      No-Tree-Checks: true
      No-Try: true
      Bug: chromium:1075999
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2440827
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Reviewed-by: 's avatarLeszek Swirski <leszeks@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#70231}
      74f3665c
    • Leszek Swirski's avatar
      [serializer] Allocate during deserialization · 5d7a29c9
      Leszek Swirski authored
      This patch removes the concept of reservations and a specialized
      deserializer allocator, and instead makes the deserializer allocate
      directly with the Heap's Allocate method.
      
      The major consequence of this is that the GC can now run during
      deserialization, which means that:
      
        a) Deserialized objects are visible to the GC, and
        b) Objects that the deserializer/deserialized objects point to can
           move.
      
      Point a) is mostly not a problem due to previous work in making
      deserialized objects "GC valid", i.e. making sure that they have a valid
      size before any subsequent allocation/safepoint. We now additionally
      have to initialize the allocated space with a valid tagged value -- this
      is a magic Smi value to keep "uninitialized" checks simple.
      
      Point b) is solved by Handlifying the deserializer. This involves
      changing any vectors of objects into vectors of Handles, and any object
      keyed map into an IdentityMap (we can't use Handles as keys because
      the object's address is no longer a stable hash).
      
      Back-references can no longer be direct chunk offsets, so instead the
      deserializer stores a Handle to each deserialized object, and the
      backreference is an index into this handle array. This encoding could
      be optimized in the future with e.g. a second pass over the serialized
      array which emits a different bytecode for objects that are and aren't
      back-referenced.
      
      Additionally, the slot-walk over objects to initialize them can no
      longer use absolute slot offsets, as again an object may move and its
      slot address would become invalid. Now, slots are walked as relative
      offsets to a Handle to the object, or as absolute slots for the case of
      root pointers. A concept of "slot accessor" is introduced to share the
      code between these two modes, and writing the slot (including write
      barriers) is abstracted into this accessor.
      
      Finally, the Code body walk is modified to deserialize all objects
      referred to by RelocInfos before doing the RelocInfo walk itself. This
      is because RelocInfoIterator uses raw pointers, so we cannot allocate
      during a RelocInfo walk.
      
      As a drive-by, the VariableRawData bytecode is tweaked to use tagged
      size rather than byte size -- the size is expected to be tagged-aligned
      anyway, so now we get an extra few bits in the size encoding.
      
      Bug: chromium:1075999
      Change-Id: I672c42f553f2669888cc5e35d692c1b8ece1845e
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2404451
      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@{#70229}
      5d7a29c9
  23. 29 Sep, 2020 1 commit
  24. 22 Sep, 2020 1 commit
  25. 18 Sep, 2020 1 commit
    • Leszek Swirski's avatar
      [snapshot] Reland recent de/serializer related changes · 695d9b64
      Leszek Swirski authored
      This reverts commit 1aa9ab73.
      
      The reverted CL chain had an issue where ThinStrings could accidentally
      end up in compilation artifacts, causing issues down the line with ICs
      that expected direct internalized strings.
      
      The reason for this bug was that forward references to internalized
      strings were resolved before PostProcessNewObject. When this happened,
      the internalized string A would be written to the field where it was
      previously deferred, then PostProcessNewObject would change string A to
      string A', and update string A to a ThinString.  This means any _future_
      back references to A would see the ThinString and follow it to receive
      A', but any _past_ forward references would keep pointing to the
      ThinString A.
      
      This reland fixes this by preventing InternalizedString deferral, so
      that all references to InternalizedStrings are back references. It also
      adds some additional verification to the heap verifier that constant
      pools and object boilerplate descriptors aren't allowed to hold thin
      strings.
      
      This patch also fixes an additional bug in the original CL, where weak
      forward refs weren't being serialized with a weak prefix.
      
      Original change's description:
      > Revert recent de/serializer related changes
      >
      > They are suspected to be causing Canary crashes, confirmed through
      > local reverts and repro attempts.
      >
      > This reverts:
      > - "Reland "[serializer] Change deferring to use forward refs""
      >   commit 76d684cc.
      > - "Reland "[serializer] Remove new space""
      >   commit 81231c23.
      > - "[serializer] Clean-up and de-macro ReadDataCase"
      >   commit c06d24b9.
      > - "[serializer] DCHECK deserializer allocations are initialized"
      >   commit fbc1f32d.
      >
      > Bug: chromium:1128872
      > Change-Id: Id2bb3b8fac526fdf9ffb033222ae08cd423f8238
      > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2414220
      > Reviewed-by: Igor Sheludko <ishell@chromium.org>
      > Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
      > Reviewed-by: Adam Klein <adamk@chromium.org>
      > Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#69955}
      
      Tbr: jgruber@chromium.org,dinfuehr@chromium.org
      Bug: chromium:1075999
      Bug: chromium:1127610
      Bug: chromium:1128848
      Bug: chromium:1128872
      Bug: chromium:1128957
      Change-Id: I8b7bbabf77eb8cb942a28316afbfaa5f9a0aa4cb
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2418101
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Reviewed-by: 's avatarIgor Sheludko <ishell@chromium.org>
      Reviewed-by: 's avatarLeszek Swirski <leszeks@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#69988}
      695d9b64
  26. 16 Sep, 2020 1 commit
  27. 10 Sep, 2020 1 commit
    • Leszek Swirski's avatar
      [serializer] Clean-up and de-macro ReadDataCase · c06d24b9
      Leszek Swirski authored
      Refactors weak prefix handling, in particular the post-hoc weak prefix
      read and HeapObjectReference creation, to a few function calls. This
      simplifies ReadDataCase sufficiently that it can be inlined into
      ReadData, which removes the need for a) having two places where we
      branch on the bytecode value (ReadData and ReadDataCase), and b)
      removes the need for the macro helper which calls ReadData. With a
      bit of refactoring we can therefore make the big switch much more
      explicit.
      
      This patch also moves that switch into a per-bytecode helper, so that
      switch entries can return the updated slot, rather than remembering to
      update in-place and continue looping.
      
      It also moves the weak prefix handling from the deserializer allocator
      to the deserializer itself, as weak prefixes don't have anything to do
      with allocation.
      
      Change-Id: I84fbda021cb65d5bfb91fc3ef27f72823acee05a
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2395557
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Reviewed-by: 's avatarJakob Gruber <jgruber@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#69800}
      c06d24b9
  28. 09 Sep, 2020 6 commits
    • Leszek Swirski's avatar
      Reland "[serializer] Remove new space" · 81231c23
      Leszek Swirski authored
      This is a reland of 1c7618ab
      
      The revert was due to an missing dependency in the incremental build,
      fixed in https://crrev.com/c/2400987.
      
      Original change's description:
      > [serializer] Remove new space
      >
      > The new space is unused in the snapshot, as we convert all new objects
      > to old space objects when serializing. This means we can get rid of
      > the snapshot new space entirely, and as a result get rid of the write
      > barrier checks.
      >
      > This also rejiggles the order of the general spaces enum so that the new
      > spaces are at the end, and can be truncated off for the SnapshotSpace
      > enum.
      >
      > As a drive by, fix a bug in an unrelated test-api test which this patch
      > exposed.
      >
      > Change-Id: If67ff8be5bf03104a3ffae7df707c22460bba3a1
      > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2390762
      > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#69761}
      
      Tbr: jgruber@chromium.org,dinfuehr@chromium.org
      Change-Id: I9fbc61a124fae09d12d6281baaca60eb6c39a6e5
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2401420Reviewed-by: 's avatarLeszek Swirski <leszeks@chromium.org>
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#69785}
      81231c23
    • Leszek Swirski's avatar
      Reland "[serializer] Change deferring to use forward refs" · 76d684cc
      Leszek Swirski authored
      This is a reland of 81577a79
      
      The revert was due to an missing dependency in the incremental build,
      fixed in https://crrev.com/c/2400987.
      
      Original change's description:
      > [serializer] Change deferring to use forward refs
      >
      > Now that we have forward references, we can replace the body deferring
      > mechanism with forward references to the entire pointer.
      >
      > This ensures that objects are always deserialized with their contents
      > (aside from themselves maybe holding forward refs), and as a result we
      > can simplify the CanBeDeferred conditions which encode the constraint
      > that some objects either need immediately have contents, or cannot be
      > deferred because their fields are changed temporarily (e.g. backing
      > store refs).
      >
      > This also means that objects with length fields (e.g. arrays) will
      > always have those length fields deserialized when the object is
      > deserialized, which was not the case when the body could be deferred.
      > This helps us in the plan to make GC possible during deserialization.
      >
      > Bug: v8:10815
      > Change-Id: Ib0e5399b9de6027765691e8cb47410a2ccc15485
      > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2390643
      > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#69760}
      
      Tbr: jgruber@chromium.org
      Bug: v8:10815
      Change-Id: I235076a97c5dfa58513e880cc477ac72a28b29e9
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2400992Reviewed-by: 's avatarJakob Gruber <jgruber@chromium.org>
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#69779}
      76d684cc
    • Sathya Gunasekaran's avatar
      Revert "[serializer] Change deferring to use forward refs" · cb1a96e5
      Sathya Gunasekaran authored
      This reverts commit 81577a79.
      
      Reason for revert: https://ci.chromium.org/p/v8/builders/ci/V8%20Linux64%20-%20shared/10544
      
      Original change's description:
      > [serializer] Change deferring to use forward refs
      > 
      > Now that we have forward references, we can replace the body deferring
      > mechanism with forward references to the entire pointer.
      > 
      > This ensures that objects are always deserialized with their contents
      > (aside from themselves maybe holding forward refs), and as a result we
      > can simplify the CanBeDeferred conditions which encode the constraint
      > that some objects either need immediately have contents, or cannot be
      > deferred because their fields are changed temporarily (e.g. backing
      > store refs).
      > 
      > This also means that objects with length fields (e.g. arrays) will
      > always have those length fields deserialized when the object is
      > deserialized, which was not the case when the body could be deferred.
      > This helps us in the plan to make GC possible during deserialization.
      > 
      > Bug: v8:10815
      > Change-Id: Ib0e5399b9de6027765691e8cb47410a2ccc15485
      > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2390643
      > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#69760}
      
      TBR=jgruber@chromium.org,leszeks@chromium.org
      
      Change-Id: I7a93a59217a2b38e2157c0f7ffc7ac648590a8d6
      No-Presubmit: true
      No-Tree-Checks: true
      No-Try: true
      Bug: v8:10815
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2398535Reviewed-by: 's avatarSathya Gunasekaran  <gsathya@chromium.org>
      Commit-Queue: Sathya Gunasekaran  <gsathya@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#69763}
      cb1a96e5
    • Sathya Gunasekaran's avatar
      Revert "[serializer] Remove new space" · 79e4420c
      Sathya Gunasekaran authored
      This reverts commit 1c7618ab.
      
      Reason for revert: https://ci.chromium.org/p/v8/builders/ci/V8%20Linux64%20-%20shared/10544
      
      Original change's description:
      > [serializer] Remove new space
      > 
      > The new space is unused in the snapshot, as we convert all new objects
      > to old space objects when serializing. This means we can get rid of
      > the snapshot new space entirely, and as a result get rid of the write
      > barrier checks.
      > 
      > This also rejiggles the order of the general spaces enum so that the new
      > spaces are at the end, and can be truncated off for the SnapshotSpace
      > enum.
      > 
      > As a drive by, fix a bug in an unrelated test-api test which this patch
      > exposed.
      > 
      > Change-Id: If67ff8be5bf03104a3ffae7df707c22460bba3a1
      > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2390762
      > Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      > Reviewed-by: Jakob Gruber <jgruber@chromium.org>
      > Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
      > Cr-Commit-Position: refs/heads/master@{#69761}
      
      TBR=jgruber@chromium.org,leszeks@chromium.org,dinfuehr@chromium.org
      
      Change-Id: Iaf2362d8cd3a17d8410030aca0dd2250c5a0a7af
      No-Presubmit: true
      No-Tree-Checks: true
      No-Try: true
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2398533Reviewed-by: 's avatarSathya Gunasekaran  <gsathya@chromium.org>
      Commit-Queue: Sathya Gunasekaran  <gsathya@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#69762}
      79e4420c
    • Leszek Swirski's avatar
      [serializer] Remove new space · 1c7618ab
      Leszek Swirski authored
      The new space is unused in the snapshot, as we convert all new objects
      to old space objects when serializing. This means we can get rid of
      the snapshot new space entirely, and as a result get rid of the write
      barrier checks.
      
      This also rejiggles the order of the general spaces enum so that the new
      spaces are at the end, and can be truncated off for the SnapshotSpace
      enum.
      
      As a drive by, fix a bug in an unrelated test-api test which this patch
      exposed.
      
      Change-Id: If67ff8be5bf03104a3ffae7df707c22460bba3a1
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2390762
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Reviewed-by: 's avatarJakob Gruber <jgruber@chromium.org>
      Reviewed-by: 's avatarDominik Inführ <dinfuehr@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#69761}
      1c7618ab
    • Leszek Swirski's avatar
      [serializer] Change deferring to use forward refs · 81577a79
      Leszek Swirski authored
      Now that we have forward references, we can replace the body deferring
      mechanism with forward references to the entire pointer.
      
      This ensures that objects are always deserialized with their contents
      (aside from themselves maybe holding forward refs), and as a result we
      can simplify the CanBeDeferred conditions which encode the constraint
      that some objects either need immediately have contents, or cannot be
      deferred because their fields are changed temporarily (e.g. backing
      store refs).
      
      This also means that objects with length fields (e.g. arrays) will
      always have those length fields deserialized when the object is
      deserialized, which was not the case when the body could be deferred.
      This helps us in the plan to make GC possible during deserialization.
      
      Bug: v8:10815
      Change-Id: Ib0e5399b9de6027765691e8cb47410a2ccc15485
      Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2390643Reviewed-by: 's avatarJakob Gruber <jgruber@chromium.org>
      Commit-Queue: Leszek Swirski <leszeks@chromium.org>
      Cr-Commit-Position: refs/heads/master@{#69760}
      81577a79