Previously, we performed "is A subtype of B?" checks by walking
A's supertypes list and comparing every found type to B.
This CL stores not just A's immediate parent type on A, but its
entire list of supertypes, and uses that list plus compile-time
knowledge of B's distance to the root type in order to compare
only exactly one of A's supertypes to B.
Bug: v8:7748
Change-Id: I0011b72c4b54440b16494918f64d8fb119bef8b0
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2527097
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Manos Koukoutos <manoskouk@chromium.org>
Cr-Commit-Position: refs/heads/master@{#71127}
This change plumbs import assertions from SourceTextModuleDescriptor's
ModuleRequestMap into SourceTextModuleInfo via a new ModuleRequest
type, where previously there had been only the specifier.
SourceTextModuleDescriptor::module_map now deduplicates module requests
using the specifier and the import assertions. Continuing to use the
specifier alone would cause a loss of information in the event that
a module imports from the same specifier multiple times using different
sets of assertions. Failing to deduplicate at all would result in
multiple requests for statements like `import {a,b,c} from "foo.js"`,
which would be a potential performance issue. See design doc at
https://docs.google.com/document/d/1yuXgNHSbTAPubT1Mg0JXp5uTrfirkvO1g5cHHCe-LmY
for more detail on this decision.
v8::internal::ModuleRequest holds the assertions as an array of the form
[key1, value1, position1, key2, value2, assertion2, ...]. However the
parser still needs to use a map, since duplicate assertion keys need to
be detected at parse time. A follow-up change will ensure that
assertions are sorted using a proper lexicographic sort.
Bug: v8:10958
Change-Id: Iff13fb9a37d58fc1622cd3cce78925ad2b7a14bb
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2493060
Commit-Queue: Dan Clark <daniec@microsoft.com>
Reviewed-by: Adam Klein <adamk@chromium.org>
Reviewed-by: Marja Hölttä <marja@chromium.org>
Cr-Commit-Position: refs/heads/master@{#71066}
This is a reland of 26f10ecd95
Change compared to original CL:
The deserializer changes StrongDescriptorArray to DescriptorArray.
Since this CL uses separate BodyDescriptors for the two kinds of
descriptor arrays, this caused a DCHECK failure when the deserializer
changes the map while the object is visited from the concurrent marking
thread. Fix this by disabling the corresponding checks.
Original change's description:
> [torque] allow exported classes with custom C++ class
>
> Introduce a new annotation @customCppClass that can be used for
> non-extern @export classes, that is, generate everything, remove
> boilerplate from all the internal lists and switches, but allow
> a custom C++ class, which in turn also allows overwriting the generated
> print and verify functions.
>
> Port DescriptorArray and StrongDescriptorArray as an example.
>
> Bug: v8:7793
> Change-Id: I744e52fb4102ac49c0097f1c95bb17d301975bf0
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2489687
> Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
> Reviewed-by: Nico Hartmann <nicohartmann@chromium.org>
> Reviewed-by: Seth Brenith <seth.brenith@microsoft.com>
> Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#70989}
Bug: v8:7793
Change-Id: I7505fb111896991d16d7d113704c8c3676669f34
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2526383
Reviewed-by: Nico Hartmann <nicohartmann@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#71048}
This reverts commit 26f10ecd95.
Reason for revert: GC stress failures:
https://crbug.com/v8/11114
Original change's description:
> [torque] allow exported classes with custom C++ class
>
> Introduce a new annotation @customCppClass that can be used for
> non-extern @export classes, that is, generate everything, remove
> boilerplate from all the internal lists and switches, but allow
> a custom C++ class, which in turn also allows overwriting the generated
> print and verify functions.
>
> Port DescriptorArray and StrongDescriptorArray as an example.
>
> Bug: v8:7793
> Change-Id: I744e52fb4102ac49c0097f1c95bb17d301975bf0
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2489687
> Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
> Reviewed-by: Nico Hartmann <nicohartmann@chromium.org>
> Reviewed-by: Seth Brenith <seth.brenith@microsoft.com>
> Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#70989}
TBR=ulan@chromium.org,tebbi@chromium.org,seth.brenith@microsoft.com,nicohartmann@chromium.org
Change-Id: I4631db66a76f41cf62b400e8ee64df27e641a320
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:7793,v8:11114
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2521911
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70994}
Introduce a new annotation @customCppClass that can be used for
non-extern @export classes, that is, generate everything, remove
boilerplate from all the internal lists and switches, but allow
a custom C++ class, which in turn also allows overwriting the generated
print and verify functions.
Port DescriptorArray and StrongDescriptorArray as an example.
Bug: v8:7793
Change-Id: I744e52fb4102ac49c0097f1c95bb17d301975bf0
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2489687
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Nico Hartmann <nicohartmann@chromium.org>
Reviewed-by: Seth Brenith <seth.brenith@microsoft.com>
Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70989}
Drive-by fixes:
- Use constexpr types to determine C++ type names.
- Fix factory constructors to not skip write barriers in old generation.
Change-Id: I0ebbfd56c06ad41d02836fb48531ae7eded166bf
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2400994
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Nico Hartmann <nicohartmann@chromium.org>
Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70921}
This commit adds the 'l' (linear) RegExp flag (as in e.g. /asdf|123/l)
that forces execution in linear time. These regexps are handled by the
experimental engine. If the experimental engine cannot handle the
pattern, an exception is thrown on creation of the regexp.
The commit also adds a new global V8 flag and changes an existing one:
* --enable-experimental-engine, which turns on recognition of the RegExp
'l' flag. Previously this flag also caused all supported regexps to
be executed by the experimental engine; this is not the case anymore.
* --default-to-experimental-regexp-engine takes over the previous
semantics of --enable-experimental-regexp-engine: We execute all
supported regexps with the experimental engine.
Cq-Include-Trybots: luci.v8.try:v8_linux64_fyi_rel_ng
Bug: v8:10765
Change-Id: I5622a89b19404105e8be280d454e9fdd63c003b3
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2461244
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Reviewed-by: Simon Zünd <szuend@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Martin Bidlingmaier <mbid@google.com>
Cr-Commit-Position: refs/heads/master@{#70892}
This CL provides synchronized get/set to feedback vector slots.
The FeedbackNexus is set up to use order preserving reads when used
on the background thread, and a lock to ensure coherent read
of information for ICKinds with two slots. The main thread takes
the lock on sets.
This test provides patterns to be followed by concurrent TurboFan.
We don't yet access the FeedbackVector on the background thread.
This CL only makes it safe to do so. The next step will come when
the optimizing compiler begins to query the the vector from the
background thread. Currently, with --concurrent-inlining turned on
this is done in bytecode serialization on the main thread. Without
concurrent inlining, it's also done on the main thread, in both
cases using the FeedbackNexus.
Bug: v8:7790
Change-Id: I49d8b8031190f91a0da1c24f375b6b6d8a9fe038
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2276210
Commit-Queue: Michael Stanton <mvstanton@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Nico Hartmann <nicohartmann@chromium.org>
Reviewed-by: Santiago Aboy Solanes <solanes@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70797}
This changes OrderedHashMap, OrderedHashSet, and OrderedNameDictionary
as follows:
- Create a dedicated allocation function AllocateEmpty to create zero-
element instances of these classes
- Fix bugs resulting from using these zero-element versions
Further, this CL
- provides a canonical empty versions of OrderedNameDictionary
- changes the types of the canonical ordered hash table and hash set
from FixedArray to the actual subclasses
Bug: v8:7569
Change-Id: I0fe1215e7d164617afa777c8b3208a0857ab6edd
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2476315
Commit-Queue: Frank Emrich <emrich@google.com>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Marja Hölttä <marja@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70604}
This is the first change in the process of implementing import
assertions per https://tc39.es/proposal-import-assertions/.
This CR adds support for the empty form of the AssertClause.
Also added is a --harmony-import-assertions flag to enable/disable
import assertions. For now, the feature is off by default.
The next change will enable the parser to handle a non-empty list
of AssertEntries.
Bug: v8:10958
Change-Id: I0832d89effc27225aa4430605a51690461daf7ad
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2468623
Reviewed-by: Marja Hölttä <marja@chromium.org>
Commit-Queue: Dan Clark <daniec@microsoft.com>
Cr-Commit-Position: refs/heads/master@{#70545}
This relands commit 3f4e9bbe43.
which was a reland of c4a062a958
which was a reland of 28a30c578c
which was a reland of 5d7a29c90e
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 76ad3ab597
[identity-map] Change resize heuristic
This relands commit 77cc96aa48
[identity-map] Cache the calculated Hash
This relands commit bee5b996aa
[serializer] Remove Deserializer::Initialize
This relands commit c8f73f2266
[serializer] Cache instance type in PostProcessNewObject
This relands commit 4e7c99abda
[identity-map] Remove double-lookups in IdentityMap
Original change's description:
> Reland^3 "[serializer] Allocate during deserialization"
>
> This is a reland of c4a062a958
> which was a reland of 28a30c578c
> which was a reland of 5d7a29c90e
>
> 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 28a30c578c
> > which was a reland of 5d7a29c90e
> >
> > 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 5d7a29c90e
> > >
> > > 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: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70366}
This reverts commit 3f4e9bbe43, along
with the following dependent changes (reverted to make this a clean revert):
76ad3ab597 [identity-map] Change resize heuristic
77cc96aa48 [identity-map] Cache the calculated Hash
bee5b996aa [serializer] Remove Deserializer::Initialize
c8f73f2266 [serializer] Cache instance type in PostProcessNewObject
4e7c99abda [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 c4a062a958
> which was a reland of 28a30c578c
> which was a reland of 5d7a29c90e
>
> 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 28a30c578c
> > which was a reland of 5d7a29c90e
> >
> > 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 5d7a29c90e
> > >
> > > 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/+/2451030
Reviewed-by: Adam Klein <adamk@chromium.org>
Commit-Queue: Adam Klein <adamk@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70328}
This is a reland of c4a062a958
which was a reland of 28a30c578c
which was a reland of 5d7a29c90e
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 28a30c578c
> which was a reland of 5d7a29c90e
>
> 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 5d7a29c90e
> >
> > 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}
This reverts commit c4a062a958.
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 28a30c578c
> which was a reland of 5d7a29c90e
>
> 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 5d7a29c90e
> >
> > 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/+/2445250
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70280}
This is a reland of 28a30c578c
which was a reland of 5d7a29c90e
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 5d7a29c90e
>
> 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}
This reverts commit 28a30c578c.
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 5d7a29c90e
>
> 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/+/2441608
Reviewed-by: Zhi An Ng <zhin@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70268}
This is a reland of 5d7a29c90e
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}
This reverts commit 5d7a29c90e.
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: Leszek Swirski <leszeks@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70231}
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}
This reverts commit 1aa9ab7384.
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 76d684cc82.
> - "Reland "[serializer] Remove new space""
> commit 81231c23a9.
> - "[serializer] Clean-up and de-macro ReadDataCase"
> commit c06d24b915.
> - "[serializer] DCHECK deserializer allocations are initialized"
> commit fbc1f32d8e.
>
> 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: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69988}
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 76d684cc82.
- "Reland "[serializer] Remove new space""
commit 81231c23a9.
- "[serializer] Clean-up and de-macro ReadDataCase"
commit c06d24b915.
- "[serializer] DCHECK deserializer allocations are initialized"
commit fbc1f32d8e.
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}
This is a reland of 81577a79e1
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/+/2400992
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69779}
This reverts commit 81577a79e1.
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/+/2398535
Reviewed-by: Sathya Gunasekaran <gsathya@chromium.org>
Commit-Queue: Sathya Gunasekaran <gsathya@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69763}
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}
Change the serialization protocol to ensure that maps are serialized
before objects using them. This ensures that as soon as we allocate
space for an object, we can immediately write the object's map into that
allocation. In the future, this will allow us to make deserialized
object visible to the GC.
Specifically, this forces map serialization to happen after emitting
a kNewObject for an object, but before allocating the space for it. We
have to serialize the map after kNewObject because otherwise the map
itself would be written into the "current" slot, into which the object
is supposed to be deserialized.
Objects whose maps are currently being deserialized are considered
"pending" -- started, but not yet allocated. The map might point to a
pending object (e.g. if an object's constructor points to the object).
This is solved by introducing a new concept of forward references, where
the field referring to the pending object is serialized as a "pending
forward reference" which is "resolved" once the object is allocated.
It might also point to itself, in the case of the meta map -- this is
simply solved by introducing a new bytecode for the meta map; this
cannot be a pending forward reference because the meta map is not yet
allocated, so its map slot cannot be registered as pending.
Finally, we may need to go to a new chunk after serializing the map; so
after the map serialization, we peek to see if there's a next chunk
bytecode before the object allocation.
Bug: v8:10815
Change-Id: Ifa8f25bdaf3b15b5d990a1d2e7be677c2fa80013
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2362953
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69636}
This reached consensus in the March 2020 TC39.
https://github.com/tc39/ecma262/pull/1908
This aligns JS with wasm, which allows atomics operations on non-shared
linear memory.
Bug: v8:10687, v8:9921
Change-Id: I7b60473b271cee6bccb342e97a4fd3781aedddb4
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2330802
Commit-Queue: Shu-yu Guo <syg@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69392}
ArrayBufferTracker was superseded by ArrayBufferList and
ArrayBufferSweeper. Now that ArrayBufferSweeper is used in production,
we can remove the unused ArrayBufferTracker mechanism.
Bug: v8:10064
Change-Id: I479169c76b6c5c634672024f77e689bb64a36504
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2339105
Reviewed-by: Simon Zünd <szuend@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Commit-Queue: Dominik Inführ <dinfuehr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69351}
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: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69270}
This is a stop-gap solution (while we wait for a proper spec)
that lets managed WasmGC objects perform round-trips through
JavaScript. On the JavaScript side, they appear as empty/opaque.
Bug: v8:7748
Change-Id: I0dd368bc14d622f3ef41871484228267359e9b5b
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2316306
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69207}
https://tc39.es/proposal-intl-segmenter/
TC39 passed Intl.Segmenter to stage 3 in Jul 21.
This CL move our earlier prototype to the current spec.
Bug: v8:6891
Change-Id: I07234beed54f671c26bdbfb3983c5bc2fa5a29b0
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2219413
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Reviewed-by: Frank Tang <ftang@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Frank Tang <ftang@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69080}
Instead allocating the bitmap with malloc, we now reserve a block
at the start of the memory chunk. This CL is a partial revert of
https://chromium-review.googlesource.com/c/v8/v8/+/1254125
Additionally it refactors field offset computation and moves them
to MemoryChunkLayout.
Having the bitmap in the memory chunk simplifies sharing of RO pages
and also solves the malloc fragmentation issues.
Bug: chromium:1073140
Change-Id: Ibc04f48921fc9496370858ce4c25c56b31c93c89
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2289979
Commit-Queue: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Dan Elphick <delphick@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68783}
By introducing a globally known map for each generic type.
These maps are never used to allocate objects, they only
serve as sentinels for generic heap types.
Bug: v8:7748
Change-Id: I950a8c712dc1510759a833fe9122b9e9a6222dc2
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2288860
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68755}
Relanding without changes, revert reason was fixed by:
https://chromium-review.googlesource.com/c/v8/v8/+/2272564
Originally reviewed at:
https://chromium-review.googlesource.com/c/v8/v8/+/2260566
Original description:
RTTs are internally represented as Maps. To store supertype information,
this patch introduces a WasmTypeInfo object, which is installed on Wasm
objects' Maps and points at both the off-heap type information and the
parent RTT.
In this patch, rtt.sub always creates a fresh RTT. The canonicalization
that the proposal requires will be implemented later.
Bug: v8:7748
Change-Id: I7fd4986efa3153ac68037ec418ea617f3f7636e8
Tbr: ulan@chromium.org
Tbr: tebbi@chromium.org
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2273123
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68581}
This reverts commit 04ce88eae5.
Reason for revert: TSAN failure: https://ci.chromium.org/p/v8/builders/ci/V8%20Linux64%20TSAN/32135
Original change's description:
> [wasm-gc] Implement rtt.sub
>
> RTTs are internally represented as Maps. To store supertype information,
> this patch introduces a WasmTypeInfo object, which is installed on Wasm
> objects' Maps and points at both the off-heap type information and the
> parent RTT.
> In this patch, rtt.sub always creates a fresh RTT. The canonicalization
> that the proposal requires will be implemented later.
>
> Bug: v8:7748
> Change-Id: I8286dd11f520966155cd95c2bd844ec34fccd131
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2260566
> Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
> Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
> Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#68564}
TBR=ulan@chromium.org,jkummerow@chromium.org,tebbi@chromium.org
Change-Id: I311732e1ced4de7a58b87d4a9b6056e0d62aa986
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:7748
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2270734
Reviewed-by: Shu-yu Guo <syg@chromium.org>
Commit-Queue: Shu-yu Guo <syg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68567}
RTTs are internally represented as Maps. To store supertype information,
this patch introduces a WasmTypeInfo object, which is installed on Wasm
objects' Maps and points at both the off-heap type information and the
parent RTT.
In this patch, rtt.sub always creates a fresh RTT. The canonicalization
that the proposal requires will be implemented later.
Bug: v8:7748
Change-Id: I8286dd11f520966155cd95c2bd844ec34fccd131
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2260566
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68564}
This change enables automatic generation of Cast<> operators for
classes that are defined in Torque.
* Cast<> macros are generated for all classes that are defined in
Torque code that are neither shapes nor marked with a new
@doNotGenerateCast annotation.
* Implicitly generated Cast macros simply call through to an
internally-defined "DownCastForTorqueClass" macro that implements
the cast using one of three strategies for efficiency. If the class
has subclasses (i.e. a range of instance types including subtypes),
the DownCastForTorqueClass checks for inclusion in the instance type
range. If the class has a single instance type (i.e. no subclasses),
then either 1) a map check is used if the class has a globally-
defined map constant or 2) an equality check for the instance type
is used.
* Added new intrinsics to introspect class information, e.g. fetching
instance type ranges for a class, accessing the globally-defined map
for a class.
* Removed a whole pile of existing explicit Cast<> operators that are
no longer needed because of the implicitly generated Cast<> macros.
* Added tests for the new Cast<> implementations.
Bug: v8:7793
Change-Id: I3aadb0c62b720e9de4e7978b9ec4f05075771b8b
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2250239
Commit-Queue: Daniel Clifford <danno@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68478}
The interpreter is only used for testing, and is now instantiated and
invoked directly instead of via the {WasmDebugInfo}, holding the
{InterpreterHandle}.
This CL removes both classes.
R=ahaas@chromium.org
Bug: v8:10389
Change-Id: Iede3feea413decae1edc28146b871a819e204768
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2237132
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#68271}
Currently, if d8 is run with the --turbo-profiling flag, it prints info
about every TurboFan-compiled function. This info includes the number of
times that each basic block in the function was run. It also includes
text representations of the function's schedule and code, so that the
person reading the output can associate counters with blocks of code.
The data about each function is currently stored in a
BasicBlockProfiler::Data instance, which is attached to a list owned by
the singleton BasicBlockProfiler. Each Data contains an
std::vector<uint32_t> which represents how many times each block in the
function has executed. The generated code for each block uses a raw
pointer into the storage of that vector to implement incrementing the
counter.
With this change, if you compile with v8_enable_builtins_profiling and
then run with --turbo-profiling, d8 will print that same info about
builtins too.
In order to generate code that can survive being serialized to a
snapshot and reloaded, this change uses counters in the JS heap instead
of a std::vector outside the JS heap. The steps for instrumentation are
as follows:
1. Between scheduling and instruction selection, add code to increment
the counter for each block. The counters array doesn't yet exist at
this point, and allocation is disallowed, so at this point the code
refers to a special marker value.
2. During finalization of the code, allocate a BasicBlockProfilingData
object on the JS heap containing data equivalent to what is stored in
BasicBlockProfiler::Data. This includes a ByteArray that is big
enough to store the counters for each block.
3. Patch the reference in the BuiltinsConstantsTableBuilder so that
instead of referring to the marker object, it now refers to this
ByteArray. Also add the BasicBlockProfilingData object to a list that
is attached to the heap roots so it can be easily accessed for
printing.
Because these steps include modifying the BuiltinsConstantsTableBuilder,
this procedure is only applicable to builtins. Runtime-generated code
still uses raw pointers into std::vector instances. In order to keep
divergence between these code paths to a minimum, most work is done
referring to instances of BasicBlockProfiler::Data (the C++ class), and
functions are provided to copy back and forth between that type and
BasicBlockProfilingData (the JS heap object).
This change is intended only to make --turbo-profiling work consistently
on more kinds of functions, but with some further work, this data could
form the basis for:
- code coverage info for fuzzers, and/or
- hot-path info for profile-guided optimization.
Bug: v8:10470, v8:9119
Change-Id: Ib556a5bc3abe67cdaa2e3ee62702a2a08b11cb61
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2159738
Commit-Queue: Seth Brenith <seth.brenith@microsoft.com>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#67944}
Frank Tang
Manos Koukoutos
Gus Caplan
Tobias Tebbi
Jakob Kummerow
Daniel Clark
Michael Achenbach
Martin Bidlingmaier
Mike Stanton
Frank Emrich
Leszek Swirski
Adam Klein
Clemens Backes
Zhi An Ng
Sathya Gunasekaran
Gus Caplan
Shu-yu Guo
Dominik Inführ
Marja Hölttä
Ulan Degenbaev
Daniel Clifford
Seth Brenith