This is a reland of fbfa9bf4ec
The arm64 was missing proper codegen for CFI, thus sizes were off.
Original change's description:
> Reland "[deoptimizer] Change deopt entries into builtins"
>
> This is a reland of 7f58ced72e
>
> It fixes the different exit size emitted on x64/Atom CPUs due to
> performance tuning in TurboAssembler::Call. Additionally, add
> cctests to verify the fixed size exits.
>
> Original change's description:
> > [deoptimizer] Change deopt entries into builtins
> >
> > While the overall goal of this commit is to change deoptimization
> > entries into builtins, there are multiple related things happening:
> >
> > - Deoptimization entries, formerly stubs (i.e. Code objects generated
> > at runtime, guaranteed to be immovable), have been converted into
> > builtins. The major restriction is that we now need to preserve the
> > kRootRegister, which was formerly used on most architectures to pass
> > the deoptimization id. The solution differs based on platform.
> > - Renamed DEOPT_ENTRIES_OR_FOR_TESTING code kind to FOR_TESTING.
> > - Removed heap/ support for immovable Code generation.
> > - Removed the DeserializerData class (no longer needed).
> > - arm64: to preserve 4-byte deopt exits, introduced a new optimization
> > in which the final jump to the deoptimization entry is generated
> > once per Code object, and deopt exits can continue to emit a
> > near-call.
> > - arm,ia32,x64: change to fixed-size deopt exits. This reduces exit
> > sizes by 4/8, 5, and 5 bytes, respectively.
> >
> > On arm the deopt exit size is reduced from 12 (or 16) bytes to 8 bytes
> > by using the same strategy as on arm64 (recalc deopt id from return
> > address). Before:
> >
> > e300a002 movw r10, <id>
> > e59fc024 ldr ip, [pc, <entry offset>]
> > e12fff3c blx ip
> >
> > After:
> >
> > e59acb35 ldr ip, [r10, <entry offset>]
> > e12fff3c blx ip
> >
> > On arm64 the deopt exit size remains 4 bytes (or 8 bytes in same cases
> > with CFI). Additionally, up to 4 builtin jumps are emitted per Code
> > object (max 32 bytes added overhead per Code object). Before:
> >
> > 9401cdae bl <entry offset>
> >
> > After:
> >
> > # eager deoptimization entry jump.
> > f95b1f50 ldr x16, [x26, <eager entry offset>]
> > d61f0200 br x16
> > # lazy deoptimization entry jump.
> > f95b2b50 ldr x16, [x26, <lazy entry offset>]
> > d61f0200 br x16
> > # the deopt exit.
> > 97fffffc bl <eager deoptimization entry jump offset>
> >
> > On ia32 the deopt exit size is reduced from 10 to 5 bytes. Before:
> >
> > bb00000000 mov ebx,<id>
> > e825f5372b call <entry>
> >
> > After:
> >
> > e8ea2256ba call <entry>
> >
> > On x64 the deopt exit size is reduced from 12 to 7 bytes. Before:
> >
> > 49c7c511000000 REX.W movq r13,<id>
> > e8ea2f0700 call <entry>
> >
> > After:
> >
> > 41ff9560360000 call [r13+<entry offset>]
> >
> > Bug: v8:8661,v8:8768
> > Change-Id: I13e30aedc360474dc818fecc528ce87c3bfeed42
> > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2465834
> > Commit-Queue: Jakob Gruber <jgruber@chromium.org>
> > Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
> > Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
> > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
> > Cr-Commit-Position: refs/heads/master@{#70597}
>
> Tbr: ulan@chromium.org, tebbi@chromium.org, rmcilroy@chromium.org
> Bug: v8:8661,v8:8768,chromium:1140165
> Change-Id: Ibcd5c39c58a70bf2b2ac221aa375fc68d495e144
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2485506
> Reviewed-by: Jakob Gruber <jgruber@chromium.org>
> Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
> Commit-Queue: Jakob Gruber <jgruber@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#70655}
Tbr: ulan@chromium.org, tebbi@chromium.org, rmcilroy@chromium.org
Bug: v8:8661
Bug: v8:8768
Bug: chromium:1140165
Change-Id: I471cc94fc085e527dc9bfb5a84b96bd907c2333f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2488682
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70672}
This is a reland of 7f58ced72e
It fixes the different exit size emitted on x64/Atom CPUs due to
performance tuning in TurboAssembler::Call. Additionally, add
cctests to verify the fixed size exits.
Original change's description:
> [deoptimizer] Change deopt entries into builtins
>
> While the overall goal of this commit is to change deoptimization
> entries into builtins, there are multiple related things happening:
>
> - Deoptimization entries, formerly stubs (i.e. Code objects generated
> at runtime, guaranteed to be immovable), have been converted into
> builtins. The major restriction is that we now need to preserve the
> kRootRegister, which was formerly used on most architectures to pass
> the deoptimization id. The solution differs based on platform.
> - Renamed DEOPT_ENTRIES_OR_FOR_TESTING code kind to FOR_TESTING.
> - Removed heap/ support for immovable Code generation.
> - Removed the DeserializerData class (no longer needed).
> - arm64: to preserve 4-byte deopt exits, introduced a new optimization
> in which the final jump to the deoptimization entry is generated
> once per Code object, and deopt exits can continue to emit a
> near-call.
> - arm,ia32,x64: change to fixed-size deopt exits. This reduces exit
> sizes by 4/8, 5, and 5 bytes, respectively.
>
> On arm the deopt exit size is reduced from 12 (or 16) bytes to 8 bytes
> by using the same strategy as on arm64 (recalc deopt id from return
> address). Before:
>
> e300a002 movw r10, <id>
> e59fc024 ldr ip, [pc, <entry offset>]
> e12fff3c blx ip
>
> After:
>
> e59acb35 ldr ip, [r10, <entry offset>]
> e12fff3c blx ip
>
> On arm64 the deopt exit size remains 4 bytes (or 8 bytes in same cases
> with CFI). Additionally, up to 4 builtin jumps are emitted per Code
> object (max 32 bytes added overhead per Code object). Before:
>
> 9401cdae bl <entry offset>
>
> After:
>
> # eager deoptimization entry jump.
> f95b1f50 ldr x16, [x26, <eager entry offset>]
> d61f0200 br x16
> # lazy deoptimization entry jump.
> f95b2b50 ldr x16, [x26, <lazy entry offset>]
> d61f0200 br x16
> # the deopt exit.
> 97fffffc bl <eager deoptimization entry jump offset>
>
> On ia32 the deopt exit size is reduced from 10 to 5 bytes. Before:
>
> bb00000000 mov ebx,<id>
> e825f5372b call <entry>
>
> After:
>
> e8ea2256ba call <entry>
>
> On x64 the deopt exit size is reduced from 12 to 7 bytes. Before:
>
> 49c7c511000000 REX.W movq r13,<id>
> e8ea2f0700 call <entry>
>
> After:
>
> 41ff9560360000 call [r13+<entry offset>]
>
> Bug: v8:8661,v8:8768
> Change-Id: I13e30aedc360474dc818fecc528ce87c3bfeed42
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2465834
> Commit-Queue: Jakob Gruber <jgruber@chromium.org>
> Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
> Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
> Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#70597}
Tbr: ulan@chromium.org, tebbi@chromium.org, rmcilroy@chromium.org
Bug: v8:8661,v8:8768,chromium:1140165
Change-Id: Ibcd5c39c58a70bf2b2ac221aa375fc68d495e144
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2485506
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70655}
This reverts commit 7f58ced72e.
Reason for revert: Segfaults on Atom_x64 https://ci.chromium.org/p/v8-internal/builders/ci/v8_linux64_atom_perf/5686?
Original change's description:
> [deoptimizer] Change deopt entries into builtins
>
> While the overall goal of this commit is to change deoptimization
> entries into builtins, there are multiple related things happening:
>
> - Deoptimization entries, formerly stubs (i.e. Code objects generated
> at runtime, guaranteed to be immovable), have been converted into
> builtins. The major restriction is that we now need to preserve the
> kRootRegister, which was formerly used on most architectures to pass
> the deoptimization id. The solution differs based on platform.
> - Renamed DEOPT_ENTRIES_OR_FOR_TESTING code kind to FOR_TESTING.
> - Removed heap/ support for immovable Code generation.
> - Removed the DeserializerData class (no longer needed).
> - arm64: to preserve 4-byte deopt exits, introduced a new optimization
> in which the final jump to the deoptimization entry is generated
> once per Code object, and deopt exits can continue to emit a
> near-call.
> - arm,ia32,x64: change to fixed-size deopt exits. This reduces exit
> sizes by 4/8, 5, and 5 bytes, respectively.
>
> On arm the deopt exit size is reduced from 12 (or 16) bytes to 8 bytes
> by using the same strategy as on arm64 (recalc deopt id from return
> address). Before:
>
> e300a002 movw r10, <id>
> e59fc024 ldr ip, [pc, <entry offset>]
> e12fff3c blx ip
>
> After:
>
> e59acb35 ldr ip, [r10, <entry offset>]
> e12fff3c blx ip
>
> On arm64 the deopt exit size remains 4 bytes (or 8 bytes in same cases
> with CFI). Additionally, up to 4 builtin jumps are emitted per Code
> object (max 32 bytes added overhead per Code object). Before:
>
> 9401cdae bl <entry offset>
>
> After:
>
> # eager deoptimization entry jump.
> f95b1f50 ldr x16, [x26, <eager entry offset>]
> d61f0200 br x16
> # lazy deoptimization entry jump.
> f95b2b50 ldr x16, [x26, <lazy entry offset>]
> d61f0200 br x16
> # the deopt exit.
> 97fffffc bl <eager deoptimization entry jump offset>
>
> On ia32 the deopt exit size is reduced from 10 to 5 bytes. Before:
>
> bb00000000 mov ebx,<id>
> e825f5372b call <entry>
>
> After:
>
> e8ea2256ba call <entry>
>
> On x64 the deopt exit size is reduced from 12 to 7 bytes. Before:
>
> 49c7c511000000 REX.W movq r13,<id>
> e8ea2f0700 call <entry>
>
> After:
>
> 41ff9560360000 call [r13+<entry offset>]
>
> Bug: v8:8661,v8:8768
> Change-Id: I13e30aedc360474dc818fecc528ce87c3bfeed42
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2465834
> Commit-Queue: Jakob Gruber <jgruber@chromium.org>
> Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
> Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
> Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#70597}
TBR=ulan@chromium.org,rmcilroy@chromium.org,jgruber@chromium.org,tebbi@chromium.org
# Not skipping CQ checks because original CL landed > 1 day ago.
Bug: v8:8661,v8:8768,chromium:1140165
Change-Id: I3df02ab42f6e02233d9f6fb80e8bb18f76870d91
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2485504
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70649}
While the overall goal of this commit is to change deoptimization
entries into builtins, there are multiple related things happening:
- Deoptimization entries, formerly stubs (i.e. Code objects generated
at runtime, guaranteed to be immovable), have been converted into
builtins. The major restriction is that we now need to preserve the
kRootRegister, which was formerly used on most architectures to pass
the deoptimization id. The solution differs based on platform.
- Renamed DEOPT_ENTRIES_OR_FOR_TESTING code kind to FOR_TESTING.
- Removed heap/ support for immovable Code generation.
- Removed the DeserializerData class (no longer needed).
- arm64: to preserve 4-byte deopt exits, introduced a new optimization
in which the final jump to the deoptimization entry is generated
once per Code object, and deopt exits can continue to emit a
near-call.
- arm,ia32,x64: change to fixed-size deopt exits. This reduces exit
sizes by 4/8, 5, and 5 bytes, respectively.
On arm the deopt exit size is reduced from 12 (or 16) bytes to 8 bytes
by using the same strategy as on arm64 (recalc deopt id from return
address). Before:
e300a002 movw r10, <id>
e59fc024 ldr ip, [pc, <entry offset>]
e12fff3c blx ip
After:
e59acb35 ldr ip, [r10, <entry offset>]
e12fff3c blx ip
On arm64 the deopt exit size remains 4 bytes (or 8 bytes in same cases
with CFI). Additionally, up to 4 builtin jumps are emitted per Code
object (max 32 bytes added overhead per Code object). Before:
9401cdae bl <entry offset>
After:
# eager deoptimization entry jump.
f95b1f50 ldr x16, [x26, <eager entry offset>]
d61f0200 br x16
# lazy deoptimization entry jump.
f95b2b50 ldr x16, [x26, <lazy entry offset>]
d61f0200 br x16
# the deopt exit.
97fffffc bl <eager deoptimization entry jump offset>
On ia32 the deopt exit size is reduced from 10 to 5 bytes. Before:
bb00000000 mov ebx,<id>
e825f5372b call <entry>
After:
e8ea2256ba call <entry>
On x64 the deopt exit size is reduced from 12 to 7 bytes. Before:
49c7c511000000 REX.W movq r13,<id>
e8ea2f0700 call <entry>
After:
41ff9560360000 call [r13+<entry offset>]
Bug: v8:8661,v8:8768
Change-Id: I13e30aedc360474dc818fecc528ce87c3bfeed42
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2465834
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70597}
CodeKind::OPTIMIZED_CODE -> TURBOFAN
Kinds are now more fine-grained and distinguish between TF, TP, NCI.
CodeKind::STUB -> DEOPT_ENTRIES_OR_FOR_TESTING
Code stubs (like builtins, but generated at runtime) were removed from
the codebase years ago, this is the last remnant. This kind is used
only for deopt entries (which should be converted into builtins) and
for tests.
Change-Id: I67beb15377cb60f395e9b051b25f3e5764982e93
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2440335
Auto-Submit: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Mythri Alle <mythria@chromium.org>
Reviewed-by: Mythri Alle <mythria@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70234}
With the new Turbofan variants (NCI and Turboprop), we need a way to
distinguish between them both during and after compilation. We
initially introduced CompilationTarget to track the variant during
compilation, but decided to reuse the code kind as the canonical spot to
store this information instead.
Why? Because it is an established mechanism, already available in most
of the necessary spots (inside the pipeline, on Code objects, in
profiling traces).
This CL removes CompilationTarget and adds a new
NATIVE_CONTEXT_INDEPENDENT kind, plus helper functions to determine
various things about a given code kind (e.g.: does this code kind
deopt?).
As a (very large) drive-by, refactor both Code::Kind and
AbstractCode::Kind into a new CodeKind enum class.
Bug: v8:8888
Change-Id: Ie858b9a53311b0731630be35cf5cd108dee95b39
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2336793
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#69244}
In preparation for adding reference types, which need an additional
parameter to indicate the referenced type.
Bug: v8:7748
Change-Id: If4023f3d9c7f42ed603b69c43356d2e8b81a0daa
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2091471
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#66687}
This is a reland of c509bb8c55
Original change's description:
> Cache native modules in the wasm engine by their wire bytes. This is to
> prepare for sharing {Script} objects between multiple {WasmModuleObject}
> created from the same bytes. This also saves unnecessary compilation
> time and memory.
>
> R=clemensb@chromium.org
>
> Bug: v8:6847
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1916603
> Commit-Queue: Thibaud Michaud <thibaudm@chromium.org>
> Reviewed-by: Clemens Backes <clemensb@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#65296}
R=clemensb@chromium.org
Bug: v8:6847
Change-Id: I8839c9ec96dc4141cf3c30916a62ccf86f5463ff
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1960287
Commit-Queue: Thibaud Michaud <thibaudm@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65500}
This reverts commit c509bb8c55.
Reason for revert: Breaks arm64 - sim - MSAN, see https://ci.chromium.org/p/v8/builders/ci/V8%20Linux%20-%20arm64%20-%20sim%20-%20MSAN/30050
Original change's description:
> [wasm] Share native modules compiled from the same bytes
>
> Cache native modules in the wasm engine by their wire bytes. This is to
> prepare for sharing {Script} objects between multiple {WasmModuleObject}
> created from the same bytes. This also saves unnecessary compilation
> time and memory.
>
> R=clemensb@chromium.org
>
> Bug: v8:6847
> Change-Id: Iad5f70efbfe3f0f134dcb851edbcec50691677e0
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1916603
> Commit-Queue: Thibaud Michaud <thibaudm@chromium.org>
> Reviewed-by: Clemens Backes <clemensb@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#65296}
TBR=clemensb@chromium.org,thibaudm@chromium.org
Change-Id: I908b0f59bce26678d0b5d7fddc986384c40b4709
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6847
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1946334
Reviewed-by: Maya Lekova <mslekova@chromium.org>
Commit-Queue: Maya Lekova <mslekova@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65297}
Cache native modules in the wasm engine by their wire bytes. This is to
prepare for sharing {Script} objects between multiple {WasmModuleObject}
created from the same bytes. This also saves unnecessary compilation
time and memory.
R=clemensb@chromium.org
Bug: v8:6847
Change-Id: Iad5f70efbfe3f0f134dcb851edbcec50691677e0
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1916603
Commit-Queue: Thibaud Michaud <thibaudm@chromium.org>
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65296}
Make WasmFeatures a proper class which uses an EnumSet under the hood.
This way, it inherits all behaviour of EnumSet like comparison, merge,
etc.
Accesses change from being simple field access into the struct to
actually bit tests in the EnumSet.
R=mstarzinger@chromium.org
Bug: v8:10019
Change-Id: I768f92b90ac0294156f4482defba5ce00bc70165
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1934334
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65184}
This makes the {code_size_estimate} computation explicit in the caller,
and removes one of the two {NewNativeModule} constructors. It turns out
that the calculation is totally off in the streaming calculation phase,
since no function bodies have been parsed yet. So all
{WasmFunction::code} fields are still empty, and we compute an estimate
that is way too low.
This CL prepares the actual fix for that (by computing a better estimate
at specific call sites).
R=ahaas@chromium.org
Bug: v8:9950
Change-Id: I68a891c97e5f65a9c7e73e21684bdfa7e261e216
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1901273
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#64845}
This allows to return bool values from Torque macros and branch on them
without performance penalty, reconstructing good control flow.
Drive-by cleanup: Delete EnsureDeferredCodeSingleEntryPoint(), since
it's no longer needed. Constructing a graph and then re-inferring
deferred blocks based on branch hints achieves this effect
automatically.
Bug: v8:7793
Change-Id: Idb6802372b407549e4760f290933d5b8f1e9d952
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1681132
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62979}
Cpplint usually checks for non-const reference arguments. They are
forbidden in the style guide, and v8 does not explicitly make an
exception here.
This CL re-enables that warning, and fixes all current violations by
adding an explicit "NOLINT(runtime/references)" comment. In follow-up
CLs, we should aim to remove as many of them as possible.
TBR=mlippautz@chromium.org
Bug: v8:9429
Change-Id: If7054d0b366138b731972ed5d4e304b5ac8423bb
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1687891
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62551}
v8memory.h does not have V8 specific definitions, and having it in base
makes it clear that every component may include the file. It also
ensures that including it does not create spurious dependencies on
v8_base.
Change-Id: I565f63b25f33a9ada19d7b2ac5990863ab17f4a7
Bug: v8:9183, v8:8855
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1657923
Commit-Queue: Sigurd Schneider <sigurds@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#62309}
Our {Vector} template provides both {start} and {begin} methods. They
return exactly the same value. Since the {begin} method is needed for
iteration, and is also what standard containers provide, this CL
switches all uses of the {start} method to use {begin} instead.
Patchset 1 was auto-generated by using this clang AST matcher:
callExpr(
callee(
cxxMethodDecl(
hasName("start"),
ofClass(hasName("v8::internal::Vector")))
),
argumentCountIs(0))
Patchset 2 was created by running clang-format. Patchset 3 then
removes the now unused {Vector::start} method.
R=jkummerow@chromium.orgTBR=mstarzinger@chromium.org,yangguo@chromium.org,verwaest@chromium.org
Bug: v8:9183
Change-Id: Id9f01c92870872556e2bb3f6d5667463b0e3e5c6
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1587381
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61081}
This CL adds all the necessary {WasmCodeRefScope}s in the code base, or
at least a good approximation. A follow-up CL will enable a check that
a {WasmCodeRefScope} exists whenever a pointer to a {WasmCode} object
is returned from the {NativeModule}. This should flush out any missing
scopes.
R=titzer@chromium.org
Bug: v8:8217
Change-Id: I54c7eb39aeb1acde38273c399396e6b1390a4cb2
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1533860
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Ben Titzer <titzer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#60566}
We need to ensure that the NativeModule stays alive while any
{BackgroundCompileScope} exists, because during that time we hold
shared ownership of the mutex in the {BackgroundCompileToken}. If the
{NativeModule} dies during that period, we would need to get exclusive
ownership of the mutex and deadlock.
This change requires holding a {std::weak_ptr<NativeModule>} in the
BackgroundCompileToken instead of a raw pointer, hence it can only be
initialized after the NativeModule was created. This is done via a
separate {InitCompilationState} method.
R=ahaas@chromium.org
Bug: v8:8979
Change-Id: Ia14bd272ea0bc47aec547024da6020608418c9d2
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1518178
Auto-Submit: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#60203}
The WasmCodeManager held a list of all Isolates that use the
WasmEngine/WasmCodeManager (those two are 1:1).
Since we want to move all isolate-specific tasks (like code logging and
compilation callbacks) to the WasmEngine, this CL moves this management
from the WasmCodeManager to the WasmEngine. We now have a bidirectional
mapping from NativeModules to the Isolates that use them, and from an
Isolate to all the NativeModules it uses (n:n).
The IsolateData struct will be extended in follow-up CLs to hold things
like the ForegroundTaskRunner. The Isolate* in the NativeModule /
CompilationState will eventually be removed.
R=mstarzinger@chromium.org
Bug: v8:8689
Change-Id: Ic2c003c3949f73ce3264dd9dac96884a5c0b9896
Reviewed-on: https://chromium-review.googlesource.com/c/1433793
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59092}
This CL splits the backend of TurboFan off into its own directory,
without changing namespaces. This makes ownership management a bit
more fine-grained with a logical separation.
R=mstarzinger@chromium.org,jarin@chromium.org,adamk@chromium.org
Change-Id: I2ac40d6ca2c4f04b8474b630aae0286ecf79ef42
Reviewed-on: https://chromium-review.googlesource.com/c/1308333
Commit-Queue: Ben Titzer <titzer@chromium.org>
Reviewed-by: Adam Klein <adamk@chromium.org>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#57437}
Instead, create it when needed and pass it down to the actual
compilation.
This saves memory by making the WasmCompilationUnit smaller and will
eventually allow us to implement the trap handler fallback correctly by
using an updated ModuleEnv in background compilation and tier up.
R=mstarzinger@chromium.org
Bug: v8:5277, v8:8343
Change-Id: I0dc3a37fb88e54eb4822dc99d58ff024f4b2a367
Reviewed-on: https://chromium-review.googlesource.com/c/1293953
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#56896}
This also makes the {AddCodeCopy} method more specific to only apply to
import wrappers, otherwise the use of {set_code} would be unprotected.
R=clemensh@chromium.org
BUG=v8:8015
Change-Id: I62561560f57e4cc235a338c0e769e50ff55ec42d
Reviewed-on: https://chromium-review.googlesource.com/1238477
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#56137}
This CL introduces a set of configuration options implemented as
a struct of booleans that together comprise the set of enabled
or detected features. The configuration options replace command-line
flags that were checked deep in the implementation. As such, it is
necessary to plumb them through multiple levels of abstraction.
R=ahaas@chromium.orgCC=mstarzinger@chromium.org
BUG=chromium:868844
Cq-Include-Trybots: luci.chromium.try:linux_chromium_rel_ng
Change-Id: I1b82f5826e4fd263f68e8cafcd923bac5818a637
Reviewed-on: https://chromium-review.googlesource.com/1163670
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Ben Titzer <titzer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#55018}
Stack parameters on arm64 require padding. Since the stack areas for
parameters and returns should not overlap, we have to pad the parameters
already during the construction of the CallDescriptor so that we can set
the correct stack offset for returns.
R=mstarzinger@chromium.org
Bug: chromium:838098
Change-Id: I23389dc35037054b750e61ea6b1bfdfc4c5bc868
Reviewed-on: https://chromium-review.googlesource.com/1150178
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#54716}
The code under test is handling of multi-return values in TurboFan and
hence actually independent of WebAssembly. The only reason to generate
WasmCode is in order to use the WebAseembly linkages. This changes the
generated code to have {STUB} kind instead of {WASM_FUNCTION} kind to
avoid having stack checks in the generated code which would require a
proper WasmInstanceObject to be allocated.
R=ahaas@chromium.org
BUG=chromium:862508
Change-Id: I4feb7bff1a42bbf59cfc5f249f2e0585ce7011ad
Reviewed-on: https://chromium-review.googlesource.com/1136438
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#54436}
This CL surfaces AssemblerOptions to CodeAssembler::GenerateCode and
to pipeline methods. To allow forward declaring AssemblerOptions,
AssemblerBase::Options was moved out of the AssemblerBase class.
Bug: v8:6666
Change-Id: If9fc50d3d4767bb5dd39a0c3b6e094021f4cae2b
Reviewed-on: https://chromium-review.googlesource.com/1127039
Commit-Queue: Sigurd Schneider <sigurds@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#54286}
Instead of storing both the {NativeModule} and the {WasmModule} in a
{Managed} object, just store the {WasmModule} in the {NativeModule}
directly. This fixes crashes that happen if the {Managed<WasmModule>}
dies before the {Managed<NativeModule>}.
R=mstarzinger@chromium.org
Bug: chromium:854794, v8:7879, v8:7889
Change-Id: I6b11729943fe7a03d225138782655ee5dafd26a6
Reviewed-on: https://chromium-review.googlesource.com/1118171
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#54090}
TurboFan returned null handles if compilation did not succeed. This CL
changes that to a MaybeHandle to make it explicit that client code needs
to handle the error.
Bug: v8:7856
Change-Id: I6087e6263faa1150b9788213dd22c398b4a2fc2d
Reviewed-on: https://chromium-review.googlesource.com/1104688
Commit-Queue: Sigurd Schneider <sigurds@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53824}
The isolate is mainly used for accounting purposes. As such, it
doesn't need a field in the WasmCodeManager, and cannot have one
if it is to be made isolate independent. Instead, pass the isolate
explicitly in the appropriate cases.
R=mstarzinger@chromium.org
BUG=v8:7424
Change-Id: I539c2b33692e57605a280530bd704ef25269ad0f
Reviewed-on: https://chromium-review.googlesource.com/1073412
Commit-Queue: Ben Titzer <titzer@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#53676}
This includes the following changes:
- Limit code space to 128 MB.
- Use direct branches wherever possible.
- Where not possible, continue using load literal followed by an indirect
branch.
- Sort RelocInfo by target_address_address for the serializer, since mixing
load literal instructions and branch instructions messes up that order.
- Ensure we always wipe out targets in the serializer (not just for the
snapshot) in order to be able to distinguish between constant pool entries
and branch instructions.
Change-Id: I1a1029ce2a5f72a3a94802daf267d14a42c7c790
Reviewed-on: https://chromium-review.googlesource.com/939175
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Yang Guo <yangguo@chromium.org>
Commit-Queue: Georgia Kouveli <georgia.kouveli@arm.com>
Cr-Commit-Position: refs/heads/master@{#52885}
Linkage-related methods were declared in wasm-compiler.h and
implemented in wasm-linkage.cc. This required all users of e.g. wasm
call descriptors to include the whole wasm compiler header. Also, some
wasm linkage information is independent of turbofan and also used
outside of the compiler directory.
This CL splits off wasm-linkage.h (with minimal includes) and puts it
in src/wasm. This allows to use that information without including
compiler headers (will clean up several uses in follow-up CLs).
R=mstarzinger@chromium.org, titzer@chromium.org
Bug: v8:7570
Change-Id: Ifcae70b4ea7932cda30953b325c2b87c4176c598
Reviewed-on: https://chromium-review.googlesource.com/1013701
Reviewed-by: Ben Titzer <titzer@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#52849}
Wasm tier-up first compiles the whole module using Liftoff, and then
using Turbofan. The idea is to achieve fast start-up times by first
running Liftoff-compiled code. In the meantime we finish compilation
with Turbofan, and replace the Liftoff-compiled code as soon
as Turbofan finished compilation, thus achieving high performance.
Tier-up is enabled through the flag FLAG_wasm_tier_up.
Bug: v8:6600
Change-Id: I70552969c53d909a591666a1e7ce1ee1419b2f34
Reviewed-on: https://chromium-review.googlesource.com/1010422
Commit-Queue: Kim-Anh Tran <kimanh@google.com>
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#52759}
First version which can compile a very basic code.
Change-Id: I3b98412a5ca39a28f8fe5b60516b82c6981dd187
Reviewed-on: https://chromium-review.googlesource.com/993232
Commit-Queue: Vincent Belliard <vincent.belliard@arm.com>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#52622}
The "Address" type is V8's general-purpose type for manipulating memory
addresses. Per the C++ spec, pointer arithmetic and pointer comparisons
are undefined behavior except within the same array; since we generally
don't operate within a C++ array, our general-purpose type shouldn't be
a pointer type.
Bug: v8:3770
Cq-Include-Trybots: luci.chromium.try:linux_chromium_rel_ng;master.tryserver.blink:linux_trusty_blink_rel
Change-Id: Ib96016c24a0f18bcdba916dabd83e3f24a1b5779
Reviewed-on: https://chromium-review.googlesource.com/988657
Commit-Queue: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Cr-Commit-Position: refs/heads/master@{#52601}
In order to clarify the difference between, e.g., InstructionStart and
instruction_start, rename as follows:
Code::instruction_start -> raw_instruction_start
Code::instruction_end -> raw_instruction_end
Code::instruction_size -> raw_instruction_size
The difference between the camel-case and raw_* function families is
in how they handle off-heap-trampoline Code objects. For example, when
called on an off-heap-trampoline: raw_instruction_start returns the
trampoline's entry point, while InstructionStart returns the off-heap
code's entry point (located in the .text section of the binary).
Some callsites were updated to call the camel-case function family as
appropriate.
Bug: v8:6666
Change-Id: I4a572f47c2d161a853599d7c17879e263b0d1a87
Reviewed-on: https://chromium-review.googlesource.com/997532
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#52387}
With the Ignition + Turbofan pipeline there is very little overlap between the data
needed for unoptimized compilation and optimized compilation. As a result, it is
cleaner to split up the CompilationInfo into UnoptimizedCompilationInfo and
OptimizedCompilationInfo.
Doing so also necessitate splitting up CompilationJob into UnoptimizedCompilationJob
and OptimizedCompilationJob - again there is not much overlap so this seems cleaner.
Change-Id: I1056ad520937b7f8582e4fc3ca8f4910742de30a
Reviewed-on: https://chromium-review.googlesource.com/995895
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#52369}
This moves source position tables associated with WasmCode objects to be
located outside the garbage-collected heap. There now is a clear link to
the source position table from code, making the one-to-one relationship
and its lifetime explicit.
R=ahaas@chromium.org
BUG=v8:7424
Change-Id: I9d0b332732508c302ba525059ef02559f45aa2f6
Reviewed-on: https://chromium-review.googlesource.com/975565
Commit-Queue: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#52178}
The multi-return tests and fuzzer used a custom call descriptor which
was based on the default RegisterConfiguration. This meant that for the
tests, all available registers could be used to pass parameters and to
return values. This caused a problem, because in some cases we need a
scratch register in the frame deconstruction.
With this CL I change both the tests and the fuzzer to use the
WebAssembly call descriptor. Thereby we only use 2 registers for
returns, and one of the other registers can be used as scratch
register.
WebAssembly is the only use case at the moment which wants to return
values not only through registers but also over the stack. Therefore
I think it's acceptable to only test the WebAssembly usecase.
R=mstarzinger@chromium.org
Bug: chromium:813288
Change-Id: I31bed757af5f3e8589d2b3dfb6f0112ddecd1a20
Reviewed-on: https://chromium-review.googlesource.com/970656
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#52099}
Apparently the fuzzer tries to create functions with more 130000
parameters, which is too much for TurboFan. For returns I use the
wasm limit because only wasm uses multiple returns.
R=clemensh@chromium.org
Bug: chromium:811070
Change-Id: Ib9a55439c1da8e82ef5f35ffb2e79cab8d4a9018
Reviewed-on: https://chromium-review.googlesource.com/913268
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51231}
The multi-return fuzzer was able to generate more than 256 parameters of
the same type. However, the fuzzer itself could not deal with so many
parameters. With this change more than 256 parameters of the same type
can be handled and tested.
R=clemensh@chromium.org
Bug: chromium:807862
Change-Id: I6941eb0ff7e78a8feebc437624fa100adeda4e3d
Reviewed-on: https://chromium-review.googlesource.com/897673
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#51089}