... as a result of merging RelocInfo::target_object() with
RelocInfo::target_object_no_host(PtrComprCageBase),
where the cage base is used for accessing compressed embedded pointers.
There are two reasons for this change:
1) the parameterless version used to compute the cage base value from
the host Code object, however, when external code space is enabled
such a base value will not work for non-Code objects, since they
require different cage base for decompressing,
2) when external code space is enabled, there must be no need to embed
compressed Code objects at all because CodeDataContainers must be
used instead.
In addition this CL introduces DCHECKs to enforce (2).
Bug: v8:11880
Change-Id: I5b504f91dea87c2bcaa1165d2dbfaada70cba7be
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3211998
Reviewed-by: Camillo Bruni <cbruni@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/main@{#77361}
Move this from macro-assembler-x64 to shared-macro-assembler, and use
this implementation for ia32 (TurboFan and Liftoff).
Bug: v8:11589
Change-Id: If851560c8db1293924ca024725609c399c553a4a
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3124099
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Cr-Commit-Position: refs/heads/main@{#76803}
1. Move Abspd, Negpd from MacroAssembler into TurboAssembler so that we
can use it in code-generator
2. Add Absps and Negps (float32 versions of the instructions in 1)
3. Refactor SSE/AVX float32/float64 abs/neg to use these macro-assembler
helpers.
4. Use these helpers in Liftoff too
This has the benefit of not requiring to set up the masks in a temporary
register, and loading the constants via an ExternalReference instead.
It does require (in ins-sel) to have the input be in a Register, since
the ExternalReference is an operand (and the instruction can only have 1
operand input).
Bug: v8:11589
Change-Id: I68fafaf31b19ab05ee391aa3d54c45d547a85b34
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3123635
Reviewed-by: Adam Klein <adamk@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/main@{#76520}
- Add new Builtin enum
- Move Builtins::Name:kXXX to Builtin::kXXX
- Update existing code
Follow CLs will unify the mix of using int builtin-ids and
Builtins::Name to only use the new Builtin enum and changing it to
an enum class.
Change-Id: Ib39aa45a25696acdf147f46392901b1e051deaa4
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2905592
Commit-Queue: Camillo Bruni <cbruni@chromium.org>
Reviewed-by: Michael Stanton <mvstanton@chromium.org>
Reviewed-by: Mythri Alle <mythria@chromium.org>
Reviewed-by: Dominik Inführ <dinfuehr@chromium.org>
Reviewed-by: Andreas Haas <ahaas@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#74995}
On x64 we can emit more compact instructions for mov(reg, imm). However
currently this only happens when using the Set method explicitly.
This CL renames Set to Move to avoid confusion and yield better code
by default.
Also use the new Move helper for Smis as well.
Change-Id: I06558e88d1142098f77fb98870f09742d494f3dc
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2874450
Reviewed-by: Zhi An Ng <zhin@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Camillo Bruni <cbruni@chromium.org>
Cr-Commit-Position: refs/heads/master@{#74512}
For consistency with the PtrComprCageBase struct and the upcoming
PtrComprCage.
Bug: v8:11460
Change-Id: I2e393331c36481ee911edeaf9fb3ff971cfdba83
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2787701
Commit-Queue: Shu-yu Guo <syg@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#73874}
This is a reland of 0c63aa9eee
Fixes the correctness fuzzing BUILD.gn breakage.
Original change's description:
> [ptr-cage] Reserve base registers on x64 (r14) and arm64 (x28)
>
> Also add a V8_COMPRESS_POINTERS_IN_SHARED_CAGE define when pointer
> compression is enabled.
>
> This CL is to get performance numbers for reserving an extra register.
> There is no actual pointer cage yet, and the base register will always
> have the same value as the root register. The pointer decompression code
> is switched to using the base register instead of the root register.
>
> Bug: v8:11460
> Change-Id: I40bae556c2098608fb6fc193a52694e3f54754bd
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2716075
> Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
> Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
> Reviewed-by: Leszek Swirski <leszeks@chromium.org>
> Commit-Queue: Shu-yu Guo <syg@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#73204}
TBR=rmcilroy@chromium.org,jkummerow@chromium.org,leszeks@chromium.org
Bug: v8:11460
Change-Id: Iecf6b783392a384b40ab33e0f4ce13538a8f81ee
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2737681
Reviewed-by: Shu-yu Guo <syg@chromium.org>
Commit-Queue: Shu-yu Guo <syg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#73207}
This reverts commit 0c63aa9eee.
Reason for revert: Breaking clusterfuzz builds
Original change's description:
> [ptr-cage] Reserve base registers on x64 (r14) and arm64 (x28)
>
> Also add a V8_COMPRESS_POINTERS_IN_SHARED_CAGE define when pointer
> compression is enabled.
>
> This CL is to get performance numbers for reserving an extra register.
> There is no actual pointer cage yet, and the base register will always
> have the same value as the root register. The pointer decompression code
> is switched to using the base register instead of the root register.
>
> Bug: v8:11460
> Change-Id: I40bae556c2098608fb6fc193a52694e3f54754bd
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2716075
> Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
> Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
> Reviewed-by: Leszek Swirski <leszeks@chromium.org>
> Commit-Queue: Shu-yu Guo <syg@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#73204}
Bug: v8:11460
Change-Id: Idebf1fc6eeeda880a21d65b6f2c674fa58690bfa
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2737812
Auto-Submit: Shu-yu Guo <syg@chromium.org>
Commit-Queue: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Bot-Commit: Rubber Stamper <rubber-stamper@appspot.gserviceaccount.com>
Cr-Commit-Position: refs/heads/master@{#73205}
Also add a V8_COMPRESS_POINTERS_IN_SHARED_CAGE define when pointer
compression is enabled.
This CL is to get performance numbers for reserving an extra register.
There is no actual pointer cage yet, and the base register will always
have the same value as the root register. The pointer decompression code
is switched to using the base register instead of the root register.
Bug: v8:11460
Change-Id: I40bae556c2098608fb6fc193a52694e3f54754bd
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2716075
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Commit-Queue: Shu-yu Guo <syg@chromium.org>
Cr-Commit-Position: refs/heads/master@{#73204}
- Use movl which clears the upper 32bits on x64
- Use xorl + movb for Smi.ptr values <= 0xFF, saving one byte over movl
Change-Id: Iacdacfbe397670667e71d1d12ef427a01994481d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2642250
Commit-Queue: Camillo Bruni <cbruni@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#72282}
This is a reland of b2a611d815
Original change's description:
> [Turboprop] Move dynamic check maps immediate args to deopt exit.
>
> Rather than loading the immediate arguments required by the
> dynamic check maps builtin into registers in the fast-path,
> instead insert them into the instruction stream in the deopt
> exit and have the builtin load them into registers itself.
>
> BUG=v8:10582
>
> Change-Id: I66716570b408501374eed8f5e6432df64c6deb7c
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2589736
> Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
> Reviewed-by: Sathya Gunasekaran <gsathya@chromium.org>
> Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#71790}
TBR=tebbi@chromium.org,gsathya@chromium.org
Bug: v8:10582
Change-Id: Ieda0295ee135bff983c67c3f04bb47115f0a2739
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2595311
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#71803}
This reverts commit b2a611d815.
Reason for revert: Several failures on https://ci.chromium.org/ui/p/v8/builders/ci/V8%20Linux%20-%20arm64%20-%20sim%20-%20CFI/3743/overview
Original change's description:
> [Turboprop] Move dynamic check maps immediate args to deopt exit.
>
> Rather than loading the immediate arguments required by the
> dynamic check maps builtin into registers in the fast-path,
> instead insert them into the instruction stream in the deopt
> exit and have the builtin load them into registers itself.
>
> BUG=v8:10582
>
> Change-Id: I66716570b408501374eed8f5e6432df64c6deb7c
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2589736
> Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
> Reviewed-by: Sathya Gunasekaran <gsathya@chromium.org>
> Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#71790}
TBR=rmcilroy@chromium.org,gsathya@chromium.org,tebbi@chromium.org
Change-Id: I4c56bee156ffcea8de0aeaff9ac1bf03e03134c9
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:10582
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2595308
Reviewed-by: Clemens Backes <clemensb@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#71793}
Rather than loading the immediate arguments required by the
dynamic check maps builtin into registers in the fast-path,
instead insert them into the instruction stream in the deopt
exit and have the builtin load them into registers itself.
BUG=v8:10582
Change-Id: I66716570b408501374eed8f5e6432df64c6deb7c
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2589736
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Sathya Gunasekaran <gsathya@chromium.org>
Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#71790}
Unifies various operators for dynamic map checks with the naming
scheme of DynamicCheckMaps (to be similar to CheckMaps.
BUG=v8:10582
Change-Id: I8ac842f55fe31cdc7b84968d077017a86ddf4442
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2567952
Auto-Submit: Ross McIlroy <rmcilroy@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#71559}
In order to reduce the codegen size of dynamic map checks, add the
ability to have an eager with resume deopt point, which can call
a given builitin to perform a more detailed check than can be done
in codegen, and then either deoptimizes itself (as if the calling
code had performed an eager deopt) or resumes execution in the
calling code after the check.
In addition, support for adding extra arguments to a
deoptimization continuation is added to enable us to pass the
necessary arguments to the DynamicMapChecks builtin.
Finally, a trampoline is added to the DynamicMapChecks which saves
the registers that might be clobbered by that builtin, to avoid
having to save them in the generated code. This trampoline also
performs the deoptimization based on the result of the
DynamicMapChecks builtin.
In order to ensure both the trampoline and DynamicMapChecks
builtin have the same call interface, and to limit the number
of registers that need saving in the trampoline, the
DynamicMapChecks builtin is moved to be a CSA builtin with a
custom CallInterfaceDescriptor, that calls an exported Torque
macro that implements the actual functionality.
All told, this changes the codegen for a monomorphic dynamic
map check from:
movl rbx,<expected_map>
cmpl [<object>-0x1],rbx
jnz <deferred_call>
resume_point:
...
deferred_call:
<spill registers>
movl rax,<slot>
movq rbx,<object>
movq rcx,<handler>
movq r10,<DynamicMapChecks>
call r10
cmpq rax,0x0
jz <restore_regs>
cmpq rax,0x1
jz <deopt_point_1>
cmpq rax,0x2
jz <deopt_point_2>
int3l
restore_regs:
<restore_regs>
jmp <resume_point>
...
deopt_point_1:
call Deoptimization_Eager
deopt_point_2:
call Deoptimization_Bailout
To: movl rax,<slot>
movl rcx,<expected_map>
movq rdx,<handler>
cmpl [<object>-0x1],rcx
jnz <deopt_point>
resume_point:
...
deopt_point:
call DynamicMapChecksTrampoline
jmp <resume_point>
BUG=v8:10582
Change-Id: Ica4927b9acc963b9b73dc62d9379a7815335650f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2560197
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Georg Neis <neis@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#71545}
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 Liftoff, we have a good estimate about how big the generated code
might get. Also, we often compile hundreds of functions which each hold
an assembler buffer alive until we finally add that code to the wasm
module.
In order to reduce memory consumption in Liftoff, this CL reduces
{AssemblerBase::kMinimalBufferSize} from 4096 to 128, and adds
{AssemblerBase::kDefaultBufferSize} to be used instead.
R=jkummerow@chromium.org
Change-Id: I7029bf501244770f4824a86b233d7f99c4b7910b
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1914559
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#64958}
The current version of SmiTag and SmiUntag was checking if the
registers were the same, copying them if not and then untagging.
We can avoid a branch and a check by having two versions of
SmiTag and SmiUntag.
Change-Id: Id89213e073cefc9f8e46fcf0e79d0c1d349342ae
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1826730
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Commit-Queue: Santiago Aboy Solanes <solanes@chromium.org>
Cr-Commit-Position: refs/heads/master@{#64021}
This CL fixes comparison operations that take into account full-word
value instead of the lower 32 bits.
Bug: v8:9706
Change-Id: I04d2708f331a65e1c73302e8c36653f9cb40706e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1824946
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Cr-Commit-Position: refs/heads/master@{#64001}
Especially for function types, this increases readability significantly.
Also the style guide recommends for 'using' over 'typedef'.
R=mstarzinger@chromium.org
Bug: v8:9183
Change-Id: If2d17863de39383f5a35e089298d37408791ce4b
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1631415
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61872}
This CL was generated by an automatic clang AST rewriter using this
matcher expression:
callExpr(
callee(
cxxMethodDecl(
hasName("operator->"),
ofClass(isSameOrDerivedFrom("v8::internal::Object"))
)
),
argumentCountIs(1)
)
The "->" at the expression location was then rewritten to ".".
R=jkummerow@chromium.orgTBR=mstarzinger@chromium.org,verwaest@chromium.org,yangguo@chromium.org
Bug: v8:9183, v8:3770
No-Try: true
No-Tree-Checks: true
Change-Id: I0a7ecabdeafe51d0cf427f5280af0c7cab96869e
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1624209
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Commit-Queue: Yang Guo <yangguo@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61764}
Use the GeneratedCode construct in the EmbeddedObj test to ensure
clang with cfi support is happy.
Bug: v8:7703
Change-Id: Ib80add62127049ee5a763c6a4320add11682395d
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1611537
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Commit-Queue: Michael Stanton <mvstanton@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61470}
On Windows, expanding the stack by more than 4 KB at a time can cause
access violations. This change fixes a few known cases (and includes
unit tests for those), and attempts to make stack expansion more
consistent overall by using the AllocateStackSpace helper method
everywhere we can, even when the offset is a small constant.
On arm64, there was already a consistent method for stack pointer
manipulation using the Claim and Drop methods, so Claim is updated to
touch every page.
Bug: v8:9017
Change-Id: I2dbbceeebbdefaf45803e9b621fe83f52234a395
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1570666
Commit-Queue: Seth Brenith <seth.brenith@microsoft.com>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61186}
New RelocInfo test needs to adapt to the new pattern.
Change-Id: I09e531d099541511f94b929de647a3e3510f748f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1588465
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61108}
Failure addressed by not exposing the new test to the jitless environment.
(jgruber@ on TBR).
New enum RelocInfo::COMPRESSED_EMBEDDED_OBJECT created to support
compressed pointers in generated code. Enum name EMBEDDED_OBJECT
changed to FULL_EMBEDDED_OBJECT.
RelocInfo::[set_]target_object() abstract away the difference between
FULL_EMBEDDED_OBJECT and COMPRESSED_EMBEDDED_OBJECT.
Compressed embedded objects can only be created at this time on
x64 with pointer compression turned on. Arm64 constant pools don't
support compressed objects at this time.
NOPRESUBMIT=true
Bug: v8:7703
TBR: jgruber@chromium.org
Change-Id: Ifff53b041bab09b4b8c3e16085e5df4aa2b99f4f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1588461
Reviewed-by: Michael Stanton <mvstanton@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Michael Stanton <mvstanton@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61104}
This reverts commit b5da9fcb51.
Reason for revert: Breaks pointer compression bot:
https://ci.chromium.org/p/v8/builders/ci/V8%20Linux64%20-%20pointer%20compression/3098
Original change's description:
> [ptr-compr] New RelocInfo for compressed pointers.
>
> New enum RelocInfo::COMPRESSED_EMBEDDED_OBJECT created to support
> compressed pointers in generated code. Enum name EMBEDDED_OBJECT
> changed to FULL_EMBEDDED_OBJECT.
>
> RelocInfo::[set_]target_object() abstract away the difference between
> FULL_EMBEDDED_OBJECT and COMPRESSED_EMBEDDED_OBJECT.
>
> Compressed embedded objects can only be created at this time on
> x64 with pointer compression turned on. Arm64 constant pools don't
> support compressed objects at this time.
>
> Bug: v8:7703
> Change-Id: I03bfd84effa33c65cf9bcefa5df680ab7eace9dd
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1547661
> Commit-Queue: Michael Stanton <mvstanton@chromium.org>
> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
> Reviewed-by: Jakob Gruber <jgruber@chromium.org>
> Reviewed-by: Igor Sheludko <ishell@chromium.org>
> Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#61076}
TBR=ulan@chromium.org,mvstanton@chromium.org,mstarzinger@chromium.org,jgruber@chromium.org,ishell@chromium.org
Change-Id: I262b2b98315fa987c5a66b1050dc726563ccdb2d
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:7703
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1588135
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61087}
New enum RelocInfo::COMPRESSED_EMBEDDED_OBJECT created to support
compressed pointers in generated code. Enum name EMBEDDED_OBJECT
changed to FULL_EMBEDDED_OBJECT.
RelocInfo::[set_]target_object() abstract away the difference between
FULL_EMBEDDED_OBJECT and COMPRESSED_EMBEDDED_OBJECT.
Compressed embedded objects can only be created at this time on
x64 with pointer compression turned on. Arm64 constant pools don't
support compressed objects at this time.
Bug: v8:7703
Change-Id: I03bfd84effa33c65cf9bcefa5df680ab7eace9dd
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1547661
Commit-Queue: Michael Stanton <mvstanton@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Reviewed-by: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61076}
In the Crankshaft days we (mis)used the Representation to also express
the various internal representations that the compiler understands. But
with TurboFan we now have proper MachineRepresentation and MachineType,
which do that independently. So there's no need to have this in the
Representation class anymore, and instead the Representation class only
needs to deal with the field representations.
Bug: v8:8749, v8:8834, v8:8865
Change-Id: I34ea9558b5fdf20d6c7939b52762eaffd4316b06
Reviewed-on: https://chromium-review.googlesource.com/c/1479954
Commit-Queue: Benedikt Meurer <bmeurer@chromium.org>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59750}
Refactor the AllocateAssemblerBuffer helper for the new Assembler API.
This is the only non-mechanical part, all other callsites that create
Assembler instances can be trivially changed to the new API. This will
be done in a separate CL.
R=mstarzinger@chromium.org
Bug: v8:8689, v8:8562
Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b
Reviewed-on: https://chromium-review.googlesource.com/c/1415490
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#58881}
in order to make it also work with optional registers.
Bug: v8:8562
Change-Id: Iaea905913cc9fd1637026b83e9356c740965e128
Reviewed-on: https://chromium-review.googlesource.com/c/1367807
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#58105}
This is a reland of a462a7854a
Original change's description:
> [turboassembler] Introduce hard-abort mode
>
> For checks and assertions (mostly for debug code, like stack alignment
> or zero extension), we had two modes: Emit a call to the {Abort}
> runtime function (the default), and emit a debug break (used for
> testing, enabled via --trap-on-abort).
> In wasm, where we cannot just call a runtime function because code must
> be isolate independent, we always used the trap-on-abort behaviour.
> This causes problems for our fuzzers, which do not catch SIGTRAP, and
> hence do not detect debug code failures.
>
> This CL introduces a third mode ("hard abort"), which calls a C
> function via {ExternalReference}. The C function still outputs the
> abort reason, but does not print the stack trace. It then aborts via
> "OS::Abort", just like the runtime function.
> This will allow fuzzers to detect the crash and even find a nice error
> message.
>
> Even though this looks like a lot of code churn, it is actually not.
> Most added lines are new tests, and other changes are minimal.
>
> R=mstarzinger@chromium.org
>
> Bug: chromium:863799
> Change-Id: I77c58ff72db552d49014614436259ccfb49ba87b
> Reviewed-on: https://chromium-review.googlesource.com/1142163
> Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#54592}
Bug: chromium:863799
Change-Id: I7729a47b4823a982a8e201df36520aa2b6ef5326
Reviewed-on: https://chromium-review.googlesource.com/1146100
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Clemens Hammacher <clemensh@chromium.org>
Cr-Commit-Position: refs/heads/master@{#54656}
