For interpreted functions, use the optimized code slot in the feedback vector
to store an optimization marker (optimize/in optimization queue) rather than
changing the JSFunction's code object. Then, adapt the self-healing mechanism
to also dispatch based on this optimization marker. Similarly, replace SFI
marking with optimization marker checks in CompileLazy.
This allows JSFunctions to share optimization information (replacing shared
function marking) without leaking this information across native contexts. Non
I+TF functions (asm.js or --no-turbo) use a CheckOptimizationMarker shim which
generalises the old CompileOptimized/InOptimizationQueue builtins and also
checks the same optimization marker as CompileLazy and
InterpreterEntryTrampoline.
Change-Id: I6826bdde7ab9a919cdb6b69bc0ebc6174bcb91ae
Reviewed-on: https://chromium-review.googlesource.com/509716
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45901}
Since code flushing is gone, we treat the code entry as a strong field.
Change-Id: Idfcaf6fbfd84f7e4435b81d30a2a0e1be71ec89d
Reviewed-on: https://chromium-review.googlesource.com/531285
Commit-Queue: Ulan Degenbaev <ulan@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Hannes Payer <hpayer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45894}
With the deprecation of Crankshaft, it's no longer necessary for
FullCodeGen to keep track of its runtime profiler ticks on the code
object, and we can instead unify the behaviour of FCG and Ignition to
both increment the SFI counter instead.
Bug: v8:6408
Change-Id: Idcdd673aa39af06fe15a0fc14dfda2afafb5e417
Reviewed-on: https://chromium-review.googlesource.com/528117
Reviewed-by: Toon Verwaest <verwaest@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45892}
This is an unexciting CL (doesn't make the build step situation any better)
but enables moving FixedArray & co next.
BUG=v8:5402,v8:6474
Change-Id: Ia36eb3973e6242f6f68e02b9f583dc552d48422f
Reviewed-on: https://chromium-review.googlesource.com/529168
Commit-Queue: Marja Hölttä <marja@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45889}
Don't treat new prototypes differently depending on how they become a
prototype. This is work towards always keeping prototypes in slow-mode.
Bug: v8:6471
Change-Id: I62de1018e21d91fda3a5da044615f32c718910b1
Reviewed-on: https://chromium-review.googlesource.com/526596
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Toon Verwaest <verwaest@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45781}
This CL implements general infrastructure for block coverage together with
initial support for if-statements.
Coverage output can be generated in lcov format by d8 as follows:
$ d8 --block-coverage --lcov=$(echo ~/simple-if.lcov) ~/simple-if.js
$ genhtml ~/simple-if.lcov -o ~/simple-if
$ chrome ~/simple-if/index.html
A high level overview of the implementation follows:
The parser now collects source ranges unconditionally for relevant AST nodes.
Memory overhead is very low and this seemed like the cleanest and simplest
alternative.
Bytecode generation uses these ranges to allocate coverage slots and insert
IncBlockCounter instructions (e.g. at the beginning of then- and else blocks
for if-statements). The slot-range mapping is generated here and passed on
through CompilationInfo, and is later accessible through the
SharedFunctionInfo.
The IncBlockCounter bytecode fetches the slot-range mapping (called
CoverageInfo) from the shared function info and simply increments the counter.
We don't collect native-context-specific counts as they are irrelevant to our
use-cases.
Coverage information is finally generated on-demand through Coverage::Collect.
The only current consumer is a d8 front-end with lcov-style output, but the
short-term goal is to expose this through the inspector protocol.
BUG=v8:6000
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_chromium_rel_ng
Review-Url: https://codereview.chromium.org/2882973002
Cr-Commit-Position: refs/heads/master@{#45737}
When the input to Function.prototype.bind is a known function, we can
inline the allocation of the JSBoundFunction into TurboFan, which
provides a 2x speed-up for several hot functions in Node streams (as
discovered by Matteo Collina). One of example of this can be found in
https://github.com/nodejs/node/pull/13322, which can be optimized and
made more readable using bind instead of closures.
R=jarin@chromium.org
Review-Url: https://codereview.chromium.org/2916063002
Cr-Commit-Position: refs/heads/master@{#45679}
This is consistent with C++ memory model and avoids confusion with GC
write barrier.
BUG=
Review-Url: https://codereview.chromium.org/2912773002
Cr-Commit-Position: refs/heads/master@{#45584}
... which caused assertion failures in --enable-slow-asserts mode.
The surrounding code treated the constructor value properly so regression
test is not necessary.
Bug: chromium:726622
Change-Id: Icd43d9117a1125bec8feca8eca5708993de2c3ef
Reviewed-on: https://chromium-review.googlesource.com/516626
Reviewed-by: Daniel Vogelheim <vogelheim@chromium.org>
Commit-Queue: Igor Sheludko <ishell@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45543}
Asynchronous context tracking mechanisms in Node.js need to store some
state on all promise objects. This change will allow embedders to
configure the number of internal fields on promises as is already done
for ArrayBuffers.
BUG=v8:6435
Review-Url: https://codereview.chromium.org/2889863002
Cr-Commit-Position: refs/heads/master@{#45496}
Reason for revert:
Speculative revert for:
https://build.chromium.org/p/client.v8/builders/V8%20Win32%20-%20debug/builds/8901
Original issue's description:
> [es2015] Precompute the descriptive string for symbols.
>
> Previously the String constructor and the Symbol.prototype.toString
> methods had to compute the descriptive string for a Symbol on the fly,
> which can produce a lot of garbage when this happens a lot, i.e. when
> the String representation of a Symbol is used often. Now instead of
> doing this on-demand we can just do it upfront when creating the Symbol.
>
> That way we also ensure that we won't throw an exception when accessing
> the descriptive string of a Symbol, due to potential String length
> overflow, but have the exception during Symbol creation upfront, which
> is a lot less surprising behavior.
>
> BUG=v8:6278,v8:6344,v8:6350
> TBR=mlippautz@chromium.org
> R=ishell@chromium.org
>
> Review-Url: https://codereview.chromium.org/2900703002
> Cr-Commit-Position: refs/heads/master@{#45479}
> Committed: e87573822eTBR=ishell@chromium.org,mlippautz@chromium.org,bmeurer@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=v8:6278,v8:6344,v8:6350
Review-Url: https://codereview.chromium.org/2903533002
Cr-Commit-Position: refs/heads/master@{#45483}
Previously the String constructor and the Symbol.prototype.toString
methods had to compute the descriptive string for a Symbol on the fly,
which can produce a lot of garbage when this happens a lot, i.e. when
the String representation of a Symbol is used often. Now instead of
doing this on-demand we can just do it upfront when creating the Symbol.
That way we also ensure that we won't throw an exception when accessing
the descriptive string of a Symbol, due to potential String length
overflow, but have the exception during Symbol creation upfront, which
is a lot less surprising behavior.
BUG=v8:6278,v8:6344,v8:6350
TBR=mlippautz@chromium.orgR=ishell@chromium.org
Review-Url: https://codereview.chromium.org/2900703002
Cr-Commit-Position: refs/heads/master@{#45479}
For the Scavenger we require a first pass over global handles for identifying
unmodified nodes because the Scavenger might have already written forwarding
pointers during scanning, making it hard to perform the proper checks.
The minor MC does not mutate the object graph during marking and can thus merge
this phase into the regular phase executed during marking roots.
Furthermore, moves processing into the parallel marking phase of the minor MC
collector.
Bug: chromium:720477, chromium:651354
Change-Id: Id33552124264e3ab0bdf34d22ac30c19c1522707
Reviewed-on: https://chromium-review.googlesource.com/509550
Commit-Queue: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45461}
Only FullCodegen code ever gets flushed by code flushing. Since we are
deprecating the old pipeline, the added complexity introduced by code
flushing is no longer worth it. This CL removes it (but keeps code aging,
which is used to unlink SFIs from the compilation cache).
BUG=v8:6389,v8:6379,v8:6409
Change-Id: I90de113a101f86dbeaaf0511c61a090ef12aa365
Reviewed-on: https://chromium-review.googlesource.com/507388
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Hannes Payer <hpayer@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45446}
The WebAssembly code now uses these new APIs to allocate memory with guard
regions. Guarded array buffers are no longer always external, which eliminates
a lot of special cases around WebAssembly memory.
Bug: chromium:720302
Change-Id: I355b74ac30a05a18c8b363bd256d57458742849f
Reviewed-on: https://chromium-review.googlesource.com/505715
Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Reviewed-by: Brad Nelson <bradnelson@chromium.org>
Commit-Queue: Eric Holk <eholk@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45436}
WebAssembly needs to be able to allocate memory with guard regions, which
requires more functionality from the array buffer allocator. This change adds
functions for reserving memory regions and changing the memory protection.
This CL also includes some minor refactoring of the code to free array buffers.
Bug: chromium:720302
Change-Id: Iab9a266003043b0d36592a79668d1eea53952abf
Reviewed-on: https://chromium-review.googlesource.com/506377
Reviewed-by: Jochen Eisinger <jochen@chromium.org>
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Commit-Queue: Eric Holk <eholk@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45407}
This patch adds HeapObject::set_map_after_allocation method that
initializes the map of the object without object layout checks.
All other map setters now check that transitions unsafe for
concurrent marking properly notify the GC.
BUG=chromium:694255
Review-Url: https://codereview.chromium.org/2885883004
Cr-Commit-Position: refs/heads/master@{#45403}
Generate the code (extra runtime calls) for --trace-ignition support at
compile time, based on a #define (similar to TRACE_MAPS). Then check for
--trace-ignition at run-time when deciding whether to actually print
anything. This should make --trace-ignition less painful to use.
Note that --trace-igition is disabled by default, even on debug builds.
It has to be enabled with the gn arg "v8_enable_trace_ignition=true"
As a drive-by, TRACE_MAPS is renamed to V8_TRACE_MAPS, for consistency,
and SFI unique index (needed both by --trace-ignition and --trace-maps)
is cleaned up to be behind another #define.
Change-Id: I8dd0c62d0e6b7ee9c75541d45eb729dc03acbee9
Reviewed-on: https://chromium-review.googlesource.com/506203
Commit-Queue: Leszek Swirski <leszeks@chromium.org>
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45346}
We currently grow the backing store to (old_capacity*1.5)+16 if we exceed capacity,
but shrink the capacity to the current length when 2*length <= capacity.
For short arrays (up to length 32), this can lead to a copy on every operation when using push/pop or push/shift.
Example:
Array of length 32, capacity 32
push
Array grown to length 33, capacity 32*1.5+16 = 64
pop
Array trimmed to length 32, capacity 32 because 2*32 <= 64
...
This CL leaves additional slag space when calling pop and restricts the trimming to backing stores with at least 16 elements to prevent excessive re-trimming on short arrays.
Bug:
Change-Id: I9dd13e5e2550c7ac819294c8e29f04c8855e02a4
Reviewed-on: https://chromium-review.googlesource.com/502911
Commit-Queue: Tobias Tebbi <tebbi@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45324}
With this CL SloppyArguments immediately go to dictionary elements on
deletion, keeping the arguments backing store packed.
Bug: v8:6251
Change-Id: I90d1972179447bf6810e7fe2b8e0bc8703b38d9d
Reviewed-on: https://chromium-review.googlesource.com/486921
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Commit-Queue: Camillo Bruni <cbruni@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45286}
Since the feedback vector is itself a native context structure, why
not store optimized code for a function in there rather than in
a map from native context to code? This allows us to get rid of
the optimized code map in the SharedFunctionInfo, saving a pointer,
and making lookup of any optimized code quicker.
Original patch by Michael Stanton <mvstanton@chromium.org>
BUG=v8:6246,chromium:718891
TBR=yangguo@chromium.org,ulan@chromium.org
Change-Id: I3bb9ec0cfff32e667cca0e1403f964f33a6958a6
Reviewed-on: https://chromium-review.googlesource.com/500134
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45234}
For small arrays, it's way faster to just move the elements instead of
doing the fairly complex and heavy-weight left-trimming. Crankshaft has
had this optimization for small arrays already; this CL more or less
ports this functionality to TurboFan, which yields a 4x speed-up when
using shift on small arrays (with up to 16 elements).
This should recover some of the regressions reported in the Node.js issues
https://github.com/nodejs/node/issues/12657
and discovered for the syncthrough module using
https://github.com/mcollina/syncthrough/blob/master/benchmarks/basic.js
as benchmark.
R=jarin@chromium.org
BUG=v8:6376
Review-Url: https://codereview.chromium.org/2874453002
Cr-Commit-Position: refs/heads/master@{#45216}
This CL introduces a new type, MapHandles, which is a STL vector of Handle<Map>.
It is now used everywhere where lists of Handle<Maps> are required, replacing
usages of V8's internal List type.
Also-By: franzih@chromium.org
BUG=v8:6333,v8:6325
LOG=N
Review-Url: https://codereview.chromium.org/2809923002
Cr-Commit-Position: refs/heads/master@{#45211}
This reverts commit 662aa425ba.
Reason for revert: Crashing on Canary
BUG=chromium:718891
Original change's description:
> Reland: [TypeFeedbackVector] Store optimized code in the vector
>
> Since the feedback vector is itself a native context structure, why
> not store optimized code for a function in there rather than in
> a map from native context to code? This allows us to get rid of
> the optimized code map in the SharedFunctionInfo, saving a pointer,
> and making lookup of any optimized code quicker.
>
> Original patch by Michael Stanton <mvstanton@chromium.org>
>
> BUG=v8:6246
> TBR=yangguo@chromium.org,ulan@chromium.org
>
> Change-Id: Ic83e4011148164ef080c63215a0c77f1dfb7f327
> Reviewed-on: https://chromium-review.googlesource.com/494487
> Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
> Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#45084}
TBR=ulan@chromium.org,rmcilroy@chromium.org,yangguo@chromium.org,jarin@chromium.org
# Not skipping CQ checks because original CL landed > 1 day ago.
BUG=v8:6246
Change-Id: Idab648d6fe260862c2a0e35366df19dcecf13a82
Reviewed-on: https://chromium-review.googlesource.com/498633
Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45174}
The concurrent marker now colors objects grey and black using
the atomic markbit operations.
The heap visitor is changed in two ways:
1) It iterates the map pointer of each object.
2) It guards object visitation with a predicate, which is overridden
in the concurrent marker with the result of GreyToBlack transition.
BUG=chromium:694255
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_chromium_rel_ng
Review-Url: https://codereview.chromium.org/2866473003
Cr-Commit-Position: refs/heads/master@{#45144}
Revert "[heap] Make non-atomic markbit operations consistent with atomic ones."
This reverts commit dd37366fb5
Revert "[heap] Use atomic marking operations in incremental marking if"
This reverts commit 1f2c3596e9
Revert "[heap] Prepare IncrementalMarking::VisitObject for concurrent marking."
This reverts commit 00d1e2cf76
Revert "[heap] Use shared markbits in the concurrent marker."
This reverts commit b0db0541eehttps://codereview.chromium.org/2857713002 blocks the current roll:
https://codereview.chromium.org/2857423002/
Doesn't revert cleanly.
NOTRY=true
TBR=ulan@chromium.org
Bug: chromium:694255
Change-Id: Iada35af5c2529cd9e604802700604b16cc30aa2d
Reviewed-on: https://chromium-review.googlesource.com/497387
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45118}
If an ArrayBuffer is setup through the WebAssembly.Memory constructor, identify these with a flag and avoid optimizations in js-typed-lowering.cc. This is needed becasue buffers associated with memory objects can be grown/detached leading to crashes.
BUG=chromium:717194
Review-Url: https://codereview.chromium.org/2862763002
Cr-Commit-Position: refs/heads/master@{#45105}
The concurrent marker now colors objects grey and black using
the atomic markbit operations.
The heap visitor is changed in two ways:
1) It iterates the map pointer of each object.
2) It guards object visitation with a predicate, which is overridden
in the concurrent marker with the result of GreyToBlack transition.
BUG=chromium:694255
Review-Url: https://codereview.chromium.org/2855003004
Cr-Commit-Position: refs/heads/master@{#45099}
Since the feedback vector is itself a native context structure, why
not store optimized code for a function in there rather than in
a map from native context to code? This allows us to get rid of
the optimized code map in the SharedFunctionInfo, saving a pointer,
and making lookup of any optimized code quicker.
Original patch by Michael Stanton <mvstanton@chromium.org>
BUG=v8:6246
TBR=yangguo@chromium.org,ulan@chromium.org
Change-Id: Ic83e4011148164ef080c63215a0c77f1dfb7f327
Reviewed-on: https://chromium-review.googlesource.com/494487
Reviewed-by: Jaroslav Sevcik <jarin@chromium.org>
Commit-Queue: Ross McIlroy <rmcilroy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#45084}