Hey folks,
I summarized the issue with the Script leak in the inspector in a doc
<https://docs.google.com/document/d/1tZcF1KZSJvSKJTeBc1Cwtt2f1uPrxW0g1yXwc030auI>,
and outlined the proposed solution (and alternatives we considered).
Please take a look.
cheers,
Benedikt
--
Benedikt
FrameState, it's just that this was the way the world worked back when we
wrote most of this code.
cheers,
Benedikt
On Tue, Sep 22, 2020 at 10:30 AM 'Victor Gomes' via v8-dev <
v8-dev@googlegroups.com> wrote:
> Thanks Benedikt!
>
> On Mon, Sep 21, 2020 at 6:04 PM Benedikt
FuzY0LLM7%2BAjoaax1usWN9Jov1eq89EvdYKdVDtdR80j5rFA%40mail.gmail.com
> <https://groups.google.com/d/msgid/v8-dev/CAGFuzY0LLM7%2BAjoaax1usWN9Jov1eq89EvdYKdVDtdR80j5rFA%40mail.gmail.com?utm_medium=email_source=footer>
> .
>
--
Benedikt Meurer
Chromium DevTools TL
bmeu...@chromium.org
Googl
devtools frontend through CDP?
> Or are we looking to replace LLDB with our own debugger interpretation?
>
> Best regards,
> Philip
>
> On Fri, Oct 11, 2019 at 8:51 AM Benedikt Meurer
> wrote:
>
>> Hey folks,
>>
>> Following up on the prototyping
I've shared an item with you:
WebAssembly Inspector Interface
https://docs.google.com/document/d/1zi5LLKwWMRdIIWN4Vgyy7RgcM0f4Xi3Ffn27QO1iMGU/edit?usp=sharing=5da06e9e
It's not an attachment -- it's stored online. To open this item, just click
the link above.
Hey folks (specifically Wasm
it describes two possible high level
solutions how to address the current problem.
Please let me know what you think.
cheers,
Benedikt
--
Benedikt Meurer
Chromium DevTools TL
bmeu...@chromium.org
Google Germany GmbH
Erika-Mann-Straße 33
80636 München
Geschäftsführer: Paul Manicle, Halimah
cussion on the web visit
>>> https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAMd%2BM7yLMvv55qcOXyW8o_g3sHe0mrYgT_WC%2BmE2GRrqKG-SuA%40mail.gmail.com
>>> <https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAMd%2BM7yLMvv55qcOXyW8o_g3sHe0mrYgT_WC%2BmE2GRrqK
LGTM!
On Wed, Apr 24, 2019, 02:05 Adam Klein wrote:
> LGTM!
>
> On Tue, Apr 23, 2019 at 4:19 PM Sathya Gunasekaran
> wrote:
>
>> Contact Emails:
>> gsat...@chromium.org
>>
>> Spec:
>> https://tc39.github.io/proposal-promise-allSettled/
>>
>> Summary:
>> Promise.allSettled returns a promise
gt;> https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAAKm-Q%2BuNU%3D%2BbSgu87THKGedRbf1OuCQSsJbmRgK5iPWAVO%2BDA%40mail.gmail.com
>> <https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAAKm-Q%2BuNU%3D%2BbSgu87THKGedRbf1OuCQSsJbmRgK5iPWAVO%2BDA%40mail.gmail.com?utm_
al.clientHttpRequest] += ' (' +
>> Math.round(Math.sqrt(
>>
>> local.resultList[local.clientHttpRequest].reduce(function (aa, bb) {
>> return aa + Math.pow(
>> (bb || 0) -
>> local.resultMean[local.clientHttpRe
sks are run in a single JS entry, whereas in v8 ToT
> you enter/exit JS thousands of times.
>
> Cheers,
>
> Yang
>
> On Fri, Sep 1, 2017 at 7:29 AM Benedikt Meurer <bmeu...@chromium.org>
> wrote:
>
>> I don't think strong-rooting a JSFunction is a good
ved regs are actually saved and restored.
>
> If I can strong-root a full JSFunction and just use the ordinary
> JSEntryStub, as Adam suggested, it may be a non-issue.
>
> On Aug 31, 2017, at 11:00 PM, Benedikt Meurer <bmeu...@chromium.org>
> wrote:
>
> I like the idea of bei
I like the idea of being able to run microtasks from CSA land. Not sure
about this JSEntryStub businesses tho, it all sounds dangerous to me. Is
this just to avoid some code duplication between CSA and C++? If so, then
I'd strongly recommend against it and just duplicate that logic for now. If
not
that Nodev9 for both ES5 and ES6 run approximately the same speed at
> 820,000 eps which is about *1.7x faster* than Nodev8 using ES5!
>
> Kudo's to the V8 team!!
>
>
> On Monday, June 12, 2017 at 9:36:01 PM UTC-7, Benedikt Meurer wrote:
>
>> With V8 5.8, that sounds
With V8 5.8, that sounds possible. In 5.8 modern JavaScript (i.e. ES2015
and later) is sent to the new compiler pipeline (based on Ignition and
TurboFan), whereas classic JavaScript is sent to the old compiler pipeline
(based on FullCodegen and Crankshaft). JavaScript code is strictly divided
by
The ObjectBuiltinsAssembler adds a couple of additional helper methods.
-- Benedikt
Benedikt Meurer | Software Engineer, V8 | Google Germany GmbH |
Erika-Mann-Str.
33, 80636 München
Registergericht und -nummer: Hamburg, HRB 86891 | Sitz der Gesellschaft:
Hamburg | Geschäftsführer: Matthew
for
the ToObject builtin (which has internal stub linkage). I don't think you
need to touch src/runtime or src/debug for this.
HTH,
Benedikt
Benedikt Meurer | Software Engineer, V8 | Google Germany GmbH |
Erika-Mann-Str.
33, 80636 München
Registergericht und -nummer: Hamburg, HRB 86891 | Sitz der
Hey Dmitriy,
Each JSObject contains at least three fields, the map, the properties
backing store and the elements backing store. The first movp in the code
you point to initializes the map, the next two initializes the backing
stores to empty arrays.
HTH,
Benedikt
Benedikt Meurer | Software
Potter <caitpotte...@gmail.com> schrieb am Fr., 3. Feb. 2017, 04:26:
> Sorry bout that, sharing is enabled now
>
> Sent from my iPhone
>
> On Feb 2, 2017, at 10:05 PM, Benedikt Meurer <bmeu...@chromium.org> wrote:
>
> Hey Caitlin,
>
> Can you enable li
Hey Caitlin,
Can you enable link sharing for the doc?
-- Benedikt
Caitlin Potter (via Google Docs) schrieb am Fr., 3. Feb.
2017, 00:36:
> Caitlin Potter has invited you to *edit* the following
> document:
> Desugaring language features in the
Indeed. We need to kill FCG+CS first. I was wondering about the same for
ToBoolean feedback in Ignition, where only Ignition knows whether it really
needs a slot.
-- Benedikt
'Ross McIlroy' via v8-dev schrieb am Mi., 7. Dez.
2016, 19:39:
> Hi Caitlin,
>
> As I
Hey Adam,
I think that this is not even consistent; as you point out, some builtins
are strict (the js ones), while others are sloppy. I think this is for
historical reasons, because initially everything was sloppy, and then we
migrated some of the js files to strict mode. But so far no one payed
Hey Deon,
No there's currently no way to trace reducers in general. The
--trace-turbo-graph flag prints out the full graph after every pass, once
all the reductions in that pass happened, but I suppose that's not what you
were looking for.
-- Benedikt
Deon Dior schrieb
Awesome! Thanks for taking care of this, Dan.
On Wed, May 11, 2016 at 7:54 AM, 'Daniel Ehrenberg' via v8-dev <
v8-dev@googlegroups.com> wrote:
> I ended up adding UseCounters and waiting for results to come in. Finally,
> they did: none of these three methods were used very much, the most
>
Hey Yang,
Currently asm.js still goes through the full JavaScript pipeline in
TurboFan, which turned out to be too expensive for AOT. Hence we start with
fullcodegen and only tier up to TurboFan if the function becomes hot.
This will change soon once the asm.js verfier is enabled. Then asm.js
Hey Deon,
The whole so called "typed pipeline" (initially enabled via --turbo-types)
is optional in theory, which means that only the "generic pipeline"
(consisting only of graph building and generic lowering) runs. This doesn't
make sense in practice though as the generated code is then
Hey Robert,
It depends on the context: if your x is a property of the global object,
then V8 still optimizes Math.pow in your example. Even if it's a local
variable, we might still be able to optimize this in many cases.
But to answer your question: The polyfill assignment won't break the
YI, another wrinkle: MSVC doesn't support inline assembly for x64
>> targets <https://msdn.microsoft.com/en-us/library/wbk4z78b.aspx>. :-|
>>
>> Sounds like the suggested workaround is intrinsics or using the assembler
>> directly.
>>
>> On Monday, Janua
Yeah, we talked about that offline before Christmas and mostly agreed that
we should do it soonish.
Yang Guo schrieb am Mi., 6. Jan. 2016, 22:02:
> Ive had thoughts about this as well, and it indeed makes sense. SGTM.
>
> Yang
>
> On Wed, Jan 6, 2016, 20:00 'Adam Klein'
Hm, skipping those tests in the simulators could easily backfire. Let's
just do the runtime function approach and see how far we can get.
-- Benedikt
Ben Smith <bi...@chromium.org> schrieb am Mo., 4. Jan. 2016, 22:05:
>
>
> On Tuesday, December 29, 2015 at 9:07:15 PM UTC-8,
22:35:
> On Thursday, December 17, 2015 at 2:17:01 PM UTC-8, Ben Smith wrote:
>>
>>
>>
>> On Thursday, December 17, 2015 at 11:26:48 AM UTC-8, Benedikt Meurer
>> wrote:
>>>
>>> Can't you use the same instruction sequence, even from C++ code?
>>&
sure to generate the
exactly the same instructions I guess?
IIRC Jaro thought about this already, maybe he can give some advice.
On Thu, Dec 17, 2015 at 8:06 PM, Ben Smith <bi...@chromium.org> wrote:
>
>
> On Wednesday, December 16, 2015 at 7:16:07 PM UTC-8, Benedikt Meurer w
You probably need the same guarantees for CS. So how about putting them
into a code stub? And calling the stub for the %_ cases in all three
compilers?
BTW wouldn't it work if FCG always uses the C++ version?
-- Benedikt
Ben Smith schrieb am Do., 17. Dez. 2015, 00:01:
> Hi
How about adding overloads for the template with pointers w/o
implementation?
On Wed, Dec 9, 2015 at 10:59 AM, 'Erik Corry' via v8-dev <
v8-dev@googlegroups.com> wrote:
> Yes. We plan to put in asserts to prevent it being used for pointer types.
>
> On 4 December 2015 at 15:30, Jochen Eisinger
This is creating a lot of noise in mainstream media, I.e.
http://m.heise.de/newsticker/meldung/JavaScript-Engine-V8-Vorsicht-vor-Math-random-3010353.html
Jakob Kummerow schrieb am Fr., 20. Nov. 2015,
16:10:
> Observe https://codereview.chromium.org/1462293002 which
SGTM
On Thu, Oct 15, 2015 at 3:21 PM, Yang Guo wrote:
> I agree, if Clusterfuzz no longer directly calls runtime functions with
> weird arguments anymore.
>
> On Thu, Oct 15, 2015 at 3:16 PM Michael Starzinger <
> mstarzin...@chromium.org> wrote:
>
>> *TL;DR*: I propose to
I had a quick look at the Promises code and I think a lot of that would
actually benefit from being rewritten in C++. I don't know yet about the
typed array code.
'Daniel Ehrenberg' via v8-dev schrieb am Fr., 2.
Okt. 2015, 03:27:
> On Sat, Sep 26, 2015 at 5:44 AM, Jakob
AFAIK everyone is doing (1) currently, and so at least for now, which is
kinda what we agreed on some time ago. I don't see why we have to change
that now. I think (3) is essentially (2) with a different directory/base
filename, and many people thought that (2) was bad, which AFAIR is why we
are
One important bit tho: The (static) methods should be somewhere close the
classes they relate to, otherwise we just go back to runtime-*.cc.
On Fri, Sep 25, 2015 at 1:37 PM, Benedikt Meurer <bmeu...@chromium.org>
wrote:
> AFAIK everyone is doing (1) currently, and so at least for n
Reviewers: Sven Panne,
Description:
[mjsunit] Remove unsupported flag --turbo-deoptimization from tests.
R=svenpa...@chromium.org
Please review this at https://codereview.chromium.org/1183123002/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+18, -43 lines):
Reviewers: Igor Sheludko,
Description:
[mjsunit] regress/regress-crbug-491062 takes too long with --always-opt.
R=ish...@chromium.org
Please review this at https://codereview.chromium.org/1183133002/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+1, -1 lines):
Reviewers: Sven Panne,
Description:
[mjsunit] Remove obsolete nosse2 tests.
R=svenpa...@chromium.org
Please review this at https://codereview.chromium.org/1186713003/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+1, -883 lines):
D
Reviewers: Michael Starzinger,
Description:
[turbofan] Remove context canonicalization hack from generic lowering.
This hack doesn't help currently, since it only applies to asm.js where
we do function context specialization and there we already have a
canonicalized context (and the
Reviewers: Michael Starzinger,
Description:
[turbofan] Remove the TryLowerDirectJSCall hack from generic lowering.
The TryLowerDirectJSCall method tried to lower to a direct JavaScript
function call depending on the type of the receiver, but only if the
target is a cosntant JSFunction. Since
Hey Mircea,
I think I just came up with a good criteria for a new register allocator:
It should be possible to add stack checks to every loop in asm.js code
without noticable performance impact.
Currently we skip the stack checks for asm.js functions because the
spilling around the stack check
Reviewers: Sven Panne,
Description:
[turbofan] Remove unused typedef VirtualRegisterSet.
TBR=svenpa...@chromium.org
Please review this at https://codereview.chromium.org/1179913003/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+0, -1 lines):
M
Reviewers: Sven Panne,
Description:
[turbofan] Mark MachineType as uint16_t.
This fixes a few funky implicit conversions for the enum (that are
inconsistent across compilers) and also helps to save space, i.e. for
the representations_ vector in the InstructionSequence.
R=svenpa...@chromium.org
Reviewers: Sven Panne,
Description:
[turbofan] Use appropriate type for NodeId.
Up until now we used int32_t for NodeId, but that was not ideal because
negative values are invalid for NodeId and we use it as an array index
for example in the NodeMarker class, where C++ compilers on x64 have to
Reviewers: Yang,
Description:
[heap] All immutable roots are either smis or immovable heap objects.
R=yang...@chromium.org
Please review this at https://codereview.chromium.org/1180043002/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+5, -2 lines):
M
Committed patchset #1 (id:1) manually as
bd3edf37e1a1ecf158b1ec57904558b5afddd5e3 (presubmit successful).
https://codereview.chromium.org/1184693002/
--
--
v8-dev mailing list
v8-dev@googlegroups.com
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed
Reviewers: jarin,
Description:
[turbofan] Inline hot functions for NodeMarkerBase.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1184693002/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+18, -32 lines):
M
Reviewers: Michael Starzinger,
Description:
[turbofan] Move RawMachineAssembler to unittests.
R=mstarzin...@chromium.org
Please review this at https://codereview.chromium.org/1177353003/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+6, -844 lines):
M
Reviewers: jarin,
Description:
[turbofan] Materialize all from non-writable roots from root array.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1175113005/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+10, -19 lines):
M
Reviewers: jarin,
Description:
[turbofan] Use RootIndexMap to speed up IsMaterializableFromRoot predicate.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1181123002/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+9, -13 lines):
M
Reviewers: jarin,
Description:
[heap] Unify the immortal immovable root detection mechanism.
Uniformly use the Heap::GetRootListIndex() and
Heap::RootIsImmortalImmovable() methods to detect immortal immovable
roots in the optimizing compilers.
R=ja...@chromium.org
Please review this at
Reviewers: jarin,
Description:
[deoptimizer] Basic support inlining based on SharedFunctionInfo.
Up until now we can only inline based on JSFunction, because of the way
the deoptimization works. With this change we will be able to inline
based on the SharedFunctionInfo and materialize the
Reviewers: Michael Starzinger,
Description:
[turbofan] Record the SharedFunctionInfo of ALL inlined functions.
Previously we only recorded the SharedFunctionInfo of inlined functions
that had at least one (lazy) deopt point left at code generation time.
R=mstarzin...@chromium.org
Please
Reviewers: jarin,
Description:
[stl] Fix ZonePriorityQueue wrapper.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1179763002/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+4, -2 lines):
M src/zone-containers.h
Index:
Committed patchset #2 (id:20001) manually as
ffa0b4007cd7de0cfd6d37079ef360e3beeb5686 (presubmit successful).
https://codereview.chromium.org/1175953002/
--
--
v8-dev mailing list
v8-dev@googlegroups.com
http://groups.google.com/group/v8-dev
---
You received this message because you are
Reviewers: Michael Starzinger,
Description:
[frames] Remove obsolete JavaScriptFrame::GetInlineCount() method.
R=mstarzin...@chromium.org
Please review this at https://codereview.chromium.org/1176473002/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+7, -33
Reviewers: jarin,
Description:
[deoptimizer] Remove uses of TranslationIterator outside the deoptimizer.
Use the new TranslatedState and friends, which work at a higher level
than the TranslationIterator, which will make it easier to change the
deoptimization commands in subsequent CLs.
Committed patchset #1 (id:1) manually as
472b147708ba585f07c0020434302d96867cbb61 (presubmit successful).
https://codereview.chromium.org/1174593003/
--
--
v8-dev mailing list
v8-dev@googlegroups.com
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed
Reviewers: jarin,
Description:
[frames] No GC is allowed while using the unhandlified TranslatedState.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1174593003/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+2, -0 lines):
M
Reviewers: jarin,
Description:
[for-in] Make ForInNext and ForInFilter deal properly with exceptions.
Both ToName() and HasProperty() can actually throw, so we need to
propagate those exceptions properly.
BUG=chromium:496331
LOG=y
R=ja...@chromium.org
Please review this at
Reviewers: jarin,
Description:
[turbofan] Initial support for the %_DateField intrinsic.
This only introduces better typing and lowering for access to the value
field.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1169723002/
Base URL:
Committed patchset #2 (id:20001) manually as
5eafd7a3d95d9ed1e5f1f11a1ea408a1415da2a8 (presubmit successful).
https://codereview.chromium.org/1169723002/
--
--
v8-dev mailing list
v8-dev@googlegroups.com
http://groups.google.com/group/v8-dev
---
You received this message because you are
Reviewers: jarin,
Description:
[x64] Smi zero can be used as an immediate.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1150853009/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+10, -1 lines):
M
Reviewers: Jakob,
Description:
[date] Refactor the %_DateField intrinsic to be optimizable.
Previously the %_DateField intrinsic would also check the object and
throw an exception if you happen to pass something that is not a valid
JSDate, which (a) violates our policy for instrinsics and (b)
Reviewers: jarin,
Description:
[turbofan] Don't lower to NumberModulus unless the inputs are numbers.
The IC for modulus is usually way faster than converting the inputs to
numbers and doing a Float64Mod on them.
R=ja...@chromium.org
Please review this at
Reviewers: Michael Starzinger,
Description:
[turbofan] Use reference equal to zero instead of a smi check.
In typed lowering we can use the ReferenceEqual simplified operator
instead of ObjectIsSmi to check for context extensions. This generates
the desired code.
TBR=mstarzin...@chromium.org
Reviewers: jarin,
Description:
[turbofan] Enable typed lowering of string addition.
Unfortunately StringAdd is not pure in V8 because we might throw an
exception if the resulting string length is outside the valid bounds, so
there's no point in having a simplified StringAdd operator.
Reviewers: jarin,
Description:
[turbofan] Specify better type for FixedArray::length field.
This avoids redundant smi checks when using the length of a FixedArray,
which is always a positve smi.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1162163003/
Base URL:
Reviewers: jarin,
Description:
[turbofan] First step towards sanitizing for-in and making it optimizable.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1160983004/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+360, -202 lines):
M
Reviewers: jarin,
Description:
[crankshaft] Fix some useless deopts in for-in.
The one in lithium codegen for x64 was probably an oversight during
refactoring.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1157423003/
Base URL:
Reviewers: jarin,
Description:
[turbofan] Remove the useless SimplifiedOperatorReducer.
The SimplifiedOperatorReducer is (mostly) unused, except for the very
rough store elimination, and just eats compilation time.
R=ja...@chromium.org
Please review this at
Reviewers: jarin,
Description:
[turbofan] Remove the JSGraph dependency from the ControlFlowOptimizer.
The control flow optimization should work independent of the JSGraph. We
used the JSGraph there because it was convinient, not because it was
necessary.
R=ja...@chromium.org
Please review
Reviewers: jarin,
Description:
[crankshaft] Record inlined shared function infos instead of closures.
The list of inlined functions is used in exactly two places - for live
edit and to prevent code flushing for inlined functions - and those are
fine with SharedFunctionInfo and don't require a
Reviewers: jarin,
Description:
[turbofan] Record SharedFunctionInfo of inlined functions.
We need the shared function info of inlined functions to prevent code
flushing for their unoptimized code, and also to make sure that liveedit
can find the proper functions to deoptimize.
Reviewers: jarin,
Description:
[deoptimizer] Materialize double values as smis whenever possible.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1156393002/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+48, -14 lines):
M
Reviewers: jarin,
Description:
[turbofan] Simplify graph construction for for-in.
This is an initial step towards a faster and less incorrect
implementation of for-in in TurboFan.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1153403002/
Base URL:
Reviewers: Michael Starzinger,
Description:
[turbofan] Use Start as sentinel for frame states.
This simplifies inlining, in that we only need to update uses of Start
and inputs of End instead of walking the whole inlinee to update all
outer frame states.
R=mstarzin...@chromium.org
Please
Reviewers: jarin,
Description:
[turbofan] Remove frame state TODOs from VisitForInBody.
Neither the increment nor the comparison can deoptimize, so we don't
need proper frame states there.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1148323003/
Base URL:
Reviewers: Michael Starzinger,
Description:
[turbofan] Optimize and || in test context.
R=mstarzin...@chromium.org
Please review this at https://codereview.chromium.org/1160683003/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+4, -0 lines):
M
Reviewers: Michael Starzinger,
Description:
[turbofan] Change End to take a variable number of inputs.
This simplifies the handling of the End node. Based on this CL we will
finally fix terminating every loop from the beginning (via Terminate
nodes) and fix inlining of Throw, Deoptimize and
Reviewers: jarin,
Description:
[turbofan] Properly kill Terminate nodes when removing loops.
BUG=chromium:491578
LOG=n
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1161583002/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+18, -7
Reviewers: Michael Starzinger,
Description:
[turbofan] Connect loops to end via Terminate during graph building.
This way we don't need to connect (potentially) non-terminating loops
later during control reduction, which saves one forward pass over the
control graph. Long term we will move the
Reviewers: Michael Starzinger,
Description:
[turbofan] Don't mix up name/property accesses.
R=mstarzin...@chromium.org
Please review this at https://codereview.chromium.org/1149663003/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+17, -78 lines):
M
Reviewers: jarin,
Description:
[turbofan] Enable deoptimization for non-asm.js TurboFan code.
Replace the --turbo-deoptimization flag with --turbo-asm-deoptimization
and enable deoptimization for non-asm.js TurboFan code unconditionally.
R=ja...@chromium.org
Please review this at
Reviewers: Michael Starzinger,
Description:
[turbofan] --turbo should not imply --turbo-type-feedback.
The --turbo flag should reflect our TurboFan staging configuration, and
for now that does not include the JSTypeFeedbackSpecializer.
R=mstarzin...@chromium.org
Please review this at
Committed patchset #1 (id:1) manually as
f2ffa6ade5e28d6ff1ed8895e320fc77bfb08c68 (presubmit successful).
https://codereview.chromium.org/1150923005/
--
--
v8-dev mailing list
v8-dev@googlegroups.com
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed
Reviewers: Michael Starzinger,
Description:
[turbofan] Fix UnifyReturn magic in the inliner.
The inliner previously assumed that there will only be returns reaching
the end node, but that's not true. This refactoring will make it
possible to also hook up Deoptimize, Throw and Terminate nodes
Reviewers: Michael Starzinger,
Description:
[turbofan] Add streaming operator for opcodes to ease debugging.
R=mstarzin...@chromium.org
Please review this at https://codereview.chromium.org/1147973002/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+10, -0
Reviewers: jarin,
Description:
[turbofan] Pass deoptimization mode to type feedback specializer.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1145143002/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+229, -160 lines):
M
Reviewers: jarin,
Description:
[turbofan] Pass deoptimization mode to intrinsic lowering.
R=ja...@chromium.org
Please review this at https://codereview.chromium.org/1148033002/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+18, -8 lines):
M
Reviewers: jarin,
Description:
[turbofan] Carefully update frame states during inlining.
During inlining we should pay attention to only rewire the outer frame
states of the inlinee, but leave any inner frame states of the inlinee
untouched. Otherwise we might run into trouble once we start
95 matches
Mail list logo