We need to check the KeyedLoadIC state to guard against potential
deoptimization loops due to out-of-bounds accesses, because the IC
system uses the MEGAMORPHIC state to also signal that there was an
out-of-bounds access already.
R=jarin@chromium.org
Review-Url: https://codereview.chromium.org/2443893002
Cr-Commit-Position: refs/heads/master@{#40525}
Fixes:
- Remove OsrGuards on frame specialization (for asm.js).
- Handle the rename in the walk for native context.
- Fix LoadContext effect wiring for Osr context chains.
Review-Url: https://codereview.chromium.org/2388303006
Cr-Commit-Position: refs/heads/master@{#40021}
Reason for revert:
Tanks the world.
Original issue's description:
> [turbofan] Osr value typing + dynamic type checks on entry.
>
> This introduces a new OsrGuard node that is inserted during graph building
> to guard the inferred type of the OSR value.
>
> The type of the OSR value is inferred by running the typer before OSR
> deconstruction, and then taking the type from the phi that takes the
> OSR value. After the deconstruction, we throw the types away.
>
> At the moment we only support the SignedSmall OSR type and we always
> pick the tagged representation. Later, we might want to support more
> types (such as Number) and pick better representations (int32/float64).
>
> This CL also removes the OSR deconstruction tests because they build
> unrealistic graph (no effect chain, no loop termination). I considered
> adding the effect chains to the tests, but this would make the tests
> even more brittle.
>
> Committed: https://crrev.com/1f5dc90a900d222da44bee3eff171a2ba1e3c076
> Cr-Commit-Position: refs/heads/master@{#39971}
TBR=bmeurer@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
Review-Url: https://codereview.chromium.org/2395783002
Cr-Commit-Position: refs/heads/master@{#39985}
This introduces a new OsrGuard node that is inserted during graph building
to guard the inferred type of the OSR value.
The type of the OSR value is inferred by running the typer before OSR
deconstruction, and then taking the type from the phi that takes the
OSR value. After the deconstruction, we throw the types away.
At the moment we only support the SignedSmall OSR type and we always
pick the tagged representation. Later, we might want to support more
types (such as Number) and pick better representations (int32/float64).
This CL also removes the OSR deconstruction tests because they build
unrealistic graph (no effect chain, no loop termination). I considered
adding the effect chains to the tests, but this would make the tests
even more brittle.
Review-Url: https://codereview.chromium.org/2384113002
Cr-Commit-Position: refs/heads/master@{#39971}
This makes sure we only replace load operations for fields on virtual
objects. Even though data flow information for non-virtual (escaping)
allocations is available, it might be inaccurate in certain situations
where object state hasn't been cleared.
R=jarin@chromium.org
TEST=mjsunit/compiler/regress-escape-analysis-indirect
Review-Url: https://codereview.chromium.org/2369953002
Cr-Commit-Position: refs/heads/master@{#39776}
Reason for revert:
Tanks EarleyBoyer.
Original issue's description:
> [compiler] Properly guard the speculative optimizations for instanceof.
>
> Add a general feedback slot for instanceof similar to what we already have
> for for-in, which basically has a fast (indicated by the uninitialized
> sentinel) and a slow (indicated by the megamorphic sentinel) mode. Now
> we can only take the fast path when the feedback slot says it hasn't
> seen any funky inputs and nothing funky appeared in the prototype chain.
> In the TurboFan code we also deoptimize whenever we see a funky object
> (i.e. a proxy or an object that requires access checks) in the prototype
> chain (similar to what Crankshaft already did).
>
> Drive-by-fix: Also make Crankshaft respect the mode and therefore
> address the deopt loop in Crankshaft around instanceof.
>
> We might want to introduce an InstanceOfIC mechanism at some point and
> track the map of the right-hand side.
>
> BUG=v8:5267
> R=mvstanton@chromium.org
>
> Committed: https://crrev.com/a0484bc6116ebc2b855de87d862945e2ae07169b
> Cr-Commit-Position: refs/heads/master@{#39718}
TBR=mvstanton@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=v8:5267
Review-Url: https://codereview.chromium.org/2365223003
Cr-Commit-Position: refs/heads/master@{#39736}
Add a general feedback slot for instanceof similar to what we already have
for for-in, which basically has a fast (indicated by the uninitialized
sentinel) and a slow (indicated by the megamorphic sentinel) mode. Now
we can only take the fast path when the feedback slot says it hasn't
seen any funky inputs and nothing funky appeared in the prototype chain.
In the TurboFan code we also deoptimize whenever we see a funky object
(i.e. a proxy or an object that requires access checks) in the prototype
chain (similar to what Crankshaft already did).
Drive-by-fix: Also make Crankshaft respect the mode and therefore
address the deopt loop in Crankshaft around instanceof.
We might want to introduce an InstanceOfIC mechanism at some point and
track the map of the right-hand side.
BUG=v8:5267
R=mvstanton@chromium.org
Review-Url: https://codereview.chromium.org/2370693002
Cr-Commit-Position: refs/heads/master@{#39718}
Extract String feedback on Add operation and utilize to lower ConsString
creation in JSTypedLowering when we know that a String addition will
definitely result in the creation of a ConsString.
Note that Crankshaft has to guard the potential length overflow of the
resulting string with an eager deoptimization exit, while we can safely
throw an exception in that case.
Also note that the bytecode pipeline does not currently provide the
String feedback for the addition, which has to be added.
BUG=v8:5267
R=jarin@chromium.org
Review-Url: https://codereview.chromium.org/2354853002
Cr-Commit-Position: refs/heads/master@{#39540}
This fixes the materialization of JSFunction objects to not rely on a
context being available. The context has been cleared because it might
be de-materiallized itself.
R=bmeurer@chromium.org
TEST=mjsunit/compiler/escape-analysis-materialize
BUG=chromium:644245
Review-Url: https://codereview.chromium.org/2320983002
Cr-Commit-Position: refs/heads/master@{#39277}
The trouble here is that the type of the induction variable might be
a bit ahead of the increment (JSAdd) operation's type. When we update
the type of the increment, we might only update the induction variable
type while the JSAdd type might be stale. If the induction variable typing
needs to fall back to normal phi typing (e.g., when the increment is not
an integer anymore), it might use the stale type.
To get around this, we fake monotonicity if we fallback to normal phi
typing. Another option would be to force re-typing of the increment
operation, but that seems to be harder to maintain.
BUG=chromium:644633
Review-Url: https://codereview.chromium.org/2320803002
Cr-Commit-Position: refs/heads/master@{#39261}
We were previously incorrectly changing:
sub r0, 0, r1
cmp r2, r0
b.cond <addr>
to:
cmn r2, r1
b.cond <addr>
for all conditions. This is incorrect for conditions involving the C (carry)
and V (overflow) flags, and in particular in the case where r1 = INT_MIN.
The optimization is still safe to perform for Equal and NotEqual since they
do not depend on the C and V flags.
BUG=
Review-Url: https://codereview.chromium.org/2318043002
Cr-Commit-Position: refs/heads/master@{#39246}
Migrate the isNaN, isFinite, Number.isFinite, Number.isInteger,
Number.isSafeInteger and Number.isNaN predicates to TurboFan
builtins and make them optimizable (for certain input types) in
JavaScript callees being optimized by TurboFan. That means both
the baseline and the optimized version is now always at maximum,
consistent performance. Especially TurboFan suffered from poor
baseline (and optimized) performance because it cannot play the
same weird tricks that Crankshaft plays for %_IsSmi.
This also adds a bunch of new tests to properly cover the use
of the Harmony predicates in optimized code.
R=franzih@chromium.org
BUG=v8:5049,v8:5267
Review-Url: https://codereview.chromium.org/2313073002
Cr-Commit-Position: refs/heads/master@{#39242}
Disable the propagation of truncations through Phi, Select or TypeGuard
if the output representation is tagged, because when the truncations are
taken we don't necessarily reflect this in the types and therefore we
might end up in a situation where we produce a word32 value, the type
says Number, and now we need to change that to tagged, which is not
possible since we don't know how to interpret the bits, i.e. whether the
value is Signed32 or Unsigned32.
BUG=chromium:644048
Review-Url: https://codereview.chromium.org/2311903002
Cr-Commit-Position: refs/heads/master@{#39186}
- Make constants more interesting.
- Add an addition to be done after the inlined call in the try-block.
- On command line, have a bit more output.
- New alternative that deopts from unoptimized code.
BUG=
R=jarin
Review-Url: https://codereview.chromium.org/2285743002
Cr-Commit-Position: refs/heads/master@{#38974}
These tests were spliced out of changelist 2216353002 and extended.
BUG=
Review-Url: https://codereview.chromium.org/2245263003
Cr-Commit-Position: refs/heads/master@{#38906}
Reason for revert:
Octane/Mandreel aborts with an exception now:
TypeError: __FUNCTION_TABLE__[(r2 >> 2)] is not a function
Original issue's description:
> [turbofan] Insert dummy values when changing from None type.
>
> Currently we choose the MachineRepresentation::kNone representation for
> values of Type::None, and when converting values from the kNone representation
> we use "impossible" conversions that will crash at runtime. This
> assumes that the impossible conversions should never be hit (the only
> way to produce the impossible values is to perform an always-failing
> runtime check on a value, such as Smi-checking a string). Note that
> this assumes that the runtime check is executed before the impossible
> convesrion.
>
> Introducing BitwiseOr type feedback broke this in two ways:
>
> - we always pick Word32 representation for bitwise-or, so the
> impossible conversion does not trigger (it only triggers with
> None representation), and we could end up with unsupported
> conversions from Word32.
>
> - even if we inserted impossible conversions, they are pure conversions.
> Since untagging, bitwise-or operations are also pure, we could hoist
> all these before the smi check of the inputs and we could hit the
> impossible conversions before we get to the smi check.
>
> This CL addresses this by just providing dummy values for conversions
> from the Type::None type. It also removes the impossible-to-* conversions.
>
> BUG=chromium:638132
>
> Committed: https://crrev.com/c83b21ab755f1420b6da85b3ff43d7e96ead9bbe
> Cr-Commit-Position: refs/heads/master@{#38883}
TBR=mstarzinger@chromium.org,jarin@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:638132
Review-Url: https://codereview.chromium.org/2280613002
Cr-Commit-Position: refs/heads/master@{#38893}
Currently we choose the MachineRepresentation::kNone representation for
values of Type::None, and when converting values from the kNone representation
we use "impossible" conversions that will crash at runtime. This
assumes that the impossible conversions should never be hit (the only
way to produce the impossible values is to perform an always-failing
runtime check on a value, such as Smi-checking a string). Note that
this assumes that the runtime check is executed before the impossible
convesrion.
Introducing BitwiseOr type feedback broke this in two ways:
- we always pick Word32 representation for bitwise-or, so the
impossible conversion does not trigger (it only triggers with
None representation), and we could end up with unsupported
conversions from Word32.
- even if we inserted impossible conversions, they are pure conversions.
Since untagging, bitwise-or operations are also pure, we could hoist
all these before the smi check of the inputs and we could hit the
impossible conversions before we get to the smi check.
This CL addresses this by just providing dummy values for conversions
from the Type::None type. It also removes the impossible-to-* conversions.
BUG=chromium:638132
Review-Url: https://codereview.chromium.org/2266823002
Cr-Commit-Position: refs/heads/master@{#38883}
Unfortunately, I was unable to produce a repro without asm.js. In normal
JavaScript, the bounds check renaming saves us.
I have not done anything about the index variable aliasing and handling
of differently sized elements yet!
BUG=chromium:639210, v8:5266
Review-Url: https://codereview.chromium.org/2270793004
Cr-Commit-Position: refs/heads/master@{#38874}
Unify the representation selection rules for NumberAdd/Subtract and
SpeculativeNumberAdd/Subtract wrt. Int32Add/Sub selection. We can
safely use Int32Add/Sub as long as the inputs are in the safe additive
integer range and the output is either truncated to Word32 or provably
in Signed32 or Unsigned32 range.
R=jarin@chromium.org
Review-Url: https://codereview.chromium.org/2253293005
Cr-Commit-Position: refs/heads/master@{#38746}
At the moment, two NumberConstant nodes get different type even if their
value is the same because we always allocate a new heap number for
each number constant. This can lead to replacing a node with a node of
disjoint type in value numbering, which can result in incorrect code
down the line because of inconsistent types.
This fix makes sure that we only replace a node with a sub-type
node. Once we introduce a proper type for number constants, we can
move back to the intersection typing in value numbering.
Unfortunately, it is quite hard to write a repro for this because we cache NumberConstant nodes. We only throw away cached values that have too many conflicts (>5), so the test has to contain values that fall into the same bucket. That's where the magic floating point numbers in the test come from (they have the same low 8-bits of their hashes).
BUG=chromium:633497
Review-Url: https://codereview.chromium.org/2251833002
Cr-Commit-Position: refs/heads/master@{#38675}
This adds a very first version of inlined Array.prototype.pop into
TurboFan optimized code. We currently limit the inlining to fast
object or smi elements, until the unclear situation around hole NaNs
is resolved and we have a clear semantics inside the compiler.
It's also probably overly defensive in when it's safe to inline
the call to Array.prototype.pop, but we can always extend that
later once we have sufficient trust in the implementation and see
an actual need to extend it.
BUG=v8:2229,v8:3952,v8:5267
R=epertoso@chromium.org
Review-Url: https://codereview.chromium.org/2239703002
Cr-Commit-Position: refs/heads/master@{#38578}
This switches the interface of the runtime profiler to use frames as
opposed to functions for performing on-stack replacement. Requests for
such replacements need to target a specific frame. This will enable us
to activate bytecode as well as baseline code for the same function.
The existing %OptimizeOsr runtime function also had to adapted and now
takes an optional stack depth to target a specific stack frame.
R=bmeurer@chromium.org
BUG=v8:4764
Review-Url: https://codereview.chromium.org/2230783004
Cr-Commit-Position: refs/heads/master@{#38548}
We now deopt when the lhs of a mod is negative and the rhs is 1 too (previously, we erroneusly returned 0 instead of -0).
BUG=v8:5278
R=bmeurer@chromium.org
Review-Url: https://codereview.chromium.org/2233713002
Cr-Commit-Position: refs/heads/master@{#38525}
If we infer loop variable bounds, we need to insert a type rename node
(sigma) to make sure that simplified lowering can choose representations
consistently.
Review-Url: https://codereview.chromium.org/2222513002
Cr-Commit-Position: refs/heads/master@{#38391}
This introduces a bunch of new tests that test various aspects of
accessor inlining in TurboFan (without the actual inlining), and does
the appropriate fixes to the AstGraphBuilder. The actual inlining CL
will land separately (so we don't need to revert the tests and fixes
if the accessor CL has to be reverted).
R=jarin@chromium.org
Review-Url: https://codereview.chromium.org/2197913002
Cr-Commit-Position: refs/heads/master@{#38191}
Drive-by fix: actually match the hint in the IsSpeculativeBinopMatcher.
Review-Url: https://codereview.chromium.org/2191883002
Cr-Commit-Position: refs/heads/master@{#38176}
This required the introduction of the CheckedNumberOrOddballAsWord32 use info, and a change in the RepresentationChanger to handle it.
BUG=
Review-Url: https://codereview.chromium.org/2184513003
Cr-Commit-Position: refs/heads/master@{#38086}
After multiplying two integers we emit code like:
if (result == 0) {
if (OR_OPERATION(rhs, lhs) < 0) {
DEOPT;
}
}
This CL allows us to eliminate the OR and comparison if either rhs or
lhs is a negative number, reducing the code to:
if (result == 0) DEOPT;
BUG=
Review-Url: https://codereview.chromium.org/2167643002
Cr-Commit-Position: refs/heads/master@{#38016}
Use the ForInFilterStub directly. Hence we will only jump to the runtime for
special receivers (instance_type <= LAST_SPECIAL_RECEIVER_TYPE) and for
converting element indices which are not in the string cache.
BUG=
Review-Url: https://codereview.chromium.org/2151773002
Cr-Commit-Position: refs/heads/master@{#37934}
This makes sure that we preserve call's tailness even if we have
introduced a loop exit between the call and the return.
BUG=chromium:628773
Review-Url: https://codereview.chromium.org/2155123002
Cr-Commit-Position: refs/heads/master@{#37832}
In int32 multiplication, if we have a positive integer as input, then we know we can't produce a -0 answer. The same is true if truncation is applied (x * y | 0). Without this information, we have to rather annoyingly check if the result of multiplication is 0, then OR the inputs to check for negativity, and possibly return -0. In TurboFan, we'll deopt in this case.
BUG=
Review-Url: https://codereview.chromium.org/2154073002
Cr-Commit-Position: refs/heads/master@{#37831}
Delaying for merges caused branch cloning using dummy effect phi inputs,
potentially splitting the effect chain at start.
We still have to delay the creation for loops because we need to break
cycles.
BUG=chromium:628403
Review-Url: https://codereview.chromium.org/2159603002
Cr-Commit-Position: refs/heads/master@{#37808}
This makes sure that the uses of PlainPrimitiveToNumber get a more
precise type (so that the uses know how to interpret the output
representation).
BUG=chromium:628516
Review-Url: https://codereview.chromium.org/2151223002
Cr-Commit-Position: refs/heads/master@{#37792}
Typed lowering now produces SpeculativeNumberShiftLeft for JSShiftLeft if the type feedback is kSignedSmall or kSigned32.
BUG=v8:4583
LOG=n
Review-Url: https://codereview.chromium.org/2150553002
Cr-Commit-Position: refs/heads/master@{#37762}
This drops the %_ValueOf intrinsic, but keeps the runtime entry
%ValueOf for now, by either migrating the functionality (mostly
Debug mirror or toString/valueOf methods) to C++ or TurboFan
builtins, or switching to the %ValueOf runtime call when it's
not performance critical anyways.
The %_ValueOf intrinsic was one of the last blockers for fixing
the unsound machine operator typing in TurboFan.
R=yangguo@chromium.org
BUG=v8:5049
Committed: https://crrev.com/293bd7882987f00e465710ce468bfb1eaa7d3fa2
Review-Url: https://codereview.chromium.org/2126453002
Cr-Original-Commit-Position: refs/heads/master@{#37512}
Cr-Commit-Position: refs/heads/master@{#37519}
Reason for revert:
[Sheriff] Breaks without i18n:
https://build.chromium.org/p/client.v8/builders/V8%20Linux%20-%20noi18n%20-%20debug/builds/8466
Original issue's description:
> [intrinsic] Drop the %_ValueOf intrinsic.
>
> This drops the %_ValueOf intrinsic, but keeps the runtime entry
> %ValueOf for now, by either migrating the functionality (mostly
> Debug mirror or toString/valueOf methods) to C++ or TurboFan
> builtins, or switching to the %ValueOf runtime call when it's
> not performance critical anyways.
>
> The %_ValueOf intrinsic was one of the last blockers for fixing
> the unsound machine operator typing in TurboFan.
>
> R=yangguo@chromium.org
> BUG=v8:5049
>
> Committed: https://crrev.com/293bd7882987f00e465710ce468bfb1eaa7d3fa2
> Cr-Commit-Position: refs/heads/master@{#37512}
TBR=yangguo@chromium.org,bmeurer@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=v8:5049
Review-Url: https://codereview.chromium.org/2117273002
Cr-Commit-Position: refs/heads/master@{#37514}
This drops the %_ValueOf intrinsic, but keeps the runtime entry
%ValueOf for now, by either migrating the functionality (mostly
Debug mirror or toString/valueOf methods) to C++ or TurboFan
builtins, or switching to the %ValueOf runtime call when it's
not performance critical anyways.
The %_ValueOf intrinsic was one of the last blockers for fixing
the unsound machine operator typing in TurboFan.
R=yangguo@chromium.org
BUG=v8:5049
Review-Url: https://codereview.chromium.org/2126453002
Cr-Commit-Position: refs/heads/master@{#37512}
The re-typer now only types a node if its inputs are all typed with the
exception of phi nodes. This works because all cycles in the graph have
to contain a phi node.
BUG=chromium:625558
Review-Url: https://codereview.chromium.org/2120243002
Cr-Commit-Position: refs/heads/master@{#37493}
Import fdlibm versions of acos, acosh, asin and asinh, which are more
precise and produce the same result across platforms (we were using
libm versions for asin and acos so far, where both speed and precision
depended on the operating system so far). Introduce appropriate TurboFan
operators for these functions and use them both for inlining and for the
generic builtin.
Also migrate the Math.imul and Math.fround builtins to TurboFan builtins
to ensure that their behavior is always exactly the same as the inlined
TurboFan version (i.e. C++ truncation semantics for double to float
don't necessarily meet the JavaScript semantics).
For completeness, also migrate Math.sign, which can even get some nice
love in TurboFan.
Drive-by-fix: Some alpha-sorting on the Math related functions, and
cleanup the list of Math intrinsics that we have to export via the
native context currently.
BUG=v8:3266,v8:3496,v8:3509,v8:3952,v8:5169,v8:5170,v8:5171,v8:5172
TBR=rossberg@chromium.orgR=franzih@chromium.org
Review-Url: https://codereview.chromium.org/2116753002
Cr-Commit-Position: refs/heads/master@{#37476}
The ARM64 instruction selector can generate code like this
negs w0, w1
b.vs deopt
but then reference the old value of w0 in the frame state, which will
obviously lead to wrong results.
R=jarin@chromium.org
BUG=v8:5158
Review-Url: https://codereview.chromium.org/2103793002
Cr-Commit-Position: refs/heads/master@{#37322}
This adds a dedicated test to make sure we don't try constant folding on
checks (in this case CheckTaggedPointer), which would generate invalid
code as we removing checks that guard the constant without knowing
whether it's safe to do so.
R=jarin@chromium.org
Review-Url: https://codereview.chromium.org/2087153002
Cr-Commit-Position: refs/heads/master@{#37168}
We cannot change x - y < 0 to x < y, because it would only be safe if
x - y cannot overflow, which we don't know in general.
R=jarin@chromium.org
BUG=v8:5129
Review-Url: https://codereview.chromium.org/2090493002
Cr-Commit-Position: refs/heads/master@{#37164}
This CheckBounds simplified operator is similar to the HBoundsCheck in
Crankshaft, and is hooked up to the new type feedback support in the
SimplifiedLowering. We use it to check the index bounds for keyed
property accesses.
Note to perf sheriffs: This will tank quite a few benchmarks, as the
operator makes some redundant branch elimination ineffective for
certain patterns of keyed accesses. This does require more serious
redundancy elimination, which we will do in a separate CL. So ignore
any regressions from this CL, we know there will be a few.
R=jarin@chromium.org
BUG=v8:4470,v8:5100
Committed: https://crrev.com/85e5567dae66a918500ae94c5568221137a0f5d4
Review-Url: https://codereview.chromium.org/2035893004
Cr-Original-Commit-Position: refs/heads/master@{#36947}
Cr-Commit-Position: refs/heads/master@{#37003}
This introduces SilenceNaN operator, which makes sure that we only
store quiet NaNs into holey arrays. We omit the NaN silencing code
at instruction selection time if the input is an operation that
cannot possibly produce signalling NaNs.
BUG=
Review-Url: https://codereview.chromium.org/2060233002
Cr-Commit-Position: refs/heads/master@{#36950}
This CheckBounds simplified operator is similar to the HBoundsCheck in
Crankshaft, and is hooked up to the new type feedback support in the
SimplifiedLowering. We use it to check the index bounds for keyed
property accesses.
Note to perf sheriffs: This will tank quite a few benchmarks, as the
operator makes some redundant branch elimination ineffective for
certain patterns of keyed accesses. This does require more serious
redundancy elimination, which we will do in a separate CL. So ignore
any regressions from this CL, we know there will be a few.
R=jarin@chromium.org
BUG=v8:4470,v8:5100
Review-Url: https://codereview.chromium.org/2035893004
Cr-Commit-Position: refs/heads/master@{#36947}
In Crankshaft we don't know reliably know that an HAdd might not turn
into a string addition later (via deoptimization), so we cannot set the
HValue::kAllowUndefinedAsNaN flag on the HAdd instruction in those
cases. It doesn't seem to affect performance if we just remove the flag
completely from the HAdd instruction, so let's stick to that approach
for now.
R=jarin@chromium.org
BUG=v8:5074
Review-Url: https://codereview.chromium.org/2048643002
Cr-Commit-Position: refs/heads/master@{#36805}
This introduces optimized number operations based on type feedback.
Summary of changes:
1. Typed lowering produces SpeculativeNumberAdd/Subtract for JSAdd/Subtract if
there is suitable feedback. The speculative nodes are connected to both the
effect chain and the control chain and they retain the eager frame state.
2. Simplified lowering now executes in three phases:
a. Propagation phase computes truncations by traversing the graph from uses to
definitions until checkpoint is reached. It also records type-check decisions
for later typing phase, and computes representation.
b. The typing phase computes more precise types base on the speculative types (and recomputes
representation for affected nodes).
c. The lowering phase performs lowering and inserts representation changes and/or checks.
3. Effect-control linearization lowers the checks to machine graphs.
Notes:
- SimplifiedLowering will be refactored to have handling of each operation one place and
with clearer input/output protocol for each sub-phase. I would prefer to do this once
we have more operations implemented, and the pattern is clearer.
- The check operations (Checked<A>To<B>) should have some flags that would affect
the kind of truncations that they can handle. E.g., if we know that a node produces
a number, we can omit the oddball check in the CheckedTaggedToFloat64 lowering.
- In future, we want the typer to reuse the logic from OperationTyper.
BUG=v8:4583
LOG=n
Review-Url: https://codereview.chromium.org/1921563002
Cr-Commit-Position: refs/heads/master@{#36674}
This prevents the compiler from optimizing
f64-to-tagged(tagged-to-f64(x)) ==> x
for non-number x (such as undefined).
Review-Url: https://codereview.chromium.org/2027593002
Cr-Commit-Position: refs/heads/master@{#36613}
Previously we first created a temporary graph for the inlinee and then
copied over all the nodes to the actual graph. This however introduces
unnecessary complexity, and we can instead just create the inlinee
inside the target graph.
R=jarin@chromium.org
Review-Url: https://codereview.chromium.org/2006353003
Cr-Commit-Position: refs/heads/master@{#36508}
This promotes the escape analysis from an experimental feature to be a
fully supported feature. The main goal is to unleach ClusterFuzz on the
implementation so that we can stabilize it.
R=jarin@chromium.org
Review-Url: https://codereview.chromium.org/1989833002
Cr-Commit-Position: refs/heads/master@{#36324}
This adds back the instanceof operator support in the backends and
introduces a @@hasInstance protector cell on the isolate that guards the
fast path for the InstanceOfStub. This way we recover the ~10%
regression on Octane EarleyBoyer in Crankshaft and greatly improve
TurboFan and Ignition performance of instanceof.
R=ishell@chromium.orgTBR=hpayer@chromium.org,rossberg@chromium.org
BUG=chromium:597249, v8:4447
LOG=n
Review-Url: https://codereview.chromium.org/1980483003
Cr-Commit-Position: refs/heads/master@{#36275}
We got the condition wrong and actually deoptimized when the typed array
was not neutered. This fixes the deopt loop in Math.random and actually
many programs that use typed arrays.
R=jarin@chromium.org
Review-Url: https://codereview.chromium.org/1970123002
Cr-Commit-Position: refs/heads/master@{#36194}
The flag in question used to fall-back to Crankshaft whenever an OSR
request couldn't be handled by TurboFan. By now OSR in TurboFan is
sufficiently stabilized that one single --use-osr flag should do it.
R=titzer@chromium.org
Review-Url: https://codereview.chromium.org/1960043002
Cr-Commit-Position: refs/heads/master@{#36102}
The problem is actually not related to try-catch, so here is a test
without try-catch.
BUG=chromium:607493
LOG=n
Review-Url: https://codereview.chromium.org/1943883002
Cr-Commit-Position: refs/heads/master@{#35985}
This prefixes the escape analysis flag with "experimental", thereby
making sure the flag in question is not being fuzzed. It will reduce
noise levels on ClusterFuzz again.
R=jarin@chromium.org
BUG=chromium:603653
LOG=n
Review URL: https://codereview.chromium.org/1894513002
Cr-Commit-Position: refs/heads/master@{#35521}
Now that ES2015 const has shipped, in Chrome 49, legacy const declarations
are no more. This lets us remove a bunch of code from many parts of the
codebase.
In this patch, I remove parser support for generating legacy const variables
from const declarations. This also removes the special "illegal declaration"
bit from Scope, which has ripples into all compiler backends.
Also gone are any tests which relied on legacy const declarations.
Note that we do still generate a Variable in mode CONST_LEGACY in one case:
function name bindings in sloppy mode. The likely fix there is to add a new
Variable::Kind for this case and handle it appropriately for stores in each
backend, but I leave that for a later patch to make this one completely
subtractive.
Review URL: https://codereview.chromium.org/1819123002
Cr-Commit-Position: refs/heads/master@{#35002}
This CL adds support for builtins with JavaScript linkage written using
the TurboFan CodeStubAssembler, but with a JSCall descriptor (which was
already supported thanks to a previous patch by Ben Smith). As a first
example, we convert the Math.sqrt builtin and thereby get rid of the
%_MathSqrt intrinsic, which causes trouble for the representation
selection pass in the JavaScript pipeline.
R=mstarzinger@chromium.org
Review URL: https://codereview.chromium.org/1824993002
Cr-Commit-Position: refs/heads/master@{#34989}
Moves the accumulator value on-heap to be restored in the
InterpreterNotifyDeopt handler rather than explicitly
setting the accumulator register. This allows it to be
materialized correctly if required.
BUG=v8:4678
LOG=N
Review URL: https://codereview.chromium.org/1707133003
Cr-Commit-Position: refs/heads/master@{#34113}
This removes support for the %Arguments and %ArgumentsLength runtime
entries and their intrinsic counterparts. If you need variable arguments
in any builtin, either use (strict) arguments object or rest parameters,
which are both compositional across inlining (in TurboFan), and not that
much slower compared to the %_Arguments hackery.
R=jarin@chromium.org
Review URL: https://codereview.chromium.org/1688163004
Cr-Commit-Position: refs/heads/master@{#33943}
The idea here is to perform the handler lookup in the deoptimizer, and then take the information from the handler table to build the catch handler frame in the deoptimizer. Specifically, we use the pc offset, context location and stack height (in full-code) to tweak the output frame.
Sadly, this still requires nasty voodoo for the liveness analyzer so that it keeps variables alive if they are used in the catch handler.
Review URL: https://codereview.chromium.org/1416543006
Cr-Commit-Position: refs/heads/master@{#33936}
Reason for revert:
No fix needed, original CL was perfectly fine!
Original issue's description:
> Revert of [interpreter] Correctly thread through catch prediction. (patchset #1 id:1 of https://codereview.chromium.org/1690973002/ )
>
> Reason for revert:
> Depends on the reverted https://codereview.chromium.org/1691723002
>
> Original issue's description:
> > [interpreter] Correctly thread through catch prediction.
> >
> > This change correctly sets the {CatchPrediction} field in exception
> > handler tables for bytecode and optimized code. It also adds tests
> > independent of promise handling for this prediction, to ensure all our
> > backends are in sync on their prediction.
> >
> > R=rmcilroy@chromium.org,yangguo@chromium.org
> > TEST=mjsunit/compiler/debug-catch-prediction
> > BUG=v8:4674
> > LOG=n
> >
> > Committed: https://crrev.com/ba55f5594cb0b4a1a1e9b35d87fe54afe2d93f3b
> > Cr-Commit-Position: refs/heads/master@{#33906}
>
> TBR=rmcilroy@chromium.org,yangguo@chromium.org,mstarzinger@chromium.org
> # Skipping CQ checks because original CL landed less than 1 days ago.
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=v8:4674
>
> Committed: https://crrev.com/c5229b311968fd638a6cd537c341b1055eb7be97
> Cr-Commit-Position: refs/heads/master@{#33922}
TBR=rmcilroy@chromium.org,yangguo@chromium.org,adamk@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=v8:4674
Review URL: https://codereview.chromium.org/1689113004
Cr-Commit-Position: refs/heads/master@{#33933}
Reason for revert:
Depends on the reverted https://codereview.chromium.org/1691723002
Original issue's description:
> [interpreter] Correctly thread through catch prediction.
>
> This change correctly sets the {CatchPrediction} field in exception
> handler tables for bytecode and optimized code. It also adds tests
> independent of promise handling for this prediction, to ensure all our
> backends are in sync on their prediction.
>
> R=rmcilroy@chromium.org,yangguo@chromium.org
> TEST=mjsunit/compiler/debug-catch-prediction
> BUG=v8:4674
> LOG=n
>
> Committed: https://crrev.com/ba55f5594cb0b4a1a1e9b35d87fe54afe2d93f3b
> Cr-Commit-Position: refs/heads/master@{#33906}
TBR=rmcilroy@chromium.org,yangguo@chromium.org,mstarzinger@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=v8:4674
Review URL: https://codereview.chromium.org/1695613002
Cr-Commit-Position: refs/heads/master@{#33922}
This change correctly sets the {CatchPrediction} field in exception
handler tables for bytecode and optimized code. It also adds tests
independent of promise handling for this prediction, to ensure all our
backends are in sync on their prediction.
R=rmcilroy@chromium.org,yangguo@chromium.org
TEST=mjsunit/compiler/debug-catch-prediction
BUG=v8:4674
LOG=n
Review URL: https://codereview.chromium.org/1690973002
Cr-Commit-Position: refs/heads/master@{#33906}
This adds test cases for exception handlers that require a context
switch when entering the catch-block or the finally-block, triggered
through nested contexts within the try-block.
R=jarin@chromium.org
Review URL: https://codereview.chromium.org/1681933002
Cr-Commit-Position: refs/heads/master@{#33845}
By now only the default %TypedArray%.prototype.sort compare function
and the JS implementation of SameValueZero were still using the odd
%_IsMinusZero intrinsic, whose semantics both included a number check
(actually HeapNumber test) plus testing if the heap number stores the
special -0 value. In both cases we already know that we deal with
number so we can reduce it to a simple number test for -0, which can
be expressed via dividing 1 by that value and checking the sign of
the result. In case of the compare function, we can be even smarter
and work with the reciprocal values in case x and y are equal to 0
(although long term we should probably rewrite the fast case for
the typed array sorting function in C++ anyway, which will be way,
way faster than our handwritten callback-style, type-feedback
polluted JS implementation).
R=yangguo@chromium.org
Review URL: https://codereview.chromium.org/1680783002
Cr-Commit-Position: refs/heads/master@{#33833}
Avoid the hacking in JSIntrinsicLowering and provide a proper simplified
operator ObjectIsReceiver instead that is used to implement %_IsJSReceiver
which is used by our JavaScript builtins and the JSInliner.
R=jarin@chromium.org
BUG=v8:4544
LOG=n
Review URL: https://codereview.chromium.org/1657863004
Cr-Commit-Position: refs/heads/master@{#33675}
The previous versions of Math.max and Math.min made it difficult to
optimize those (that's why we already have custom code in Crankshaft),
and due to lack of ideas what to do about the variable number of
arguments, we will probably need to stick in special code in TurboFan
as well; so inlining those builtins is off the table, hence there's no
real advantage in having them around as "not quite JS" with extra work
necessary in the optimizing compilers to still make those builtins
somewhat fast in cases where we cannot inline them (also there's a
tricky deopt loop in Crankshaft related to Math.min and Math.max, but
that will be dealt with later).
So to sum up: Instead of trying to make Math.max and Math.min semi-fast
in the optimizing compilers with weird work-arounds support %_Arguments
%_ArgumentsLength, we do provide the optimal code as native builtins
instead and call it a day (which gives a nice performance boost on some
benchmarks).
R=jarin@chromium.org
Review URL: https://codereview.chromium.org/1641083003
Cr-Commit-Position: refs/heads/master@{#33582}
This patch moves the semantics of 'const' in sloppy mode to match those
in strict mode, that is, const makes lexical (let-like) bindings, must
have an initializer, and does not create properties of the global object.
R=adamk
LOG=Y
BUG=v8:3305
CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_chromium_rel_ng;tryserver.blink:linux_blink_rel
Review URL: https://codereview.chromium.org/1571873004
Cr-Commit-Position: refs/heads/master@{#33218}
Adds support for the CallRuntimeForPair bytecode to the Bytecode Graph
Builder. Modifies the FrameState support to allow updating of output
registers.
Also adds Eval tests to test-run-bytecode-graph-builder since these are
enabled by CallRuntimeForPair support.
BUG=v8:4280
LOG=N
Review URL: https://codereview.chromium.org/1570623007
Cr-Commit-Position: refs/heads/master@{#33186}
Deopt support is added on two levels. On the IR level,
a new ObjectState node is added, which represenents an
object to be materialized. ObjectState nodes appear as
inputs of FrameState and StateValues nodes. On the
instruction select/code-generation level, the
FrameStateDescriptor class handles the nesting
introduced by ObjectState, and ensures that deopt code
with CAPTURED_OBJECT/DUPLICATED_OBJECT entries are
generated similarly to what crankshaft's escape
analysis does.
Two unittests test correctness of the IR level implementation.
Correctness for instruction selection / code generation
is tested by mjsunit tests.
R=jarin@chromium.org,mstarzinger@chromium.org
BUG=v8:4586
LOG=n
Review URL: https://codereview.chromium.org/1485183002
Cr-Commit-Position: refs/heads/master@{#33115}
Bugfixes and improvements in escape analysis include:
* Handling of ObjectIsSmi (non-escaping)
* Handling of nested phi replacements
* Handling of phis with arity > 2
* Resilience against effectful nodes dangling from start
* Allocations escape now, if non-const load/store is performed
* Fixed a bug where non-allocated objects where tracked
* Allow fixed double arrays to be tracked
R=mstarzinger@chromium.org
BUG=v8:4586
LOG=n
Review URL: https://codereview.chromium.org/1510973006
Cr-Commit-Position: refs/heads/master@{#32833}
This patch improves escape analysis and fixes bugs
triggered by clusterfuzz. Impovements include:
* Handling of LoadElement/StoreElement if index is a
constant
* Handling of JSStoreProperty: invalidate all information,
as the store could have altered any field.
* Treat phis that use an allocation as escaping
* Improve resolution of replacements
R=mstarzinger@chromium.org
BUG=v8:4586
LOG=n
Review URL: https://codereview.chromium.org/1499143002
Cr-Commit-Position: refs/heads/master@{#32656}
Reason for revert:
Blocks the roll:
https://codereview.chromium.org/1497763004/
Original issue's description:
> [es6] Correctify and unify ArrayBuffer and SharedArrayBuffer constructors.
>
> The ArrayBuffer and SharedArrayBuffer constructors should raise an
> exception when called with no arguments or undefined length. Also
> unified the ArrayBuffer and SharedArrayBuffer implementations as C++
> builtins, and removed some (now) obsolete runtime entries.
>
> R=yangguo@chromium.org
>
> Committed: https://crrev.com/3235ccbb7826ceec2188f6ebab98fc851b54f60e
> Cr-Commit-Position: refs/heads/master@{#32590}
TBR=yangguo@chromium.org,cbruni@chromium.org,adamk@chromium.org,bmeurer@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
Review URL: https://codereview.chromium.org/1501673002
Cr-Commit-Position: refs/heads/master@{#32606}
