Turns out this was hurting performance instead of helping it.
By removing the inline capacity, we shrink the size of ExecutionContext
by 512 bytes, which substantially reduces the stack pressure created by
JS recursion (each call creates a new ExecutionContext on the stack).
4.4% speed-up on the entire Kraken benchmark :^)
Previously every file that included Executable.h (which is pretty much
most LibJS and LibHTML files, given that VM.h needs it) had the whole
definition of LibRegex, which was slowing down source parsing.
When we hit the cache in GetGlobal, we don't need the identifier string
at all, so let's defer fetching it until after the cache miss.
7% speed-up on Kraken/imaging-gaussian-blur.js :^)
If we have a cached environment coordinate that hasn't been screwed
by eval(), we can get the value directly without instantiating a
Reference.
15% speed-up on Octane/zlib.js :^)
The functions for registering and unregistering MarkedVector, Handle,
etc. were quite prominent in benchmark profiles.
4% speed-up on the entire Kraken benchmark :^)
(including: 7% speed-up on Kraken/imaging-gaussian-blur.js, the current
slowest subtest)
These functions all have a very common case that can be dealt with a
very simple inline check, often avoiding the need to call an out-of-line
function. This patch moves the common case to inline functions in a new
ValueInlines.h header (necessary due to header dependency issues..)
8% speed-up on the entire Kraken benchmark :^)
For example, the locale "fr-FR" will have the preferred hour cycle list
of "H hB", meaning h23 and h12-with-day-periods. Whether date-times are
actually formatted with day-periods is up to the user, but we need to
parse the hour cycle as h12 to know that the FR region supports h12.
This bug was revealed by LibJS no longer blindly falling back to h12 (if
the `hour12` option is true) or h24 (if the `hour12` option is false).
These are normative changes in the ECMA-402 spec. See:
896ffccaf4ec46e25c455
(This combines the above commits into one patch as they each do not work
on their own).
This is an editorial change in the ECMA-262 spec. See:
73926a5
The idea here is to reduce duplication of these AOs between ECMA-262,
ECMA-402, and Temporal. This patch contains only the ECMA-262 changes.
We currently only return primary time zones, i.e. time zones that are
not a Link. LibJS will require knowledge of Link entries, and whether
each entry is or is not a Link.
This patch adds a fast path to the PutByValue bytecode op that bypasses
a ton of things *if* a set of assumptions hold:
- The property key must be a non-negative Int32
- The base object must not interfere with indexed property access
- The base object must have simple indexed property storage
- The property key must already be present as an own property
- The existing value must not have any accessors defined
If this holds (which it should in many common cases), we can skip all
kinds of checks and poke directly at the property storage, saving time.
16% speed-up on the entire Kraken benchmark :^)
(including: 88% speed-up on Kraken/imaging-desaturate.js)
(including: 55% speed-up on Kraken/audio-fft.js)
(including: 54% speed-up on Kraken/audio-beat-detection.js)
This patch adds a fast path to the GetByValue bytecode op that bypasses
a ton of things *if* a set of assumptions hold:
- The property key must be a non-negative Int32
- The base object must not interfere with indexed property access
- The property key must already be present as an own property
- The existing value must not have any accessors defined
If this holds (which it should in the common case), we can poke directly
at the indexed property storage and save a boatload of time.
10% speed-up on the entire Kraken benchmark :^)
(including: 31% speed-up on Kraken/audio-dft.js)
(including: 23% speed-up on Kraken/stanford-crypto-aes.js)
This function must return true if the object may intercept and customize
access to indexed properties (properties where the property name is a
non-negative integer.)
This will be used to implement fast path optimizations for array-like
accesses in subsequent commits.
The strings will get deduplicated when actually turned into
PrimitiveString objects at runtime anyway, and keeping the string
tables deduplicated was actually wasting a lot of time.
4.4% speed-up on Kraken/stanford-crypto-ccm.js :^)
The following snippet would cause "i" to be incremented twice(!):
let a = []
let i = 0
a[++i] += 0
This patch solves the issue by remembering the base object and property
name for computed MemberExpression LHS in codegen. We the store the
result of the assignment to the same object and property (instead of
computing the LHS again).
3 new passes on test262. :^)
When the spec says to call "! TrimString", we should use MUST instead
of TRY. (We were previously using TRY in order to propagate OOM errors,
but we don't care about such OOMs anymore.)
- Convert to FlatPtr instead of doing pointer arithmetic on a too-large
pointer type in find_min_and_max_block_addresses(). This makes the
range more accurate.
- Untag possible cell pointers in add_possible_value() before doing the
rejection. Otherwise we end up rejecting most pointers since the tags
sit in the highest bits!
This fixes a crash when running the Speedometer benchmark.
Just using Vector::resize() meant that we allocated exact capacity
instead of leaving padding at the end. This patch adds a call to
grow_capacity() before resize(), which ensures that we grow with the
usual extra padding.
This change adds a check to discard pointers that are lower than the
minimum address of all allocated blocks or higher than the maximum
address of all blocks. By doing this we avoid executing plenty of set()
operations on the HashMap in the add_possible_value().
With this change gather_conservative_roots() run 10x times faster in
Speedometer React-Redux-TodoMVC test.
We can use `ensure_capacity` for binding vectors if we know their sizes
in advance. This ensures that binding vectors aren't reallocated during
the `function_declaration_instantiation` execution.
With this change, `try_grow_capacity()` and `shrink_to_fit()` are no
longer visible in the `function_declaration_instantiation()` profiles
when running React-Redux-TodoMVC from Speedometer.
If any of binding pattern entry's name is expession
`contains_expression()` should return true.
For example:
```js
function evalInComputedPropertyKey(
{[eval("var x = 'inner'")]: ignored}
) {}
```
`contains_expression()` should return true for the binding param in
this function.
We should initialize jump targets when constructing the jump instruction
instead of doing it later. This was already the case in all construction
sites but one. This first patch converts all those sites to pass final
targets to the constructor directly.
This change makes LibJS correctly report a syntax error when a unary
expression is followed by exponentiation, as the spec requires.
Apparently this is due to that expression being ambiguous ordering.
Strangely this check does not seem to apply in the same way for '++' and
'--' for reasons that I don't fully understand. For example
```
let x = 5;
++x ** 2
```
Since `--5` and `++5` on it's own results in a syntax error anyway, it
seems we do not need to perform this exponentiation check in those
places.
Diff Tests:
+6 ✅ -6 ❌
This reduces the minimum size of a basic block from 4 KiB to 0 bytes.
With this change, memory usage at the end of Speedometer is 1.2 GiB,
down from 1.8 GiB.
Instead of calling out to helper functions for flow control (and then
checking control flags on every iteration), we now simply inline those
ops in the interpreter loop directly.
If we don't have a local unwind context to handle the exception, we can
just return right away. This allows us to remove one check from the
inner loop.
Nuke all the per-instruction bounds checking when iterating instructions
by using raw pointers instead of indexing into a ReadonlyBytes.
The interpreter loop already checks that we're in-bounds anyway.
