1879997 - Use __builtin_clz or std::countl_one in InflateUTF8ToUTF16

Status: RESOLVED FIXED

(Core :: JavaScript Engine, enhancement, P3)

Description

[Tooru Fujisawa [:arai]]

https://phabricator.services.mozilla.com/D201330#inline-1116243

https://searchfox.org/mozilla-central/rev/471301f58df4e9aa3496ce387d2340af5e3502e4/js/src/vm/CharacterEncoding.cpp#278-282

// Non-ASCII code unit.  Determine its length in bytes (n).
uint32_t n = 1;
while (v & (0x80 >> n)) {
  n++;
}

The code counts the number of leading 1 bits, as part of decoding UTF-8 code units.

Either of the following can be used:
https://gcc.gnu.org/onlinedocs/gcc/Other-Builtins.html#index-_005f_005fbuiltin_005fclz
https://en.cppreference.com/w/cpp/numeric/countl_one

Comment 1

[Mayank Bansal]

There is also this: https://github.com/simdutf/simdutf

Comment 2

[Jon Coppeard (:jonco)]

There's an MFBT function for this: CountLeadingZeroes32

Comment 3

[Steve Fink [:sfink] [:s:]]

Attachment: Bug 1879997 - use CountLeadingZeroes32

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/695de01232f0
use CountLeadingZeroes32 r=spidermonkey-reviewers,arai

Comment 5

[Cosmin Sabou [:CosminS]]

Backed out for SM assertion failures on MathAlgorithms.

Push with failures: https://treeherder.mozilla.org/jobs?repo=autoland&resultStatus=retry%2Ctestfailed%2Cbusted%2Cexception%2Cusercancel&revision=695de01232f0b2d0f75cd8f82b2588a1c638a9ac

Failure log: https://treeherder.mozilla.org/logviewer?job_id=447783641&repo=autoland

Backout link: https://hg.mozilla.org/integration/autoland/rev/53f7baad39fc2408bc5c1ab6df97372a395730a4

Comment 6

[Steve Fink [:sfink] [:s:]]

I uploaded a fix and requested re-review.

Comment 7

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/533127adaec7
use CountLeadingZeroes32 r=spidermonkey-reviewers,arai

Comment 8

[Natalia Csoregi [:nataliaCs]]

https://hg.mozilla.org/mozilla-central/rev/533127adaec7