JIT simplifications after JSScript/LazyScript unification
Categories
(Core :: JavaScript Engine, enhancement, P2)
Tracking
()
| Tracking | Status | |
|---|---|---|
| firefox76 | --- | fixed |
People
(Reporter: tcampbell, Assigned: tcampbell)
References
Details
Attachments
(2 files)
A number of comments and helper functions can be simplified in IonMonkey now that LazyScript has a jitCodeRaw_. See review comments on https://phabricator.services.mozilla.com/D38255
|
Assignee | |
Comment 1•4 years ago
|
||
Re-purposing this bug a bit...
We should also fix MacroAssembler::loadJitCodeMaybeNoArgCheck which no longer needs to worry about lambda clones being lazy when the template is not. This method no longer needs to be "Maybe".
|
|
Assignee | |
Comment 2•4 years ago
|
||
These comments refer to concerns from when lambda clones might still be
pointing to a lazy script. Now, the ICs will be tossed before relazification
happens and if we observed any clone having bytecode, then they all will
agree.
|
|
Assignee | |
Comment 3•4 years ago
|
||
Ion only uses this entry for known targets, but previously had to worry about
lambda clones still having a pointer to lazy script. With unified scripts, we
know all clones of same target will have bytecode. On relazification the Ion
code will be tossed as well.
Depends on D69206
Pushed by tcampbell@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/eebbf5a0f70e Remove stale jit comments about relazification. r=jandem https://hg.mozilla.org/integration/autoland/rev/beb1529f957d Make no-arg-check JIT entry unconditional. r=jandem
|
||
Comment 5•4 years ago
|
||
| bugherder | ||
https://hg.mozilla.org/mozilla-central/rev/eebbf5a0f70e
https://hg.mozilla.org/mozilla-central/rev/beb1529f957d
Description
•