Bug 226264 – [JSC] cloop again broken on 32-bit big endian in 2.32.x

• Status: RESOLVED
• Resolution: FIXED
• Component: JavaScriptCore

[Nina Kolesa]

It seems a change was missed. Patch incoming.

[Nina Kolesa]

Created attachment 429747 [details]
patch

patch

[Caio Lima]

Comment on attachment 429747 [details]
patch

r=me

[Caio Lima]

I’m a bit surprised that we are running JSC on 32-bits big endian platforms. Maybe we should add a post-commit buildbot at some point? There’s a couple of places I worked on 32-bits LLInt where I considered little-endian only, so we might have other bugs like that around. Also, without test feedback, it will be quite likely that we will introduce bugs like that in the future.

[Nina Kolesa]

i'll definitely run the tests at one point (on powerpc 64le, 64be, 32be) though at this point the big endian platforms are a best-effort thing in my distribution, so i'm not able to invest more time than right now into them...

but yeah things are known to work okay currently (at very least, it never crashes the web process, there may be buggy javascript on some sites - e.g. google maps js throws exceptions when trying to render tiles right now - but some quite complex webapps are known to work okay)

[Adrian Perez]

Should we cq+ this?

[Adrian Perez]

Comment on attachment 429747 [details]
patch

Setting cq+ to land in trunk, so I can backport it to the release branch

[EWS]

Committed r278157 (238201@main): <https://commits.webkit.org/238201@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 429747 [details].

[Radar WebKit Bug Importer]

<rdar://problem/78560223>