[Webkit-unassigned] [Bug 44329] SH4 JIT SUPPORT
bugzilla-daemon at webkit.org
bugzilla-daemon at webkit.org
Fri Oct 15 04:56:31 PDT 2010
https://bugs.webkit.org/show_bug.cgi?id=44329
--- Comment #27 from Gabor Loki <loki at webkit.org> 2010-10-15 04:56:30 PST ---
> We can work on 64-bit , after the 32-bit port is accepted.
There are not so much changes to get JSValue32_64 working (if you have a JSValue32 version).
> The common solution should work fine, but with our solution we reduce the number of instructions.
Well, we have to sacrifice some performance to have a common masm and jit. Unfortunately the same thing is true for each architecture.
> In the common code, there are a lot of branches and the offsets are known when linking.
> when branching to solwcase, the offset is not reachable, so we need to load it in the constantPool, otherwise there is no need. that's why I added in branch functions a parameter to check if we need to load a constant or not.
On ARMv5 and below the situation is the same, there are lots of pc relative loads which read a pointer or a big number. There are only very few cases when we are able to encode a constant into an instruction (most of those constants appear in the arithmetic part of jit).
--
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.
More information about the webkit-unassigned
mailing list