[Webkit-unassigned] [Bug 44329] SH4 JIT SUPPORT

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=44329





--- Comment #8 from thouraya <thouraya.andolsi at st.com>  2010-09-16 03:40:44 PST ---
Hello,

(In reply to comment #7)
> > Created an attachment (id=67675)
 --> (https://bugs.webkit.org/attachment.cgi?id=67675) [details] [details]
> 
> Hmmm, you have implemented your own constant/literal pool for SH4. Why? Is the AssemblerBufferWithConstantPool is not sufficent for SH4?
> 
> As I see you follow a very similar way to attach the data and pointers into the instruction stream as we have done in AssemblerBufferWithConstantPool. At first sight I did not see any reason why the SH4 should have its own pool implementation. I suggest you try to use the AssemblerBufferWithConstantPool interface for SH4. If you have any problem or question about the interface, you can contact me at IRC.

We implemented our constant pool before you add AssemblerBufferWithConstantPool.

I'll try to use it and let you know.

> 
> In addition, I agree with Patrick about the style issue:
> - use static const ints, enums and inline functions instead of macros
> - fix tabs, unnecessary parenthesis and other trivial style errors which are reported by the check-webkit-style script.
> 
> One more question:
> What was the problem with the generatePatternCharacterPair function? Why do you have to read a simple character? (I am not familiar with the SH4 architecture)

We have to read a simple character because the address should be correctly aligned for a long-word otherwise we'll get a misaligned address.

Regards,
Thouraya.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.