[Webkit-unassigned] [Bug 28886] Build error using JIT with gcc 4.1.2 and ARM v5

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=28886





--- Comment #23 from Andre Pedralho <apedralho at gmail.com>  2009-09-14 12:01:56 PDT ---
Sorry by the delay!

(In reply to comment #19)
> I think Andre Pedralho or someone else should not only check that it compiles
> and links, but also that the __clear_cache actually clears the instruction
> cache. As it was pointed in the Android related discussion
> (http://code.google.com/p/android/issues/detail?id=1803) the __clear_cache call
> could be a no-op, which would than break the JIT functionality.
>
I could run it using strace and it calls both operation (the g++ __clear_cache
and th system __asm __volatile...) defined in
http://code.google.com/p/android/issues/detail?id=1803

cacheflush(0, 0, 0, 0x1, 0x40022db0)    = 0
cacheflush(0, 0, 0, 0xf0002, 0x40022db0) = 0

> Also maybe WebKit can do a bit better guarding the __clear_cache call and the
> fallback assembly code. Fill post a patch for consideration.


(In reply to comment #20)
> Created an attachment (id=39430)
 --> (https://bugs.webkit.org/attachment.cgi?id=39430) [details]
> Be more specific with guards.
> 
> 2./ It might be better to use the CLEAR_INSN_CACHE guard to test if
> __clear_cache is available than testing for a specific GCC version. Andre can
> you see if CLEAR_INSN_CACHE is defined in your environment.
>
No, it is not defined. So I don't know if __clear_cache is really clearing
cache. The code below builds fine:

#ifdef CLEAR_INSN_CACHE
    xxxxxx
#endif

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