[Webkit-unassigned] [Bug 162705] Offline asm should not output masm assembly when using a x86_64 asm backend

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Thu Sep 29 15:47:16 PDT 2016


https://bugs.webkit.org/show_bug.cgi?id=162705

--- Comment #14 from Mark Lam <mark.lam at apple.com> ---
(In reply to comment #13)
> (In reply to comment #12)
> 
> Thanks for your feedback Mark, 
> 
> > It looks like this change is the only one that is of consequence.  All the
> > other changes below only change the code to use a local var (initialized by
> > a function) instead of a global var.  Are they really necessary?  Or did I
> > misread something?
> 
> What was happening there was that the activebackend was not being set until
> later on in the code when some node operations are done in backends.rb. In
> ruby, class and instance level variables are resolved when the file is
> included which was causing the generate offsets script to fail too.

OK.  Sounds good.

> > 2. change isMSVC to be based on isWin.
> The same problem as above is seen here, isWin is called from asm.rb before
> the backend is set. For wincairo and other platforms that use cmake, the
> correct file name is set in cmake which causes emitWinAsm to be set
> correctly. DWARF2 annotations will still be incorrectly set for cross
> compiling to a x86_(_64)_win backend. It would be a nice change to have but
> it is much more involved. 
> 
> I agree with removing isGCC, I can make a patch for that. Let me know if you
> think it's still worthwhile to remove isMSVC.

You can just remove isGCC in this patch.  Add a comment in the ChangeLog to say that it is unused.

You should not remove isMSVC.  I think you should update it to use isWin like you did for isIntelSyntax.

You should remove isCompiledForWindows since it will be now unused after the above changes.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20160929/26f1e514/attachment.html>


More information about the webkit-unassigned mailing list