[Mono-bugs] [Bug 324134] Add support for Mac OS X / ppc64

[bugzilla_noreply at novell.com]

https://bugzilla.novell.com/show_bug.cgi?id=324134

User andreas.faerber at web.de added comment
https://bugzilla.novell.com/show_bug.cgi?id=324134#c41





--- Comment #41 from Andreas Färber <andreas.faerber at web.de>  2008-10-26 03:33:59 MDT ---
Concerning the state of the patchset and its testability, it has suffered a
little under the Linear IR introduction - we were in a debugging stage before
(yes, for the JIT, not the interpreter), now first code handling the new long
opcodes needs to be added before we get there again (assertion failures for
unhandled opcodes at bootstrap).

Some guidance on how to best handle the varying code sizes in mini would be
helpful, too. Geoff didn't like my local code size defines but didn't come up
with a better suggestion. Such things could be decided and committed for
ppc(32) before any of the ppc64 stuff to gradually reduce the ppc64 patch size.

I recently had some doubts about my ppc_ld/ppc_std macros: The intructions
require word-aligned addresses and don't store the two trailing zero bits.
Currently, my macros accept the value to be serialized, requiring <<2 in lots
of places in mini. I would propose that I change my macro to accept the real
value, optionally assert its alignment, and do a mask there. What do you think?

Regarding the atomic builtins, I understood Olaf in such a way that those were
generic helpers introduced by GCC 4.1+ and not specific to ppc(64). If that
were correct, it might relieve the Mono team of adding and maintaining atomic
functions for newer platforms. Just as a clarification of my reasons to apply
his patch to my ppc64 branch, orthogonal to my own changes.


-- 
Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.