[cfe-dev] Patch: Add support for pusha/popa

[Chris Lattner]

On Jun 22, 2010, at 10:34 PM, Nico Weber wrote:

> Done.

Looks good to me, please commit (you do have commit access to llvm as well as clang) with two comment typo fixes:

+      // ignore (pushes all GP registoers onto the stack)
..
+      // ignore (pushes all GP registoers onto the stack)

Thanks Nico!

-Chris

> 
> On Tue, Jun 22, 2010 at 10:17 PM, Daniel Dunbar <daniel at zuster.org> wrote:
>> Hi Nico,
>> 
>> This looks fine to me, but please add a test case to x86_32-new-encoder.s or so.
>> 
>>  - Daniel
>> 
>> On Tue, Jun 22, 2010 at 9:56 PM, Nico Weber <thakis at chromium.org> wrote:
>>> Hi,
>>> 
>>> the attached patch adds support for "pusha" and "popa" in asm blocks.
>>> It's strictly speaking an llvm patch, but I'm not on llvm-dev.
>>> 
>>> The binary produced by clang with for the two instructions matches gcc
>>> in i386 mode. In x86_64 mode, gcc correctly points out that the
>>> instruction is 32-bit only, while clang doesn't do this yet. nicholas
>>> on irc says that that's probably because clang doesn't check for
>>> 32/64bit mismatches yet for inline assembly.
>>> 
>>> Let me know what you think.
>>> 
>>> Thanks,
>>> Nico
>>> 
>>> _______________________________________________
>>> cfe-dev mailing list
>>> cfe-dev at cs.uiuc.edu
>>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
>>> 
>>> 
>> 
> <llvm-pusha.patch>_______________________________________________
> cfe-dev mailing list
> cfe-dev at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev