[cfe-commits] r102019 - in /cfe/trunk: lib/CodeGen/TargetInfo.cpp test/CodeGen/x86_32-arguments.c test/CodeGen/x86_64-arguments.c

Daniel Dunbar daniel at zuster.org
Fri Apr 23 14:25:10 PDT 2010


On Wed, Apr 21, 2010 at 9:09 PM, Chris Lattner <clattner at apple.com> wrote:
>
> On Apr 21, 2010, at 12:49 PM, Daniel Dunbar wrote:
>
>> +// CHECK: define void @f20(%struct.s20* byval align 32 %x)
>> +struct __attribute__((aligned(32))) s20 {
>> +  int x;
>> +  int y;
>> +};
>> +void f20(struct s20 x) {}
>
> That is really crazy.  While it makes perfect sense, I had no idea you could do that! :)

To be fair, its just a crazy test case I cooked up to see if the
backend would handle it. It appears to do the right thing with
aligning to something over the stack alignment, although I didn't look
super closely. GCC seems to give up on this example, FWIW.

 - Daniel

> -Chris
>
>




More information about the cfe-commits mailing list