[M3devel] Darwin Intel Data Alignment

[Tony Hosking]

Hmm.  I am not familiar with this aspect of things.  One thing would  
be to check if things work for you on LINUXLIBC6.  Then we just need  
to figure out what is different between the Darwin and Linux Intel  
targets.  Can you send me some test code that demonstrates the problem?

On May 22, 2007, at 9:13 AM, Darko wrote:

> Hi,
>
> I'm having problems trying to use byte aligned fields in structures  
> in the new Darwin Intel compiler. My understanding is that the  
> processor accepts accesses on (word, etc) data not aligned on their  
> normal boundaries. Is there something in the back end which means  
> this is not possible?
>
> The <*LAZYALIGN*> pragma doesn't work properly and I get:
>
> "../src/mac/PalPlat.m3", line 1525: ** INTERNAL CG ERROR ***  
> unaligned store_indirect  type=11  s/a=32/16
> "../src/mac/PalPlat.m3", line 2977: ** INTERNAL CG ERROR *** unable  
> to find integer type?  type=Int.32  s/o/a=32/64/16
>
> when I compile. I think the platform independent alignment  
> calculation code is correct, so I'm thinking this is a back end  
> limitation.
>
> Any thoughts?
>
>
> Cheers,
> Darko.