Re: How should _Decimal64 and _Decimal128 be aligned on stack?

["H.J. Lu" <hjl dot tools at gmail dot com>]

Hi David,

I added gcc mailing list.

H.J.
On Tue, Apr 1, 2008 at 10:53 AM, Dave Kreitzer
<david.l.kreitzer@intel.com> wrote:
>
>  > I think we should align _Decimal64 and _Decimal128 to their natural
>  > alignments when passing a function. The same should apply to x86-64
>  > when they are passed on stack.
>  >
>
>  Rather than aligning _Decimal64 to its natural boundary, I think it is
>  more sensible to be consistent with the existing 8-byte scalar types,
>  i.e. double and long long.  That would mean aligning _Decimal64 to 4
>  bytes within structures and when passing to a function.
>

That means to change the default alignment of  _Decimal64 to 4 bytes in
32bit mode.

>  For _Decimal128, I agree with the conclusion to align to 16-bytes
>  within structures and as outgoing arguments.  That behavior is
>  consistent with the existing data types __m128, __m128d, and __m128i.
>
>  Dave Kreitzer
>  IA-32 and Intel 64 Code Generation
>  Intel Compiler Lab
>
>
>
>  --~--~---------~--~----~------------~-------~--~----~
>  You received this message because you are subscribed to the Google Groups "IA32 System V Application Binary Interface" group.
>  To post to this group, send email to ia32-abi@googlegroups.com
>  To unsubscribe from this group, send email to ia32-abi-unsubscribe@googlegroups.com
>  For more options, visit this group at http://groups.google.com/group/ia32-abi?hl=en
>  -~----------~----~----~----~------~----~------~--~---
>
>