Re: Data structure alignment

[Jim Langston <jimlangston@xxxxxxxxxxxxxxxx>]

Wow, thanks a lot for the information Ken!  This is *very* useful information!

Excellent research, kudos :)

Regards,

Jim Langston

"Sims, Ken" wrote:
> 
> Hi Peter, Richard, Jon, et.al. -
> 
> >How can one tell what boundary a data structure, or any other field for
> that
> >matter, is aligned on?
> 
> For my test programs, I put debug(*yes) in the H-spec.
> I used eval [pointerfield] = %addr[field or data structure]
> followed by dump to get a nice readable output with the addresses of the
> fields and data structures.
> 
> For the V4R4 RPG IV compiler, my observations are:
> 
> 1. all data structures are currently 16-byte aligned, even ones containing
> just a simple character field.  However as Barbara has said, you shouldn't
> count on this always being the case for data structures that do not contain
> pointers.
> 
> 2. for multiple occurence data structures, only the first occurrence is
> necessarily 16-byte aligned.  If the data structure contains pointers, all
> occurrences are 16-byte aligned of course.  If a data structure contains
> float or integer fields and the ALIGN keyword is not used, there are no gaps
> between occurrences and so those fields can be aligned in some occurrences
> and not aligned in other occurrences.  If the ALIGN keyword is used, there
> will be gaps between occurrences as needed so that all of the occurrences
> will start on the boundary needed to force alignment, but not necessarily on
> a 16-byte boundary.
> 
> 3. for individual fields, float and integer fields are aligned on the
> appropriate boundary for performance as expected.  Packed and zoned fields
> are not aligned in any particular way.  Interestingly, it appears that
> individual character fields are always aligned to the boundary of the power
> of 2 greater than or equal to the length of the field, up to a maximum of
> 16-byte alignment!  (Character fields in a data structure just follow each
> other as expected.)  The order of fields in memory will be rearranged from
> the order specified in the program to take advantage of the gaps left by
> alignment.
> 
> But again, one should not count anything of the things mentioned above
> (except pointer alignment) as being something that one can count on to stay
> the same from release to release.
> 
> Ken
> Southern Wine and Spirits of Nevada, Inc.
> Opinions expressed are my own and do not necessarily represent the views of
> my employer or anyone in their right mind.
+---
| This is the RPG/400 Mailing List!
| To submit a new message, send your mail to RPG400-L@midrange.com.
| To subscribe to this list send email to RPG400-L-SUB@midrange.com.
| To unsubscribe from this list send email to RPG400-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---