[ofa-general] Re: [PATCH] Update mad formatting functions.
Sasha Khapyorsky
sashak at voltaire.com
Wed Apr 15 17:09:34 PDT 2009
On 14:03 Wed 15 Apr , Ira Weiny wrote:
> >
> > ib_mad_f table also has a name string field. I think it can be useful
> > too - will help to unify outputs. Of course this can be done as
> > subsequent patch.
>
> Yes but I don't think we should force users to use any specific output. If
> they want to print the "name" of a field that should be a separate specifier
> __not__ automatic. Is this what you mean?
Yes, exactly - using separate convesion symbol, not both in %F.
> > Lids are printed as decimals.
>
> Well I thought I copied the output from the example but I see that it is
> printing decimal. So?? :-/ I fixed it.
>
> As an aside, not all LID's are decimal. Should we change this?
Yes, it is better to have things unified.
> from fields.c
> ...
> {BE_OFFS(256, 16), "DrSmpDLID", mad_dump_hex},
> {BE_OFFS(272, 16), "DrSmpSLID", mad_dump_hex},
> ...
> {BITSOFFS(224, 16), "RedirectLID", mad_dump_hex},
> {BITSOFFS(480, 16), "TrapLID", mad_dump_hex},
> ...
> {BITSOFFS(320, 16), "PathRecDLid", mad_dump_hex},
> {BITSOFFS(336, 16), "PathRecSLid", mad_dump_hex},
> ...
> {BITSOFFS(288, 16), "McastMemMLid", mad_dump_hex},
Old stuff...
> > Now instead of reimplementing *printf() functions with potential need
> > to follow their extensions/conventions/update/etc wouldn't it be easier
> > (and in long term safer) to just rebuild format string by resolving
> > known %X conversions and then to pass it with rest parameters to
> > standard libc's *printf()?
> >
> > In this way we will support all what *printf()s know + our conversions.
> >
>
> I thought about that but decided not to do it. I can't remember why though...
> ;-) So maybe I agree with you, let me try and remember and if I can't I will
> change it.
Thanks.
Sasha
More information about the general
mailing list