Experimental Physics and
Industrial Control System

Benjamin Franksen <[email protected]> · Mon, 27 Jun 2011 23:09:09 +0200

Am Montag, 27. Juni 2011, um 18:12:11 schrieb J. Lewis Muir:
> On 6/27/11 7:48 AM, Benjamin Franksen wrote:
> >> This means you shouldn't try to
> >> rely on EPICS_HOST_ARCH containing the string "64" when you're building
> >> for a 64-bit CPU, and even if it does it isn't necessarily going to be
> >> /only/ building for a 64-bit CPU.
> >
> > Hm, does that also mean I can not even rely on parsing the current
> > directory to find out which target arch I am on?
>
> That's correct.  For example, if I have EPICS_HOST_ARCH set to
> "linux-x86" in my environment and ARCH_CLASS set to "i386
> x86_64" in configure/os/CONFIG_SITE.Common.darwin-x86 in EPICS
> Base, then when I do make in seq-snapshot-2011-06-24, the only
> O.* directories in src/seq are O.Common and O.darwin-x86.  So,
> the O.darwin-x86 contains "fat" object files, libraries, or
> binaries for both i386 and x86_64.

I must admit that I went into this discussion w/o any idea what "fat binary"
means. I know better now, and I understand the problem: with fat binaries one
has different target architectures inside a single binary. However, could we
not create a separate pseudo target architecture for such fat binaries? Maybe
with another separator character; I think '+' would be fine (and self-
explaining), so that we would have e.g.

  T_A = darwin-x68+x69_64+pps

> > Could I test the value of ARCH_CLASS
> > in the Makefile to decide whether a test should fail or not?
>
> That might work, but maybe not in the way you think.  Say I
> specify a 32-bit architecture *and* a 64-bit architecture in
> ARCH_CLASS (e.g. i386 and x86_64).  In this case, I think the
> test would actually pass when building the i386 machine code but
> fail when building the x86_64 machine code.  As Andrew noted,
> the EPICS build system does not invoke gcc for each ARCH_CLASS,
> that's happening automatically behind the scenes, so I think you
> would just see failures when building because the fat binary
> includes the x86_64 machine code.  And maybe that's fine?  If a
> fat binary (or library) would contain code that would be broken
> if run on a particular architecture, then fail even though the
> code would actually be OK for some of the architectures (i.e.
> i386 in this example).

Yes, that was the idea. However, with the current build rules for TESTPRODUCTs
I see no way to communicate the distinction from the Makefile to the test
script. By parsing the last part of the cwd I could find out the T_A, and if we
adopted the above proposal, I could match it with '64' inside the test
script...

Cheers
Ben

________________________________

Helmholtz-Zentrum Berlin für Materialien und Energie GmbH

Mitglied der Hermann von Helmholtz-Gemeinschaft Deutscher Forschungszentren e.V.

Aufsichtsrat: Vorsitzender Prof. Dr. Dr. h.c. mult. Joachim Treusch, stv. Vorsitzende Dr. Beatrix Vierkorn-Rudolph
Geschäftsführer: Prof. Dr. Anke Rita Kaysser-Pyzalla, Dr. Ulrich Breuer

Sitz Berlin, AG Charlottenburg, 89 HRB 5583

Postadresse:
Hahn-Meitner-Platz 1
D-14109 Berlin

http://www.helmholtz-berlin.de

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System