[OpenAFS-devel] RH 7.1 spec file: archlist
Rudolph T Maceyko
rtm@cert.org
Wed, 25 Jul 2001 16:52:11 -0400
Derek,
--On Wednesday, July 25, 2001 15:53:39 -0400 Derek Atkins
<warlord@MIT.EDU> wrote:
> I'm glad it worked, however....
>
> Rudolph T Maceyko <rtm@cert.org> writes:
>
>> $ cat /usr/vice/etc/modload/SymTable
>> # SymTable, automatically generated
>> # symbol version cpu module
>>
>> iget4 smp_d7538340 - 2.4.3-12-athlon.mp
>> iget4 0ac9f202 - 2.4.3-12-athlon
>> iget4 6fc4bdd5 - 2.4.3-12-i386
>> iget4 smp_c0971f1d - 2.4.3-12-i586.mp
>> iget4 67b2bf7c - 2.4.3-12-i586
>> iget4 smp_c0971f1d - 2.4.3-12-i686.mp
>> iget4 67b2bf7c - 2.4.3-12-i686
>
> It looks like I may not need to build different modules for both 586
> and 686. Can you 'md5sum' the 586 and 686 modules and see if they are
> the same or different? According to this, they are the same.
The iget4 symbol is the same but the libafs modules themselves are NOT:
$ cmp libafs-2.4.3-12-i586.o libafs-2.4.3-12-i686.o
libafs-2.4.3-12-i586.o libafs-2.4.3-12-i686.o differ: char 33, line 1
$ md5sum *.o | sort
02d9da2f7d7f815e9c0bf1dc46819884 libafs-2.4.3-12-i386.mp.o
100ef3329ea8dee4eb05ebc7fe5a88d2 libafs-2.4.3-12-athlon.o
3d6ca0375137e781084806c3f1635a5e libafs-2.4.2-2-i686.o
4ca10a5d6ac80ec8563c0bd7c3f1ff12 libafs-2.4.3-12-i386.o
4cca7558230708db3d9a9e57114cdb1d libafs-2.4.3-12-athlon.mp.o
513ec06cbdfa01b516776289e0882f57 libafs-2.4.3-12-i686.mp.o
5b5715e257703d62f1e07eda74ea5c19 libafs-2.4.2-2-i686.mp.o
5d9a4ead0bd80a5c62fcc11cb3e19edb libafs-2.4.2-2-i586.o
8a81c8188930c6c6f49fa69aab470a8e libafs-2.4.3-12-i586.o
8e1ef58231840acf8701cf63e21bdb2b libafs-2.4.3-12-i686.o
93752db2f7ffc1cdaa6fec3a0b8028a0 libafs-2.4.3-12-i586.mp.o
b833f8d959cfa4ac2f27396f77e0f795 libafs-2.4.2-2-athlon.mp.o
b833f8d959cfa4ac2f27396f77e0f795 libafs-2.4.2-2-i386.mp.o
d6ef15260912c34f5df1de5936e8ce67 libafs-2.4.2-2-athlon.o
d6ef15260912c34f5df1de5936e8ce67 libafs-2.4.2-2-i386.o
ebad50ffbf5dc3d841cb3cd9b0d23cd6 libafs-2.4.2-2-i586.mp.o
Note that libafs-2.4.2-2-{i386,athlon}.o are the same though as are
libafs-2.4.2-2-{i386,athlon}.mp.o...
>> Looks good!
>>
>> I think we may want to add "enterprise" to the arch list for Red Hat
>> systems, BTW.
>
> I was thinking about that, but unfortunately 'enterprise' doesn't
> quite fit into my current spec design. The problem is that
> 'enterprise' is almost another axis, because I believe you can have
> enterprise-i586, enterprise-i686, enterprise-i586+mp,
> enterprise-i686+mp, and you _may_ even get enterprise-athlon and
> enterprise-athlon+mp.
Yeah, I just figured that out too when I tried to add it to the spec
file :-) For now we're just running the normal SMP kernel on those
boxes...
> Granted, currently enterprise is only built on i686. If you can tell
> me what /boot/kernel.h looks like on an 'enterprise' installation, I
> can try to fit it in for future versions.
I'll send you the /boot/kernel.h from one of them once they are
actually running the enterprise kernel again...
> -derek
Thanks again,
Rudy