Re: [Freeipmi-users] SUNBMC failure

[Al Chu]

Hey Dave,

I put out a beta release:

http://ftp.gluster.com/pub/freeipmi/qa-release/freeipmi-0.8.6.beta0.tar.gz

That should compile the sunbmc driver automatically even if bmc_intf.h
is not found (and assuming other build conditions are met).  Could you
give it a shot and LMK if it works?

Al

On Mon, 2010-05-03 at 13:12 -0700, Al Chu wrote:
> Hey Dave,
> 
> On Mon, 2010-05-03 at 12:52 -0700, Dave Love wrote:
> > Al Chu <address@hidden> writes:
> > 
> > > GPL and the CDDL (the Sun license) have incompatibilities with each
> > > other (see http://**www.**gnu.org/licenses/license-list.html), which is 
> > > why
> > > I did it this way.  However, it appears the incompatibilities may exist
> > > only w/ linking and not with code.  So presumably I could copy the code
> > > in.  But I'm not sure.  I've been trying to figure this out, but as you
> > > can expect, licenses are hard to understand :-)  I'll ask around, see if
> > > I can tweak this or not.
> > 
> > Sure if it was CDDL code, you couldn't distribute it to combine with
> > GPL'ed code unless the copyright holders on all the GPL code allowed an
> > exception.  However, as I understand it, a header with no code isn't a
> > problem.  Anyway, what I pointed to is BSD-ish, not CDDL, which is why I
> > think there's no problem -- you'd just need to add the copyright notice
> > at most.  That version is trivially different to the one in OpenSolaris.
> 
> Ahh!  That was a subtlety I missed before.  There is the bmc_intf.h from
> OpenSolaris and the bmc_intf.h from ipmitool.  I had always assumed
> ipmitool copied it from OpenSolaris and thus both were CDDL.  But that
> is not the case.  Ipmitool's bmc_intf.h appears to be "forked" from an
> earlier version, so I could freely use that version.
> 
> Thanks for the catch.  I'll look into it a bit more deeply.  Assuming no
> snags, I can copy that file in.  I'll make a beta, and if you have the
> chance, could you try out the beta for me?
> 
> Al
> 
> > > Hmmm.  I think the reason it fails like that on Linux is b/c /dev/bmc
> > > doesn't exist.  This is in contrast to Solaris where /dev/bmc does
> > > exist, but something doesn't allow the code to work with it (thus
> > > "device not found" would not be a good error code).  Looking through the
> > > error code list, IPMI_ERR_DEVICE_NOT_SUPPORTED is probably the better
> > > one.  Sound like a good error code to change to?
> > 
> > Right, that sounds right -- definitely better than an internal error.
> > If you can just distribute the header, it's moot anyway.
> > 
> > For what it's worth, here's the meat of the licence on bmc_intf.h from
> > ipmitool:
> > 
> >  * Redistribution and use in source and binary forms, with or without
> >  * modification, are permitted provided that the following conditions
> >  * are met:
> >  * 
> >  * Redistribution of source code must retain the above copyright
> >  * notice, this list of conditions and the following disclaimer.
> >  * 
> >  * Redistribution in binary form must reproduce the above copyright
> >  * notice, this list of conditions and the following disclaimer in the
> >  * documentation and/or other materials provided with the distribution.
> > 
-- 
Albert Chu
address@hidden
Computer Scientist
High Performance Systems Division
Lawrence Livermore National Laboratory