[OpenAFS-devel] Question on possible change in OpenAFS compilation for UML

[Raymond Wong]

Kris, 

I think your idea is good.  Kernel headers reflect a better picture than
/proc/cpuinfo.

Ray


-----Original Message-----
From: openafs-devel-admin@openafs.org
[mailto:openafs-devel-admin@openafs.org] On Behalf Of Kris Van Hees
Sent: Thursday, July 10, 2003 2:36 AM
To: openafs-devel@openafs.org
Subject: [OpenAFS-devel] Question on possible change in OpenAFS
compilation for UML


Currently, the support for UML (User Mode Linux) depends on it being
compiled on UML itself, because it detects the host using information in
/proc/cpuinfo. That of course causes an issue if you would want to do
the actual compilation on another Linux system (a real one, not UML) for
performance reasons or just to centralize your compilations.  OpenAFS
offers the --with-linux-kernel-headers option to configure to (possibly)
compile against a Linux source tree that is not identical to the host
the compilation is running on.

So the question is...  Do you think that it would be better to have the
code in configure for detecting the appropriate sysname use the
information found in the Linux kernel headers (look for CONFIG_USERMODE
set to 1 in <autoconf.h>) rather than the current check on
/proc/cpuinfo?

My guess would be a resounding 'YES' for using the kernel headers, but
I'd like to confirm that before sending in a patch for it :)

	Kris
-- 
Never underestimate a Mage with:
 - the Intelligence to cast Magic Missile,
 - the Constitution to survive the first hit, and
 - the Dexterity to run fast enough to avoid being hit a second time.
_______________________________________________
OpenAFS-devel mailing list
OpenAFS-devel@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-devel