Re: [PATCH] cpu_exclusive sched domains on partial nodes temp fix

Re: [PATCH] cpu_exclusive sched domains on partial nodes temp fix

[Dave Jones]

On Tue, Aug 23, 2005 at 09:01:52PM -0700, Linux Kernel wrote:
 > tree c81c261274011d301dfbcfd1a3e13480b93c167e
 > parent ae75784bc576a1af70509c2f3ba2b70bb65a0c58
 > author Paul Jackson <pj@sgi.com> Tue, 23 Aug 2005 15:04:27 -0700
 > committer Linus Torvalds <torvalds@g5.osdl.org> Wed, 24 Aug 2005 10:02:52 -0700
 > 
 > [PATCH] cpu_exclusive sched domains on partial nodes temp fix

This broke ppc64 for me.
 
kernel/cpuset.c: In function 'update_cpu_domains':
kernel/cpuset.c:648: error: invalid lvalue in unary '&'
kernel/cpuset.c:648: error: invalid lvalue in unary '&'

		Dave

Re: [PATCH] cpu_exclusive sched domains on partial nodes temp fix

[Paul Jackson]

Dave wrote:
> > [PATCH] cpu_exclusive sched domains on partial nodes temp fix
> 
> This broke ppc64 for me.


Thanks to Paul Mackerras's report, I sent a patch for this a few hours ago:

  [PATCH 2.6.13-rc6] cpu_exclusive sched domains build fix

It just makes a local copy of the cpumask_t in a local variable on the stack.

I'm still a couple of hours from actually verifying that ppc64 builds with
this - due to unrelated confusions on my end.  Perhaps you or Mackerras will
report in first, to verify if this patch works as advertised.

-- 
                  I won't rest till it's the best ...
                  Programmer, Linux Scalability
                  Paul Jackson <pj@sgi.com> 1.925.600.0401

Re: [PATCH] cpu_exclusive sched domains on partial nodes temp fix

[Dave Jones]

On Wed, Aug 24, 2005 at 12:13:40PM -0700, Paul Jackson wrote:

 > I'm still a couple of hours from actually verifying that ppc64 builds with
 > this - due to unrelated confusions on my end.  Perhaps you or Mackerras will
 > report in first, to verify if this patch works as advertised.

It does build, but I'm unable to boot test it.

		Dave

Re: [PATCH] cpu_exclusive sched domains on partial nodes temp fix

[Paul Jackson]

Dave wrote:
> It does build, but I'm unable to boot test it.

Ok - thanks.

Looks like Nick might be leaning toward the simpler path of disabling
this feature (defining dynamic sched domains using exclusive cpusets)
for 2.6.13, or some such.

We'll see.

-- 
                  I won't rest till it's the best ...
                  Programmer, Linux Scalability
                  Paul Jackson <pj@sgi.com> 1.925.600.0401