On Thu, Jul 31, 2008 at 09:57:13AM +0100, Mel Gorman wrote:
> On (31/07/08 11:08), David Gibson didst pronounce:
> > On Wed, Jul 30, 2008 at 09:04:53AM +0100, Mel Gorman wrote:
> > > On (30/07/08 13:37), David Gibson didst pronounce:
> > > > While checking some other things, I noticed today that with a kernel
> > > > built from today's Linus tree the 'counters' testcase fails, at least
> > > > on powerpc.
> > > >
> > > > counters (32): FAIL Line 375: Bad HugePages_Total: expected 0,
> > > > actual 1
> > > > counters (64): FAIL Line 375: Bad HugePages_Total: expected 0,
> > > > actual 1
> > > >
> > > > I had a look at the code, but since I don't really understand how the
> > > > dynamic pool resizing stuff works, I couldn't figure out what was
> > > > going on. Somebody who understands this better should probably take a
> > > > look at this.
> > > >
> > >
> > > It's not dynamic pool resizing.
> > >
> > > Anything older than libhugetlbfs 2.0-pre1 will fail the counters
> > > test on 2.6.27-rc1. Private reserves are now accounted for and
> > > mmap() will return NULL if enough huge pages are not available to
> > > satisfy the mapping. The counters test expects no reserves to be
> > > taken for private mappings so it fails.
> >
> > Um... but I thought I had 2.0-pre1. I pulled it yesterday.
> >
>
> If you do have it, the test should be passing for 2.6.27-rc1. Check if
> tests/counters.c has a call to "kernel_has_private_reservations(fd)" in
> it. If it does and the test still fails, can you tell me what sort of
> machine and exact kernel version this is happening on and I'll try
> reproducing it.
I did another pull today. tests/counters.c has that call, and it's
still failing. Kernel reports as 2.6.26, but I suspect it's actually
some random git version between 2.6.26 and 2.6.27-rc1, running on a
POWER6 machine. I'll dig up more details when I get the chance.
--
David Gibson | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_
| _way_ _around_!
http://www.ozlabs.org/~dgibson
-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
Libhugetlbfs-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libhugetlbfs-devel