failing LOCK1 smbtorture test

[MCCALL,DON (HP-USA,ex1)]

Hi Andrew,
Yes, I like that.  I am on holiday at the moment with my kids, but I expect
to be working with the hp-ux lab on the mmap issue (sometimes returning 0
instead of MAP_FAILED) when I get back on Aug 6th.
Thanks,

don

-----Original Message-----
From: Andrew Tridgell [mailto:tridge at valinux.com]
Sent: Thursday, July 26, 2001 23:52
To: don_mccall at hp.com
Cc: anton at samba.org; jeremy at valinux.com; don_mccall at hp.com;
samba-technical at lists.samba.org
Subject: Re: failing LOCK1 smbtorture test


> For instance, the issue with HP-UX not having a unified
> buffer/page cache could be addressed by HP in the near future, and then we
> would PASS the test that Jeremy is proposing, and once again be able to
take
> advantage of the mmap performance gains, without a code modification.  If
> however, we set up a 'blacklist', where we hardcode ifdef's against
> particular os'es, when they have addressed the issues, then a code
> modification is required.

the problem is that this test would need to be 100% reliable, and that
is very hard. I'm sure that HPUX would occasionally pass a mmap
coherency test as the pages are backed by the file, and if the system
happens to be low on ram and happens to flush them to disk at the
right time then a test will pass.

Maybe we need a "use mmap" option in smb.conf? Then we could default
it to off on the blacklist OSes, and sysadmins who know they have a
fixed system could enable it. This would also fix the problems we have
with people running Samba on NFS, where on some systems mmap on NFS
is broken.