On Tue, Aug 26, 2003 at 10:44:04AM -0500, Eric Sandeen wrote: > FWIW, the other solution seems to be to get the latest RPMs from rpm.org > (4.2.x) which turn off O_DIRECT in the rpm source. So using rpm-4.2, 4.1.1 or 4.0.5 should be fine (see: http://www.geocrawler.com/lists/3/Red-Hat-Linux/87/0/10430019/) > I still don't know what the underlying problem actually is. On Tue, Aug 26, 2003 at 10:56:59AM -0500, Eric Sandeen wrote: > On Tue, 2003-08-26 at 10:37, Axel Thimm wrote: > > > underlying kernel - or have you merged 1.2.0 up to 2.4.20-19.9? > > > > Yes, the atrpms kernels track latest RH errata and had (up to now) XFS > > 1.2.0. > > Ah, but I'm sure in your latest errata kernel, patch 1300 was still in > place, and this takes out O_DIRECT very early. That's true and makes sense. > But for your kernels, I'd leave patch 1300 in place, disabling > O_DIRECT entirely. This is no worse than the original RH kernel. > Yes, it cripples XFS, but it's probably the simplest path. If rpm >= 4.2 does the job, I'd rather put a Requires: line in the kernel specfile than crippling XFS. After all I have backported rpm 4.2 for RH8.0 and RH7.3, so it is available from the same source. I'll do some more testing with rpm 4.2. Kai, could you also try it out? -- Axel.Thimm@physik.fu-berlin.de
Attachment:
pgp00012.pgp
Description: PGP signature