Re: Lockup safety in rpm-4.1.1 and rpm-4.2

[Jeff Johnson <jbj redhat com>]

On Mon, Apr 07, 2003 at 08:07:01PM +0300, Panu Matilainen wrote:
> On Mon, 2003-04-07 at 19:51, Jeff Johnson wrote:
> > On Mon, Apr 07, 2003 at 07:21:05PM +0300, Panu Matilainen wrote:
> > > 
> > > Ah.. that explains the thing then. After upgrading to rpm-4.2-1: no more
> > > stale locks and rpm hangs caused by randon ctrl-C'ing when apt is being
> > > run. Thanks for clarifying this!
> > > 
> > > The next question then: when can we expect errata rpm release for RHL 9?
> > 
> > When it's done, the errata currently needs to be redone because of stupid
> > O_DIRECT breakage that the kernel folks choose to inflict on previously
> > working code on already released distro's.
> 
> Care to elaborate: is something bad expected to happen if I use the
> current rpm-4.2-1 package from www.rpm.org on RH9, compared to the
> problems of rpm-4.2-0.69 ? I'd rather upgrade rpm to better version than
> use dirty tricks for the cure.
> 

O_DIRECT used when opening /var/lib/rpm/__db.001 fails, 1st
write returns EINVAL because pointer is not page (actually 512b) aligned.

What's really, really silly is that the kernel folks don't yet have an
O_DIRECT implementation, and are deliberately breaking legacy code according
to recently invented "kernel specifications".  Not good, even though fix is
trivial.

I'm waiting for the dust to clear a bit.

> > 
> > No matter what: Positive reports (presumably accurate) as above will only
> > expedite errata release.
> 
> FWIW it's 100% reproducible both ways (breaks if rpm downgraded, fixed
> if upgrade) here...
> 

I believe you, and now I can point the finger elsewhere ;-)

73 de Jeff

-- 
Jeff Johnson	ARS N3NPQ
jbj@redhat.com (jbj@jbj.org)
Chapel Hill, NC