bugs/conflicts with XFree86 3.3.5

bugs/conflicts with XFree86 3.3.5

Post by Randall J. Par » Fri, 08 Oct 1999 04:00:00

I posted this message half a dozen places attached to X and SCSI issues.
Your's is the first response of any kind that I have gotten.

Nor have I seen any other responses which address these SCSI issues.

This greatly concerns me. I'm very concerned that an XFree86 update caused
such major SCSI problems. That seems to indicate a fragility of the
SCSI/file system that is truly scary. I have some pressing X related
performance issues but feel it would be very dangerous to upgrade my
XFree86 and/or upgrade to RH6.1.

If you hear of anything related to these SCSI and/or X problems please
post it and/or let me know.

Sorry I couldn't be more help.


P.S. Every notice how some posts you throw out there just never get a
response from anyone? I expected something on this one even if only
"you're full of beans". Go figure.

> Randall,
>   Read your note in comp.os.linux.x about a conflict
> between XFree86 3.3.5 (from RedHat) and SCSI systems.

> I also noticed these SCSI errors yesterday and again this
> morning.  Mine manifest themselves when StarOffice is running,
> oddly enough.

> I have a Adaptec 2940UW SCSI controller installed, with a
> Seagate ST34520W SCSI hard drive and a Yamaha CRW4416 CD-RW
> on the scsi chain.

> Did you hear any solutions or patches other than to downgrade
> the installed copy of XFree86?

> Thanks for any pointers

> --
> ---------- ---------- ---------- ---------- ---------- ----------
> ----------
>  Art Mulder,               Sr Systems Administrator, Imaging Research
> Labs

> Institute
>  www.irus.rri.on.ca/~amulder/                 London, Ontario, Canada,
> Eh!


1. bug/conflict in XFree86 3.3.5 on RH6?

I think there may be a very bad bug and/or conflict in XFree86 3.3.5
installed from the rpm's on RedHat 6.0.

Does anyone know if that is true?

I have RedHat 6.0 installed on one machine and had applied the updates
available as of the as of the end of Aug from updates.redhat.com PLUS I
had installed XFree86 3.3.4 from rawhide.redhat.com so I could use a
Matrox G400. That machine is running just fine. It is using Adaptec
2940U2W controllers, IBM 18G drives, and an HP DAT.

I then installed RedHat 6.0 on a 2nd machine also with 2940U2W
controllers, IBM 18G drives and the same HP DAT and an older Matrox
Millenium video board. Everything was fine. I then installed the the
updates available as of Sept 20th or so from updates.redhat.com. These
now included XFree86 3.3.5 (which I installed).

I immediately noticed that it went from a couple of seconds to 30 to 120
seconds for the X server / gdm login to come up at the end of the boot.
It also took 15 to 60 seconds for the gdm login to reappear when logging
out. I put up with it and tried to finish the configuration and find out
what was taking it so long.

At about this same time I started having sever problems with the SCSI
drives and controllers. I kept getting into SCSI time out loops which
could only be broken by powering down the machine (with the attendent
file checks, loss of files, etc.).

I also noticed that this machine took a couple of minutes or more to
pass the NFS mountd but I thought it was just that it had half a dozen
large file systems to export.

When I finally tracked this all down I found a BUNCH of different
variations on filesystem/access timeouts where at the root of all these

I found the NFS mount was hanging at the rpc.rquotad phase.

I also found Oracle was consistently failing to find message files
during boot up but was fine when run manually.

I spent several days trying to figure out what was wrong with the SCSI
system and or drivers.

I finally realized that the scsi timeout loops most often occured when I
was logging in and especially if I had a failed login. In the
/var/log/messages I found that the scsi timeouts loops where
consistently starting right after gdm would fail to create :0.xauth
file/link. The scsi problems lessened if I used xdm instead of gdm. They
were also better if I slowed the SCSI buss down; disconnected the DAT,

Ah, hah, says I, (it's getting late). I updated the system forcing use
of the original XFree86 3.3.3 rpms from the install CD. All of my login,
SCSI timeout problems, NFS mount problems magically and immediately went
away. X server / gdm (or xdm) pop up in a second or two (with no error
messages in /var/log/messages). NFS mount phase flies right by (and
works). I pushed the SCSI speed back up, reattached the DAT drive, etc.
and haven't had a single hick-up in about 24 hours.

THUS, I think there is a bug and/or bad conflict in the XFree86 3.3.5
rpms that I installed from updates.redhat.com

Temporal Arts

2. How to download Yellow Dog Linux CS 1.2 ???

3. xfree86 bug? (or java bug?)

4. Is this racy?

5. VC keys and XFree86 conflict?

6. any command to trace IP packets ?

7. Kernel bug - conflict with C-250 Jumbo Tape drive and Bus Mouse

8. root crontab need default for Solaris 7

9. TGUI9680 problems under XFree86 3.2, IRQ-conflicts ?

10. UMSDOS and XFree86 Conflict

11. Suse 8.0 Update / XFree86 Conflict?

12. gpm conflict with XFree86?

13. Cache-attribute conflict bug in the kernel exposed on newer A MD Athlon CPUs