Red Hat Bugzilla – Bug 55244
RH7.2 aic7xxx on Intel PR440FX motherboard
Last modified: 2007-04-18 12:37:49 EDT
Description of Problem:
When I tried to install RH7.2 on a box with an Intell PR44FX mother board
it hangs and reports a timeout error in the SCSI bus. It also incorrectly
reports the card speed (ie.; 20mbps (fast) as opposed to 40mbps (fast
Version-Release number of selected component (if applicable):
Steps to Reproduce:
The motherboard has two intel overdrive PII333 processors.
I am having a similar problem with a very similar motherboard. I have a
VALinux 1150 (1U server)... user guide:
The Intel motherboard uses the Intel 82443GX+ chipset. My SCSI card is an
onboard Adaptec AIC-7896 Dual Channel.
To clarify my previous post... the 7.2 install appears to go fine, however upon
rebooting the computer, after the GRUB page, during all the hardware
initializations, I get the SCSI error and can't get past it. It appears to
recognize my 2 scsi drives, but then I start getting a bunch of scsi timeout
errors as the OS tries to communicate with all of the 16 scsi IDs.
My issue is during the install and not after the install on subsiquent restarts
of the system.
Any one working on this? I see that it has been assigned but I have not
recieved a contact.
I am using an INTEL PR440FX board too, and everything went fine for me (QUANTUM
Atlas IV 9.1 GB, PIONEER DR-U16S, FUJITSU MCE-3064SS). The QUANTUM drive is the
-last- one in the chain and terminates all the 16 data lines!
However, I must confirm, that the QUANTUM drive is -reported- to operate only at
20 MBytes/s. May be it's simply an error in the text output and should rather
read 20 MWords/s (when a word is defined to be 16 bit wide in UW mode):
Attached scsi disk sda at scsi0, channel 0, id 0, lun 0
(scsi0:0:0:0) Synchronous at 20.0 Mbyte/sec, offset 15.
The HD is an U2W disk, so in principle, one would expect it to operate at 40.0
NO! None of the Quantum Atlas IV drives have built in termination (in fact,
*NO* LVD drive has built in termination). Your speeds are so low because the
SCSI bus can't support going any faster when it's unterminated like yours is.
Additionally, when the driver says 20MByte/s, it means 20MByte/s, no more. And,
a U2W drive is good for 80MByte/s when connected to a U2W controller and when
the bus is properly terminated and no Single-Ended devices are placed on the
bus. If you put even one Single-Ended drive on the same cable as the LVD
drives, then the whole bus is slowed to a maximum speed of 40MByte/s and forced
to operate in Single-Ended instead of LVD mode (which is less reliable and has a
much shorter cable length limit).
Your problem is the same as reported in bug #29555. However, in order to make
your machine work properly after install, you need to either use an SMP kernel
or pass the apic command line option to the UP kernel. I would suggest going
into the boot loader configuration file (for which ever boot loader your machine
uses) and adding the apic option to the standard kernel parameters that are
passed to the kernels on your machine. In order to get into the machine after
the install you can manually specify the apic option of course.
First, the driver reports the speed that it is actually running at in MByte/s,
not MBits/s. Further, there hasn't been a problem in the speed reporting in
this driver in years. I would suspect something else is impacting the speed
your drives are reporting, and it is likely the same thing that is causing all
the resets. I would go through and verify that the SCSI bus is properly cabled
and terminated and that the Adaptec BIOS on your SCSI controller has all of your
drives configured to use the maximum allowed speed.