Bug 69660 - SMP Kernel cannot load ncr53c8xx SCSI driver on Proliant server
Summary: SMP Kernel cannot load ncr53c8xx SCSI driver on Proliant server
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Red Hat Linux
Classification: Retired
Component: kernel
Version: 7.3
Hardware: i586
OS: Linux
medium
medium
Target Milestone: ---
Assignee: Arjan van de Ven
QA Contact: Brian Brock
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2002-07-24 09:07 UTC by Jesse Keating
Modified: 2007-04-18 16:44 UTC (History)
1 user (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2003-11-01 18:42:27 UTC
Embargoed:


Attachments (Terms of Use)

Description Jesse Keating 2002-07-24 09:07:33 UTC
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.0) Gecko/20020606

Description of problem:
When using the 2.4.18-5smp kernel, my proliant 1500 PC is unable to get past
loading the ncr53c9xx device.  The error goes:

ncr53c8xx_reset: pid=X reset_flags2 serial_number=X serial_number_at_timeout=X
scsi : aborting command due to timeout : pid X, scsi0, channel0, id4,  lun 0
Inquiry 00 00 00 ff 00
ncr53c8xx_abort: pid=X serial_number=X serial_number_at_timeout=X

X is a increasing number, up until pid=18, serial_number=19,
serial_number_at_timeout=19.

If I boot the 2.4.18-5 uniprocessor kernel, I don't have any errors, and the
system works fine. (except for only one CPU is used)

Version-Release number of selected component (if applicable):


How reproducible:
Always

Steps to Reproduce:
1.  Install Red Hat Linux 7.3
2.  install errata (including kernel)
3.  Select 2.4.18-5smp kernel image from grub

	

Actual Results:  Boot process errors w/ above error

Expected Results:  A clean boot

Additional info:

This problem does occur w/ the stock kernel images as well.

In order to get this machine to even boot, I do have to pass memory mappings
into the kernel:

mem=exactmap mem=640K@0 mem=191M@1M

Comment 1 Arjan van de Ven 2002-07-24 09:10:44 UTC
There's an alternative symbios driver now, maybe that's worth a shot?
(the new driver just as "_2" appended to the name)

Comment 2 Jesse Keating 2002-07-24 09:19:50 UTC
I'm curious, will I have to build a new initrd image for using this module, or
can I just modify /etc/modules.conf ?  This controller does control the drive
that has all my file systems on it.

(BTW, thanx for the instant reply!!!)

Shocked in Seattle, 
jkeating
j2Solutions

Comment 3 Arjan van de Ven 2002-07-24 09:22:01 UTC
you will have to do both; eg adjust /etc/modules.conf and then make a new initrd
with mkinitrd (I'd recommend making a second file so you can get grub to
dual-boot between the two :)

Comment 4 Jesse Keating 2002-07-24 09:24:33 UTC
After further inspection, I only see one such ncr53c8xx.o driver, but there are
2 sym53c8xx drivers.  sym53c8xx.o and sym53c8xx_2.  Are these suitable for a
ncr53c825-0 SCSI controller?

jkeating

Comment 5 Arjan van de Ven 2002-07-24 09:26:44 UTC
ok you won't like the answer "depends a bit"
80% of the ncr/symbios cards work with the "ncr" driver
80% with the "sym" one
but not the same 80%.....


Comment 6 Jesse Keating 2002-07-24 09:40:23 UTC
Ok, sym53c8xx_2 failed misreably... but it could be due to my initrd.  Perhaps
there are more than just that module to load?  Here is the output:

/lib/sym53c8xx_2.0: unresolved symbol: unresolved symbol
scsicam_bios_param_Rsmp_a0402c7f
/lib/sym53c8xx_2.0: unresolved symbol: unresolved symbol
scsi_unregister_module-Rsmp_81d85a75
/lib/sym53c8xx_2.0: unresolved symbol: unresolved symbol
scsi_unregister_Rsmp_1ea5cd66
/lib/sym53c8xx_2.0: unresolved symbol: unresolved symbol
scsi_register_module_Rsmp_fa20b7b0
/lib/sym53c8xx_2.0: unresolved symbol: unresolved symbol scsi_register_Rsmp_9d254550
ERROR: /bin/insmod exited abnormally!

My mkinitrd command was:

mkinitrd --preload sym53c8xx_2 --with=sym53c8xx_2 -v
/boot/initrd-2.4.18-5smp.img.test 2.4.18-5smp

and it ran without errors.

Thank you for continued support!!

jkeating

Comment 7 Arjan van de Ven 2002-07-24 10:05:54 UTC
this looks like the initrd loads the sym module before the scsi layer...

easiest way to do this is to NOT provide options just change your
/etc/modules.conf temporarily. mkinitrd will do the right thing automatic then

Comment 8 Jesse Keating 2002-07-24 17:01:39 UTC
I seem to be running out of options.  Your method for making initrds helped, but
unfortionatly, neither of the sym drivers worked.  The straight one (not the _2)
panics w/  aunsupported device, the _2 one gets a bit more interesting:

sym0: <825> rev 0x2 on pci bus 0 device 10 function 0 irc 10
sym0: No NVRAM, ID 7, Fast-10, SE, parity checking
sym0: SCSI BUS has been reset
scsi0 : sym-2.1.17a

And then it goes into a bunch of sym0:1:0: ABORT operation started
sym0:1:0: ABORT operation timed-out.
sym0:1:0 DEVICE RESET operation started
sym0:1:0 DEVICE RESET operation timed-out.

scsi: device set offline - not ready or command retry failed after bus reset:
host 0 channel 0 id 2 lun 0

Got any more magic tricks to pull out of your hat?

jkeating

Comment 9 Arjan van de Ven 2002-07-24 19:48:27 UTC
Only one more I can think of right now:
the "noapic" kernel option (needs to be added to the vmlinuz line in
/boot/grub/grub.confO

Comment 10 Jesse Keating 2002-07-24 20:03:10 UTC
Arjan, you magician you!!!

adding "noapic" to my already lengthy kernel line in grub seems to have solved
my problems.  I want to thank you immensely for the help you have given me in
this problem.  Once again, it is proven to me, Red Hat is _the_ Linux Distribution!!

jkeating
j2Solutions


Note You need to log in before you can comment on or make changes to this bug.