Bug 144621

Summary: disk partitions on IBM x346
Product: [openSUSE] SUSE Linux 10.1 Reporter: Ray Carlino <rcarlino>
Component: InstallationAssignee: Hannes Reinecke <hare>
Status: RESOLVED INVALID QA Contact: Klaus Kämpf <kkaempf>
Severity: Blocker    
Priority: P5 - None CC: kgw, snwint
Version: Beta 1   
Target Milestone: ---   
Hardware: x86-64   
OS: Other   
Whiteboard:
Found By: Other Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---
Attachments: hwinfo --disk results
fdisk results
hwinfo results
lspci results
hwinfo --storage-ctrl results
y2log

Description Ray Carlino 2006-01-21 16:13:19 UTC
When installing all scsi drives show up as drives instead of just the 2 raid 5 containers. So instead of having just 2 drives to setup it thinks it has 8, 2 virtual and 6 physical. During install grub will not install because the partition is invalid. looking thru the logs the install is formating the physical drives. This is using server-raid 7k. The same hardware works fine with suse 10.0 and only shows the 2 containers.
Comment 1 Martin Lasarsch 2006-01-23 09:57:47 UTC
please provide some information:

/var/log/YaST/y2log

the output of:
hwinfo --disk
hwinfo --storage-ctrl
fdisk -l
lspci

please attach the files to the bug, thanks.
Comment 2 Ray Carlino 2006-01-23 13:50:02 UTC
Created attachment 64493 [details]
hwinfo --disk results
Comment 3 Ray Carlino 2006-01-23 13:50:22 UTC
Created attachment 64494 [details]
fdisk results
Comment 4 Ray Carlino 2006-01-23 13:50:39 UTC
Created attachment 64495 [details]
hwinfo results
Comment 5 Ray Carlino 2006-01-23 13:50:54 UTC
Created attachment 64496 [details]
lspci results
Comment 6 Ray Carlino 2006-01-23 13:51:17 UTC
Created attachment 64497 [details]
hwinfo --storage-ctrl results
Comment 7 Ray Carlino 2006-01-23 13:51:43 UTC
Created attachment 64498 [details]
y2log
Comment 8 Thomas Fehr 2006-01-23 17:00:34 UTC
Steffen "hwinfo --disk" shows 8 disks (sda to sdh) as usable to storage module.
According to the user it should only show two disks.
The softraid flag is not set for any of the disks.
Comment 9 Steffen Winterfeldt 2006-01-23 17:06:08 UTC
Not much I can do here. Looks like the ips module works differently now
and reports 8 disks.
Comment 11 Hannes Reinecke 2006-02-01 08:03:13 UTC
I'll need the output of 'dmesg' or /var/log/messages containing the initialisation messages. It looks as if the 'nondasd' support has been enabled.
Comment 12 Hannes Reinecke 2006-02-01 09:07:24 UTC
Configuration issue. dmesg output:

ACPI: PCI Interrupt 0000:01:08.1[A] -> GSI 20 (level, low) -> IRQ 177
AAC0: kernel 2.8-0[6095] 
AAC0: monitor 2.8-0[6095]
AAC0: bios 2.8-0[6095]
AAC0: serial 7eb410d3
AAC0: ROMB RAID/SCSI mode enabled
AAC0: Non-DASD support enabled.
scsi0 : percraid

This corresponds to the following initialisation code:

	/*
	 * If the firmware supports ROMB RAID/SCSI mode and we are currently
	 * in RAID/SCSI mode, set the flag. For now if in this mode we will
	 * force nondasd support on. If we decide to allow the non-dasd flag
	 * additional changes changes will have to be made to support
	 * RAID/SCSI.  the function aac_scsi_cmd in this module will have to be
	 * changed to support the new dev->raid_scsi_mode flag instead of
	 * leaching off of the dev->nondasd_support flag. Also in linit.c the
	 * function aac_detect will have to be modified where it sets up the
	 * max number of channels based on the aac->nondasd_support flag only.
	 */
	if ((dev->adapter_info.options & AAC_OPT_SCSI_MANAGED) &&
	    (dev->adapter_info.options & AAC_OPT_RAID_SCSI_MODE)) {
		dev->nondasd_support = 1;
		dev->raid_scsi_mode = 1;
	}
	if (dev->raid_scsi_mode != 0)
		printk(KERN_INFO "%s%d: ROMB RAID/SCSI mode enabled\n",
				dev->name, dev->id);
		
	if(nondasd != -1) {  
		dev->nondasd_support = (nondasd!=0);
	}
	if(dev->nondasd_support != 0){
		printk(KERN_INFO "%s%d: Non-DASD support enabled.\n",dev->name, dev->id);
	}

So this 'failure' appears to be a feature, triggered by some setting in the driver BIOS. As this driver moved over to a common RAID infrastructure lately it looks as if this move triggered the different behaviour.

However, by simply setting the module option 'nondasd=0' for aacraid the old behaviour should be restored. I don't think we need to invest more time here.

Please reopen if setting the above option does not solve the problem.
Comment 13 Hannes Reinecke 2006-02-01 10:07:29 UTC
*** Bug 144814 has been marked as a duplicate of this bug. ***