Hi, I just discovered a weird "bug"/"interaction" between the ncr53c8xx, aic7xxx scsi modules when it comes to disk allocation. Has anyone else seen this ? I have a server, HP Netserver LPR and it boots from the NCR chipset. It had one disk, a 9GB 10KRPM Cheetah - HP firmware. The aic7xxx is a separate PCI card. It has an external LVD rack connected to it with initially four hard drives. This all worked and I had a nice RAID5 configuration working quite well with this. Now, I installed a further three hard drives and the driver and the scsi subsystem detected them fine. However, the SCSI disk system did not detect the disks/partitions. So I get letters assigned to the disks f,g and h but I am unable to access them. Apparently the hardware is unavailable. Only the first four hard drives on that SCSI chain are accessible from within Linux. As a test, I disabled the NCR and moved the boot drive to the AIC7XXX. Now all drives are there and everything is just fine. I can see all 8 hard drives without any problems, fdisk lets me access them and partition them without incident and I can even configure them into a RAID5 array. So my guess is that there is some sort of over allocation bug between the NCR and AIC drivers so that the NCR is taking too many resources and leaving the AIC with only four usable drive slots. I'm not a kernel hacker so it's hard for me to figure this out by looking at the source. If no-one else has heard of this problem, I'll try reporting it to the SCSI mailing list. Thanks Yan Li Hewlett-Packard Company
