Description of problem: On a test system I have a drive which has partition from /dev/sda1 to /dev/sda15. Unfortunately anaconda creates only /dev/sda1 to /dev/sda14 nodes and when booting "rescue" 20070525 images, and an installation which happens to use /dev/sda15, it fails to mount a corresponding partition producing an error. In a similar manner /dev/sdb1 to /dev/sdb14 and /dev/sdc1 to /dev/sdc14 are also created for two other disks which happen to actually use much smaller numbers of partitions so there you never bump into that issue. BTW - this is not an issue with libata.ignore_hpa=1 from bug 241288. When booted to a shell prompt it is possible to do mknod /dev/sda15 lvm vgchange -a y (as this partition happens to be used by an lvm volume) and after that corresponding volume "magically" appears in /dev and its logicals can be mounted without any further troubles. Version-Release number of selected component (if applicable): anaconda from 20070525 images (it does not show its version) How reproducible: always
Created attachment 161152 [details] check if the device exists and create if it doesn't
Created attachment 161179 [details] Fix the missing /dev/sda15 device There can be only 15 of scsi partitions right now, thanks to libata.. but there was a small error... range(1,15) didn't create the 15th device. Additionaly and for sure, check ensure device exists when used to normal mount and label testing.
Created attachment 161257 [details] patch cleanup Patch cleanup, better error checking and reporting (to log).
Commited to upstream repository, will be included in next build.