Re: Strange boot problems on amd64-current (6.99.40)

[christos%zoulas.com@localhost (Christos Zoulas)]

On May 12,  1:53am, cube%cubidou.net@localhost (Quentin Garnier) wrote:
-- Subject: Re: Strange boot problems on amd64-current (6.99.40)

| My analysis is that rf_netbsdkintf.c:1.306-8 introduced a call to
| findroot() (via cpu_rootconf() changed later to cpu_bootconf()),
| which sets booted_partition to e because it matched wd0e and wd1e to
| the physical boot device.
| 
| But when booted_device is later changed to raid0, booted_partition is
| not touched.  I am not 100% clear about what changed there, but in a
| way the old behaviour was wrong, too:  it doesn't seem like a good
| idea to change booted_device without resetting booted_partition to a
| proper value.
| 
| Now, what should rf_buildroothack do?  Set it to 'a' unconditionnally?
| Maybe set it to whichever partition is at the beginning of the device?

That's what the "wedge" part of the code does... It is easy to do
the same in the non-wedge part. How do we find the root partition of a
RAID set?

christos