https://bugzilla.novell.com/show_bug.cgi?id=794505
https://bugzilla.novell.com/show_bug.cgi?id=794505#c2
--- Comment #2 from Peter Benie 2012-12-14 19:43:38 GMT ---
The card has a single disk attached, and there were no devices under
/dev/mapper (first thing I checked!).
The problem appears to happen when the first partition doesn't start right at
the beginning of the disk and the start of the region for the first partition
(as reported in disk_sda.info) is also not zero.
When both these conditions are true, installer treats the free space at the
start of the disk as available, and this space is not marked as 'exists'
(get_gaps).
Free space after the end of data is also not marked as 'exists'.
Later on, when calculating the size of the gaps, it adds together the free
spaces that don't exist, but these spaces can't in fact be added
(normalize_gaps).
While I was looking at this, I spotted an anomaly in do_weighting - for gaps
that don't 'exist', the following adjustments are made:
cylinders==0 => no change
0 weight reduced by 15
disk_size/20<=cylinders => weight reduced by 5
It seems unlikely that the 0 case (which is a valid value) was intended to
behave like this.
On Monday I'll construct a simple config file to demonstrate the problem,
however, the exact details to demonstrate it will depend on the disk geometry.
--
Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.