Bug#426737: lilo-installer fails on HP SmartArray controllers (/dev/ida)
On 06/18/2007 05:41 PM, Frans Pop wrote:
> On Wednesday 30 May 2007 18:36, Alexander 'Leo' Bergolth wrote:
>> P.S.: Note that the patch only fixes installation in the MBR,
>> installing lilo in a partition would need further fixes.
> Could you elaborate on this comment?
The problem is that the current script is not very modular. There are
many places where the device-name is matched against [hs]d[a-z]. Many
checks also imply that a device ending with a number is a partition, not
a disk. Both assumptions are not correct with SmartArray devices
(/dev/ida/c0d0, /dev/ida/c0d0p1 and /dev/cciss/...).
I have only fixed some of those checks, those that were in the code path
for installing lilo in the MBR.
> Adding support without adding full support (or at least checking for
> the non-supported case in the dialogs that allow selecting a
> different partition) does not sound like a really good idea.
Note that HP/Compaq Smartarray controllers are very common devices. I
believe it's better to support them for the most common case than not to
support them at all. Currently the postinst script just exits with an
error. (Handling the non-supported case would be fine, of course.)
Of course the best solution would be encapsulating and centralizing the
checks into functions but that's a rewrite of the whole script.
e-mail ::: Alexander.Bergolth (at) wu-wien.ac.at
fax ::: +43-1-31336-906050
location ::: Computer Center | Vienna University of Economics | Austria