Bug#318194: Installation report

[Jamie Tatum <jtatum@gmail.com>]

Hi All, 

Thanks for all the replies.  I've had a chance to look at this today
and have some information.

First, some more detail about the system.  This system is running the
new standard SATA BIOS Dell uses on desktops.  We use all Dells here
at work and we have been getting SATA systems with an identical BIOS
since January.  The BIOS doesn't have a lot of configuration in most
areas but has a lot of choices for SATA.  The choices are in a menu
labeled SATA Operation (maybe this will help someone):

RAID Autodetect / AHCI
This one requires the AHCI module for the hard drives.  The ATA
connected DVD drive shows up with the ATA driver as hda.

RAID Autodetect / ATA
This setting will work with the ata_piix driver.  DVD still shows up on hda.

RAID On
This setting mandates their ... undoubtedly high quality ... software
RAID.  I didn't play with it.  There is some effort to create drivers
for it, I think.

Combination
This was a strange one.  Again, ata_piix picked up the hard drives. 
The DVD drive for some reason came up as hdc.

Once I found this, I ran lspci with the AHCI and ATA modes and diffed
the result.  I was surprised that there is a difference in the lspci
output but I guess it makes sense.

AHCI:
0000:00:1f.2 0106: Intel Corp.: Unknown device 27c1 (rev 01)

ATA:
0000:00:1f.2 IDE interface: Intel Corp.: Unknown device 27c0 (rev 01)

Part of the reason I am playing with this is because I am having
problems with the AHCI driver.  It's very new and as soon as it loads,
it generates constant activity on my sdb even with nothing mounted. 
It didn't affect data integrity so I don't feel like debugging it.  I
was able to generate a new initrd and cut over to the ata_piix driver
with little pain.

Now, to the specific questions:

> -first go up to the moment the HD detection fails
> -switch to VT3 (Alt+F3)
> -run "lsmod". Note the list
> -modprobe ahci
> -note the list (and check differences)
> -switch back to VT1 (Alt+F1)
> -do what you did (back twice, re-run HW detection)
> -as soon as the HD is detected, go back to VT3 and run lsmod again
> -check the differences....this should help

Thanks for clarifying this.  I have the files available and was going
to post a diff, but I'll just post what happens.  Let me know if you
would like any of these files.  After modprobe, libata loads and
appears as a user of scsi_mod, ahci loads and appears as a user of
libata and scsi_mod.  After the new device detection, sd_mod loads and
appears as a user of scsi_mod.

> > -- Rescue Mode --
[snip]
> Is there anything interesting near the end of /var/log/syslog or
> /var/log/messages about this?

syslog has some mention.  This is 2.6 rescue mode, by the way.  Here's the tail:

Jul 14 15:30:25 main-menu[1336]: (process:7060): File descriptor 6 left open
Jul 14 15:30:25 main-menu[1336]: (process:7060):
Jul 14 15:30:25 main-menu[1336]: (process:7060): No volume groups found
Jul 14 15:30:25 main-menu[1336]: (process:7060):
Jul 14 15:30:25 main-menu[1336]: WARNING **: Configuring 'rescue-mode'
failed with error code 1
Jul 14 15:30:25 main-menu[1336]: WARNING **: Menu item 'rescue-mode' failed.

Bear in mind that, at this point, no driver is loaded for the hard
drives.  Could this be related to the missing partition warning
screen?

On that subject, I took some notes about the partitioning screen from
the regular install.  As I mentioned, there is no red screen warning
that there are no partitionable disks.  I get to a screen titled "[!!]
Partition disks".  For "Partitioning method:", I am offered only one
choice - "Manually edit partition table".  The dialog also has the
"<go back>" button.

Hope this helps.  Mahalo

~J