Bug#271181: installation-reports: complete failure with 2.6 kernel on compaq proliant 3000
Package: installation-reports
Version: N/A; reported 2004-09-11
Severity: critical
Justification: breaks the whole system
Debian-installer-version: sarge-i386-netinst.iso RC1, downloaded 2004-09-10
uname -a: n/a
Date: 2004-09-11
Method: netinst CD based install with "expert26"
Machine: Compaq Proliant 3000
Processor: 2x PII-450
Memory: 768MB
Root Device: SCSI, /dev/sda1
Root Size/partition table:
8 0 2055000 scsi/host0/bus0/target0/lun0/disc 8598 41254 99716 58900 8542 149865 317002 2158920 0 141590 2217820
8 1 1927768 scsi/host0/bus0/target0/lun0/part1 8594 41248 99690 58820 8542 149865 317002 2158920 0 141510 2217740
8 2 1 scsi/host0/bus0/target0/lun0/part2 0 0 0 0 0 0 0 0 0 0 0
8 5 120456 scsi/host0/bus0/target0/lun0/part5 1 0 8 20 0 0 0 0 0 20 20
Output of lspci and lspci -n:
0000:00:00.0 Host bridge: ServerWorks CNB20-LE Host Bridge (rev 04)
0000:00:00.1 Host bridge: ServerWorks CNB20-LE Host Bridge (rev 02)
0000:00:03.0 System peripheral: Compaq Computer Corporation Advanced System Management Controller
0000:00:06.0 PCI bridge: Intel Corp. 80960RP [i960 RP Microprocessor/Bridge] (rev 05)
0000:00:06.1 Memory controller: Intel Corp. 80960RP [i960RP Microprocessor] (rev 05)
0000:00:09.0 Network controller: Compaq Computer Corporation Netelligent 10/100 TX PCI UTP (rev 10)
0000:00:0f.0 ISA bridge: ServerWorks OSB4 South Bridge (rev 4d)
0000:00:11.0 Host bridge: ServerWorks CNB20-LE Host Bridge (rev 04)
0000:00:11.1 Host bridge: ServerWorks CNB20-LE Host Bridge (rev 02)
0000:01:00.0 VGA compatible controller: ATI Technologies Inc 3D Rage IIC 215IIC [Mach64 GT IIC] (rev 7a)
0000:02:04.0 SCSI storage controller: LSI Logic / Symbios Logic 53c875 (rev 14)
0000:02:04.1 SCSI storage controller: LSI Logic / Symbios Logic 53c875 (rev 14)
0000:02:08.0 Ethernet controller: Intel Corp. 82557/8/9 [Ethernet Pro 100] (rev 05)
0000:00:00.0 0600: 1166:0007 (rev 04)
0000:00:00.1 0600: 1166:0005 (rev 02)
0000:00:03.0 0880: 0e11:a0f0
0000:00:06.0 0604: 8086:0960 (rev 05)
0000:00:06.1 0580: 8086:1960 (rev 05)
0000:00:09.0 0280: 0e11:ae32 (rev 10)
0000:00:0f.0 0601: 1166:0200 (rev 4d)
0000:00:11.0 0600: 1166:0007 (rev 04)
0000:00:11.1 0600: 1166:0005 (rev 02)
0000:01:00.0 0300: 1002:4756 (rev 7a)
0000:02:04.0 0100: 1000:000f (rev 14)
0000:02:04.1 0100: 1000:000f (rev 14)
0000:02:08.0 0200: 8086:1229 (rev 05)
Base System Installation Checklist:
[O] = OK, [E] = Error (please elaborate below), [ ] = didn't try it
Initial boot worked: [O]
Configure network HW: [E]
Config network: [O]
Detect CD: [O]
Load installer modules: [O]
Detect hard drives: [O]
Partition hard drives: [O]
Create file systems: [O]
Mount partitions: [O]
Install base system: [O]
Install boot loader: [O]
Reboot: [E]
Comments/Problems:
The problem is with hardware auto-discovery -- the system has a compaq
"remote insight board" (lights out management hardware) at PCI
0000:00:06.0 and 0000:00:06.1, which is mis-detected as a megaraid
controller, and "insmod megaraid" then hangs.
This happens during non-expert install, but can be worked around by
using expert mode and deselecting the megaraid driver when is is
detected. Still, when rebooting from hard disk (grub in the drive's
boot sector), the autodetection in the initrd makes the same error, so
the system never completes the boot.
This means it's impossible to get a working 2.6 sarge install with
this card in the machine, since the "rescue" entry in the grub menu
isn't able to get around the initrd hw-detection stuff, and even
mounting the root fs from a knoppix boot and deleting the hw-detection
stuff doesn't help, as it is still in the initrd.
Note that the hardware detection seems to be severely broken WRT the
megaraid card, as it also shows up in the list of modules offered
after "detect network hardware" (together with the modules for the two
NICs that are correctly detected; no other modules are mis-detected as
network cards).
In the end, I booted the 2.4 installer and installed the 2.4.26
kernel, and it worked. Still had to deselect the megaraid module
during the install, but the installed OS boots fine.
As a personal comment: I think this "hold the user's hand" stuff is
going in the wrong direction for debian -- If I need a distro which
thinks it is smarter than me (thereby breaking stuff, as demonstrated
here), I can use SuSE, fedora or mandrake.
I would strongly suggest either
- removing hardware autodetection from the initrd and instead going
with the fixed list of modules which were detected/used with d-i.
- providing some way to get around the autodetection in initrd (maybe
there is such a way, but the d-i manual and FAQ don't describe it)
cm.
Reply to: