LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Linux - Hardware (http://www.linuxquestions.org/questions/linux-hardware-18/)
-   -   Device IDs and IDE Raids (PCI and Onboard) (http://www.linuxquestions.org/questions/linux-hardware-18/device-ids-and-ide-raids-pci-and-onboard-305059/)

Micah 03-23-2005 08:37 AM

Device IDs and IDE Raids (PCI and Onboard)
 
I'm curious if any of you have any ideas...
Note: I don't know the brand of board, or any information without looking at it and I'm at work so I can't look at the moment.... I can provide this information later - i doubt it will make a difference though...

I have a Linux Box with this setup:
Onboard IDE Controller
Onboard IDE Raid Controller (Highpoint Controller)
PCI IDE Raid Controller SiL0680 (Silicon Image)

Now the fun part...
an lspci command shows something odd in the Device ID's...

Code:

00:09:xx (PCI IDE Raid)
 00:11:xx (Onboard IDE)
 00:13:xx (Onboard IDE Raid)

See something odd there already? The Device ID of the PCI Card is lower than the Onboard ones. In fact All of the PCI Cards have lower IDs than the onboad items. (Video, TV Card, Sound Card, NIC)

In curiosity I bought an Adaptec IDE Raid controller to find out it uses the same Silicon Image 0680 chipset and its device ID is 00:0d:xx..

I am unsure if moving them to different PCI slots will make a difference BUT I will mention again, ALL PCI Devices (Have 6 of 8 atm) have lower IDs than the onboard devices so it doesnt' matter much anyhow...

Whats the problem? I like compiling things into the kernel as it makes things REALLY nice once it's working. Now when I compile the SiL0680 into the kernel something happens...

It reorders my drive 'letters' or 'device ids' (/dev/hda becomes /dev/hde and so forth)

With means then to modify LILO you must modify lilo.con to point top /dev/hde1 instead of /dev/hda1... (There is something else I have to do as well, but I don't remember... something with the boot device and running lilo twice with two configs to make everything work...)

It also means, if you boot from a LiveCD or any kernel without builtin raid (into the kernel itself and not a module) it marks my raids as needing to be rebuilt so... it's bad if you know what that does...

The Options I see are:
1: Compile PCI IDE Raid as Module
2: Live with the above conditions

Drawbacks:
1: As a module, originally (6 months ago) I couldn't start two different modules for two different Raid Controllers - the SiL0680 would be found (the HighPoint worked fine though) -- I'll be trying again soon

2: With the above conditions, if something drastic happened I'd have to wait 2-3 hours for the Raids to rebuild - even just looking at a Live CD caused the Raids to be marked as "BAD" -- At least Linux runs stable most of the time =)

Question:
Does anyone else have any ideas or thoughts on this as it's annoying and I'm rebuilding the box now... (New Case and Hardware changes)

t3gah 04-11-2005 08:32 AM

Re: Device IDs and IDE Raids (PCI and Onboard)
 
Quote:

Originally posted by Micah
I'm curious if any of you have any ideas...
Note: I don't know the brand of board, or any information without looking at it and I'm at work so I can't look at the moment.... I can provide this information later - i doubt it will make a difference though...

I have a Linux Box with this setup:
Onboard IDE Controller
Onboard IDE Raid Controller (Highpoint Controller)
PCI IDE Raid Controller SiL0680 (Silicon Image)

Now the fun part...
an lspci command shows something odd in the Device ID's...

Code:

00:09:xx (PCI IDE Raid)
 00:11:xx (Onboard IDE)
 00:13:xx (Onboard IDE Raid)

See something odd there already? The Device ID of the PCI Card is lower than the Onboard ones. In fact All of the PCI Cards have lower IDs than the onboad items. (Video, TV Card, Sound Card, NIC)

In curiosity I bought an Adaptec IDE Raid controller to find out it uses the same Silicon Image 0680 chipset and its device ID is 00:0d:xx..

I am unsure if moving them to different PCI slots will make a difference BUT I will mention again, ALL PCI Devices (Have 6 of 8 atm) have lower IDs than the onboard devices so it doesnt' matter much anyhow...

Whats the problem? I like compiling things into the kernel as it makes things REALLY nice once it's working. Now when I compile the SiL0680 into the kernel something happens...

It reorders my drive 'letters' or 'device ids' (/dev/hda becomes /dev/hde and so forth)

With means then to modify LILO you must modify lilo.con to point top /dev/hde1 instead of /dev/hda1... (There is something else I have to do as well, but I don't remember... something with the boot device and running lilo twice with two configs to make everything work...)

It also means, if you boot from a LiveCD or any kernel without builtin raid (into the kernel itself and not a module) it marks my raids as needing to be rebuilt so... it's bad if you know what that does...

The Options I see are:
1: Compile PCI IDE Raid as Module
2: Live with the above conditions

Drawbacks:
1: As a module, originally (6 months ago) I couldn't start two different modules for two different Raid Controllers - the SiL0680 would be found (the HighPoint worked fine though) -- I'll be trying again soon

2: With the above conditions, if something drastic happened I'd have to wait 2-3 hours for the Raids to rebuild - even just looking at a Live CD caused the Raids to be marked as "BAD" -- At least Linux runs stable most of the time =)

Question:
Does anyone else have any ideas or thoughts on this as it's annoying and I'm rebuilding the box now... (New Case and Hardware changes)

What kernel is this?

Do you have the kernel patches loaded for the RAID controllers?

What drivers are you loading for the devices in question?

Micah 06-17-2005 12:57 PM

I should read the forum more often...

Kernel is 2.4.28
No patches for RAID devices as it's in the kernel (Don't know where I'd find them anyhow)

I can't name the exact drivers (at work)
the modprobes are:
modprobe siimage
modprobe hpt7xx (I think...)


All times are GMT -5. The time now is 01:39 AM.