3.3 & 3.2 kernel panic: extent_free: region not found

[4ltern4te]

I've got a new Dell PowerEdge 600SC-
2.4ghz P4, two 40gig IDE's
with a ServerWorks CMIC_SL Host board.
Every attempt to boot OpenBSD 3.2 or 3.3 ends with:

extent_free: start 0x36e, end 0x371
panic:extent_free: region not found

After this, any key pressed results in a reboot of the system. I can't find this specific board on the hardware list, but NetBSD 1.6 supports it, so would it be a huge undertaking to add support for this board from the NetBSD source tree? (I'm an OpenBSD newbie)
Could I manually add the relevant support/lines to the generic /bsd file on my cd before burning and booting?
I don't have a dmesg, I can't even drop to a shell-

Thanks for your help-

 

[4ltern4te]

I answered my own question. The fix was simple.
The hard drives were jumpered to "cable select",
and the cdrom drive was "cable select" on the third IDE controller. I set the hard drives in a master/slave config,
changed the cdrom to be the secondary master, and she booted right up.

 

[Shutterbug79]

4ltern4te,

I have exactly the same issue as you. The same error message etc. Your solution didn´t solve my issue. Have you got some more details to share?

Regards,

Johan

 

[4ltern4te]

Johan-

I think it is a DMA issue. The a.out version of 3.3(stable) does not seem to recognize the correct DMA modes for the IDE drives.
Have you tried disabling DMA mode entirely?
at the boot> prompt, enter "-c" (no quotes)
to drop into the UKC (user kernel configurator, I think)
[ Or, just do this (as root) #config -e -o /bsd.new /bsd ]
At the UKC> prompt,
type "change pciide*", (no quotes)
change the "flags" to 0x0fffc
do this for the following devices:
pciide*
atapiscsi*
wd*
cd*
type quit to exit the UKC
This should at least enable you to boot the machine and continue the install. Remember, UKC changes are not permanent, and will be lost on your next reboot. If the machine boots sucessfully after changing these device flags, write them to /bsd like this:
config -e -o /bsd.new /bsd (make UKC changes to dev flags)
cp /bsd.new /bsd
reboot

Hope this helps-

4ltern4te

 

[Shutterbug79]

4ltern4te,

Thanks for trying to help me solve my issue. I think you’re right; seems to be a DMA issue. I followed your instruction and disabled DMA mode but without result. I got the same error message;
pciide0: chanel 1 ignored (not responding; disabled or no drivs?)
extent_free: start 0x36e, end 0x371
panic:extent_free: region not found

My machine is a Dell PowerEdge 600SC out of the box. I´m trying to install the 3.3 stable.
In your first post, you wrote about driver jumpering in your solution. Does it matter?
When using “cable select” I don’t get the error message above, the machine just freeze.

I have tried to find a solution through misc etc with out result. Any further ideas?

Thanks in advance

-Johan

 

[4ltern4te]

Johan-

Ok. Have you opened up the server yet?
Dell ships the poweredge with the hard drives attached to the third IDE controller. Windows doesn't mind this, but *nix chokes on it.
Open the case, and verifiy that your hard drives are jumpered to master/slave configuration on the primary IDE channel (IDE_0 on the board) Next, make sure that any additional IDE drives are in master/slave config on the secondary IDE channel. I haven't even experimented with using the third IDE channel with BSD, so I can't help you with it. I had to make these changes to my 600SC to even get it to boot, so this is probably your problem, too.
I'm not entirely sure if it matters all the time, as I have OpenBSD running on drives in "cable select" configuration (on 100% hardware-compliant boxes), but when working around any hardware issues, it always seems that the more you can explicitly define, (including in CMOS, are your IDE drives set to be auto-discovered?) the better the results.

4ltern4te

 

[robertostraussman]

Hello,

I am new to posting, so please forgive any incorrect procedures, and the length of this first post.

I have been following this thread, and am very appreciative to those who have responded with helpful information. Here is the situation:

First, I found this post because a client of mine received a PE 600sc, and was unable to boot. He found the same issues as the first post above (ide-3, and cable select for both). Once he switched the drives to master, and stopped using ide-3, his machine booted without issue.

Now, another client is having the same exact problem, however with the new PE 400sc's. In the 400sc's, there is no ide-3, therefore this is not part of the problem. I changed the drives to master, and tried to boot. Same error. At that point, I tried disabling the DMA per 4ltern4te's suggestion, however, when leaving the UKC, the machine automatically reboots, therefore discarding the changes I just made.

Two questions:

1) any thoughts on leaving the UKC without rebooting....
2) Above, Johan says he tried the disabling DMA, without success, and then 4ltern4te responded about the ide-3..adn also mentions the cmos: 4ltern4te, in the cmos, do I leave the ide config to AUTO?

Thank you in advance for any help that anyone might be able to provide.

Roberto

 

[4ltern4te]

Roberto-

Your machine reboots when you exit the UKC?

4ltern4te

 

[4ltern4te]

Have you tried specifying /bsd.new at the boot> prompt?

4ltern4te

 

[robertostraussman]

Hello 4ltern4te,

Thank you for your response.

To your first question, yes the machine automatically reboots after typing quit at the UKC prompt.

To your second quesiton, no I have not tried that because I was unable to get it to boot at all. Where would the /bsd.new be if I was unable to issue that command you suggested: config -e -o /bsd.new /bsd? Should I try that command from the boot prompt? or the UKC prompt?

Thank you again for your help!

 

[4ltern4te]

If the machine doesn't continue to boot after leaving the UKC,
there must be no mountable, writable media available for the UKC to write the /bsd.new file. (Or no recognized drive devices on the IDE channel)
Is(are) the drive(s) specified as type "auto" in the BIOS?
What is your current config on this machine?
Try explicilty defining the hard drive params in the bios.
(most new bios's will discover the IDE drive params for you, then just save and use that info (instead of 'auto') in the general bios settings - )
Make sure you have them properly jumpered for master/slave, and on the primary IDE channel.
If you can get the UKC to be able to mount the / and write the /bsd.new, you're set.
You might want to look for a setting in your bios to see if you can manually turn off DMA mode there, too.

4ltern4te

BTW- are we talking about 3.3-STABLE, or a -CURRENT snapshot?

 

[halligas]

I am having the same problem that robertostraussman describes with the 400sc's. Specifically:

1) There is no tertiary ide (like the 600sc) on the board or in bios to disable

-BUT-

2) The error comes up after detecting pciide1. You may guess that pciide1 is the secondary channel, but it is not. It works like this:

pciide0 channel 0 is the primary ide (so in the default hardware config of the 400sc this the HD would be channel 0 device 0 (master))
pciide0 channel 1 is the secondary ide ( with device 0 being the cd-drive)

So what is this pciide1 that the boot up to the openbsd 3.3 install is finding?

I have no idea. But there seems to be no way to disable it in the bios or on the board, and since the pciide line in the kernel conf is pciide*, you can't disable it there either (without disabling pciide altogether)

-steve

 

[halligas]

Ok, I have the 400sc figured out.

The phantom pciide1 that the install boot is finding is the Intel SATA controller, which shows up in the pci bus scan even though it is disabled in the bios.

The solutions therefore are:

400sc: Use a snapshot of OpenBSD current or OpenBSD 3.4 when it comes out.

600sc: The above works, but you can also do the following and get 3.2 or 3.3 to work. Move the CD-drive to the secondary ide channel and disable the tertiary IDE channel in bios.

-steve