Anyone Using CentOS 6 On PC104 Stacks?

Home » CentOS » Anyone Using CentOS 6 On PC104 Stacks?
CentOS 3 Comments

Is anyone using CentOS 6 on PC104 stacks?
I have tried a couple of different CPU cards with CentOS 6.5 [no updates]. I can install with the Install DVD to the PATA hard drive just fine, but they each hang up while the installed system is trying to get UDEV going. And by “hang up” I mean a) leave the machine setting for hours and it never gets udev started, AND b) the keyboard has gone non-responsive (power switch is the only way out).

Note1: each board and drive works fine under CentOS 5.9. Note2: for the purposes of this test all other boards have been removed, it is just the CPU board and an power supply. Granted that does leave the video chip, serial chip, audio chip, usb chip, and memory (which was memtest86 for 50 hours in this config).

Has 6 dropped support for ISA somehow?

Any ideas on what might be wrong or where to start debugging?

I have tried:
Setting selinux to permissive (there was a udev issue in 6.0 where this was needed)

Removing quiet and RHGB from the boot, so I can see what is going on

In /etc/udev/udev.conf set udev_log=”debug”, but this seems to just information overload, AND
It never stops at the same set of messages. Even booted with udevtrace.

Syslog has not started, so there is nothing in /var/log/messages

Moved /etc/udev/rules.d/* to /etc/udev/rules.distro/ so that rules created at install time, under the install dvd’s kernel, are not causing issues. Though I did move one file per boot attempt, newest first. (yes I have reached desperate stage)

Thanks for any pointers.

Even when this disclaimer is not here:
I am not a contracting officer. I do not have authority to make or modify the terms of any contract.

3 thoughts on - Anyone Using CentOS 6 On PC104 Stacks?

  • Todd, can you give us more details on the PC104 CPU hardware?
    Manufacturer and model would be nice, too.

    My only PC104 CPU’s here are 486’s, and thus C6 is not going to boot (C6
    requires PAE and i686).

  • Sorry for the delay, I had a couple of other interrupts to deal with. As can be seen below, both CPU cards include PAE support. Both were running with just the CPU card (with its built-in peripheral interfaces), hard-drive and power supply. While collecting the information below, I played with kernel options and happened to find that with the RTD cme37786hx I could get it to boot reliably from the hard-drive if I included
    ” noapic nomodeset udevtrace” (in that order) replacing the “quiet rhgb” *AND* either use the CentOS 6.3 install kernel or update to the latest CentOS 6.5 kernel. The CentOS 6.5 DVD installed kernel would pause forever just after trying ‘edd’.

    As for the VL-EPMs-21b, even though it worked OK in CentOS 5.9, it would always stop in the UDEV
    startup of CentOS 6.5 even after updating to the latest 6.5 kernel and using the above kernel options. Interestingly though while working with it, I noticed a little brown cube on the ESD bench where the board had been setting. Who would think that a missing capacitor might stop a new kernel, with a much faster IO engine, from accessing all the hardware in parallel (max power pull) at entry into user space? :)
    Beginning to think this one is not CentOS’s fault. :)
    Fortunately I am expecting to get a new, not used before I got it, version … sometime. :)

    So I guess it is sort of solved for now. Thanks for taking a look.
    —

  • I would have thought this one would have worked, but a missing cap might cause that failure. I have seen cap failures do strange things like this; once I was troubleshooting a radio station automation computer that passed POST and several diags but would hang going into Windows XP
    (it hung going into a liveCD of a Linux distribution, too, so it wasn’t Windows’ fault). I asked the DJ if anything unusual had happened when it first locked-up, to which she said, ‘well, not really, except that gunshot I heard.’ Turned out a capacitor had shot its casing with enough force that it dented the steel case cover; but the machine would run fine until enough power was being drawn to shut down the regulators. Again note that this machine not only passed POST but also memtest86.

    The VIA Eden isn’t officially support, but yours appears to have cmov, which is the usual problem with them and CentOS 5+.

    I haven’t personally tried an ISA-capable motherboard in a while with any modern CentOS……