On Wed, Feb 04, 2010 at 1:01:38AM Jacob Meuser wrote:
>>On Wed, Feb 03, 2010 at 01:06:15PM -0500, Predrag Punosevac wrote:
>> This is the brief preliminary report on Wyse C90LE. I downloaded
>> yesterday the latest snapshot of OpenBSD 4.7 beta for i386.
>>
>> I installed on 2.0 USB of 2GB.
>> I tested the live USB by booting DeLL optiplex 960 (Intel 2 core).
>> Everything works perfect including X server.
>>
>> Conclusion is that USB media is OK.
> are you sure it really booted off the USB?
I am 1000% sure I booted from USB. The Thin Client comes pre-installed
with WindowsXP embedded.
>
>> I broke into the BIOS of Wyse C90LE by holding DEL key. BIOS requires
>> password. The default password is Fireport (found on the web-site of
>> manufacturer). I changed the boot order so that PXE boot is the first
>> and USB 2.0 is the second. I didn't see any other options I could play
>> with in BIOS (like legacy USB or similar).
>>
>> Boot process goes fine and then stops. It is looking for the root device
>> I manually typed wd0a and wd0b for swap (original partition was done
>> by accepting defaults from the installer just for test).
>> The boot process is aborted.
>>
>> ddb{0}>trace
>> Debugger(50,d08cf780,d0a32f48,0,0) at Debugger +0x4
>> panic(d07543e8,1,ffffffff,0,d0863178) at panic +0x55
>> dk_mountroot(2,2,2fecc,d0a32f9c,d04ca223) at dk_mountroot +0x1a7
>> main(0,0,0,0,0) at main +0x4eb
>I've never seen a USB disk come up as wd*, always sd*.
Me neither! I tried to put sd0a as a root directory. I think, I have
very clear idea what is wrong. Look at my second message where
I describe the boot process. Problems occur after four messages
ehci_sync_hs:tsleep()=35
ehci as you know is the driver for USB 2.0. I looked through archives
and I found that people have reported problems with sheety USB
controller on VIA mother boards. Somebody submitted the patch last year.
Obviously even the best software can not fixed crapy hardware.
The OpenBSD aborts the boot because USB is not anymore seen by kernel.
>so your crash makes sense, since you told it to boot off something that
>doesn't yet exist.
Absolutely right. Kernel doesn't see USB drive from that point because
USB is not 2.0 as advertised but it is 1.1. Crapy hardware. Seems that
uhci doesn't pick up from ehci during the boot process.
> sounds like your USB image is not right. it should be booting off
> sd0. when the kernel boots, do you see sd0 in the dmesg?
> you got trace output, so where's the dmesg?
There is no dmesg. The boot process aborts. My hunch is that I would be
able to install OpenBSD on internal flash drive possibly even by booting
with USB. I booted bsd.rd from USB without problems but I didn't want
to erase embedded image because these units are not paid. They were
given to us for evaluation. That is way I wanted to use Live USB.
Jake, thanks a lot for taking the time to think about my problem.
Predrag
