DragonFlyBSD

How far did it go, such as whether you saw the twirling bar, the DragonFly
menu, the copyright notice, or so on. If you can get to the boot menu,
please try booting with ACPI disabled to see if it changes anything.
I know it's hard because our boot loader can't properly initialize
the keyboard (and I see it on other OSes, too). You'll have to try
rebooting until the keyboard works.

It boots fine on MacBook Pro(core 2 duo) here, FYI. It may or may not
have something to do with the firmware version. Updating the firmware
on a Mac is a scary operation, though.

Cheers.

It comes until some stage of device initialization. The last thing I
could see was probing of uhub or something like that.

So my guess is that the usb keyboard doesn't attach correctly and panics
the kernel. ddb pops up and can't read from they console, thus leading
to an endless db> output loop.

[to avoid confusion: I was standing next to Daniel when he was trying
to boot]

That's very good news! Can you work with it or are there still some
major problems?

cheers
simon

for the record, I'm guessing this is via boot camp?

Just to throw another bit of data in there: here's my MacBook Pro (first
gen) attempting to boot 1.10.1 without ACPI on:

http://www.shiningsilence.com/df_on_mbp.jpg (2048x1536 jpeg)

The same last few lines are what show with ACPI enabled. I don't get the
db> prompt - it just stops on those last few livelock lines, and goes no
further for the minute or so that I've sat and watched it.

Yes, and it boots on Parallels, too.
Too bad I don't have a digital camera to capture the screen.

Cheers.

Can scroll back to find which driver is sharing irq 10?
I remember seeing this stray interrupts/interrupt storm messages when
I was trying to port keyboard changes from FreeBSD, and I don't remember
when I stopped seeing them. I've been under the impression that some
changes in usb driver at that time have fixed that...

Cheers.

Yes via bootcamp and disabling ACPI doesn't change anything. These keyboards are
usb (ukbd), btw is there any way to boot "step-by-step" in debug mode ? (I know
it sounds like "shift+f8", but it would help a bit :P)

cheers,
Daniel

On Tue, September 11, 2007 11:03 pm, YONETANI Tomokazu wrote:

was trying to port keyboard changes from FreeBSD, and I don't remember
when I stopped seeing them. I've been under the impression that some
changes in usb driver at that time have fixed that...

I booted with a Dell USB keyboard plugged in, as there's no scroll lock or
pg up/pg down keys to let me scroll back. No serial port, either...
There's a few different behaviors:

- If I boot with ACPI off, and don't touch anything on the attached USB
keyboard, it has the same behavior as before - stops on that livelock
message. The USB keyboard is powered at that point - the num lock or
scroll lock lights will light up.

- If I boot with ACPI off, and start hitting the scroll lock key or
anything else during boot to see what happens, I get the runaway db>
prompt as the original poster described. The keyboard will not light up
for any of the lock buttons.

- Booting with ACPI on, it stops with:

intr 11 at 40001/40000 hz, livelocked limit engaged!
intr 10 at 12342/20000 hz, livelock removed

Booting with ACPI on and with that Dell keyboard attached, I see it
loading up fwohci0 and firewire0 - no uhci messages as before, and it
pauses at

ata0: <ATA channel 0> on atapci0
intr 10 at 2205/20000 hz, livelock removed

and then after a while (perhaps because I left it alone while typing this)
it continues booting. There's a lot of the 'intr 10' livelock
engaged/removed messages - probably 1-2 per second.

Anyway, since I can see dmesg.boot, here's what says IRQ 10:

pcib1: <PCI to PCI bridge (vendor=8086 device=27a1)> irq 10 at device 1.0
on pci0
pci1: <ATI model 71c5 graphics accelerator> at 0.0 irq 10
pci0: <unknown card> (vendor=0x8086, dev=0x27a3) at 7.0 irq 10
pcib3: <PCI to PCI bridge (vendor=8086 device 27d2)> irq 10 at device 28.1
on pci0
uhci3: <UHCI (generic) USB controller> port 0x4040-0x405f irq 10 at device
29.3 on pci0

I haven't rebooted yet to see if I can repeat this semi-successful boot.

We don't have msk(4); it seemed to work without a problem on FreeBSD 7
(which has been installed until I tried to install DragonFly). We don't
have a networking without it.
The installer went fine until it failed installing the boot manager.
It seemed that fdisk was complaining about something was not multiple
of sector size(sorry, forgot writing down the log).

Cheers.

I am also experiencing these problems with my MacBook (13 inch) with
and without ACPI disabled. the db> prompt loops and impossible to get
a back trace in both cases.

Regards,
Jonas Trollvik