"console [tty01] enabled, bootconsole disabled" boot failure

[zogness]

My machine is a UltraSparc 5 with 1GB ram. I'm trying to boot from the same scsi card and disk I had used to run FreeBSD on this computer a couple years ago. Yes, the OBP recognizes the scsi card. The disk has no Sun firmware on it, but it does have a Sun disklabel. The disk is only 18GB.

The subject line of my post is the last line that prints to screen no matter what I do to the kernel config.

I tried changing the default boot args in my latest kernel config and saw the same behavior. The boot stops after, "console [tty01] enabled, bootconsole disabled."

I am not using a separate /boot partition because the install guide says not to. /boot is just a directory on my root partition.

livecd / # fdisk -l /dev/sda

Disk /dev/sda (Sun disk label): 64 heads, 32 sectors, 17501 cylinders
Units = cylinders of 2048 * 512 bytes

Device Flag Start End Blocks Id System
/dev/sda1 0 768 786432 83 Linux native
/dev/sda2 768 4864 4194304 83 Linux native
/dev/sda3 0 17501 17921024 5 Whole disk
/dev/sda4 u 4864 6912 2097152 82 Linux swap
/dev/sda5 6912 17501 10843136 83 Linux native

lspci -k |grep -i scsi
02:01.0 SCSI storage controller: LSI Logic / Symbios Logic 53c875 (rev 14)
Subsystem: LSI Logic / Symbios Logic LSI53C876/E PCI to Dual Channel SCSI Controller
02:01.1 SCSI storage controller: LSI Logic / Symbios Logic 53c875 (rev 14)
Subsystem: LSI Logic / Symbios Logic LSI53C876/E PCI to Dual Channel SCSI Controller

# My silo.conf
partition = 1
root = /dev/sda1
timeout = 10
image = /vmlinuz
label = linux

cd /boot ; ls -la
total 5940
drwxr-xr-x 2 root root 4096 Oct 19 17:08 .
drwxr-xr-x 18 root root 4096 Oct 18 09:29 ..
-rw-r--r-- 1 root root 0 Oct 10 22:34 .keep
-rw-r--r-- 1 root root 1024 Oct 18 01:34 fd.b
-rw-r--r-- 1 root root 512 Oct 18 01:34 first.b
-rw-r--r-- 1 root root 1024 Oct 18 01:34 generic.b
-rw-r--r-- 1 root root 724 Oct 18 01:34 ieee32.b
-rw-r--r-- 1 root root 7720 Oct 18 01:34 isofs.b
-rwxr-xr-x 1 root root 5854336 Oct 18 20:54 linux-2.6.34-gentoo-r11
-rw-r--r-- 1 root root 7680 Oct 18 20:55 old.b
-rw-r--r-- 1 root root 77824 Oct 19 16:26 second.b
-rw-r--r-- 1 root root 91 Oct 19 17:08 silo.conf
-rw-r--r-- 1 root root 75575 Oct 18 01:34 silotftp.b
-rw-r--r-- 1 root root 512 Oct 18 01:34 ultra.b
lrwxrwxrwx 1 root root 23 Oct 19 16:25 vmlinuz -> linux-2.6.34-gentoo-r11

Thanks for any help.

[zogness]

OK. I have the right frame buffers in my kernel and the right module for the on-board ATI Rage Pro 3D but

STILL

the friggin kernel insists on switching to a serial console instead of staying on the monitor.

And for some unknown reason, I can't ssh in. I did enable sshd to start at the default boot level.

This does suck.

[zogness]

<code>
# Automatically generated make config: don't edit
I ended up using the config.gz kernel config from the install cdrom.
</code>

[alexbuell]

Have you checked your OpenPROM settings, in particular the screen variable?

[JFGI_123]

I get the exact same problem using the TFTP boot image. Stops at the same place and everything

Only I don't get the chance to play around with the kernel options

Screen variable?

[alexbuell]

I get the exact same problem using the TFTP boot image. Stops at the same place and everything

Only I don't get the chance to play around with the kernel options

Screen variable?

Yeah, at the OpenProm prompt. type in printenv screen and see what it says. Also, check to see if it's not redirecting console to the serial ports.

[JFGI_123]

output redirection was correct, very strange

However I've got around the problem by re-directing the output a Wyse terminal I had lying around

I'll get on with building the machine and see if I can get it output to the screen with my kernel

[alexbuell]

output redirection was correct, very strange

However I've got around the problem by re-directing the output a Wyse terminal I had lying around

I'll get on with building the machine and see if I can get it output to the screen with my kernel

Yeah, when you rebuild the kernel, you can enable the framebuffer driver required for your platform, and on boot it should correctly bring up the consle.

[zogness]

Yeah, my OBP was (and is) set to screen for the output device. I guess I'd be fine with having output sent to a serial terminal but so far I have not been able to connect using Minicom and the serial cables I have here.

Yeah, I could prolly find a dumb terminal at the MIT flea market but I have enough old computer junk in the house already.

Now I have a new problem.

I have "local-mac-address? = true" in the OBP so both NICs should have their own MAC addresses. Seems like udev is not reading the OBP correctly although I have full OBP support in my kernel. In dmesg I see both NICs (one is on PC bus a and the other on PCI bus b) but ifconfig and ethtool can't see the second NIC. I tried writing a udev rule for the invisible NIC but that failed. Any ideas? Has anyone ever run the FreeBSD installer from Linux? Oh, the Gentoo install disk (cdrom) CAN see both NICs and I'm using a kernel .config from the install iso as the basis for my own .config.

[zogness]

My console problem is fixed but I'm stuck on the MAC address issue. Both NICS are assigned the same MAC address even though local-mac-address?=true in the OBP.

[mattst88]

Please edit your post to put code tags around your kernel .config.

[zogness]

I guess it wouldn't matter if both network interfaces have the same MAC address as log as udevd knew how to deal with that.

The first NIC, which I want to call eth0 is being renamed by udevd to eth0_rename.

I could use the name udevd gives me and change the links in /etc/init.d and the definition in /etc/conf.d/net to eth0_rename but prolly udevd would choke on that too.

Any advice on writing a udev rule to tell udev to ingnore the duplicate MAC address?

Thanks.

[zogness]

I found a workaround for the system-wide MAC address issue:

localhost ~ # for dir in /sys/class/net/* ; do
> [ -e $dir/device ] && {
> basename $dir ; readlink -f $dir/device
> }
> done
eth0
/sys/devices/root/f005f2f8/pci0000:00/0000:00:01.0/0000:02:03.1
eth1
/sys/devices/root/f005f2f8/pci0000:00/0000:00:01.1/0000:01:01.1

So I define the NICs by their PCI bus IDs.

localhost ~ # cat /etc/udev/rules.d/70-persistent-net.rules
ACTION=="add", SUBSYSTEM=="net", BUS=="pci", KERNELS=="0000:02:03.1", NAME="eth0"
ACTION=="add", SUBSYSTEM=="net", BUS=="pci", KERNELS=="0000:01:01.1", NAME="eth1"

But wait! There's more fun ahead!

cat /var/log/messages
Oct 25 14:20:08 sparc udevd[502]: BUS= will be removed in a future udev version, please use SUBSYSTEM= to match the event device, or SUBSYSTEMS= to match a parent device, in /etc/udev/rules.d/70-persistent-net.rules:2
Oct 25 14:31:24 sparc udevd[502]: BUS= will be removed in a future udev version, please use SUBSYSTEM= to match the event device, or SUBSYSTEMS= to match a parent device, in /etc/udev/rules.d/70-persistent-net.rules:1
Oct 25 14:31:24 sparc udevd[502]: BUS= will be removed in a future udev version, please use SUBSYSTEM= to match the event device, or SUBSYSTEMS= to match a parent device, in /etc/udev/rules.d/70-persistent-net.rules:2

[dogshu]

I'm seeing the same "console [tty01] enabled, bootconsole disabled" problem with install-sparc64-minimal-20110509.iso on my blade 100. Unfortunately I don't have a Wyse terminal sitting around. My output-device is set to "screen:r1024x768x60". I've tried the boot parameter video=atyfb:1024x768@60 with no difference.

It seems like "bootconsole disabled" also disables my keyboard, which means I can't blindly type in an ifconfig command. Is there a way to forcibly enable bootconsole?

Also I found this while searching for solutions to this issue, is it possible I'm being affected by the same bug?

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=609466

[dogshu]

Just booted from the positively ancient install-sparc64-minimal-2005.1-r1.iso: it worked fine, I'm setting up the network with my keyboard right now. Is this a bug with newer livecds?

[armin76]

its bug https://bugs.gentoo.org/show_bug.cgi?id=354919

[dogshu]

FYI, new kernels have a "keep_bootcon" kernel parameter which prevents Linux from disabling the boot console. I used this with a 3.0.3 kernel and found that my Blade 100 was kernel panicking:

https://bugzilla.kernel.org/show_bug.cgi?id=41522

[iMike]

I am having the same problem as the original poster with my Ultra 10. Is the solution then, if I can summarize, to:
(1) get the latest firmware (and probably have to use tftp to install it), and
(2) passing the kernel parameter "keep_bootcon"
Is that right? Reading a little about bootcon, e.g., http://us.generation-nt.com/answer/patc ... 21192.html makes it sound like it's really intended for debugging, not so much for regular use.

An alternative seem to be to remove the RSC card, e.g., http://forums.gentoo.org/viewtopic-t-84 ... art-0.html I'm not sure if I have one, but will of course check. However, I never had to remove it before to get screen output.

Anyone have any ideas? I didn't see anything about it in the sparc guide.

I recently upgraded to 3.8.8 and although I get nothing on the screen, I can ssh in and things seem fine. After several hours, though, it seems to crash, and I haven't found any log files with info about why. This is one of the main reason I want to get the boot console working again.

Thanks!

[dogshu]

I ended up having to reset all openprom environment variables to their default in order to boot gentoo properly. I don't remember off the top of my head what the command was to do that, it might be model specific.

[iMike]

Thanks for the answer, dogshu. Unfortunately, I did a "reset-all" via L1-N (Stop-N) during boot up, and the only thing that has gotten me a bit further in tracking down what is wrong with the kernel is using the keep_bootcon kernel parameter. That lets me see that the initramfs is getting loaded and everything, until is stops giving output to the screen, is fine. After that, I can ssh in and see nothing wrong in dmesg or messages. After a certain time running, however, the machine hangs. When I go back to see what went wrong (looking at log files), there is nothing there. So, I still have no idea why the hold up.

If I net boot one of the gentoo tftpboot images, I do get output to the terminal just as one would expect. I wish I knew what setting were used to make that kernel do the right thing.

Still stuck.

PS I somehow figured Oracle would have newer OBP available, since mine is 3.15, but it seems you have to become a paying customer to get that. Anybody know of any old update-OBP tftpboot images around?

[dogshu]

You'll need to know the patch number of the openprom update you're looking for. You should be able to figure it out with some creative googling. Then check to see if its available here (an openprom update for my sun blade 100 is in here somewhere):
http://ftp.sunet.se/pub/security/vendor ... _unsigned/