Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
Frame buffer console resolution on Alpha [SOLVED]
View unanswered posts
View posts from last 24 hours

Goto page Previous  1, 2  
Reply to topic    Gentoo Forums Forum Index Desktop Environments
View previous topic :: View next topic  
Author Message
Altivo
n00b
n00b


Joined: 01 Aug 2015
Posts: 49
Location: Northern Illinois

PostPosted: Mon Aug 10, 2015 3:15 am    Post subject: Reply with quote

The 'nofb' option did nothing used as a keyword. Should I have used 'video=nofb' instead?

The genkernel completed after many hours. I changed my aboot.conf first line to boot from the newly created kernel and initramfs. It did the same thing every custom kernel I've tried has done. It booted up, inspected memory and the bus, loaded the ramdisk, found the scsi devices, etc. Then it arrived at

    Freeing unused kernel memory 240K ...

And froze. Nothing more. No network connection, console dead, had to power down and restart.

Now I've really got a problem. It would appear that this time some module got replaced that keeps the original kernal from the LiveCD from booting. That now does the same thing, stops at the same point. The only difference there is that it is freeing 442K instead of 240K.

I can no longer boot to my root fs. I had to boot the LiveCD itself and mount the root partition in order to retrieve the data you requested.

Here is the config for the LiveCD kernel, retrieved from /proc/config.gz

The new kernel, initrd, and modules were generated this way:
Code:

cd /root
zcat /proc/config.gz > gentoo-config
genkernel --kernel-config=/root/gentoo-config --menuconfig --install all


This ran for several hours. At the beginning it did a 'make oldconfig' and a 'make mrproper' so I assume it rebuilt everything. It did create new kernel and new initramfs and place them in /boot along with a matching System.map. And It did install new modules.

Here is the resulting /usr/src/linux/.config

Note that the Gentoo version in the LiveCD config is 3.17.7, while the new kernel config is 4.0.5. In the menuconfig, I made only a couple of small changes (srm_conf changed to resident in kernel from module, ext3 file system made resident, ALSA turned on, and a couple of the compression methods for the initrd turned off.)

Here is the output of lspci.The first IDE has a Toshiba CD-ROM on it. The second is unused. The ethernet comes up as enp0s3: after boot. There is a floppy drive, attached somehow to the 82c693 I believe. Neither the CD-ROM nor the floppy disk are visible in the Gentoo system after it boots. (Debian is able to use both, but OpenBSD also fails to find them.) Standard PS/2 type mouse and keyboard are plugged in but do not appear here.

Code:

00:03.0 Ethernet controller: Digital Equipment Corporation DECchip 21142/43 (rev 30)
00:07.0 ISA bridge: Contaq Microsystems 82c693
00:07.1 IDE interface: Contaq Microsystems 82c693
00:07.2 IDE interface: Contaq Microsystems 82c693
00:07.3 USB controller: Contaq Microsystems 82c693
00:14.0 PCI bridge: Digital Equipment Corporation DECchip 21152 (rev 02)
01:04.0 SCSI storage controller: QLogic Corp. ISP1020 Fast-wide SCSI (rev 05)
01:09.0 VGA compatible controller: Texas Instruments TVP4020 [Permedia 2] (rev 01)


The fdisk provided in the LiveCD is unable to recognize or read the OSF/BSD disk label. I used the Debian fdisk to retrieve that, and here it is.
Code:

BSD disklabel command (m for help): p

4 partitions:
#       start       end      size     fstype   [fsize bsize   cpg]
  a:        1*      131*      130*      swap                     
  b:      131*      141*       10*      ext2                     
  c:        1      2213      2213     unused        0     0       
  d:      141*     2213      2072*      ext2                     



Partition a (/dev/sdd1) is swap. Partition b (/dev/sdd2) is /boot. Partition d (/dev/sdd4) is the root file system. Partition c, as normal for a BSD label, is the entire disk excluding the bootstrap cylinder. BSD partitions can overlap, but overlapping partitions are not to be used at the same time. I don't know the reason for this rule. I just don't ever mount partition c. The drive is a Seagate Cheetah 18 GB. Note that partition d, the root, is actually ext3 but the BSD file label has no code to indicate this. So when I mount it, I always state explicitly that it is type ext3. The label is being displayed here using cylinder numbers for the partition boundaries, though they were allocated by sector counts instead. Where a cylinder number has a * after it, that means that part of the cylinder is included, and the rest belongs to the prior or following partition. Sorry, I forgot that was the default display and should have changed it to sectors. Nothing overlaps though, except for partition c.

Here is the boot parameter list from aboot.conf:
Code:

0:2/vmlinux0809 initrd=initrd0809 console=ttyS0 rootfstype=ext3 root=/dev/sdd4 vga=0x305 video=pm2fb:1024x768-16@70


In case you are unfamiliar with the format, the "0:" just identifies this configuration as the default boot (nine others may be included and selected on the boot command.) The 2 is the partition number (equivalent to b) of the boot disk that contains the boot files (/boot) and is followed by the path and file name for the kernel. The rest of the options are passed to the kernel just as given. A typical boot command to the SRM console monitor would be "boot -flags 0 dka3" which means to boot according to line "0" of /etc/aboot.conf in the root partition on device "dka3" (VMS-speak for /dev/sdd in Gentoo terms.) For the purpose of this command, the root partition is identified by a field in the boot sector, and can be changed if necessary by a utility command. The aboot loader requires that boot and root partitions be in either ext2 or ext3 format. In order to use anything else (jfs or ext4 for instance) another partition (or several) must be made and mounted after bootup. (Normally done via /etc/fstab.) The root partition must contain /etc and the mount points for the other partitions in that case.

And now that I can't boot the machine, I have turned it off for the night. Hope you can figure out what I did wrong. I'm at a loss here. An attempt to boot the LiveCD kernel, using its own initramfs but still mounting my boot partition now freezes up at the same point, as I said. This is the way I have been booting for several days until now.

Actually, I do have two full backups of the root and boot partitions, made yesterday and three days ago. If all else fails, I can restore from one of those rather than starting all over again.
_________________
Have patience with me, I'm only half-baked.
--Altivo
Back to top
View user's profile Send private message
Altivo
n00b
n00b


Joined: 01 Aug 2015
Posts: 49
Location: Northern Illinois

PostPosted: Mon Aug 10, 2015 2:02 pm    Post subject: Reply with quote

Addendum: I restored the rootfs from my backup, and it is bootable again using the LiveCD kernel.

I will await advice on generating a functional kernal, since I've failed at least six times now. Perhaps this tail end to the framebuffer thread ought to be moved to a different spot? (Since it's no longer about the framebuffer, though that issue remains unresolved.)

While booting up after the the restore this morning I tried 'vga=0x305 video=nofb' and still got the same issues on the CRT display, except this time they were at 640x480 and 8 bit color. Tux had green feet. I think it was still the framebuffer, because ordinary VGA on this card works fine for every other OS. The kernel from the LiveCD does try to start the framebuffer.
_________________
Have patience with me, I'm only half-baked.
--Altivo
Back to top
View user's profile Send private message
NeddySeagoon
Administrator
Administrator


Joined: 05 Jul 2003
Posts: 43755
Location: 56N 3W

PostPosted: Mon Aug 10, 2015 7:52 pm    Post subject: Reply with quote

Altivo,

This is a good start. Your HDD uses a BSD disk label and allows the kernel to read it.
Code:
CONFIG_MSDOS_PARTITION=y
CONFIG_BSD_DISKLABEL=y


Code:
CONFIG_DEVTMPFS=y
CONFIG_DEVTMPFS_MOUNT=y
This tells the kernel to populate /dev

Code:
CONFIG_SCSI_MOD=y
# CONFIG_RAID_ATTRS is not set
CONFIG_SCSI=y
CONFIG_SCSI_DMA=y
and
Code:
CONFIG_BLK_DEV_SD=y
CONFIG_CHR_DEV_ST=m
# CONFIG_CHR_DEV_OSST is not set
CONFIG_BLK_DEV_SR=y
CONFIG_BLK_DEV_SR_VENDOR=y
CONFIG_CHR_DEV_SG=y
is all good stuff. CONFIG_CHR_DEV_ST=m is for a SCSI tape drive.

Code:
CONFIG_SCSI_QLOGIC_1280=y
is for your SCSI controller. All the other things in the Low Level SCSI menu can be off. It will save kernel build time.
Code:
CONFIG_PATA_CYPRESS=y
looks like your 00:07.1 IDE interface: Contaq Microsystems 82c693

Code:
# File systems
#
CONFIG_EXT2_FS=y
CONFIG_EXT2_FS_XATTR=y
CONFIG_EXT2_FS_POSIX_ACL=y
CONFIG_EXT2_FS_SECURITY=y
CONFIG_EXT3_FS=y
# CONFIG_EXT3_DEFAULTS_TO_ORDERED is not set
CONFIG_EXT3_FS_XATTR=y
CONFIG_EXT3_FS_POSIX_ACL=y
CONFIG_EXT3_FS_SECURITY=y
covers the root fs.

and all the useful Pseudo filesystems too

Code:
CONFIG_PROC_FS=y
CONFIG_PROC_KCORE=y
CONFIG_PROC_SYSCTL=y
CONFIG_PROC_PAGE_MONITOR=y
CONFIG_KERNFS=y
CONFIG_SYSFS=y
CONFIG_TMPFS=y


A few odds and ends, all set correctly
Code:
CONFIG_BINFMT_SCRIPT=y
CONFIG_FHANDLE=y
CONFIG_NET=y

genkernel builds a fully modular kernel. From the above, it looks like you have flipped =m to =y for everything you need to get booted, so you don't need an initrd.

The next line that you don't see, should be processing uevents.

What version of udev do you nave installed?
Its a little picky about kernel versions that it works with. It also needs to be in the sysinit runlevel, but root needs to be mounte to get init to run.

Your kernel looks good.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.
Back to top
View user's profile Send private message
Altivo
n00b
n00b


Joined: 01 Aug 2015
Posts: 49
Location: Northern Illinois

PostPosted: Mon Aug 10, 2015 9:55 pm    Post subject: Reply with quote

The machine isn't running at the moment and I have to go feed horses and put them in for the night, but I'll check on udev and get back to you as soon as I can.

I did not install it directly, though I would expect it was included in the "stage3" tarball that I unpacked onto the root partition as part of the installation, if it wasn't already in the portion that came from the LiveCD.

I have noticed in the dmesg from booting with the LiveCD kernel (which was pastebinned earlier in this thread) that this message appears at about the 32 second mark:

    systemd-udevd[2167]: starting version 216


Is that the version number you needed? I'll see what else I can find in a bit.

Thanks for checking the config for me. I was sure it made sense, and you're right that I hoped ultimately to boot without the initrd, but when it just wouldn't work I figured I had better try the ramdisk method.

When I try to boot the kernel I configured without initrd, though, it gets a kernel panic. The message is not clear to me but it appears not to be able to find /dev/sdd4 (the rootfs) or else the mount point. And this is after several messages that seem to confirm that the proper volume and partition has in fact been located and even read (which had to happen much earlier too in order to read in the kernel parameter string.)
_________________
Have patience with me, I'm only half-baked.
--Altivo
Back to top
View user's profile Send private message
Altivo
n00b
n00b


Joined: 01 Aug 2015
Posts: 49
Location: Northern Illinois

PostPosted: Mon Aug 10, 2015 11:48 pm    Post subject: Reply with quote

It appears udev is version 216, though I find it difficult to confirm. I expected to find /sbin/udevd, ask it for a version, and get an answer. This works on my aging Ubuntu i386 box.

However, on gentoo-alpha I find no udevd, nor any systemd-udevd to inquire of. Running find on /usr/portage comes up with at least a dozen versions of udev and eudev.

Now I admit ignorance. How do I find out what version is on this machine?
_________________
Have patience with me, I'm only half-baked.
--Altivo
Back to top
View user's profile Send private message
Altivo
n00b
n00b


Joined: 01 Aug 2015
Posts: 49
Location: Northern Illinois

PostPosted: Tue Aug 11, 2015 11:59 am    Post subject: Reply with quote

OK, got it this morning. systemd-udevd is hiding in /lib/systemd rather than /usr/lib/systemd. The installed version is indeed 216.
_________________
Have patience with me, I'm only half-baked.
--Altivo
Back to top
View user's profile Send private message
NeddySeagoon
Administrator
Administrator


Joined: 05 Jul 2003
Posts: 43755
Location: 56N 3W

PostPosted: Tue Aug 11, 2015 5:24 pm    Post subject: Reply with quote

Altivo,

Thats a new enough udev for the 4.0.5 kernel.

A few things to check ...
.config:
CONFIG_UEVENT_HELPER=y
CONFIG_UEVENT_HELPER_PATH="/sbin/hotplug"
Do you have a /sbin/hotplug?
The right answer in no. udev takes care of this now.

/sbin/hotplug was a shell script. If you have it, it will be run once for every kernel event generated.
This can take time and use a lot of RAM. If you have it, move it out of they way.

In your chroot, what does
Code:
ls -l /etc/runlevels/*
show?
Code:
/etc/runlevels/sysinit:
total 0
lrwxrwxrwx 1 root root 17 Jul  9  2011 devfs -> /etc/init.d/devfs
lrwxrwxrwx 1 root root 17 Jul  9  2011 dmesg -> /etc/init.d/dmesg
lrwxrwxrwx 1 root root 18 Oct 20  2012 sysfs -> //etc/init.d/sysfs
lrwxrwxrwx 1 root root 24 Sep 20  2013 tmpfiles.dev -> /etc/init.d/tmpfiles.dev
lrwxrwxrwx 1 root root 16 Jul  9  2011 udev -> /etc/init.d/udev
lrwxrwxrwx 1 root root 24 Aug  2 11:06 udev-trigger -> /etc/init.d/udev-trigger
are all needed to get you started.
They are all symlinks and they must all be Cyan (pale blue).
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.
Back to top
View user's profile Send private message
Altivo
n00b
n00b


Joined: 01 Aug 2015
Posts: 49
Location: Northern Illinois

PostPosted: Tue Aug 11, 2015 8:21 pm    Post subject: Reply with quote

Ahh!

/usr/src/linux/.config has
Code:

CONFIG_UEVENT_HELPER=y
CONFIG_UEVENT_HELPER_PATH=""


So the helper path really should say "/sbin/hotplug"?

That may be the issue.

No, there is no actual /sbin/hotplug in the root file system now.

Now as for the question about chroot, I confess you've lost me. By "In your chroot" are you referring to the contents of the initrd file system? Something on the actual final root file system? I'm not sure where to look, and in the case of the initrd, not sure HOW to go about looking.

I did connect up another computer in place of the VT220 on the console port, and run through the failing boot command this morning so I could capture the actual messages. Since we have no dmesg, perhaps these will help clarify what is going on.

Failed boot of custom kernel on Alpha PWS 433au

Aside from the fact that the machine doesn't complete the initialization, I see that it doesn't find its sound card or floppy drive. It may have found the Toshiba CD-ROM, I'm not sure. You're right that PATA-Cypress should be the correct module to get to this stuff. One other oddity I see, around 30 lines down, is

    Booting on Miata using machine vector Miata from MILO

MILO is the boot loader used with ARC BIOS, rather than aboot with SRM BIOS which was actually used here. This could be because I selected machine architecture Miata in the config, rather than leaving it as undefined. Miata is indeed the architecture here, but it can be used with either ARC/MILO or SRM/aboot for booting.
_________________
Have patience with me, I'm only half-baked.
--Altivo
Back to top
View user's profile Send private message
Altivo
n00b
n00b


Joined: 01 Aug 2015
Posts: 49
Location: Northern Illinois

PostPosted: Wed Aug 12, 2015 2:37 am    Post subject: Reply with quote

Thanks so much, Neddy. That (/sbin/hotplug) was the key.

Recompiled the kernel with that change, and also turned on "Boot from SRM" which I had somehow missed.

Still using initrd, may not have to, but it boots up and runs. The MILO message changed to SRM, and the SRM EEPROM values are available in /proc/srm_environment where they should be.

The floppy drive appears to be available now. The internal CD-ROM still is not, and will require some research.

While adjusting the kernel, I turned off the framebuffer choices. I now have six text console screens as I expected, and I can switch to the local keyboard and CRT to boot the machine as well.

The next step should be easy: add ntpd to keep the clock accurate.

After that I have to look into getting X11 operating.

Your help has been most invaluable. I had run diffs between configurations repeatedly and still somehow missed that uevent hotplug issue. I'm not sure how it got blanked out, but blank it was.

I may yet be able to use this Alpha for a regular desktop machine. I've tried OpenVMS, Debian, and OpenBSD but none of them were up to the task. Gentoo looks promising once I get it all installed and configured.

:wink:
_________________
Have patience with me, I'm only half-baked.
--Altivo
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Desktop Environments All times are GMT
Goto page Previous  1, 2
Page 2 of 2

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum