Grub Error Collection [Part 8] [POST GRUB QUESTIONS HERE!]

Message

keenblade · Post by **keenblade** » Sun Jan 06, 2008 9:57 pm

NeddySeagoon wrote:keenblade,

I don't know the Pheonix BIOS. Another possibility ...
The hard drive addressing mode must be set to Logical Block Addressing (LBA) not to Large or ... I forget the 3rd option.

For accessing mode (I guess it is addressing mode) there is two option. One is large the other is auto. None of them works.

bhagany · Post by **bhagany** » Sun Jan 06, 2008 10:27 pm

Hi Neddy,

So, I went into my BIOS to look for the logical block addressing, and it was set correctly. However, I found a "SATA mode" setting that was set to RAID instead of IDE. This must have been set automatically when I originally was playing around with the hardware RAID controller (it's completely separate from the BIOS). Anyway, I reinstalled grub after I discovered this, and now it works. I appreciate all your patience and advice. My many thanks.

gentoo_fan · Post by **gentoo_fan** » Wed Jan 09, 2008 6:10 am

NeddySeagoon wrote:gentoo_fan,

Nope, this has no effect on windows loading.
Error 17 suggests you used root (hdx,y) rather than rootnoverify ((hdx,y) in your Windows boot Stanza.
Look in the grub.conf.example file.

I got rid of that error, and a new one popped up.

Code: Select all

root (hd0,5)
 Filesystem type is fat, partition type 0xb
makeactive 

Error 12: Invalid device requested

Its finding my windows partition with windows already installed, but I have no idea why it wont go any farther.

grub.conf:

Code: Select all

default 0
timeout 30

title=Gentoo Main
root (hd0,0)
kernel /boot/2.6.23-tuxonice-r6 root=/dev/hda3


title=Windows XP
rootnoverify (hd0,5)
makeactive
chainloader +1

fdisk output:

Code: Select all

  Device Boot      Start         End      Blocks   Id  System
/dev/hda1   *           1           5       40131   83  Linux
/dev/hda2               6          68      506047+  82  Linux swap / Solaris
/dev/hda3              69        2559    20008957+  83  Linux
/dev/hda4            2560        9729    57593025    5  Extended
/dev/hda5            2560        3805    10008463+  83  Linux
/dev/hda6            3806        5051    10008463+   b  W95 FAT32
/dev/hda7            5052        9729    37576003+  83  Linux

Post by **NeddySeagoon** » Wed Jan 09, 2008 8:25 pm

gentoo_fan,

I think the Windows C:\ partition must be a primary partition ... check that.
If that turns out to be true, then it must be on one partition 1..4

keenblade · Post by **keenblade** » Fri Jan 11, 2008 5:31 pm

I booted with amd64 Gentoo 2007.0 Minimal CD. Then chrooted, installed grub, setup grub and finally reboot. But no change. Any idea?
I also wonder why grub works within minimal cd but not within my hard drive.

psomas · Post by **psomas** » Fri Jan 11, 2008 5:44 pm

keenblade wrote:I booted with amd64 Gentoo 2007.0 Minimal CD. Then chrooted, installed grub, setup grub and finally reboot. But no change. Any idea?
I also wonder why grub works within minimal cd but not within my hard drive.

what do you mean that grub works within minimal cd?...

did you install it in the mbr of the drive you boot?...

keenblade · Post by **keenblade** » Fri Jan 11, 2008 5:54 pm

psomas wrote: what do you mean that grub works within minimal cd?...

did you install it in the mbr of the drive you boot?...

I mean I can boot the minimal cd.
I installed it in the mbr like this:

Code: Select all

grub> root (hd0,5)
grub> setup (hd0)

It is not a new install. I upgraded mb bios. Then grub stop working. It just gives a blank screen after I select any kernel from its menu.

depontius · Post by **depontius** » Fri Jan 11, 2008 6:12 pm

I have an A7N8Xe-Deluxe based dual-boot (Win98SE and Gentoo) system which was originally brought up, and has been in use for several years with a now-crowded IDE drive. I've installed a new SATA drive and installed both OSs on it, though I may need to reinstall Win98SE. I'm currently using grub installed on the IDE disk to boot Gentoo on the SATA or Win98SE on the IDE. This works, though by hiding the IDE C and D partitions, I haven't been able to boot Win98SE on the SATA, which is why I think I may need to reinstall it. I may also have done the installation wrong for adding SATA support to Win98SE, according to the recently-relocated motherboard manual.

I'd like to remove the IDE drive, dedicate it to another purpose, and run this machine with just the SATA.

To that end, I went into BIOS and turned off the primary IDE controller, which should make the IDE drive "go away." Then I booted a LiveCD and installed grub on the SATA drive, the "normal way."

This system won't boot. First the display is black, but when I press "Enter" I get an "Error 18" message, the classic 1024 cylinder problem. Then after the error message, I get back to the grub menu that didn't display at first. Only problem is that the video is horribly distorted. It looks like it's in text mode, but with a vertical bar about a pixel wide in each character cell on the screen. Trying to boot anything gets the same "Error 18" message.

I was beginning to despair, thinking that the SATA BIOS sucks, and that I just can't do this, until I remembered that grub on the IDE boots the SATA just fine, and somehow grub has found it menu. The grub menus on the SATA and IDE drives are different, and after the first boot failure, the grub menu (badly) displayed is clearly the one from the SATA drive. Furthermore, on the SATA drive the borked Win98SE install chews up over 3000 cylinders, and the menu is on the first Linux partition after that, yet grub is appearing to read it, even after the "Error 18" complaint. That is, unless grub also tucks a copy of the menu in the MBR.

I'd like to clear this up, and have several trial-and-error things I can try, but if there's experience here that can save me some time, I'd love it. Conversely, whatever false roads I travel, I'll try to help others avoid.

1: Though I've turned off the primary IDE controller in BIOS, it's possible that it's only mostly dead, and that grub can see it, and gets confused. To this end, I should try unplugging the IDE drive prior to booting, and see if it gets better.

BUT...

2: If grub can see the turned-off drive, then I'm sure the LiveCD saw it. In that case the grub installation may have been messed up. In this case, I should unplug the IDE drive, boot the LiveCD, reinstall grub in the MBR, and then try to boot off of the SATA.

3: I've seen things about a "boot-floppy / native" install for grub. Is there anything really different about this vs the normal root/setup from a LiveCD or chrooted into the SATA Gentoo install?

4: I've also seen something about telling grub to ignore drive information it gets from BIOS, and passing that information in manually. Problem is, I've only seen it in passing, and can't reproduce the information on demand. Nor can I remember it. But the fact that the IDE grub boots the SATA drive makes me think this shouldn't be the problem.

5: Apparently I can also choose SCSI boot on my motherboard. At the moment when trying to run SATA-only I set it to boot only CDROM or floppy, and even set the "other boot device" to off. It still tries to boot the SATA. I'm wondering if the SCSI boot path might be different from the behind-the-scenes SATA boot path.

I'll start more trial-and-error this weekend, but any insights that might save me time would be appreciated.

keenblade · Post by **keenblade** » Fri Jan 11, 2008 7:39 pm

I am getting somewhere. If I remove the vga=792 computer boots. But this time I have other problems:

Code: Select all

/etc/init.d/cpufreqd start
/var/lib/init.d/depcache: line 621: config: command not found
/var/lib/init.d/depcache: line 2811: config: command not found
 * cpufreqd requires the kernel to be configured with CONFIG_CPU_FREQ
 * Make sure that the appropiate drivers for your CPU are available.

and

Code: Select all

/etc/init.d/cpufrequtils start
 * Enabling ondemand cpufreq governor on CPU0 ...                                                                                             [ !! ]unknown or unhandled CPU?

but some info!!!

Code: Select all

zgrep CONFIG_CPU_FREQ /proc/config.gz
CONFIG_CPU_FREQ=y
CONFIG_CPU_FREQ_TABLE=m
# CONFIG_CPU_FREQ_DEBUG is not set
CONFIG_CPU_FREQ_STAT=m
# CONFIG_CPU_FREQ_STAT_DETAILS is not set
# CONFIG_CPU_FREQ_DEFAULT_GOV_PERFORMANCE is not set
CONFIG_CPU_FREQ_DEFAULT_GOV_USERSPACE=y
CONFIG_CPU_FREQ_GOV_PERFORMANCE=m
CONFIG_CPU_FREQ_GOV_POWERSAVE=m
CONFIG_CPU_FREQ_GOV_USERSPACE=y
CONFIG_CPU_FREQ_GOV_ONDEMAND=m
CONFIG_CPU_FREQ_GOV_CONSERVATIVE=m

Code: Select all

/etc/init.d/acpid start
 * Starting acpid ...
acpid: can't open /proc/acpi/event: Device or resource busy

Code: Select all

cat /proc/cpuinfo
processor       : 0
vendor_id       : AuthenticAMD
cpu family      : 15
model           : 47
model name      : AMD Athlon(tm) 64 Processor 3200+
stepping        : 0
cpu MHz         : 2010.341
cache size      : 512 KB
fpu             : yes
fpu_exception   : yes
cpuid level     : 1
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 syscall nx mmxext fxsr_opt lm 3dnowext 3dnow rep_good pni lahf_lm
bogomips        : 4021.80
TLB size        : 1024 4K pages
clflush size    : 64
cache_alignment : 64
address sizes   : 40 bits physical, 48 bits virtual
power management: ts fid vid ttp tm stc

depontius · Post by **depontius** » Fri Jan 11, 2008 8:17 pm

keenblade,

Now that your booting, I think your problems are minor.

First up, getting cpufreq stuff isn't just a matter of building governors, you've got to have the driver for your cpu/chipset. I think I saw somewhere upstream that this is an nForce4 board. If so, you need to be building CONFIG_X86_POWERNOW_K8 as well as the generic cpufreq and governor stuff. I have an nForce4 system, and am happily running "ondemand", but come to think of it, first boot after 2.6.23 I found that cpufreq didn't work, and I had to go back and set CONFIG_X86_POWERNOW_K8, myself. I'm not sure what changed, but something did. Trivially fixed with a kernel rebuild.

As for the acpid, I don't have this problem, but I seem to remember hearing something about it being a conflict between hald and acpid. They both want the events, hald gets there first, and won't share them. The suggestion I saw was to run acpid in the boot runlevel. Apparently it will share. I'm not running hald on my nForce4 machine, so I can't tell for that. Oddly enough, on the machine I'm using now I'm running both acpid and hald at the default runlevel, and having no problems. I just checked some old logs, and can see acpid start, then dsnmasq, then hald. I'm guessing that hald should be started after acpid, but there's no relationship coded into the initscript, so the sequence depends on what else you've got running that forces the issue. Looking again at the initscripts, hald says "need acpid" and acpid says "before hald", so I guess the problem I saw is old.

Is your system up to date? Check your acpid and hald in /etc/init.d and look for the need/before I mention.

keenblade · Post by **keenblade** » Fri Jan 11, 2008 8:37 pm

depontius, thanks. For acpid, situation exactly as you said. If acpid starts before hald, it is ok.
For cpufreq, the drivers and governor stuf is already right:

Code: Select all

zgrep CONFIG_X86_POWERNOW_K8 /proc/config.gz
CONFIG_X86_POWERNOW_K8=m
CONFIG_X86_POWERNOW_K8_ACPI=y

I can't understand why cpufreq does not work.

depontius · Post by **depontius** » Fri Jan 11, 2008 9:31 pm

You also need to put "powernow-k8" in "/etc/modules.autoload.d/kernel-2.6". It doesn't load automagically.

keenblade · Post by **keenblade** » Sat Jan 12, 2008 1:56 am

powernow-k8" is already in "/etc/modules.autoload.d/kernel-2.6, but it is not loading:

Code: Select all

modprobe powernow-k8
FATAL: Error inserting powernow_k8 (/lib/modules/2.6.23-gentoo-r5/kernel/arch/x86_64/kernel/cpufreq/powernow-k8.ko): No such device

I am doing a revdep-rebuild.

depontius · Post by **depontius** » Sat Jan 12, 2008 2:30 am

Still stuck...

I've unplugged the IDE drive, so the system has only the floppy, cdrom, and SATA hard drive.
I've told BIOS to boot SCSI.

No dice, same symptoms.

I've booted a Gentoo minimal cdrom, and chrooted into the Gentoo install on the SATA drive.

From there I ran the grub/root/setup commands.

No dice, same symptoms.

I realized that there was no device.map file, so I created an appropriate one, then reran grub/root/setup commands.

Just for jollies, while here I did "find /grub/stage1" and it correctly said, "(hd0,4)".

No dice, same symptoms.

The only thing left is a grub boot floppy, so I'll try that next. It's a good thing that this system actually has a floppy drive on it, none of the newer systems do. At this point, I'm rather pessimistic. What I don't understand is why grub on the IDE drive is able to boot the SATA drive, when I'm having such a tough time getting the SATA drive to boot itself.

Still accepting suggestions.

keenblade · Post by **keenblade** » Sat Jan 12, 2008 3:18 am

depontius, did you try to upgrade your mobo bios? (I saw v1013 for your mobo).
Maybe what gives me trouble gives you success

Also if there is splashimage statement, try removing it in the grub menu.

depontius · Post by **depontius** » Sat Jan 12, 2008 1:19 pm

I have the BIOS, but haven't applied it, yet. Looking at the ASUS "changelogs" the only BIOS updates since the one that came with the board have been recognition for Semperon processors and more kinds of memory. That doesn't mean that there might not be other changes, I'll admit. I'm just a little fearful of flashing BIOS, because it can be a good way to brick a board.

Last night I build a grub boot floppy based off of the grub install on the SATA disk, and got the exact same symptoms. Then I thought to check the grub install on the IDE drive against the one on the SATA install. The stage1.5 and stage2 files are different. I tried copying these over to the grub floppy, and the system tried to boot, before the kernel panicked. Because the text screen was garbled, I couldn't tell what had happened.

Thanks for the suggestion on commenting the splashimage line. I'd heard of the "text mode" suggestion, but with casual checking hadn't found the command to force it. I guess "no splashimage means text mode" sounds decent enough.

I've got a few more ideas, now.

I'm going to compare my grub files from several different machines. It's leaving me thinking that maybe some machine configuration is compiled into the grub stage files, and that's really my problem. The SATA install was done in a chroot from the IDE install. Maybe what I really need to do is unplug the IDE drive, boot from CDROM, chroot into the SATA, and re-emerge grub.

I'm also going to re-try the floppy boot experiment with the splashimage line commented out.

Thanks, I'll be back.

keenblade · Post by **keenblade** » Sat Jan 12, 2008 2:59 pm

With the new bios everthing goes bad for linux (pci interrupts, usb stop working, frozen grub vs.) Xp worked fine.
Btw, I tried to downgrade bios with asus update tool. It did not proceed because of the old bios has an older date! How nonsense disabling downgrade. Anyway I hex edited the bios changing the date of it. It worked and flashed the bios successfully. When I reboot it gave checksum error. Then I created a cd containing awdflash and the good old bios. This time it flashed from cmd line and now everyting is fine exept cpufreq. Gonna try building kernel again.

depontius · Post by **depontius** » Sat Jan 12, 2008 3:47 pm

I haven't tried my stuff yet, but I'll look hard before flashing 1013, even though we're talking different board/BIOS here.

Did you get any sort of messages when you tried to modprobe powernow-k8? Does dmesg tell you anything?

keenblade · Post by **keenblade** » Sat Jan 12, 2008 4:04 pm

depontius wrote: Did you get any sort of messages when you tried to modprobe powernow-k8? Does dmesg tell you anything?

here it is:

Code: Select all

modprobe powernow-k8
FATAL: Error inserting powernow_k8 (/lib/modules/2.6.23-gentoo-r5/kernel/arch/x86_64/kernel/cpufreq/powernow-k8.ko): No such device

If you are going to flash your mobo, make sure to create a cd and a floppy that includes your current bios and upgrade tool.
I later found many posts for my mobo about the new bios does not work for linux. I wish I would have searched before flashing.

depontius · Post by **depontius** » Sun Jan 13, 2008 11:46 pm

I appear to have had every possible mistake in my grub.conf. In particular, if you get the path to the splash image wrong, it doesn't just fail to put up a splash image, it badly corrupts the video. Incidentally, somewhere along the line, Gentoo began putting a ". ->boot" symlink in the /boot directory that increases the forgiveness of the system. That caught me, but in addition to correcting the path, I added the symlink myself.

At any rate, I've got most, if not all of the grub.conf errors corrected, I'm able to bring up a civil grub menu with the splash in graphical mode, and I can start to boot. In fact, I can even boot that started-but-never-finished Win98SE install. I just can't boot Linux.

This isn't verbatim, since I don't have a serial console set up to log from, but:

Code: Select all

mknod: /newroot/dev/console /newroot/dev/tty1 : no such file or directory
Booting (initramfs) ... switchroot: bad console /dev/console
kernel panic - not syncing:
attempted to kill init!

I also rebuild my kernel/modules/initramfs, same results.

I've really carefully checked the spelling of my initrd line in grub.conf, but I'll check it again. I searched and found some people having problems with genkernel, but I've been using it for years with no problems. But next, I'm going to connect the IDE drive again and reboot, then cross-check the IDE and SATA grub.conf files. In the forum entries I also found about ripping open the initramfs to examine what's inside, so I'll give that a try.

Curiously, searching against the terms (not quoted string) "mknod /newroot/dev/console no such file or directory" all of the hits were Gentoo or Sabayon. At least this is no longer a SATA or BIOS problem.

Still open to suggestions.

Post by **NeddySeagoon** » Sun Jan 13, 2008 11:56 pm

depontius,

Its not a grub issue either.

Grub loads the kernel, loads the initd and jumps to the kernel start address. The kernel is in charge from that point.
Your error comes after the kernel has mounted the initrd as its root and is running the initrc script.

Its something in (or missing from) your kernel or initrd but I don't know what.

depontius · Post by **depontius** » Mon Jan 14, 2008 2:06 am

Found it. Another typo. I was passing the wrong partition with real_root, giving it the boot partition instead of the root partition. I'm running, now.

The SATA drive was done from an existing IDE installation, with many /etc files copied over, and grub.conf was copied/tweaked, as well. But the IDE installation didn't have a separate /boot partition, it was just part of /root. I decided to have separate /boot and /root partitions on the SATA drive, so there were numerous mixed references laying around. This was just the last of the problems. The differences were subtle, and looked right, not syntactically wrong at all, which was why it has been such a trial and error spree.

Thanks to all.

(keenblade - Ever find out why powernow-k8 wasn't loading?)

keenblade · Post by **keenblade** » Mon Jan 14, 2008 3:09 pm

depontius wrote:Found it. Another typo. I was passing the wrong partition with real_root, giving it the boot partition instead of the root partition. I'm running, now...

Thanks to all.

(keenblade - Ever find out why powernow-k8 wasn't loading?)

Nice to see that you solved your problem.
For me, I haven't found a solution , yet. I have three different kernels that was booting fine before mobo flash. Rebuilding kernel is useless. I don't understand why powernow driver fails to detect my cpu, since the mobo has good old bios again.

gentoo_fan · Post by **gentoo_fan** » Tue Jan 15, 2008 7:59 am

NeddySeagoon wrote:gentoo_fan,

I think the Windows C:\ partition must be a primary partition ... check that.
If that turns out to be true, then it must be on one partition 1..4

'Tis true indeed. Oh windows, how you like to make my life difficult. Thanks again for all your help!

dpetka2001 · Post by **dpetka2001** » Mon Jan 21, 2008 9:01 pm

i switched to "ahci mode" and played a little with the boot sequence in BIOS and finally i think i managed to have to correct boot order...but when grub tries to load it says "grub loading stage1.5
grub loading
error 22"...so what should i do from here on?? any help please?? thanks in advance...

Grub Error Collection [Part 8] [POST GRUB QUESTIONS HERE!]

A7N8Xe-Deluxe having problems moving from IDE to SATA

keenblade

A7N8Xe-Deluxe having problems moving from IDE to SATA

completely new symptoms - kernel panic on boot