Urgent help needed! Can't boot into X [RESOLVED]

[Havin_it]

<panic panic panic>

 

General notes:

Laptop: Toshiba Satellite A30, dual-boot WinXP and Mandy 10.0 OE (FTP upgrade from 10.0 CE)

Kernel: 2.6.3-16mdk

bootloader: GRUB

 

So, I activated ACPI suspend fuunctions in KLaptop with the goal of being able to do a hibernate from my power button. First, tried doing it from the KLaptop menu. Short scroll of text, power off, OK.

 

Boot back in from GRUB, first part looks like normal boot, then it cuts to a different readout. Something about pages and waiting for DMAs to settle down. Then cuts out and goes back to the pre-BIOS manufacturer screen. Repeat. Same thing. Repeat. Just get the word GRUB and a blinking cursor.

 

Boot into old kernel 2.6.3-15, get through all the normal initialisation checks with a few things amiss - localhost can't be mounted, some new output about Vtund (too quick to read), then goes to console login. Tried typing 'kde' but just get a bunch of output most of which disappears off the screen before I can read it. Type 'exit' and logs out, shuts down but doesn't power-off.

 

How can I proceed? It seems like the stored hibernation info is causing the problem, but hey, I'm no expert. Can I fix this from the console? I have a FAT32 partition that's shared between Win and Lin if it's necessary to import other files/tools, and I still have the 10.0CE install CDs.

 

<please please help>

Edited November 30, 2004 by Havin_it

[pmpatrick]

It sounds pretty buggered. Unless someone has a better suggestion, you can try doing an upgrade install with a set of 10.0-OE disks if you can get them. Boot off install cd-1 and select "Upgrade" for the install option. Don't select any new packages; the upgrade install will reconfigure your hardware and might get it to the point where you can boot into X. I guess you could try doing the upgrade install off your 10.0-CE disks but I don't know if it will let you do that if you already have OE.

If you have data you want to save on there, you might want to consider getting a livecd like knoppix or slax and copying the data over to your FAT32 partition and doing a fresh install.

[Havin_it]

Oh God. Serious?

 

Isn't there anything extreme like emptying the swap partition or deleting a hibernation file (like Windoze hiberfil.sys) I could try?

 

At the end of the day, I could retrieve my data (once I remember where it all is) using command-line then do a clean install once 10.1OE isos come out, so I guess it's not the worst. But I'd welcome a chance to get back in there properly...

 

:Hurry Up With The ISOs!!:

Edited November 17, 2004 by Havin_it

[theYinYeti]

Were I you, I would boot an emergency bootable linux CD. From there, I would mount read-only all my disk partitions, and backup as much personnal data and system settings as possible, in case something goes wrong in the process.

Then I would unmount all the disk partitions and run a filesystem-check on each one. Note that in case of filesystem corruption, using fsck is intended to repair things, but it might break things too, hence the backup.

After all filesystems are checked and repaired if needed, I would use mkswap to reformat the swap partition(s).

 

Then I would boot as usual.

 

Yves.

[Youssou75]

Hi,

 

I had the same disaster as you two days ago when I tried to hibernate my laptop. I couldn't get Linux to boot correctely and ended up with a new fresh install. I was surprised though when reinstalled Linux since I couldn't recover my data (something must happen with the partitions, but I don't really know).

I think you should follow pmpatrick's advice and try to backup as much as you can and reinstall (30 min).

 

This is the warning you get when activating the advanced ACPI options

''The /usr/bin/klaptop_acpi_helper application does not seem to have the same size or checksum as when it was compiled we do NOT recommend you proceed with making it setuid-root without further investigation''

 

Since I am not an expert I would just recommend to avoid that in case.

 

Good luck

Yassine

[Havin_it]

Thanks guys. I did a little googling myself and it seems the problem has cropped up a few times. This thread has a lot of info (from a lot of anxious contributors) on the specific problem:

 

http://www.linuxquestions.org/questions/sh...p?postid=836764

 

My reading of the info there is that:

a) The root of the problem may well be not enough swap space. My swap partition is about 512MB (same as my amount of RAM), but in XP my hiberfil.sys is about 700-800MB.

 

B) I can probably get back into the 2.6.3-16 kernel boot by adding 'noresume' to the config or typing it at the GRUB prompt. I haven't tried this yet because of c)

 

c) It's mentioned that the filesystems will not have been unmounted properly, so I should do a filesystem check during booting. Can anyone tell me how to do that? It's done this automatically every 20 boots or so, but I'm not sure how to force it.

 

Another thought of my own: I wonder if it would be wise/very unwise to purge the swap partition from Windows? I believe I could do that using Partition Magic...

Edited November 21, 2004 by Havin_it

[Havin_it]

Further info: there is a bug for this in the Mandrake anthill:

 

http://bugs.mandrakelinux.com/query.php?bug=559

 

The info there talks about hacking suspend scripts etc. - bit out of my league.

 

They mention using 'mkswap' to purge the hibernation file, but I can't recall the mount-point of my swap partition. Is there a command I can use to identify it?

[aRTee]

If your issue is that the laptop tries to continue from hibernate, this has happened to me and the fix is easy - just don't know for grub, but with lilo, all that I needed to do is:

1- boot to lilo menu, hit esc, then boot linux with this command:

linux noresume

which booted to linux as a fresh boot, and swap was not mounted.

2- went into diskdrake, formatted my swap partition, mounted it.

 

After that all was fine.

[Havin_it]

I've done the equivalent for GRUB - removed the 'resume=/dev/hda6' item from the boot command.

 

Before doing this, I booted into the old kernel and did 'mkswap /dev/hda6'

(Don't know if this is the same as formatting...)

 

Result: can now boot new kernel, but only to command-line. Tried 'kde' but it complains about an undefined screen 'screen2' not found. Again, some of this scrolls off the screen before I can read it.

 

The errors in the boot process seem to be centred on the lo interface, vtund and CUPS. Is there anyway I can log the boot info so I can post it here?

[Havin_it]

Ne'mind, I have got myself back in. It seems there was an additional problem with XFree86 config, which I'd also been playing with before the shutdown.

 

I found a config for 'monitor2' in the config (I'd been trying to use the laptop's TV-out connection) which couldn't be found and was choking XFree.

 

So, one crash-course in vi later (god that was confusing. Who would use it?) I've got to the point where I can type 'kde' and get back in. Yet to see if it'll do so when I reboot, but hey, the days are ticking down 'til clean-installing 10.1OE anyhoo...

 

Just glad not to be marooned in Windoze in the meantime!

[Havin_it]

Just got 10.1OE fresh-intalled and I'm in NO HURRY to try hibernating again, thank you very much...! But oddly, before I did the install, the attempted-resume happened again! Also when I looked in grub.conf I found the

 

resume=/dev/hda6

 

item had been replaced! Don't know how the dickens that happens, quite annoying.

 

I'm wondering now: is there a documented way of calculating how much swap-space is required to avoid problems? I've seen one quote off-site somewhere that recommended 1.5 times your RAM size, therefore 768MB for me, which would be about the same size as my Windoze pagefile, but a thread here said it wasn't worth making the swap bigger than 512MB if you have that much RAM.

 

Can anyone clear that up? I know it's only one of many possible causes of the problem I've had, but if I've made my swap far too small then I guess I should rectify this anyway...

[Youssou75]

Can anyone clear that up? I know it's only one of many possible causes of the problem I've had, but if I've made my swap far too small then I guess I should rectify this anyway...

<{POST_SNAPBACK}>

 

I am not sure that the problem has something to do with the swap since I have 1.5 GB (512 MB RAM) and got the same stupid problem with hibernate. As for the swap I used to set it to twice the RAM (saw it somewhere). But this time I increased it because I am running calculations with my laptop that takes a lot of RAM.

[Havin_it]

latest: hibernate actually seems to work now, but not from the power button or lid (I've given up on those for the time being).

 

This might be because I added the noapic and nolapic options to the boot command, as suggested elsewhere.

 

When I use the hibernate option from Klaptop, there is a warning and something about USB, and 'doing a suspend instead'. Nevertheless, it writes to swap and FULLY powers-off. Power back on is fine too (after 2 attempts with a reboot between). GRUB starts loading, then it finds the image on swap and resumes from there.

 

Sidenote: don't trust the '...and lock' options in klaptop. On resume, the screen was visible long enough to read everythin in my console before the screensaver kicked-in!

 

It's interesting about the 'suspend instead' line. For a start, it was clearly still using swsusp after that point, and also, when I actually tried the 'suspend' option, that's when I DID get problems. I powered-on and just got a black screen - no Toshiba BIOS splash, no GRUB menu - and couldn't find any way to resume. I left enough time for a full reboot to occur, then tapped my power-button (which during normal operation does a normal shutdown, skipping the KDE prompt), but still nothing, so I was forced to do a hard reboot.

 

Conclusions: resolved, but I ain't about to start doing endurance-testing on it.