Problem 2 - System won't start again after CMOS battery failure
I have a fairly infrequently used system in our lounge that I use occasionally when I want to display stuff on the telvision. Yesterday it failed to boot when I tried to use it so, after a few failed retries I went and got my laptop. As per "Problem 1" *that* wouldn't connect to the WiFi. To get round the immediate problem I got a long patch cable and connected that way, so we could see the pictures I wanted to show on the TV! So, I'd like to get the dead CMOS system back up and running. I've replaced the CMOS battery and loaded "Optimised defaults" into the CMOS. That didn't initially work at all because it didn't enable the SATA ROM and the system, though a few years old, has just one disk drive which is a SATA disk. I've now enabled the SATA disk and the SATA ROM in the CMOS and it gets a bit further, GRUB comes up and offers me three or four kernel versions plus memtest. Memory checks OK. If I just let it go on to the default kernel it pops up the disk ID, thinks for a while, then pops up the disk ID again with "Starting system" below, and that's it. I tried the recovery mode and it looks as if it can't read the SATA disk drive, though obviously the CMOS can as otherwise it wouldn't boot into GRUB at all. It goes through a load of timeout messages and then there's a kernel panic. I was going to try and produce some detail by copying to my laptop but that's waiting for the solution to "Problem 1"! Is there any way to copy/save the boot messages produced by GRUB in recovery mode so I can Google for them and/or show them here so that someone might be able to decipher what's wrong? ... and/or can anyone suggest something deep in the CMOS settings that might have been lost and is necessary for Linux to boot from SATA successfully? It's running Xubuntu 10.10 (I think!). -- Chris Green
On 30 October 2011 13:15, Chris Green <cl@isbd.net> wrote:
I tried the recovery mode and it looks as if it can't read the SATA disk drive, though obviously the CMOS can as otherwise it wouldn't boot into GRUB at all. It goes through a load of timeout messages and then there's a kernel panic. I was going to try and produce some detail by copying to my laptop but that's waiting for the solution to "Problem 1"!
You might have originally had the disk controller in PATA compatibility mode (or words to that effect) rather than native AHCI. This might not be the problem, but it can certainly send Windows off into the long, blue, grass. Can you boot off a live CD? Good luck, Tim.
On Sun, Oct 30, 2011 at 05:37:27PM +0000, Tim Green wrote:
On 30 October 2011 13:15, Chris Green <cl@isbd.net> wrote:
I tried the recovery mode and it looks as if it can't read the SATA disk drive, though obviously the CMOS can as otherwise it wouldn't boot into GRUB at all. It goes through a load of timeout messages and then there's a kernel panic. I was going to try and produce some detail by copying to my laptop but that's waiting for the solution to "Problem 1"!
You might have originally had the disk controller in PATA compatibility mode (or words to that effect) rather than native AHCI. This might not be the problem, but it can certainly send Windows off into the long, blue, grass.
Can you boot off a live CD?
I can certainly try and see what happens. -- Chris Green
On Sun, Oct 30, 2011 at 08:33:26PM +0000, Chris Green wrote:
On Sun, Oct 30, 2011 at 05:37:27PM +0000, Tim Green wrote:
On 30 October 2011 13:15, Chris Green <cl@isbd.net> wrote:
I tried the recovery mode and it looks as if it can't read the SATA disk drive, though obviously the CMOS can as otherwise it wouldn't boot into GRUB at all. It goes through a load of timeout messages and then there's a kernel panic. I was going to try and produce some detail by copying to my laptop but that's waiting for the solution to "Problem 1"!
You might have originally had the disk controller in PATA compatibility mode (or words to that effect) rather than native AHCI. This might not be the problem, but it can certainly send Windows off into the long, blue, grass.
Can you boot off a live CD?
I can certainly try and see what happens.
Booted up OK. I then went back and tried the older kernel versions in the grub menu and the 2.6.32 kernel one worked OK (the latest, non-working, one being 2.6.35). So I've copied /home off the system and I'm going to install a new xubuntu from scratch, there's virtually no customisation on the system so that shouldn't cause me any problems at all. (famous last words!) -- Chris Green
On Sun, Oct 30, 2011 at 09:35:23PM +0000, Chris Green wrote:
On Sun, Oct 30, 2011 at 08:33:26PM +0000, Chris Green wrote:
On Sun, Oct 30, 2011 at 05:37:27PM +0000, Tim Green wrote:
On 30 October 2011 13:15, Chris Green <cl@isbd.net> wrote:
I tried the recovery mode and it looks as if it can't read the SATA disk drive, though obviously the CMOS can as otherwise it wouldn't boot into GRUB at all. It goes through a load of timeout messages and then there's a kernel panic. I was going to try and produce some detail by copying to my laptop but that's waiting for the solution to "Problem 1"!
You might have originally had the disk controller in PATA compatibility mode (or words to that effect) rather than native AHCI. This might not be the problem, but it can certainly send Windows off into the long, blue, grass.
Can you boot off a live CD?
I can certainly try and see what happens.
Booted up OK. I then went back and tried the older kernel versions in the grub menu and the 2.6.32 kernel one worked OK (the latest, non-working, one being 2.6.35).
So I've copied /home off the system and I'm going to install a new xubuntu from scratch, there's virtually no customisation on the system so that shouldn't cause me any problems at all. (famous last words!)
... and it was (famous last words) too! :-( The installation starts up OK, recognises the hardware, including the hard disk but then gives an error as soon as it tries to wite to the hard disk, it's something like:- IO error when reading/writing hard disk. It seems as if kernels up to 2.6.32 work with the SATA controller but ones from 2.6.35 onwards don't. It's a VIA VT2620 controller, I've Googled a bit but have found nothing very significant. -- Chris Green
Well, I'm even more convinced this is a problem that has something to do with later kernels. I've tried both 32-bit and 64-bit xubuntu 11.10 installations and they both do exactly the same. Installation starts fine, hardware detected (including 500Gb SATA drive) but as soon as the installation attempts to write to the disk I get "input/output error during write on /dev/sda" So I tried an xubuntu 10.04 installation and that works OK with no problems. As I said originally (before I tried re-installing) the old 10.04 installation booted OK if I selected an older (2.6.32) kernel but failed to boot with the latest (2.6.35) kernel. This suggests to me that something in the kernel has changed and has broken the drivers for the VIA VT2620 controller. However I can't find anything that seems directly relevent by Googling. If anyone has any bright ideas or suggestions they would be very welcome. -- Chris Green
On 31/10/11 10:47, Chris Green wrote:
So I tried an xubuntu 10.04 installation and that works OK with no problems.
As I said originally (before I tried re-installing) the old 10.04 installation booted OK if I selected an older (2.6.32) kernel but failed to boot with the latest (2.6.35) kernel.
This suggests to me that something in the kernel has changed and has broken the drivers for the VIA VT2620 controller. However I can't find anything that seems directly relevent by Googling.
If anyone has any bright ideas or suggestions they would be very welcome.
As I think someone else suggested...Look in the BIOS and flick the SATA mode between legacy and AHCI mode...in legacy mode pretty much anything should be able to work with the controller but performance will be sub-optimal. My betting is that given this worked before your CMOS was reset, this setting has defaulted into whatever mode it wasn't in before. It may be that later kernels expect that controller to be in AHCI mode. NOTE: Some BIOS/chipset combinations need a cold hardware reset to go from legacy/AHCI so don't rely on the warm reboot when the BIOS saves and resets...save BIOS settings and powercycle the machine. Finally ISTR you talking about loading optimised defaults which will include aggressive bus and memory timings...what happens if you load the "failsafe" defaults and then go back in and re-enable the SATA chipset ?
On Mon, Oct 31, 2011 at 11:45:59PM +0000, Wayne Stallwood wrote:
On 31/10/11 10:47, Chris Green wrote:
So I tried an xubuntu 10.04 installation and that works OK with no problems.
As I said originally (before I tried re-installing) the old 10.04 installation booted OK if I selected an older (2.6.32) kernel but failed to boot with the latest (2.6.35) kernel.
This suggests to me that something in the kernel has changed and has broken the drivers for the VIA VT2620 controller. However I can't find anything that seems directly relevent by Googling.
If anyone has any bright ideas or suggestions they would be very welcome.
As I think someone else suggested...Look in the BIOS and flick the SATA mode between legacy and AHCI mode...in legacy mode pretty much anything should be able to work with the controller but performance will be sub-optimal.
There's no such thing in the BIOS on the system in question. It's *very* early SATA, the system dates from around 2004 I think. The only SATA options in the BIOS are to turn the SATA on/off and to enable the SATA ROM. If you enable the SATA ROM then you get RAID things to set up but that's all (and it doesn't work anyway, i.e. all you can do is view it).
My betting is that given this worked before your CMOS was reset, this setting has defaulted into whatever mode it wasn't in before. It may be that later kernels expect that controller to be in AHCI mode.
I *think* the CMOS failure is a red herring, it still worked after replacing the CMOS battery with kernel 2.6.32 or older, it's just the later kernels that don't work.
NOTE: Some BIOS/chipset combinations need a cold hardware reset to go from legacy/AHCI so don't rely on the warm reboot when the BIOS saves and resets...save BIOS settings and powercycle the machine.
Finally ISTR you talking about loading optimised defaults which will include aggressive bus and memory timings...what happens if you load the "failsafe" defaults and then go back in and re-enable the SATA chipset ?
It's certainly worth trying the failsafe settings, I'll go and try it. -- Chris Green
On 01/11/11 10:20, Chris Green wrote:
I*think* the CMOS failure is a red herring, it still worked after replacing the CMOS battery with kernel 2.6.32 or older, it's just the later kernels that don't work.
But presumably it was able to run the later kernel before the CMOS battery went flat as it had a later kernel installed ? AFAIK xubuntu doesn't change kernel versions except when there is a new release..you only get security updates to the release kernel. So unless you did a dist upgrade and never rebooted and then let the CMOS battery go flat I'd suggest that the CMOS failure *isn't* a red herring. Or did I lose track of what was happening somewhere ?
On Tue, Nov 01, 2011 at 07:41:56PM +0000, Wayne Stallwood wrote:
On 01/11/11 10:20, Chris Green wrote:
I*think* the CMOS failure is a red herring, it still worked after replacing the CMOS battery with kernel 2.6.32 or older, it's just the later kernels that don't work.
But presumably it was able to run the later kernel before the CMOS battery went flat as it had a later kernel installed ?
No, I don't think so.
AFAIK xubuntu doesn't change kernel versions except when there is a new release..you only get security updates to the release kernel. So unless you did a dist upgrade and never rebooted and then let the CMOS battery go flat I'd suggest that the CMOS failure *isn't* a red herring.
Or did I lose track of what was happening somewhere ?
Well as far as I can see 10.04 certainly went from 2.6.32 to 2.6.35 and I'm pretty sure it had some older versions than that too. -- Chris Green
On 01/11/11 20:43, Chris Green wrote:
Well as far as I can see 10.04 certainly went from 2.6.32 to 2.6.35 and I'm pretty sure it had some older versions than that too.
Does it not track the Ubuntu kernels then ? Ubuntu released 10.04 on 2.6.32 but afaik didn't have 2.6.35 until 10.10, in fact they even went to the effort of backporting DRM from 2.6.33 to 2.6.32 rather than change the kernel version. Again AFAIK Ubuntu never change the kernel version mid release.
On Tue, Nov 01, 2011 at 09:21:25PM +0000, Wayne Stallwood wrote:
On 01/11/11 20:43, Chris Green wrote:
Well as far as I can see 10.04 certainly went from 2.6.32 to 2.6.35 and I'm pretty sure it had some older versions than that too.
Does it not track the Ubuntu kernels then ?
Ubuntu released 10.04 on 2.6.32 but afaik didn't have 2.6.35 until 10.10, in fact they even went to the effort of backporting DRM from 2.6.33 to 2.6.32 rather than change the kernel version.
Again AFAIK Ubuntu never change the kernel version mid release.
Well a quick Google turns up:- http://www.ubuntugeek.com/kernel-2-6-35-officially-available-for-ubuntu-10-0... Although I must admit most pages seem to say that each Ubuntu version sticks with one kernel major version. I'm confused then, why do I see a list of different kernel versions available when I boot, even though I haven't upgraded to the next Ubuntu version? -- Chris Green
On Tue, Nov 01, 2011 at 10:03:18PM +0000, Chris Green wrote:
On Tue, Nov 01, 2011 at 09:21:25PM +0000, Wayne Stallwood wrote:
On 01/11/11 20:43, Chris Green wrote:
Well as far as I can see 10.04 certainly went from 2.6.32 to 2.6.35 and I'm pretty sure it had some older versions than that too.
Does it not track the Ubuntu kernels then ?
Ubuntu released 10.04 on 2.6.32 but afaik didn't have 2.6.35 until 10.10, in fact they even went to the effort of backporting DRM from 2.6.33 to 2.6.32 rather than change the kernel version.
Again AFAIK Ubuntu never change the kernel version mid release.
Well a quick Google turns up:-
http://www.ubuntugeek.com/kernel-2-6-35-officially-available-for-ubuntu-10-0...
Although I must admit most pages seem to say that each Ubuntu version sticks with one kernel major version. I'm confused then, why do I see a list of different kernel versions available when I boot, even though I haven't upgraded to the next Ubuntu version?
On my current system synaptic offers me both 2.6.32 and 2.6.35 linux src files (this is xubuntu 10.10). -- Chris Green
On Tue, Nov 01, 2011 at 10:09:18PM +0000, Chris Green wrote:
On my current system synaptic offers me both 2.6.32 and 2.6.35 linux src files (this is xubuntu 10.10).
I reckon that you have either the backports repo enabled or you have a kernel ppa configured as I think a backport of the Maverick kernel was made available for 10.04 but you had to explicitly install it. Adam -- New signature required, suggestions on a post card please.
participants (4)
-
Adam Bower -
Chris Green -
Tim Green -
Wayne Stallwood