SEO

I have an 8 GB harddisk I've been using for a very long time of running
Gentoo (1.4).
It has never presented a problem I couldn't fix, so far, but now I'm
stumbled.

I've been
compiling kernels for a while now, I'm not new to compiling, installing
and configuring GRUB, so I know what I'm doing most of the time.

Though this time it's different.

Roughly a week ago, I compiled the 2.6.7 kernel (straight from
kernel.org), without any problems.
Now, I have a new monitor, and wanted to reconfigure X11, and install
the latest NVidia driver taht was released a few days ago.
So, I configured X okay, and was installing the NVidia driver, when it
halted on an error, saying I needed to switch off 'rivafb' in my kernel,
thus having to recompile the kernel.

After the kernel compiled (and modules), I placed the files (bzImage and
System.map) in /boot, as I always do, and rebooted.

Now here is where it started happening.

Once booting into 2.6.7, it gave me an error saying (from the tops of my
head) "Kernel panic: VFS: Unable to mount root fs /dev/hda3".

Though, I know hda3 is my root
filesystem.

So I rebooted again, and luckily I still had my 2.6.6 kernel, so I
booted into that (where I am now).
The first thing I did was check my kernel settings, which were all okay
in my eyes.
To be safe, I deleted the .config file, and grabbed my backed up copy of
my 2.6.6 .config file. I did a 'make
mrproper', copied the backup .config file, set up the "new" features,
'make menuconfig' to check to be certain (and re-removed the 'rivafb',
hey, I'm not that stupid, heh), then compiled the kernel and modules,
and installed the lot.

Rebooted again, and, the same problem.

After doing the same for a few hundred time, I got so sick of it, that I
re-downloaded the 2.6.7 kernel completely, and set that one up manually.
But STILL the same damn error occurd.

This does NOT make sense, at all.

This is a part of my grub.config file:

---
title Linux; Gentoo 1.4 (Kernel 2.6.7)
root (hd0,0)
kernel (hd0,0)/boot/kernel-2.6.7 root=/dev/hda3
video=vesafb:ywrap,mtrr vga=0x317
initrd=(hd0,0)/boot/initrd-1024x768
title Linux; Gentoo 1.4 (Kernel 2.6.6)
root (hd0,0)
kernel (hd0,0)/boot/kernel-2.6.6 root=/dev/hda3
video=vesafb:ywrap,mtrr vga=0x317
initrd=(hd0,0)/boot/initrd-1024x768
---

Before going on; Yes, I did try and boot with no initrd and only "kernel
(hd0,0)/boot/kernel-2.6.7 root=/dev/hda3", in fact, I even tried typing
it out myself in the GRUB command-line.

Now there's another thing, which I think is the main issue here.
This is the result of "fdisk -l":

---
Disk /dev/hda: 8399 MB, 8399978496 bytes
16 heads, 63 sectors/track, 16276 cylinders
Units = cylinders of 1008 * 512 = 516096 bytes

Device Boot Start End Blocks Id System
/dev/hda1 * 1 63 31720+ 83 Linux
/dev/hda2 64 1056 500472 82 Linux swap
Partition 2 does not end on cylinder boundary.
/dev/hda3 1057 16276 7670880 83 Linux
---

Is it possible that the bad cylinder boundary on partition 2 is causing
this?
If so, how do I fix it?

Oh, both HDA1 and HDA3 are EXT3, and yes, I did turn the switch on in
the kernel as well.



Any help would be much appreciated.

Alessandro Gabriele Isidoro Lo-Presti enlightened us with:
> Once booting into 2.6.7, it gave me an error saying (from the tops of my
> head) "Kernel panic: VFS: Unable to mount root fs /dev/hda3".

Are you 100% sure you've got support for your IDE chipset? I've had
the same problem with another kernel (can't remember the version)
where they split up a generic IDE module into two more specific
modules, which I had turned off.

> To be safe, I deleted the .config file, and grabbed my backed up
> copy of my 2.6.6 .config file. I did a 'make mrproper', copied the
> backup .config file, set up the "new" features, 'make menuconfig' to
> check to be certain

Next time, do a "make oldconfig" instead of "make menuconfig" - you'll
only be asked the NEW questions. Saves a lot of time.

> Is it possible that the bad cylinder boundary on partition 2 is
> causing this?

I doubt it. If that were the case, why would 2.6.6 run properly?

Sybren
--
The problem with the world is stupidity. Not saying there should be a
capital punishment for stupidity, but why don't we just take the
safety labels off of everything and let the problem solve itself?

Hey,

Thanks for replying so quickly.

Yes, I'm 100% sure I have proper support for my IDE chipset.
I've used this kernel configuration since 2.6.5.
I could attach my .config if you'd like to have a glance at it.

Also, I forgot to write that I indeed also did "make oldconfig".
That's what I meant with 'copied the backup .config file, set up the
"new" features'.
Sorry if I wasn't being to clear about that.

And, yeah, I was wondering that myself, but it's teh only thing that
seems off.
I've checked to see if deleting it and re-adding the swap partition
would correct that error, and it did.
Though I have not yet written the partition table to the harddisk
because I'm not too convinced it won't touch the rest of the harddisk
(will it?).


Kind regards,
Alessandro Lo-Presti




Sybren Stuvel wrote:
> Alessandro Gabriele Isidoro Lo-Presti enlightened us with:
>
>>Once booting into 2.6.7, it gave me an error saying (from the tops of my
>>head) "Kernel panic: VFS: Unable to mount root fs /dev/hda3".
>
>
> Are you 100% sure you've got support for your IDE chipset? I've had
> the same problem with another kernel (can't remember the version)
> where they split up a generic IDE module into two more specific
> modules, which I had turned off.
>
>
>>To be safe, I deleted the .config file, and grabbed my backed up
>>copy of my 2.6.6 .config file. I did a 'make mrproper', copied the
>>backup .config file, set up the "new" features, 'make menuconfig' to
>>check to be certain
>
>
> Next time, do a "make oldconfig" instead of "make menuconfig" - you'll
> only be asked the NEW questions. Saves a lot of time.
>
>
>>Is it possible that the bad cylinder boundary on partition 2 is
>>causing this?
>
>
> I doubt it. If that were the case, why would 2.6.6 run properly?
>
> Sybren

Alessandro Gabriele Isidoro Lo-Presti enlightened us with:
> Yes, I'm 100% sure I have proper support for my IDE chipset. I've
> used this kernel configuration since 2.6.5.

My point was: kernel configuration options change. If you're only sure
because it worked in previous versions, you're *not* sure.

> I could attach my .config if you'd like to have a glance at it.

No thanks. No attachments in this group.

> Also, I forgot to write that I indeed also did "make oldconfig".
> That's what I meant with 'copied the backup .config file, set up the
> "new" features'. Sorry if I wasn't being to clear about that.

Always be explicit about what you do.

> And, yeah, I was wondering that myself, but it's teh only thing that
> seems off.

What are you referring to? You're top-posting, which makes it very
difficult to keep track of what you are replying to. Next time, remove
the unwanted quited text, and reply in-line, as I'm doing now.

> I've checked to see if deleting it and re-adding the swap partition
> would correct that error, and it did.

Ok, that's good.

> Though I have not yet written the partition table to the harddisk

Then how can you be certain it corrected the error? You must be
talking about a different error than the kernel panic, which is the
subject of this entire thread. Wat are you talking about?

> because I'm not too convinced it won't touch the rest of the harddisk
> (will it?).

Don't know. You haven't told us *exactly* what you did, so how can we
answer?

Please, next time, think about the way you post. Many people don't
think about this and put their reply at the top of their post, quoting
all of the original post below it. Now take a look at this:

---------------------------------------------------------------------
> Why is top posting so hard to read?
> > > What is one of the most annoying responses on Usenet?
> > When they top post.
In the English speaking world we read left to right, top to bottom.
---------------------------------------------------------------------

If people do it the other way around, things will clear up instantly:

---------------------------------------------------------------------
> > > What is one of the most annoying responses on Usenet?
> > When they top post.
> Why is top posting so hard to read?
In the English speaking world we read left to right, top to bottom.
---------------------------------------------------------------------

The best response type is "in-line" posting, that is, remove all the
quoted text that does not apply to your response, adding your comments
in context underneath as you go.

The distributed nature of Usenet is causing that not everyone sees all
posts in a particular thread in the right order or or even at all. If
you make sure you put your reply in a top-to-bottom order, people will
be able to easily read through it from top to bottom.

Be sure to remove all excess quoted text though. There is nothing more
annoying than having to scroll down three pages of quoted text you
already read, just to read a two-lined reply ;-)

A more complete guide to posting to Usenet can be found at
http://www.netmeister.org/news/learn2quote.html

The relevant section on good posting form is here
http://www.netmeister.org/news/learn2quote2.html#ss2.3



Sybren
--
The problem with the world is stupidity. Not saying there should be a
capital punishment for stupidity, but why don't we just take the
safety labels off of everything and let the problem solve itself?

Sybren Stuvel wrote:
> Alessandro Gabriele Isidoro Lo-Presti enlightened us with:
>
>>Yes, I'm 100% sure I have proper support for my IDE chipset. I've
>>used this kernel configuration since 2.6.5.
>
>
> My point was: kernel configuration options change. If you're only sure
> because it worked in previous versions, you're *not* sure.

First of all, I ran the 2.6.7 kernel fine, for about a week.
After I then re-compiled with the current and working 2.6.7
configuration, minus the 'rivafb' module, it gave me that error.

Otherwise, how can I be sure?
I have no idea what to do next because I have nothing to work with.
The error message itself is vague as it is, because hda3 is my root fs. :\

>
>>And, yeah, I was wondering that myself, but it's teh only thing that
>>seems off.
>
>
> What are you referring to? You're top-posting, which makes it very
> difficult to keep track of what you are replying to. Next time, remove
> the unwanted quited text, and reply in-line, as I'm doing now.

I'm referring to the fact that the only thing that seems off that could
possibly cause this (after a day of trouble-shooting and kernel
switches), is the disk boundary being off.
I don't know what else it could be.

>
>>I've checked to see if deleting it and re-adding the swap partition
>>would correct that error, and it did.
>
>
> Ok, that's good.
>
>
>>Though I have not yet written the partition table to the harddisk
>
>
> Then how can you be certain it corrected the error? You must be
> talking about a different error than the kernel panic, which is the
> subject of this entire thread. Wat are you talking about?

I'm talking about me trying to fix the disk boundary, because I thought
that was causing the kernel panic.
After I deleted the swap partition, and re-added it, I checked with 'p',
to print out the current partition table.
And the error message was gone after doing so.
But I didn't actually write to the disk yet.

>
>>because I'm not too convinced it won't touch the rest of the harddisk
>>(will it?).
>
>
> Don't know. You haven't told us *exactly* what you did, so how can we
> answer?
>
> Please, next time, think about the way you post. Many people don't
> think about this and put their reply at the top of their post, quoting
> all of the original post below it. Now take a look at this:
>
> {...}

Okay, okay, I understand, I'm sorry.
I'm doing it the way it's preferred, look. ^



So, what could it possibly be?
I've tried configuring the kernel from the grounds up, without importing
the old configuration file.
I've tried a LOT of kernel switches, and rebooted a few dozen times.
All to no avail.


Regards,
Alessandro Lo-Presti

Alessandro Gabriele Isidoro Lo-Presti enlightened us with:
> First of all, I ran the 2.6.7 kernel fine, for about a week.

Ok, you never said that. You told us you *compiled* the kernel just
fine - you never talked about using it just fine. How can we help you
if you don't give us the correct information?

> After I then re-compiled with the current and working 2.6.7
> configuration, minus the 'rivafb' module, it gave me that error.

Have you tried running a filesystem check?

> I have no idea what to do next because I have nothing to work with.
> The error message itself is vague as it is, because hda3 is my root
> fs. :\

Are you using devfs?

> I'm referring to the fact that the only thing that seems off that
> could possibly cause this (after a day of trouble-shooting and
> kernel switches), is the disk boundary being off. I don't know what
> else it could be.

Have you browsed through the kernel's changelog?

Sybren
--
The problem with the world is stupidity. Not saying there should be a
capital punishment for stupidity, but why don't we just take the
safety labels off of everything and let the problem solve itself?

Hey,

Thanks for your help, I appreciate it.
I found my answer on IRC.

It seems that I needed to switch on "CONFIG_MSDOS_PARTITION".
Which is weird, because in my 2.6.6 kernel, I did not set it myself, in
fact, I don't remember setting it at all.

Either way, it's fixed now.


Again; thanks for your help.



Regards,
Alessandro Lo-Presti


Sybren Stuvel wrote:
> {...}

Alessandro Gabriele Isidoro Lo-Presti wrote:

> Any help would be much appreciated.

i'm not so experienced with linux, anyway i suppose that every required
component for booting properly is compiled inside kernel, istn't it?

otherwise why don't you try to do mkinitrd by the hand... try to see what
does it contain... and of course - try not to use the same initrd for both
2.6.6 and 2.6.7, but create two distinct.

--
remove mailto__ on reply