[2.5.51] Failure to mount ext3 root when ext2 compiled in

[2.5.51] Failure to mount ext3 root when ext2 compiled in

[Rusty Russell]

Just noticed this (usually ext2 is compiled as a module, but was
testing a patch with CONFIG_MODULES=n).  Reverted to plain 2.5.51, and
it's still there:

	VFS: Cannot open root device "301" or 03:01
	Please append a correct "root=" boot option
	Kernel panic: VFS: Unable to mount root fs on 03:01

Now, I have an ext3 root, but when CONFIG_EXT3_FS=y and
CONFIG_EXT2_FS=y, I get this failure.  Turning off CONFIG_EXT2_FS
"fixes" it.

Hope this helps,
Rusty.
--
  Anyone who quotes me in their sig is an idiot. -- Rusty Russell.

Re: [2.5.51] Failure to mount ext3 root when ext2 compiled in

[Andrew Morton]

Rusty Russell wrote:
> 
> Just noticed this (usually ext2 is compiled as a module, but was
> testing a patch with CONFIG_MODULES=n).  Reverted to plain 2.5.51, and
> it's still there:
> 
>         VFS: Cannot open root device "301" or 03:01
>         Please append a correct "root=" boot option
>         Kernel panic: VFS: Unable to mount root fs on 03:01
> 
> Now, I have an ext3 root, but when CONFIG_EXT3_FS=y and
> CONFIG_EXT2_FS=y, I get this failure.  Turning off CONFIG_EXT2_FS
> "fixes" it.
> 

In the past year I've booted about 1,000,000,000 kernels with
CONFIG_EXT2_FS=y and CONFIG_EXT3_FS=y.  Across that period, 
maybe five or ten times I have seen this problem.

As soon as I get down to debug it it goes away.  I once traced it
as far as seeing ext3_fill_super() return failure, then I lost it.

Rebuilding the kernel, even if you "didn't change anything" makes
it go away.

I assume that in your case a `make clean' will not fix it.   You
lucky duck.   Can you stick a printk right at the end of
ext3_fill_super()?

Re: [2.5.51] Failure to mount ext3 root when ext2 compiled in

[Wayne Willson]

I have seen the exact same problem.  Turning off CONFIG_EXT2_FS fixes it for
me as well.



----- Original Message -----
From: "Rusty Russell" <rusty@rustcorp.com.au>
To: <akpm@zip.com.au>; <viro@math.psu.edu>
Cc: <linux-kernel@vger.kernel.org>
Sent: Thursday, December 12, 2002 9:48 PM
Subject: [2.5.51] Failure to mount ext3 root when ext2 compiled in


> Just noticed this (usually ext2 is compiled as a module, but was
> testing a patch with CONFIG_MODULES=n).  Reverted to plain 2.5.51, and
> it's still there:
>
> VFS: Cannot open root device "301" or 03:01
> Please append a correct "root=" boot option
> Kernel panic: VFS: Unable to mount root fs on 03:01
>
> Now, I have an ext3 root, but when CONFIG_EXT3_FS=y and
> CONFIG_EXT2_FS=y, I get this failure.  Turning off CONFIG_EXT2_FS
> "fixes" it.
>
> Hope this helps,
> Rusty.
> --
>   Anyone who quotes me in their sig is an idiot. -- Rusty Russell.
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: [2.5.51] Failure to mount ext3 root when ext2 compiled in

[Mohamed El Ayouty]

Hi,

This sounds more like the bug I have opened:

http://bugme.osdl.org/show_bug.cgi?id=110

where if CONFIG_DEVFS_FS = Y and CONFIG_DEVFS_MOUNT = Y, you will get:

VFS: Cannot open root device "hda2" or 03:02
Please append a correct "root=" boot option
Kernel panic: VFS: Unable to mount root fs on 03:02

I worked around it by enabling CONFIG_UNIX98_PTYS = Y under the
character devices.

But, a recent update to the bug shows that a patch was posted but nobody
cared.

Personally, I think the patch should be merged.

Mohamed

Re: [2.5.51] Failure to mount ext3 root when ext2 compiled in

[Andrew Morton]

Mohamed El Ayouty wrote:
> 
> Hi,
> 
> This sounds more like the bug I have opened:
> 
> http://bugme.osdl.org/show_bug.cgi?id=110
> 
> where if CONFIG_DEVFS_FS = Y and CONFIG_DEVFS_MOUNT = Y, you will get:
> 
> VFS: Cannot open root device "hda2" or 03:02
> Please append a correct "root=" boot option
> Kernel panic: VFS: Unable to mount root fs on 03:02
> 
> I worked around it by enabling CONFIG_UNIX98_PTYS = Y under the
> character devices.
> 
> But, a recent update to the bug shows that a patch was posted but nobody
> cared.
> 
> Personally, I think the patch should be merged.

You mean this one?

	http://www.lkml.org/archive/2002/12/13/50/index.html

It appears to simply disable internal mounting of devfs.

In which kernel did this problem first appear?   There were
devfs changes in 2.5.51.

Re: [2.5.51] Failure to mount ext3 root when ext2 compiled in

[Mohamed El Ayouty]

I started having that problem since 2.5.48.

I just commented out those 4 lines in init/do_mounts.c:

#ifdef CONFIG_DEVFS_FS
        sys_mount("devfs", "/dev", "devfs", 0, NULL);
        do_devfs = 1;
#endif

and 2.5.51 now works.

Mohamed

Re: [2.5.51] Failure to mount ext3 root when ext2 compiled in

[Andrew Morton]

Mohamed El Ayouty wrote:
> 
> I started having that problem since 2.5.48.
> 
> I just commented out those 4 lines in init/do_mounts.c:
> 
> #ifdef CONFIG_DEVFS_FS
>         sys_mount("devfs", "/dev", "devfs", 0, NULL);
>         do_devfs = 1;
> #endif
> 
> and 2.5.51 now works.
> 

OK, the above code was added in 2.5.48.   Thanks.  I'll bug Mr Viro
about it when the power comes back on over on the East coast...

If you replace CONFIG_DEVFS_FS with CONFIG_DEVFS_MOUNT there, does
that fix it?

Re: [2.5.51] Failure to mount ext3 root when ext2 compiled in

[Mohamed El Ayouty]

I tried your suggestion to replace CONFIG_DEVFS_FS with 

CONFIG_DEVFS_MOUNT in the code snippet:

#ifdef CONFIG_DEVFS_FS
        sys_mount("devfs", "/dev", "devfs", 0, NULL);
        do_devfs = 1;
#endif

But it gave me the same panic on boot.

Mohamed

Re: [2.5.51] Failure to mount ext3 root when ext2 compiled in

[Rusty Russell]

In message <3DF95D90.DEE68C66@digeo.com> you write:
> Rebuilding the kernel, even if you "didn't change anything" makes
> it go away.
> 
> I assume that in your case a `make clean' will not fix it.   You
> lucky duck.   Can you stick a printk right at the end of
> ext3_fill_super()?

You have cursed me.  Now it works.  Looks like a build problem?

Rusty.
--
  Anyone who quotes me in their sig is an idiot. -- Rusty Russell.