i replaced 'read-only' in lilo with 'read-write'. and it worked. the system worked well.
but why it worked int pre1 and pre2,but not in rc1 and rc2, is there some modification to this in the kernel?
thanks.
>At 15:02 18-7-02, you wrote:
>
>> i met some problem with linux2.4.19. i devided my disk to three
>> partitions: hda1,hda2 and hda3. hda3 is the swap. both hda1 and hda2 are
>> ext3 file system. i boot the system with a initrd image which is saved on hda1.
>> when i use linux-2.4.19-pre1 and pre2, the system worked well. but when
>> i test linux-2.4.19-pre9,pre10,rc1 and rc2, the same problem happened to
>> all these verions. the system boot up, but the filesystem is readonly.the
>> following is what on the screen(not exactly):
>>..
>>mount root as readonly
>>INIT version is 2.78
>>swap on(swap priority -1)
>>..
>
>this looks like normal behaviour (output from dmesg):
><snip>
>VFS: Mounted root (reiserfs filesystem) readonly.
><snip>
>
>> is there some bugs in linux-2.4.19 with the filesystem? or with the
>> initrd boot? or i made some mistakes?
>
>the last 2 lines of your output are NOT kernel messages, when you see 'INIT
>version ...' it means that the kernel has booted and userspace bootup begins.
>
>your bootscripts should first check the root filesystem (and then it must
>be mounted readonly) and then remount it read/write
>
>another possiblity is to change your lilo.conf and remove the line which
>says 'read-only' or replace it with 'read-write'
>
>if you do not have lilo, then check the man-page of your bootloader on how
>to force the kernel to mount root rw
>
> Rudmer
On Thu, 18 Jul 2002, zhengchuanbo wrote:
>
> i replaced 'read-only' in lilo with 'read-write'. and it worked.
No! The file-system must be mounted read-only upon startup! There
are exceptions in embedded systems and special systems that build
file-systems (root file-system ram-disks) upon startup.
> ><snip>
> >VFS: Mounted root (reiserfs filesystem) readonly.
Correct.
The init scripts should check the file-systems (using fsck) and
then mount them read-write. If you (or init) executes fsck
on r/w mounted file-systems, you may (read will) destroy them.
Look in /etc/rc.d to see what happens upon startup. Something
like `fsck -A -V -a` gets executed. Then, after than happens,
something like `mount -n -o remount,rw /` gets executed. Then,
to update /etc/mtab, somewhere there will be a `mount -f /`.
Cheers,
Dick Johnson
Penguin : Linux version 2.4.18 on an i686 machine (797.90 BogoMips).
Windows-2000/Professional isn't.
On Thursday 18 July 2002 16:18, Richard B. Johnson wrote:
> On Thu, 18 Jul 2002, zhengchuanbo wrote:
> > i replaced 'read-only' in lilo with 'read-write'. and it worked.
>
> No! The file-system must be mounted read-only upon startup! There
> are exceptions in embedded systems and special systems that build
> file-systems (root file-system ram-disks) upon startup.
True.
in my previous mail i forgot to mention that it is not recommended to mount
root read-write on startup...
> The init scripts should check the file-systems (using fsck) and
> then mount them read-write. If you (or init) executes fsck
> on r/w mounted file-systems, you may (read will) destroy them.
> Look in /etc/rc.d to see what happens upon startup. Something
> like `fsck -A -V -a` gets executed. Then, after than happens,
> something like `mount -n -o remount,rw /` gets executed. Then,
> to update /etc/mtab, somewhere there will be a `mount -f /`.
you really should check if this gets executed in one of your bootscripts and
otherwise change them to do it this way.
Rudmer