On Thu, 7 Aug 2003, Robert de Bath wrote:
>
> On Thu, 7 Aug 2003, Tom Oehser wrote:
>
> >
> > > AFAIK even 1.7.361 is able to mount any ext2/3 filesystem, however, it
> >
> > I think ext2 can only mount *clean* ext3 filesystems, if there is
> > journalling to apply, doesn't it set a bit that stops it? [...]
>> The 1.7.361 should be able to mount it if it is clean.
>
> Oh yes, "couldn't mount because of unsupported optional features";
> and you need tune2fs to remove that flag.
>
> Or debugfs:
I can confirm that 1.7.361 is able to mount an ext3 filesystem that is clean. I have one question on this - is it safe to mount it rw? I tried it (on a not-very-precious fs) and it appeared ok but mount said it was ext2. I suppose it is "at your own risk". Anyway, ro is all I need for the particular filesystem I am trying to recover. I didn't try the debugfs/tune2fs/e2fsck ideas. I was lucky that I was able to run a complete cycle of boot, fsck-st-start, shutdown on the dodgy disk.
On Thu, 7 Aug 2003, Tom Oehser wrote:
> > How about moving the dodgy disk to the machine that 2.0.103 boots in? >
Ah - the dodgy-disk machine is a thinkpad with little itty-bitty disk drives that would rattle around to no useful effect in my old PII workstation.
On Thu, 7 Aug 2003, Tom Oehser wrote:
> > One thing to try is pulling out spurious hardware, such as network > cards and sound cards and such, and see if it will boot then, in case > one of them is what is locking up on the probe. >
Don't think I can in a thinkpad (not safely anyway).
Going back to 2.0.103 tomsrtbt - I tried this again on the thinkpad with a new, clean, /dev/hda drive and (of course) it still hung after that
FDC 0 is a National Semi...
line. I am quite keen to get a 2.0.103 that works on the thinkpad. Do you suppose that's possible? Well, of course it's possible - I mean, can you offer any suggestions for debugging/fixing? >
> > The problem is there are a lot of auto-detects happening here, any one
> > of them could be upsetting some piece of hardware and most of them say
> > nothing if they find nothing.
>
As I mentioned in my original note:
> The old 1.7.361 gets past that point and (with a slight difference in that
> it prints only a single line for the hda) continues on with messages about:
> scsi: 0 hosts
> scsi: detected xxx
Can you confirm (or tell me how to confirm) that the kernel in 2.0.103 auto-detects devices in the same order as the one in 1.7.361? Which would then imply that the problem is definitely with the scsi detection. I am roughly thinking along the lines of -
. unpack.s
. replace the scsi piece of the kernel with the equivalent piece from the thinkpad's own kernel (I did say "roughly")
. buildit.s
John
