I have a turbolinux system with only one hard drive: /dev/hda

My partitions were like so:

/dev/hda1 for win 98
/dev/hda2 for linux
/dev/hda3 for swap

Well, while trying to create a second ext2 partition something went wrong and now my partition table seems scrambled. I cannot access it with fdisk or cfdisk.

fdisk says unable to read /dev/hda
cfdisk says something about bad partition table
sfdisk is way over my head

Now, the weird thing is that the system still works fine as far as LILO booting into win 98 or linux and both systems running ok.

But in linux I now cannot access /dev/fd0 or /dev/cdrom or /dev/windows (which has been working fine). I get Segmentation faults on all. Is this related?

I have tried gpart and rescuept and I cannot figure them out.

Is there anything else I can try?

I am assuming the system is going to fail hard sometime soon if I cannot fix the partition table.

Thanks

seems to me it could be your fstab file...
have you checked it.. /etc/fstab
post it here if you can.

/etc/fstab seems normal. Here it is:

/dev/hda2 / ext2 defaults 1 1
/dev/hda3 swap swap defaults 0 0
/dev/fd0 /mnt/floppy ext2 noauto 0 0
/dev/cdrom /mnt/cdrom iso9660 noauto,ro 0 0
proc /proc proc defaults 0 0
none /dev/pts devpts mode=0622 0 0

Actually, what is that last line? Is that normal?

Bret

Yes, it's normal. As short description is here. Not sure what else is going on though.

what shows up if you run "gpart -dvv -l (logfilename)"? and "gpart -cvv -l (logfilename)"?

Ok. I ran these commands and I have a logfile from each but they are pretty long. Should I post them?

The first line of each is:

Error: invalid extended ptbl found at sector (13076910)

and the second logfile (with the -c option) mentions the logical partition in hda4 is bigger than something (the line ends with $).

Now, I shouldn't have a logical partition or a /dev/hda4 for that matter.

That must be left over from me trying to make a new linux partition (where the whole problem started).

Should I try and get rid of it (hda4)?

Thanks again.

Alright.

I have to apologize.

The Segmentation fault error with mounts I mentioned in the first post is unrelated to the ptbl problem.

I forgot I had recently compiled a new kernel and have never tried to mount anything with it. Don't know why I have THAT problem, but that can wait until another day.

I went back to the original kernel and that problem is fixed.

But the partition problem remains.

Sorry about that.

Ok, forget gpart. Search for an app called "testdisk" its by a frenchman named "Grenier", IIRC.
Compile or if youve got Dos use the dos executable.
Make sure you know your CHS (cylinder, head, sector) data from either BIOS or a free app like Partition Magic's Partinfo, or dmesg.
Make sure you know approximately your partition layout.
(Unpack, compile) start testdisk and feed it the CHS values just to be sure. Turn to the harddisk and let it search for partitions.
Now for each partition you know is good, turn the [D] to [P] or [E]. Since your partition table is messed up, you wouldnt want to back it up, but if you do do "dd if=/dev/hdx of=/tmp/pt.hdx bs=1 count=64 skip=446 seek=446" where hdx is the harddisk.
Now let testdisk write the new partition table.

This works better than cfdisk which moans about 0 sized and invalid partitions but wont delete em.

HTH somehow.

Well, I tried all of the above with testdisk and testdisk can only see one extended partition (the one that should have been deleted in the first place).

And the situation did get worse.

LILO cannot see the partitions now, and yet I can still mount my linux part running tomsrtbt.

So my stuff is still there but nothing seems to be able to correct the partition table.

I may just start over with a new system.

Thanks to all.