WinXP NTSF corrupted

[Hard Ball]

OK,

Last night, on my workstation, which I have on dual boot between Fedora C4_64 and Win XP Prof something akin to hell broke loose.
Here are the HW specs:

Antec True550
Asus K8N-DL
2 X Opteron 275
4.0 GB PC3200 ECC Reg.
GF 6800 NU
1 X 160GB SATA Hitachi
1 X 80GB SATA Seagate
1 X 20GB PATA Deskstar
...


The machine was partitioned as such:
80GB SATA = sda
160GB SATA = sdb
20GB PATA = hda

sda1 Win XP Pro C:/ boot NTFS ~40GB;
sda2 Win G:/ data FAT32 ~20GB;
sda3 Linux /home/ ~50GB;
sda4 Linux Swap /dev/shm ~6GB;

sdb1 Win D:/ apps/data NTFS ~30GB
sdb2 Win H:/ data(backup) FAT32 ~15GB
sdb3 Linux / (root) ~25GB

hda1 Linux /usr

And the boot loader in use is GRUB, with the config:

default=0
timeout=5
splashimage=(hd0,2)/boot/grub/splash.xpm.gz
hiddenmenu
title Fedora Core (2.6.11-1.1369_FC4smp)
root (hd0,2)
kernel /boot/vmlinuz-2.6.11-1.1369_FC4smp ro root=LABEL=/ ide=nodma rhgb quiet
initrd /boot/initrd-2.6.11-1.1369_FC4smp.img
title Fedora Core-up (2.6.11-1.1369_FC4)
root (hd0,2)
kernel /boot/vmlinuz-2.6.11-1.1369_FC4 ro root=LABEL=/ ide=nodma rhgb quiet
initrd /boot/initrd-2.6.11-1.1369_FC4.img
title Windows XP Professional
map (hd0) (hd1)
map (hd1) (hd0)
rootnoverify (hd0,0)
chainloader +1
makeactive
title Other
rootnoverify (hd2,0)
chainloader +1


The machine spends most of its time booted into Linux; and I havn't really used Win XP for a couple of week. It's been in this configuration since last summer, and I have even upgraded the processor and changed a number of other HW components on this machine without any problem. And there was no cause for concern for me either, since the Windows is seldomly used on my machine, and well protected by AV. The only possibility was that there was a couple of short power outages earlier in the day, yesterday; but the machine was booted into Linux at the time anyway.

But last night, when I booted into WinXP, certain registry corruption messages begin to pop up, such as mmc.exe not found. I tried to look at systems log, and nothing relevant was found; I tried to run chkdsk , which would not run because chkdsk.exe was supposedly corrupted. Then I thought a reboot may help with the matter. So the first time I rebooted, the machine hung a few seconds into the Win boot process (after selecting Win XP in GRUB); and forces me to shutdown after a few minutes, while HDD access LED remained on the whole time. Then I tried to reboot a second time, this time, I was able to get into my Win account, but as before, a number of messages about corruption in Win registries popped up, and eventually, the machine hung again in Windows, forcing me to reboot a third time.

This time, after Grub, Win does not boot up at all, but instead tells me that HAL.DLL is missing. I tried a couple of more times, and used the XP CD to do a recovery. During recovery, I tried to look at the partitions and their file systems, but it turns out, that the original D: partiion on sdb(80 GB SATA) has changed drive letter to C:, and the original C: partition on sda(160GB SATA) has become D:, and the file structure, instead of NTSF, has become "Not Recognized". I tried
fixboot D:
with no luck, and ran chkdsk, which showed that the partition was partially filled, but file system unrecognized.

Does anyone have any idea what happened?

And what would be the best way to fix it? I guess I can just reinstall XP , and then run Linux recovery to reinstall the boot loader; but is there any better way of fixing this?

I have to admit; I don't know that much about window file systems; most work I have done and systems that I have dealt with were Linux based; andy suggestions or pointers would be appreciated;

Thanks in advance
--HB

 

[Hard Ball]

com one; No one ??

 

[Patrick Wolf]

Kind of a similar thing happened to me last night. Except it was mostly my fauilt I think. My PC was defraging and running other programs, including memtest-which found 2 errors so I wanted to restart my PC and tweak the BIOS. It wouldn't restart so I hit the restart button on my case, it restarted, I changed some settings, saved, exited, windows loading...computer restarts, XP loading...restarts. Damn XP wouldn't even load. I tweaked BIOS settings again and again until it finally loaded windows. But unfortunely some corruption had occured and stuff wasn't working right. Now I'm all messed up.

I just installed XP on the same partition and am using it untill my new HD comes, than I'm going install XP onto it and transfer all my important files over and wipe this HD clean.

Damn I'm pissed. :|

Sorry OP, the only advice I can give is format and start from scratch.

 

[oldman420]

the weird errors you are pointing out seem to me to be hardware related often times when I oc too much I get the same errors you are getting except for the hal error.
each time I have seen that one I end up re-installing the OS.
I hope you are good enough to reinstall and still maintain grub I always break grub and end up reinstalling suse 10.
so you may want to boot into Linux and save any user files in the ntfs partition you can and burn em to CD.
it doesn't look great

 

[kylef]

The most likely culprit for the scenario you are describing is massive disk corruption. Whether this is caused by bad hardware (i.e., bad hard drive or faulty RAM) or software (something in Linux started writing to your Windows partition mistakenly) is anyone's guess at this point.


Did you ever mount your NTFS volume (sda1) in Linux? I have personally never trusted the NTFS file system drivers in Linux.

To eliminate the hard disk hardware as a culprit, you should run one of those low-level disk diagnostics tools from your hard drive manufacturer. They typically run off of a boot floppy and inspect the drive's internal logging mechanisms to report bad blocks/regions, using S.M.A.R.T. (self-monitoring analysis and reporting technology). The drive in question is obviously your 160GB Hitachi, which contains the corrupted NTFS boot partition.

I'd also run memtest86 for a day to rule out RAM errors. Faulty RAM can corrupt the filesystem cache, causing bad data to be written to disk. If your Windows partition was never mounted in Linux however and Linux was the primary boot OS, then this is less likely.

 

[kylef]

I forgot to mention another strong possibility: faulty SATA drivers. What kind of SATA controller do you have, and what version of drivers are you using for the controller?

 

[Hard Ball]

Thanks Guys,

These are all good suggestions, please keep them coming, if you can think of anything else;

I don't think it's the RAM, because I actually tested all of the memory ( 4 X 1GB ) sticks in Jan, after I upgrade the machine; and I actually sent back one stick of faulty RAM to Newegg, and the all of the RAM that I have now have been working fine.

I don't think it's OC related, since I am not really OCing at the moment at all. When I first rebuilt the machine, actually I did try a little OCing (this was a few months back), and it didn't seem to hurt anything in either OS, and both have been running fine until now. I finally decided to run everything stock for greater stability rather than a little added performance.

I ran chkdsk in Windows recovery mode, and the partition that was wiped out seemed to have no problem passing; but I should probably do that a couple of more times. Could there be a faulty sector? The suggestion of the SATA driver seems pretty promising. Although I'm still mystified as to how Linux is running stable, and what ever happened to Windows happened with very little premonition.

Edit: by the way, none of the windows partitions was ever mounted in Linux.

 

[kylef]

Hmm, if the partition was never mounted in Linux then we can rule out the Linux NTFS drivers.

But it would still be possible that a bad Linux SATA driver could have written to portions of the disk outside of its assigned partition. Although this is unlikely.

What about Windows? What did you do the last time you booted Windows? Have you changed your hardware at all? Swapped your drives around maybe?

When you installed Windows in this setup, how did you configure your SATA controller? Did you hit "F6" during Windows setup to specify an additional mass storage controller and insert a floppy disk to load a driver? If so, which one?

I would still run one of the low-level disk scan utilities from Hitachi. Looks like this program called Drive Fitness Test should do the trick.

 

[Hard Ball]

Originally posted by: kylef
Hmm, if the partition was never mounted in Linux then we can rule out the Linux NTFS drivers.

But it would still be possible that a bad Linux SATA driver could have written to portions of the disk outside of its assigned partition. Although this is unlikely.

What about Windows? What did you do the last time you booted Windows?

I don't remember doing anything out of the ordinary; it's possible that I did something that I don't remember, I believe I simply used the boot loader to select the OS, and that's it.

Have you changed your hardware at all? Swapped your drives around maybe?

Nothing like that; I definitely didn't change the connection between the drives and the mobo; even when I do do that, simply changing disk boot order in BIOS would allow GRUB to run, and load up any OS with no problem.

When you installed Windows in this setup, how did you configure your SATA controller? Did you hit "F6" during Windows setup to specify an additional mass storage controller and insert a floppy disk to load a driver? If so, which one?

I don't remember very well. I think that this last time I installed Windows, back in Oct or so (remeber, I don't use Win very often), the version of Windows simply contained the right SATA driver, so that I didn't need to use F6 and floppy to load it. I could be wrong though, since I don't recall that very well. I do remeber that when I got my first SATA drive in Spring of 04, I did need to load a driver from a floppy, but that was a different system.

I would still run one of the low-level disk scan utilities from Hitachi. Looks like this program called Drive Fitness Test should do the trick.

Good idea, you never know with such high failure rates of some of these HDDs; although Linux seems to be running fine (/home directory) on that drive.

 

[JohnBernstein]

First thing to do if you wish to coexist with Windows is to throw away Fedora Core and get a distribution that installs NTFS read only by default.

Fedora Core, Redhat, are deliberately crippled in many ways.

Throw it out.

Clearly, the NTFS linux drivers are not at fault because Fedora Core doesn't ship with any.

 

[kylef]

To my knowledge, Windows XP Sp2 doesn't have any inbox SATA drivers, so you must have either selected F6 and provided a driver disk, or you selected "disable nvRAID" in your BIOS, which (I think) makes your SATA drive appear to Windows like a regular IDE drive (in which case you don't need drivers).

Either way, I would double check that either the latest nVidia SATA drivers are installed, or the latest Asus BIOS has been flashed (I have heard reports that older Asus BIOSes for other boards had problems with the SATA-to-IDE compatibility mode).

But I'm just speculating at this point.

 

[JohnBernstein]

Second thing to do (I guess you wish to keep Fedora Core), is to download and compile the NTFS programs from:

http://www.linux-ntfs.org/content/view/19/37/

Or just grab the RPM. Either way is easy.

Then see what the program ntfsfix can do for you.

I have never used it, but you never know. Besides this process will give you Read and (limited) Write access to your NTFS partition.

 

[JohnBernstein]

Originally posted by: kylef
To my knowledge, Windows XP Sp2 doesn't have any inbox SATA drivers

That is correct.

Windows XP Sp2 doesn't have any inbox SATA drivers.

HOWEVER:

Recent Linux distros support SATA.

 

[kylef]

Originally posted by: JohnBernstein

Originally posted by: kylef
To my knowledge, Windows XP Sp2 doesn't have any inbox SATA drivers

That is correct.

Windows XP Sp2 doesn't have any inbox SATA drivers.

HOWEVER:

Recent Linux distros support SATA.

Right... I don't think anyone is disputing that, are they?

The reason I brought it up is that in order to get Windows to boot from a SATA drive, you'd either have to load the SATA driver during setup (using F6) or enable a BIOS feature which supports IDE compatibility mode. It might help to know which one of these was used here. (Which SATA driver, or which BIOS version).

If you're trying to get to the bottom of why your Windows partition got hosed, it doesn't really help to know that the Linux kernel now comes with builtin SATA support. Unless, of course, you think that is the culprit here.

 

[Hard Ball]

THanks, kylef and Johnberstein,

I'll look into all of these suggestions this weekend, when I have more time,

I will get back to you on how the system is doing.

--HB

 

[JohnBernstein]

If you're trying to get to the bottom of why your Windows partition got hosed, it doesn't really help to know that the Linux kernel now comes with builtin SATA support.

Errrr,... true. But sometimes a little extra info is useful.

 

[Comanche]

This happened to me on another system dual boot. Win xp pro and Win xp pro 64. I found that when I was loading windows, either one of them, the operating system was putting files onto the other drive. So that when the other drive was loaded, the info from the first load was erased. It was real pain figuring it out. So, I disconected drives, loaded one operating system, then uplugged those and replugged the other and loaded the other system. I have had no trouble since.

I don't know this is what happened with yours, but it might be worth a try.