HDD GURU FORUMS

The following was discovered by me when attempting to WRITE to an IBM native DeskStar AVV2 drive.

While doing some testing with PENDING SECTORS i discovered that the WRITE TEST on IBM (IBM-HITACHI) drives DOESN'T WORK AS IT SHOULD - DOESN'T WRITE ANYTHING AT ALL TO THE SECTOR !!!!

This affects IBM NATIVE drives, was tested with AVV2 but i'm pretty sure it will affect all other IBM based drives as well.

1 - Write something to LBA 0 with disk editor or (i've used HRT) :

2 - Check LBA 0 with Victoria :



3 - WRITE TEST with Victoria :





All DATA will be lost !!!! Yeah, right ..... Or so we think .....

Aparently the Victoria have written to the sectors including LBA 0 without any issue ....

Think again

4 - Re-Check LBA 0 :



DATA IS STILL THERE !!!

This of course, apply to ALL SECTORS OF THE DRIVE !!! BELIEVE ME, I'VE TESTED IT.

If i do the same on Samsung drives, for example, Victoria will work just fine.

Problem is with version 4.46b of Victoria for windows in PIO mode but most likely will affect all other versions.

Possible cause - way VICTORIA IMPLEMENTS the quick write.

All of my HRT tools that are IBM based have this option :

IBM_OLD
IBM_AVV2
IBM_AVV3
IBM_BOOK_NEW
IBM_Book_OLD



Quick Write is usefull to speed up write time but will only correct headers/ecc and it can be used when you add defects to P-List and you want to correct the headers for new translator so the sectors can be read/verify again, otherwise you will get error from the place where you have the translator shift till the end.

I think VICTORIA is doing just this on IBM drives, just erase the header and correct ecc but doesn't write data on the sector at all.

Let's write with HRT now.

REMOVE THE QUICK WRITE CHECKMARK and write to the drive :

Write forward till lba 4352



Check with Victoria :





HRT writes it's pattern of data to the LBAs without any problem. Do the WRITE TEST again with VICTORIA and again the DATA IS NOT ERASED !!!!

Now even worse ....

Try a LONG WRITE to the drive ...



And a SCAN !!!



Victoria for windows on my IBM drive when doing the LONG write the sectors got ..... fuck***

Now if i check PENDING SECTORS, the LONG write + SCAN makes the sectors accounted as PENDING because now EECC is badly written (because of VICTORIA).



If I READ the sectors that were ERASED/WRITTEN with victoria i get this :



And if i force DUMP of the Buffer :



Pattern that HRT WRITE test have written to the drive is STILL THERE meaning that VICTORIA WRITE TEST didn't wrote anything to the sector even with LONG WRITE and even worse damage the sector ECC.

Now we check S.M.A.R.T. :



2 Pending sectors - no bad sectors.

Let's fix the drive by writting to the sectors. As the pending sectors are ok and just have bad ecc because of buggy victoria for windows writting to them should remove them from the pending list, correct ?

- Write to a bunch of sectors :



- Check S.M.A.R.T. :



PENDING SECTORS STILL THERE !!!!

SO ... NOT EVEN CORRECT ECC WAS WRITTEN TO THE SECTORS AS THE SECTORS ARE STILL AWAITING TO BE WRITTEN AGAIN TO CHECK IF THEY ARE REALLY BAD OR NOT.

What if i do the write with something like HRT ?



And check S.M.A.R.T. ?



Pending sectors are removed because HRT writes correct data to the sector and ecc and header !!!


THIS PROBLEM MIGHT APPLY TO OTHER TOOLS, NOT JUST VICTORIA !!!!

Problem - You WRITE something to IBM-Hitachi drive to ERASE it with some tool like Victoria, you think that the data is gone and in fact .... IT'S STILL THERE !!!



Do not trust software/hardware to do the job whithout CHECKING FIRST.

Whatever you do, if you are ERASING OLD DATA check if the data is really gone .... or you might have a BIG PROBLEM LATER .....

NOT ONLY SECURE ERASE DOESN'T WORK AS EXOPECTED ON SOME DRIVES NOW THE WRITE TEST AND WRITE FUNCTION MIGHT NOT WORK AS IT SHOULD ON SOME TOOLS FOR SOME DRIVES.


SAME VICTORIA TOOL WILL WRITE/ERASE SECTORS WITHOUT PROBLEM ON OTHER BRAND OF DRIVES !!!

The actual culprit might be Windows after all.

I remember one article linked on here where it said that the reason why the great part of people use LINUX for HDD operations on hardware level is that you cannot access this lowest level with any programming language under Windows. It just won't let you do that, "for reasons of security" as the official saying goes.

You might want to try Victoria for DOS on that old IBM drive. I would love to know about your results!
From simple logic, what fails under V. for Windows MUST work under DOS.
That's because ISTR that V. for DOS is written in assembly language.
And in assembly language, you CAN get to the level I spoke of.
But even if you used a C++ program in Windows that contains some __asm__ bits, it would still deny you direct access to lots of things.
That's the reason why most BIOS flashing tools cannot work if Windows is running - to this day.

Victoria for windows have a direct I/O Driver and i was doing a PIO mode detection of the drive.
If Victoria were to send the correct ATA standard command to write to the sector/lba it should have worked without any issue, or so i think...

Also if i can format the drive under windows and put data there with windows there are no logic reason at all for VICTORIA not to write to the drive...

Spildit wrote:Also if i can format the drive under windows and put data there with windows there are no logic reason at all for VICTORIA not to write to the drive...

Are you sure this isn't just a limitation of the freeware version?
I might be wrong here but I seem to remember reading somewhere this version does not erase or write.

Could it be that Victoria stored your read data pattern and used that pattern during the write? Basically only one buffer used for both read/write?

dick wrote:

Spildit wrote:Also if i can format the drive under windows and put data there with windows there are no logic reason at all for VICTORIA not to write to the drive...

Are you sure this isn't just a limitation of the freeware version?
I might be wrong here but I seem to remember reading somewhere this version does not erase or write.

I don't think so as it aparently works on the other brand of drives that i have, like Samsung, it clears the sectors and they are all zero filled.

guru wrote:Could it be that Victoria stored your read data pattern and used that pattern during the write? Basically only one buffer used for both read/write?

I don't think so, because if that were to be the case the pending sectors would have been removed from the pending list and the sectors would be considered good. Also the long write damaged the sectors. So it's writting something to the ECC that causes the sectors to be bad....

I'm curious!
I will try the same next week and see what's happening

guru wrote:I'm curious!
I will try the same next week and see what's happening

I've figured this one out The HRT card itself is doing "something" when handling the IBM drives.

If i plug the drive DIRECTLY to the motherboard it wipes just ok.

It's the combination of VCTORIA + IBM + HRT HARDWARE CARD !



The hardware card itself "changes" the commands send by victoria intro something else and sends that something else to the drive...

If i use the HRT software itself it works as it should.

I would bet that it's related to that quick erase thing.

This is happening only with the combination of the hrt card + IBM drive + Victoria + Write.

If i put in any other drive that it's not IBM based Victoria writes just ok, even if the drive is plugged on the HRT card.

I would say that this Quick Write thing is implemented on the HRT card itself :



So when you plug an IBM drive to the HRT card the card firmware knows it's IBM and issuing write commands at software level will result in the HRT card sending "quick write" commands to the drive, that don't write at all except on the header of the sector....

Hi. I take back what I said earlier. The freeware version does allow writing to the drive!

Like you I also tested this with an AVV2 drive (74gb).

The only real difference in my setup was that I used an old PC3000 pci card.

Victoria would both write and erase sector 0 without any difficulty!

Therefore your conclusion there is some interference by the HRT card is probably correct.

dick wrote:Hi. I take back what I said earlier. The freeware version does allow writing to the drive!

Like you I also tested this with an AVV2 drive (74gb).

The only real difference in my setup was that I used an old PC3000 pci card.

Victoria would both write and erase sector 0 without any difficulty!

Therefore your conclusion there is some interference by the HRT card is probably correct.

I tested with the old PC-3000 PCI card as well and Victora works fine with that.

Good work Spildit Frend.

pikus_09 wrote:Good work Spildit Frend.

Yeah !