MHDD log slow (>100mS) sectors to file, makebad only LBA28

[icezone]

I have 4-5 harddisks of the 500GB - 1TB size and I scanned them all with MHDD.

I want to mark as UNC (bad) all sectors with scan time over 100mS to not stress further those HDD areas.

The problems:

1) MHDD will log to file only the very slow sectors (over 500mS I think, useless in this situation here).

2) MHDD will freeze if the sector wich I want to mark as bad is at LBA over 128GB.

For security reasons, I want to mark 20 sectors around that point as "bad" too, even if they are good now.

The solutions:

1) Maybe there it is an option in mhdd.cfg to lower the timeout for the green/red sectors? I tried to dissasemble mhdd.exe to modify the conditional jump for that timeout, to make it lower, but mhdd.exe is packed, so no luck

Also there it is no documentation for mhdd.cfg to enable full scan logging.

2) With hdparm we can write UNC sectors with LBA48, so the second problem may be solved, but we still need the list of slower than 100mS sectors...


What we can do?

[Vulcan]

I want to mark as UNC (bad) all sectors with scan time over 100mS to not stress further those HDD areas.

Depending on what you do after this, then I'm not sure how your plan will completely achieve that objective - but perhaps this is not clear due to the missing details...

Also, remember that the reported scan time in MHDD is not related to a sector.

Also there it is no documentation for mhdd.cfg to enable full scan logging.

I have not tried to do this, but if I was trying to do what you describe then I would investigate using the "MAKELOG" command. Then use your favourite method (perl, nawk etc.) to parse advlog.bin containing the scan times. I can forsee some other challenges with that technique, but it is something that I would investigate, based on your comments, if you don't find any simpler way.

[icezone]

Thank you for suggestions!

I found later (by reading this forum) that MHDD can no longer be developed, because its source code now belongs to Seagate or WD.

Anyway, instead of trying to analyze the binary logs of MHDD, I had a better idea: to make my own program (to reinvent the wheel).

What I have produced until now (I am an Assembly Language programmer):
- I can interogate BIOS for LBA support and present to the user a choice with each detected disk and its size.
- I can read or verify disk sectors (127 buc at once).
- I can create a file and write to it some data

What I need to do more to complete the program:

- analyze the timing of each read or verify to discover the near faulty disk sectors and log them to a file.

The scanning speed obtained was in excess of 100MB/sec (using the good old BIOS, not direct I/O)!