Programming

Wow, I'm glad that worked! Since I was completely out of ideas...

Updates!!

Unfortunately, even the SG_IO method of sending write command (technically they are SAT commands, though the core is ATA commands) could not be failed for a bad sector.

I have confirmed the bad sector presence with the help of MHDD, Winhex and HDDSCAN utilities, but the linux program is unable to identify them as bad and the write passes through without any issues.

I even tried the sg_dd command from sg3-utils package which does the same write method as what I have attempted and even that fails to identify the bad sectors. (with or without direct io enabled at /proc/scsi/sg/allow_dio).

I have tried sgio, dio, dsync, fsync and almost all the combinations of multiple blocks per transfer ranging from a single sector to 512 sectors. (1, 8, 16,32,64, 128, 256 and 512 sectors per transfer). None could identify the bad sectors.

I am completly out of options unless if somebody could direct me else where!!

Testing takes a LOT of time even for a smaller HDD (20 GB).

My sincere thanks to Corona688 and fpmurphy for all their valuble comments, critics and suggestions!

Why are you doing writes to a dying drive? Try reads.

Furthermore you must know about sector remapping.

Nothing left but to talk to the drive raw with ATA commands. You want raw I/O, do raw I/O. Not portable, what you're asking for just isn't.

My requirement is simple.

I want to zero fill a drive before I do "something" to it.

I want to know how many bad sectors on it as my decision of doing "something" is dependent on it.

That's all. I wanted to do in Linux and I am unable to do it with whatever

I could get it in DOS / WINDOWS, but not in Linux.

"Something" translates to destroying / reusing it.

Reading alone will not suffice as even read fails to identify bad sectors correctly.

I know about sector remapping and the triggering mechanism for a bad sector too.

Thanks.

Well that's new. Thanks for telling. According to earlier conversation you couldn't get it in windows without a custom device driver. which brings me back to my earlier suggestion of the linux kernel mailing list. If you haven't contacted them earlier, now would be a great time, as this thread's done a great job of rattling out what your requirements are.

I guess, we are debating on what I wrote on this thread rather. I had mentioned that I wanted to zero fill a drive in my first post.

Let's not go over into frenzy again!

I am able to do it on Windows or in DOS without any special device drivers. If I want to send commands to the hardware directly, I am able to send commands to the hardware directly. It works fine in Windows and in DOS (Well, I quote DOS for historical reasons as the initial erasure program was developed in DOS about 14 years ago!).

The last try I had with Linux was with the SG_IO commands (I believe that is the method of sending RAW ATA COMMANDS to the drive) and that did not work too. The command, say sg_dd is unable to identify the bad blocks / sectors on the hdd during the write process (on a drive which has failed with Windows, DOS and another Windows based disk erasure applications).

I guess, I am clear now, if not please let me know.

What APIs are you using to do this in Windows?