Slow & bad sectors

[ipx]

I have been meaning to ask this for ages

First, what constitutes a slow sector.. does it mean, that the reading takes place, but the data read is not perfect (erroneous to some extent) and extra time is spent on ECC recovery and that is what shows up as a delay in sector's access time (i.e. mhdd)? Am I right to think, that when the sector takes about 500ms to read - the firmware is doing some massive ECC in the background and it was successfully recovered in the end (otherwise, it would have been an UNC)? Or is the sector reading slowness not obligatory related to ECC, but is actually influenced by something else (what?).. ? Of course, by that theory, just overwriting a 500ms sector should return it to a "normal" state with rapid response time (which I believe to be false). So where is the truth here?

Also, the "soft" and the "hard" bad sectors.. I am a bit confused with this designation.. Can anyone explain the difference?

I have noticed that, i.e., when this attribute is positive (non-zero):

#197 Current_Pending_Sector_Count 1

and at the same time this attrib. is zero:

#5 Reallocated_Sector_Count 0

And say, I will write 0x00 to the LBA referenced in #197 (which should presumably cause it to reallocate) - the reallocated sector count (#5) does NOT change! It stays at 0 (or doesn't increase)! So it's like I was able to write data into the "bad" sector, even though that should have caused it to be remapped? But most of the times, ofcourse, a different picture is seen - #5 is increased, instead. So what's the deal (my suspicion is "soft" vs "hard" bad blocks )?

10x

[Vulcan]

Based on my experience working with disk drive manufacturers on things like root cause analysis of drive problems, I'll just point out why IMHO you shouldn't be focussing on ECC. Of course other people might have the time to give longer answers than me

First, what constitutes a slow sector.. does it mean, that the reading takes place, but the data read is not perfect (erroneous to some extent) and extra time is spent on ECC recovery

There are many, many different techniques used by disk drives to try to read data - ECC (either on-the-fly, which is frequently happening, or firmware-assisted) is typically just the first of those recovery techniques, and those are exhausted relatively quickly. It's the other techniques which usually take more time. The details are proprietary and vary between different manufacturers & drive families but in short, yes, a slow sector usually does mean that extra error recovery attempts are being done.