Quote:

There may not be that many bad sectors -- again, only the S.M.A.R.T. logs will know for sure. Linux tends to overreport bad sectors, due to incomplete low-level error handling in (all) drivers. If there's one bad sector in a single sequential I/O for, say, 1000 sectors, Linux will usually report read-errors for the bad-sector and all which followed it in the same I/O request, not realizing the err in its ways..

OK, showing my age (as someone who grew up with Good Old UNIX - as opposed to all this new-fangled Linux stuff ) - does the Linux driver/kernel do buffering even when you do a basic dd with BS=512? Is there a linux equivalent of a raw driver?