Feature Desc riptions
6- 8 Maxt or Atl as 10K III
Max tor implement s multiple LOG SEN SE pages wh ich logically group tog ether
parameters related to disk an d sub system activities. Monitoring this data enables users
to observe drive performa nce and possibly predict failures.
The mode pages control all error recovery. Any error recovery that is enabled is
attempted without host intervention.
S ee k er ror recov ery is automa ti c if th e hardware is ph ys i cally ab l e to position to the
correct track.
Data read errors are recov e red , if poss i ble, by one o r more of th e fo llowi n g me t hod s :
•Correct ion of th e data “on-the-fly” with ECC hardware.
•Re-try the operation.
•Relo ad read chann el regist ers to calibrated values.
•Off-track Recovery. Use different off-track offsets and re-try.
•U se a fixed gain while retrying.
•Margin servo thresholds to facilitate seek and detent.
•Margin th e erro r to le rance o f syn c mark det e ct i on.
•Force sync mark detection while retrying.
•Skip Sector Rereads. Retries, each without one of eight earlier servo
blocks.
•C o rre ction of the bad d ata with th e s o ftwar e E C C a l gorith m .
•Switching the bias current of the MR head between retries.
•TA baseline compensation, and other TA recovery methods.
Most of the spare blocks are located on the inner cylinders of the drive. All defective
bloc ks are in-line spared at the factory or at format time. Some spare b locks are
allocated at the end of each band, and are reserved for grown, revec tored blocks. All
grown defects are reallocated to one of the spare blocks, using a “nearest neighbor”
algorithm.
If bad block replacement is enabled (via the AWRE and ARRE bits in the Read-
Write Error Recovery Page), blocks that are difficult to read or write may be replaced
after mult iple re tries.
If the bad block data is recoverable, the drive first verifies that the original block
location is bad with mu ltiple write tests. If the wr ite tests fail, the original data is
reassigned to a new location.