Organization of the file system
the integrity of the file system. They can run after loading or after the crash
and carry out a variety of scanning multiple file system data structures in
search of contradictions. There are also heuristic test. For example, to find
the index node number, which exceeds the number on the disk or finding files in
user directories belonging roots.
The Office of bad blocks. The existence of bad blocks on the disk normal. A bad
blocks is usually described as disk blocks for which an amount calculated
Control-readable data does not match the amount held by a control. Often
appearing in the operation. Sometimes they already exist in the supply, together
with the list because very difficult for suppliers to make disk completely free
of defects. Two solution to the problem of bad blocks, one at a different level
sensor OS kernel level. The first way is to keep the list of bad blocks in the
disk controller. When the controller is initialized, it reads bad blocks and
replace the defective unit standby, marking the display of the list of bad
blocks. All real requests will be to go to the contingency block. It should be
noted that this mechanism lift (the most common mechanism for processing
requests to the disk block) will be working inefficiently. The fact is that
there is a strategy of priority requests to the disk (see Chapter O). The
strategy dictates the direction of the head disk reader to the correct cylinder.
Usually blocks placed on standby outside the cylinder. If a bad block is located
at the internal cylinder, and predictive controller implements a transparent
manner, the apparent movement of the ball will be done to the internal cylinder,
but the actual to the outside. This is a violation of the strategy and,
consequently, less the scheme. Decision-level OS might be as follows. First, it
is necessary to carefully design a file containing bad blocks. Then they are
removed from the list of free blocks. Then you need to make to this file has not
been hits. If possible, the problem is solved.
The performance file system The most typical technology to speed up work with
the disk cache. Resorting to disk usually 100000 times slower than to memory. (Address
to the memory of several hundred-nanosecond, and reading from the disk-block
tens of milliseconds). Through information in the disk cache memory can reduce
the number of disk operations, keeping part of the disk blocks, which were made
before this treatment, in a special area of memory called a cache buffer (cache).
This is possible because of the inherent properties of localization OS (localization
properties of much discussion in the chapters dealing with the description of
the work management system memory).
Next part
|