Ext file system mechanism principle

https://blog.csdn.net/li_wen01/article/details/105313344

There are many types of file systems ,ext234,windows On NTFS, The CD-ROM file system iso9660,MAC Hybrid file systems HFS, Network file system NFS,Oracle R & D btrfs, Old fashioned FAT/FAT32 etc. .

1 The components of the file system

block appear

Read and write at the bottom of the hard disk io One sector at a time , If you want to read and write a lot of files , It must be slow in sectors , Consumption performance , So hard drives use a concept called logical blocks . Logical blocks are logical , The disk drive is responsible for maintenance and operation , It is not physically partitioned like a sector , The size of a logical block may contain one or more sectors , Each logical block has a unique address , be called LBA Logic Block Address. With logical blocks , The disk controller operates on data in logical blocks , Read and write one logic block at a time , The disk controller knows how to translate logical blocks into corresponding sectors and read and write data .

here we are Linux Operating system hierarchy , A read / write unit, also known as a block, is provided through the file system , The size of the file system data block is generally 1024bytes,1k,2048bytes 2k, or 4096bytes 4k, File system data block is also a logical concept , File system level maintenance , The logical data block of the disk is maintained by the disk controller , File system io The manager knows how to translate its data block into the address of the data block maintained by the disk LBA. For those who use the file system io In terms of operation , For example, reading and writing documents , these io The basic unit of is the data block on the file system , Read and write one file system block at a time . For example, when you need to read one or more blocks , File system io The manager first calculates which disk data blocks these file system blocks correspond to , That is to calculate LBA, Then tell the disk controller which blocks to read , The disk controller translates these blocks into sector addresses , Then read the data from the sector , Then, these sector data are reorganized and written into memory through the hard disk controller .

This article is about file system , Naturally, the focus is on the file system rather than the disk , So what will happen later block Both mean to add the data block of your system , instead of , Logical blocks maintained by disk .

file system block The appearance of makes the reading and writing performance on the file system level greatly improved , And a lot less debris . But its side effect is that it can waste space . Because the file system uses block It's a read-write unit , Even if only 1k size , Will also occupy one block, The rest of the space is completely wasted . Under some business requirements, a large number of small files may be stored , This time a lot of space is wasted .

Despite its shortcomings , But the advantages are obvious enough , Nowadays, hard drives are cheap and performance oriented , Use block It must be .

bmap