cromfs - Copyright (C) 1992,2011 Bisqwit (http://iki.fi/bisqwit/) License: GPL3 Homepage: http://bisqwit.iki.fi/source/cromfs.html Understanding how files are stored in cromfs archives. -------------------------------- This simple example shows how mkcromfs stores files. Let's assume the following input parameters: Our block size = 5 bytes (-b option in mkcromfs) Our fblock size = 20 bytes (-f option in mkcromfs) Our input is two files, containing the following data: File1: "My hissing little kitten that ate a big fish" File2: "Jim's big dog is chasing cars" (Let's assume there are no directories etc. to be encoded) First, this data is split into blocks. File1 data: "My hi" "ssing" " litt" "le ki" "tten " "that " "ate a" " big " "fish" File2 data: "Jim's" " big " "dog i" "s cha" "sing " "cars" Then, mkcromfs figures out how to all block data exactly once. The resulting fblocks are: 01234567890123456789 Fblock 0: "My hissing little ki" Fblock 1: "tten that ate a big " Fblock 2: "fishJim'sdog is cha" Fblock 3: "cars" The resulting block table is: Block 0: Fblock 0, offset 0 ("My hi") Block 1: Fblock 0, offset 5 ("ssing") Block 2: Fblock 0, offset 10 (" litt") Block 3: Fblock 0, offset 15 ("le ki") Block 4: Fblock 1, offset 0 ("tten ") Block 5: Fblock 1, offset 5 ("that ") Block 6: Fblock 1, offset 10 ("ate a") Block 7: Fblock 1, offset 15 (" big ") Block 8: Fblock 2, offset 0 ("fish") Block 9: Fblock 2, offset 4 ("Jim's") Block 10: Fblock 2, offset 9 ("dog i") Block 11: Fblock 2, offset 14 ("s cha") Block 12: Fblock 0, offset 6 ("sing ") Block 13: Fblock 3, offset 0 ("cars") How? First, "My hi" is stored. It has never been seen before, so it's just appended to a fblock, Fblock 0. Similarly for all the subsequent blocks of File1. When " big " is met the second time (in File2), mkcromfs knows it has already encoded this block (see BlockIndexing.txt), so it will just reuse the same block pointer (Block 7, see above). When "sing " is met in File2, mkcromfs finds that the string "sing " is already found in Fblock 0, because it was saved there as part of File1, so Block 12 becomes a pointer to the existing data. The resulting inodes corresponding to File1 and File2 are: File1: length 44 bytes, block list: 0, 1, 2, 3, 4, 5, 6, 7, 8 File2: length 29 bytes, block list: 9, 7, 10, 11, 12, 13 Last, the cromfs archive is constructed by compressing the inode table, the block table, and each fblock.