On Mon, Dec 01, 2014 at 12:08:35AM +0000, Hendrik Boom wrote: > The option -t random specifies that "the block should be filled with a > random bit pattern". > > Now, just how random is that bit pattern. > > Does it choose a random byte and fill the entire hard drive with it? If I'm reading the source correctly, Read-Write testing happens in the function test_rw()[1] which, in turn, calls pattern_fill()[2] at the start of a pass. This pattern is stored in a buffer (by default 64 blocks) and used for the whole pass. > > Does it make up a random disk block and write that to the whole disk? > > Or does each block get its own randomly chosen data, presumably generated > by a pseudorandom generator so as to catch bad seeks? In this case, is > the seed always the same, so I could take a disk I wrote on with -t > random and check it next week, the computer being off in between? The random() call is seeded[3] by the system time, so it's unlikely you could get the PRNG to generate the same value. If you desire to do this, it's better to supply your own test_pattern. > > -- hendrik > [1]: http://sources.debian.net/src/e2fsprogs/1.42.8-1/misc/badblocks.c/?hl=2#L584 [2]: http://sources.debian.net/src/e2fsprogs/1.42.8-1/misc/badblocks.c/?hl=2#L311 [3]: http://sources.debian.net/src/e2fsprogs/1.42.8-1/misc/badblocks.c/?hl=2#L1060
Attachment:
signature.asc
Description: Digital signature