4 Nov 2024

Badblocks

Another tips/reminder post for when I’ll need it again in the future.

When I get a new hard drive, or I’m getting rid of one, I like to make sure the drive is properly wiped-out and that it’s in a good working state. It turns out that there’s a software doing both of that: badblocks. Note that wiping out an SSD is different and this doesn’t really apply.

Checking hard drive’s block size

This is an optional part, but seems useful in some cases to specify the disk’s block size to badblocks.

Install blockdev first if necessary, on Debian: sudo apt install util-linux

Get the block size: sudo blockdev --getbsz /dev/sdX (replace sdX by the hard drive you want to check)

Installing badblocks

On Debian: sudo apt install e2fsprogs

Using badblocks

Note: the data on the drive will be lost as testing it requires writing on the whole disk.

I usually run badblocks this way:

sudo badblocks -s -v -w -b 4096 -c 65536 -t 0xaa -t 0x55 -t 0xff -t 0x00 -t random -o /tmp/badblocks.txt /dev/sdX

Here is a detail of the used options:

-s: show progress.
-v: verbose, display the number of errors.
-w: write mode, this is DESTRUCTIVE, all content on the hard drive will be lost.
-b: specify the block size.
-c: specify the number of blocks to test at a time, I usually use a high value to increase speed, adjust to your own needs.
-t: specify the patterns to write on the drive, if not specified badblocks will by default use 0xaa, 0x55, 0xff and 0x00 but since I want to add the random pattern too, I have to put all of them again. Basically, badblocks write the pattern on the whole drive, then read it again, the read pattern should match the written one. Trying different patterns to cover all the combinations to make sure all bits of the drive work:
- 0xaa being 10101010 ;
- 0x55 being 01010101 ;
- 0xff being 11111111 ;
- 0x00 being 00000000 ;
- random being a random pattern chosen when starting badblocks.
-o: writes the bad blocks found in a file.
/dev/sdx: the drive to test.