Dd

Template:UnixTemplate:Lowercase dd (doesn't stand for anything that matters) is a powerful and dangerous unix program that performs raw block copies of data. In particular, it is often used for copying hard disk data raw partition-to-partition, partition-to-file, and/or file-to-partition.

Synopsis

Often the invocation is as simple as

dd if=<input file = stdin> of=<output file = stdout> bs=<block size = 512> count=<block count = until end of input>

That is, from the beginning of if copy count blocks of bs bytes to the file of.

Block sizes

bs can use human suffixes. Use e.g. kB, MB, GB for decimal meanings (1000^n), but use e.g. K, M, G or KiB, MiB, GiB for the binary equivalents (1024^n).

bs can be replaced by ibs and obs to set in and out block sizes separately.

Writing to files in-place

See conv=notrunc, seek=, skip=.

Hard disk recovery

When recovering a failing drive, among the recommendations are:

• Use a small block size, such as 512 bytes
• Use the conv option noerror, which causes the program to continue even if a read fails
• Use the conv option sync, which causes short (erroneously read) blocks to be padded out so that any good output still aligns correctly

In other words:

dd bs=512 if=/dev/<some device> of=output.img conv=noerror,sync

Print status messages

By default, dd keeps quiet about its progress. In most versions, there is a way to send it a signal to request a progress report.

In Linux, this signal is the USR1 signal. The following command will cause all running dd processes to echo status every second:

while killall -USR1 dd; do sleep 1; done

killall can be replaced with the equivalent kill command and pid to signal only one process.

dcfldd

The program dcfldd is a fork of dd that has special security- and display-related options. The USR1 signal is not necessary to produce a readout; by default, a status is printed to stderr every 256 blocks. Set statusinterval=<number of blocks> to change the interval.