According with the FHS about /dev at:

it contains:

The following devices must exist under /dev.

      All data written to this device is discarded. A read from this device will return an EOF condition.

      This device is a source of zeroed out data. All data written to this device is discarded. A read from this device will return as many bytes containing the value zero as was requested.

Observe that both have:

All data written to this device is discarded

I read many tutorials where /dev/null is always used to discard data. But because both have the same purpose about writing (discard)


  • When is mandatory use /dev/zero over /dev/null for write/discard purpose?

BTW for other differences - practically mostly about read - we have available:

  • 2
    Why do you think it would ever be mandatory? (Well, apart from constructed cases like a chroot environment that doesn't have null but has zero, I guess.)
    – ilkkachu
    Commented Apr 14, 2022 at 20:29
  • Perhaps mandatory for some scenarios where performance and blocking would be involved. Commented Apr 14, 2022 at 21:37

3 Answers 3


If you're using Linux, it's never "mandatory" to redirect to /dev/null instead of /dev/zero. As you've noticed, you'll get the same result either way.

That said, you should always redirect to /dev/null if you're discarding data. Because everyone understands that writing to /dev/null means throwing the data away; it's expressing your intention.

On the other hand, writing to /dev/zero also throws your data away, but it's not immediately obvious that that's what you're trying to do.

Besides that, I'd be concerned whether writes to /dev/zero are allowed on other Unices, like the BSDs etc. I don't think /dev/zero is even required by POSIX, while /dev/null is. So using /dev/null for its intended purpose is maximally portable; doing anything else is sacrificing portability for no gain.

  • 1
    Interesting - here comes a secondary question: What was the purpose to allow write/discard to dev/zero if /dev/null offers the same approach Commented Apr 14, 2022 at 12:48
  • 2
    @ManuelJordan I don't know for sure, but probably because not allowing it would've taken extra work, while allowing it worked by default and didn't cause issues.
    – Egor Hans
    Commented Apr 14, 2022 at 13:04
  • 12
    If I had to construct a reason on the fly, I'd say it would have to do with the ability to open /dev/zero in read-write mode. You can at least read something from /dev/zero, something you can't do with /dev/null.
    – chepner
    Commented Apr 14, 2022 at 13:18
  • 6
    I'll bet they actually share much of the code in the kernel drivers. The only difference between them is how they respond to reads.
    – Barmar
    Commented Apr 14, 2022 at 14:07
  • 6
    @ManuelJordan: I went looking for a justification when it was implemented. I found that the kernel function for writes to /dev/zero was unified with the one handling writes to /dev/null in this patch in 1993 and hasn't been touched since. The ability to write to /dev/zero seems to have been added in this commit in October 1992 without comment. Commented Apr 14, 2022 at 18:01

There is a use case: if a program detects its stdout is directed into /dev/null and alters its behavior (which is not very Unix-way, in my opinion), and you don't want that.

I know only one example: tar. Imagine you want to ensure all files in some directory are actually readable from the disk (no I/O errors, etc.). One way to do it is to tar them to /dev/null:

tar -cv dir/ > /dev/null

But if you try it you may find that tar doesn't actually read the contents of the files, only the metadata (the command runs way too fast). And

tar -cv dir/ > /dev/zero

does read the files' contents (runs slower, with more disk reads).

  • Hngh, yep, that's a hilarious behaviour. Another way to work around it is to pipe to /dev/null through cat. Also works if you want prevent the software from detecting it's connected to a terminal.
    – ilkkachu
    Commented Apr 16, 2022 at 16:18

There is no mandatory cases, but there are use cases:

  • /dev/null: is often used when you need to provide a file name, but you do not care about the data.

    Often this is used on stdout: you do not want output of a program, so you redirect the output to /dev/null (so it is discarded). Very convenient on scripts (and cronjob), where you check the status, but text will just confuse.

    You can read /dev/null to create a empty file (but other syntaxes may be better e.g. : > empty_file_name), or when I just want the headers of a table, but I do not want to deliver data

  • /dev/zero: I use it when I should *physically" erase data, e.g. with dd command: I override an hard-disk with zeros. Zeros is never ending, so it will override all the hard disk, without having a original file (or occupy space). Alternate it is to use /dev/random but it will much slower.

    I do not remember I wrote to /dev/zero (it seems /dev/null has a better name).

Note: /dev/zero will continue to feed data (and it tell program that it has data). /dev/null instead (when used as input) just tell the program that there are no more data).

It is interesting the documentation you linked. As you see /dev/null is usually used to write, and /dev/zero to read (it according with the first case for each one listed on documentation).

In general: when we are creating a script, we may find new way to use them. It depends on what I'm doing, and what parameters are required by a specific programs. We have also yes command, which deliver us y (with a new line), when we need to feed many confirmation to programs.

  • 6
    I can't see a reason to write to /dev/zero here.
    – ilkkachu
    Commented Apr 14, 2022 at 20:27

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .