Pipes allow to build one-directional communication channels between
related processes. They re available in shells through the
| symbol, or as pipe() and
pipe2() syscalls. Pipes can be represented as disk
references using named pipes or FIFO files, which are created using
mkfifo.
An important feature of pipes is the max size of an atomic write. The
PIPE_BUF constant (man 7 pipe) determines it and sets it to
4096 bytes. Please, read the man carefully if you want to rely on this
guarantee. The default size of the pipe buffer (which is a ring buffer
with slots), is the number of slots multiplied by the
PIPE_BUF constant, which is 16 and 4KiB by default, giving
us 64KiB. Pipes get used instead of files due to their more convenient
API.
SIGPIPEA process receives SIGPIPE when writing to a pipe with a
dead reader. Default behaviour in such case is to exit, sending
SIGPIPE to the next producer in line. Thus exits cascade
through pipelines. When a pipeline terminates, shell sets
$? to the return code of the last command in the pipeline
(i.e. rightmost) and populates $PIPESTATUS array with all
return codes of the pipeline. $? containing the result of
the last command can lead to unexpected results, where an important
command in the pipeline fails without it getting reflected in the return
of the pipeline. Use set -o pipefail to propagate any
fails in the pipeline to the return value.
In some situations you can improve performance of low-level code by
using zero-copy operations like splice(),
vmsplice() and tee() and treating pipes as
64KiB kernel space buffers.
splice() - moves data from the buffer to an arbitrary
file descriptor, or vice versa, or from one buffer to another
(man 2 splice)vmsplice() - "copies" data from user space into the
buffer (man 2 vmsplice)tee() - allocates internal kernel buffer
(man 2 tee)