SC2312 – ShellCheck Wiki

See this page on GitHub

Sitemap

Consider invoking this command separately to avoid masking its return value (or use '|| true' to ignore).

Optional - check-extra-masked-return

This is an optional rule, which means that it has a special "long name" and is not enabled by default. See the optional page for more details. In short, you have to enable it with the long name instead of the "SC" code like you would with a normal rule:

.shellcheckrc

enable=check-extra-masked-return # SC2312

Problematic code:

set -e
cd "$(get_chroot_dir)/etc"
tar xf "${config}"

Correct code:

set -e
dir="$(get_chroot_dir)"
cd "${dir}/etc"
tar xf "${config}"

Correct code: (with correction)

set -e
dir="$(get_chroot_dir)"
[[ -d "${dir}" ]] || exit 1
cd "${dir}/etc"
tar xf "${config}"

Rationale:

In the problematic example, the exit code for get_chroot_dir is ignored because it is used in a command substitution in the argument of another command.

If the command shows error: Can't determine chroot and exits with failure without outputting a directory, then the command being run will be cd "/etc" and the script will proceed to overwrite the host system's configuration.

By assigning it to a variable first, the exit code of the command will propagate into the exit code of the assignment, so that it can be checked explicitly with if or implicitly with set -e.

Exceptions:

If you don't care about the command's exit status, already handle it through a side channel like <(cmd; echo $? > status), or (in the case of background processes and process substitution) wait on the result like <(cmd) ; wait $!, then you can either ignore the suggestion with a directive, or use || true (or || :) to suppress it.

Handling process substitution failures

Note that you can combine file descriptor duplication with wait to reference process substitution output (or input) while retaining the exit codes of those processes. For example:

generate_data() {
    declare i
    for (( i = 0 ; i < 5 ; ++i ))
    do
        date -d "$RANDOM hours"
    done
}

consume_data() {
    declare line
    while IFS= read -r line
    do
        echo Consuming line: "$line"
    done
}

declare \
    input_file_descriptor \
    process

# The following statement
#
# - uses process substitution to allow us to read the output of `generate_data`
#   via a filename and
# - duplicates the file descriptor for that file so that it is not
#   immediately closed
#
# Note that process substitution uses either `pipe(2)` or named pipes (FIFOs)
# with `O_RDONLY` or `O_WRONLY`, and so the file descriptor that is duplicated
# via `[N]<&WORD` is only opened for reads
exec {input_file_descriptor}< <(
    generate_data
)
process=$!

# Returns non-zero if `consume_data` does
consume_data <&"$input_file_descriptor"

# Returns non-zero if `generate_data` does
wait "$process"

This can be particularly helpful with readarray for robust array handling

https://mywiki.wooledge.org/BashPitfalls#cmd1_.26.26_cmd2_.7C.7C_cmd3

ShellCheck is a static analysis tool for shell scripts. This page is part of its documentation.