scripts: get_feat.pl: use /usr/bin/env to find perl

I tried running `make pdfdocs` on NixOS, but it failed because
get_feat.pl uses a shebang line with /usr/bin/perl, and that file path
doesn't exist on NixOS. Using the more portable /usr/bin/env perl fixes
the problem.

Signed-off-by: David Reaver <me@davidreaver.com>
Link: https://lore.kernel.org/r/20220625211548.1200198-1-me@davidreaver.com
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

scripts/get_feat.pl: allow output the parsed file names

Such output could be helpful while debugging it, but its main
goal is to tell kernel_feat.py about what files were used
by the script. Thie way, kernel_feat.py can add those as
documentation dependencies.

Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Link: https://lore.kernel.org/r/11b438ee01e00c866f5ea197d6aecc26e9f86945.1648290305.git.mchehab@kernel.org
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

scripts: get_feat.pl: reduce table width for all features output

Auto-adjust the table columns width to better fit under
terminals, by breaking the description on multiple lines
and auto-estimating the minimal size for the
per-architecture status.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/9d39ac3fd51f1360aecc328c01558be88a1d6930.1607095090.git.mchehab+huawei@kernel.org
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

scripts: get_feat.pl: change the group by order

Right now, arch compatibility is grouped by status at the
alphabetical order from A to Z, and then from a to z, e. g:.

---
TODO
ok

Revert the order, in order to print first the OK results,
then TODO, and, finally, the not compatible ones.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/46d53d138eab8e4a55124323ceb5b212c6eedd08.1607095090.git.mchehab+huawei@kernel.org
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

scripts: get_feat.pl: make complete table more coincise

Currently, there are too many white spaces at the tables,
and the information is very sparsed on it.

Make the format a lot more compact.

Suggested-by: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/8165ff379313e63a69898db19d790e4436224ffd.1607095090.git.mchehab+huawei@kernel.org
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

scripts: get_feat.pl: use its implementation for list-arch.sh

Add support for the same output format as the bash script,
and use its implementation instead of the previous one.

I opted to do such patch in order to have a single script
responsible for parsing Documentation/features and
produce different outputs.

As someone may rely on the past format, which is easy
to parse it, get_feat.pl now gains a new command with
the same output format as the previous script.

As a side effect, the perl script is a lot faster, as it reads
each file only once, instead of parsing files several times
via a for command and grep commands inside it.

This patch also changes the features list order to be
case-insensitive, in order to better match the output of
the existing script.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/a97f49677805ad4e6b982d02c0db8c9dfbbd20a6.1606748711.git.mchehab+huawei@kernel.org
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

scripts: get_feat.pl: improve matrix output

Instead of producing a too wide table, let's split it per
subsystem, and use a better notation in order to make easier
for the reader to identify how a feature is supported on
multiple architectures.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/9c3e8c813e8146c5c30e10fa75974f8fbfe6016a.1606748711.git.mchehab+huawei@kernel.org
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

scripts: get_feat.pl: add a script to handle Documentation/features

The Documentation/features contains a set of parseable files.
It is not worth converting them to ReST format, as they're
useful the way it is. It is, however, interesting to parse
them and produce output on different formats:

1) Output the contents of a feature in ReST format;

2) Output what features a given architecture supports;

3) Output a matrix with features x architectures.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/8b0c1ad06d689283a6d78c4ccd188a02c3acc0de.1606748711.git.mchehab+huawei@kernel.org
Signed-off-by: Jonathan Corbet <corbet@lwn.net>