tools/nxstyle: fix most of the incompatibilities with the coding standard

[raiden00pl]

Summary

This PR fixes a lot of gaps in the nxstyle tool. Unfortunately, the codebase contains many inconsistencies with the coding standard. Fixing everything might mess up git history a bit, so I'm not sure if it's worth it.

For now I only fixed nxstyle.c, sched, audio and net

Impact

better compatibility with nuttx coding style.

This solves some of the issues related to nxstyle, but I didn't check which ones: https://github.com/apache/nuttx/issues?q=is%3Aissue%20state%3Aopen%20nxstyle

Testing

Output from /tools/nxstyle_sweep.sh that checks all files:

nxstyle-errors.txt:  101060 diagnostics                    
nxstyle-files.txt: 2690 files

Checked 17165 files, 2690 failed

[github-actions]

MemBrowse Memory Report

arduino-mega2560

• flash: .text +108 B (+0.2%, 64,394 B / 262,144 B, total: 25% used)

esp32-devkitc

• ROM: .flash.text +80 B (+0.1%, 124,476 B / 4,194,272 B, total: 3% used)
• irom0_0_seg: .flash.text +80 B (+0.1%, 88,680 B / 3,342,304 B, total: 3% used)

hifive1-revb

• flash: .text +68 B (+0.1%, 83,376 B / 4,194,304 B, total: 2% used)

mirtoo

• kseg0_progmem: .text +132 B (+0.2%, 67,008 B / 131,072 B, total: 51% used)

rx65n-rsk2mb

• ROM: .text +64 B (+0.1%, 86,496 B / 2,097,152 B, total: 4% used)

s698pm-dkit

• Code: .text +240 B (+0.1%, 363,648 B)

stm32-nucleo-f103rb

• flash: .text +76 B (+0.2%, 34,116 B / 131,072 B, total: 26% used)
  No memory changes detected for:
• qemu-armv8a
• qemu-intel64

[acassis]

@raiden00pl please fix conflict

[raiden00pl]

I'll be adding changes to nxstyle in separate PRs. I'm still unsure whether to fix all the nxstyle upstream issues, or leave it for the future and fix it step by step.

[JianyuWang0623]

While debugging a CI failure on an unrelated PR (apache/nuttx-apps#3751), I found a false positive in the Bad left brace alignment check that the regression-introducing commit bdeb262b8d3 ("tools/nxstyle: indent code against its enclosing brace, not modulo 4") added, and confirmed this PR already fixes it.

Root cause: the exemption for a macro-that-takes-a-block (e.g. list_for_every_entry(...)) only compares against prevcodeindent (the indentation of the previous line):

else if (prevlastcode == ')' && indent == prevcodeindent + 2)
  {
  }

When the macro call wraps its arguments onto a continuation line, prevcodeindent becomes the indentation of that continuation line instead of the line where the macro call started, so the exemption never matches and the brace is flagged.

Minimal repro (built master's tools/nxstyle.c as of this comment, nxstyle repro.c):

void add_ready(FAR struct action_manager_s *am,
                FAR struct action_s *ready)
{
  list_for_every_entry(&am->ready_actions, ready, struct action_s,
                        ready_node)
    {
      do_something(ready);
    }
}

$ ./nxstyle repro.c
repro.c:19:4: error: Bad left brace alignment

This isn't an isolated case — the same false positive currently fires on existing, unmodified code such as drivers/vhost/vhost.c on master.

I confirmed this branch (nxstyle_fixes) already resolves it. The relevant change extends the exemption with a second alternative, stmt_lineindent + 2, which covers the continuation-line case:

else if (prevlastcode == ')' &&
         (indent == prevcodeindent + 2 ||
          indent == stmt_lineindent + 2))
  {
  }

$ ./nxstyle repro.c   # built from this branch
$ echo $?
0

Given the scope of this PR is large and it currently has merge conflicts against master, would it be worth splitting out just the nxstyle.c fix (or at least this specific brace-alignment exemption) into its own smaller PR so the regression can be addressed independently? Happy to help if useful.

[cederom]

Thank you @raiden00pl looks like this one is really important and welcome as it fixes CI checks in PRs that does not touch the impacted parts :-)

[raiden00pl]

@JianyuWang0623 the fix for this should be in 7ead748

I created PR for this commit: #19955 but first #19934 must be merged

[cederom]

@JianyuWang0623 the fix for this should be in 7ead748

I created PR for this commit: #19955 but first #19934 must be merged

#19934 is now merged :-)