linux-kernel - Re: [PATCH 2/3] checkpatch: tweak extern in C warning

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CAJfuBxz7DMfFOwyvUvN7F9Jc=meQs7ro5OFhiycq1GxMNHRPzw@mail.gmail.com>
Date:   Sat, 26 Jun 2021 21:47:11 -0600
From:   jim.cromie@...il.com
To:     Joe Perches <joe@...ches.com>
Cc:     LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 2/3] checkpatch: tweak extern in C warning

On Sat, Jun 26, 2021 at 12:46 PM Joe Perches <joe@...ches.com> wrote:
>
> On Fri, 2021-06-25 at 21:40 -0600, Jim Cromie wrote:
> > The extern-in-C rule has one important exception: the symbol is
> > defined in/by the linker script.  By convention, these almost always
> > contain: _start, _stop, _end.  Suppress the warning on such symbols.
> []
> > diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
> > []
> > @@ -6910,7 +6910,8 @@ sub process {
> >                   $stat =~ /^.\s*extern\s+/)
> >               {
> >                       WARN("AVOID_EXTERNS",
> > -                          "externs should be avoided in .c files\n" .  $herecurr);
> > +                          "externs should be avoided in .c files\n($stat)\n" .  $herecurr)
> > +                         unless $stat =~ /_start|_stop|_end/;
>
> nak.
>
> As far as I can tell, there's no reason these symbols
> should not be in .h files.
>

judging from the codebase, it has been a case-by-case decision,
with 8/10 of the linker-vars extern'd into C files, not headers.

[jimc@...do wk-test]$ perl -ne '/(\w*_(?:start|stop|end)(?:_\w+))/ and
print "$1\n"' `find . -name \*.lds.h` > symbols
[jimc@...do wk-test]$ wc symbols
  99   99 2112 symbols
[jimc@...do wk-test]$ grep -n -r --exclude-dir=builds/ -f symbols .  |
grep -E '\.c:' | grep extern | wc
     79     331    6402
[jimc@...do wk-test]$ grep -n -r --exclude-dir=builds/ -f symbols .  |
grep -E '\.h' | grep extern | wc
     19      81    1581

8/10 cases dont expose these symbols in headers,
Makes sense to me, mostly theyre internal, and often double-underscored too.
2/10 are presumably in headers for specific reasons.


> besides that:
>
> output is single line, $stat should not be used and
> using unless is not desired.
>

could you clarify ?
style issues are easy, std if form...
$stat is already used, it must contain extern to get here.
checking it for a likely-linker-symbol seems fair.