linux-kernel - Re: [PATCH] checkpatch: Require commit text and warn on long commit text lines

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]

Date:   Fri, 13 Jul 2018 17:08:37 -0700
From:   Joe Perches <joe@...ches.com>
To:     pheragu@...eaurora.org, Andrew Morton <akpm@...ux-foundation.org>,
        Greg KH <gregkh@...uxfoundation.org>
Cc:     apw@...onical.com, linux-kernel@...r.kernel.org,
        tsoni@...eaurora.org, bryanh@...eaurora.org,
        ckadabi@...eaurora.org, David Keitel <dkeitel@...eaurora.org>
Subject: Re: [PATCH] checkpatch: Require commit text and warn on long commit
 text lines

On Fri, 2018-07-13 at 16:28 -0700, pheragu@...eaurora.org wrote:
> On 2018-07-13 14:46, Joe Perches wrote:
> > On Fri, 2018-07-13 at 14:40 -0700, Prakruthi Deepak Heragu wrote:
> > > Commit text is almost always necessary to explain why a change is 
> > > needed.
> > 
> > This bit seems sensible, but perhaps it should just count the
> > number of lines after the end of email headers and before any
> > Signed-off-by:/Signature line
> > 
> 
> While committing the changes, one can just write the subject and not 
> write
> the commit text at all. So, if we just count the lines between email 
> headers
> and signed-off, we still do count lines which form the subject, but the
> commit text is still absent. Also, subject can be longer than one line. 
> So,
> just counting lines doesn't really guarantee the presence of commit 
> text.

Not true.
Look at $in_header_lines and $in_commit_log.

> > > Also, warn on commit text lines longer than 75 characters. The commit 
> > > text
> > > are indented and may wrap on a terminal if they are longer than 75
> > > characters.
> > 
> > This is already exists via
> > 
> > # Check for line lengths > 75 in commit log, warn once
> > 		if ($in_commit_log && !$commit_log_long_line &&
> > 		    length($line) > 75 &&
> > 
> 
> True, but this patch points out every line of the commit text that is
> exceeding the limit.

Which is bad because things like dump_stack() are added in
commit logs and those are already allowed to be > 75 chars.

Anyway, something like this probably works.  Please test.
---
 scripts/checkpatch.pl | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl
index b5c875d7132b..8b5f3dae31c9 100755
--- a/scripts/checkpatch.pl
+++ b/scripts/checkpatch.pl
@@ -2240,6 +2240,7 @@ sub process {
 	my $in_header_lines = $file ? 0 : 1;
 	my $in_commit_log = 0;		#Scanning lines before patch
 	my $has_commit_log = 0;		#Encountered lines before patch
+	my $commit_log_lines = 0;	#Number of commit log lines
 	my $commit_log_possible_stack_dump = 0;
 	my $commit_log_long_line = 0;
 	my $commit_log_has_diff = 0;
@@ -2497,6 +2498,18 @@ sub process {
 
 		$cnt_lines++ if ($realcnt != 0);
 
+# Verify the existence of a commit log if appropriate
+# 2 is used because a $signature is counted in $commit_log_lines
+		if ($in_commit_log) {
+			if ($line !~ /^\s*$/) {
+				$commit_log_lines++;	#could be a $signature
+			}
+		} else if ($has_commit_log && $commit_log_lines < 2) {
+			WARN("COMMIT_MESSAGE",
+			     "Missing commit description - Add an appropriate one\n");
+			$commit_log_lines = 2;	#warn only once
+		}
+
 # Check if the commit log has what seems like a diff which can confuse patch
 		if ($in_commit_log && !$commit_log_has_diff &&
 		    (($line =~ m@^\s+diff\b.*a/[\w/]+@ &&