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
| ||
|
Message-ID: <9de78c46ad8da9e4c59ba6206d9dcc51bc984db1.camel@perches.com> Date: Tue, 27 Sep 2022 13:39:16 -0700 From: Joe Perches <joe@...ches.com> To: Kees Cook <keescook@...omium.org>, ditya Srivastava <yashsri421@...il.com> Cc: Andy Whitcroft <apw@...onical.com>, Dwaipayan Ray <dwaipayanray1@...il.com>, Lukas Bulwahn <lukas.bulwahn@...il.com>, linux-kernel@...r.kernel.org, linux-hardening@...r.kernel.org Subject: Re: [PATCH] checkpatch: Don't count URLs for "line too long" On Tue, 2022-09-27 at 07:07 -0700, Kees Cook wrote: > On Mon, Sep 26, 2022 at 10:40:50PM -0700, Joe Perches wrote: > > On Mon, 2022-09-26 at 19:15 -0700, Kees Cook wrote: > > > URLs (not in a Link: field) should be ignored for "line too long" > > > warnings. Allow any line containing "https://" or "http://". > > > > More generally, this should be for any URI > > > > > diff --git a/scripts/checkpatch.pl b/scripts/checkpatch.pl > > [] > > > @@ -3175,6 +3175,8 @@ sub process { > > > # file delta changes > > > $line =~ /^\s*(?:[\w\.\-\+]*\/)++[\w\.\-\+]+:/ || > > > # filename then : > > > + $line =~ /https?:\/\// || > > > + # URLs > > > $line =~ /^\s*(?:Fixes:|Link:|$signature_tags)/i || > > > # A Fixes: or Link: line or signature tag line > > > $commit_log_possible_stack_dump)) { > > > > And I don't recollect why this wasn't applied, but I think it's a rather better, > > more complete, patch: > > > > https://lore.kernel.org/lkml/20210114073513.15773-2-yashsri421@gmail.com/ > > > > I believe I did ack it and forward it to Andrew Morton, but I don't > > see it on a lore list. > > Many patches have internal references in the commit log (that aren't > appropriate for "Link:") like: > > [1] https://some.domain/path/ > > It seems more flexible to just ignore lines with "http" in it... URLs vs URIs which can have multiple forms. Take a look at the results of this grep: $ git log -1000000 --format=email --no-merges --grep='://' | grep '://' | \ grep -v -P 'https?://' | grep -P -oh '\b\w+://' | \ sort | uniq -c | sort -rn 422 git:// 46 ftp:// 15 file:// 10 smb:// 5 c:// 3 usbip:// 2 svn:// 2 rsync:// 2 pvr:// 1 ttp:// 1 systemd:// 1 server:// 1 radio:// 1 nbd:// 1 irc:// 1 intel_bts:// 1 htp:// 1 40176:// The more inclusve search pattern finds many uses of git:// and ftp:// (and probably a few typos too) Assuming the URI scheme uses ://, it's perhaps reasonable to change the suggested search in Aditya's patch from + if ($line =~ /^\s*([a-z][\w\.\+\-]*:\/\/\S+)/i) { to maybe something like this so the URI's use of :// is more obvious + if ($line =~ m@^.*\b([a-z][\w\.\+\-]*://\S+)@i) {
Powered by blists - more mailing lists