linux-kernel - Re: [PATCH] vt: properly ignore xterm-256 colour codes

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: <20130909164623.GA12525@angband.pl>
Date:	Mon, 9 Sep 2013 18:46:23 +0200
From:	Adam Borowski <kilobyte@...band.pl>
To:	David Herrmann <dh.herrmann@...il.com>
Cc:	linux-kernel <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] vt: properly ignore xterm-256 colour codes

On Mon, Sep 09, 2013 at 05:53:19PM +0200, David Herrmann wrote:
> On Fri, Jul 12, 2013 at 10:23 PM, Adam Borowski <kilobyte@...band.pl> wrote:
> >  drivers/tty/vt/vt.c | 26 ++++++++++++++++++++------
> >  1 file changed, 20 insertions(+), 6 deletions(-)
> >
> > diff --git a/drivers/tty/vt/vt.c b/drivers/tty/vt/vt.c
> > index c677829..f7aaa28 100644
> > --- a/drivers/tty/vt/vt.c
> > +++ b/drivers/tty/vt/vt.c
> > @@ -1300,13 +1300,27 @@ static void csi_m(struct vc_data *vc)
[...]
> > -                       case 38: /* ANSI X3.64-1979 (SCO-ish?)
> > -                                 * Enables underscore, white foreground
> > -                                 * with white underscore (Linux - use
> > -                                 * default foreground).
> > -                               vc->vc_color = (vc->vc_def_color & 0x0f) | (vc->vc_color & 0xf0);
> > -                               vc->vc_underline = 1;
> 
> You break the old behavior here. _Iff_ this is what you want, then
> please do that in another commit. Explicitly state that "38" is used
> for 256color and shouldn't turn on underline+default-col. The SCO-ish
> behavior is weird, indeed, but breaking it silently is not ok.

This is implied by the description; none among modern terminal emulators
support this.  Would an additional comment in the commit message be
enough, or do I need to change the replacement into a pair of commits?
 
> > +                               i++;
> > +                               if (i > vc->vc_npar)
> 
> This should be ">=", but the for()-loop does allow your ">". So unless
> someone fixes the for-loop to use "<" (do a ++vc->vc_npar before it,
> if it's correct. But blindly doing "<=" is really irritating) I think
> this is ok.

The loop this switch is in does:
	for (i = 0; i <= vc->vc_npar; i++)
which is obviously contrary to what we're used to, but I did not want to
rewrite nearby code to match my preferences.

The change you suggest would deoptimize the code by a single unnecessary
dereference and increment, which is negligible, but since the whole cost
of speedier version is having <= instead of < in the loop, I'm not so
certain this is a good idea.

[...]
> Btw., you should put Greg Kroah-Hartman and Andrew Morton on CC. Both
> are the most likely to pick this up.

Thanks for the suggestion.  I've sent the patch two days ago to Jiri Slaby
(listed as a maintainer besides Greg) together with a newbie question, but
he's apparently busy.

I've got more changes for the vt, but there's no hurry, I wanted to test
the waters with a single minor one in 3.12 first.

-- 
ᛊᚨᚾᛁᛏᚣ᛫ᛁᛊ᛫ᚠᛟᚱ᛫ᚦᛖ᛫ᚹᛖᚨᚲ
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/