linux-kernel - Re: [PATCH] console UTF-8 fixes

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: <20070619144230.GA14609@uhulinux.hu>
Date:	Tue, 19 Jun 2007 16:42:31 +0200
From:	Egmont Koblinger <egmont@...linux.hu>
To:	Bodo Eggert <7eggert@....de>
Cc:	Andrew Morton <akpm@...ux-foundation.org>,
	"H. Peter Anvin" <hpa@...or.com>,
	Jan Engelhardt <jengelh@...ux01.gwdg.de>,
	Alan Cox <alan@...rguk.ukuu.org.uk>,
	Behdad Esfahbod <behdad@...dad.org>,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] console UTF-8 fixes

On Tue, Jun 19, 2007 at 03:54:52PM +0200, Bodo Eggert wrote:

> Does the FLUSH DTRT by design, or does it just shrink and hide the original
> race?

I haven't deeply studied this aspect of the source, don't know what
_exactly_ this FLUSH does, but of course I have an inner feeling about this.
Kind of this macro does the _real_ print on the screen. It actually calls
vc->vc_sw->con_putcs, whereas I guess vc_sw refers to the particular
implementation (vga, fb) and con_putcs might be doing the real job.

Without these two new FLUSHes the behavior was very funny: applications
(e.g. text editors) highlighed incorrect cells, but either switching between
VTs, or gpming over these cells corrected the display. So the logical
content behind the scenes (reflected by /dev/vcsa*) was maintained
correctly, the bug is somewhere when the content is first displayed.

I saw that this FLUSH was already used 3 times inside do_con_write(), I
placed it here 2 more times (when changing color attr to inverse and when
reverting to normal). I think it cannot hurt, and apparently it fixes the
problem.

But you may be right: yes, it might be a bug (or misfeature) in the FB code,
too. Could you please look after it? I don't think I'd be able to do it and
have time for this.

> And why wasn't VGA affected, too?

I guess tt must be some difference between the actual VGA and FB code that
is called from here. I don't know whether it's a bug in FB or just some
difference.

Perhaps it's that the FB driver's con_putcs() call uses vc->vc_attr (the
current attr for the console) for each char cell to be printed, instead of
the own attr values of the cells themselves. Just an idea...

-- 
Egmont
-
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/