linux-kernel - [Bisected] nouveau: v3.7-rc breaks multi-monitor setup

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>] [day] [month] [year] [list]

Message-ID: <508D99F1.9080406@Gmail.com>
Date:	Sun, 28 Oct 2012 15:47:45 -0500
From:	Calvin Owens <jcalvinowens@...il.com>
To:	David Airlie <airlied@...ux.ie>, Ben Skeggs <bskeggs@...hat.com>,
	Martin Peres <martin.peres@...ri.fr>
CC:	Calvin Owens <jcalvinowens@...il.com>,
	dri-devel@...ts.freedesktop.org, linux-kernel@...r.kernel.org
Subject: [Bisected] nouveau: v3.7-rc breaks multi-monitor setup

Hello all,

I have a somewhat unusual setup with 3 monitors, two connected to an
NVIDIA GT9800, and one connected to the onboard graphics card (a Radeon
HD4250, which is selected as the primary card in the BIOS). This has
worked well for quite some time using nouveau, but the 3.7 merge window
changes cause the two monitors attached to the NV card to appear blank.

The interesting thing is that X doesn't seem to even notice the
difference under 3.7-rc: the screens are blank, but it reports no errors
at all. Even the driver itself seems to happily go along writing to it's
framebuffer (see log below).

Also, the NV card *does* still work when I turn off the onboard gpu
(by setting the NV card as the primary (The BIOS masks the onboard
gpu in this case, so the 3-monitor configuration doesn't work)).

It would seem as though there is some important initialization step that
the BIOS does that nouveau used to do too, but isn't doing anymore.
(Usually when kernel modesetting happens (or when the BIOS comes up if
it's the primary card), the fan on the card quiets down, but now it just
stays in blow-dryer mode.)

Here is a nouveau debug loglevel "spam" bootup under 3.7-rc2+:
	* http://pastebin.com/Fysifmta
(It was 180k lines long - I tried to condense it by cutting out large
blocks of the same message)

I've bisected this to 77145f1 (which appears to be the last commit in a
big restructuring of the nouveau source tree). Unfortunately, the 40
commits before that one cause build errors, so narrowing it down will be
more time-consuming...

Other possibly relevant info:
lspci:		http://pastebin.com/6yjUv5du
Kconfig:	http://pastebin.com/eUPpNyJZ

Information from "git-bisect":
git output:	http://pastebin.com/DMRSr5AA
git bisect log: http://pastebin.com/d8MRXRAC
(77145f1 is the earliest build-able commit that causes the regression)

Let me know if there's anything I can do to help, or any more info that
would be useful to you.

Cheers,
Calvin Owens
--
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/