linux-kernel - Re: i915/kms/backlight-combo mode problem (was: Re: Linux 2.6.39-rc5)

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:	Sat, 21 Jan 2012 10:32:09 -0800
From:	Keith Packard <keithp@...thp.com>
To:	Chris Diamand <chris.diamand@...il.com>,
	linux-kernel@...r.kernel.org
Cc:	tiwai@...e.de, melchior.franz@...il.com
Subject: Re: i915/kms/backlight-combo mode problem (was: Re: Linux 2.6.39-rc5)

On Sat, 21 Jan 2012 17:09:25 +0000, Chris Diamand <chris.diamand@...il.com> wrote:
> Hi,
> 
> I have the problem described in the thread here 
> (https://lkml.org/lkml/2011/4/29/217) where the backlight
> values are reversed so although stuff is displayed on the screen, the 
> backlight is off. This occurs with kernels from about 2.6.39 onwards.
> 
> After booting a "broken" kernel (>~2.6.39), the backlight can be lit 
> temporarily with:
> setpci -s 00:02.0 F4.B=0
> and then turned off with:
> setpci -s 00:02.0 F4.B=FF

Can you experiment with other values? I'd like to know if it's
completely reversed, or if there's some other interaction here.

Testing 0x80, 0x7f, 0x01, 0xfe would all be interesting to me.

I suspect that writes to the LPBC value are being trapped by SMI and
'doing things' underneath.

> Yesterday I tried this patch, https://lkml.org/lkml/2011/4/30/37 by 
> manually applying it to the 3.2.1
> kernel source - this fixes the problem.

Can you still turn the backlight off with this patch in place?

> ** Unfortunately the LKML thread stops after this patch and given that 
> the problem is still present in the latest source, this code wasn't 
> pushed into the kernel. Is there a reason for this or was it just 
> forgotten about? What can I do about this? I am happy to make a git 
> patch with the latest source if this will help.

The patch would completely bypass LPBC-based brightness controls if the
backlight happened to be off when the driver started, which isn't
exactly what you want.

-- 
keith.packard@...el.com

Content of type "application/pgp-signature" skipped