lkml.org 
[lkml]   [2010]   [Jul]   [20]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
SubjectRe: Blank (disconnected) screen during boot with latest 2.6.35-rcX kernels
From
On Tue, Jul 20, 2010 at 12:47 PM, trapDoor <trapdoor6@gmail.com> wrote:
> Hi Alex,
>
>>
>> Probably the oem made two similar boards one with dual DVI and one
>> with DVI + VGA + HDMI and used the same connector table since the
>> encoder routing is still correct even if the connectors are named
>> wrong.  We can add a quirk to fix it.
>>
>
> I plugged my monitor into the other (white) DVI port and now system
> recognises it as DVI:
>
> [    18.418] (II) RADEON(0): Output HDMI-0 disconnected
> [    18.418] (II) RADEON(0): Output VGA-0 disconnected
> [    18.418] (II) RADEON(0): Output DIN disconnected
> [    18.418] (II) RADEON(0): Output DVI-0 connected
>
> As I mentioned before, I guess that on my card HDMI signal is
> supported only on the yellow DVI port (through DVI > HDMI adapter),
> and even if a device is connected to it directly the system will still
> recognise it as HDMI.
>
>>
>> So the patch fixes the issue?  I'll send it on then.  Looks like your
>> systems needs a connector table quirk.  Can you send me the pci ids
>> for your GPU (lspci -vnn) and a copy of your vbios?  To dump your
>> vbios (as root):
>> cd /sys/bus/pci/devices/<pci bus id>
>> echo 1 > rom
>> cat rom > /tmp/vbios.rom
>> echo 0 > rom
>>
>
> Yes, I confirm it does fix it.

Good.

> Full lspci output attached and below pasting what's related to my graphics card:
> --------------
> 04:00.0 VGA compatible controller [0300]: ATI Technologies Inc
> Mobility Radeon HD 3600 Series [1002:9598]
>        Subsystem: ASUSTeK Computer Inc. Device [1043:01e4]
>        Flags: bus master, fast devsel, latency 0, IRQ 43
>        Memory at d0000000 (64-bit, prefetchable) [size=256M]
>        Memory at fbfe0000 (64-bit, non-prefetchable) [size=64K]
>        I/O ports at e000 [size=256]
>        Expansion ROM at fbfc0000 [disabled] [size=128K]
>        Capabilities: <access denied>
>        Kernel driver in use: radeon
>
> 04:00.1 Audio device [0403]: ATI Technologies Inc RV635 Audio device
> [Radeon HD 3600 Series] [1002:aa20]
>        Subsystem: ASUSTeK Computer Inc. Device [1043:aa20]
>        Flags: bus master, fast devsel, latency 0, IRQ 45
>        Memory at fbffc000 (64-bit, non-prefetchable) [size=16K]
>        Capabilities: <access denied>
>        Kernel driver in use: HDA Intel
> --------------
>
> Vbios dump for bus ID 0000:04:00.0 attached.
>

The attached patch should fix the connectors to report properly.
There are a number of similar ASUS cards with the same quirk.

>>
>> As to the performance regressions, 2.6.35 added proper support for the
>> glx vblank stuff, so you are likely seeing apps synchronized to the
>> vrefresh.
>>
>
> I guess you were referring to the glxgears 'test tool' that used to
> show (before 2.6.35-rc) much higher number of fps than it does now
> when it corresponds to vertical-refresh like it's supposed to. I
> noticed the change and assumed it is correct. And I never considered
> glxgears as a proper performance test anyway so I wouldn't rely on the
> results it gives.
>
> Where I see a real 3D performance drop now is for example Armagetron
> Advanced - a game I use just for testing. I enabled certain features
> using 3D in it. Without going into details I just say that with kernel
> 2.6.34.1 the overall 3D performance in AA is about twice better. The
> game runs quite smooth, number of fps doesn't go under 60. With kernel
> 2.6.35-rc5-git3 on the same system, with the same drivers, mesa, drm
> etc., the game runs slower and not so smooth, certain details look
> slightly worse and I won't get more than 35 fps.
>
> Apart from that I don't notice any regressions on my desktop, with
> compiz effects etc. I'd run some other tests if you could recommend
> some tools, but not too complex and rather some specialised software
> just for testing purposes, not games.

If it's just the kernel that changed is there any chance you could
bisect what caused the performance regression?

Alex
From ebdffc386fc3d90f2185e95d9429248151939c6e Mon Sep 17 00:00:00 2001
From: Alex Deucher <alexdeucher@gmail.com>
Date: Tue, 20 Jul 2010 13:15:37 -0400
Subject: [PATCH] drm/radeon/kms: add quirk for ASUS HD 3600 board

Connector is actually DVI rather than HDMI.

Reported-by: trapDoor <trapdoor6@gmail.com>
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
---
drivers/gpu/drm/radeon/radeon_atombios.c | 9 +++++++++
1 files changed, 9 insertions(+), 0 deletions(-)

diff --git a/drivers/gpu/drm/radeon/radeon_atombios.c b/drivers/gpu/drm/radeon/radeon_atombios.c
index 3f89c9b..0a97aeb 100644
--- a/drivers/gpu/drm/radeon/radeon_atombios.c
+++ b/drivers/gpu/drm/radeon/radeon_atombios.c
@@ -280,6 +280,15 @@ static bool radeon_atom_apply_quirks(struct drm_device *dev,
}
}

+ /* ASUS HD 3600 board lists the DVI port as HDMI */
+ if ((dev->pdev->device == 0x9598) &&
+ (dev->pdev->subsystem_vendor == 0x1043) &&
+ (dev->pdev->subsystem_device == 0x01e4)) {
+ if (*connector_type == DRM_MODE_CONNECTOR_HDMIA) {
+ *connector_type = DRM_MODE_CONNECTOR_DVII;
+ }
+ }
+
/* ASUS HD 3450 board lists the DVI port as HDMI */
if ((dev->pdev->device == 0x95C5) &&
(dev->pdev->subsystem_vendor == 0x1043) &&
--
1.7.1.1
\
 
 \ /
  Last update: 2010-07-20 19:25    [W:0.101 / U:0.104 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site