1 |
On Tue, Jan 26, 2010 at 2:59 AM, Peter Humphrey |
2 |
<peter@××××××××××××××.org> wrote: |
3 |
> On Tuesday 26 January 2010 01:13:59 Mark Knecht wrote: |
4 |
> |
5 |
>> mtrr: type mismatch for c0000000,10000000 old: write-back new: |
6 |
>> write-combining [drm] MTRR allocation failed. Graphics performance may |
7 |
>> suffer. |
8 |
> |
9 |
> This rings a bell. Your kernel line in grub.conf has something like |
10 |
> "video=inteldrmfb:mtrr:3,ywrap", no? The 3 is causing the error above, or |
11 |
> you may have a 4. Google should be able to find you the docs on this. The |
12 |
> options are write-combining and write-back; if one doesn't work the other |
13 |
> should. |
14 |
> |
15 |
> -- |
16 |
> Rgds |
17 |
> Peter. |
18 |
> |
19 |
> |
20 |
|
21 |
Hi all, |
22 |
OK - it's working and gentoo-amd64 is the first email recipient |
23 |
from my new i5-661 running Gentoo Linux and XFCE4. THANKS TO EVERYONE! |
24 |
|
25 |
I'll need to document what I did to get it going as well as archive |
26 |
files so as to protect myself from some sort of infant mortality or |
27 |
operator error. None the less at least it's up and usable. |
28 |
|
29 |
This seems to be one of those cases where I cannot recognize |
30 |
exactly who had the final answer as there were a lot of things I did |
31 |
last night to get the machine going, both based on information from |
32 |
this thread as well as a thread at intel-gfx. Clearly I'm still |
33 |
learning/confused about the exact technology here so I want to iron |
34 |
that out over the next few days. |
35 |
|
36 |
The main points, subject to me getting schooled on what's really |
37 |
happening here: |
38 |
|
39 |
1) The new i5-661 processor and i915 running X are ONLY supported |
40 |
using KMS so there seem to be two ways to do this: |
41 |
a) Build the kernel with AGP, DRM and KMS into the kernel, or |
42 |
b) Build the kernel with AGP, DRM and KMS modular and then use |
43 |
i915.modeset=1 on the boot command line |
44 |
|
45 |
I am using b) at this time. |
46 |
|
47 |
2) Use drm.debug=0x06 to get lots of nice messages from DRM about |
48 |
what's going on. |
49 |
|
50 |
3) Configure the kernel to support frame buffers but turn off |
51 |
everything except these 4, and possibly only the first and last ones: |
52 |
(grep for "FB" and "FRAMEBUFFER") |
53 |
|
54 |
CONFIG_FB=y |
55 |
CONFIG_FB_CFB_FILLRECT=m |
56 |
CONFIG_FB_CFB_COPYAREA=m |
57 |
CONFIG_FB_CFB_IMAGEBLIT=m |
58 |
CONFIG_FRAMEBUFFER_CONSOLE=y |
59 |
|
60 |
The CFB entries are probably not necessary but I haven't figured out |
61 |
where they are in the kernel yet to turn them off. They do load |
62 |
modules so they might be required. |
63 |
|
64 |
4) In make.conf use these lines among others: |
65 |
|
66 |
CFLAGS="-O2 -march=native -pipe" |
67 |
USE="hal dts mmx sse sse2 ssse3 sse4 -gnome -kde" |
68 |
INPUT_DEVICES="evdev" |
69 |
VIDEO_CARDS="intel fbdev" |
70 |
|
71 |
to set up for X and get a normal text boot which I think is taking |
72 |
place in the frame buffer. |
73 |
|
74 |
5) As root run |
75 |
|
76 |
Xorg -configure |
77 |
Xorg -config /root/xorg.conf.new |
78 |
|
79 |
If things go right I'm to the point where Drake was correct. X is up |
80 |
but the screen is blank. Copy xorg.conf.new to /etc/X/xorg.conf |
81 |
|
82 |
X -retro does work. Thanks Drake! |
83 |
|
84 |
6) emerge xfce4-meta and then as user put "exec startxfce4" in |
85 |
.xinitrc, and then startx |
86 |
|
87 |
Assuming I haven't made any omissions or stupid mistakes that gets |
88 |
me into xfce4. I don't know if I'm running X over a frame buffer or |
89 |
using a more native VGA. |
90 |
|
91 |
I will of course review this in more depth as I document it for |
92 |
real but I wanted to say thanks for the answers as they were all |
93 |
helpful. |
94 |
|
95 |
Cheers, |
96 |
Mark |