Bug 3280

Created attachment 2670 [details]
/etc/X11/xorg.conf

xorg.conf for PowerPC Mac mini, ATI Radeon 9200, ViewSonic A50.

Created attachment 2671 [details]
/var/log/Xorg.0.log

Xorg.0.log for PowerPC Mac mini, ATI Radeon 9200, ViewSonic A50 CRT

Perhaps selinux targeted enforcing mode is getting in the way.

When I run "kudzu -p -b DDC" [as root], the output is
-----
Framebuffer ioctl failed. Exiting.
-
class: MONITOR
bus: DDC
detached: 0
driver: unknown
desc: "A50"
id: VSC455a
horizSyncMin: 30
horizSyncMax: 56
vertRefreshMin: 50
vertRefreshMax: 120
mode: 960x960
mode: 960x720
-----

and also /var/log/audit/audit.log has
-----
type=KERNEL msg=audit(1115993381.897:0): avc:  denied  { read } for  name=fb0
dev=tmpfs ino=1641 scontext=root:system_r:kudzu_t
tcontext=system_u:object_r:framebuf_device_t tclass=chr_file
-----

I can't seem to repeat this with pretty much the exact same hardware. Did
system-config-display work for you? (It did for me). Various differences I can
point out is the: 

* Option      "UseFBDev" "true"
* DefaultDepth     16 (as opposed to your default 24)

I too have SELinux enabled, and its just some regular el-cheapo 15" monitor. And
my kudzu output:
[root@mini ~]# kudzu -p -b DDC
Framebuffer ioctl failed. Exiting.
-
class: MONITOR
bus: DDC
detached: 0
driver: unknown
desc: "(null)"
id: ___9e56
horizSyncMin: 31
horizSyncMax: 69
vertRefreshMin: 48
vertRefreshMax: 120
mode: 960x960
mode: 960x720
mode: 960x720
mode: 1272x1017

Created attachment 2686 [details]
xorg.diff between mine and whats available in this bugzilla attachment

Just in case you wanted to apply it to see if it'll work. Its generated from
s-c-display fwiw

Created attachment 2756 [details]
xorg.conf for Samsung 730B; video switching works

Created attachment 2757 [details]
Xorg.0.log for Samsung 730B; video switching works

$ cat /proc/version
Linux version 2.6.11-1.1340_FC4 (bhcompile@sizzler.build.redhat.com) (gcc
version 4.0.0 20050519 (Red Hat 4.0.0-8)) #1 Sun May 22 03:19:02 EDT 2005
$ rpm -qa | grep xorg-x11
xorg-x11-6.8.2-31
$ 

# /usr/sbin/kudzu -p -b DDC
Framebuffer ioctl failed. Exiting.
-
class: MONITOR
bus: DDC
detached: 0
driver: unknown
desc: "SyncMaster"
id: SAM9101
horizSyncMin: 30
horizSyncMax: 81
vertRefreshMin: 56
vertRefreshMax: 75
mode: 1272x1017
mode: 760x608
mode: 880x660
# 


So, at this point the difference between working and not working is the kernel
version (and gcc version used to compile it), and a brand new monitor with
latest DDC; versus a 6-year old monitor with early DDC, and a kernel version
that is about one week behind.

This problem probably is caused in part (or made worse) by a hardware bug with
the Mac mini: bad circuit design/use by Apple, and/or bad ATI 9200 chip
(bug/power/thermal/noise/etc), and/or bad mechanical design (unreliable in
consumer use), etc.  Many independent users report that several DVI displays
don't work with Mac mini:
http://www.freewebs.com/themagius/macintoshmini.htm
http://discussions.info.apple.com/ > Mac mini > Displays > "Running tally of DVI
display reports"

My current state with VGA cable (both Samsung 730B LCD and ViewSonic A50 CRT) is
that Fedora Core 4 test 3 hangs in boot, with the Tux+VGA text console showing
"agpgart: aperture is 16M @0" and "Serial: ..." as the last two lines
[paraphrased].  Using DVI cable, the Samsung 730B LCD display detects no signal.

"This problem probably is caused in part (or made worse) by a hardware bug with
the Mac mini"

I'm not so sure, I have the same problem with a imac g4 (geforce 2 mx). It seems
to me that I started having this problem with the 6.8.x release, all pre-6.8.x
releases worked fine, as well as the xfree branch.

I also the error when I resume my monitor, if i have it suspended.

I have the same problem on linux + thinkpad + Samtron 71S  ( my screen is black
with "video mode not supported"). It also does not work when I am not anymore
with the Samtron 71S But if I boot without the SAMTRON then it is also a strange
console. 

My version is: 
Linux myhost 2.6.12-1.1387_FC4 #1 Fri Jul 1 12:41:40 EDT 2005 i686 i686 i386
GNU/Linux
with x version: 6.8.2 release 37 for Fedora Core 4

 my Xorg.conf contains two DISPLAY section:
  SubSection "Display"
                Viewport   0 0
                Depth     16
                Modes    "800x600" "640x480"
        EndSubSection
  SubSection "Display"
                Viewport   0 0
                Depth     24
                Modes    "1280x1024" "1152x864" "1024x768" "800x600" "640x480"
        EndSubSection
and /usr/sbin/kudzu -p -b DDC:

-
class: VIDEO
bus: DDC
detached: 0
driver: unknown
desc: "S3 Incorporated. VBE 2.0"
mem: 15168
-
class: MONITOR
bus: DDC
detached: 0
driver: unknown
desc: "S/T 71S"
id: STN0014
horizSyncMin: 30
horizSyncMax: 81
vertRefreshMin: 56
vertRefreshMax: 76
mode: 1280x1024
mode: 1024x768
mode: 800x600
mode: 640x480

List of DVI monitors vs. macmini:
http://www.freewebs.com/themagius/macintoshmini.htm

J.R.: Can you please summarize what works and what does not?  What are the
remaining problems?

(In reply to comment #9)
> J.R.: Can you please summarize what works and what does not?  What are the
> remaining problems?

In the 4 months since my last update: the original Samsung 730B DVI port was
found to be bad (did not work in direct A:B swap with working other DVI monitor
+ DVI_cable + EVI machine), the ViewSonic A50 CRT [VGA] died, and I have a new
Samsung 730B with working DVI port.  The new Samsung 730B works under Apple
Macintosh OS X 10.3.9 using DVI port, and also on x86_64 + ATI 9250 + FC4. 
However, Mac mini + Fedora Core 4 Linux + Xorg has gotten worse: X11 does not
start at all (not even after generating a new xorg.conf file); only text
terminal mode works.

I will attach xorg.conf, Xorg.0.log, and miscellaneous info (/proc/version, list
of Fedora Core 4 xorg*.rpm, dmesg | grep radeon, kudzu -p -b DDC).

Created attachment 3401 [details]
/proc/version, installed xorg*.rpm, dmesg | grep radeon, kudzu -p -b DDC

Created attachment 3402 [details]
xorg.conf for Samsung 730B; X11 does not start

Created attachment 3403 [details]
Xorg.0.log;  Screen(s) found, but none have a usable configuration

Apple Macintosh mini (radeon 9200)

Created attachment 3420 [details]
xorg.conf on working x86_64, ATI 9200PRO, DVI

Working configuration on similar system: same Samsung SyncMaster 730B monitor
using DVI port, x86_64, ATI9200PRO card.  Perhaps there are enough similarities
to be useful.

Created attachment 3421 [details]
Xorg.0.log on working x86_64, ATI9200PRO, DVI

Working configuration with same Samsun SyncMaster 730B on DVI port, ATI9200PRO
card, x86_64 machine.  Perhaps there are enough similarities to be useful.

During boot of the operating system, there is a console message:
   radeonfb (0000:00:10.0) Invalid ROM signature 0 should be0xaa55
which might be relevant.

Does this still happen with a current version of xf86-video-ati?

(In reply to comment #17)
> Does this still happen with a current version of xf86-video-ati?

Console switching between graphics and text virtual terminals works correctly as
expected (no problems) running Xorg 7.0.0 (21 December 2005),  "Driver ati" in
/etc/X11/xorg.conf, "ATI driver (version 6.5.7)" and "R128" in
/var/log/Xorg.0.log, .rpm of xorg-x11-server-Xorg-1.0.1-9.fc5 on Fedora Core 5
with kernel-2.6.16-1.2096_FC5.