EXTERNAL: Re: No Device Node and .ko files for xf86-video-ati

Chen, Simon simon.chen at lmco.com
Thu Jun 18 06:26:51 PDT 2015

Hi Michel,

Here's what I get from doing  dmesg | less. The mailing list does not seem to like how big my full dmesg result was, unless you'd like me to "grep" something?

[    0.000000] Initializing cgroup subsys cpuset
[    0.000000] Initializing cgroup subsys cpu
[    0.000000] Initializing cgroup subsys cpuacct
[    0.000000] Linux version 3.14.19-yocto-standard (dev3l at dev3l) (gcc version 4.9.1 (GCC) ) #2 SMP PREEMPT Wed Jun 17 09:29:44 EDT 2015
[    0.000000] Command line: BOOT_IMAGE=/vmlinuz LABEL=boot root=/dev/ram0
[    0.000000] e820: BIOS-provided physical RAM map:
[    0.000000] BIOS-e820: [mem 0x0000000000000000-0x0000000000057fff] usable
[    0.000000] BIOS-e820: [mem 0x0000000000058000-0x0000000000058fff] reserved
[    0.000000] BIOS-e820: [mem 0x0000000000059000-0x000000000009efff] usable
[    0.000000] BIOS-e820: [mem 0x000000000009f000-0x000000000009ffff] reserved
[    0.000000] BIOS-e820: [mem 0x0000000000100000-0x0000000099098fff] usable
[    0.000000] BIOS-e820: [mem 0x0000000099099000-0x00000000990a9fff] ACPI data
[    0.000000] BIOS-e820: [mem 0x00000000990aa000-0x00000000a5786fff] usable
[    0.000000] BIOS-e820: [mem 0x00000000a5787000-0x00000000a5aaefff] reserved
[    0.000000] BIOS-e820: [mem 0x00000000a5aaf000-0x00000000a5faefff] ACPI NVS
[    0.000000] BIOS-e820: [mem 0x00000000a5faf000-0x00000000a5fe9fff] ACPI data
[    0.000000] BIOS-e820: [mem 0x00000000a5fea000-0x00000000a5ffffff] usable
[    0.000000] BIOS-e820: [mem 0x00000000e00f8000-0x00000000e00f8fff] reserved
[    0.000000] BIOS-e820: [mem 0x00000000fed1c000-0x00000000fed1ffff] reserved
[    0.000000] BIOS-e820: [mem 0x00000000ffc00000-0x00000000ffffffff] reserved
[    0.000000] BIOS-e820: [mem 0x0000000100000000-0x000000014dffffff] usable
[    0.000000] NX (Execute Disable) protection: active

That is a very fascinating point. It appears that the .la and .so for the xorg driver shows up just fine in the /usr portion of the file system. It seems that the xorg driver is doing what its supposed to do. Perhaps the problem lies in a missing kernel radeon driver. I enabled the ATI Radeon kernel configuration flag, but it seems my embedded OS is missing a lot of .ko's not just the radeon.ko.

The embedded OS I'm using by the way is Yocto (poky-dizzy-12.0.1), a flavor of OpenEmbedded. So far I haven't gotten a response from anyone in their mailing list.


-----Original Message-----
From: Michel Dänzer [mailto:michel at daenzer.net] 
Sent: Wednesday, June 17, 2015 9:35 PM
To: Chen, Simon
Cc: xorg-driver-ati at lists.x.org
Subject: EXTERNAL: Re: No Device Node and .ko files for xf86-video-ati

On 18.06.2015 00:12, Chen, Simon wrote:
> I am trying to integrate the xf86-video-ati 7.4.0 for an external AMD 
> Radeon E8860 onto an embedded system. I have enabled the kernel 
> configuration flag for ATI Radeon, and I have installed the 
> xf86-video-ati successfully without any dependency errors. However, 
> the embedded OS does not seem to use the graphics card upon start 
> time. In the Xorg.0.log file that I’ve attached, the OS was able to 
> load out a .so for the driver. However, it later complains about a 
> missing
> dev/dri/card0 device node. After doing some research into how this 
> device node gets made, it seems udev or atleast some kind of 
> init-script is responsible for executing a modprobe in the Linux 
> filesystem to search for driver .ko files. It extracts a device ID 
> from this driver .ko and tries to match it with the device ID’s from 
> the external hardware detected on the board. So far it seems the OS is 
> capable of detecting the graphics card, and this is evident when I do “$ lspci –k.”
> After matching the device IDs, a device node is supposed to be created 
> and associated with the driver and hardware. The OS however is missing 
> the appropriate .ko’s and the MODALIAS device ID.
> My question is...is this process executed anywhere in the 
> xf86-video-ati installation package? I’ve read that drivers can be 
> implemented either as a kernel built-in or as kernel modules. In the 
> case with the xf86-video-ati driver however, I am a bit confused as to 
> which category it falls under. The necessity for a device node and a 
> .ko file seems to presume that the xf86-video-ati is supposed to be 
> modularized, and yet the installation package has no script that sets these up.

The xf86-video-ati radeon driver is a userspace driver for Xorg. It depends on the kernel radeon driver (and the corresponding
/dev/dri/card* device node) having been set up correctly.

Please provide the output of dmesg.

Earthling Michel Dänzer               |               http://www.amd.com
Libre software enthusiast             |             Mesa and X developer

More information about the xorg-driver-ati mailing list