| View previous topic :: View next topic |
| Author |
Message |
LukasGraf
n00b
Joined: 23 Jun 2006
Posts: 8
Location: Bern, Switzerland
|
 Posted: Wed Jul 05, 2006 3:44 pm Post subject: radeon fails to initialize AGP,radeon_cp_init error [SOLVED] |
 |
|
Hi everybody!
I'm trying to get a ATI Radeon 9600 to work with the opensource radeon driver from X.org, but the driver fails to initialize AGP, and in dmesg I get a "radeon_cp_init called without lock held" error every time I start X. Below are the details about my setup, and what I already did and tried. Let's start with the errors I get:
After I run startx, this shows up in /var/log/Xorg.0.log:
| Code: | (WW) RADEON(0): [agp] AGP not available
(EE) RADEON(0): [agp] AGP failed to initialize. Disabling the DRI.
(II) RADEON(0): [agp] You may want to make sure the agpgart kernel module is loaded before the radeon kernel module.
|
But agpgart is loaded, before radeon. So I checked dmesg:
| Code: | [drm:radeon_cp_init] *ERROR* radeon_cp_init called without lock held, held 0 owner 00000000 dea4d800
[drm:drm_unlock] *ERROR* Process 1974 using kernel context 0
|
These two error messages are generated once every time I start X, reproducibly. Until I attempt to start X, dmesg doesn't show anything suspicous.
Setup:
Motherboard: ASUS K8V Deluxe (Host Bridge: VIA VT8385 [K8T800 AGP])
VGA: ATI Radeon 9600 (GPU: RV350)
| Code: | # lspci -v | egrep "AGP|VGA"
00:00.0 Host bridge: VIA Technologies, Inc. VT8385 [K8T800 AGP] Host Bridge (rev 01)
Capabilities: [80] AGP version 3.5
01:00.0 VGA compatible controller: ATI Technologies Inc RV350 AP [Radeon 9600] (prog-if 00 [VGA])
Capabilities: [58] AGP version 3.0
|
BIOS Settings:
| Code: | Search for MDA Resources [no]
AGP Mode [AGP 4X]
AGP Fast Write [disabled]
Graphics Aperture Size [64MB]
|
Kernel: 2.6.16-gentoo-r9
Settings:
| Code: | Device Drivers --->
Character devices --->
<M> /dev/agpgart (AGP Support)
<M> VIA chipset support
< > Direct Rendering Manager (XFree86 4.1.0 and higher DRI support)
Power management options (ACPI, APM) --->
--- Power Management support
[*] Legacy Power Management API
|
Modules:
Instead of the kernel modules I use x11-base/x11-drm to provide DRM. I set VIDEO_CARDS="radeon vesa" in make.conf:
(I also tried the kernel modules earlier, same errors)
| Code: | # emerge -pv x11-drm
[ebuild R ] x11-base/x11-drm-20060608 VIDEO_CARDS="radeon -i810 -mach64 -mga -nv -r128 -savage -sis -tdfx -via" 0 kB
|
(x11-base/x11-drm20050502 didn't compile for me, with a non-trivial error, 4.3.0-r7 doesn't support 2.6 kernels, so I had to use a ~x86 version (on an otherwise stable x86 system))
All modules are loaded in the correct order and with no error messages whatsoever:
| Code: | # grep -v '#' /etc/modules.autoload.d/kernel-2.6
agpgart
via-agp
drm
radeon
|
| Code: | # lsmod | egrep "radeon|agp"
radeon 115296 0
drm 74968 1 radeon
via_agp 8320 0
agpgart 29644 2 drm,via_agp
|
| Code: | # # (After booting the system, before starting X:
# dmesg | egrep -i "error|fail|warn|unable|cannot|fatal"
#
|
X.org: x11-base/xorg-x11-7.0-r
| Code: | # emerge -pv xorg-server xorg-x11
[ebuild R ] x11-base/xorg-server-1.0.2-r7 USE="dri -debug -ipv6 -minimal -xprint" 0 kB
[ebuild R ] x11-base/xorg-x11-7.0-r1 USE="-3dfx" INPUT_DEVICES="keyboard mouse [-*] " VIDEO_CARDS="radeon vesa [-*]" 0 kB
|
/etc/X11/xorg.conf: (relevant parts only)
| Code: | Section "Module"
:
Load "glx" # libglx.a
Load "dri" # libdri.a
Load "drm" # libdrm.a
:
EndSection
Section "dri"
Mode 0666
EndSection
Section "Device"
Identifier "ATI Radeon 9600"
Driver "radeon"
Option "AGPMode" "4" # As set in BIOS
Option "AGPFastWrite" "false" # As set in BIOS
Option "UseInternalAGPGART" "no"
#Option "RenderAccel" "on" # Commented that out to lower complexity for debugging
BusID "PCI:1:0:0" # As reported by lspci
Screen 0
EndSection
Section "Screen"
Identifier "Screen0"
Device "ATI Radeon 9600"
:
EndSection |
This is about it. I'm not 100% sure the radeon_cp_init error actually leads to AGP failing, but I pretty much think so. At least, these errors sound very nasty  Also, the DRI Troubleshooting Guide says, the message | Code: | | [drm:radeon_cp_init] *ERROR* radeon_cp_init called without lock held |
is due to missing agpgart and specific AGP chipset modules. But I compiled and loaded both agpgart and via-agp modules.
What makes me suspicous, is that the help in menuconfig states that AGP_VIA provides support for "VIA MVP3/Apollo Pro chipsets". But I have neither of those, I have got a VIA K8T800 chipset (VT8385 northbridge). So is the list of supported chipsets for AGP_VIA just not complete, or is my chipset actually not supported? But if so, I should see messages like | Code: | agpgart: Unsupported Via chipset (device id: 0282), you might want to try agp_try_unsupported=1.
agpgart: no supported devices found. |
in dmesg, which I don't.
I'm confused, clueless and tired, please help!
Last edited by LukasGraf on Thu Jul 06, 2006 1:32 am; edited 2 times in total |
|
| Back to top |
|
 |
whiskas
n00b
Joined: 11 Sep 2003
Posts: 58
Location: Bucharest, Romania
|
| Posted: Wed Jul 05, 2006 4:06 pm Post subject: |
|
|
Well, it's obvious: the AGP kernel module has no support for your particular chipset, and DRM / DRI is not avaiable without AGP support.
You might want to try a 2.6.17-based kernel (or perhaps the -mm branch), I'm pretty sure they added support for newer VIA northbridges. |
|
| Back to top |
|
|
LukasGraf
n00b
Joined: 23 Jun 2006
Posts: 8
Location: Bern, Switzerland
|
| Posted: Wed Jul 05, 2006 4:29 pm Post subject: |
|
|
Which one do you mean by "AGP kernel module"? The generic agpgart module, or the specifc via-agp module?
What you say makes perfect sense for me, but then, why don't I get a "agpgart: Unsupported Via chipset..." message in dmesg? As far as I can tell, agpgart loads fine, without complaining...
Hmm, unstable kernel or -mm sources is not really an option for me, I'd like to run a stable system. Maybe I can find a kernel patch for my VIA nothbridge.
Anyway, thanks for your reply! |
|
| Back to top |
|
|
whiskas
n00b
Joined: 11 Sep 2003
Posts: 58
Location: Bucharest, Romania
|
| Posted: Wed Jul 05, 2006 4:38 pm Post subject: |
|
|
Ok, I completely misread you, for which I am very sorry 
Now, back to the problem at hand, I would try the following steps, in order:- reemerge x11-base/x11-drm;
- do not autoload any of the related kernel modules, instead, load only via-agp by hand, then startx as usual;
- try a newer kernel (kernel.org says that the latest stable version of the Linux kernel is 2.6.17.3).
And that's about all I can think of right now...
Edit: it seems that the newest kernel release has quite a few fixes for AGP and RADEON / DRM, so you might just want to give it a shot (including this one). |
|
| Back to top |
|
|
gami
Apprentice
Joined: 02 Jun 2006
Posts: 297
|
| Posted: Wed Jul 05, 2006 4:50 pm Post subject: |
|
|
| Can you please post the output of |
|
| Back to top |
|
|
LukasGraf
n00b
Joined: 23 Jun 2006
Posts: 8
Location: Bern, Switzerland
|
| Posted: Wed Jul 05, 2006 4:54 pm Post subject: |
|
|
| Code: | # lspci -n
00:00.0 0600: 1106:3188 (rev 01)
00:01.0 0604: 1106:b188
00:07.0 0c00: 1106:3044 (rev 46)
00:0a.0 0200: 11ab:4320 (rev 13)
00:0c.0 0200: 10b7:9200 (rev 78)
00:0f.0 0104: 1106:3149 (rev 80)
00:0f.1 0101: 1106:0571 (rev 06)
00:10.0 0c03: 1106:3038 (rev 81)
00:10.1 0c03: 1106:3038 (rev 81)
00:10.2 0c03: 1106:3038 (rev 81)
00:10.3 0c03: 1106:3038 (rev 81)
00:10.4 0c03: 1106:3104 (rev 86)
00:11.0 0601: 1106:3227
00:11.5 0401: 1106:3059 (rev 60)
00:18.0 0600: 1022:1100
00:18.1 0600: 1022:1101
00:18.2 0600: 1022:1102
00:18.3 0600: 1022:1103
01:00.0 0300: 1002:4150
01:00.1 0380: 1002:4170 |
|
|
| Back to top |
|
|
LukasGraf
n00b
Joined: 23 Jun 2006
Posts: 8
Location: Bern, Switzerland
|
| Posted: Wed Jul 05, 2006 5:05 pm Post subject: |
|
|
| whiskas wrote: | | Ok, I completely misread you, for which I am very sorry |
No problem, I almost thought so
| Quote: | Now, back to the problem at hand, I would try the following steps, in order:- reemerge x11-base/x11-drm;
- do not autoload any of the related kernel modules, instead, load only via-agp by hand, then startx as usual;
- try a newer kernel (kernel.org says that the latest stable version of the Linux kernel is 2.6.17.3).
And that's about all I can think of right now... |
* Reemerged x11-drm
* Loaded modules by hand: First agpgart, then via-agp, drm and at last radeon
Didn't change anything, the error messages in dmesg and Xorg.0.log stay exactly the same.
* Loaded only via-agp by hand (which pulls in agpgart): Still the same result.
New kernel: I'll give it a shot, but I really don't like the idea of running a production system with an unstable kernel. The lastest stable kernel from upstream (kernel.org) isn't necessarily considered stable on gentoo yet. You're right, a lot in the are of AGP / Radeon seems to have changed with new kernels, I'll try a 2.6.17.x, and if it's only to proove that the problem is lying there..
Last edited by LukasGraf on Wed Jul 05, 2006 5:15 pm; edited 1 time in total |
|
| Back to top |
|
|
gami
Apprentice
Joined: 02 Jun 2006
Posts: 297
|
| Posted: Wed Jul 05, 2006 5:07 pm Post subject: |
|
|
| LukasGraf wrote: | | Code: | # lspci -n
00:00.0 0600: 1106:3188 (rev 01)
[...]
|
|
This is the entry the via-agp driver looks for. It is defined in linux/pci_ids.h as PCI_DEVICE_ID_VIA_8385_0. In the 2.6.16 series this device is not recognized by via-agp. I have had a similar problem with a different VIA chipset in the past. See this topic for the patch I needed.
You can try to create a similar patch for your chipset. Adding a new agp_device_ids entry to via_agp_device_ids and an ID entry to the agp_via_pci_table in via-agp.c should do the trick.
Once you have the patch working you might want to submit it to the agpgart maintainer Dave Jones. |
|
| Back to top |
|
|
LukasGraf
n00b
Joined: 23 Jun 2006
Posts: 8
Location: Bern, Switzerland
|
| Posted: Wed Jul 05, 2006 5:20 pm Post subject: |
|
|
Ahh, this sounds very promising, thank you very much!
I will take some time to read all the stuff, and try to come up with a patch that works. In case of sucess or failure, I'll report here |
|
| Back to top |
|
|
LukasGraf
n00b
Joined: 23 Jun 2006
Posts: 8
Location: Bern, Switzerland
|
| Posted: Wed Jul 05, 2006 10:36 pm Post subject: |
|
|
Ok, let's call it partial success
The good thing is, my AGP bridge seems to be recognized by via-agp. The bad thing is, my system locks up when trying to start X... 
This is the patch I applied to linux/drivers/char/agp/via-agp.c:
| Code: | --- /usr/src/linux-2.6.16-gentoo-r9/drivers/char/agp/via-agp.c.orig 2006-07-05 22:38:08.000000000 +0200
+++ /usr/src/linux-2.6.16-gentoo-r9/drivers/char/agp/via-agp.c 2006-07-05 23:42:49.000000000 +0200
@@ -285,6 +285,14 @@
.chipset_name = "KT266/KY266x/KT333",
},
+
+ /* VT8385 */
+ {
+ .device_id = PCI_DEVICE_ID_VIA_8385_0,
+ .chipset_name = "K8T800",
+ },
+
+
/* VT8633 (for CuMine/ Celeron) */
{
.device_id = PCI_DEVICE_ID_VIA_8653_0,
@@ -501,6 +509,7 @@
ID(PCI_DEVICE_ID_VIA_8363_0),
ID(PCI_DEVICE_ID_VIA_8753_0),
ID(PCI_DEVICE_ID_VIA_8367_0),
+ ID(PCI_DEVICE_ID_VIA_8385_0),
ID(PCI_DEVICE_ID_VIA_8653_0),
ID(PCI_DEVICE_ID_VIA_XM266),
ID(PCI_DEVICE_ID_VIA_862X_0), |
When loading the patched via-agp module, dmesg now gives me
| Code: | Linux agpgart interface v0.101 (c) Dave Jones
agpgart: Detected VIA K8T800 chipset
agpgart: AGP aperture is 64M @ 0xf0000000 |
instead of just the first line. Yay! 
K8T800 is probably the .chipset_name I chose, but the apterture size is definitely correct. It is set to 64MB, and when I change it in the BIOS settings to 128MB, it shows up here. Nice 
However, starting X results in the system locking up completely (the power button is also dead). But hey, that means we're now at least accessing hardware, right?
Xorg.0.log contains this: (From the beginning of the DRM stuff to EOF)
| Code: | [drm] Initialized drm 1.0.1 20051102
ACPI: PCI Interrupt 0000:01:00.0[A] -> GSI 16 (level, low) -> IRQ 16
[drm] Initialized radeon 1.25.0 20060524 on minor 0:
[drm] Used old pci detect: framebuffer loaded
(II) RADEON(0): [drm] DRM interface version 1.2
(II) RADEON(0): [drm] created "radeon" driver at busid "pci:0000:01:00.0"
(II) RADEON(0): [drm] added 8192 byte SAREA at 0xe1887000
(II) RADEON(0): [drm] mapped SAREA 0xe1887000 to 0xb7925000
(II) RADEON(0): [drm] framebuffer handle = 0xd0000000
(II) RADEON(0): [drm] added 1 reserved context for kernel
(II) RADEON(0): [agp] Mode 0x1f000a09 [AGP 0x1106/0x3188; Card 0x1002/0x4150]
(II) RADEON(0): [agp] 8192 kB allocated with handle 0x00000001
(II) RADEON(0): [agp] ring handle = 0xf0000000
(II) RADEON(0): [agp] Ring mapped at 0xa76e9000
(II) RADEON(0): [agp] ring read ptr handle = 0xf0101000
(II) RADEON(0): [agp] Ring read ptr mapped at 0xa76e8000
(II) RADEON(0): [agp] vertex/indirect buffers handle = 0xf0102000
(II) RADEON(0): [agp] Vertex/indirect buffers mapped at 0xa74e8000
(II) RADEON(0): [agp] GART texture map handle = 0xf0302000
(II) RADEON(0): [agp] GART Texture map mapped at 0xa7008000
(II) RADEON(0): [drm] register handle = 0xfd200000
(II) RADEON(0): [dri] Visual configs initialized
(**) RADEON(0): DRI New memory map param
(**) RADEON(0): RADEONInitMemoryMap() :
(**) RADEON(0): mem_size : 0x10000000
(**) RADEON(0): agp_size : 0x000006b8
(**) RADEON(0): agp_base : 0xb7e52a20
(**) RADEON(0): MC_FB_LOCATION : 0xdfffd000
(**) RADEON(0): MC_AGP_LOCATION : 0xffffffc0
(**) RADEON(0): RADEONModeInit()
1280x1024 135.00 1280 1296 1440 1688 1024 1025 1028 1066 (24,32) +H +V
1280x1024 135.00 1280 1296 1440 1688 1024 1025 1028 1066 (24,32) +H +V
(**) RADEON(0): Pitch = 10485920 bytes (virtualX = 1280, displayWidth = 1280)
(**) RADEON(0): dc=13500, of=27000, fd=120, pd=2
(**) RADEON(0): RADEONInit returns 0x821b770
(**) RADEON(0): RADEONRestoreMode()
(**) RADEON(0): RADEONRestoreMode(0x821b770)
(**) RADEON(0): RADEONRestoreMemMapRegisters() :
(**) RADEON(0): MC_FB_LOCATION : 0xdfffd000
(**) RADEON(0): MC_AGP_LOCATION : 0xffffffc0
(**) RADEON(0): Map Changed ! Applying ... |
Looks quite ok to me so far, it seems that the problem occurs while applying the map. But... which map? To what?
In order to get X up again, I tried to unload via-agp:
Oops:
| Code: | Oops: 0000 [#1]
PREEMPT SMP
Modules linked in: radeon drm via_agp agpgart snd_pcm_oss snd_mixer_oss snd_seq_oss snd_seq_midi_event snd_seq snd_via82xx snd_mpu401_uart snd_rawmidi snd_seq_device snd_intel8x0 snd_ac97_codec snd_ac97_bus snd_pcm snd_timer snd snd_page_alloc
CPU: 0
EIP: 0060:[<e18f04c2>] Not tainted VLI
EFLAGS: 00010246 (2.6.16-gentoo-r9 #5)
EIP is at drm_lastclose+0x82/0x400 [drm]
eax: 00000000 ebx: fffffff4 ecx: 00000000 edx: df3ae8dc
esi: 00000001 edi: e1956780 ebp: df3ae800 esp: d8cfdf04
ds: 007b es: 007b ss: 0068
Process rmmod (pid: 10304, threadinfo=d8cfc000 task=c1645580)
Stack: <0>df3ae95c 00000000 00000880 df3ae800 00000001 e1956780 00000880 e18f7a35
df3ae800 c025736e dfe9b0e8 df3ae800 e18f7c3e df3ae800 d8cfdf44 e1956f00
00000000 00000000 e194e86f e1956780 c013ba4c e1956f00 00000880 d8cfdf64
Call Trace:
[<e18f7a35>] drm_cleanup+0x25/0x1c0 [drm]
[<c025736e>] kobject_put+0x1e/0x30
[<e18f7c3e>] drm_exit+0x6e/0xc0 [drm]
[<e194e86f>] radeon_exit+0xf/0x172 [radeon]
[<c013ba4c>] sys_delete_module+0x18c/0x1c0
[<c015714f>] do_munmap+0x11f/0x180
[<c0157201>] sys_munmap+0x51/0x80
[<c0103105>] syscall_call+0x7/0xb
Code: 09 00 00 8d 85 5c 01 00 00 89 04 24 e8 b8 98 83 de 8b 55 00 85 d2 0f 85 1b 02 00 00 8b 85 dc 00 00 00 8d 95 dc 00 00 00 8d 58 f4 <8b> 7b 0c 89 54 24 08 83 ef 0c 39 d0 74 46 8d b5 cc 00 00 00 89 |
Doesn't tell me much. drm_lastclose() is defined in linux-2.6.16-gentoo-r9/drivers/char/drm/drm_drv.c, but I have no idea what is going wrong here.
Thanks a lot for the help so far, I still think this is progress |
|
| Back to top |
|
|
LukasGraf
n00b
Joined: 23 Jun 2006
Posts: 8
Location: Bern, Switzerland
|
| Posted: Thu Jul 06, 2006 1:31 am Post subject: |
|
|
[SOLVED]
Phew! After my own attempt of a patch to via-agp.c did not work out too well, I thought I'd look if there maybe is already some working being done on this in the 2.6.17 tree. But I found nothing, so I googled if there is any use of the PCI_DEVICE_ID_VIA_8385_0 somewhere at all. And yes there is, in the amd64-agp module. This module seems to use the on-CPU northbridge of AMD64 processors for GART. Which is fine for me, I've got an Athlon64 and a supported AGP bridge (which is still needed).
So, although there is a VIA chipset present, I don't want to use it, so I disable it, and use the amd64-agp module (not the AMD irongate stuff):
| Code: | <M> /dev/agpgart (AGP Support)
< > AMD Irongate, 761, and 762 chipset support
<M> AMD Opteron/Athlon64 on-CPU GART support
< > VIA chipset support
|
Update /etc/modules.autoload.d/kernel-2.6:
| Code: | agpgart
amd64-agp
drm
radeon
|
Now, the addional kernel messages from dmesg look reeeally friendly :
| Code: | agpgart: Found an AGP 3.5 compliant device at 0000:00:00.0.
agpgart: Putting AGP V3 device at 0000:00:00.0 into 4x mode
agpgart: Putting AGP V3 device at 0000:01:00.0 into 4x mode
[drm] Setting GART location based on new memory map
[drm] Loading R300 Microcode
[drm] writeback test succeeded in 1 usecs
|
Also Xorg.0.log looks promising:
| Code: | (II) RADEON(0): [drm] DRM interface version 1.2
(II) RADEON(0): [drm] created "radeon" driver at busid "pci:0000:01:00.0"
(II) RADEON(0): [drm] added 8192 byte SAREA at 0xe1887000
(II) RADEON(0): [drm] mapped SAREA 0xe1887000 to 0xb78c6000
(II) RADEON(0): [drm] framebuffer handle = 0xd0000000
(II) RADEON(0): [drm] added 1 reserved context for kernel
(II) RADEON(0): [agp] Mode 0x1f000a09 [AGP 0x1106/0x3188; Card 0x1002/0x4150]
(II) RADEON(0): [agp] 8192 kB allocated with handle 0x00000001
(II) RADEON(0): [agp] ring handle = 0xf0000000
(II) RADEON(0): [agp] Ring mapped at 0xa768a000
(II) RADEON(0): [agp] ring read ptr handle = 0xf0101000
(II) RADEON(0): [agp] Ring read ptr mapped at 0xb78c5000
(II) RADEON(0): [agp] vertex/indirect buffers handle = 0xf0102000
(II) RADEON(0): [agp] Vertex/indirect buffers mapped at 0xa748a000
(II) RADEON(0): [agp] GART texture map handle = 0xf0302000
(II) RADEON(0): [agp] GART Texture map mapped at 0xa6faa000
(II) RADEON(0): [drm] register handle = 0xfd200000
(II) RADEON(0): [dri] Visual configs initialized
...
(**) RADEON(0): Map Changed ! Applying ...
(**) RADEON(0): Map applied, resetting engine ...
...
(II) RADEON(0): [drm] installed DRM signal handler
(II) RADEON(0): [DRI] installation complete
...
(II) RADEON(0): Direct rendering enabled
|
Sounds good. But when I check with glxinfo, also Xorg says "enabled", I get "direct rendering: no":
| Code: | # glxinfo | grep "direct rendering"
direct rendering: No
|
So I enabled verbose debugging output for libGLand ran glxinfo again:
| Code: | # export LIBGL_VERBOSE="" LIBGL_DEBUG=""
# glxinfo
name of display: :0.0
libGL error: dlopen /usr/lib//modules/dri/r300_dri.so failed (/usr/lib//modules/dri/r300_dri.so: cannot open shared object file: No such file or directory)
...
|
Hmm.. r300_dri.so exists, but in /usr/lib/xorg... So I just symlinked it, and that seems to do the trick:
| Code: |
euklid ~ # ln /usr/lib/xorg/modules/dri/r300_dri.so /usr/lib/modules/dri/r300_dri.so
euklid ~ # glxinfo | grep "direct rendering"
direct rendering: Yes
|
glxgears immediately shows the effect of the change: From ~150fps before to ~1200fps after
Thanks for the help guys! |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Kernel & Hardware
