Failure with ATI Radeon 9600

Failure with ATI Radeon 9600

[Jacob Gorm Hansen]

hi,

I just purchased an ATI Radeon 9600 128MB for use with my Xen testbox. 
However, upon loading the intel-agp module, the following test fails:

in drivers/char/agp/backend.c:

     if (bridge->driver->create_gatt_table()) {
         printk(KERN_ERR PFX
             "unable to get memory for graphics translation table.\n");
         rc = -ENOMEM;
         goto err_out;
     }

The card runs fine (i.e. accellerated) with vanilla linux 2.6.10.

Any suggestions?

Jacob


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl

Re: Failure with ATI Radeon 9600

[Jacob Gorm Hansen]

Jacob Gorm Hansen wrote:
> hi,
> 
> I just purchased an ATI Radeon 9600 128MB for use with my Xen testbox. 
> However, upon loading the intel-agp module, the following test fails:
> 
> in drivers/char/agp/backend.c:
> 
>     if (bridge->driver->create_gatt_table()) {
>         printk(KERN_ERR PFX
>             "unable to get memory for graphics translation table.\n");
>         rc = -ENOMEM;
>         goto err_out;
>     }

I just checked, and the output of /proc/mtrr is the same on both 
configurations (0 - 1GB, even though dom0 was only given 256 megs of 
memory).

The chipset is an Intel 875.

Jacob


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl

Re: Failure with ATI Radeon 9600

[Jacob Gorm Hansen]

Jacob Gorm Hansen wrote:

> The chipset is an Intel 875.

It appears I had not compile the fglrx module with the correct symbols 
defined (it seems to like CONFIG_X86), so after some hacking this is 
what I have:


(from dmesg after loading intel-agp and fglrx):


Linux agpgart interface v0.100 (c) Dave Jones
agpgart: Detected an Intel i875 Chipset.
agpgart: Maximum main memory to use for agp memory: 440M
agpgart: unable to get memory for graphics translation table.
agpgart: agp_backend_initialize() failed.
agpgart-intel: probe of 0000:00:00.0 failed with error -12
fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies, 
Starnberg, GERMANY' taints kernel.
[fglrx] Maximum main memory to use for locked dma buffers: 430 MBytes.
PCI: Obtained IRQ 16 for device 0000:01:00.0
[fglrx] module loaded - fglrx 3.14.6 [Oct 30 2004] on minor 0


(from serial console after attempting to start X):


(XEN) (file=/mnt/hda3/home/jacobg/xeno-unstable.bk/xen/include/asm/mm.h, 
line=157) Error pfn 00001020: ed=fc599260, sd=fc5996a0, caf=80000002, 
taf=f0000001
Oops: 0000 [#1]
PREEMPT DEBUG_PAGEALLOC
Modules linked in: fglrx intel_agp agpgart
CPU:    0
EIP:    0061:[<e08b5083>]    Tainted: P      VLI
EFLAGS: 00213202   (2.6.10-xen0)
EIP is at vm_shm_nopage+0x73/0x100 [fglrx]
eax: 01f2a000   ebx: 00000000   ecx: 00000064   edx: 00000382
esi: dafb9250   edi: 00000000   ebp: e08b5010   esp: df2d1e60
ds: 0069   es: 0069   ss: 0069
Process X (pid: 6628, threadinfo=df2d0000 task=da286b20)
Stack: dafb9250 00000005 0000000c e08e1160 e08b47c8 dafb9250 da266b7c 
c014bb13
        dafb9250 b7d63000 df2d1ea8 00000001 00000000 000e0819 dafb9250 
de0f3f74
        00000000 de0f3f74 00000001 df2d0000 da266b7c dab40ddc b7d63000 
c014c056
Call Trace:
  [<e08b47c8>] __ke_malloc+0x18/0x20 [fglrx]
  [<c014bb13>] do_no_page+0xc3/0x3c0
  [<c014c056>] handle_mm_fault+0x126/0x250
  [<c011475c>] do_page_fault+0x41c/0x687
  [<c0111acd>] old_mmap+0xfd/0x110
  [<c0109eab>] page_fault+0x3b/0x40
Code: 34 24 29 c3 e8 6f 02 00 00 8d 0c 03 a1 fc 28 43 c0 89 ca c1 ea 16 
8b 04 90 a8 01 74 43 c1 e9 0a 25 00 f0 ff ff 81 e1 fc 0f 00 00 <8b> 84 
01 00 00 00 c0 a8 81 74 13 8b 15 d0 a2 51 c0 c1 e8 0c c1


Non-preemtible kernel gives a similar error.

Has anyone played with this before? I am using the latest xeno-unstable 
from bkbits.

Jacob


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl

Re: Failure with ATI Radeon 9600

[Jacob Gorm Hansen]

Jacob Gorm Hansen wrote:

> Linux agpgart interface v0.100 (c) Dave Jones
> agpgart: Detected an Intel i875 Chipset.
> agpgart: Maximum main memory to use for agp memory: 440M
> agpgart: unable to get memory for graphics translation table.
> agpgart: agp_backend_initialize() failed.
> agpgart-intel: probe of 0000:00:00.0 failed with error -12

The reason agp_generic_create_gatt_table() fails is that 
ioremap_nocache() always return NULL. (Whether or not the subsequent X 
crash has anything to do with this is not clear -- the ATI wrapper has 
quite a few references to virt_to_phys(), so likely some problems there 
as well.)

Is there a good reason for this / what will it take to implement 
ioremap_nocache correctly?

Jacob


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl

Re: Failure with ATI Radeon 9600

[Jacob Gorm Hansen]

Jacob Gorm Hansen wrote:
> The reason agp_generic_create_gatt_table() fails is that 
> ioremap_nocache() always return NULL.

Oops, not if I have physdev-access, please disregard.

Jacob


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl