Bug 23990

Summary: kms drm-next commit 4aac047323e3082d0866b8ad3784236632105af4 causes ring test fail with agpmode=-1
Product: xorg Reporter: Andy Furniss <adf.lists>
Component: Driver/RadeonAssignee: xf86-video-ati maintainers <xorg-driver-ati>
Status: RESOLVED FIXED QA Contact: Xorg Project Team <xorg-team>
Severity: normal    
Priority: medium    
Version: git   
Hardware: x86 (IA32)   
OS: Linux (All)   
Whiteboard:
i915 platform: i915 features:
Attachments:
Description Flags
lspci -vnnxx
none
dmesg failed with agpmode=-1
none
dmesg OK with revert
none
dmesg normal agp fail + oops
none
Patch to avoid r600 getting rv370 gart functions. none

Description Andy Furniss 2009-09-17 04:49:30 UTC
Created attachment 29619 [details]
lspci -vnnxx

Sapphire Radeon HD3850 AGP card on nforce2 mobo.

Been testing kms and until 4aac047323e3082d0866b8ad3784236632105af4 it worked OK (barring nits) for me doing modprobe radeon modeset=1 agpmode=-1. Current git also works with that reverted.

Without revert I get a ring test fail with agpmode=-1.

It has always failed ringtest and oops without agpmode=-1.

Attached are 

lspci -vnnxx

dmesg failing with current git and agpmode=-1

dmesg working current with revert.

dmesg showing agp fail + oops
Comment 1 Andy Furniss 2009-09-17 04:51:17 UTC
Created attachment 29620 [details]
dmesg failed with agpmode=-1
Comment 2 Andy Furniss 2009-09-17 04:53:00 UTC
Created attachment 29621 [details]
dmesg OK with revert
Comment 3 Andy Furniss 2009-09-17 04:55:03 UTC
Created attachment 29622 [details]
dmesg normal agp fail + oops
Comment 4 Andy Furniss 2009-09-22 13:50:22 UTC
Created attachment 29774 [details] [review]
Patch to avoid r600 getting rv370 gart functions.
Comment 5 Andy Furniss 2009-09-22 13:52:41 UTC
(In reply to comment #0)

> Been testing kms and until 4aac047323e3082d0866b8ad3784236632105af4 it worked
> OK (barring nits) for me doing modprobe radeon modeset=1 agpmode=-1.

Looking at the diff I could see that R600 using agpmode=-1 was now getting rv370 functions set.

Just stopping that with the patch attached gets me a working again.

Is the correct thing to do set everything up like < R600 & handling RV770 as well? 
Comment 6 Andy Furniss 2009-09-22 17:05:18 UTC
(In reply to comment #3)
> Created an attachment (id=29622) [details]
> dmesg normal agp fail + oops
> 

With more recent drm-next using AGP I no longer get an oops, ring test fails, but it carries on and (I think) sets up in PCIE mode, which works OK.

Comment 7 Andy Furniss 2009-09-22 17:08:43 UTC
(In reply to comment #5)
> (In reply to comment #0)

> Is the correct thing to do set everything up like < R600 & handling RV770 as
> well? 

I also notice that with agpmode=-1 the 32 bit vs 40 bit dma setup will choose 40 rather than 32.
Comment 8 Andy Furniss 2009-09-25 17:45:58 UTC
(In reply to comment #0)

Fixed now in drm-next

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.