1 |
On Fri, Apr 10, 2009 at 11:42 AM, Paul Hartman < |
2 |
paul.hartman+gentoo@×××××.com <paul.hartman%2Bgentoo@×××××.com>> wrote: |
3 |
|
4 |
> On Fri, Apr 10, 2009 at 10:26 AM, Nikos Chantziaras <realnc@×××××.de> |
5 |
> wrote: |
6 |
> > Harry Holt wrote: |
7 |
> >> |
8 |
> >> On Fri, Apr 10, 2009 at 11:06 AM, Nikos Chantziaras <realnc@×××××.de |
9 |
> >> <mailto:realnc@×××××.de>> wrote: |
10 |
> >> |
11 |
> >> I don't know if this error exists upstream or not, but I may try |
12 |
> >> installing the 9.3 version from ATI just to find out (if I can |
13 |
> >> get through it). I know the Debian version in "squeeze" works |
14 |
> >> but it's a newer version as far as I can tell (version numbers |
15 |
> >> are really inconsistent). |
16 |
> >> |
17 |
> >> Anyway, has anyone else had this problem? Is there a way around |
18 |
> >> it without turning on PCI_LEGACY in the kernel? |
19 |
> >> |
20 |
> >> |
21 |
> >> No. You have to read the comments of that bug report and do as they |
22 |
> >> say. The driver needs to use a kernel feature that is disabled in |
23 |
> >> your kernel. Debian has it enabled. |
24 |
> >> |
25 |
> >> |
26 |
> >> Sorry, but that's just not true. If I compile a 2.6.27 kernel in debian |
27 |
> >> with PCI_LEGACY=N, I can still install the "9.2" deb ATI Driver package. |
28 |
> >> But the same kernel configured the same way in Gentoo causes the |
29 |
> >> ati-drivers compile to fail. |
30 |
> > |
31 |
> > Then I guess you can open a new bug report about it. Maybe that option |
32 |
> is |
33 |
> > not needed anymore. On Gentoo, old stuff tends to accumulate over time |
34 |
> and |
35 |
> > no one really checks if it's still needed. It's very possible that many |
36 |
> > kernel options the ebuild checks for are not needed at all anymore. |
37 |
> |
38 |
> That's a good idea. Try to comment out the check in the ebuild and see |
39 |
> what happens. I use the x11 radeon driver so I can't try it. |
40 |
> |
41 |
> |
42 |
Well I'm still working on trying various things. I decided to go ahead and |
43 |
enable PCI_LEGACY to get xorg, etc. all in place and then try disabling to |
44 |
see if things still worked. Unfortunately I ran into this: |
45 |
|
46 |
* The current ati-drivers don't compile when having |
47 |
* paravirtualization active due to GPL symbol export |
48 |
* restrictions. |
49 |
* Please disable it: |
50 |
* CONFIG_PARAVIRT=n |
51 |
* in /usr/src/linux/.config or |
52 |
* Processor type and features --> |
53 |
* [ ] Paravirtualization support (EXPERIMENTAL) |
54 |
* in 'menuconfig' |
55 |
|
56 |
Which is crazy, because the whole point of this box is to run VMs. |
57 |
|
58 |
Does the "fglrx" VIDEO_CARDS flag do anything other than pulling in a |
59 |
dependency? If so I could just disable that get everything in place. |