Re: [Fwd: Re: Dual head on G4 Powerbook 400MHz]
On Thu, 26 Feb 2004 at 09:17:49 +1100, Benjamin Herrenschmidt wrote: In the normal case, we really want all displays opened. This make sure everything gets properly initialized by the firmware. The fact that m3mirror hack doesn't work in this case is a side effect ;) It could probably be fixed by other means though. I compared the radeonfb.c to aty128fb.c. in 2.6.3 It looks like the radeonfb.c initialzes the displays by toggling a few flags in some registers for CRTC2_GEN_CNTL and CRTC_EXT_CNTL. However, aty128fb.c only deals with the CRTC_EXT_CNTL register. I am sure there are lots of other differences too, but it is late. In which kernel version was dual head mode first working with the radeonfb driver? What kind of tweaks for dual head were needed in the radeonfb driver? Also, I am a little confused about the framebuffer. Is that only used for the primary display? It only seems to be setup with color, and mode information for the primary display in both drivers. I haven't looked at the X drivers yet, I am sure that will answer my questions. Again what version was dual head support introduced for the radeon driver so I can see what was changes were needed. -- David Stanaway [EMAIL PROTECTED]
Re: Dual head on G4 Powerbook 400MHz
On Tue, 2004-02-24 at 13:18, Julien BLACHE wrote: There was a patch for the 2.4 kernel, but it doesn't work on 2.6. On 2.4, you could also replace the prom_init.c file with the one from 2.4.20 and that would allow you to use m3mirror to enable the crt output. I have had a look at the differences, and the only things I can see are some POWER4 additions, and improving the process for opening the displays in check_display. When I boot, the prom_init output scrolls past pretty fast, and I think I just see one display get opened before the LCD gets reset to a different graphics mode. My device-tree has a parent node ATY,[EMAIL PROTECTED] with 2 children displays: ATY,RageM3p12A and ATY,RageM3p12B A is LCD and B is the CRT. Would a hack workaround be to change the sorting of display nodes so that the CRT is opened first? Also, is there a safe way to put a pause in before opening the displays in prom_init to review the displays found in the first part of check_displays? I haven't looked through the aty128fb driver yet, but what is involved in fixing up dualhead there? -- David Stanaway [EMAIL PROTECTED]
Re: Dual head on G4 Powerbook 400MHz
On Wed, 2004-02-25 at 06:18, Julien BLACHE wrote: Maybe it's possible to reuse prom_init.c from 2.4.20 with a 2.6 kernel, I haven't tried (yet). No but you can obtain the same result by commenting out one line :) Just force check_display() to only open the first fisplay. The proper solution would be to add dualhead support to aty128fb.c. (or pester BenH until he either writes the code or gives us enough information to do it -- now that radeonfb is mostly fixed, maybe we can get a chance ? ;) Or to XFree r128 driver... I don't have time to work on that and the information I have is under NDA, so while I can use it to write code, I can't just publish it... Sorry... Ben.
Re: Dual head on G4 Powerbook 400MHz
I am trying to get dual head, or even just CRT only use from the Rage 128 M3 on my TiBook without much luck. The good news is, the problem is known; the bad news is, nobody is working on it. [...] The proper solution would be to add dualhead support to aty128fb.c. (or pester BenH until he either writes the code or gives us enough information to do it -- now that radeonfb is mostly fixed, You described at least three different fixed for 2.4 - I would presume that should provide enough information to do it already? So it can't really be lack of information here. Do what I did to get mirror support for the Rage LT Pro: Use the information available (from 2.4 or other sources) to write the necessary code. Michael
[Fwd: Re: Dual head on G4 Powerbook 400MHz]
I got this from BenH, looks like he ment to CC the list but typoed :P. -- David Stanaway [EMAIL PROTECTED] ---BeginMessage--- I could make it just open the first display if there were 2 and they were both M3s, but is there are more general test to see if this is necessary? In the normal case, we really want all displays opened. This make sure everything gets properly initialized by the firmware. The fact that m3mirror hack doesn't work in this case is a side effect ;) It could probably be fixed by other means though. The proper solution would be to add dualhead support to aty128fb.c. (or pester BenH until he either writes the code or gives us enough information to do it -- now that radeonfb is mostly fixed, maybe we can get a chance ? ;) Or to XFree r128 driver... Is is much of a hack to the build scripts (debian ones) for X to just have the r128 driver rebuilt and the .o files relinked as needed? Otherwise it will take ages and ages for itterative builds? I don't have time to work on that and the information I have is under NDA, so while I can use it to write code, I can't just publish it... Sorry... No problem. I will have a look at it, but I don't have alot of spare time. -- Benjamin Herrenschmidt [EMAIL PROTECTED] ---End Message---
Dual head on G4 Powerbook 400MHz
I am trying to get dual head, or even just CRT only use from the Rage 128 M3 on my TiBook without much luck. I just got a Studio 21 display and I am not having any luck using it in linux yet. I am using sid with X 4.3.0-2 and kernel 2.6.3 (I also build the drm trunk from the source package in sid if that is relevant). Should I be using framebuffer and the ati driver in X, or should I be using the fb driver? I can see the CRT and the LCD device entries in /proc/device-tree and I can turn the lcd and the crt off and on with m3mirror, I just can't get X to use the CRT. -- David Stanaway [EMAIL PROTECTED]