On 04/09/2015 04:08 AM, John Paul Adrian Glaubitz wrote: > On 04/09/2015 12:38 AM, Milan Kupcevic wrote: >>> Well, yes. As you could see from the bug report you linked above, >>> compiling radeonfb into the kernel broke X on basically all >>> Radeon cards. >>> > >> It seems that radeonfb was compiled into the kernel since the dawn >> of time. > > On the powerpc Debian kernel, yes, on the other architectures, no. > radeonfb has been deprecated long time ago and should no longer be > used. > > The only reason it was compiled into the powerpc kernel without anyone > complaining is the fact that the xf86-video-ati (= Radeon X.Org) driver > in Wheezy was still old enough [1] to support userland mode-setting > which was dropped upstream in version 7.0.0 almost three years ago [2]. > > This is no longer true in Jessie and therefore we have to use radeon, > not radeonfb. Otherwise you won't get any X display. What did you do to ensure the radeon module is included in debian installer? > >>> The reason is that the radeon KMS module does not work once >>> radoenfb has been loaded once. Even if you unload radeonfb later, >>> the radeon module will still refuse to work meaning X will not be >>> usable. >>> > >> This is good to know. > > I did thorough testing for the bug report you linked above which is > why I am a bit surprised you could convince Ben so easily to partly > revert my change. I think I also explained in the other bug tracker > that you have to use KMS these days as UMS is being phased out. > There is no revert of your change at all. We are just trying to add a module to debian installer to enable installation on machines with no native offb support, the machines you have skipped in your thorough testing. Any input on the best way to achieve this is welcome. >>> Well, I think this can be answered straight-forward: radeonfb is >>> the wrong solution as radeonfb does not support X which means you >>> can use the text-based installer only. But I guess you never >>> tested that as you stated in your bug report. >>> How is that X display in debian installer on PowerPC platform working for you? >>> The proper fix for your problem is not to revert my previous fix >>> but to use the proper driver, thus reopening. >>> > >> There was no "my previous fix," just reporting about the state of >> the things in RC1 and RC2. I'm trying to help as I have access to >> more than a few power and powerpc machines. > > Sure, but please ask people in the future before you are asking for > changes which you don't understand. Your understanding of the consequences of your changes is trully stellar. > I also have tons of different > hardware available here, including several powerpc machines and > way beyond that (m68k, sh4, mips etc). > b > I think you should at least discuss such changes with some more people > before you ask for them to be integrated. Well, I've asked, no one complained until you showed up. The resulting discussion spreads very interesting scent. > For anyone who follows X.Org > and kernel development, it is common knowledge that all the non-KMS > drivers in the kernel are being deprecated and that you should no > longer use them. This also applies to X.Org. > > It would be nice if you could CC me in the future if you are having > issues with non-x86 hardware, especially anything that's powerpc, > Macintosh, or - like in this case - Amiga-related hardware. Are you sure you want to know about every non-x86 issue I face on daily basis? M
Attachment:
signature.asc
Description: OpenPGP digital signature