On Thu, Sep 23, 2010 at 8:48 AM, Peter Boy <pboy@xxxxxxxxxxxxxxxxxxxxx> wrote: > Am Mittwoch, den 22.09.2010, 11:15 +0200 schrieb Marko Vojinovic: >> For all issues with the catalyst driver you should contact ATI (and >> good luck with that). It's closed source, they are the only ones who >> know how to fix it. > > In general you are right, but this time it seems not to be ATI but > kernel and associated kmod and wrapper magic. As I noted in my post: > Same ATI binary driver in combination with kernel-2.6.32.21-166 works > just fine. Some stuff in kernel-2.6.32.21-168 introduces regression. In most cases it is not a regression, but rather the usual kernel API/ABI change. This requires a rebuild or and/or source code modification to all affected kernel modules, sometimes the graphics drivers (and in this case apparently VirtualBox kernel module, and maybe some others as well). The kernel does require such changes occasionally, and it is the responsibility of third-party module suppliers to keep up and adjust to the new API/ABI. Depending on the severity of the kernel change, it might be enough just to recompile the module against the new kernel, or it might require interventions at the source code level, in which case the module providers usually push a new version of their software. So old modules keep working with the old kernel, but fail to be compatible with the new one. But this is not a regression of the kernel, but rather its rapid advancement. And it is the responsibility of module providers to adapt to the new changes. Now in this case, given that the catalyst driver is closed source, only ATI folks can make any relevant changes, if source code needs to be modified. Otherwise, rpmfusion folks can just recompile the module against new kernel and push it as an update. The same thing happens to closed source nVidia drivers. With open source drivers you don't have this problem because they get rebuilt and updated along with the new kernel, so the whole problem is transparent to the end-user. >> For buying ATI hardware, you have only yourself to blame. :-) > > Years ago I changed my Hardware from Nvidia to ATI because the free ATI > driver supported dual monitor xinerama those times, the free nvidia > driver did not. The ATI hardware and free driver combo worked for years > perfectly well. > > ATI is not as bad as some rumour may spread. :-) Well, if ATI works for your needs, fine, I don't mind. But my experience is that both open and closed source ATI drivers tend to break quite often, compared to nVidia and Intel. That makes every ATI graphics card unreliable from a user's perspective, and it is often a gamble whether the latest drivers will work with any particular model or not. And fuurthermore, when something doesn't work with the open source nouveau driver, the devs have a good excuse since they need to RE it. But when something doesn't work with the open source radeon driver, the devs have no such excuse, since all specs are (supposedly) available. Or otherwise ATI is lying to its Linux customers. :-) Marko -- users mailing list users@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines
