| |||||||||||
Linux breaks desktop barrier in 2004: Torvalds (LinuxWorld.com.au)Linux breaks desktop barrier in 2004: Torvalds (LinuxWorld.com.au)Posted Jan 16, 2004 23:32 UTC (Fri) by mmarq (guest, #2332)Parent article: Linux breaks desktop barrier in 2004: Torvalds (LinuxWorld.com.au) OK... but when can "we" have a ALVA (v->video) or a ALGA (g->graphics) ? Isn't it possible to have an advance system similar to ALSA , in kernel land, that encompasses 3D and 2D, i.e. DRI, FRAMEBUFFER and GENERAL DRIVER, living X for windowing only ? There are a few projects around libGGI, DRM, framebufer guys, XFree, Xouvert,... wouldn't it be just the case of putting everybody working on a code base with all those encompassing functionality ? Having X server with network transparency is already a win over Windows GDI,..., no need for every functionality to be attached to it!... This means giving Fresco a chance, in the good old spirit of Open-Source.
(Log in to post comments)
Kernel API issues Posted Jan 17, 2004 4:38 UTC (Sat) by proski (subscriber, #104) [Link] As I understand it, video cards are much more sophisticated devices than audio cards. It's much harder to come with a kernel API that would allow access to most functions of several different video cards. It would be too complex for the kernel and prone to security issues.For instance, if video cards had acceleration for fonts (maybe some cards have it already), would you move font support to the kernel?
Kernel API issues Posted Jan 17, 2004 16:17 UTC (Sat) by mmarq (guest, #2332) [Link] But that for the sake of Linux and Open-Spouce is the all point...Open-Source needs a Advance (..) Arquitecture for Video and or graphics, if he wants not to be hostage of DRM strict schemes, and in consequence of M$... As to specfications, IMHO, would be nice to ear what GPU manufactors have to say,..., in Ottawa Symposium i supose...
Kernel API issues Posted Jan 17, 2004 16:19 UTC (Sat) by mmarq (guest, #2332) [Link] "if he wants not to be hostage of DRM strict schemes"Here i mean Digital Rights Management.
Kernel API issues Posted Jan 22, 2004 13:22 UTC (Thu) by ortalo (subscriber, #4654) [Link] IMHO: of course not! The problem is not to decide if all graphic board features should be implemented in the kernel. That would be stupid of course. But, let me reverse the question: do you think it is reasonable that graphics-related hardware be the only type of hardware that "should" be directly controlled by a user space program like X11? I do not find this reasonable either. What I'd like to see is some simple and efficient way of controlling access to graphics hardware in the kernel. If you look at KGI you'll see that the kernel driver does not export so many things.
Linux breaks desktop barrier in 2004: Torvalds (LinuxWorld.com.au) Posted Jan 17, 2004 10:22 UTC (Sat) by nix (subscriber, #2304) [Link] Having X server with network transparency is already a win over Windows GDI,..., no need for every functionality to be attached to it! If there's a choice between putting something in the X server and putting it in the kernel, and whatever-it-is doesn't need mediation between multiple processes accessing the video card at the same time, put it in the X server. It's not much of a security advantage yet, what with the X server having CAP_RAWIO and running as iopl()ed root, but at least it can be swapped out, and at least at some point in the future it *can* be moved somewhere less prone to holes. (The only bit of the X server that runs in kernel space, DRM, is there because its job *is* mediation between multiple processes accessing the video card at the same time.)
Linux breaks desktop barrier in 2004: Torvalds (LinuxWorld.com.au) Posted Jan 18, 2004 19:35 UTC (Sun) by garloff (subscriber, #319) [Link] The fact that the Kernel developers work in a much better way than thevarious X11 projects out there at the moment is no reason to put graphics drivers in the kernel.
|
|||||||||||