On Maw, 2004-02-10 at 00:30, Jon Smirl wrote: > There are lots of solutions to this: > 1) queue the printk's
Not a solution. The box crashed, game over, where is my data > 2) add an in-kernel path for write to console that cooperates with the normal > user space one Ditto > 3) if you know you are going to be debugging code like this, use an alternative > solution like the serial port or a second video card running framebuffer., Makes life hard for 99.9% of the debugging people > You can't see a kernel oops from an interrupt when running XWindows any way. Untrue if you are running a suitable console. One thing you can get out of this with kernel side mode switching is *more* ability to get back into a sane mode and dump data. (BTW there is a patch to use int10 16bit emulation hacks to drop back into 16 bit mode on oops and bios clear to display the crash) > I'm sure an expert kernel hacker can come up with 10 more solutions. Let's work > together to try and solve these problems. Why the hell do you want console in user space anyway ? If the console layers are built properly then it comes for free near enough Your bottom layer is a PCI bus driver that spots cards and handles them appearing and leaving, and deals with command queues, describing resource types and which are safe to access. It also handles revocation and hotplug event paths. It might do some or all the mode switching but as Ben pointed out to me hotplug can handle that in some situations. It should also export a description of the frame buffer memory layout, palette and the like. (Akin to what DGA exports basically). It may also offer some minimal accelerations for use by the fb driver but primarily it wants to export the descriptions and the command engines. On top of that you can load a frame buffer driver, unaccelerated and not hard to deal with. The kernel console code is pretty decent there although it too lacks hotplug paths, even though the tty layer above it is quite happy. It also lacks multi-console and/or you can load a DRI module which uses the queue interfaces to provide direct render service and/or X server render services It all starts to look much like DirectFB, and I think there is some wheel redesigning going on around here that perhaps should be avoided by looking at directfb in detail too as a basis for the final thing ------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click -- _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
