On Mon, May 13, 2002 at 09:41:04PM -0700, Beaker wrote: > >But the X driver might need kernel support, like agpgart. > > This got me curious so I took a look around at www.netbsd.org but, damm! - > not > a dang thing could I find about AGP. So I posted to the netbsd-help list and > got this > reply: > > --- reply from Michael Core via muc.lists.netbsd.help --- > Beaker wrote: > > > How DOES a user determine AGP support on NetBSD? I looked around in the > > 'supported hardware' section at http://www.netbsd.org/Ports but didn't > > see AGP mentioned. > > NetBSD-current has AGP support in general. > > man agp(4) mentions following chipsets > > - ALI M1541 host-to-AGP bridge > - AMD 751 and 761 host-to-AGP bridges > - Intel 82810, 82810-DC100, 82810E, and 82815 SVGA > controllers > - SiS 5591 host-to-AGP bridge > - VIA > > > A Google search pulled up a somewhat dated reference > > to a kernel module but again, I didn't find anything about an AGP module > > in the 'Documentation' sections. > > Support for video cards is X's problem. An exception is DRI which is not > supported by NetBSD at the moment. > > BTW, man agp(4) looks a little strange: > > HISTORY > The agp driver first appeared in FreeBSD 4.1. It was adopted in > NetBSD 1.6. > > NetBSD 1.5.2 September 11, 2001 > > --- end of reply --- > > I knew the bit about X but was rather surprised at the difficulty in finding > info on > AGP support. But there it is: Since the various free BSDs borrow stuff from > eachother > I suspect OpenBSD should have similar AGP support. Check you chipset.
Actually, we have a machine with an 82810-DC100, that runs OpenBSD. Sergey Smitienko ported the NetBSD code to OpenBSD 3.0. There was a rumor that the patches would be integrated at that time. In the mean time, the disk machine with the 82810 started grinding, so I said, "What the heck, I'd rather run OBSD on this thing anyway." I used Sergey's patches and X did indeed work, but there were some glitches. Also since OBSD 3.0 was released, XFree86 4.2.0 had been released and the release notes said there were fixes in the i810 driver, so, of course, time to try -current :) I had to make a few minor changes to resolve the diff conflicts, but once I got it compiled, I found the glitches were gone (but I now know that xdm will shutdown after running for a few days + several hours of idle time). I thought it would be nice if the patches would be integrated so I wouldn't have a bunch of ?s and Ms in my cvs update logs :) While I was fixing the conflicts, I noticed the code was not up to KNF standards (http://www.openbsd.org/cgi-bin/man.cgi?query=style), and __P macros were taken out of the -current tree, so I made a few more changes; added a function prototype here, changed spaces to tabs there, etc. Then one day there was a thread about running X on OpenBSD on an i810. I ended up tarring up the diffs I had and put them on my web server and posted the link to [EMAIL PROTECTED] A few days later some of (actually an inconsequential part) the diffs were added to the OBSD XF4 tree. I email the OBSD X maintainer, and he said some of the kernel hackers were looking at the patches. That was about 2 weeks ago, and I haven't seen any related commits from source-changes@ ... On OBSD, there's no agp device, and no agp(4) manpage. It's an extension of pcivga. pci0 at mainbus0 bus 0: configuration mode 1 (no bios) pchb0 at pci0 dev 0 function 0 "Intel 82810" rev 0x02: rng active, 6Kb/sec, potential AGP device vga1 at pci0 dev 1 function 0 "Intel 82810 Graphics" rev 0x02 OpenBSD AGP port from NetBSD: aperture at 0xe0000000, size 0x4000000 wsdisplay0 at vga1: console (80x25, vt100 emulation) pci0 at mainbus0 bus 0: configuration mode 1 (no bios) pchb0 at pci0 dev 0 function 0 "VIA VT8363 Host" rev 0x03, potential AGP device ppb0 at pci0 dev 1 function 0 "VIA VT8363 PCI-AGP" rev 0x00 pci1 at ppb0 bus 1 pchb1 at pci0 dev 7 function 4 "VIA VT82C686 SMBus" rev 0x30, potential AGP device vga1 at pci0 dev 9 function 0 "3DFX Interactive Voodoo3" rev 0x01 OpenBSD AGP port from NetBSD: aperture at 0x0, size 0x10000000 wsdisplay0 at vga1: console (80x25, vt100 emulation) pci0 at mainbus0 bus 0: configuration mode 1 (no bios) pchb0 at pci0 dev 0 function 0 "Intel 82443BX PCI-AGP" rev 0x02, potential AGP device ppb0 at pci0 dev 1 function 0 "Intel 82443BX AGP" rev 0x02 pci1 at ppb0 bus 1 vga1 at pci1 dev 0 function 0 "Intel 82740 AGP" rev 0x21 OpenBSD AGP port from NetBSD: aperture at 0xe0000000, size 0x8000000 wsdisplay0 at vga1: console (80x25, vt100 emulation) The Voodoo3 is a PCI model, the VGA slot is empty on that machine. > Cheers, > Beaker - daemon advocate - running NetBSD-1.5.2/i386 -- <[EMAIL PROTECTED]>
