to, 2007-09-06 kello 02:34 +0300, Antti J. Huhtala kirjoitti: > Hi list, > > I've just filed a bug in > > https://bugzilla.redhat.com/show_bug.cgi?id=278721 > > to report about agpgart aperture check failure after two recent kernel > updates in my AMD Athlon 64 3000+ desktop box with ASUS K8N MB, Nvidia > nForce3 chipset and ATI Radeon 9200 PRO controller. > The problem appears in trying to restart the computer (warm boot). Then > the booting process freezes right after 'Booting the kernel.' appears on > screen. It won't go on from that point. > > To be able to use the box at all I must perform a cold boot, i.e. power > the system completely off. A couple of seconds off is sufficient. When > booted after power-off, four lines of agpgart messages including > > "No usable aperture found." > "Consider rebooting with iommu=memaper=2 to get a good aperture." > > are printed but booting continues past the 'freeze point' and eventually > ends up in usual Gnome login screen. From then on the box runs normally. > > agpgart aperture check has worked in this box through all kernels since > FC4 in June, 2005 until 2.6.22.4-65 update on 25th Aug. apgart version > is still the same 1.02 which has worked just fine for at least several > months. > > A more detailed description can be read from bugzilla. If anyone on this > list has similar experience and box components, please add your comments > to the above bug. If someone knows a quick remedy, that's welcome, too. > > Thank you! > > Antti > Although nobody on the list has commented my original bug report, I consider it appropriate to announce also here that a solution was found for my AGP aperture detection problem and my consequent inability to warm-boot the computer since kernel 2.6.22.4-65.fc7.x86_84 update. My desktop box (only) has 512 M of RAM. However, I have to put mem=510M in /etc/grub.conf kernel line to facilitate both computer restart (warm boot) and AGP aperture detection. The latter is important e.g. in using GoogleEarth and other graphics-intensive programs (i.e. for speed). My uneducated guess is that if no upper RAM memory limit is given, a memory overflow will occur and newer kernels crash. I don't know why this didn't happen until kernel 2.6.22.4-65 update came about. I still have one minor problem: if I cold boot my box, AGP aperture is not detected but a simple reboot rectifies that. I suspect this is a timing problem in cold boot process but it may be just my memory chips. Antti
