It did not work out too well with mem=nopentium. The boot process did hang 3 times due to a problem specific to my BM: MS-6570 K7N2 Delta For some unknown reason the "Athlon Powersaving mode" has been enabled by the BIOS or simply by turning on the power. This facility is simply unstable with MS-6570. The CPU stops completely, and I have to turn the power off and on.
My solution is to call the athcool program in rc.local: athcool off athcool version 0.3.0 enabling/disabling Athlon Powersaving mode nVIDIA nForce2 (10de 01e0) found disabling 'Halt Disconnect and Stop Grant Disconnect' bit ... done Address 0x6F : 0x1F -> 0x0F This solves the problem. It should really be done at the very beginning of the boot sequence, but I don't know how to do that. It looks as if the "mem=nopentium" somehow is increasing the instability. -- Bjarne On Fri, 2003-09-19 at 09:49, [EMAIL PROTECTED] wrote: > On Fri, 19 Sep 2003, Bjarne Thomsen wrote: > > > I have had about 4 cases where the keyboard > > and graphics freeze since RC2 using gnome 2.4 > > (with updates from cooker). The cursor can be moved. > > I can login via ssh. Everything seems to be working > > except X is using 99.7% of CPU time. > > I killed X but the keyboard and screen was stil > > frozen, but now with white horizontal stripes. > > I had to shutdown the PC via ssh. > > > > > > kernel 2.4.22-9mdk-i686-up-4GB > > XFree86-4.3-23mdk > > NVIDIA_kernel-1.0-4363 > > > > > > The problem could be with the nvidia driver as > > previously suggested on this list. > > The only problem is that I have been using the > > nvidia 4363 driver since April without a single > > freeze, and now I have had 4 within a week with RC2. > > I'm pretty sure it's nvidia. It used to be fixable by disabling > XRenderaccel in the XF86config, but that was not the real cause of the > problem (they have never fixed it). > Symtems are exactly as you describe, and if you remote login and trace X > it is in a loop (lots of SIGALRMs). > It could even be a real X bug triggered by nvidia, but nobody is going to > debug it without the driver source I think. > > > d. > -- Bjarne Thomsen <[EMAIL PROTECTED]> Institute of Physics & Astronomy
