In have had this problem with a standard athlon t-bird 1.4 It started when gnome2/gcc3.2 came out. Found that cflags, specifying i686 instead of athlon helped but didnt cure the problem. Extra cooling also didnt do a thing, despite a 10 degree C temp drop under full load. A curious problem was nautilus and setiathome created instant hard lockup, other applications took awhile under load plus there were random lockups under no load.
Cure was winding the clock speed back to 1.3G (underclocking). Very unsatisfactory, and I am still suspicious of something in gnome2/gcc3.2.x has caused the problem - otherwise, why would a system that was perfectly stable change at the exact time of this update. Unfortunately, I have to use this system as my desktop/workstation, so have left it at 1.3G. BillK On Mon, 2003-02-10 at 09:51, Bruno Lustosa wrote: > Hello, list. > > I have tried everything I can think of, but still can't find an answer > myself. So thought someone here might be having the same problem. > I'm running Gentoo 1.4 on an Athlon XP 1800+, a Soyo DragonPlus MB, and > some other miscellaneous cards. > > Problem is, if I let the system running and the load comes to a high > point (load avg at about 2-3), system will eventually come to a crash. I > can reproduce it every time. > Sometimes, there's a message from the kernel (a call to BUG(), an "Aiee! > Killing interrupt handler" message and the system is frozen). > I thought there might be a bug with all the patches applied to the > standard gentoo kernel sources, becuase I tried lots of configurations, > but they all lead to the same hanging. > So, I emerge'd sys-kernel/vanilla-sources, compiled a plain kernel > without any fancy options, but alas... same behaviour. > > I have used Linux before on this computer (Debian and Slackware), and > have WinXP installed as well. Gentoo is the only one who's getting > trouble with the machine, so I don't think this could be a hardware > issue (wouldn't windows or the other distros hang as well?). > At the moment, I'm trying recompiling glibc with different CFLAGS, as I > used somewhat high optimization flags to build the majority of the > system. > > Instead of using: > CFLAGS="-march=athlon-xp -m3dnow -msse -mfpmath=sse -mmmx -O3 -pipe -fforce-addr >-fomit-frame-pointer -funroll-loops -frerun-cse-after-loop -frerun-loop-opt >-falign-functions=4 -maccumulate-outgoing-args -ffast-math -fprefetch-loop-arrays" > > I'm using: > > CFLAGS="-mcpu=athlon-xp -O3 -pipe" > > Could this be the problem? -- William Kenworthy <[EMAIL PROTECTED]> -- [EMAIL PROTECTED] mailing list
