Hi, i write to tell you about weird hangs i experienced. The machine is an Ultra 5 333 Mhz, running debian testing. I compiled a vanilla 2.6.0 kernel. At the beginning this machine only had one of the two banks of ram used with two sticks of 64 MB each. Runing kernel 2.6.0 went just fine witht hat ram configuration, i got an uptime of about 30 days and everything was quite smooth. I upgraded the memory, putting 4 sticks of 128 Megs each,that should be the max memory supported by ultra 5. I rebooted just fine. After few hours the machine hanged. No kernel panics, no trace at all. I just rebooted. Few hours and again the machine hanged.
At this point i tried the diagnostics in openboot. So i ran those tests and the memory and the rest of the hardware, according those tests, were just fine. I did a search on the net and i read many peope suggesting not to rely on openboot diags that much. So i started a memtest. Memtest ran for about 2 hours without problem. To run memtest i booted the machine. Ran top and i could see about 20 megs of memory used and the rest free. I started memtest telling it to lock about 400 Megs of memory. It started and ran for two hoiurs just fine. I thought memtest would have locked 256 Megs as maximum, i mean i thought the test could be conducted on one bank completely not used. Few hours later the machine hanged again. At this point i compiled a vanilla 2.4.24 and booted it. Ran memtest again under 2.4.24. Top shows about 20 megs of memory used again. And once again i started memtest, telling it to lock 400 Megs of memory. This time memtestrefused to do it and just allocated 256 megs (a completely "free" bank?). It ran for 12 hours, all fine. I waited some days to report this, now the machine has been up for 7 days. I just wanted to report these weird hangs with 2.6.0. If you think i can be of help tracing a bug let me know what to do. Should i write to kernel list too? Have a nice day, marco -- marco <[EMAIL PROTECTED]> ppcnerds.org
