This question relates to errors reported by memtest86+-1.55.1-1. The documentation warns that sometimes the program gets the wrong answer when calculating the address ranges used by memory. I am suspicious that this may be the case and I would like a second opinion. Anyone else every see something similar? Last week I lost a power supply on my FC4 machine at home. I put in a new one and everything was find for a day or so. It then froze. /var/log/messages showed a kernel BUG at include/linux/list.h:166. That line is sanity checking a double linked list. I dropped back to a previous kernel and after a while got the same result after a day or so. I loaded the memtest86+ rpm and booted it. It ran fine till test 5 which is copying large blocks of data. I got 762 errors. The box has 2 256 Meg dimms. I pulled one out and ran the test again. Got errors at the same address ranges (at least the ones I saved). I swapped in the other memory dimm and ran again. Same errors, same addresses. Very suspicious. I put everything back together and booted normally. Swaping the memory around of course reseated the boards. The MB has a 1.2 GHz AMD Duron processor. To complicate the issue, I had one other change. I have been playing with Solaris X86 V10 on another machine and have a cron job automounting some overlapping directories on the FC4 machine. I turned that off after the problems and have not seen any problem for a few days. Thanks for your time. Robert E. Styma Principal Engineer (DMTS) Lucent Technologies, Phoenix Email: stymar@xxxxxxxxxx / styma@xxxxxxxxxx Phone: 623-582-7323 Cell: 602-478-0114 Company: http://www.lucent.com Personal: http://www.swlink.net/~styma
