ehrlich@psuvax1.psu.edu (Dan Ehrlich) (10/28/87)
Of late our 4.3BSD UNIX kernel has been doing crash and burns of the "panic: protection fault" variety. Not being an kernel debugging wizard I have a few questions. First background. The only thing that has "changed" in the last 3-4 weeks is we are now heavily using a program the runs IBM's RSCS protocol over UDP (N.B. it is blindingly fast and beats a 9600 baud synch modem but a mile). We use it to connect BITnet hosts on our campus together. It can generate quit a large number of packets in a short time and I have a gut feeling that we mave have uncovered a bug in the UDP code. Not wanting to point an acusing fingers without enough eveidence, if anyone can answer the following for me I would appreciate hearing from you. 1) How does one determine where in the kernel the protection fault occured using adb? 2) I have been told that the scripts in /usr/lib/adb "do not work". I find this hard to believe as they were being pushed by Karl McKusick in the "Advanced 4.3 Internals" course at the Phoenix USENIX. Any comments? 3) Are there any "friendlier" tools than adb for doing crash dump analysis? Any help would be greatly appreciated. Thanks in advance.