when running rhn_register on a 386, the screen is filled with the blue background and then I get a Segmentation fault (core dumped) [root@gw /root]# file core core: ELF 32-bit LSB core file of 'rhn_register' (signal 11), Intel 80386, version 1 I assume that it is python which is crashing but I don't know anything about python. I have the following RPM's installed python-xmlrpc-1.4-0.6.x rpm-python-4.0.2-6x python-1.5.2-27.6.x I have upgraded to rhn_register-1.3.2-0.6.x and the problem still persists.
Jay: can you verify? JN: Do you get the segfault right away, or does it go through a couple of screens first and then segfault?
Not in our lab. Oldest thing that we have is a 486SX.
Created attachment 15464 [details] script output of crash
When rhn_register is started, there is about 20 seconds of disk grindage (it is a 386 after all, 7.94 bogomips). Then the screen colors are set to white on blue, the screen is cleared and then the crash. I have attached a typescript which probably won't help much. One thing it could be is the CPU does have the popad bug but I would imagine that the compiler would not use that instruction if it was known to be buggy (not to mention extreemly inefficient)
If this still happens with the latest versions, could you attach a copy of the core dump? I really havent a clue what is causing this atm, but suspect it is a bug somewhere either deep in python, or perhaps rpm. Have you had any similar crashes with rpm on this machine? How about any other python or gtk apps?
I have had problems on the machine before with RH6.0 because something (glibc?) was compiled with i486 instructions. This however caues 'illegal instruction' errors. SIGILL. I have now upgraded the box so I can not test it immediatly but I still have the machine if you want to persue this. I did not have any problems with RPM. The only python app I ran was rhn_register. Is there a pyhthon test suite?