core dumps being overwritten
Alexander Best
alexbestms at math.uni-muenster.de
Thu Jul 23 23:15:07 UTC 2009
i was thinking about this issue the other day again. if /var/crash/bounds gets
corrupted vital crash/debug information could be lost (like in my case).
why don't we check for existing core dumps instead of using bounds? doing
something like in this pseudo code:
i = 0;
while(exist("vmcore."i) || exist("core.txt."i) || exist("info."i))
i++;
write("vmcore."i);
write(""core.txt."i);
write("info."i);
or how about using /dev/random to create unique filenames so core dumps never
get overwritten?
cheers.
More information about the freebsd-current
mailing list