3rd party kernel modulesegfault
08-21-2005 ITYM urban legends. That has not been my experience. Untrue. BAT files are not generally where configuration data are stored. Instead you have to edit, e.g...
This is a post from an intermediate level Linux user, who is in his final year...
I am using a 3rd party kernel module that has some glaring deficiencies, but I don't really have enough experience to be confident in my buttertion that the deficiencies are so horrible. The two that I find most annoying are:
1) At least one of their library calls will cause a kernel panic if given invalid parameters. (ie, a kernel panic will happen within a few seconds of making the call.)
2) Several of their library calls print to stderr instead of returning an error value.
I'm listing the second point merely because I take it as evidence that the implementers didn't have a clue. Is that being too harsh?
However, my immediate concern is that a simple program in which I am using their library calls is occasionally segfaulting. By "occasionally", I mean that I can force the condition after about 10 iterations of 'start process, start 2nd process, kill 1st process, repeat'. The backtrace shows the segfault coming from a malloc called from syslog.
My first thought is: "buggy kernel module, so my entire system is unreliable, I should dump this vendor and find another solution." My second thought is: "segfault is in malloc via syslog; it's not really their fault".
How should I proceed? Am I correct in thinking that the segfault is probably caused by their mis-handling of memory in kernel space? At the time of the segfault, I've used their calls to allocate several data structures, and then used their calls to free the memory used, and I have a gut feeling that says they aren't freeing the memory correctly; how can I track that down without spending huge amounts of time doing their job?