joris at giovannangeli.fr
Mon Jun 24 02:47:08 PDT 2013
I've not made a lot of progress in the code this week, i've been stuck
with a panic in my new procesdesc code. I spent a lot of time learning
how to debug things whithout a debugger, since gdb segfault when I try
to print a backtrace or to examine memory of my faulty vkernel, I
started using trace points, KASSERTS, and tools like addr2line. The
panic occurs because the refcount on file structure sometimes become
null during a closef call, but there is still a knote referencing it for
kqueue. Hence, the file structure is freed and kqueue panic.
I've started the procdesc code in the meantime, and have made good
progress in understanding what is needed. I think i'll be done with it
in the next few days. This panic has slow me down a lot, but the good
part is that I now understand the descriptor code very well. Hopefully,
by reading the diff carefully once again, I'll understand the panic soon.
More information about the Kernel