After some hours of configuration, I could successfully use our
-current XF86 4.2 sources with my iBook. Everything is accelerated and
it's neat.

 However, I have two serious problems (which seem to be triggered by X
but are likely to be kernel bugs):

- When I start xdm (wscons is enabled) from rc.conf, the X server
  catches a SIGSEGV (as can be seen in XFree86.0.log), creates a
  0-length /XFree86.core but the X server is stuck in state "biowait"
  obviously waiting for its core to be dumped. Of course it can't be
  killed because it's already dead.

 Problem is that the screen is black by then, and I can only ssh from
 another computer. From there, if I reboot, the disk is marked dirty
 anyway (some flush certainly doesn't happen correctly) and as another
 datapoint, "ls -i /" (/ is where the empty core lies) as a normal
 user makes the machine crash (dunno how or why, screen is black and
 I got no kernel dump).

 Note that if I don't activate xdm in rc.conf but rather log in and
 then "forcestart" it directly, everything is fine.

 The problem is at least two-fold: why does X get a sigsegv when
 started "too early" or "with too much disk activity going on"? Why
 does the kernel let itself be fooled by this situation?

- Another bug in the X server is that when I logout from my login
  xterm, the X server often tries to quit but seems to be stuck in an
  infinite loop, taking almost all the CPU, and apparently doing a lot
  of syscalls because most of the CPU percentage is taken by the
  system. Remotely killing the X server works here and xdm starts
  another server.


All this is with -current sources from march, 26th and thus with new
pmap.

I'd like to know if anyone has stumbled on this and/or has a (better)
analysis so that I can file one or two PRs in the right categories.

thanks,
 Aymeric