[Xquartz-dev] Fwd: 2.4.0_beta4 and beta5

[Harald Hanche-Olsen]

+ Jeremy Huddleston <jeremyhu at apple.com>:

> > Actually, 1.6.2-apple1 also didn't work well on top of 2.4.0_rc1. Once
> > I switched from X11 to Terminal and back, the xterm was unresponsive,
> > and I had to quit X11.
> 
> I haven't seen that yet.  Do you have any traces of the problem?

Well, below is a sample of Xquartz. Not very interesting, but perhaps
illuminating none the less?

Sampling process 44745 for 1 seconds with 1 millisecond of run time between samples
Sampling completed, processing symbols...
Analysis of sampling X (pid 44745) every 1 millisecond
Call graph:
    835 Thread_2507
      835 start
        835 main
          835 start_x11_server
            835 mach_msg
              835 mach_msg_trap
                835 mach_msg_trap

Total number in stack (recursive counted multiple, when >=5):

Sort by top of stack, same collapsed (when >= 5):
        mach_msg_trap        835
Sample analysis of process 44745 written to file /dev/stdout

Just before the hang, I find this in system.log:

Jul 28 16:16:54 mach org.x.startx[44696]: font_cache: Done
Jul 28 16:16:55 mach org.x.privileged_startx[44703]: font_cache: Done
Jul 28 16:16:56 mach org.x.startx[44696]: X11.app Handing off fd to server thread via DarwinListenOnOpenFD(5)
Jul 28 16:16:56 mach org.x.startx[44696]: DarwinListenOnOpenFD: 5
Jul 28 16:16:56 mach org.x.startx[44696]: Calling ListenOnOpenFD() for new fd: 5

I don't know what other sorts of traces I can come up with. There used
to be a whole pile of ready made scripts for use with dtrace
somewhere, I think, but I can't find them. (And I haven't learned to
use raw dtrace on my own.)

- Harald