I’m having some severe problems with Neutrino 6.2.1. I’m attempting to
port a very large QNX 4 application to the new OS. The app is composed
of several resource managers and many clients; for development purposes,
I’m currently running two resource managers and a single client (which
is actually composed of three processes).
I develop and run the app via telnet sessions into the Neutrino machine.
Currently io-net is crashing about every third time I try to run the
app. In the past the machine would spontaneously reboot after the
crash, but it hasn’t rebooted in several days; only io-net seems to be
affected. Needless to say, development via telnet is very difficult
when I spend half my time walking to the machine to reboot it.
As part of the client startup, it sets up a block of shared memory. It
then forks twice, resulting in three processes sharing the same shared
memory. There is some evidence that commenting out the shared memory
eliminates the io-net crashes, but I have no hard evidence and it could
just be coincidence. It seems rather unlikely that there is any
connection to me.
I found a thread via Google in which people were reporting io-net
crashes that were fixed by altering the io-net command to
‘set(IONET_CMD, io-net -ptcpip stacksize=11648)’. I tried that, but
io-net is still failing.
I realize I haven’t given many details, but I’m clueless where to start.
If anyone has any ideas or would like more information, just ask.