Re: What's missing/wrong in these test programs?

[Svante Signell]

On Sun, 2015-04-12 at 20:19 +0200, Samuel Thibault wrote:
> Hello,
> 
>  Now you need to check that the
> proc_user/server_identify RPC properly passes ports to the proc server
> so that the port name gets the same there.

Attached are the following files:
process.defs.new (new RPCs proc_user/server_identify)

Can somebody please give an explanation to the confusing number of
arguments is generated by MIG for the implementation of the S_*
functions to be correct. poly/simple/etc, e.g mach_port_send_t,
mach_port_poly_t?? Yes, I've read the few docs available, but they don't
give the complete story (for me). Studying the MIG source code and the
Hurd RPC's is also tough.

identify.c with printouts to stderr:
Both proc_user_identify() and proc_server_identify() gets stuck at
pthread_hurd_cond_wait_np()

Bypassing in both seems to indicate that the rendezvous port is the
same. Why is there a hang?? to move to something else by now for
sanity  

Obviously something is not correctly implemented, pleas help me to find
out where the errors are.

I have also written similar code implementing a identhandle as in auth.c
but the same hang persists.

process_U.h
processUser.c
process_S.h
processServer.c

Justus:
I don't yet have any way to publish my code via git by a public repo.
Therefore the new code is sent in whole this time. Test code is alerady
published in this thread, change #if 0 to #if 1.

Thank you for your time :)

identify.c
Description: Text Data

process.defs.new
Description: Text Data

process_S.h
Description: Text Data

processServer.c
Description: Text Data

process_U.h
Description: Text Data

processUser.c
Description: Text Data