perl build (was: Re: Stability of 2.0_RCx?)

Some perl build experiences:

(with Izumi's patch)

> I also started a pkg build of perl and a kernel compilation over the
> night. They both aborted with a signal 11 in the compiler, but the
> system didn't reboot. I restarted them and the perl compilation is
> currently still running.

I had to restart the perl build 2 or 3 times, since the compiler
aborted with signal 11 each time. But finally the whole perl was
built.

"make test" reports some errors:

------
t/op/taint................................[1]   Bad system call (core dumped) ./perl -I. -MTes...
FAILED at test 150
------
ext/Encode/t/Aliases......................#     Failed test (../ext/Encode/t/Aliases.t at line 126)
#          got: ''
#     expected: 'cp949'
alias was ks_c_5601-1987 at ../ext/Encode/t/Aliases.t line 126.
FAILED at test 165
------
ext/IPC/SysV/t/ipcsysv....................SIGSYS caught.
It may be that your kernel does not have SysV IPC configured.

FAILED at test 1
ext/IPC/SysV/t/msg........................FAILED at test 0
ext/IPC/SysV/t/sem........................FAILED at test 0
------
ext/POSIX/t/sigaction.....................FAILED at test 5
------

The SysV errors are correct, I don't have SYSVxxx enabled in the
kernel config file.


Later on the make test apparently hangs at

ext/threads/t/stress_cv...................

"ps ax" shows a perl instance

 2391 p0    UxLa+  1:41.51 ./perl -I. -MTestInit ../ext/threads/t/stress_cv.t 

and on the console I got a message

sa_switchcall(2391.2 flag 4d00004): Not enough stacks.


Not quite prime time :-/

What actually is the "U" status in the ps output?
Why is there a "bad system call"?

I will reboot and re-run "make test" right now...

regards,
chris