apparently, disabled POSIX scheduler in snapshot images makes ports fail

[Christopher J. Ruwe]

On Wed, 26 Feb 2014 07:54:32 -0700
Ian Lepore <ian at FreeBSD.org> wrote:

> On Wed, 2014-02-26 at 13:03 +0100, Christopher J. Ruwe wrote:
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA512
> > 
> > I am trying to build postgresql-client for ARMv6 Raspberry Pi with
> > machines installed from the images provided on the official servers
> > and a custom image from the crochet.
> > 
> > That fails because in the configure stage, the test for thread
> > safety fails and dumps core.
> > 
> >       checking thread safety of required library functions... Abort
> > \ trap (core dumped) 
> >       no
> >       configure: error: thread test program failed
> >       This platform is not thread-safe.  Check the file
> > 'config.log' \ or compile 
> >       and run src/test/thread/thread_test for the exact reason.
> >       Use --disable-thread-safety to disable thread safety.
> >       ===>  Script "configure" failed unexpectedly.
> >       Please report the problem to pgsql at FreeBSD.org [maintainer]
> > and attach the
> > "/usr/ports/databases/postgresql90-client/work/postgresql-9.0.15/config.log"
> > including the output of the failure of your make command. Also, it
> > might be  a good idea to provide an overview of all packages
> > installed on your system (e.g. a /usr/local/sbin/pkg-static info -g
> > -Ea). *** Error code 1
> > 
> > Investigating, I found a similar problem where somebody failed to
> > compile postgresql ports due to  disabled KPOSIX_PRIORITY_SCHEDULING
> > in the kernel.
> > 
> > As in that situation, in the KPOSIX scheduler is disabled:
> > 
> >    #  strings /boot/kernel/kernel | grep POSIX
> >    nfscl_filllockowner: not F_POSIX or F_FLOCK
> >    POSIX P1003.1B realtime extensions
> >    The version of POSIX 1003.2 with which the system attempts to \
> >    comply 
> >    Version of POSIX attempting to comply to
> >    POSIX real time signal
> >    POSIX shared memory
> > 
> > The expected "options _KPOSIX_PRIORITY_SCHEDULING" is obviously
> > missing, which explains the build failure.
> > 
> > Yet, in stable/10, KPOSIX seems to be enabled:
> > 
> >      [cjr at dijkstra:/usr/src]$ grep POSIX sys/arm/conf/RPI-B
> >      options _KPOSIX_PRIORITY_SCHEDULING #Posix P1003_1B real-time \
> >      extensions 
> > 
> > Why is _KPOSIX_PRIORITY_SCHEDULING then disabled in the snapshots?
> > Is my thinking wrong?
> > 
> > 
> > Thanks and cheers,
> > - -- 
> > Christopher 
> 
> Your method of grepping for the string isn't reliable.  To see if
> posix priority scheduling is available, use:
> 
> sysctl kern.features.kposix_priority_scheduling
> 
> It's almost certainly going to be set to one.  Then the question
> becomes "what is the actual abort that's happening during configure?"
> 
> -- Ian
> 

The answer is yes, [...].kposix_priority_scheduling=1. Hmm ... I hoped
for an easy solution ...

Thanks,
-- 
Christopher

TZ:         GMT + 1h
GnuPG/GPG:  0xE8DE2C14
 
FreeBSD 10.0-STABLE #0 r262291+cf0fa48(stable/10): Sat Feb 22 02:01:26
CET 2014 cjr at dijkstra.hb22.cruwe.de:/usr/obj/usr/src/sys/GENERIC 
 
Punctuation matters:
"Lets eat Grandma or Lets eat, Grandma" - Punctuation saves lives.
"A panda eats shoots and leaves" or "A panda eats, shoots, and leaves" -
Punctuation teaches proper biology.

"With sufficient thrust, pigs fly just fine. However, this is not
necessarily a good idea. It is hard to be sure where they are going to
land, and it could be dangerous sitting under them as they fly
overhead." (RFC 1925)