Re [solved]: Prashant's oddly hanging build ...

This issue is solved.

Thanks to Stephen (sberg) and Michael (mmeeks) for their involvement and
assistance :)

Regards
Prashant

On Wed, Mar 6, 2013 at 5:25 PM, Michael Meeks <michael.meeks@suse.com>wrote:

> Drat, sent this early with ctrl-enter:
>
> On Wed, 2013-03-06 at 11:49 +0000, Michael Meeks wrote:
> >       A new thread - 31830 - which does exactly nothing except fiddling
> with
> > futexes:
> >
> > 31830 1362225668.627412 futex(0x2b15cd5c3360, FUTEX_WAKE_PRIVATE, 1) = 0
> > 31830 1362225668.627477 futex(0x2b15cd5c35cc, FUTEX_WAIT_BITSET_PRIVATE|
> > FUTEX_CLOCK_REALTIME, 2663, {1362225678, 627393000}, ffffffff) = -1
> > ETIMEDOUT (Connection timed out)
>         for the rest of it's life - and I assume is the ultimate cause of
> the
> gnumake hang in wait() - it's waiting for this child to be cleaned up.
>
>         meanwhile the libtest_sw_filters_test.so seems to go into a
> slow-spin-rinse cycle of endless ;-)
>
> 31829 1362212360.132295 pread(16,  <unfinished ...>
> 31829 1362212360.132314 <... pread resumed> "", 4096, 35) = 0
>
>         Which I would tend to not expect either. Endless short reads from
> socket 16 which is apparently an XML file containing only an XML header:
>
> 31829 1362212360.126681 open("/tmp/kDxAZ6", O_RDONLY|O_EXCL
> <unfinished ...>
> 31829 1362212360.126703 <... open resumed> ) = 16
> 31829 1362212360.126733 fstat(16,  <unfinished ...>
> 31829 1362212360.126752 <... fstat resumed> {st_mode=S_IFREG|0600,
> st_size=25, ...}) = 0
> 31829 1362212360.126821 pread(16,  <unfinished ...>
> 31829 1362212360.126835 <... pread resumed> "<?xml version\302
> \205='1.0'?>\r\n", 4096, 0) = 25
> 31829 1362212360.126867 pread(16,  <unfinished ...>
> 31829 1362212360.126882 <... pread resumed> "\n", 4096, 24) = 1
> 31829 1362212360.126909 pread(16,  <unfinished ...>
> 31829 1362212360.126924 <... pread resumed> "", 4096, 25) = 0
>
>         Which is also odd.
>
>         Prashant - if this happens again - you really need to attach gdb
> to the
> hung unit test process - which will be called 'cppunittester' and get a
> stack trace for us.
>
>         How are things going now ?
>
>         Thanks,
>
>                 Michael.
>
> --
> michael.meeks@suse.com  <><, Pseudo Engineer, itinerant idiot