ports/184210: [FIX] multimedia/tstools requires MAKE_JOBS_UNSAFE=yes

[Oliver Fromme]

>Number:         184210
>Category:       ports
>Synopsis:       [FIX] multimedia/tstools requires MAKE_JOBS_UNSAFE=yes
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    freebsd-ports-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          change-request
>Submitter-Id:   current-users
>Arrival-Date:   Sun Nov 24 12:40:00 UTC 2013
>Closed-Date:
>Last-Modified:
>Originator:     Oliver Fromme
>Release:        FreeBSD 9.1-STABLE amd64
>Organization:
secnetix GmbH & Co. KG
		http://www.secnetix.de/bsd
>Environment:

The behaviour described occurs on my 6-core AMD Phenom II X6
with SMP enabled.  However, I could not reproduce it on a
2-core Intel Atom system.

>Description:

Building the port multimedia/tstools on my 6-way SMP system
breaks in various, non-reproducible ways, most of the time.
Apparently it is not able to work correctly with multiple
make jobs.  For example:

cc obj/pcapreport.o -o bin/pcapreport -g -lm   -Llib -ltstools 
obj/pcapreport.o: In function `main':
pcapreport.c:(.text+0x453): undefined reference to `ethernet_packet_from_pcap'
gmake: *** [bin/pcapreport] Error 1

Or:

cc obj/ts2es.o -o bin/ts2es -g -lm   -Llib -ltstools 
obj/esfilter.o: In function `main':
esfilter.c:(.text+0x126e): undefined reference to `tswrite_open'
gmake: *** [bin/esfilter] Error 1

Or various other ways.  Sometimes the build succeeds (about
24% of the time).

Adding the following line to the Makefile fixes it:

MAKE_JOBS_UNSAFE= yes

With that line, the port builds fine reproducible on all
systems that I've tried.

>How-To-Repeat:

Try to build multimedia/tstools on a 6-way SMP system.
However, whether the problem occurs or not seems to depend
on the number of cores, and maybe it also depends on timing
and other things.  Sometimes it even succeeds on the same
system.

On my 6-way system (AMD Phenom II X6) the problem occured
about 75% of the time.  On a 2-way system (Intel Atom) that
I've tried, however, the port built just fine on three
consecutive attempts.

>Fix:

Add this line to the port's Makefile:

MAKE_JOBS_UNSAFE= yes
>Release-Note:
>Audit-Trail:
>Unformatted: