[libvirt] [PATCH] tests: Avoid double linking some libraries

[John Ferlan]

On 07/03/2014 12:47 PM, John Ferlan wrote:
> 
> 
> On 07/03/2014 08:44 AM, John Ferlan wrote:
>>
>>
>> On 07/03/2014 08:34 AM, Michal Privoznik wrote:
>>> The problem is, since 614581f32b domaincapstest is linked with
>>> $(LDADDS) by default. Then, since 94e3f23e8a7 the test may be
>>> conditionally linked with $(qemu_LDADDS) which already contains
>>> $(LDADDS). And some linkers doesn't cope with this nicely:
>>>
>>>   CCLD     domaincapstest
>>> ../src/libvirt_probes.o:(.probes+0x0): multiple definition of `libvirt_event_poll_add_handle_semaphore'
>>> ../src/libvirt_probes.o:(.probes+0x0): first defined here
>>> ../src/libvirt_probes.o:(.probes+0x2): multiple definition of `libvirt_event_poll_update_handle_semaphore'
>>> ../src/libvirt_probes.o:(.probes+0x2): first defined here
>>> ../src/libvirt_probes.o:(.probes+0x4): multiple definition of `libvirt_event_poll_remove_handle_semaphore'
>>> ../src/libvirt_probes.o:(.probes+0x4): first defined here
>>> ../src/libvirt_probes.o:(.probes+0x6): multiple definition of `libvirt_event_poll_dispatch_handle_semaphore'
>>> ../src/libvirt_probes.o:(.probes+0x6): first defined here
>>>
>>> And so on.
>>>
>>> Signed-off-by: Michal Privoznik <mprivozn at redhat.com>
>>> ---
>>>
>>> I'd push this as build breaker, but there's another approach to consider,
>>> create qemu_BARE_LDADDS and fill it with the qemu_impl, etc. Then, let:
>>> qemu_LDADDS = $(qemu_BARE_LDADDS) $(LDADDS)
>>>
>>> What's your preference?
>>>
>>
>> What is below works for me - so to that degree I'd ACK.  However, if
>> someone had a preference for whatever it is you're describing as an
>> option, then I suppose that would be fine too.  Makefile's are mostly
>> black magic to me :-)
>>
>> John
> 
> Let me amend my last statement - something is not quite working...
> 
> I tried to apply this patch to other work I'm doing and things fell
> apart with missing symbols.  If I try to reset my environment back to
> the top of the tree, then the build seems to work fine... If I apply
> this patch, then things fall apart in the same way.
> 
> It's kind of difficult to describe, but there seems to be some
> dependency that isn't quite right or doesn't get reset if this patch is
> removed from the environment.
> 
> I think the following steps make things reproducible.
> 
> 1. Start with a "clean" top...
> 2. Build, will get the multiple definition failure...
> 3. Apply this patch (git am)
> 4. Build - get lots of failures
> 5. git reset HEAD^
> 6. git co tests/Makefile.am
> 7. Build cleanly
> 8. git am the first change from my series
> 9. Build, results from step2
> 10. Repeat steps 3->7 for each/all of the patches from my series.
> 
> I'm still wondering why steps 3 & 4 worked for me before - perhaps
> because I was testing something intermediary.
> 
> Like I said before these Makefiles are black magic to me!
> 
> John
> 


The following works for me:

$ git diff tests/Makefile.am
diff --git a/tests/Makefile.am b/tests/Makefile.am
index a262c7b..2441742 100644
--- a/tests/Makefile.am
+++ b/tests/Makefile.am
@@ -833,12 +833,12 @@ vircaps2xmltest_LDADD = $(LDADDS)

 domaincapstest_SOURCES = \
        domaincapstest.c testutils.h testutils.c
-domaincapstest_LDADD = $(LDADDS)
-
 if WITH_QEMU
 domaincapstest_SOURCES += testutilsqemu.c testutilsqemu.h
-domaincapstest_LDADD += $(qemu_LDADDS)
-endif WITH_QEMU
+domaincapstest_LDADD = $(qemu_LDADDS)
+else ! WITH_QEMU
+domaincapstest_LDADD = $(LDADDS)
+endif ! WITH_QEMU

 if WITH_LIBVIRTD
 libvirtdconftest_SOURCES = \
$

Not sure if it's the "right" solution, but I'm at least able to avoid
the cycle of apply, build, remove, build.


John