[OE-core] [PATCH 1/1] dbus: fix ptest runtime failures

Kang Kai Kai.Kang at windriver.com
Tue Sep 5 01:56:47 UTC 2017


On 2017年09月05日 04:04, Richard Purdie wrote:
> On Mon, 2017-09-04 at 17:39 +0800, kai.kang at windriver.com wrote:
>> From: Kai Kang <kai.kang at windriver.com>
>>
>> When run dbus ptest cases, there are some failures:
>>
>>> ./test/test-dbus-daemon: relocation error: ./test/test-dbus-daemon:
>>> symbol
>>> _dbus_real_assert, version LIBDBUS_PRIVATE_1.10.20 not defined in
>>> file
>>> libdbus-1.so.3 with link time reference
>>> FAIL: test/test-dbus-daemon
>>> [   24.943454] test-dbus-daemo[251]: segfault at 0
>>> ip           (null) sp
>>> 00007ffddabe6818 error 14 in test-dbus-daemon-eavesdrop[
>>> 400000+a000]
>> It disables tests in dbus recipe which cases symbols missing and
>> segment
>> fault. When 'ptest' in DISTRO_FEATURES, it should enable some
>> configure
>> options to make the symbols visible to ptest cases.
>>
>> It also need to correct paths in ptest configure files.
> Does this have a runtime effect on dbus, i.e. are these runtime asserts
> which we don't want in production code?

HI Richard,

I am sorry that did some copy and paste errors for packageconfig ptest. 
It should disable them all if 'ptest' is disabled.
Did that make myself misunderstood?


If not, please:

It gives warnings in configure.ac for enable embedded_tests and asserts:

if test x$enable_embedded_tests = xyes; then
         echo "NOTE: building with unit tests increases the size of the 
installed library and renders it insecure."
fi

if test x$enable_asserts = xyes; then
         echo "NOTE: building with assertions increases library size and 
decreases performance."
fi


It seems true that enable asserts will affect runtime performance. But 
the configure options are required for ptest cases.
Would is it not ok that only enable them when ptest in DISTRO_FEATURES?


I'll send V2 to correct the typos. Sorry for inconvenience.


Regards,
Kai

> Cheers,
>
> Richard
>
>

-- 
Regards,
Neil | Kai Kang




More information about the Openembedded-core mailing list