bug#16335: Segmentation fault when using cp -a with SELinux and fakeroot

[Pádraig Brady]

On 01/13/2014 05:09 PM, Pádraig Brady wrote:
> On 01/13/2014 03:27 PM, Bernhard Voelker wrote:
>> On 01/13/2014 03:57 PM, Pádraig Brady wrote:
>>> On 01/13/2014 02:50 PM, Pádraig Brady wrote:
>>>> +# Then compile/link it:
>>>> +$CC -shared -fPIC -O2 k.c -o k.so \
>>>> +  || framework_failure_ 'failed to build SELinux shared library'
>>>
>>> I'll change that to a || skip_ ...
>>> so that we avoid issues with no (stub) <selinux/selinux.h> being available.
>>
>> LD_PRELOADed tests are sometimes a bit tricky, so doing
>> double checks is a good idea: I'd add a
>>   fclose(fopen("x"));
>> inside the dummies, and check if that file has really been
>> created. Otherwise, you can't be sure if replacing the functions
>> really worked.
> 
> Right, I'll skip_ in that case to warn
> about stale tests.
> 
>> Furthermore, when I added a LD_PRELOADed test a while ago,
>> I think Paul suggested to add -ldl for some non-GNU/Linux
>> platforms.
> 
> Right. I'll refactor all those calls to a gcc_shared_() for consistency.
> 
>> I'd also specify 'gcc' hardcoded ... probably with -Wall.
> 
> Hmm, icc and clang support this gcc interface,
> so I'm inclined to leave it as $CC so as not
> preclude those from this part of the testing matrix.
> We can always beef up require_gcc_shared_() if
> this ever becomes an issue.

Pushing the attached 2 patches in a while.

thanks,
Pádraig.

cp-selinux-segfault.patch
Description: Text Data