[LLVMdev] [cfe-dev] [lldb-dev] [3.7 Release] RC1 has been tagged, Testing Phase I begins

Ben Pope benpope81 at gmail.com
Wed Jul 22 04:48:37 PDT 2015


On Wednesday, July 22, 2015 04:44 AM, Hans Wennborg wrote:
> On Tue, Jul 21, 2015 at 7:39 AM, Ben Pope <benpope81 at gmail.com> wrote:
>> On Tuesday, July 21, 2015 05:51 AM, Hans Wennborg wrote:
>>>
>>> I don't think the test failures are correlated to not getting a
>>> tarball. It's probably the "rpath problem" that Dimitry mentioned
>>> further up in the thread. Try using his patch for it
>>
>>
>> It still failed, I used test-release.sh r242767 which appears to have the
>> rpath change.
>
> Weird. Did you get any output suggesting what the error was?

My normal (using cmake) builds on linux usually put the rpath on, 
although I thought the cmake install step then strips them - I just 
checked and that's not true for my latest build from trunk which both 
have RUNPATH, not RPATH like the rc1.  Weirder and weirder.

The rpath step in test-release may have been added or changed back 
around release 3.3; it looks like that discussion didn't hit the mailing 
list, nor did it include you.  I can forward it to you if you like.

Anyway, cmake supports customisation of the rpath handling, so we should 
probably just use that and remove it from the test-release script.

The only errors I can see are:

CMakeFiles/check-all.dir/build.make:49: recipe for target 
'CMakeFiles/check-all' failed
make[3]: *** [CMakeFiles/check-all] Error 1
make[3]: Target 'CMakeFiles/check-all.dir/build' not remade because of 
errors.
CMakeFiles/Makefile2:129: recipe for target 
'CMakeFiles/check-all.dir/all' failed
make[2]: *** [CMakeFiles/check-all.dir/all] Error 2
CMakeFiles/Makefile2:137: recipe for target 
'CMakeFiles/check-all.dir/rule' failed
make[1]: *** [CMakeFiles/check-all.dir/rule] Error 2
make[1]: Target 'check-all' not remade because of errors.
Makefile:171: recipe for target 'check-all' failed
make: *** [check-all] Error 2

So it appears to be -o pipefail picking up the sanitizer failures.

Ben





More information about the llvm-dev mailing list