[compiler-rt] r252742 - [ASan] Improve portability of new tests.

[Evgenii Stepanov via llvm-commits]

The patch looks fine, go ahead and submit it.
Btw, I've just disabled this test on android (did not notice this
message). Please make sure to remove the UNSUPPORTED line.


On Thu, Nov 12, 2015 at 8:42 AM, Yury Gribov <y.gribov at samsung.com> wrote:
> On 11/12/2015 12:40 AM, Evgenii Stepanov wrote:
>>
>> I've removed the test in r252805 to unblock the bots.
>>
>> On Wed, Nov 11, 2015 at 12:12 PM, Evgenii Stepanov
>> <eugeni.stepanov at gmail.com> wrote:
>>>
>>> Also, this test hangs on linux starting with:
>>> http://lab.llvm.org:8011/builders/sanitizer-x86_64-linux/builds/21209
>
>
> Again, thanks a lot for unblocking the build. I've indeed found a silly bug
> in reporting code. The problem is that I can not repro the error locally
> (although I've been rerunning make check-all for couple of hours on my
> x86_64 Ubuntu 14). How would you suggest to proceed?

I think it's OK to submit the fix and look at the bots. For me the
problem reproduced on the 19th run of the signals test on a GCE x86_64
machine under load (it was running the buildbot in parallel).

>
>>> On Wed, Nov 11, 2015 at 12:01 PM, Evgenii Stepanov
>>> <eugeni.stepanov at gmail.com> wrote:
>>>>
>>>> On Android rand_r appears in API 21, and we need to support at least API
>>>> 19.
>>>> You may need to use locked rand() or even implement a simple generator
>>>> right in the test.
>
>
> I've attached the patch for this. Does it look sane or should I do a
> full-blown review?
>
> Best regards,
> Yury Gribov
>