On 18 Jun 2019, at 07:54, Viktor Ashirov <vashirov(a)redhat.com>
wrote:
On Tue, Jun 18, 2019 at 1:30 AM Simon Pichugin <spichugi(a)redhat.com> wrote:
>
> Hi team,
> I'm in the process of creating a Vagrant file which is close to the
customer's ENV.
> It is heavilly based on Viktor's beaker task.
> I use it for building and testing my code. And it is pretty important to build with
ASAN.
>
> Currently, what I do is:
> 1. Set 'ASAN_ON = 1' in rpm.mk
> 2. Run `make -f rpm.mk srpms` target
> 3. Build the RPM using `mock -q my_generated.srpm`
> 4. Install it
>
> Then I've tried running `dscreate` manually or running tests with py.test.
> Every time I have the same error here: /run/dirsrv/ns-slapd-standalone1.asan.XXXXX
>
> ==22487==LeakSanitizer has encountered a fatal error.
> ==22487==HINT: For debugging, try setting environment variable
LSAN_OPTIONS=verbosity=1:log_threads=1
> ==22487==HINT: LeakSanitizer does not work under ptrace (strace, gdb, etc)
Ludwig also recently had this issue. Looks like you're hitting this
bug:
https://github.com/google/sanitizers/issues/723
We're using posix_memalign() in a few places and LeakSanitizier can't handle it.
There is a workaround in the last comment. I did the builds for gcc8
and gcc9 in copr, both internal and fedora one, but they failed (not
related to the patch).
So I did a local build with the patch and it worked like a charm. I
will share the links to the rpms for you to try.
Which patch?
Perhaps we should review our usage of posix_memalign() or convince the
upstream to implement a proper fix for this.
Memalign is pretty important - the short version is "we can not remove it".
There are some structures in the code that rely on this for performance to guarantee that
they memory is aligned to a page boundary, or cache line boundary. In some cases it's
required to allow the atomics to work in nunc-stans (well, lfds, but the value of that
today is questionable when the rust version is possibly safer and faster).
>
> I've tried setting `export LSAN_OPTIONS=verbosity=1:log_threads=1` and run once
again.
> Same issue.
>
> Did anybody encountered the issue? Maybe, Viktor or William, could you please check?
> I'm putting the Vagrantfile to the attachments so you can reproduce.
> Just run: `ASAN=on vagrant up` from the directory with Vagrantfile.
>
> William, I think, libvirt is present on SUSE so you should have no issues with this
too...
It is, but I run on a mac so ... my setup is fun fun fun :)
I would normally run this on my home lab, but I'm a few thousand km's away ....
>
> Thanks,
> Simon
> _______________________________________________
> 389-devel mailing list -- 389-devel(a)lists.fedoraproject.org
> To unsubscribe send an email to 389-devel-leave(a)lists.fedoraproject.org
> Fedora Code of Conduct:
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
> List Guidelines:
https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives:
https://lists.fedoraproject.org/archives/list/389-devel@lists.fedoraproje...
--
Viktor
_______________________________________________
389-devel mailing list -- 389-devel(a)lists.fedoraproject.org
To unsubscribe send an email to 389-devel-leave(a)lists.fedoraproject.org
Fedora Code of Conduct:
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines:
https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives:
https://lists.fedoraproject.org/archives/list/389-devel@lists.fedoraproje...
—
Sincerely,
William Brown
Senior Software Engineer, 389 Directory Server
SUSE Labs