Win 2012 failures in Ninja

[Robert Relyea]

It looks like we are getting persistent Win 2012 failures in Ninja from
the following patch:

https://hg.mozilla.org/projects/nss/rev/b83ad33acd678c78c91b836ad42310e75cac3361

A quick look looks like it's a problem in an MPI test program which gets
RNG_RNGInit() added to it even though RNG is part of freebl and not part
of MPI (From a layering point of view, MPI is independent and shouldn't
be calling back into freebl).

The test program is not actually used anywhere, but it is causing ninja
Windows 2012 builds to fail, which will cause issues with any mozilla
Windows 2012. I suspect the program isn't even built in Windows 2012
Make builds, and that other platforms are tolerant with the unreferenced
symbol.

We probably need to restructure the patch in the future to keep the
proper layering, but that's not as critical. In the mean time we should
either drop the mpi test program from the ninja builds, or remove the
RNG_RNGInit call from the program so that at least Windows 2012 builds
before we release NSS 3.77.

bob

[John Schanck]

Thanks, Bob. Dennis noticed that build failure last week, and I tried to land a fix after verifying on try (https://hg.mozilla.org/projects/nss-try/rev/9c4a364239cf49a5aa7a57ba99c8b5408faaef56). Looks like I re-landed the original patch instead of the new one. We'll get this sorted today.

Cheers,

John

--
You received this message because you are subscribed to the Google Groups "dev-tec...@mozilla.org" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dev-tech-cryp...@mozilla.org.
To view this discussion on the web visit https://groups.google.com/a/mozilla.org/d/msgid/dev-tech-crypto/5246074d-ee0b-62fa-0d64-835a77644c93%40redhat.com.

[Robert Relyea]

On 3/30/22 10:09 AM, John Schanck wrote:

Thanks, Bob. Dennis noticed that build failure last week, and I tried to land a fix after verifying on try (https://hg.mozilla.org/projects/nss-try/rev/9c4a364239cf49a5aa7a57ba99c8b5408faaef56). Looks like I re-landed the original patch instead of the new one. We'll get this sorted today.

Cheers,

John

Sounds good! Watching for failures is tricky these days because there are a lot of environmental failures (container platforms too small?) you kind of have to watch multiple builds to see if the failures are 'sticky':(. I wonder if we can ask the treeherder owners to look into what may be causing our intermittent issues?

bob

To view this discussion on the web visit https://groups.google.com/a/mozilla.org/d/msgid/dev-tech-crypto/CAFgAd7HxW1ZMbZodhkdTYzmx0OgAXjD5R2oe0-WhwXGFUDeTvQ%40mail.gmail.com.