Ben Greear wrote:

With both of Stephen's patches included in the latest -rc6 source,
I re-ran the test and it seems to be working (I added printks so
that I would know the new code was being exercise

I had 2000 or so mac-vlans configured, with 10 of them being
re-configured concurrently, while also deleting groups of 20-100
mac-vlans in my test.  This was locking up reliably before,
and now it seems to be working fine.

Here's the kernel log showing the ERESTARTSYS in action.  I don't
have an easy way to check to see if the VFS (or whatever) retried
the call properly, but will let you all know if I see any indication
that isn't working.

I only saw the ipv6 fixup in my logs, but maybe my test case just
doesn't hit the other...

This looks like its working fine. Despite the non-desirable active
spinning, this seems like the best fix (actually much simpler than
I expected to be possible) at this time. If we just could avoid
the spinning when unnecessary, it would be perfect :)