GVPR segfaults in CI on CentOS

Today I merged !1472 which subsequently failed CI on CentOS6 and CentOS7. Three of the GVPR tests segfault there.

This is somewhat confusing as (a) this MR passed CI prior to merge and (b) I don’t see any relevant GVPR changes in master that would cause this. Any ideas? I haven’t yet reverted this because it seemed like a CI fault to me at first, but if this is blocking you please feel free to revert it.

Can it be a latent bug that previously was ignored, but after your vmalloc/malloc changes (or something else) now makes GVPR crash?

Yes, good call, you’re right. ASan catches a NULL dereference when running gvpr -f cmd/gvpr/lib/color </dev/null that was not there before. Bisecting…

1 Like