Hi, On Wed, 2016-07-27 at 18:16 +0200, Helmut Grohne wrote: > > I'm wondering if I should just disable the biarch tests on all > > arches and remove the gcc-multilib Depends completly. > > I won't disagree with that. > > Personally, I'd like to see biarch/multilib to die as soon as possible. > Unfortunately, it looks like we'll have to support it for quite some > time yet. Still we don't have to add it for new architectures (like > ppc64el).
As upstream hacker I would like to see the native/biarch tests been run if at all possible. They do show if we do the right thing when a 64-bit process uses elfutils to inspect a 32-bit process running on the system. If we get that wrong I do like to know about it. But some of the native/biarch tests are a little fragile (failures can occur not just for the cross 32/64 word issue, but also because of ptrace/proc, eh_frame issues, symbol table completeness, etc.) So if there is anything we can do to make these tests easier/simpler/better please let us/upstream know. Cheers, Mark
