On 02/02/21 15:17, Valentin Schneider wrote: > That is, rather than building overlapping groups and fixing them whenever > that breaks (distance > 2), we could have: > - the local group being the child domain's span (as always) > - all non-local NUMA groups spanning a single node each, with the right sgc > cross-referencing. > > Thoughts? >
Hmph I'm thinking this can be broken by domain degeneration, as nothing will come fix up the ->sgc link and we'd be back to having a reference to a group that doesn't get updated...
