On Mon Jul 3, 2023 at 10:03 PM AEST, Nicholas Piggin wrote:
ppc currently silently accepts invalid real address access. Catch
these and turn them into machine checks on POWER9/10 machines.
Would there be any objections to merging this and the checkstop patch?
We could disable this one before release if it turns out to cause
breakage.
I don't think it needs to rebase, and passes clang build and make check
here. Just messed up the separator on the changelog of the checkstop
patch.
Thanks,
Nick
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
Since v1:
- Only implement this for POWER9/10. Seems like previous IBM processors
may not catch this, trying to get info.
Since v2:
- Split out from larger series since it is independent.