John Baldwin
John Baldwin
Replaced with upstream PR at https://github.com/freebsd/freebsd-src/pull/636.
It also violates the principle of intentional use (one of the two underlying principles of CHERI).
The other error cases in the loop after the realloc all call free, so this is more consistent.
Would maybe s/first loop cycle/previous loop iteration/ in the log message. "again" might also be confusing as you aren't calling free multiple times, so I would perhaps omit it from...
I think the superpages commit can certainly wait for the refactoring to land. The only bit I think can maybe be easily extracted as a separate commit perhaps is the...
#2120 has been merged
This seems sensible to me in light of the earlier discussion in #1083.
The arm64 fix is straightforward since arm64 supplies FAR for a capability abort. For RISC-V it is messier and the approach here is a simple fixup. Arguably it is fine...
These have all landed upstream now it looks like.
@mhorne , this is maybe fallout from your earlier commit, and if it is correct it is probably applicable to other architectures such as RISC-V and arm64 as well.