I went through the series. Patch 1 correctly updates HK_TYPE_KTHREAD to reflect the post-041ee6f3727a behavior where kthreads follow HK_TYPE_DOMAIN. Keeping it mapped to HK_TYPE_KERNEL_NOISE is no longer accurate, so the alias is the right fix.
Patch 2 properly introduces RCU protection for accessing the HK_TYPE_KTHREAD/HK_TYPE_DOMAIN cpumask in IPVS. Since the mask can now change at runtime, guarding against potential UAF is required. The changes are minimal and the helper wrappers keep call sites clean. Reviewed-by: David Dull <[email protected]>
