> On 5. Mar 2020, at 23:48, Andrew Doran <ad%netbsd.org@localhost> wrote: > > Hi, > > I'd like to merge the changes on the ad-namecache branch, and would > appreciate any code review. The motivation for these changes is, as > you might guess, performance. I put an overview below. I won't be > merging this for a couple of weeks in any case. [snip] > vfs_vnode.c: > > Namecache related changes, and the change to not block new vnode > references across VOP_INACTIVE() mentioned on tech-kern. I don't > think this blocking is needed at least for well behaved FSes like > ffs & tmpfs. I suppose this blocking is no longer needed. It originates from the early steps to clean up the vnode lifecycle. If anything goes wrong it is quite simple to undo this change. Blocking further references should only be needed for vrecycle(). We have to terminate vrelel() early if we got new references, diff attached. -- J. Hannken-Illjes - hannken%eis.cs.tu-bs.de@localhost - TU Braunschweig
Attachment:
addendum.diff
Description: Binary data
Attachment:
signature.asc
Description: Message signed with OpenPGP