Date: Thu, 5 Oct 2023 12:44:51 GMT From: Mateusz Guzik <mjg@FreeBSD.org> To: src-committers@FreeBSD.org, dev-commits-src-all@FreeBSD.org, dev-commits-src-main@FreeBSD.org Subject: git: 8b622172ba44 - main - vfs cache: add 2 more optimizaiton ideas Message-ID: <202310051244.395CipkC027506@gitrepo.freebsd.org>
next in thread | raw e-mail | index | archive | help
The branch main has been updated by mjg: URL: https://cgit.FreeBSD.org/src/commit/?id=8b622172ba44f914092c1d44cecb2b101eb8e17b commit 8b622172ba44f914092c1d44cecb2b101eb8e17b Author: Mateusz Guzik <mjg@FreeBSD.org> AuthorDate: 2023-10-05 12:38:32 +0000 Commit: Mateusz Guzik <mjg@FreeBSD.org> CommitDate: 2023-10-05 12:44:43 +0000 vfs cache: add 2 more optimizaiton ideas --- sys/kern/vfs_cache.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/sys/kern/vfs_cache.c b/sys/kern/vfs_cache.c index cc2f405d5281..99e293d5a873 100644 --- a/sys/kern/vfs_cache.c +++ b/sys/kern/vfs_cache.c @@ -276,6 +276,20 @@ * It could become part of the API contract to *always* have a rootvnode set in * mnt_rootvnode. Such vnodes are annotated with VV_ROOT and vnlru would have * to be modified to always skip them. + * + * === inactive on v_usecount reaching 0 + * + * VOP_NEED_INACTIVE should not exist. Filesystems would indicate need for such + * processing with a bit in usecount. + * + * === v_holdcnt + * + * Hold count should probably get eliminated, but one can argue it is a useful + * feature. Even if so, handling of v_usecount could be decoupled from it -- + * vnlru et al would consider the vnode not-freeable if has either hold or + * usecount on it. + * + * This would eliminate 2 atomics. */ static SYSCTL_NODE(_vfs, OID_AUTO, cache, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?202310051244.395CipkC027506>