Date: Fri, 3 Oct 2025 19:22:18 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: 94aae0513896 - main - nullfs: remove the vhold/vdrop cycle around unlock Message-ID: <202510031922.593JMIgN047299@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=94aae0513896de398a1feda6cd97f7efa2a5dc2e commit 94aae0513896de398a1feda6cd97f7efa2a5dc2e Author: Mateusz Guzik <mjg@FreeBSD.org> AuthorDate: 2025-10-01 11:09:50 +0000 Commit: Mateusz Guzik <mjg@FreeBSD.org> CommitDate: 2025-10-03 19:14:30 +0000 nullfs: remove the vhold/vdrop cycle around unlock Both lower vnode and null data are safe while the lock is held, at the same time neither is touched after unlock. While here remove stale comment about interlock handling. It is no longer legal to pass to unlock. Reviewed by: kib Tested by: pho (previous version) Differential Revision: https://reviews.freebsd.org/D38761 --- sys/fs/nullfs/null_vnops.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-) diff --git a/sys/fs/nullfs/null_vnops.c b/sys/fs/nullfs/null_vnops.c index e9d598014a2f..64e5b004abca 100644 --- a/sys/fs/nullfs/null_vnops.c +++ b/sys/fs/nullfs/null_vnops.c @@ -840,11 +840,6 @@ null_lock(struct vop_lock1_args *ap) return (error); } -/* - * We need to process our own vnode unlock and then clear the - * interlock flag as it applies only to our vnode, not the - * vnodes below us on the stack. - */ static int null_unlock(struct vop_unlock_args *ap) { @@ -853,11 +848,20 @@ null_unlock(struct vop_unlock_args *ap) struct vnode *lvp; int error; + /* + * Contrary to null_lock, we don't need to hold the vnode around + * unlock. + * + * We hold the lock, which means we can't be racing against vgone. + * + * At the same time VOP_UNLOCK promises to not touch anything after + * it finishes unlock, just like we don't. + * + * vop_stdunlock for a doomed vnode matches doomed locking in null_lock. + */ nn = VTONULL(vp); if (nn != NULL && (lvp = NULLVPTOLOWERVP(vp)) != NULL) { - vholdnz(lvp); error = VOP_UNLOCK(lvp); - vdrop(lvp); } else { error = vop_stdunlock(ap); }
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?202510031922.593JMIgN047299>