d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix for a bug caused by a race when 2 threads lookup the same

Browse Source 
file. Leave the loser's lock(s) initialized, so the reclaim logic can
unconditionally destroy them when that race occurs (or if the vfs hash
insert happened to fail for some other reason). Thanks to ups@ for a
careful review of the code.
Reported by : Kris Kennaway
This commit is contained in:
Mohan Srinivasan 2006-11-27 19:06:43 +00:00
parent 10d8cebc7a
commit 88d5725c38
Notes: svn2git 2020-12-20 02:59:44 +00:00 
svn path=/head/; revision=164684
1 changed files with 7 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
sys/nfsclient/nfs_node.c
View File

@ -148,6 +148,13 @@ nfs_nget(struct mount *mntp, nfsfh_t *fhp, int fhsize, struct nfsnode **npp, int
vp->v_bufobj.bo_ops = &buf_ops_nfs;
vp->v_data = np;
np->n_vnode = vp;
/*
* Initialize the mutex even if the vnode is going to be a loser.
* This simplifies the logic in reclaim, which can then unconditionally
* destroy the mutex (in the case of the loser, or if hash_insert happened
* to return an error no special casing is needed).
*/
mtx_init(&np->n_mtx, "NFSnode lock", NULL, MTX_DEF);
/*
* NFS supports recursive and shared locking.
*/
@ -168,7 +175,6 @@ nfs_nget(struct mount *mntp, nfsfh_t *fhp, int fhsize, struct nfsnode **npp, int
np->n_fhp = &np->n_fh;
bcopy((caddr_t)fhp, (caddr_t)np->n_fhp, fhsize);
np->n_fhsize = fhsize;
mtx_init(&np->n_mtx, "NFSnode lock", NULL, MTX_DEF);
*npp = np;
return (0);