Don't use LK_DRAIN before calling VOP_FSYNC() in the panic case for

audit trail failure -- this doesn't contribute anything, and might
arguably be wrong.

MFC after:	1 week
Requested by:	attilio
This commit is contained in:
Robert Watson 2008-05-21 13:05:06 +00:00
parent 82f4d64035
commit bf7baa9eca
Notes: svn2git 2020-12-20 02:59:44 +00:00
svn path=/head/; revision=179176

View File

@ -242,7 +242,7 @@ audit_record_write(struct vnode *vp, struct ucred *cred, void *data,
*/
if (audit_in_failure) {
if (audit_q_len == 0 && audit_pre_q_len == 0) {
VOP_LOCK(vp, LK_DRAIN | LK_INTERLOCK);
VOP_LOCK(vp, LK_EXCLUSIVE | LK_RETRY);
(void)VOP_FSYNC(vp, MNT_WAIT, curthread);
VOP_UNLOCK(vp, 0);
panic("Audit store overflow; record queue drained.");