Bug 27776
| Summary: | rpc.lockd panic with FreeBSD-5.0 | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | Base System | Reporter: | Jean-Luc.Richier <Jean-Luc.Richier> | ||||||
| Component: | kern | Assignee: | freebsd-bugs (Nobody) <bugs> | ||||||
| Status: | Closed FIXED | ||||||||
| Severity: | Affects Only Me | ||||||||
| Priority: | Normal | ||||||||
| Version: | Unspecified | ||||||||
| Hardware: | Any | ||||||||
| OS: | Any | ||||||||
| Attachments: |
|
||||||||
About the, proposed patch, I see I made a mistake as my diff contains a diff which is not for the bug, but for an attempt to prot to IPv6 The correct patch is simpler, cf the attached file -- Jean-Luc RICHIER (Jean-Luc.Richier@Imag.Fr richier@imag.fr) Laboratoire Logiciels, Systemes et Reseaux (LSR-IMAG) IMAG-CAMPUS, BP 72, F-38402 St Martin d'Heres Cedex Tel : +33 4 76 82 72 32 Fax : +33 4 76 82 72 87 State Changed From-To: open->closed Committed, thanks! Jean-Luc.Richier@imag.fr wrote: [...] > ((ansp->la_msg_ident.msg_seq != -1) && > (timevalcmp(&p->p_nlminfo->pid_start, > &ansp->la_msg_ident.pid_start, !=) || > ! p->p_nlminfo->msg_seq != ansp->la_msg_ident.msg_seq))) { > return (EPIPE); > + PROC_UNLOCK(p); > + } I'm not sure that this section is going to work too well... Cheers, -Peter -- Peter Wemm - peter@FreeBSD.org; peter@yahoo-inc.com; peter@netplex.com.au "All of this is for nothing if we don't go to the stars" - JMS/B5 |
On a FreeBSD 5-0, source dated may 2001, kernel compiled with witness option, activating rpc.lockd causes a panic (witness violation on holding locks when syscall returns) Fix: The problem is in nfslockdans in nfs/nfs_lock.c This function call pfind which return a pointer to a locked proc. But ther is no calls to PROC_UNLOCK done. A patch is: p->p_nlminfo->retcode = ansp->la_errno; p->p_nlminfo->set_getlk_pid = ansp->la_set_getlk_pid; --- 261,270 ---- ((ansp->la_msg_ident.msg_seq != -1) && (timevalcmp(&p->p_nlminfo->pid_start, &ansp->la_msg_ident.pid_start, !=) || ! p->p_nlminfo->msg_seq != ansp->la_msg_ident.msg_seq))) { return (EPIPE); + PROC_UNLOCK(p); + } p->p_nlminfo->retcode = ansp->la_errno; p->p_nlminfo->set_getlk_pid = ansp->la_set_getlk_pid; *************** (void)wakeup((void *)p->p_nlminfo); + PROC_UNLOCK(p); return (0); }--fm6L0UCm01KQHkwZUwZW5GxJAjRuj31SI225AHpVUClf0ktK Content-Type: text/plain; name="file.diff" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="file.diff" *** /usr/src/P50/sys/nfs/nfs_lock.c.DIST Sat May 26 01:36:46 2001 --- /usr/src/P50/sys/nfs/nfs_lock.c Wed May 30 11:58:06 2001 *************** *** 128,134 **** /* * XXX -- I think this is wrong for anything other AF_INET. */ ! msg.lm_addr = *(VFSTONFS(vp->v_mount)->nm_nam); msg.lm_fh_len = NFS_ISV3(vp) ? VTONFS(vp)->n_fhsize : NFSX_V2FH; bcopy(VTONFS(vp)->n_fhp, msg.lm_fh, msg.lm_fh_len); msg.lm_nfsv3 = NFS_ISV3(vp); --- 128,135 ---- /* * XXX -- I think this is wrong for anything other AF_INET. */ ! bcopy(VFSTONFS(vp->v_mount)->nm_nam, &msg.lm_addr, ! min(sizeof msg.lm_addr, VFSTONFS(vp->v_mount)->nm_nam->sa_len)); msg.lm_fh_len = NFS_ISV3(vp) ? VTONFS(vp)->n_fhsize : NFSX_V2FH; bcopy(VTONFS(vp)->n_fhp, msg.lm_fh, msg.lm_fh_len); msg.lm_nfsv3 = NFS_ISV3(vp); *************** *** 260,267 **** ((ansp->la_msg_ident.msg_seq != -1) && (timevalcmp(&p->p_nlminfo->pid_start, &ansp->la_msg_ident.pid_start, !=) || ! p->p_nlminfo->msg_seq != ansp->la_msg_ident.msg_seq))) return (EPIPE); How-To-Repeat: make a kernel with WITNESS option, current file nfs/nfs_lock.c (I checked with files dated 18 may 2001 or 26 may), start rpc.statd and rpc.lockd, and make some locks on nfs shared files The system will panic with witness error: system call nfssvc returning with mutex(s) held