Return-Path: Received: from mail-yb0-f194.google.com ([209.85.213.194]:34487 "EHLO mail-yb0-f194.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1729763AbeGMQUt (ORCPT ); Fri, 13 Jul 2018 12:20:49 -0400 MIME-Version: 1.0 In-Reply-To: <8fc65417ebda26a198e01e8c82f41ae08ab3c803.camel@kernel.org> References: <1531491744-13277-1-git-send-email-amir73il@gmail.com> <8fc65417ebda26a198e01e8c82f41ae08ab3c803.camel@kernel.org> From: Amir Goldstein Date: Fri, 13 Jul 2018 19:05:32 +0300 Message-ID: Subject: Re: [PATCH] nfsd: fix leaked file lock with nfs exported overlayfs To: Jeff Layton Cc: "J . Bruce Fields" , Miklos Szeredi , Eddie Horng , overlayfs , Linux NFS Mailing List , linux-fsdevel Content-Type: text/plain; charset="UTF-8" Sender: linux-nfs-owner@vger.kernel.org List-ID: On Fri, Jul 13, 2018 at 6:40 PM, Jeff Layton wrote: > On Fri, 2018-07-13 at 17:22 +0300, Amir Goldstein wrote: >> nfsd and lockd call vfs_lock_file() to lock/unlock the inode >> returned by locks_inode(file). >> >> Many places in nfsd/lockd code use the inode returned by >> file_inode(file) for lock manipulation. With Overlayfs, file_inode() >> (the underlying inode) is not the same object as locks_inode() (the >> overlay inode). This can result in "Leaked POSIX lock" messages >> and eventually to a kernel crash as reported by Eddie Horng: >> https://marc.info/?l=linux-unionfs&m=153086643202072&w=2 >> >> Fix all the call sites in nfsd/lockd that should use locks_inode(). >> This is a correctness bug that manifested when overlayfs gained >> NFS export support in v4.16. >> >> Reported-by: Eddie Horng >> Tested-by: Eddie Horng >> Cc: Jeff Layton >> Fixes: 8383f1748829 ("ovl: wire up NFS export operations") >> Signed-off-by: Amir Goldstein >> --- >> >> Hi Bruce, >> >> For the purpose of locks, nfsd/lockd should look at locks_inode() >> just like vfs lock functions. >> >> Hopefully, Miklos's work on stacked overlayfs file operations will >> be merged soon and locks_inode() will become the same as file_inode(), >> but we will still need this fix for stable kernels v4.16 through v4.18. >> >> Beyond testing by Eddie, I also verified no regressions when running >> nfstest with exported overlayfs. >> >> Thanks, >> Amir. >> >> fs/lockd/clntlock.c | 2 +- >> fs/lockd/clntproc.c | 2 +- >> fs/lockd/svclock.c | 16 ++++++++-------- >> fs/lockd/svcsubs.c | 4 ++-- >> fs/nfsd/nfs4state.c | 2 +- >> include/linux/lockd/lockd.h | 4 ++-- >> 6 files changed, 15 insertions(+), 15 deletions(-) >> >> diff --git a/fs/lockd/clntlock.c b/fs/lockd/clntlock.c >> index 96c1d14c18f1..c2a128678e6e 100644 >> --- a/fs/lockd/clntlock.c >> +++ b/fs/lockd/clntlock.c >> @@ -187,7 +187,7 @@ __be32 nlmclnt_grant(const struct sockaddr *addr, const struct nlm_lock *lock) >> continue; >> if (!rpc_cmp_addr(nlm_addr(block->b_host), addr)) >> continue; >> - if (nfs_compare_fh(NFS_FH(file_inode(fl_blocked->fl_file)) ,fh) != 0) >> + if (nfs_compare_fh(NFS_FH(locks_inode(fl_blocked->fl_file)), fh) != 0) >> continue; >> /* Alright, we found a lock. Set the return status >> * and wake up the caller >> diff --git a/fs/lockd/clntproc.c b/fs/lockd/clntproc.c >> index a2c0dfc6fdc0..d20b92f271c2 100644 >> --- a/fs/lockd/clntproc.c >> +++ b/fs/lockd/clntproc.c >> @@ -128,7 +128,7 @@ static void nlmclnt_setlockargs(struct nlm_rqst *req, struct file_lock *fl) >> char *nodename = req->a_host->h_rpcclnt->cl_nodename; >> >> nlmclnt_next_cookie(&argp->cookie); >> - memcpy(&lock->fh, NFS_FH(file_inode(fl->fl_file)), sizeof(struct nfs_fh)); >> + memcpy(&lock->fh, NFS_FH(locks_inode(fl->fl_file)), sizeof(struct nfs_fh)); >> lock->caller = nodename; >> lock->oh.data = req->a_owner; >> lock->oh.len = snprintf(req->a_owner, sizeof(req->a_owner), "%u@%s", >> diff --git a/fs/lockd/svclock.c b/fs/lockd/svclock.c >> index 3701bccab478..74330daeab71 100644 >> --- a/fs/lockd/svclock.c >> +++ b/fs/lockd/svclock.c >> @@ -405,8 +405,8 @@ nlmsvc_lock(struct svc_rqst *rqstp, struct nlm_file *file, >> __be32 ret; >> >> dprintk("lockd: nlmsvc_lock(%s/%ld, ty=%d, pi=%d, %Ld-%Ld, bl=%d)\n", >> - file_inode(file->f_file)->i_sb->s_id, >> - file_inode(file->f_file)->i_ino, >> + locks_inode(file->f_file)->i_sb->s_id, >> + locks_inode(file->f_file)->i_ino, >> lock->fl.fl_type, lock->fl.fl_pid, >> (long long)lock->fl.fl_start, >> (long long)lock->fl.fl_end, >> @@ -511,8 +511,8 @@ nlmsvc_testlock(struct svc_rqst *rqstp, struct nlm_file *file, >> __be32 ret; >> >> dprintk("lockd: nlmsvc_testlock(%s/%ld, ty=%d, %Ld-%Ld)\n", >> - file_inode(file->f_file)->i_sb->s_id, >> - file_inode(file->f_file)->i_ino, >> + locks_inode(file->f_file)->i_sb->s_id, >> + locks_inode(file->f_file)->i_ino, >> lock->fl.fl_type, >> (long long)lock->fl.fl_start, >> (long long)lock->fl.fl_end); >> @@ -566,8 +566,8 @@ nlmsvc_unlock(struct net *net, struct nlm_file *file, struct nlm_lock *lock) >> int error; >> >> dprintk("lockd: nlmsvc_unlock(%s/%ld, pi=%d, %Ld-%Ld)\n", >> - file_inode(file->f_file)->i_sb->s_id, >> - file_inode(file->f_file)->i_ino, >> + locks_inode(file->f_file)->i_sb->s_id, >> + locks_inode(file->f_file)->i_ino, >> lock->fl.fl_pid, >> (long long)lock->fl.fl_start, >> (long long)lock->fl.fl_end); >> @@ -595,8 +595,8 @@ nlmsvc_cancel_blocked(struct net *net, struct nlm_file *file, struct nlm_lock *l >> int status = 0; >> >> dprintk("lockd: nlmsvc_cancel(%s/%ld, pi=%d, %Ld-%Ld)\n", >> - file_inode(file->f_file)->i_sb->s_id, >> - file_inode(file->f_file)->i_ino, >> + locks_inode(file->f_file)->i_sb->s_id, >> + locks_inode(file->f_file)->i_ino, >> lock->fl.fl_pid, >> (long long)lock->fl.fl_start, >> (long long)lock->fl.fl_end); >> diff --git a/fs/lockd/svcsubs.c b/fs/lockd/svcsubs.c >> index 4ec3d6e03e76..899360ba3b84 100644 >> --- a/fs/lockd/svcsubs.c >> +++ b/fs/lockd/svcsubs.c >> @@ -44,7 +44,7 @@ static inline void nlm_debug_print_fh(char *msg, struct nfs_fh *f) >> >> static inline void nlm_debug_print_file(char *msg, struct nlm_file *file) >> { >> - struct inode *inode = file_inode(file->f_file); >> + struct inode *inode = locks_inode(file->f_file); >> >> dprintk("lockd: %s %s/%ld\n", >> msg, inode->i_sb->s_id, inode->i_ino); >> @@ -414,7 +414,7 @@ nlmsvc_match_sb(void *datap, struct nlm_file *file) >> { >> struct super_block *sb = datap; >> >> - return sb == file_inode(file->f_file)->i_sb; >> + return sb == locks_inode(file->f_file)->i_sb; >> } >> >> /** >> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c >> index 857141446d6b..4a17fad93411 100644 >> --- a/fs/nfsd/nfs4state.c >> +++ b/fs/nfsd/nfs4state.c >> @@ -6293,7 +6293,7 @@ check_for_locks(struct nfs4_file *fp, struct nfs4_lockowner *lowner) >> return status; >> } >> >> - inode = file_inode(filp); >> + inode = locks_inode(filp); >> flctx = inode->i_flctx; >> >> if (flctx && !list_empty_careful(&flctx->flc_posix)) { >> diff --git a/include/linux/lockd/lockd.h b/include/linux/lockd/lockd.h >> index 4fd95dbeb52f..b065ef406770 100644 >> --- a/include/linux/lockd/lockd.h >> +++ b/include/linux/lockd/lockd.h >> @@ -299,7 +299,7 @@ int nlmsvc_unlock_all_by_ip(struct sockaddr *server_addr); >> >> static inline struct inode *nlmsvc_file_inode(struct nlm_file *file) >> { >> - return file_inode(file->f_file); >> + return locks_inode(file->f_file); >> } >> > > fwiw, I think the above delta turns out to be the real fix here, as that > affects nlm_file_inuse. > >> static inline int __nlm_privileged_request4(const struct sockaddr *sap) >> @@ -359,7 +359,7 @@ static inline int nlm_privileged_requester(const struct svc_rqst *rqstp) >> static inline int nlm_compare_locks(const struct file_lock *fl1, >> const struct file_lock *fl2) >> { >> - return file_inode(fl1->fl_file) == file_inode(fl2->fl_file) >> + return locks_inode(fl1->fl_file) == locks_inode(fl2->fl_file) >> && fl1->fl_pid == fl2->fl_pid >> && fl1->fl_owner == fl2->fl_owner >> && fl1->fl_start == fl2->fl_start > > I think this looks OK. I was a bit hesitant as I wasn't sure whether > this would change the filehandles presented by the nfsd, but I don't > think it will. > Well, if nfsd were to ask overlayfs to encode a file handle, nfsd would need to pass in an overlay inode (a.k.a. locks_inode()) not an underlying inode, so if anything, this patch reduces the chance of file handle encoding bugs. In any case, we will all be better off when file_inode() is an overlay inode. > Reviewed-by: Jeff Layton Thanks, Amir.