From 8c39a39e28b86a4021d9be314ce01019bafa5fdc Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 19 Jul 2019 13:48:44 -0400 Subject: NFSv4: Fix a credential refcount leak in nfs41_check_delegation_stateid It is unsafe to dereference delegation outside the rcu lock, and in any case, the refcount is guaranteed held if cred is non-zero. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 39896afc6edf..a6d73609b163 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -2778,8 +2778,7 @@ static void nfs41_check_delegation_stateid(struct nfs4_state *state) if (status == -NFS4ERR_EXPIRED || status == -NFS4ERR_BAD_STATEID) nfs_finish_clear_delegation_stateid(state, &stateid); - if (delegation->cred) - put_cred(cred); + put_cred(cred); } /** -- cgit v1.2.3 From 5eb8d18ca0e001c6055da2b7f30d8f6dca23a44f Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 19 Jul 2019 14:08:37 -0400 Subject: NFSv4: Fix delegation state recovery Once we clear the NFS_DELEGATED_STATE flag, we're telling nfs_delegation_claim_opens() that we're done recovering all open state for that stateid, so we really need to ensure that we test for all open modes that are currently cached and recover them before exiting nfs4_open_delegation_recall(). Fixes: 24311f884189d ("NFSv4: Recovery of recalled read delegations...") Signed-off-by: Trond Myklebust Cc: stable@vger.kernel.org # v4.3+ --- fs/nfs/delegation.c | 2 +- fs/nfs/delegation.h | 2 +- fs/nfs/nfs4proc.c | 25 ++++++++++++------------- 3 files changed, 14 insertions(+), 15 deletions(-) (limited to 'fs') diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c index 0ff3facf81da..0af854cce8ff 100644 --- a/fs/nfs/delegation.c +++ b/fs/nfs/delegation.c @@ -153,7 +153,7 @@ again: /* Block nfs4_proc_unlck */ mutex_lock(&sp->so_delegreturn_mutex); seq = raw_seqcount_begin(&sp->so_reclaim_seqcount); - err = nfs4_open_delegation_recall(ctx, state, stateid, type); + err = nfs4_open_delegation_recall(ctx, state, stateid); if (!err) err = nfs_delegation_claim_locks(state, stateid); if (!err && read_seqcount_retry(&sp->so_reclaim_seqcount, seq)) diff --git a/fs/nfs/delegation.h b/fs/nfs/delegation.h index 5799777df5ec..9eb87ae4c982 100644 --- a/fs/nfs/delegation.h +++ b/fs/nfs/delegation.h @@ -63,7 +63,7 @@ void nfs_reap_expired_delegations(struct nfs_client *clp); /* NFSv4 delegation-related procedures */ int nfs4_proc_delegreturn(struct inode *inode, const struct cred *cred, const nfs4_stateid *stateid, int issync); -int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid, fmode_t type); +int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid); int nfs4_lock_delegation_recall(struct file_lock *fl, struct nfs4_state *state, const nfs4_stateid *stateid); bool nfs4_copy_delegation_stateid(struct inode *inode, fmode_t flags, nfs4_stateid *dst, const struct cred **cred); bool nfs4_refresh_delegation_stateid(nfs4_stateid *dst, struct inode *inode); diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index a6d73609b163..21e3c159bc69 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -2177,12 +2177,10 @@ static int nfs4_handle_delegation_recall_error(struct nfs_server *server, struct case -NFS4ERR_BAD_HIGH_SLOT: case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION: case -NFS4ERR_DEADSESSION: - set_bit(NFS_DELEGATED_STATE, &state->flags); nfs4_schedule_session_recovery(server->nfs_client->cl_session, err); return -EAGAIN; case -NFS4ERR_STALE_CLIENTID: case -NFS4ERR_STALE_STATEID: - set_bit(NFS_DELEGATED_STATE, &state->flags); /* Don't recall a delegation if it was lost */ nfs4_schedule_lease_recovery(server->nfs_client); return -EAGAIN; @@ -2203,7 +2201,6 @@ static int nfs4_handle_delegation_recall_error(struct nfs_server *server, struct return -EAGAIN; case -NFS4ERR_DELAY: case -NFS4ERR_GRACE: - set_bit(NFS_DELEGATED_STATE, &state->flags); ssleep(1); return -EAGAIN; case -ENOMEM: @@ -2219,8 +2216,7 @@ static int nfs4_handle_delegation_recall_error(struct nfs_server *server, struct } int nfs4_open_delegation_recall(struct nfs_open_context *ctx, - struct nfs4_state *state, const nfs4_stateid *stateid, - fmode_t type) + struct nfs4_state *state, const nfs4_stateid *stateid) { struct nfs_server *server = NFS_SERVER(state->inode); struct nfs4_opendata *opendata; @@ -2231,20 +2227,23 @@ int nfs4_open_delegation_recall(struct nfs_open_context *ctx, if (IS_ERR(opendata)) return PTR_ERR(opendata); nfs4_stateid_copy(&opendata->o_arg.u.delegation, stateid); - nfs_state_clear_delegation(state); - switch (type & (FMODE_READ|FMODE_WRITE)) { - case FMODE_READ|FMODE_WRITE: - case FMODE_WRITE: + if (!test_bit(NFS_O_RDWR_STATE, &state->flags)) { err = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE); if (err) - break; + goto out; + } + if (!test_bit(NFS_O_WRONLY_STATE, &state->flags)) { err = nfs4_open_recover_helper(opendata, FMODE_WRITE); if (err) - break; - /* Fall through */ - case FMODE_READ: + goto out; + } + if (!test_bit(NFS_O_RDONLY_STATE, &state->flags)) { err = nfs4_open_recover_helper(opendata, FMODE_READ); + if (err) + goto out; } + nfs_state_clear_delegation(state); +out: nfs4_opendata_put(opendata); return nfs4_handle_delegation_recall_error(server, state, stateid, NULL, err); } -- cgit v1.2.3 From 86dbd08b32838dc8940ccab2eebd47cf80224494 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 22 Jul 2019 09:44:04 +0100 Subject: NFSv4: Print an error in the syslog when state is marked as irrecoverable When error recovery fails due to a fatal error on the server, ensure we log it in the syslog. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4state.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index 9afd051a4876..a71a61e5fe2c 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -1463,7 +1463,7 @@ void nfs_inode_find_state_and_recover(struct inode *inode, nfs4_schedule_state_manager(clp); } -static void nfs4_state_mark_open_context_bad(struct nfs4_state *state) +static void nfs4_state_mark_open_context_bad(struct nfs4_state *state, int err) { struct inode *inode = state->inode; struct nfs_inode *nfsi = NFS_I(inode); @@ -1474,6 +1474,8 @@ static void nfs4_state_mark_open_context_bad(struct nfs4_state *state) if (ctx->state != state) continue; set_bit(NFS_CONTEXT_BAD, &ctx->flags); + pr_warn("NFSv4: state recovery failed for open file %pd2, " + "error = %d\n", ctx->dentry, err); } rcu_read_unlock(); } @@ -1481,7 +1483,7 @@ static void nfs4_state_mark_open_context_bad(struct nfs4_state *state) static void nfs4_state_mark_recovery_failed(struct nfs4_state *state, int error) { set_bit(NFS_STATE_RECOVERY_FAILED, &state->flags); - nfs4_state_mark_open_context_bad(state); + nfs4_state_mark_open_context_bad(state, error); } -- cgit v1.2.3 From c34fae003c79570b6c930b425fea3f0b7b1e7056 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 22 Jul 2019 09:54:29 +0100 Subject: NFSv4: When recovering state fails with EAGAIN, retry the same recovery If the server returns with EAGAIN when we're trying to recover from a server reboot, we currently delay for 1 second, but then mark the stateid as needing recovery after the grace period has expired. Instead, we should just retry the same recovery process immediately after the 1 second delay. Break out of the loop after 10 retries. Fixes: 35a61606a612 ("NFS: Reduce indentation of the switch statement...") Signed-off-by: Trond Myklebust --- fs/nfs/nfs4state.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index a71a61e5fe2c..d03b9cf42bd0 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -1607,6 +1607,7 @@ static int __nfs4_reclaim_open_state(struct nfs4_state_owner *sp, struct nfs4_st static int nfs4_reclaim_open_state(struct nfs4_state_owner *sp, const struct nfs4_state_recovery_ops *ops) { struct nfs4_state *state; + unsigned int loop = 0; int status = 0; /* Note: we rely on the sp->so_states list being ordered @@ -1633,8 +1634,10 @@ restart: switch (status) { default: - if (status >= 0) + if (status >= 0) { + loop = 0; break; + } printk(KERN_ERR "NFS: %s: unhandled error %d\n", __func__, status); /* Fall through */ case -ENOENT: @@ -1648,6 +1651,10 @@ restart: break; case -EAGAIN: ssleep(1); + if (loop++ < 10) { + set_bit(ops->state_flag_bit, &state->flags); + break; + } /* Fall through */ case -NFS4ERR_ADMIN_REVOKED: case -NFS4ERR_STALE_STATEID: -- cgit v1.2.3 From 731c74dd987e4f1f3179223314c039ce38dd4f19 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 22 Jul 2019 18:06:17 +0100 Subject: NFSv4: Report the error from nfs4_select_rw_stateid() In pnfs_update_layout() ensure that we do report any fatal errors from nfs4_select_rw_stateid(). Fixes: d9aba2b40de6 ("NFSv4: Don't use the zero stateid with layoutget") Signed-off-by: Trond Myklebust --- fs/nfs/pnfs.c | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c index 75bd5b552ba4..4525d5acae38 100644 --- a/fs/nfs/pnfs.c +++ b/fs/nfs/pnfs.c @@ -1903,12 +1903,6 @@ lookup_again: goto out_unlock; } - if (!nfs4_valid_open_stateid(ctx->state)) { - trace_pnfs_update_layout(ino, pos, count, iomode, lo, lseg, - PNFS_UPDATE_LAYOUT_INVALID_OPEN); - goto out_unlock; - } - /* * Choose a stateid for the LAYOUTGET. If we don't have a layout * stateid, or it has been invalidated, then we must use the open @@ -1939,6 +1933,7 @@ lookup_again: iomode == IOMODE_RW ? FMODE_WRITE : FMODE_READ, NULL, &stateid, NULL); if (status != 0) { + lseg = ERR_PTR(status); trace_pnfs_update_layout(ino, pos, count, iomode, lo, lseg, PNFS_UPDATE_LAYOUT_INVALID_OPEN); -- cgit v1.2.3 From 27a30cf64a5cbe2105e4ff9613246b32d584766a Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 22 Jul 2019 18:32:59 +0100 Subject: NFSv4.1: Fix open stateid recovery The logic for checking in nfs41_check_open_stateid() whether the state is supported by a delegation is inverted. In addition, it makes more sense to perform that check before we check for expired locks. Fixes: 8a64c4ef106d1 ("NFSv4.1: Even if the stateid is OK,...") Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 65 ++++++++++++++++++++++++++++++++----------------------- 1 file changed, 38 insertions(+), 27 deletions(-) (limited to 'fs') diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 21e3c159bc69..c9e14ce0b7b2 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -1683,6 +1683,14 @@ static void nfs_state_set_open_stateid(struct nfs4_state *state, write_sequnlock(&state->seqlock); } +static void nfs_state_clear_open_state_flags(struct nfs4_state *state) +{ + clear_bit(NFS_O_RDWR_STATE, &state->flags); + clear_bit(NFS_O_WRONLY_STATE, &state->flags); + clear_bit(NFS_O_RDONLY_STATE, &state->flags); + clear_bit(NFS_OPEN_STATE, &state->flags); +} + static void nfs_state_set_delegation(struct nfs4_state *state, const nfs4_stateid *deleg_stateid, fmode_t fmode) @@ -2074,13 +2082,7 @@ static int nfs4_open_recover(struct nfs4_opendata *opendata, struct nfs4_state * { int ret; - /* Don't trigger recovery in nfs_test_and_clear_all_open_stateid */ - clear_bit(NFS_O_RDWR_STATE, &state->flags); - clear_bit(NFS_O_WRONLY_STATE, &state->flags); - clear_bit(NFS_O_RDONLY_STATE, &state->flags); /* memory barrier prior to reading state->n_* */ - clear_bit(NFS_DELEGATED_STATE, &state->flags); - clear_bit(NFS_OPEN_STATE, &state->flags); smp_rmb(); ret = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE); if (ret != 0) @@ -2156,6 +2158,8 @@ static int nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *sta ctx = nfs4_state_find_open_context(state); if (IS_ERR(ctx)) return -EAGAIN; + clear_bit(NFS_DELEGATED_STATE, &state->flags); + nfs_state_clear_open_state_flags(state); ret = nfs4_do_open_reclaim(ctx, state); put_nfs_open_context(ctx); return ret; @@ -2697,6 +2701,7 @@ static int nfs40_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *st { /* NFSv4.0 doesn't allow for delegation recovery on open expire */ nfs40_clear_delegation_stateid(state); + nfs_state_clear_open_state_flags(state); return nfs4_open_expired(sp, state); } @@ -2739,13 +2744,13 @@ out_free: return -NFS4ERR_EXPIRED; } -static void nfs41_check_delegation_stateid(struct nfs4_state *state) +static int nfs41_check_delegation_stateid(struct nfs4_state *state) { struct nfs_server *server = NFS_SERVER(state->inode); nfs4_stateid stateid; struct nfs_delegation *delegation; const struct cred *cred = NULL; - int status; + int status, ret = NFS_OK; /* Get the delegation credential for use by test/free_stateid */ rcu_read_lock(); @@ -2753,20 +2758,15 @@ static void nfs41_check_delegation_stateid(struct nfs4_state *state) if (delegation == NULL) { rcu_read_unlock(); nfs_state_clear_delegation(state); - return; + return NFS_OK; } nfs4_stateid_copy(&stateid, &delegation->stateid); - if (test_bit(NFS_DELEGATION_REVOKED, &delegation->flags)) { - rcu_read_unlock(); - nfs_state_clear_delegation(state); - return; - } if (!test_and_clear_bit(NFS_DELEGATION_TEST_EXPIRED, &delegation->flags)) { rcu_read_unlock(); - return; + return NFS_OK; } if (delegation->cred) @@ -2776,8 +2776,24 @@ static void nfs41_check_delegation_stateid(struct nfs4_state *state) trace_nfs4_test_delegation_stateid(state, NULL, status); if (status == -NFS4ERR_EXPIRED || status == -NFS4ERR_BAD_STATEID) nfs_finish_clear_delegation_stateid(state, &stateid); + else + ret = status; put_cred(cred); + return ret; +} + +static void nfs41_delegation_recover_stateid(struct nfs4_state *state) +{ + nfs4_stateid tmp; + + if (test_bit(NFS_DELEGATED_STATE, &state->flags) && + nfs4_copy_delegation_stateid(state->inode, state->state, + &tmp, NULL) && + nfs4_stateid_match_other(&state->stateid, &tmp)) + nfs_state_set_delegation(state, &tmp, state->state); + else + nfs_state_clear_delegation(state); } /** @@ -2847,21 +2863,12 @@ static int nfs41_check_open_stateid(struct nfs4_state *state) const struct cred *cred = state->owner->so_cred; int status; - if (test_bit(NFS_OPEN_STATE, &state->flags) == 0) { - if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0) { - if (nfs4_have_delegation(state->inode, state->state)) - return NFS_OK; - return -NFS4ERR_OPENMODE; - } + if (test_bit(NFS_OPEN_STATE, &state->flags) == 0) return -NFS4ERR_BAD_STATEID; - } status = nfs41_test_and_free_expired_stateid(server, stateid, cred); trace_nfs4_test_open_stateid(state, NULL, status); if (status == -NFS4ERR_EXPIRED || status == -NFS4ERR_BAD_STATEID) { - clear_bit(NFS_O_RDONLY_STATE, &state->flags); - clear_bit(NFS_O_WRONLY_STATE, &state->flags); - clear_bit(NFS_O_RDWR_STATE, &state->flags); - clear_bit(NFS_OPEN_STATE, &state->flags); + nfs_state_clear_open_state_flags(state); stateid->type = NFS4_INVALID_STATEID_TYPE; return status; } @@ -2874,7 +2881,11 @@ static int nfs41_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *st { int status; - nfs41_check_delegation_stateid(state); + status = nfs41_check_delegation_stateid(state); + if (status != NFS_OK) + return status; + nfs41_delegation_recover_stateid(state); + status = nfs41_check_expired_locks(state); if (status != NFS_OK) return status; -- cgit v1.2.3 From ad11408970df79d5f481aa9964e91f183133424c Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 26 Jul 2019 14:40:53 +0100 Subject: NFSv4.1: Only reap expired delegations Fix nfs_reap_expired_delegations() to ensure that we only reap delegations that are actually expired, rather than triggering on random errors. Fixes: 45870d6909d5a ("NFSv4.1: Test delegation stateids when server...") Signed-off-by: Trond Myklebust --- fs/nfs/delegation.c | 23 +++++++++++++++++------ 1 file changed, 17 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c index 0af854cce8ff..071b90a45933 100644 --- a/fs/nfs/delegation.c +++ b/fs/nfs/delegation.c @@ -1046,6 +1046,22 @@ void nfs_test_expired_all_delegations(struct nfs_client *clp) nfs4_schedule_state_manager(clp); } +static void +nfs_delegation_test_free_expired(struct inode *inode, + nfs4_stateid *stateid, + const struct cred *cred) +{ + struct nfs_server *server = NFS_SERVER(inode); + const struct nfs4_minor_version_ops *ops = server->nfs_client->cl_mvops; + int status; + + if (!cred) + return; + status = ops->test_and_free_expired(server, stateid, cred); + if (status == -NFS4ERR_EXPIRED || status == -NFS4ERR_BAD_STATEID) + nfs_remove_bad_delegation(inode, stateid); +} + /** * nfs_reap_expired_delegations - reap expired delegations * @clp: nfs_client to process @@ -1057,7 +1073,6 @@ void nfs_test_expired_all_delegations(struct nfs_client *clp) */ void nfs_reap_expired_delegations(struct nfs_client *clp) { - const struct nfs4_minor_version_ops *ops = clp->cl_mvops; struct nfs_delegation *delegation; struct nfs_server *server; struct inode *inode; @@ -1088,11 +1103,7 @@ restart: nfs4_stateid_copy(&stateid, &delegation->stateid); clear_bit(NFS_DELEGATION_TEST_EXPIRED, &delegation->flags); rcu_read_unlock(); - if (cred != NULL && - ops->test_and_free_expired(server, &stateid, cred) < 0) { - nfs_revoke_delegation(inode, &stateid); - nfs_inode_find_state_and_recover(inode, &stateid); - } + nfs_delegation_test_free_expired(inode, &stateid, cred); put_cred(cred); if (nfs4_server_rebooted(clp)) { nfs_inode_mark_test_expired_delegation(server,inode); -- cgit v1.2.3 From e3c8dc761ead061da2220ee8f8132f729ac3ddfe Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 29 Jul 2019 18:25:00 +0100 Subject: NFSv4: Check the return value of update_open_stateid() Ensure that we always check the return value of update_open_stateid() so that we can retry if the update of local state failed. This fixes infinite looping on state recovery. Fixes: e23008ec81ef3 ("NFSv4 reduce attribute requests for open reclaim") Signed-off-by: Trond Myklebust Cc: stable@vger.kernel.org # v3.7+ --- fs/nfs/nfs4proc.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index c9e14ce0b7b2..3e0b93f2b61a 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -1915,8 +1915,9 @@ _nfs4_opendata_reclaim_to_nfs4_state(struct nfs4_opendata *data) if (data->o_res.delegation_type != 0) nfs4_opendata_check_deleg(data, state); update: - update_open_stateid(state, &data->o_res.stateid, NULL, - data->o_arg.fmode); + if (!update_open_stateid(state, &data->o_res.stateid, + NULL, data->o_arg.fmode)) + return ERR_PTR(-EAGAIN); refcount_inc(&state->count); return state; @@ -1981,8 +1982,11 @@ _nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data) if (data->o_res.delegation_type != 0) nfs4_opendata_check_deleg(data, state); - update_open_stateid(state, &data->o_res.stateid, NULL, - data->o_arg.fmode); + if (!update_open_stateid(state, &data->o_res.stateid, + NULL, data->o_arg.fmode)) { + nfs4_put_open_state(state); + state = ERR_PTR(-EAGAIN); + } out: nfs_release_seqid(data->o_arg.seqid); return state; -- cgit v1.2.3 From c77e22834ae9a11891cb613bd9a551be1b94f2bc Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sat, 3 Aug 2019 10:11:27 -0400 Subject: NFSv4: Fix a potential sleep while atomic in nfs4_do_reclaim() John Hubbard reports seeing the following stack trace: nfs4_do_reclaim rcu_read_lock /* we are now in_atomic() and must not sleep */ nfs4_purge_state_owners nfs4_free_state_owner nfs4_destroy_seqid_counter rpc_destroy_wait_queue cancel_delayed_work_sync __cancel_work_timer __flush_work start_flush_work might_sleep: (kernel/workqueue.c:2975: BUG) The solution is to separate out the freeing of the state owners from nfs4_purge_state_owners(), and perform that outside the atomic context. Reported-by: John Hubbard Fixes: 0aaaf5c424c7f ("NFS: Cache state owners after files are closed") Signed-off-by: Trond Myklebust --- fs/nfs/nfs4_fs.h | 3 ++- fs/nfs/nfs4client.c | 5 ++++- fs/nfs/nfs4state.c | 27 ++++++++++++++++++++++----- 3 files changed, 28 insertions(+), 7 deletions(-) (limited to 'fs') diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h index d778dad9a75e..3564da1ba8a1 100644 --- a/fs/nfs/nfs4_fs.h +++ b/fs/nfs/nfs4_fs.h @@ -465,7 +465,8 @@ static inline void nfs4_schedule_session_recovery(struct nfs4_session *session, extern struct nfs4_state_owner *nfs4_get_state_owner(struct nfs_server *, const struct cred *, gfp_t); extern void nfs4_put_state_owner(struct nfs4_state_owner *); -extern void nfs4_purge_state_owners(struct nfs_server *); +extern void nfs4_purge_state_owners(struct nfs_server *, struct list_head *); +extern void nfs4_free_state_owners(struct list_head *head); extern struct nfs4_state * nfs4_get_open_state(struct inode *, struct nfs4_state_owner *); extern void nfs4_put_open_state(struct nfs4_state *); extern void nfs4_close_state(struct nfs4_state *, fmode_t); diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c index 616393a01c06..da6204025a2d 100644 --- a/fs/nfs/nfs4client.c +++ b/fs/nfs/nfs4client.c @@ -758,9 +758,12 @@ out: static void nfs4_destroy_server(struct nfs_server *server) { + LIST_HEAD(freeme); + nfs_server_return_all_delegations(server); unset_pnfs_layoutdriver(server); - nfs4_purge_state_owners(server); + nfs4_purge_state_owners(server, &freeme); + nfs4_free_state_owners(&freeme); } /* diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index d03b9cf42bd0..a4e866b2b43b 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -624,24 +624,39 @@ void nfs4_put_state_owner(struct nfs4_state_owner *sp) /** * nfs4_purge_state_owners - Release all cached state owners * @server: nfs_server with cached state owners to release + * @head: resulting list of state owners * * Called at umount time. Remaining state owners will be on * the LRU with ref count of zero. + * Note that the state owners are not freed, but are added + * to the list @head, which can later be used as an argument + * to nfs4_free_state_owners. */ -void nfs4_purge_state_owners(struct nfs_server *server) +void nfs4_purge_state_owners(struct nfs_server *server, struct list_head *head) { struct nfs_client *clp = server->nfs_client; struct nfs4_state_owner *sp, *tmp; - LIST_HEAD(doomed); spin_lock(&clp->cl_lock); list_for_each_entry_safe(sp, tmp, &server->state_owners_lru, so_lru) { - list_move(&sp->so_lru, &doomed); + list_move(&sp->so_lru, head); nfs4_remove_state_owner_locked(sp); } spin_unlock(&clp->cl_lock); +} - list_for_each_entry_safe(sp, tmp, &doomed, so_lru) { +/** + * nfs4_purge_state_owners - Release all cached state owners + * @head: resulting list of state owners + * + * Frees a list of state owners that was generated by + * nfs4_purge_state_owners + */ +void nfs4_free_state_owners(struct list_head *head) +{ + struct nfs4_state_owner *sp, *tmp; + + list_for_each_entry_safe(sp, tmp, head, so_lru) { list_del(&sp->so_lru); nfs4_free_state_owner(sp); } @@ -1865,12 +1880,13 @@ static int nfs4_do_reclaim(struct nfs_client *clp, const struct nfs4_state_recov struct nfs4_state_owner *sp; struct nfs_server *server; struct rb_node *pos; + LIST_HEAD(freeme); int status = 0; restart: rcu_read_lock(); list_for_each_entry_rcu(server, &clp->cl_superblocks, client_link) { - nfs4_purge_state_owners(server); + nfs4_purge_state_owners(server, &freeme); spin_lock(&clp->cl_lock); for (pos = rb_first(&server->state_owners); pos != NULL; @@ -1899,6 +1915,7 @@ restart: spin_unlock(&clp->cl_lock); } rcu_read_unlock(); + nfs4_free_state_owners(&freeme); return 0; } -- cgit v1.2.3 From 09a54f0ebfe263bc27c90bbd80187b9a93283887 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sat, 3 Aug 2019 10:28:18 -0400 Subject: NFSv4: Fix an Oops in nfs4_do_setattr If the user specifies an open mode of 3, then we don't have a NFSv4 state attached to the context, and so we Oops when we try to dereference it. Reported-by: Olga Kornievskaia Fixes: 29b59f9416937 ("NFSv4: change nfs4_do_setattr to take...") Signed-off-by: Trond Myklebust Cc: stable@vger.kernel.org # v4.10: 991eedb1371dc: NFSv4: Only pass the... Cc: stable@vger.kernel.org # v4.10+ --- fs/nfs/nfs4proc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 3e0b93f2b61a..12b2b65ad8a8 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -3214,7 +3214,7 @@ static int _nfs4_do_setattr(struct inode *inode, if (nfs4_copy_delegation_stateid(inode, FMODE_WRITE, &arg->stateid, &delegation_cred)) { /* Use that stateid */ - } else if (ctx != NULL) { + } else if (ctx != NULL && ctx->state) { struct nfs_lock_context *l_ctx; if (!nfs4_valid_open_stateid(ctx->state)) return -EBADF; -- cgit v1.2.3 From dea1bb35c5f35e0577cfc61f79261d80b8715221 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sat, 3 Aug 2019 13:39:24 -0400 Subject: NFS: Fix regression whereby fscache errors are appearing on 'nofsc' mounts People are reporing seeing fscache errors being reported concerning duplicate cookies even in cases where they are not setting up fscache at all. The rule needs to be that if fscache is not enabled, then it should have no side effects at all. To ensure this is the case, we disable fscache completely on all superblocks for which the 'fsc' mount option was not set. In order to avoid issues with '-oremount', we also disable the ability to turn fscache on via remount. Fixes: f1fe29b4a02d ("NFS: Use i_writecount to control whether...") Link: https://bugzilla.kernel.org/show_bug.cgi?id=200145 Signed-off-by: Trond Myklebust Cc: Steve Dickson Cc: David Howells --- fs/nfs/fscache.c | 7 ++++++- fs/nfs/fscache.h | 2 +- fs/nfs/super.c | 1 + 3 files changed, 8 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/nfs/fscache.c b/fs/nfs/fscache.c index 53507aa96b0b..3800ab6f08fa 100644 --- a/fs/nfs/fscache.c +++ b/fs/nfs/fscache.c @@ -114,6 +114,10 @@ void nfs_fscache_get_super_cookie(struct super_block *sb, const char *uniq, int struct rb_node **p, *parent; int diff; + nfss->fscache_key = NULL; + nfss->fscache = NULL; + if (!(nfss->options & NFS_OPTION_FSCACHE)) + return; if (!uniq) { uniq = ""; ulen = 1; @@ -226,10 +230,11 @@ void nfs_fscache_release_super_cookie(struct super_block *sb) void nfs_fscache_init_inode(struct inode *inode) { struct nfs_fscache_inode_auxdata auxdata; + struct nfs_server *nfss = NFS_SERVER(inode); struct nfs_inode *nfsi = NFS_I(inode); nfsi->fscache = NULL; - if (!S_ISREG(inode->i_mode)) + if (!(nfss->fscache && S_ISREG(inode->i_mode))) return; memset(&auxdata, 0, sizeof(auxdata)); diff --git a/fs/nfs/fscache.h b/fs/nfs/fscache.h index 25a75e40d91d..ad041cfbf9ec 100644 --- a/fs/nfs/fscache.h +++ b/fs/nfs/fscache.h @@ -182,7 +182,7 @@ static inline void nfs_fscache_wait_on_invalidate(struct inode *inode) */ static inline const char *nfs_server_fscache_state(struct nfs_server *server) { - if (server->fscache && (server->options & NFS_OPTION_FSCACHE)) + if (server->fscache) return "yes"; return "no "; } diff --git a/fs/nfs/super.c b/fs/nfs/super.c index 628631e2e34f..703f595dce90 100644 --- a/fs/nfs/super.c +++ b/fs/nfs/super.c @@ -2260,6 +2260,7 @@ nfs_compare_remount_data(struct nfs_server *nfss, data->acdirmin != nfss->acdirmin / HZ || data->acdirmax != nfss->acdirmax / HZ || data->timeo != (10U * nfss->client->cl_timeout->to_initval / HZ) || + (data->options & NFS_OPTION_FSCACHE) != (nfss->options & NFS_OPTION_FSCACHE) || data->nfs_server.port != nfss->port || data->nfs_server.addrlen != nfss->nfs_client->cl_addrlen || !rpc_cmp_addr((struct sockaddr *)&data->nfs_server.address, -- cgit v1.2.3 From 67e7b52d44e3d539dfbfcd866c3d3d69da23a909 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 7 Aug 2019 07:31:27 -0400 Subject: NFSv4: Ensure state recovery handles ETIMEDOUT correctly Ensure that the state recovery code handles ETIMEDOUT correctly, and also that we set RPC_TASK_TIMEOUT when recovering open state. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 2 ++ fs/nfs/nfs4state.c | 7 +++++-- 2 files changed, 7 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 12b2b65ad8a8..1406858bae6c 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -2179,6 +2179,7 @@ static int nfs4_handle_delegation_recall_error(struct nfs_server *server, struct case -ENOENT: case -EAGAIN: case -ESTALE: + case -ETIMEDOUT: break; case -NFS4ERR_BADSESSION: case -NFS4ERR_BADSLOT: @@ -2499,6 +2500,7 @@ static int nfs4_run_open_task(struct nfs4_opendata *data, if (!ctx) { nfs4_init_sequence(&o_arg->seq_args, &o_res->seq_res, 1, 1); data->is_recover = true; + task_setup_data.flags |= RPC_TASK_TIMEOUT; } else { nfs4_init_sequence(&o_arg->seq_args, &o_res->seq_res, 1, 0); pnfs_lgopen_prepare(data, ctx); diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index a4e866b2b43b..cad4e064b328 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -1529,6 +1529,7 @@ restart: switch (status) { case 0: break; + case -ETIMEDOUT: case -ESTALE: case -NFS4ERR_ADMIN_REVOKED: case -NFS4ERR_STALE_STATEID: @@ -1682,11 +1683,13 @@ restart: case -NFS4ERR_EXPIRED: case -NFS4ERR_NO_GRACE: nfs4_state_mark_reclaim_nograce(sp->so_server->nfs_client, state); + /* Fall through */ case -NFS4ERR_STALE_CLIENTID: case -NFS4ERR_BADSESSION: case -NFS4ERR_BADSLOT: case -NFS4ERR_BAD_HIGH_SLOT: case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION: + case -ETIMEDOUT: goto out_err; } nfs4_put_open_state(state); @@ -1971,7 +1974,6 @@ static int nfs4_handle_reclaim_lease_error(struct nfs_client *clp, int status) return -EPERM; case -EACCES: case -NFS4ERR_DELAY: - case -ETIMEDOUT: case -EAGAIN: ssleep(1); break; @@ -2600,7 +2602,7 @@ static void nfs4_state_manager(struct nfs_client *clp) } /* Now recover expired state... */ - if (test_and_clear_bit(NFS4CLNT_RECLAIM_NOGRACE, &clp->cl_state)) { + if (test_bit(NFS4CLNT_RECLAIM_NOGRACE, &clp->cl_state)) { section = "reclaim nograce"; status = nfs4_do_reclaim(clp, clp->cl_mvops->nograce_recovery_ops); @@ -2608,6 +2610,7 @@ static void nfs4_state_manager(struct nfs_client *clp) continue; if (status < 0) goto out_error; + clear_bit(NFS4CLNT_RECLAIM_NOGRACE, &clp->cl_state); } nfs4_end_drain_session(clp); -- cgit v1.2.3