From 102a0743326a03cd1a1202ceda21e175b7d3575c Mon Sep 17 00:00:00 2001
From: hc <hc@nodka.com>
Date: Tue, 20 Feb 2024 01:20:52 +0000
Subject: [PATCH] add new system file
---
kernel/fs/nfsd/nfs4callback.c | 342 ++++++++++++++++++++++++++++++++++++++++++--------------
1 files changed, 254 insertions(+), 88 deletions(-)
diff --git a/kernel/fs/nfsd/nfs4callback.c b/kernel/fs/nfsd/nfs4callback.c
index 7ee417b..f5b7ad0 100644
--- a/kernel/fs/nfsd/nfs4callback.c
+++ b/kernel/fs/nfsd/nfs4callback.c
@@ -38,7 +38,9 @@
#include "nfsd.h"
#include "state.h"
#include "netns.h"
+#include "trace.h"
#include "xdr4cb.h"
+#include "xdr4.h"
#define NFSDDBG_FACILITY NFSDDBG_PROC
@@ -58,16 +60,6 @@
/* res */
int status;
};
-
-/*
- * Handle decode buffer overflows out-of-line.
- */
-static void print_overflow_msg(const char *func, const struct xdr_stream *xdr)
-{
- dprintk("NFS: %s prematurely hit the end of our receive buffer. "
- "Remaining buffer length is %tu words.\n",
- func, xdr->end - xdr->p);
-}
static __be32 *xdr_encode_empty_array(__be32 *p)
{
@@ -105,6 +97,7 @@
OP_CB_WANTS_CANCELLED = 12,
OP_CB_NOTIFY_LOCK = 13,
OP_CB_NOTIFY_DEVICEID = 14,
+ OP_CB_OFFLOAD = 15,
OP_CB_ILLEGAL = 10044
};
@@ -238,7 +231,6 @@
*status = nfs_cb_stat_to_errno(be32_to_cpup(p));
return 0;
out_overflow:
- print_overflow_msg(__func__, xdr);
return -EIO;
out_unexpected:
dprintk("NFSD: Callback server returned operation %d but "
@@ -307,7 +299,6 @@
hdr->nops = be32_to_cpup(p);
return 0;
out_overflow:
- print_overflow_msg(__func__, xdr);
return -EIO;
}
@@ -435,7 +426,6 @@
cb->cb_seq_status = status;
return status;
out_overflow:
- print_overflow_msg(__func__, xdr);
status = -EIO;
goto out;
}
@@ -523,11 +513,9 @@
if (unlikely(status))
return status;
- if (cb != NULL) {
- status = decode_cb_sequence4res(xdr, cb);
- if (unlikely(status || cb->cb_seq_status))
- return status;
- }
+ status = decode_cb_sequence4res(xdr, cb);
+ if (unlikely(status || cb->cb_seq_status))
+ return status;
return decode_cb_op_status(xdr, OP_CB_RECALL, &cb->cb_status);
}
@@ -615,11 +603,10 @@
if (unlikely(status))
return status;
- if (cb) {
- status = decode_cb_sequence4res(xdr, cb);
- if (unlikely(status || cb->cb_seq_status))
- return status;
- }
+ status = decode_cb_sequence4res(xdr, cb);
+ if (unlikely(status || cb->cb_seq_status))
+ return status;
+
return decode_cb_op_status(xdr, OP_CB_LAYOUTRECALL, &cb->cb_status);
}
#endif /* CONFIG_NFSD_PNFS */
@@ -674,14 +661,107 @@
if (unlikely(status))
return status;
- if (cb) {
- status = decode_cb_sequence4res(xdr, cb);
- if (unlikely(status || cb->cb_seq_status))
- return status;
- }
+ status = decode_cb_sequence4res(xdr, cb);
+ if (unlikely(status || cb->cb_seq_status))
+ return status;
+
return decode_cb_op_status(xdr, OP_CB_NOTIFY_LOCK, &cb->cb_status);
}
+/*
+ * struct write_response4 {
+ * stateid4 wr_callback_id<1>;
+ * length4 wr_count;
+ * stable_how4 wr_committed;
+ * verifier4 wr_writeverf;
+ * };
+ * union offload_info4 switch (nfsstat4 coa_status) {
+ * case NFS4_OK:
+ * write_response4 coa_resok4;
+ * default:
+ * length4 coa_bytes_copied;
+ * };
+ * struct CB_OFFLOAD4args {
+ * nfs_fh4 coa_fh;
+ * stateid4 coa_stateid;
+ * offload_info4 coa_offload_info;
+ * };
+ */
+static void encode_offload_info4(struct xdr_stream *xdr,
+ __be32 nfserr,
+ const struct nfsd4_copy *cp)
+{
+ __be32 *p;
+
+ p = xdr_reserve_space(xdr, 4);
+ *p++ = nfserr;
+ if (!nfserr) {
+ p = xdr_reserve_space(xdr, 4 + 8 + 4 + NFS4_VERIFIER_SIZE);
+ p = xdr_encode_empty_array(p);
+ p = xdr_encode_hyper(p, cp->cp_res.wr_bytes_written);
+ *p++ = cpu_to_be32(cp->cp_res.wr_stable_how);
+ p = xdr_encode_opaque_fixed(p, cp->cp_res.wr_verifier.data,
+ NFS4_VERIFIER_SIZE);
+ } else {
+ p = xdr_reserve_space(xdr, 8);
+ /* We always return success if bytes were written */
+ p = xdr_encode_hyper(p, 0);
+ }
+}
+
+static void encode_cb_offload4args(struct xdr_stream *xdr,
+ __be32 nfserr,
+ const struct knfsd_fh *fh,
+ const struct nfsd4_copy *cp,
+ struct nfs4_cb_compound_hdr *hdr)
+{
+ __be32 *p;
+
+ p = xdr_reserve_space(xdr, 4);
+ *p++ = cpu_to_be32(OP_CB_OFFLOAD);
+ encode_nfs_fh4(xdr, fh);
+ encode_stateid4(xdr, &cp->cp_res.cb_stateid);
+ encode_offload_info4(xdr, nfserr, cp);
+
+ hdr->nops++;
+}
+
+static void nfs4_xdr_enc_cb_offload(struct rpc_rqst *req,
+ struct xdr_stream *xdr,
+ const void *data)
+{
+ const struct nfsd4_callback *cb = data;
+ const struct nfsd4_copy *cp =
+ container_of(cb, struct nfsd4_copy, cp_cb);
+ struct nfs4_cb_compound_hdr hdr = {
+ .ident = 0,
+ .minorversion = cb->cb_clp->cl_minorversion,
+ };
+
+ encode_cb_compound4args(xdr, &hdr);
+ encode_cb_sequence4args(xdr, cb, &hdr);
+ encode_cb_offload4args(xdr, cp->nfserr, &cp->fh, cp, &hdr);
+ encode_cb_nops(&hdr);
+}
+
+static int nfs4_xdr_dec_cb_offload(struct rpc_rqst *rqstp,
+ struct xdr_stream *xdr,
+ void *data)
+{
+ struct nfsd4_callback *cb = data;
+ struct nfs4_cb_compound_hdr hdr;
+ int status;
+
+ status = decode_cb_compound4res(xdr, &hdr);
+ if (unlikely(status))
+ return status;
+
+ status = decode_cb_sequence4res(xdr, cb);
+ if (unlikely(status || cb->cb_seq_status))
+ return status;
+
+ return decode_cb_op_status(xdr, OP_CB_OFFLOAD, &cb->cb_status);
+}
/*
* RPC procedure tables
*/
@@ -703,6 +783,7 @@
PROC(CB_LAYOUT, COMPOUND, cb_layout, cb_layout),
#endif
PROC(CB_NOTIFY_LOCK, COMPOUND, cb_notify_lock, cb_notify_lock),
+ PROC(CB_OFFLOAD, COMPOUND, cb_offload, cb_offload),
};
static unsigned int nfs4_cb_counts[ARRAY_SIZE(nfs4_cb_procedures)];
@@ -743,27 +824,60 @@
static int max_cb_time(struct net *net)
{
struct nfsd_net *nn = net_generic(net, nfsd_net_id);
- return max(nn->nfsd4_lease/10, (time_t)1) * HZ;
+
+ /*
+ * nfsd4_lease is set to at most one hour in __nfsd4_write_time,
+ * so we can use 32-bit math on it. Warn if that assumption
+ * ever stops being true.
+ */
+ if (WARN_ON_ONCE(nn->nfsd4_lease > 3600))
+ return 360 * HZ;
+
+ return max(((u32)nn->nfsd4_lease)/10, 1u) * HZ;
}
-static struct rpc_cred *get_backchannel_cred(struct nfs4_client *clp, struct rpc_clnt *client, struct nfsd4_session *ses)
+static struct workqueue_struct *callback_wq;
+
+static bool nfsd4_queue_cb(struct nfsd4_callback *cb)
+{
+ return queue_work(callback_wq, &cb->cb_work);
+}
+
+static void nfsd41_cb_inflight_begin(struct nfs4_client *clp)
+{
+ atomic_inc(&clp->cl_cb_inflight);
+}
+
+static void nfsd41_cb_inflight_end(struct nfs4_client *clp)
+{
+
+ if (atomic_dec_and_test(&clp->cl_cb_inflight))
+ wake_up_var(&clp->cl_cb_inflight);
+}
+
+static void nfsd41_cb_inflight_wait_complete(struct nfs4_client *clp)
+{
+ wait_var_event(&clp->cl_cb_inflight,
+ !atomic_read(&clp->cl_cb_inflight));
+}
+
+static const struct cred *get_backchannel_cred(struct nfs4_client *clp, struct rpc_clnt *client, struct nfsd4_session *ses)
{
if (clp->cl_minorversion == 0) {
- char *principal = clp->cl_cred.cr_targ_princ ?
- clp->cl_cred.cr_targ_princ : "nfs";
- struct rpc_cred *cred;
+ client->cl_principal = clp->cl_cred.cr_targ_princ ?
+ clp->cl_cred.cr_targ_princ : "nfs";
- cred = rpc_lookup_machine_cred(principal);
- if (!IS_ERR(cred))
- get_rpccred(cred);
- return cred;
+ return get_cred(rpc_machine_cred());
} else {
- struct rpc_auth *auth = client->cl_auth;
- struct auth_cred acred = {};
+ struct cred *kcred;
- acred.uid = ses->se_cb_sec.uid;
- acred.gid = ses->se_cb_sec.gid;
- return auth->au_ops->lookup_cred(client->cl_auth, &acred, 0);
+ kcred = prepare_kernel_cred(NULL);
+ if (!kcred)
+ return NULL;
+
+ kcred->fsuid = ses->se_cb_sec.uid;
+ kcred->fsgid = ses->se_cb_sec.gid;
+ return kcred;
}
}
@@ -784,14 +898,17 @@
.program = &cb_program,
.version = 1,
.flags = (RPC_CLNT_CREATE_NOPING | RPC_CLNT_CREATE_QUIET),
+ .cred = current_cred(),
};
struct rpc_clnt *client;
- struct rpc_cred *cred;
+ const struct cred *cred;
if (clp->cl_minorversion == 0) {
if (!clp->cl_cred.cr_principal &&
- (clp->cl_cred.cr_flavor >= RPC_AUTH_GSS_KRB5))
+ (clp->cl_cred.cr_flavor >= RPC_AUTH_GSS_KRB5)) {
+ trace_nfsd_cb_setup_err(clp, -EINVAL);
return -EINVAL;
+ }
args.client_name = clp->cl_cred.cr_principal;
args.prognumber = conn->cb_prog;
args.protocol = XPRT_TRANSPORT_TCP;
@@ -800,7 +917,6 @@
} else {
if (!conn->cb_xprt)
return -EINVAL;
- clp->cl_cb_conn.cb_xprt = conn->cb_xprt;
clp->cl_cb_session = ses;
args.bc_xprt = conn->cb_xprt;
args.prognumber = clp->cl_cb_session->se_cb_prog;
@@ -811,24 +927,22 @@
/* Create RPC client */
client = rpc_create(&args);
if (IS_ERR(client)) {
- dprintk("NFSD: couldn't create callback client: %ld\n",
- PTR_ERR(client));
+ trace_nfsd_cb_setup_err(clp, PTR_ERR(client));
return PTR_ERR(client);
}
cred = get_backchannel_cred(clp, client, ses);
- if (IS_ERR(cred)) {
+ if (!cred) {
+ trace_nfsd_cb_setup_err(clp, -ENOMEM);
rpc_shutdown_client(client);
- return PTR_ERR(cred);
+ return -ENOMEM;
}
+
+ if (clp->cl_minorversion != 0)
+ clp->cl_cb_conn.cb_xprt = conn->cb_xprt;
clp->cl_cb_client = client;
clp->cl_cb_cred = cred;
+ trace_nfsd_cb_setup(clp);
return 0;
-}
-
-static void warn_no_callback_path(struct nfs4_client *clp, int reason)
-{
- dprintk("NFSD: warning: no callback path to client %.*s: error %d\n",
- (int)clp->cl_name.len, clp->cl_name.data, reason);
}
static void nfsd4_mark_cb_down(struct nfs4_client *clp, int reason)
@@ -836,7 +950,7 @@
if (test_bit(NFSD4_CLIENT_CB_UPDATE, &clp->cl_flags))
return;
clp->cl_cb_state = NFSD4_CB_DOWN;
- warn_no_callback_path(clp, reason);
+ trace_nfsd_cb_state(clp);
}
static void nfsd4_mark_cb_fault(struct nfs4_client *clp, int reason)
@@ -844,26 +958,36 @@
if (test_bit(NFSD4_CLIENT_CB_UPDATE, &clp->cl_flags))
return;
clp->cl_cb_state = NFSD4_CB_FAULT;
- warn_no_callback_path(clp, reason);
+ trace_nfsd_cb_state(clp);
}
static void nfsd4_cb_probe_done(struct rpc_task *task, void *calldata)
{
struct nfs4_client *clp = container_of(calldata, struct nfs4_client, cl_cb_null);
+ trace_nfsd_cb_done(clp, task->tk_status);
if (task->tk_status)
nfsd4_mark_cb_down(clp, task->tk_status);
- else
+ else {
clp->cl_cb_state = NFSD4_CB_UP;
+ trace_nfsd_cb_state(clp);
+ }
+}
+
+static void nfsd4_cb_probe_release(void *calldata)
+{
+ struct nfs4_client *clp = container_of(calldata, struct nfs4_client, cl_cb_null);
+
+ nfsd41_cb_inflight_end(clp);
+
}
static const struct rpc_call_ops nfsd4_cb_probe_ops = {
/* XXX: release method to ensure we set the cb channel down if
* necessary on early failure? */
.rpc_call_done = nfsd4_cb_probe_done,
+ .rpc_release = nfsd4_cb_probe_release,
};
-
-static struct workqueue_struct *callback_wq;
/*
* Poke the callback thread to process any updates to the callback
@@ -872,6 +996,7 @@
void nfsd4_probe_callback(struct nfs4_client *clp)
{
clp->cl_cb_state = NFSD4_CB_UNKNOWN;
+ trace_nfsd_cb_state(clp);
set_bit(NFSD4_CLIENT_CB_UPDATE, &clp->cl_flags);
nfsd4_run_cb(&clp->cl_cb_null);
}
@@ -888,6 +1013,7 @@
spin_lock(&clp->cl_lock);
memcpy(&clp->cl_cb_conn, conn, sizeof(struct nfs4_cb_conn));
spin_unlock(&clp->cl_lock);
+ trace_nfsd_cb_state(clp);
}
/*
@@ -895,9 +1021,12 @@
* If the slot is available, then mark it busy. Otherwise, set the
* thread for sleeping on the callback RPC wait queue.
*/
-static bool nfsd41_cb_get_slot(struct nfs4_client *clp, struct rpc_task *task)
+static bool nfsd41_cb_get_slot(struct nfsd4_callback *cb, struct rpc_task *task)
{
- if (test_and_set_bit(0, &clp->cl_cb_slot_busy) != 0) {
+ struct nfs4_client *clp = cb->cb_clp;
+
+ if (!cb->cb_holds_slot &&
+ test_and_set_bit(0, &clp->cl_cb_slot_busy) != 0) {
rpc_sleep_on(&clp->cl_cb_waitq, task, NULL);
/* Race breaker */
if (test_and_set_bit(0, &clp->cl_cb_slot_busy) != 0) {
@@ -906,7 +1035,29 @@
}
rpc_wake_up_queued_task(&clp->cl_cb_waitq, task);
}
+ cb->cb_holds_slot = true;
return true;
+}
+
+static void nfsd41_cb_release_slot(struct nfsd4_callback *cb)
+{
+ struct nfs4_client *clp = cb->cb_clp;
+
+ if (cb->cb_holds_slot) {
+ cb->cb_holds_slot = false;
+ clear_bit(0, &clp->cl_cb_slot_busy);
+ rpc_wake_up_next(&clp->cl_cb_waitq);
+ }
+}
+
+static void nfsd41_destroy_cb(struct nfsd4_callback *cb)
+{
+ struct nfs4_client *clp = cb->cb_clp;
+
+ nfsd41_cb_release_slot(cb);
+ if (cb->cb_ops && cb->cb_ops->release)
+ cb->cb_ops->release(cb);
+ nfsd41_cb_inflight_end(clp);
}
/*
@@ -925,11 +1076,8 @@
*/
cb->cb_seq_status = 1;
cb->cb_status = 0;
- if (minorversion) {
- if (!cb->cb_holds_slot && !nfsd41_cb_get_slot(clp, task))
- return;
- cb->cb_holds_slot = true;
- }
+ if (minorversion && !nfsd41_cb_get_slot(cb, task))
+ return;
rpc_call_start(task);
}
@@ -949,7 +1097,7 @@
* the submission code will error out, so we don't need to
* handle that case here.
*/
- if (task->tk_flags & RPC_TASK_KILLED)
+ if (RPC_SIGNALLED(task))
goto need_restart;
return true;
@@ -971,7 +1119,7 @@
break;
case -ESERVERFAULT:
++session->se_cb_seq_nr;
- /* Fall through */
+ fallthrough;
case 1:
case -NFS4ERR_BADSESSION:
nfsd4_mark_cb_fault(cb->cb_clp, cb->cb_seq_status);
@@ -992,17 +1140,16 @@
}
break;
default:
+ nfsd4_mark_cb_fault(cb->cb_clp, cb->cb_seq_status);
dprintk("%s: unprocessed error %d\n", __func__,
cb->cb_seq_status);
}
- cb->cb_holds_slot = false;
- clear_bit(0, &clp->cl_cb_slot_busy);
- rpc_wake_up_next(&clp->cl_cb_waitq);
+ nfsd41_cb_release_slot(cb);
dprintk("%s: freed slot, new seqid=%d\n", __func__,
clp->cl_cb_session->se_cb_seq_nr);
- if (task->tk_flags & RPC_TASK_KILLED)
+ if (RPC_SIGNALLED(task))
goto need_restart;
out:
return ret;
@@ -1011,8 +1158,10 @@
ret = false;
goto out;
need_restart:
- task->tk_status = 0;
- cb->cb_need_restart = true;
+ if (!test_bit(NFSD4_CLIENT_CB_KILL, &clp->cl_flags)) {
+ task->tk_status = 0;
+ cb->cb_need_restart = true;
+ }
return false;
}
@@ -1021,8 +1170,7 @@
struct nfsd4_callback *cb = calldata;
struct nfs4_client *clp = cb->cb_clp;
- dprintk("%s: minorversion=%d\n", __func__,
- clp->cl_minorversion);
+ trace_nfsd_cb_done(clp, task->tk_status);
if (!nfsd4_cb_sequence_done(task, cb))
return;
@@ -1038,10 +1186,12 @@
rpc_restart_call_prepare(task);
return;
case 1:
- break;
- case -1:
- /* Network partition? */
- nfsd4_mark_cb_down(clp, task->tk_status);
+ switch (task->tk_status) {
+ case -EIO:
+ case -ETIMEDOUT:
+ case -EACCES:
+ nfsd4_mark_cb_down(clp, task->tk_status);
+ }
break;
default:
BUG();
@@ -1053,9 +1203,9 @@
struct nfsd4_callback *cb = calldata;
if (cb->cb_need_restart)
- nfsd4_run_cb(cb);
+ nfsd4_queue_cb(cb);
else
- cb->cb_ops->release(cb);
+ nfsd41_destroy_cb(cb);
}
@@ -1089,6 +1239,7 @@
*/
nfsd4_run_cb(&clp->cl_cb_null);
flush_workqueue(callback_wq);
+ nfsd41_cb_inflight_wait_complete(clp);
}
/* requires cl_lock: */
@@ -1106,6 +1257,12 @@
return NULL;
}
+/*
+ * Note there isn't a lot of locking in this code; instead we depend on
+ * the fact that it is run from the callback_wq, which won't run two
+ * work items at once. So, for example, callback_wq handles all access
+ * of cl_cb_client and all calls to rpc_create or rpc_shutdown_client.
+ */
static void nfsd4_process_cb_update(struct nfsd4_callback *cb)
{
struct nfs4_cb_conn conn;
@@ -1119,9 +1276,10 @@
* kill the old client:
*/
if (clp->cl_cb_client) {
+ trace_nfsd_cb_shutdown(clp);
rpc_shutdown_client(clp->cl_cb_client);
clp->cl_cb_client = NULL;
- put_rpccred(clp->cl_cb_cred);
+ put_cred(clp->cl_cb_cred);
clp->cl_cb_cred = NULL;
}
if (clp->cl_cb_conn.cb_xprt) {
@@ -1162,6 +1320,9 @@
container_of(work, struct nfsd4_callback, cb_work);
struct nfs4_client *clp = cb->cb_clp;
struct rpc_clnt *clnt;
+ int flags;
+
+ trace_nfsd_cb_work(clp, cb->cb_msg.rpc_proc->p_name);
if (cb->cb_need_restart) {
cb->cb_need_restart = false;
@@ -1176,8 +1337,7 @@
clnt = clp->cl_cb_client;
if (!clnt) {
/* Callback channel broken, or client killed; give up: */
- if (cb->cb_ops && cb->cb_ops->release)
- cb->cb_ops->release(cb);
+ nfsd41_destroy_cb(cb);
return;
}
@@ -1186,11 +1346,13 @@
*/
if (!cb->cb_ops && clp->cl_minorversion) {
clp->cl_cb_state = NFSD4_CB_UP;
+ nfsd41_destroy_cb(cb);
return;
}
cb->cb_msg.rpc_cred = clp->cl_cb_cred;
- rpc_call_async(clnt, &cb->cb_msg, RPC_TASK_SOFT | RPC_TASK_SOFTCONN,
+ flags = clp->cl_minorversion ? RPC_TASK_NOCONNECT : RPC_TASK_SOFTCONN;
+ rpc_call_async(clnt, &cb->cb_msg, RPC_TASK_SOFT | flags,
cb->cb_ops ? &nfsd4_cb_ops : &nfsd4_cb_probe_ops, cb);
}
@@ -1211,5 +1373,9 @@
void nfsd4_run_cb(struct nfsd4_callback *cb)
{
- queue_work(callback_wq, &cb->cb_work);
+ struct nfs4_client *clp = cb->cb_clp;
+
+ nfsd41_cb_inflight_begin(clp);
+ if (!nfsd4_queue_cb(cb))
+ nfsd41_cb_inflight_end(clp);
}
--
Gitblit v1.6.2