case -NFS4ERR_DELAY:
nfs_inc_server_stats(server, NFSIOS_DELAY);
case -NFS4ERR_GRACE:
+ case -NFS4ERR_RECALLCONFLICT:
exception->delay = 1;
return 0;
struct nfs4_layoutget *lgp = calldata;
struct nfs_server *server = NFS_SERVER(lgp->args.inode);
struct nfs4_session *session = nfs4_get_session(server);
- int ret;
dprintk("--> %s\n", __func__);
- /* Note the is a race here, where a CB_LAYOUTRECALL can come in
- * right now covering the LAYOUTGET we are about to send.
- * However, that is not so catastrophic, and there seems
- * to be no way to prevent it completely.
- */
- if (nfs41_setup_sequence(session, &lgp->args.seq_args,
- &lgp->res.seq_res, task))
- return;
- ret = pnfs_choose_layoutget_stateid(&lgp->args.stateid,
- NFS_I(lgp->args.inode)->layout,
- &lgp->args.range,
- lgp->args.ctx->state);
- if (ret < 0)
- rpc_exit(task, ret);
+ nfs41_setup_sequence(session, &lgp->args.seq_args,
+ &lgp->res.seq_res, task);
+ dprintk("<-- %s\n", __func__);
}
static void nfs4_layoutget_done(struct rpc_task *task, void *calldata)
{
struct nfs4_layoutget *lgp = calldata;
+
+ dprintk("--> %s\n", __func__);
+ nfs41_sequence_done(task, &lgp->res.seq_res);
+ dprintk("<-- %s\n", __func__);
+}
+
+static int
+nfs4_layoutget_handle_exception(struct rpc_task *task,
+ struct nfs4_layoutget *lgp, struct nfs4_exception *exception)
+{
struct inode *inode = lgp->args.inode;
struct nfs_server *server = NFS_SERVER(inode);
struct pnfs_layout_hdr *lo;
- struct nfs4_state *state = NULL;
- unsigned long timeo, now, giveup;
+ int status = task->tk_status;
dprintk("--> %s tk_status => %d\n", __func__, -task->tk_status);
- if (!nfs41_sequence_done(task, &lgp->res.seq_res))
- goto out;
-
- switch (task->tk_status) {
+ switch (status) {
case 0:
goto out;
* retry go inband.
*/
case -NFS4ERR_LAYOUTUNAVAILABLE:
- task->tk_status = -ENODATA;
+ status = -ENODATA;
goto out;
/*
* NFS4ERR_BADLAYOUT means the MDS cannot return a layout of
* length lgp->args.minlength != 0 (see RFC5661 section 18.43.3).
*/
case -NFS4ERR_BADLAYOUT:
- goto out_overflow;
+ status = -EOVERFLOW;
+ goto out;
/*
* NFS4ERR_LAYOUTTRYLATER is a conflict with another client
* (or clients) writing to the same RAID stripe except when
* the minlength argument is 0 (see RFC5661 section 18.43.3).
+ *
+ * Treat it like we would RECALLCONFLICT -- we retry for a little
+ * while, and then eventually give up.
*/
case -NFS4ERR_LAYOUTTRYLATER:
- if (lgp->args.minlength == 0)
- goto out_overflow;
- /*
- * NFS4ERR_RECALLCONFLICT is when conflict with self (must recall
- * existing layout before getting a new one).
- */
- case -NFS4ERR_RECALLCONFLICT:
- timeo = rpc_get_timeout(task->tk_client);
- giveup = lgp->args.timestamp + timeo;
- now = jiffies;
- if (time_after(giveup, now)) {
- unsigned long delay;
-
- /* Delay for:
- * - Not less then NFS4_POLL_RETRY_MIN.
- * - One last time a jiffie before we give up
- * - exponential backoff (time_now minus start_attempt)
- */
- delay = max_t(unsigned long, NFS4_POLL_RETRY_MIN,
- min((giveup - now - 1),
- now - lgp->args.timestamp));
-
- dprintk("%s: NFS4ERR_RECALLCONFLICT waiting %lu\n",
- __func__, delay);
- rpc_delay(task, delay);
- /* Do not call nfs4_async_handle_error() */
- goto out_restart;
+ if (lgp->args.minlength == 0) {
+ status = -EOVERFLOW;
+ goto out;
}
- break;
+ /* Fallthrough */
+ case -NFS4ERR_RECALLCONFLICT:
+ nfs4_handle_exception(server, -NFS4ERR_RECALLCONFLICT,
+ exception);
+ status = -ERECALLCONFLICT;
+ goto out;
case -NFS4ERR_EXPIRED:
case -NFS4ERR_BAD_STATEID:
+ exception->timeout = 0;
spin_lock(&inode->i_lock);
if (nfs4_stateid_match(&lgp->args.stateid,
&lgp->args.ctx->state->stateid)) {
spin_unlock(&inode->i_lock);
/* If the open stateid was bad, then recover it. */
- state = lgp->args.ctx->state;
+ exception->state = lgp->args.ctx->state;
break;
}
lo = NFS_I(inode)->layout;
pnfs_free_lseg_list(&head);
} else
spin_unlock(&inode->i_lock);
- goto out_restart;
+ status = -EAGAIN;
+ goto out;
}
- if (nfs4_async_handle_error(task, server, state, &lgp->timeout) == -EAGAIN)
- goto out_restart;
+
+ status = nfs4_handle_exception(server, status, exception);
+ if (exception->retry)
+ status = -EAGAIN;
out:
dprintk("<-- %s\n", __func__);
- return;
-out_restart:
- task->tk_status = 0;
- rpc_restart_call_prepare(task);
- return;
-out_overflow:
- task->tk_status = -EOVERFLOW;
- goto out;
+ return status;
}
static size_t max_response_pages(struct nfs_server *server)
};
struct pnfs_layout_segment *
-nfs4_proc_layoutget(struct nfs4_layoutget *lgp, gfp_t gfp_flags)
+nfs4_proc_layoutget(struct nfs4_layoutget *lgp, long *timeout, gfp_t gfp_flags)
{
struct inode *inode = lgp->args.inode;
struct nfs_server *server = NFS_SERVER(inode);
.flags = RPC_TASK_ASYNC,
};
struct pnfs_layout_segment *lseg = NULL;
+ struct nfs4_exception exception = { .timeout = *timeout };
int status = 0;
dprintk("--> %s\n", __func__);
return ERR_PTR(-ENOMEM);
}
lgp->args.layout.pglen = max_pages * PAGE_SIZE;
- lgp->args.timestamp = jiffies;
lgp->res.layoutp = &lgp->args.layout;
lgp->res.seq_res.sr_slot = NULL;
if (IS_ERR(task))
return ERR_CAST(task);
status = nfs4_wait_for_completion_rpc_task(task);
- if (status == 0)
- status = task->tk_status;
+ if (status == 0) {
+ status = nfs4_layoutget_handle_exception(task, lgp, &exception);
+ *timeout = exception.timeout;
+ }
+
trace_nfs4_layoutget(lgp->args.ctx,
&lgp->args.range,
&lgp->res.range,
&lgp->res.stateid,
status);
+
/* if layoutp->len is 0, nfs4_layoutget_prepare called rpc_exit */
if (status == 0 && lgp->res.layoutp->len)
lseg = pnfs_layout_process(lgp);
{ PNFS_UPDATE_LAYOUT_FOUND_CACHED, "found cached" }, \
{ PNFS_UPDATE_LAYOUT_RETURN, "layoutreturn" }, \
{ PNFS_UPDATE_LAYOUT_BLOCKED, "layouts blocked" }, \
+ { PNFS_UPDATE_LAYOUT_INVALID_OPEN, "invalid open" }, \
+ { PNFS_UPDATE_LAYOUT_RETRY, "retrying" }, \
{ PNFS_UPDATE_LAYOUT_SEND_LAYOUTGET, "sent layoutget" })
TRACE_EVENT(pnfs_update_layout,
u64 count,
enum pnfs_iomode iomode,
struct pnfs_layout_hdr *lo,
+ struct pnfs_layout_segment *lseg,
enum pnfs_update_layout_reason reason
),
- TP_ARGS(inode, pos, count, iomode, lo, reason),
+ TP_ARGS(inode, pos, count, iomode, lo, lseg, reason),
TP_STRUCT__entry(
__field(dev_t, dev)
__field(u64, fileid)
__field(enum pnfs_iomode, iomode)
__field(int, layoutstateid_seq)
__field(u32, layoutstateid_hash)
+ __field(long, lseg)
__field(enum pnfs_update_layout_reason, reason)
),
TP_fast_assign(
__entry->layoutstateid_seq = 0;
__entry->layoutstateid_hash = 0;
}
+ __entry->lseg = (long)lseg;
),
TP_printk(
"fileid=%02x:%02x:%llu fhandle=0x%08x "
"iomode=%s pos=%llu count=%llu "
- "layoutstateid=%d:0x%08x (%s)",
+ "layoutstateid=%d:0x%08x lseg=0x%lx (%s)",
MAJOR(__entry->dev), MINOR(__entry->dev),
(unsigned long long)__entry->fileid,
__entry->fhandle,
(unsigned long long)__entry->pos,
(unsigned long long)__entry->count,
__entry->layoutstateid_seq, __entry->layoutstateid_hash,
+ __entry->lseg,
show_pnfs_update_layout_reason(__entry->reason)
)
);
test_bit(NFS_LAYOUT_BULK_RECALL, &lo->plh_flags);
}
-int
-pnfs_choose_layoutget_stateid(nfs4_stateid *dst, struct pnfs_layout_hdr *lo,
- const struct pnfs_layout_range *range,
- struct nfs4_state *open_state)
-{
- int status = 0;
-
- dprintk("--> %s\n", __func__);
- spin_lock(&lo->plh_inode->i_lock);
- if (pnfs_layoutgets_blocked(lo)) {
- status = -EAGAIN;
- } else if (!nfs4_valid_open_stateid(open_state)) {
- status = -EBADF;
- } else if (list_empty(&lo->plh_segs) ||
- test_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags)) {
- int seq;
-
- do {
- seq = read_seqbegin(&open_state->seqlock);
- nfs4_stateid_copy(dst, &open_state->stateid);
- } while (read_seqretry(&open_state->seqlock, seq));
- } else
- nfs4_stateid_copy(dst, &lo->plh_stateid);
- spin_unlock(&lo->plh_inode->i_lock);
- dprintk("<-- %s\n", __func__);
- return status;
-}
-
/*
-* Get layout from server.
-* for now, assume that whole file layouts are requested.
-* arg->offset: 0
-* arg->length: all ones
-*/
+ * Get layout from server.
+ * for now, assume that whole file layouts are requested.
+ * arg->offset: 0
+ * arg->length: all ones
+ */
static struct pnfs_layout_segment *
send_layoutget(struct pnfs_layout_hdr *lo,
struct nfs_open_context *ctx,
+ nfs4_stateid *stateid,
const struct pnfs_layout_range *range,
- gfp_t gfp_flags)
+ long *timeout, gfp_t gfp_flags)
{
struct inode *ino = lo->plh_inode;
struct nfs_server *server = NFS_SERVER(ino);
lgp->args.type = server->pnfs_curr_ld->id;
lgp->args.inode = ino;
lgp->args.ctx = get_nfs_open_context(ctx);
+ nfs4_stateid_copy(&lgp->args.stateid, stateid);
lgp->gfp_flags = gfp_flags;
lgp->cred = lo->plh_lc_cred;
- return nfs4_proc_layoutget(lgp, gfp_flags);
+ return nfs4_proc_layoutget(lgp, timeout, gfp_flags);
}
static void pnfs_clear_layoutcommit(struct inode *inode,
.offset = pos,
.length = count,
};
- unsigned pg_offset;
+ unsigned pg_offset, seq;
struct nfs_server *server = NFS_SERVER(ino);
struct nfs_client *clp = server->nfs_client;
- struct pnfs_layout_hdr *lo;
+ struct pnfs_layout_hdr *lo = NULL;
struct pnfs_layout_segment *lseg = NULL;
+ nfs4_stateid stateid;
+ long timeout = 0;
+ unsigned long giveup = jiffies + rpc_get_timeout(server->client);
bool first;
if (!pnfs_enabled_sb(NFS_SERVER(ino))) {
- trace_pnfs_update_layout(ino, pos, count, iomode, NULL,
+ trace_pnfs_update_layout(ino, pos, count, iomode, lo, lseg,
PNFS_UPDATE_LAYOUT_NO_PNFS);
goto out;
}
if (iomode == IOMODE_READ && i_size_read(ino) == 0) {
- trace_pnfs_update_layout(ino, pos, count, iomode, NULL,
+ trace_pnfs_update_layout(ino, pos, count, iomode, lo, lseg,
PNFS_UPDATE_LAYOUT_RD_ZEROLEN);
goto out;
}
if (pnfs_within_mdsthreshold(ctx, ino, iomode)) {
- trace_pnfs_update_layout(ino, pos, count, iomode, NULL,
+ trace_pnfs_update_layout(ino, pos, count, iomode, lo, lseg,
PNFS_UPDATE_LAYOUT_MDSTHRESH);
goto out;
}
lo = pnfs_find_alloc_layout(ino, ctx, gfp_flags);
if (lo == NULL) {
spin_unlock(&ino->i_lock);
- trace_pnfs_update_layout(ino, pos, count, iomode, NULL,
+ trace_pnfs_update_layout(ino, pos, count, iomode, lo, lseg,
PNFS_UPDATE_LAYOUT_NOMEM);
goto out;
}
/* Do we even need to bother with this? */
if (test_bit(NFS_LAYOUT_BULK_RECALL, &lo->plh_flags)) {
- trace_pnfs_update_layout(ino, pos, count, iomode, lo,
+ trace_pnfs_update_layout(ino, pos, count, iomode, lo, lseg,
PNFS_UPDATE_LAYOUT_BULK_RECALL);
dprintk("%s matches recall, use MDS\n", __func__);
goto out_unlock;
/* if LAYOUTGET already failed once we don't try again */
if (pnfs_layout_io_test_failed(lo, iomode)) {
- trace_pnfs_update_layout(ino, pos, count, iomode, lo,
+ trace_pnfs_update_layout(ino, pos, count, iomode, lo, lseg,
PNFS_UPDATE_LAYOUT_IO_TEST_FAIL);
goto out_unlock;
}
- first = list_empty(&lo->plh_segs);
- if (first) {
- /* The first layoutget for the file. Need to serialize per
+ lseg = pnfs_find_lseg(lo, &arg);
+ if (lseg) {
+ trace_pnfs_update_layout(ino, pos, count, iomode, lo, lseg,
+ PNFS_UPDATE_LAYOUT_FOUND_CACHED);
+ 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
+ * stateid.
+ */
+ if (lo->plh_stateid.seqid == 0 ||
+ test_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags)) {
+
+ /*
+ * The first layoutget for the file. Need to serialize per
* RFC 5661 Errata 3208.
*/
if (test_and_set_bit(NFS_LAYOUT_FIRST_LAYOUTGET,
wait_on_bit(&lo->plh_flags, NFS_LAYOUT_FIRST_LAYOUTGET,
TASK_UNINTERRUPTIBLE);
pnfs_put_layout_hdr(lo);
+ dprintk("%s retrying\n", __func__);
goto lookup_again;
}
+
+ first = true;
+ do {
+ seq = read_seqbegin(&ctx->state->seqlock);
+ nfs4_stateid_copy(&stateid, &ctx->state->stateid);
+ } while (read_seqretry(&ctx->state->seqlock, seq));
} else {
- /* Check to see if the layout for the given range
- * already exists
- */
- lseg = pnfs_find_lseg(lo, &arg);
- if (lseg) {
- trace_pnfs_update_layout(ino, pos, count, iomode, lo,
- PNFS_UPDATE_LAYOUT_FOUND_CACHED);
- goto out_unlock;
- }
+ nfs4_stateid_copy(&stateid, &lo->plh_stateid);
}
/*
pnfs_clear_first_layoutget(lo);
pnfs_put_layout_hdr(lo);
dprintk("%s retrying\n", __func__);
+ trace_pnfs_update_layout(ino, pos, count, iomode, lo,
+ lseg, PNFS_UPDATE_LAYOUT_RETRY);
goto lookup_again;
}
- trace_pnfs_update_layout(ino, pos, count, iomode, lo,
+ trace_pnfs_update_layout(ino, pos, count, iomode, lo, lseg,
PNFS_UPDATE_LAYOUT_RETURN);
goto out_put_layout_hdr;
}
if (pnfs_layoutgets_blocked(lo)) {
- trace_pnfs_update_layout(ino, pos, count, iomode, lo,
+ trace_pnfs_update_layout(ino, pos, count, iomode, lo, lseg,
PNFS_UPDATE_LAYOUT_BLOCKED);
goto out_unlock;
}
if (arg.length != NFS4_MAX_UINT64)
arg.length = PAGE_ALIGN(arg.length);
- lseg = send_layoutget(lo, ctx, &arg, gfp_flags);
+ lseg = send_layoutget(lo, ctx, &stateid, &arg, &timeout, gfp_flags);
+ trace_pnfs_update_layout(ino, pos, count, iomode, lo, lseg,
+ PNFS_UPDATE_LAYOUT_SEND_LAYOUTGET);
if (IS_ERR(lseg)) {
- if (lseg == ERR_PTR(-EAGAIN)) {
+ switch(PTR_ERR(lseg)) {
+ case -ERECALLCONFLICT:
+ if (time_after(jiffies, giveup))
+ lseg = NULL;
+ /* Fallthrough */
+ case -EAGAIN:
+ pnfs_put_layout_hdr(lo);
if (first)
pnfs_clear_first_layoutget(lo);
- pnfs_put_layout_hdr(lo);
- goto lookup_again;
- }
-
- if (!nfs_error_is_fatal(PTR_ERR(lseg))) {
- pnfs_layout_clear_fail_bit(lo, pnfs_iomode_to_fail_bit(iomode));
- lseg = NULL;
+ if (lseg) {
+ trace_pnfs_update_layout(ino, pos, count,
+ iomode, lo, lseg, PNFS_UPDATE_LAYOUT_RETRY);
+ goto lookup_again;
+ }
+ /* Fallthrough */
+ default:
+ if (!nfs_error_is_fatal(PTR_ERR(lseg))) {
+ pnfs_layout_clear_fail_bit(lo, pnfs_iomode_to_fail_bit(iomode));
+ lseg = NULL;
+ }
}
} else {
pnfs_layout_clear_fail_bit(lo, pnfs_iomode_to_fail_bit(iomode));
}
atomic_dec(&lo->plh_outstanding);
- trace_pnfs_update_layout(ino, pos, count, iomode, lo,
- PNFS_UPDATE_LAYOUT_SEND_LAYOUTGET);
out_put_layout_hdr:
if (first)
pnfs_clear_first_layoutget(lo);
projects / linux-2.6-microblaze.git / commitdiff