Skip to contentSkip to conversation

zfcp fixes for 2.6.28

PATCH In review Christof Schmitt 2 participants 11 messages Nov 4, 2008 – Nov 6, 2008

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover zfcp fixes for 2.6.28

James,

here is the series of zfcp fixes for 2.6.28. The patches apply on top of the current linux-2.6 git tree.

Christof

#1 lore raw

PATCH 1/8 zfcp: Dont clear reference from SCSI device to unit

From: Christof Schmitt <[email protected]>

It is possible that a remote port has a problem, the SCSI device gets deleted after the rport timeout and then the timeout for pending SCSI commands trigger an abort. For this case, don't delete the reference from the SCSI device to the zfcp unit, so that we can still have the reference to issue an abort request.

---
 drivers/s390/scsi/zfcp_scsi.c |   12 ++++--------
 1 file changed, 4 insertions(+), 8 deletions(-)
drivers/s390/scsi/zfcp_scsi.c +4 -8
@@ -24,14 +24,10 @@ char *zfcp_get_fcp_sns_info_ptr(struct f
 static void zfcp_scsi_slave_destroy(struct scsi_device *sdpnt)
 {
 	struct zfcp_unit *unit = (struct zfcp_unit *) sdpnt->hostdata;
-	WARN_ON(!unit);
-	if (unit) {
-		atomic_clear_mask(ZFCP_STATUS_UNIT_REGISTERED, &unit->status);
-		sdpnt->hostdata = NULL;
-		unit->device = NULL;
-		zfcp_erp_unit_failed(unit, 12, NULL);
-		zfcp_unit_put(unit);
-	}
+	atomic_clear_mask(ZFCP_STATUS_UNIT_REGISTERED, &unit->status);
+	unit->device = NULL;
+	zfcp_erp_unit_failed(unit, 12, NULL);
+	zfcp_unit_put(unit);
 }
 
 static int zfcp_scsi_slave_configure(struct scsi_device *sdp)
#2 lore raw

PATCH 2/8 zfcp: fix req_list_locking.

In reply to Christof Schmitt

From: Heiko Carstens <[email protected]>

The per adapter req_list_lock must be held with interrupts disabled, otherwise we might end up with nice deadlocks as lockdep tells us (see below).

zfcp 0.0.1804: QDIO problem occurred.

=========================================================
[ INFO: possible irq lock inversion dependency detected ] 2.6.27-rc8-00035-g4a77035-dirty #86
---------------------------------------------------------
swapper/0 just changed the state of lock:
(&adapter->erp_lock){++..}, at: [<00000000002c82ae>] zfcp_erp_adapter_reopen+0x4e/0x8c
but this lock took another, hard-irq-unsafe lock in the past:
(&adapter->req_list_lock){-+..}

and interrupts could create inverse lock ordering between them.

[tons of backtraces, but only the interesting part follows]

the second lock's dependencies:
-> (&adapter->req_list_lock){-+..} ops: 2280627634176 {
initial-use at:

                        [<0000000000071f10>] __lock_acquire+0x504/0x18bc
                        [<000000000007335c>] lock_acquire+0x94/0xbc
                        [<00000000003d7224>] _spin_lock_irqsave+0x6c/0xb0
                        [<00000000002cf684>] zfcp_fsf_req_dismiss_all+0x50/0x140
                        [<00000000002c87ee>] zfcp_erp_adapter_strategy_generic+0x66/0x3d0
                        [<00000000002c9498>] zfcp_erp_thread+0x88c/0x1318
                        [<000000000001b0d2>] kernel_thread_starter+0x6/0xc
                        [<000000000001b0cc>] kernel_thread_starter+0x0/0xc
   in-softirq-W at:
                        [<0000000000072172>] __lock_acquire+0x766/0x18bc
                        [<000000000007335c>] lock_acquire+0x94/0xbc
                        [<00000000003d7224>] _spin_lock_irqsave+0x6c/0xb0
                        [<00000000002ca73e>] zfcp_qdio_int_resp+0xbe/0x2ac
                        [<000000000027a1d6>] qdio_kick_inbound_handler+0x82/0xa0
                        [<000000000027daba>] tiqdio_inbound_processing+0x62/0xf8
                        [<0000000000047ba4>] tasklet_action+0x100/0x1f4
                        [<0000000000048b5a>] __do_softirq+0xae/0x154
                        [<0000000000021e4a>] do_softirq+0xea/0xf0
                        [<00000000000485de>] irq_exit+0xde/0xe8
                        [<0000000000268c64>] do_IRQ+0x160/0x1fc
                        [<00000000000261a2>] io_return+0x0/0x8
                        [<000000000001b8f8>] cpu_idle+0x17c/0x224
   hardirq-on-W at:
                        [<0000000000072190>] __lock_acquire+0x784/0x18bc
                        [<000000000007335c>] lock_acquire+0x94/0xbc
                        [<00000000003d702c>] _spin_lock+0x5c/0x9c
                        [<00000000002caff6>] zfcp_fsf_req_send+0x3e/0x158
                        [<00000000002ce7fe>] zfcp_fsf_exchange_config_data+0x106/0x124
                        [<00000000002c8948>] zfcp_erp_adapter_strategy_generic+0x1c0/0x3d0
                        [<00000000002c98ea>] zfcp_erp_thread+0xcde/0x1318
                        [<000000000001b0d2>] kernel_thread_starter+0x6/0xc
                        [<000000000001b0cc>] kernel_thread_starter+0x0/0xc
 }
 ... key      at: [<0000000000e356c8>] __key.26629+0x0/0x8

---

 drivers/s390/scsi/zfcp_fsf.c |    5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
drivers/s390/scsi/zfcp_fsf.c +3 -2
@@ -770,13 +770,14 @@ static int zfcp_fsf_req_send(struct zfcp
 {
 	struct zfcp_adapter *adapter = req->adapter;
 	struct zfcp_qdio_queue *req_q = &adapter->req_q;
+	unsigned long flags;
 	int idx;
 
 	/* put allocated FSF request into hash table */
-	spin_lock(&adapter->req_list_lock);
+	spin_lock_irqsave(&adapter->req_list_lock, flags);
 	idx = zfcp_reqlist_hash(req->req_id);
 	list_add_tail(&req->list, &adapter->req_list[idx]);
-	spin_unlock(&adapter->req_list_lock);
+	spin_unlock_irqrestore(&adapter->req_list_lock, flags);
 
 	req->qdio_outb_usage = atomic_read(&req_q->count);
 	req->issued = get_clock();
#3 lore raw

PATCH 3/8 zfcp: fix mempool usage for status_read requests

In reply to Christof Schmitt

From: Christof Schmitt <[email protected]>

When allocating fsf requests without qtcb, store the pointer to the mempool in the fsf requests for later call to mempool_free. This codepath is only used by the status_read requests.

---
 drivers/s390/scsi/zfcp_fsf.c |    1 +
 1 file changed, 1 insertion(+)
drivers/s390/scsi/zfcp_fsf.c +1 -0
@@ -683,6 +683,7 @@ static struct zfcp_fsf_req *zfcp_fsf_all
 	if (!req)
 		return NULL;
 	memset(req, 0, sizeof(*req));
+	req->pool = pool;
 	return req;
 }
 
maintainer
#4 lore raw

Re: [patch 3/8] zfcp: fix mempool usage for status_read requests

Quoted 9 lines · Christof Schmitt

plain text document attachment (fix-status-read-alloc.diff)
From: Christof Schmitt <[email protected]>

When allocating fsf requests without qtcb, store the pointer to the mempool in the fsf requests for later call to mempool_free. This codepath is only used by the status_read requests.

I put this one in, but can you explain this last signoff better? You sent me the patch, so your signoff should appear last. If Swen altered the patch, his signoff should appear before yours (with a small explanation of the alteration in square brackets). If he's just approving it, it should be Acked-by: not Signed-off-by:

James

#5 lore raw

Re: [patch 3/8] zfcp: fix mempool usage for status_read requests

Quoted 8 lines · James Bottomley
Quoted 7 lines · Christof Schmitt

plain text document attachment (fix-status-read-alloc.diff)
From: Christof Schmitt <[email protected]>

When allocating fsf requests without qtcb, store the pointer to the mempool in the fsf requests for later call to mempool_free. This codepath is only used by the status_read requests.

Quoted 8 lines
Quoted 2 lines

I put this one in, but can you explain this last signoff better? You sent me the patch, so your signoff should appear last. If Swen altered the patch, his signoff should appear before yours (with a small explanation of the alteration in square brackets). If he's just approving it, it should be Acked-by: not Signed-off-by:

I wrote the patch, Swen reviewed it and i sent it as part of the patch series. It looks like this did not follow the exact guidelines. For future patches, i will document the review with the "Acked-by:".

Christof

#6 lore raw

PATCH 4/8 zfcp: Fix request list handling in error path

In reply to Christof Schmitt

From: Christof Schmitt <[email protected]>

Fix the handling of the request list in the error path:
- Use irqsave for the lock as in the good path.
- Before removing the request, check if it is still in the list, a call to dismiss_all might have changed the list in between.
- zfcp_qdio_send does not change the queue counters on failure, trying revert something is wrong, so remove this.

---
 drivers/s390/scsi/zfcp_fsf.c |   17 ++++++-----------
 1 file changed, 6 insertions(+), 11 deletions(-)
drivers/s390/scsi/zfcp_fsf.c +6 -11
@@ -770,7 +770,6 @@ static struct zfcp_fsf_req *zfcp_fsf_req
 static int zfcp_fsf_req_send(struct zfcp_fsf_req *req)
 {
 	struct zfcp_adapter *adapter = req->adapter;
-	struct zfcp_qdio_queue *req_q = &adapter->req_q;
 	unsigned long flags;
 	int idx;
 
@@ -780,19 +779,15 @@ static int zfcp_fsf_req_send(struct zfcp
 	list_add_tail(&req->list, &adapter->req_list[idx]);
 	spin_unlock_irqrestore(&adapter->req_list_lock, flags);
 
-	req->qdio_outb_usage = atomic_read(&req_q->count);
+	req->qdio_outb_usage = atomic_read(&adapter->req_q.count);
 	req->issued = get_clock();
 	if (zfcp_qdio_send(req)) {
-		/* Queues are down..... */
 		del_timer(&req->timer);
-		spin_lock(&adapter->req_list_lock);
-		zfcp_reqlist_remove(adapter, req);
-		spin_unlock(&adapter->req_list_lock);
-		/* undo changes in request queue made for this request */
-		atomic_add(req->sbal_number, &req_q->count);
-		req_q->first -= req->sbal_number;
-		req_q->first += QDIO_MAX_BUFFERS_PER_Q;
-		req_q->first %= QDIO_MAX_BUFFERS_PER_Q; /* wrap */
+		spin_lock_irqsave(&adapter->req_list_lock, flags);
+		/* lookup request again, list might have changed */
+		if (zfcp_reqlist_find_safe(adapter, req))
+			zfcp_reqlist_remove(adapter, req);
+		spin_unlock_irqrestore(&adapter->req_list_lock, flags);
 		zfcp_erp_adapter_reopen(adapter, 0, 116, req);
 		return -EIO;
 	}
#7 lore raw

PATCH 5/8 zfcp: Fix cast warning

In reply to Christof Schmitt

From: Christof Schmitt <[email protected]>

Fix leftover from last typecast patch:

drivers/s390/scsi/zfcp_aux.c: In function ‘zfcp_port_enqueue’:
drivers/s390/scsi/zfcp_aux.c:629: warning: format ‘%016llx’ expects type ‘long long unsigned int’, but argument 3 has type ‘u64’

---
 drivers/s390/scsi/zfcp_aux.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/s390/scsi/zfcp_aux.c +2 -1
@@ -610,7 +610,8 @@ struct zfcp_port *zfcp_port_enqueue(stru
 	atomic_set_mask(status | ZFCP_STATUS_COMMON_REMOVE, &port->status);
 	atomic_set(&port->refcount, 0);
 
-	dev_set_name(&port->sysfs_device, "0x%016llx", wwpn);
+	dev_set_name(&port->sysfs_device, "0x%016llx",
+		     (unsigned long long)wwpn);
 	port->sysfs_device.parent = &adapter->ccw_device->dev;
 
 	port->sysfs_device.release = zfcp_sysfs_port_release;
Signature · 4 lines
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
#8 lore raw

PATCH 6/8 zfcp: Wait for port scan to complete when setting adapter online

In reply to Christof Schmitt

From: Christof Schmitt <[email protected]>

Attaching a unit immediately after setting the adapter online should be possible. The problem right now is that the port_scan runs from a workqueue and has not finished when the set_online call returns and the sysfs structures for the ports are not available yet. Fix that by waiting for the port scan to complete.

---
 drivers/s390/scsi/zfcp_ccw.c |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/s390/scsi/zfcp_ccw.c +3 -1
@@ -116,7 +116,9 @@ static int zfcp_ccw_set_online(struct cc
 	zfcp_erp_adapter_reopen(adapter, ZFCP_STATUS_COMMON_ERP_FAILED, 85,
 				NULL);
 	zfcp_erp_wait(adapter);
-	goto out;
+	up(&zfcp_data.config_sema);
+	flush_work(&adapter->scan_work);
+	return 0;
 
  out_scsi_register:
 	zfcp_erp_thread_kill(adapter);
#9 lore raw

PATCH 7/8 zfcp: fix erp timeout cleanup for port open requests

In reply to Christof Schmitt

From: Martin Petermann <[email protected]>

If an open port fsf request times out (in erp) the corresponding erp_action member of the fsf request need to set to NULL. If the port structure will be removed later-on there will be still a reference in the fsf request to the non existing erp_action otherwise.

---

 drivers/s390/scsi/zfcp_erp.c |    1 +
 1 file changed, 1 insertion(+)
drivers/s390/scsi/zfcp_erp.c +1 -0
@@ -472,6 +472,7 @@ static void zfcp_erp_strategy_check_fsfr
 				   ZFCP_STATUS_ERP_TIMEDOUT)) {
 			act->fsf_req->status |= ZFCP_STATUS_FSFREQ_DISMISSED;
 			zfcp_rec_dbf_event_action(142, act);
+			act->fsf_req->erp_action = NULL;
 		}
 		if (act->status & ZFCP_STATUS_ERP_TIMEDOUT)
 			zfcp_rec_dbf_event_action(143, act);
#10 lore raw

PATCH 8/8 zfcp: Fix hexdump data in s390dbf traces

In reply to Christof Schmitt

From: Christof Schmitt <[email protected]>

Fix multiple problems found in the hexdump data:
- length calculation was wrong, traces were incomplete
- FC payloads were dumped in different record than the output function tried to read
- minor fixes in output
- allow complete RSCN traces (up to 1024 bytes according to spec)

---
 drivers/s390/scsi/zfcp_dbf.c |   42 ++++++++++++++++--------------------------
 drivers/s390/scsi/zfcp_dbf.h |    8 ++------
 2 files changed, 18 insertions(+), 32 deletions(-)
drivers/s390/scsi/zfcp_dbf.c +16 -26
@@ -30,7 +30,7 @@ static void zfcp_dbf_hexdump(debug_info_
 		dump->offset = offset;
 		dump->size = min(from_len - offset, room);
 		memcpy(dump->data, from + offset, dump->size);
-		debug_event(dbf, level, dump, dump->size);
+		debug_event(dbf, level, dump, dump->size + sizeof(*dump));
 	}
 }
 
@@ -108,7 +108,7 @@ static int zfcp_dbf_view_header(debug_in
 			     t.tv_sec, t.tv_nsec);
 		zfcp_dbf_out(&p, "cpu", "%02i", entry->id.fields.cpuid);
 	} else	{
-		zfcp_dbf_outd(&p, NULL, dump->data, dump->size, dump->offset,
+		zfcp_dbf_outd(&p, "", dump->data, dump->size, dump->offset,
 			      dump->total_size);
 		if ((dump->offset + dump->size) == dump->total_size)
 			p += sprintf(p, "\n");
@@ -366,6 +366,7 @@ static void zfcp_hba_dbf_view_response(c
 			break;
 		zfcp_dbf_out(p, "scsi_cmnd", "0x%0Lx", r->u.fcp.cmnd);
 		zfcp_dbf_out(p, "scsi_serial", "0x%016Lx", r->u.fcp.serial);
+		p += sprintf(*p, "\n");
 		break;
 
 	case FSF_QTCB_OPEN_PORT_WITH_DID:
@@ -465,7 +466,8 @@ static int zfcp_hba_dbf_view_format(debu
 	else if (strncmp(r->tag, "berr", ZFCP_DBF_TAG_SIZE) == 0)
 		zfcp_hba_dbf_view_berr(&p, &r->u.berr);
 
-	p += sprintf(p, "\n");
+	if (strncmp(r->tag, "resp", ZFCP_DBF_TAG_SIZE) != 0)
+		p += sprintf(p, "\n");
 	return p - out_buf;
 }
 
@@ -880,6 +882,7 @@ void zfcp_san_dbf_event_ct_request(struc
 	struct ct_hdr *hdr = sg_virt(ct->req);
 	struct zfcp_san_dbf_record *r = &adapter->san_dbf_buf;
 	struct zfcp_san_dbf_record_ct_request *oct = &r->u.ct_req;
+	int level = 3;
 	unsigned long flags;
 
 	spin_lock_irqsave(&adapter->san_dbf_lock, flags);
@@ -896,9 +899,10 @@ void zfcp_san_dbf_event_ct_request(struc
 	oct->options = hdr->options;
 	oct->max_res_size = hdr->max_res_size;
 	oct->len = min((int)ct->req->length - (int)sizeof(struct ct_hdr),
-		       ZFCP_DBF_CT_PAYLOAD);
-	memcpy(oct->payload, (void *)hdr + sizeof(struct ct_hdr), oct->len);
-	debug_event(adapter->san_dbf, 3, r, sizeof(*r));
+		       ZFCP_DBF_SAN_MAX_PAYLOAD);
+	debug_event(adapter->san_dbf, level, r, sizeof(*r));
+	zfcp_dbf_hexdump(adapter->san_dbf, r, sizeof(*r), level,
+			 (void *)hdr + sizeof(struct ct_hdr), oct->len);
 	spin_unlock_irqrestore(&adapter->san_dbf_lock, flags);
 }
 
@@ -914,6 +918,7 @@ void zfcp_san_dbf_event_ct_response(stru
 	struct ct_hdr *hdr = sg_virt(ct->resp);
 	struct zfcp_san_dbf_record *r = &adapter->san_dbf_buf;
 	struct zfcp_san_dbf_record_ct_response *rct = &r->u.ct_resp;
+	int level = 3;
 	unsigned long flags;
 
 	spin_lock_irqsave(&adapter->san_dbf_lock, flags);
@@ -929,9 +934,10 @@ void zfcp_san_dbf_event_ct_response(stru
 	rct->expl = hdr->reason_code_expl;
 	rct->vendor_unique = hdr->vendor_unique;
 	rct->len = min((int)ct->resp->length - (int)sizeof(struct ct_hdr),
-		       ZFCP_DBF_CT_PAYLOAD);
-	memcpy(rct->payload, (void *)hdr + sizeof(struct ct_hdr), rct->len);
-	debug_event(adapter->san_dbf, 3, r, sizeof(*r));
+		       ZFCP_DBF_SAN_MAX_PAYLOAD);
+	debug_event(adapter->san_dbf, level, r, sizeof(*r));
+	zfcp_dbf_hexdump(adapter->san_dbf, r, sizeof(*r), level,
+			 (void *)hdr + sizeof(struct ct_hdr), rct->len);
 	spin_unlock_irqrestore(&adapter->san_dbf_lock, flags);
 }
 
@@ -954,7 +960,7 @@ static void zfcp_san_dbf_event_els(const
 	rec->u.els.ls_code = ls_code;
 	debug_event(adapter->san_dbf, level, rec, sizeof(*rec));
 	zfcp_dbf_hexdump(adapter->san_dbf, rec, sizeof(*rec), level,
-			 buffer, min(buflen, ZFCP_DBF_ELS_MAX_PAYLOAD));
+			 buffer, min(buflen, ZFCP_DBF_SAN_MAX_PAYLOAD));
 	spin_unlock_irqrestore(&adapter->san_dbf_lock, flags);
 }
 
@@ -1008,8 +1014,6 @@ static int zfcp_san_dbf_view_format(debu
 				    char *out_buf, const char *in_buf)
 {
 	struct zfcp_san_dbf_record *r = (struct zfcp_san_dbf_record *)in_buf;
-	char *buffer = NULL;
-	int buflen = 0, total = 0;
 	char *p = out_buf;
 
 	if (strncmp(r->tag, "dump", ZFCP_DBF_TAG_SIZE) == 0)
@@ -1029,9 +1033,6 @@ static int zfcp_san_dbf_view_format(debu
 		zfcp_dbf_out(&p, "gs_subtype", "0x%02x", ct->gs_subtype);
 		zfcp_dbf_out(&p, "options", "0x%02x", ct->options);
 		zfcp_dbf_out(&p, "max_res_size", "0x%04x", ct->max_res_size);
-		total = ct->len;
-		buffer = ct->payload;
-		buflen = min(total, ZFCP_DBF_CT_PAYLOAD);
 	} else if (strncmp(r->tag, "rctc", ZFCP_DBF_TAG_SIZE) == 0) {
 		struct zfcp_san_dbf_record_ct_response *ct = &r->u.ct_resp;
 		zfcp_dbf_out(&p, "cmd_rsp_code", "0x%04x", ct->cmd_rsp_code);
@@ -1039,23 +1040,12 @@ static int zfcp_san_dbf_view_format(debu
 		zfcp_dbf_out(&p, "reason_code", "0x%02x", ct->reason_code);
 		zfcp_dbf_out(&p, "reason_code_expl", "0x%02x", ct->expl);
 		zfcp_dbf_out(&p, "vendor_unique", "0x%02x", ct->vendor_unique);
-		total = ct->len;
-		buffer = ct->payload;
-		buflen = min(total, ZFCP_DBF_CT_PAYLOAD);
 	} else if (strncmp(r->tag, "oels", ZFCP_DBF_TAG_SIZE) == 0 ||
 		   strncmp(r->tag, "rels", ZFCP_DBF_TAG_SIZE) == 0 ||
 		   strncmp(r->tag, "iels", ZFCP_DBF_TAG_SIZE) == 0) {
 		struct zfcp_san_dbf_record_els *els = &r->u.els;
 		zfcp_dbf_out(&p, "ls_code", "0x%02x", els->ls_code);
-		total = els->len;
-		buffer = els->payload;
-		buflen = min(total, ZFCP_DBF_ELS_PAYLOAD);
 	}
-
-	zfcp_dbf_outd(&p, "payload", buffer, buflen, 0, total);
-	if (buflen == total)
-		p += sprintf(p, "\n");
-
 	return p - out_buf;
 }
 
drivers/s390/scsi/zfcp_dbf.h +2 -6
@@ -163,8 +163,6 @@ struct zfcp_san_dbf_record_ct_request {
 	u8 options;
 	u16 max_res_size;
 	u32 len;
-#define ZFCP_DBF_CT_PAYLOAD	24
-	u8 payload[ZFCP_DBF_CT_PAYLOAD];
 } __attribute__ ((packed));
 
 struct zfcp_san_dbf_record_ct_response {
@@ -174,15 +172,11 @@ struct zfcp_san_dbf_record_ct_response {
 	u8 expl;
 	u8 vendor_unique;
 	u32 len;
-	u8 payload[ZFCP_DBF_CT_PAYLOAD];
 } __attribute__ ((packed));
 
 struct zfcp_san_dbf_record_els {
 	u8 ls_code;
 	u32 len;
-#define ZFCP_DBF_ELS_PAYLOAD	32
-#define ZFCP_DBF_ELS_MAX_PAYLOAD 1024
-	u8 payload[ZFCP_DBF_ELS_PAYLOAD];
 } __attribute__ ((packed));
 
 struct zfcp_san_dbf_record {
@@ -196,6 +190,8 @@ struct zfcp_san_dbf_record {
 		struct zfcp_san_dbf_record_ct_response ct_resp;
 		struct zfcp_san_dbf_record_els els;
 	} u;
+#define ZFCP_DBF_SAN_MAX_PAYLOAD 1024
+	u8 payload[32];
 } __attribute__ ((packed));
 
 struct zfcp_scsi_dbf_record {