Skip to contentSkip to conversation

zfcp: small changes for 5.10

PATCH Applied Benjamin Block 2 participants 6 messages Sep 10, 2020 – Sep 22, 2020

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

cover zfcp: small changes for 5.10

Hello Martin, James,

here are some small changes for zfcp I'd like to include in 5.10 if possible. They apply cleanly on Martin's scsi-queue, and James' misc branches.

Both patches make the driver a bit cleaner, and hopefully easier to maintain.

Both have been in our CI for quite a while now, running every night. I also gave them a separate regression run just now with I/O and error-injects such as cable pulls and other external error sources. So I am rather confident that they don't break anything for us - apart from that they're straight forward code changes.

As always, feedback and further reviews are appreciated :-)

Julian Wiedmann (2):
zfcp: use list_first_entry_or_null() in zfcp_erp_thread()
zfcp: clarify access to erp_action in zfcp_fsf_req_complete()

 drivers/s390/scsi/zfcp_erp.c |  8 +++-----
 drivers/s390/scsi/zfcp_fsf.c | 10 ++++++++--
 2 files changed, 11 insertions(+), 7 deletions(-)
Signature · 1 line
2.26.2
#1 lore raw

PATCH 1/2 zfcp: use list_first_entry_or_null() in zfcp_erp_thread()

From: Julian Wiedmann <[email protected]>

Use the right helper to avoid poking around in the list's internals.

---
 drivers/s390/scsi/zfcp_erp.c | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)
drivers/s390/scsi/zfcp_erp.c +3 -5
index 59e662df5774..78d52a4c55f5 100644
@@ -1607,7 +1607,6 @@ static enum zfcp_erp_act_result zfcp_erp_strategy(
 static int zfcp_erp_thread(void *data)
 {
 	struct zfcp_adapter *adapter = (struct zfcp_adapter *) data;
-	struct list_head *next;
 	struct zfcp_erp_action *act;
 	unsigned long flags;
 
@@ -1620,12 +1619,11 @@ static int zfcp_erp_thread(void *data)
 			break;
 
 		write_lock_irqsave(&adapter->erp_lock, flags);
-		next = adapter->erp_ready_head.next;
+		act = list_first_entry_or_null(&adapter->erp_ready_head,
+					       struct zfcp_erp_action, list);
 		write_unlock_irqrestore(&adapter->erp_lock, flags);
 
-		if (next != &adapter->erp_ready_head) {
-			act = list_entry(next, struct zfcp_erp_action, list);
-
+		if (act) {
 			/* there is more to come after dismission, no notify */
 			if (zfcp_erp_strategy(act) != ZFCP_ERP_DISMISSED)
 				zfcp_erp_wakeup(adapter);
Signature · 1 line
2.26.2
#2 lore raw

PATCH 2/2 zfcp: clarify access to erp_action in zfcp_fsf_req_complete()

In reply to Benjamin Block

From: Julian Wiedmann <[email protected]>

While reviewing
commit 936e6b85da04 ("scsi: zfcp: Fix panic on ERP timeout for previously dismissed ERP action"),
I stumbled over zfcp_fsf_req_complete() and wondered whether it has similar issues wrt concurrent modification of req->erp_action by zfcp_erp_strategy_check_fsfreq().

But a closer look shows that both its two callers [zfcp_fsf_reqid_check(), zfcp_fsf_req_dismiss_all()] remove the request from the adapter's req_list under the req_list's lock.
Hence we can trust that if zfcp_erp_strategy_check_fsfreq() concurrently looks up the corresponding req_id, it won't find this request and is thus unable to modify it while it's being processed by zfcp_fsf_req_complete().

Add a code comment that hopefully makes this easier for future readers, and condense the two accesses to ->erp_action that made me trip over this code path in the first place.

---
 drivers/s390/scsi/zfcp_fsf.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)
drivers/s390/scsi/zfcp_fsf.c +8 -2
index c795f22249d8..d9de26157da2 100644
@@ -426,9 +426,14 @@ static void zfcp_fsf_protstatus_eval(struct zfcp_fsf_req *req)
  * or it has been dismissed due to a queue shutdown, this function
  * is called to process the completion status and trigger further
  * events related to the FSF request.
+ * Caller must ensure that the request has been removed from
+ * adapter->req_list, to protect against concurrent modification
+ * by zfcp_erp_strategy_check_fsfreq().
  */
 static void zfcp_fsf_req_complete(struct zfcp_fsf_req *req)
 {
+	struct zfcp_erp_action *erp_action;
+
 	if (unlikely(zfcp_fsf_req_is_status_read_buffer(req))) {
 		zfcp_fsf_status_read_handler(req);
 		return;
@@ -439,8 +444,9 @@ static void zfcp_fsf_req_complete(struct zfcp_fsf_req *req)
 	zfcp_fsf_fsfstatus_eval(req);
 	req->handler(req);
 
-	if (req->erp_action)
-		zfcp_erp_notify(req->erp_action, 0);
+	erp_action = req->erp_action;
+	if (erp_action)
+		zfcp_erp_notify(erp_action, 0);
 
 	if (likely(req->status & ZFCP_STATUS_FSFREQ_CLEANUP))
 		zfcp_fsf_req_free(req);
Signature · 1 line
2.26.2
#3 lore raw
In reply to Benjamin Block

Benjamin,

Quoted 2 lines

here are some small changes for zfcp I'd like to include in 5.10 if possible.

Applied to 5.10/scsi-staging, thanks!

Signature · 1 line
Martin K. Petersen	Oracle Linux Engineering
#4 lore raw
In reply to Benjamin Block
Quoted 8 lines · Benjamin Block

here are some small changes for zfcp I'd like to include in 5.10 if possible. They apply cleanly on Martin's scsi-queue, and James' misc branches.

Both patches make the driver a bit cleaner, and hopefully easier to maintain.

[...]

Applied to 5.10/scsi-queue, thanks!

[1/2] scsi: zfcp: Use list_first_entry_or_null() in zfcp_erp_thread()

      https://git.kernel.org/mkp/scsi/c/addf13729615
[2/2] scsi: zfcp: Clarify access to erp_action in zfcp_fsf_req_complete()
      https://git.kernel.org/mkp/scsi/c/d251193d1732
Signature · 1 line
Martin K. Petersen	Oracle Linux Engineering
#5 lore raw
Quoted 9 lines · Martin K. Petersen
Quoted 7 lines · Benjamin Block

here are some small changes for zfcp I'd like to include in 5.10 if possible. They apply cleanly on Martin's scsi-queue, and James' misc branches.

Both patches make the driver a bit cleaner, and hopefully easier to maintain.

Quoted 9 lines
Quoted 1 line

[...]

Applied to 5.10/scsi-queue, thanks!

[1/2] scsi: zfcp: Use list_first_entry_or_null() in zfcp_erp_thread()

      https://git.kernel.org/mkp/scsi/c/addf13729615
[2/2] scsi: zfcp: Clarify access to erp_action in zfcp_fsf_req_complete()
      https://git.kernel.org/mkp/scsi/c/d251193d1732

Thanks Martin!

Signature · 4 lines
Best Regards, Benjamin Block  / Linux on IBM Z Kernel Development / IBM Systems
IBM Deutschland Research & Development GmbH    /    https://www.ibm.com/privacy
Vorsitz. AufsR.: Gregor Pillen         /        Gesch�ftsf�hrung: Dirk Wittkopp
Sitz der Gesellschaft: B�blingen / Registergericht: AmtsG Stuttgart, HRB 243294