]> git.proxmox.com Git - mirror_ubuntu-hirsute-kernel.git/commitdiff
scsi: zfcp: Clarify access to erp_action in zfcp_fsf_req_complete()
authorJulian Wiedmann <jwi@linux.ibm.com>
Thu, 10 Sep 2020 19:49:16 +0000 (21:49 +0200)
committerMartin K. Petersen <martin.petersen@oracle.com>
Tue, 15 Sep 2020 22:01:58 +0000 (18:01 -0400)
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.

Link: https://lore.kernel.org/r/c500eac301fcbba5af942bbd200f2d6b14e46994.1599765652.git.bblock@linux.ibm.com
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: Benjamin Block <bblock@linux.ibm.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
drivers/s390/scsi/zfcp_fsf.c

index 140186fe1d1e0b418246fd4478662fb0740706f9..6cb963a06777148bbf993ff13c912ea332110783 100644 (file)
@@ -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);