nvmet: don't check iosqes,iocqes for discovery controllers

BugLink: https://bugs.launchpad.net/bugs/1923214

commit d218a8a upstream.

From the base spec, Figure 78:

  "Controller Configuration, these fields are defined as parameters to
   configure an "I/O Controller (IOC)" and not to configure a "Discovery
   Controller (DC).

   ...
   If the controller does not support I/O queues, then this field shall
   be read-only with a value of 0h

Just perform this check for I/O controllers.

Fixes: a07b497 ("nvmet: add a generic NVMe target")
Reported-by: Belanger, Martin <Martin.Belanger@dell.com>
Signed-off-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Kamal Mostafa <kamal@canonical.com>
Signed-off-by: Stefan Bader <stefan.bader@canonical.com>

Author: sagigrimberg

drivers/nvme/target/core.c

@@ -1031,9 +1031,20 @@ static void nvmet_start_ctrl(struct nvmet_ctrl *ctrl)
 {
 	lockdep_assert_held(&ctrl->lock);
 
-	if (nvmet_cc_iosqes(ctrl->cc) != NVME_NVM_IOSQES ||
-	    nvmet_cc_iocqes(ctrl->cc) != NVME_NVM_IOCQES ||
-	    nvmet_cc_mps(ctrl->cc) != 0 ||
+	/*
+	 * Only I/O controllers should verify iosqes,iocqes.
+	 * Strictly speaking, the spec says a discovery controller
+	 * should verify iosqes,iocqes are zeroed, however that
+	 * would break backwards compatibility, so don't enforce it.
+	 */
+	if (ctrl->subsys->type != NVME_NQN_DISC &&
+	    (nvmet_cc_iosqes(ctrl->cc) != NVME_NVM_IOSQES ||
+	     nvmet_cc_iocqes(ctrl->cc) != NVME_NVM_IOCQES)) {
+		ctrl->csts = NVME_CSTS_CFS;
+		return;
+	}
+
+	if (nvmet_cc_mps(ctrl->cc) != 0 ||
 	    nvmet_cc_ams(ctrl->cc) != 0 ||
 	    nvmet_cc_css(ctrl->cc) != 0) {
 		ctrl->csts = NVME_CSTS_CFS;