Commit 47f2a233 authored by Darek Stojaczyk's avatar Darek Stojaczyk Committed by Jim Harris
Browse files

nvme: improve probe error handling 



The underlying probe might fail, but we don't check its
return code. Right now we ignore the failure and in secondary
process we even continue referencing a locally-unitialized
controller struct. Then, a few calls later, we fail on assert
because current process is not on the ctrlr->active_procs list.

Change-Id: I65a59a9515a8e0196b60a181cee2af33434784dc
Signed-off-by: default avatarDarek Stojaczyk <dariusz.stojaczyk@intel.com>
Reviewed-on: https://review.gerrithub.io/431486


Tested-by: default avatarSPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: default avatarJim Harris <james.r.harris@intel.com>
Reviewed-by: default avatarShuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: default avatarBen Walker <benjamin.walker@intel.com>
parent 7f128c75

lib/nvme/nvme.c

+6 −1
Original line number Diff line number Diff line
@@ -518,7 +518,12 @@ spdk_nvme_probe_internal(const struct spdk_nvme_transport_id *trid, void *cb_ctx 


	nvme_robust_mutex_lock(&g_spdk_nvme_driver->lock);



	nvme_transport_ctrlr_scan(trid, cb_ctx, probe_cb, remove_cb, direct_connect);

	rc = nvme_transport_ctrlr_scan(trid, cb_ctx, probe_cb, remove_cb, direct_connect);

	if (rc != 0) {

		SPDK_ERRLOG("NVMe ctrlr scan failed\n");

		nvme_robust_mutex_unlock(&g_spdk_nvme_driver->lock);

		return -1;

	}



	/*

	 * Probe controllers on the shared_attached_ctrlrs list