Commit 751e2812 authored by Jim Harris's avatar Jim Harris
Browse files

nvme: do not abort reqs in multi-process cleanup path 



When a process cleans up IO qpairs from another crashed
process in a multi-process environment, we must not try to
abort reqs for that IO qpair.  Any reqs will contain callbacks
for the crashed process which we must not try to execute in
a different process.

Fixes issue #1509.

Signed-off-by: default avatarJim Harris <james.r.harris@intel.com>
Change-Id: I5e58cce7bdb86e3feb4084733815c086901f867e

Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/3536


Community-CI: Mellanox Build Bot
Reviewed-by: default avatarShuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: default avatarBen Walker <benjamin.walker@intel.com>
Tested-by: default avatarSPDK CI Jenkins <sys_sgci@intel.com>
parent 62b3b171

lib/nvme/nvme_ctrlr.c

+10 −1
Original line number Diff line number Diff line
@@ -563,7 +563,16 @@ spdk_nvme_ctrlr_free_io_qpair(struct spdk_nvme_qpair *qpair) 


	/* Do not retry. */

	nvme_qpair_set_state(qpair, NVME_QPAIR_DESTROYING);



	/* In the multi-process case, a process may call this function on a foreign

	 * I/O qpair (i.e. one that this process did not create) when that qpairs process

	 * exits unexpectedly.  In that case, we must not try to abort any reqs associated

	 * with that qpair, since the callbacks will also be foreign to this process.

	 */

	if (qpair->active_proc == nvme_ctrlr_get_current_process(ctrlr)) {

		nvme_qpair_abort_reqs(qpair, 1);

	}



	nvme_robust_mutex_lock(&ctrlr->ctrlr_lock);



	nvme_ctrlr_proc_remove_io_qpair(qpair);

lib/nvme/nvme_transport.c

+8 −1
Original line number Diff line number Diff line
@@ -278,7 +278,14 @@ nvme_transport_ctrlr_create_io_qpair(struct spdk_nvme_ctrlr *ctrlr, uint16_t qid 
int

nvme_transport_ctrlr_delete_io_qpair(struct spdk_nvme_ctrlr *ctrlr, struct spdk_nvme_qpair *qpair)

{

	return qpair->transport->ops.ctrlr_delete_io_qpair(ctrlr, qpair);

	const struct spdk_nvme_transport *transport = nvme_get_transport(ctrlr->trid.trstring);



	/* Do not rely on qpair->transport.  For multi-process cases, a foreign process may delete

	 * the IO qpair, in which case the transport object would be invalid (each process has their

	 * own unique transport objects since they contain function pointers).  So we look up the

	 * transport object in the delete_io_qpair case.

	 */

	return transport->ops.ctrlr_delete_io_qpair(ctrlr, qpair);

}



int