SPDK 21.01.2-pre

Signed-off-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
Change-Id: I473719d4142702572ff4a51858bcf8cb86ef65ea
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6684
Reviewed-by: Ben Walker <benjamin.walker@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Jim Harris <james.r.harris@intel.com>
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>

CHANGELOG.md

@@ -1,6 +1,39 @@
 # Changelog
 
-## v21.01: (Upcoming Release)
+## v21.01.2: (Upcoming Release)
+
+## v21.01.1:
+
+### dpdk
+
+Added `rte_ethdev` and `rte_net` dependencies for all builds with DPDK.
+
+Fixed compatibility issues with DPDK 19.11.
+
+### iscsi
+
+A security vulnerability has been identified and fixed in the SPDK iSCSI target.
+A TEXT PDU with no data, but CONTINUE flag set, would result in a NULL pointer dereference
+and crash the SPDK iSCSI target process. All users of the SPDK iSCSI target
+are recommended to update. All SPDK versions <= v21.01 are affected.
+
+### nbd
+
+Fixed kernel hang when bdev is removed by always setting NBD_SET_TIMEOUT.
+
+### nvme
+
+Fixed segfault when removing qpair when transport connection fails (issue #1777).
+
+### ocssd
+
+Fixed the bug that no media event is pushed to the target bdev.
+
+### sock
+
+Added `enable_quickack` and `enable_placement_id` when saving JSON configuration.
+
+## v21.01: Dynamic scheduler, vfio-user, ZNS Zone Append
 
 ### bdev
 
@@ -75,6 +108,10 @@ The `--pci-blacklist` command line option has been deprecated, replaced with
 The `--pci-whitelist/-W` command line options have been deprecated, replaced with
 `--pci-allowed/-A`.
 
+Added new experimental `dynamic` scheduler that rebalances idle threads, adjusts CPU frequency
+using dpdk_governor and turns idle reactor cores to interrupt mode. Please see
+[scheduler documentation](https://www.spdk.io/doc/scheduler.html) for details.
+
 ## ioat
 
 The PCI BDF whitelist option has been removed from the `ioat_scan_accel_engine` RPC.

CONFIG

@@ -121,7 +121,7 @@ CONFIG_VHOST=y
 CONFIG_VIRTIO=y
 
 # Build custom vfio-user transport for NVMf target and NVMe initiator.
-CONFIG_VFIO_USER=y
+CONFIG_VFIO_USER=n
 
 # Build with PMDK backends
 CONFIG_PMDK=n

app/spdk_top/spdk_top.c

@@ -2027,6 +2027,7 @@ show_thread(uint8_t current_page)
 
 	get_data();
 
+	assert(thread_number < g_threads_stats.threads.threads_count);
 	for (i = 0; i < g_threads_stats.threads.threads_count; i++) {
 		thread_info[i] = &g_threads_stats.threads.thread_info[i];
 	}
@@ -2067,6 +2068,7 @@ show_core(uint8_t current_page)
 
 	get_data();
 
+	assert(core_number < g_cores_stats.cores.cores_count);
 	for (i = 0; i < g_cores_stats.cores.cores_count; i++) {
 		core_info[i] = &g_cores_stats.cores.core[i];
 	}
@@ -2178,6 +2180,7 @@ show_poller(uint8_t current_page)
 	get_data();
 
 	prepare_poller_data(current_page, pollers, &count, current_page);
+	assert(poller_number < count);
 
 	poller_win = newwin(POLLER_WIN_HEIGHT, POLLER_WIN_WIDTH,
 			    (g_max_row - poller_counter) / 2, (g_max_col - POLLER_WIN_HOR_POS) / 2);
@@ -2300,6 +2303,7 @@ show_stats(void)
 			} else {
 				active_tab = THREADS_TAB;
 			}
+			g_selected_row = 0;
 			current_page = 0;
 			switch_tab(active_tab);
 			break;

autorun.sh

@@ -4,7 +4,8 @@ set -e
 
 rootdir=$(readlink -f $(dirname $0))
 
-conf=~/autorun-spdk.conf
+default_conf=~/autorun-spdk.conf
+conf=${1:-${default_conf}}
 
 # If the configuration of tests is not provided, no tests will be carried out.
 if [[ ! -f $conf ]]; then

configure

@@ -454,36 +454,22 @@ else
 fi
 BUILD_CMD+=(-I/usr/local/include -L/usr/local/lib)
 
-function set_os_id_version() {
-	if [[ -f /etc/os-release ]]; then
-		source /etc/os-release
-	fi
-
-	OSID=$ID
-	OSVERSION=$VERSION_ID
-
-	echo "OS-ID: $OSID | OS-Version: $OSVERSION"
-}
-
 if [[ "${CONFIG[VFIO_USER]}" = "y" ]]; then
-	set_os_id_version
 
-	if [[ $arch != x86_64* ]] || [[ $sys_name == "FreeBSD" ]]; then
+	if ! hash cmake; then
-		echo "Non x86_64 and Linux platform, disable CONFIG_VFIO_USER"
+		echo "ERROR: --with-vfio-user requires cmake"
-		CONFIG[VFIO_USER]="n"
+		echo "Please install then re-run this script"
-	# disable tests on ubuntu16 due to lack of macro definition in pci_regs.h
+		exit 1
-	elif [[ $OSID == "ubuntu" ]] && [[ $OSVERSION == "16.04" ]]; then
+	fi
-		echo "ubuntu16 OS, disable CONFIG_VFIO_USER"
+	if [[ ! -d /usr/include/json-c ]] && [[ ! -d /usr/local/include/json-c ]]; then
-		CONFIG[VFIO_USER]="n"
+		echo "ERROR: --with-vfio-user requires json-c-devel"
-	elif ! hash cmake; then
+		echo "Please install then re-run this script"
-		echo "cmake not installed, disable CONFIG_VFIO_USER"
+		exit 1
-		CONFIG[VFIO_USER]="n"
+	fi
-	elif [[ ! -d /usr/include/json-c ]] && [[ ! -d /usr/local/include/json-c ]]; then
+	if [[ ! -e /usr/include/cmocka.h ]] && [[ ! -e /usr/local/include/cmocka.h ]]; then
-		echo "json-c-devel not installed, disable CONFIG_VFIO_USER"
+		echo "ERROR: --with-vfio-user requires libcmocka-devel"
-		CONFIG[VFIO_USER]="n"
+		echo "Please install then re-run this script"
-	elif [[ ! -e /usr/include/cmocka.h ]] && [[ ! -e /usr/local/include/cmocka.h ]]; then
+		exit 1
-		echo "libcmocka-devel not installed, disable CONFIG_VFIO_USER"
-		CONFIG[VFIO_USER]="n"
 	fi
 fi
 

doc/Doxyfile

@@ -835,6 +835,7 @@ INPUT                  += \
                          peer_2_peer.md \
                          pkgconfig.md \
                          porting.md \
+                         scheduler.md \
                          shfmt.md \
                          spdkcli.md \
                          spdk_top.md \

doc/general.md

@@ -1,5 +1,6 @@
 # General Information {#general}
 
 - @subpage event
+- @subpage scheduler
 - @subpage logical_volumes
 - @subpage accel_fw

doc/jsonrpc.md

@@ -78,7 +78,7 @@ A detailed description of each RPC method and its parameters is also available.
 scripts/rpc.py bdev_nvme_attach_controller --help
 ~~~
 
-### Generate JSON-RPC methods for current configuration
+### Generate JSON-RPC methods for current configuration {#jsonrpc_generate}
 
 An initial configuration can be specified for an SPDK application via the '-c' command line parameter.
 The configuration file is a JSON file containing all of the JSON-RPC method invocations necessary
@@ -206,6 +206,18 @@ Finally, call the rpc.py script with '--plugin' parameter to provide above pytho
 ./scripts/rpc.py --plugin rpc_plugin bdev_example_create 10 4096
 ~~~
 
+### Converting from legacy configuration {#jsonrpc_convert}
+
+Starting with SPDK 20.10, legacy configuration file support has been removed.
+Users with extensive configuration files already running in SPDK application,
+can [generate JSON-RPC for current configuration](@ref jsonrpc_generate).
+
+If binary for deploying the application is unavailable, the legacy configuration
+file can be converted to JSON-RPC using python tool:
+~~~
+./scripts/config_converter.py < config.ini > config.json
+~~~
+
 # App Framework {#jsonrpc_components_app}
 
 ## spdk_kill_instance {#rpc_spdk_kill_instance}

doc/nvmf.md

@@ -116,10 +116,6 @@ An NVMe over Fabrics target can be configured using JSON RPCs.
 The basic RPCs needed to configure the NVMe-oF subsystem are detailed below. More information about
 working with NVMe over Fabrics specific RPCs can be found on the @ref jsonrpc_components_nvmf_tgt RPC page.
 
-Using .ini style configuration files for configuration of the NVMe-oF target is deprecated and should
-be replaced with JSON based RPCs. .ini style configuration files can be converted to json format by way
-of the new script `scripts/config_converter.py`.
-
 ## FC transport support {#nvmf_fc_transport}
 
 To build nvmf_tgt with the FC transport, there is an additional FC LLD (Low Level Driver) code dependency.

doc/scheduler.md

@@ -0,0 +1,82 @@
+# Scheduler {#scheduler}
+
+SPDK's event/application framework (`lib/event`) now supports scheduling of
+lightweight threads. Schedulers are provided as plugins, called
+implementations. A default implementation is provided, but users may wish to
+write their own scheduler to integrate into broader code frameworks or meet
+their performance needs.
+
+This feature should be considered experimental and is disabled by default. When
+enabled, the scheduler framework gathers data for each spdk thread and reactor
+and passes it to a scheduler implementation to perform one of the following
+actions.
+
+## Actions
+
+### Move a thread
+
+`spdk_thread`s can be moved to another reactor. Schedulers can examine the
+suggested cpu_mask value for each lightweight thread to see if the user has
+requested specific reactors, or choose a reactor using whatever algorithm they
+deem fit.
+
+### Switch reactor mode
+
+Reactors by default run in a mode that constantly polls for new actions for the
+most efficient processing. Schedulers can switch a reactor into a mode that
+instead waits for an event on a file descriptor. On Linux, this is implemented
+using epoll. This results in reduced CPU usage but may be less responsive when
+events occur. A reactor cannot enter this mode if any `spdk_threads` are
+currently scheduled to it. This limitation is expected to be lifted in the
+future, allowing `spdk_threads` to enter interrupt mode.
+
+### Set frequency of CPU core
+
+The frequency of CPU cores can be modified by the scheduler in response to
+load. Only CPU cores that match the application cpu_mask may be modified. The
+mechanism for controlling CPU frequency is pluggable and the default provided
+implementation is called `dpdk_governor`, based on the `rte_power` library from
+DPDK.
+
+#### Known limitation
+
+When SMT (Hyperthreading) is enabled the two logical CPU cores sharing a single
+physical CPU core must run at the same frequency. If one of two of such logical
+CPU cores is outside the application cpu_mask, the policy and frequency on that
+core has to be managed by the administrator.
+
+## Scheduler implementations
+
+The scheduler in use may be controlled by JSON-RPC. Please use the
+[framework_set_scheduler](jsonrpc.md/#rpc_framework_set_scheduler) RPC to
+switch between schedulers or change their options.
+
+[spdk_top](spdk_top.md#spdk_top) is a useful tool to observe the behavior of
+schedulers in different scenarios and workloads.
+
+### static [default]
+
+The `static` scheduler is the default scheduler and does no dynamic scheduling.
+Lightweight threads are distributed round-robin among reactors, respecting
+their requested cpu_mask, and then they are never moved. This is equivalent to
+the previous behavior of the SPDK event/application framework.
+
+### dynamic
+
+The `dynamic` scheduler is designed for power saving and reduction of CPU
+utilization, especially in cases where workloads show large variations over
+time.
+
+Active threads are distributed equally among reactors, taking cpu_mask into
+account. All idle threads are moved to the main core. Once an idle thread becomes
+active, it is redistributed again.
+
+When a reactor has no scheduled `spdk_thread`s it is switched into interrupt
+mode and stops actively polling. After enough threads become active, the
+reactor is switched back into poll mode and threads are assigned to it again.
+
+The main core can contain active threads only when their execution time does
+not exceed the sum of all idle threads. When no active threads are present on
+the main core, the frequency of that CPU core will decrease as the load
+decreases. All CPU cores corresponding to the other reactors remain at maximum
+frequency.

include/spdk/version.h

@@ -54,7 +54,7 @@
  * Patch level is incremented on maintenance branch releases and reset to 0 for each
  * new major.minor release.
  */
-#define SPDK_VERSION_PATCH	0
+#define SPDK_VERSION_PATCH	2
 
 /**
  * Version string suffix.

lib/env_dpdk/env.mk

@@ -51,7 +51,7 @@ DPDK_INC := -I$(DPDK_INC_DIR)
 DPDK_LIB_LIST = rte_eal rte_mempool rte_ring rte_mbuf rte_pci rte_bus_pci rte_mempool_ring
 
 ifeq ($(OS),Linux)
-DPDK_LIB_LIST += rte_power
+DPDK_LIB_LIST += rte_power rte_ethdev rte_net
 endif
 
 # DPDK 20.05 eal dependency
@@ -101,7 +101,7 @@ endif
 LINK_HASH=n
 
 ifeq ($(CONFIG_VHOST),y)
-DPDK_LIB_LIST += rte_vhost rte_net
+DPDK_LIB_LIST += rte_vhost
 LINK_HASH=y
 ifneq ($(DPDK_FRAMEWORK),y)
 DPDK_LIB_LIST += rte_cryptodev

lib/env_dpdk/memory.c

@@ -1238,7 +1238,11 @@ vtophys_iommu_device_event(const char *device_name,
 		RTE_DEV_FOREACH(dev, "bus=pci", &dev_iter) {
 			if (strcmp(dev->name, device_name) == 0) {
 				struct rte_pci_device *pci_dev = RTE_DEV_TO_PCI(dev);
+#if RTE_VERSION < RTE_VERSION_NUM(20, 11, 0, 0)
+				if (pci_dev->kdrv == RTE_KDRV_VFIO) {
+#else
 				if (pci_dev->kdrv == RTE_PCI_KDRV_VFIO) {
+#endif
 					/* This is a new PCI device using vfio */
 					g_vfio.device_ref++;
 				}
@@ -1267,7 +1271,11 @@ vtophys_iommu_device_event(const char *device_name,
 		RTE_DEV_FOREACH(dev, "bus=pci", &dev_iter) {
 			if (strcmp(dev->name, device_name) == 0) {
 				struct rte_pci_device *pci_dev = RTE_DEV_TO_PCI(dev);
+#if RTE_VERSION < RTE_VERSION_NUM(20, 11, 0, 0)
+				if (pci_dev->kdrv == RTE_KDRV_VFIO) {
+#else
 				if (pci_dev->kdrv == RTE_PCI_KDRV_VFIO) {
+#endif
 					/* This is a PCI device using vfio */
 					g_vfio.device_ref--;
 				}
@@ -1362,7 +1370,11 @@ vtophys_iommu_init(void)
 	RTE_DEV_FOREACH(dev, "bus=pci", &dev_iter) {
 		struct rte_pci_device *pci_dev = RTE_DEV_TO_PCI(dev);
 
+#if RTE_VERSION < RTE_VERSION_NUM(20, 11, 0, 0)
+		if (pci_dev->kdrv == RTE_KDRV_VFIO) {
+#else
 		if (pci_dev->kdrv == RTE_PCI_KDRV_VFIO) {
+#endif
 			/* This is a PCI device using vfio */
 			g_vfio.device_ref++;
 		}

lib/event/reactor.c

@@ -1010,8 +1010,10 @@ void
 spdk_reactors_start(void)
 {
 	struct spdk_reactor *reactor;
+	struct spdk_cpuset tmp_cpumask = {};
 	uint32_t i, current_core;
 	int rc;
+	char thread_name[32];
 
 	g_rusage_period = (CONTEXT_SWITCH_MONITOR_PERIOD * spdk_get_ticks_hz()) / SPDK_SEC_TO_USEC;
 	g_reactor_state = SPDK_REACTOR_STATE_RUNNING;
@@ -1030,6 +1032,14 @@ spdk_reactors_start(void)
 				assert(false);
 				return;
 			}
+
+			/* For now, for each reactor spawn one thread. */
+			snprintf(thread_name, sizeof(thread_name), "reactor_%u", reactor->lcore);
+
+			spdk_cpuset_zero(&tmp_cpumask);
+			spdk_cpuset_set_cpu(&tmp_cpumask, i, true);
+
+			spdk_thread_create(thread_name, &tmp_cpumask);
 		}
 		spdk_cpuset_set_cpu(&g_reactor_core_mask, i, true);
 	}

lib/iscsi/param.c

@@ -315,6 +315,16 @@ iscsi_parse_params(struct iscsi_param **params, const uint8_t *data,
 	char *p;
 	int i;
 
+	/* Spec does not disallow TEXT PDUs with zero length, just return
+	 * immediately in that case, since there is no param data to parse
+	 * and any existing partial parameter would remain as-is.
+	 */
+	if (len == 0) {
+		return 0;
+	}
+
+	assert(data != NULL);
+
 	/* strip the partial text parameters if previous PDU have C enabled */
 	if (partial_parameter && *partial_parameter) {
 		for (i = 0; i < len && data[i] != '\0'; i++) {

lib/nbd/nbd.c

@@ -50,6 +50,7 @@
 #define GET_IO_LOOP_COUNT		16
 #define NBD_BUSY_WAITING_MS		1000
 #define NBD_BUSY_POLLING_INTERVAL_US	20000
+#define NBD_IO_TIMEOUT_S		60
 
 enum nbd_io_state_t {
 	/* Receiving or ready to receive nbd request header */
@@ -922,6 +923,17 @@ nbd_start_complete(struct spdk_nbd_start_ctx *ctx)
 		goto err;
 	}
 
+#ifdef NBD_SET_TIMEOUT
+	rc = ioctl(ctx->nbd->dev_fd, NBD_SET_TIMEOUT, NBD_IO_TIMEOUT_S);
+	if (rc == -1) {
+		SPDK_ERRLOG("ioctl(NBD_SET_TIMEOUT) failed: %s\n", spdk_strerror(errno));
+		rc = -errno;
+		goto err;
+	}
+#else
+	SPDK_NOTICELOG("ioctl(NBD_SET_TIMEOUT) is not supported.\n");
+#endif
+
 #ifdef NBD_FLAG_SEND_TRIM
 	rc = ioctl(ctx->nbd->dev_fd, NBD_SET_FLAGS, NBD_FLAG_SEND_TRIM);
 	if (rc == -1) {

lib/nvme/nvme_ctrlr.c

@@ -418,8 +418,12 @@ spdk_nvme_ctrlr_alloc_io_qpair(struct spdk_nvme_ctrlr *ctrlr,
 	rc = spdk_nvme_ctrlr_connect_io_qpair(ctrlr, qpair);
 	if (rc != 0) {
 		SPDK_ERRLOG("nvme_transport_ctrlr_connect_io_qpair() failed\n");
+		nvme_robust_mutex_lock(&ctrlr->ctrlr_lock);
+		nvme_ctrlr_proc_remove_io_qpair(qpair);
 		TAILQ_REMOVE(&ctrlr->active_io_qpairs, qpair, tailq);
+		spdk_bit_array_set(ctrlr->free_io_qids, qpair->id);
 		nvme_transport_ctrlr_delete_io_qpair(ctrlr, qpair);
+		nvme_robust_mutex_unlock(&ctrlr->ctrlr_lock);
 		return NULL;
 	}
 

lib/nvmf/rdma.c

@@ -2,7 +2,7 @@
  *   BSD LICENSE
  *
  *   Copyright (c) Intel Corporation. All rights reserved.
- *   Copyright (c) 2019, 2020 Mellanox Technologies LTD. All rights reserved.
+ *   Copyright (c) 2019-2021 Mellanox Technologies LTD. All rights reserved.
  *
  *   Redistribution and use in source and binary forms, with or without
  *   modification, are permitted provided that the following conditions
@@ -944,6 +944,11 @@ nvmf_rdma_resize_cq(struct spdk_nvmf_rdma_qpair *rqpair, struct spdk_nvmf_rdma_d
 	}
 
 	if (rpoller->num_cqe != num_cqe) {
+		if (device->context->device->transport_type == IBV_TRANSPORT_IWARP) {
+			SPDK_ERRLOG("iWARP doesn't support CQ resize. Current capacity %u, required %u\n"
+				    "Using CQ of insufficient size may lead to CQ overrun\n", rpoller->num_cqe, num_cqe);
+			return -1;
+		}
 		if (required_num_wr > device->attr.max_cqe) {
 			SPDK_ERRLOG("RDMA CQE requirement (%d) exceeds device max_cqe limitation (%d)\n",
 				    required_num_wr, device->attr.max_cqe);
@@ -4106,19 +4111,24 @@ nvmf_rdma_qpair_abort_request(struct spdk_nvmf_qpair *qpair,
 	struct spdk_nvmf_rdma_transport *rtransport;
 	struct spdk_nvmf_transport *transport;
 	uint16_t cid;
-	uint32_t i;
+	uint32_t i, max_req_count;
-	struct spdk_nvmf_rdma_request *rdma_req_to_abort = NULL;
+	struct spdk_nvmf_rdma_request *rdma_req_to_abort = NULL, *rdma_req;
 
 	rqpair = SPDK_CONTAINEROF(qpair, struct spdk_nvmf_rdma_qpair, qpair);
 	rtransport = SPDK_CONTAINEROF(qpair->transport, struct spdk_nvmf_rdma_transport, transport);
 	transport = &rtransport->transport;
 
 	cid = req->cmd->nvme_cmd.cdw10_bits.abort.cid;
+	max_req_count = rqpair->srq == NULL ? rqpair->max_queue_depth : rqpair->poller->max_srq_depth;
 
-	for (i = 0; i < rqpair->max_queue_depth; i++) {
+	for (i = 0; i < max_req_count; i++) {
-		if (rqpair->resources->reqs[i].state != RDMA_REQUEST_STATE_FREE &&
+		rdma_req = &rqpair->resources->reqs[i];
-		    rqpair->resources->reqs[i].req.cmd->nvme_cmd.cid == cid) {
+		/* When SRQ == NULL, rqpair has its own requests and req.qpair pointer always points to the qpair
-			rdma_req_to_abort = &rqpair->resources->reqs[i];
+		 * When SRQ != NULL all rqpairs share common requests and qpair pointer is assigned when we start to
+		 * process a request. So in both cases all requests which are not in FREE state have valid qpair ptr */
+		if (rdma_req->state != RDMA_REQUEST_STATE_FREE && rdma_req->req.cmd->nvme_cmd.cid == cid &&
+		    rdma_req->req.qpair == qpair) {
+			rdma_req_to_abort = rdma_req;
 			break;
 		}
 	}

lib/sock/sock.c

@@ -832,6 +832,8 @@ spdk_sock_write_config_json(struct spdk_json_write_ctx *w)
 			spdk_json_write_named_uint32(w, "send_buf_size", opts.send_buf_size);
 			spdk_json_write_named_bool(w, "enable_recv_pipe", opts.enable_recv_pipe);
 			spdk_json_write_named_bool(w, "enable_zerocopy_send", opts.enable_zerocopy_send);
+			spdk_json_write_named_bool(w, "enable_quickack", opts.enable_quickack);
+			spdk_json_write_named_bool(w, "enable_placement_id", opts.enable_placement_id);
 			spdk_json_write_object_end(w);
 			spdk_json_write_object_end(w);
 		} else {

lib/util/fd_group.c

@@ -85,8 +85,8 @@ int
 spdk_fd_group_add(struct spdk_fd_group *fgrp,
 		  int efd, spdk_fd_fn fn, void *arg)
 {
-	struct event_handler *ehdlr;
+	struct event_handler *ehdlr = NULL;
-	struct epoll_event epevent;
+	struct epoll_event epevent = {0};
 	int rc;
 
 	/* parameter checking */

lib/vhost/rte_vhost_compat.c

@@ -46,6 +46,7 @@
 #include "spdk/barrier.h"
 #include "spdk/vhost.h"
 #include "vhost_internal.h"
+#include <rte_version.h>
 
 #include "spdk_internal/vhost_user.h"
 
@@ -331,7 +332,11 @@ vhost_register_unix_socket(const char *path, const char *ctrl_name,
 		}
 	}
 
+#if RTE_VERSION < RTE_VERSION_NUM(20, 8, 0, 0)
 	if (rte_vhost_driver_register(path, 0) != 0) {
+#else
+	if (rte_vhost_driver_register(path, RTE_VHOST_USER_ASYNC_COPY) != 0) {
+#endif
 		SPDK_ERRLOG("Could not register controller %s with vhost library\n", ctrl_name);
 		SPDK_ERRLOG("Check if domain socket %s already exists\n", path);
 		return -EIO;

lib/vhost/vhost.c

@@ -888,6 +888,7 @@ static int
 vhost_parse_core_mask(const char *mask, struct spdk_cpuset *cpumask)
 {
 	int rc;
+	struct spdk_cpuset negative_vhost_mask;
 
 	if (cpumask == NULL) {
 		return -1;
@@ -904,6 +905,16 @@ vhost_parse_core_mask(const char *mask, struct spdk_cpuset *cpumask)
 		return -1;
 	}
 
+	spdk_cpuset_copy(&negative_vhost_mask, &g_vhost_core_mask);
+	spdk_cpuset_negate(&negative_vhost_mask);
+	spdk_cpuset_and(&negative_vhost_mask, cpumask);
+
+	if (spdk_cpuset_count(&negative_vhost_mask) != 0) {
+		SPDK_ERRLOG("one of selected cpu is outside of core mask(=%s)\n",
+			    spdk_cpuset_fmt(&g_vhost_core_mask));
+		return -1;
+	}
+
 	spdk_cpuset_and(cpumask, &g_vhost_core_mask);
 
 	if (spdk_cpuset_count(cpumask) == 0) {

module/bdev/nvme/bdev_ocssd.c

@@ -967,7 +967,7 @@ bdev_ocssd_push_media_events(struct nvme_bdev_ns *nvme_ns,
 	TAILQ_FOREACH(nvme_bdev, &nvme_ns->bdevs, tailq) {
 		ocssd_bdev = SPDK_CONTAINEROF(nvme_bdev, struct ocssd_bdev, nvme_bdev);
 		if (bdev_ocssd_lba_in_range(ocssd_bdev, ocssd_ns, chunk_entry->lba)) {
-			return;
+			break;
 		}
 	}
 

pkg/spdk.spec

@@ -2,12 +2,12 @@
 %bcond_with doc
 
 Name: spdk
-Version: master
+Version: 21.01.x
 Release: 0%{?dist}
 Epoch: 0
 URL: http://spdk.io
 
-Source: https://github.com/spdk/spdk/archive/master.tar.gz
+Source: https://github.com/spdk/spdk/archive/v21.01.x.tar.gz
 Summary: Set of libraries and utilities for high performance user-mode storage
 
 %define package_version %{epoch}:%{version}-%{release}
@@ -22,7 +22,7 @@ License: BSD
 ExclusiveArch: x86_64
 
 BuildRequires: gcc gcc-c++ make
-BuildRequires: dpdk-devel, numactl-devel
+BuildRequires: dpdk-devel, numactl-devel, ncurses-devel
 BuildRequires: libiscsi-devel, libaio-devel, openssl-devel, libuuid-devel
 BuildRequires: libibverbs-devel, librdmacm-devel
 %if %{with doc}
@@ -32,6 +32,7 @@ BuildRequires: doxygen mscgen graphviz
 # Install dependencies
 Requires: dpdk >= 19.11, numactl-libs, openssl-libs
 Requires: libiscsi, libaio, libuuid
+Requires: python3-configshell, python3-pexpect
 # NVMe over Fabrics
 Requires: librdmacm, librdmacm
 Requires(post): /sbin/ldconfig

scripts/vagrant/autorun-spdk.conf

@@ -9,6 +9,7 @@ SPDK_TEST_ISCSI_INITIATOR=0
 SPDK_TEST_NVME=0
 SPDK_TEST_NVME_CLI=0
 SPDK_TEST_NVMF=1
+SPDK_TEST_VFIOUSER=1
 SPDK_TEST_RBD=0
 SPDK_TEST_CRYPTO=0
 SPDK_TEST_OCF=0

test/common/autotest_common.sh

@@ -81,6 +81,8 @@ export SPDK_TEST_NVME_CLI
 export SPDK_TEST_NVME_CUSE
 : ${SPDK_TEST_NVMF=0}
 export SPDK_TEST_NVMF
+: ${SPDK_TEST_VFIOUSER=0}
+export SPDK_TEST_VFIOUSER
 : ${SPDK_TEST_NVMF_TRANSPORT="rdma"}
 export SPDK_TEST_NVMF_TRANSPORT
 : ${SPDK_TEST_RBD=0}
@@ -422,6 +424,10 @@ function get_config_params() {
 		config_params+=' --with-raid5'
 	fi
 
+	if [ $SPDK_TEST_VFIOUSER -eq 1 ]; then
+		config_params+=' --with-vfio-user'
+	fi
+
 	# Check whether liburing library header exists
 	if [ -f /usr/include/liburing/io_uring.h ] && [ $SPDK_TEST_URING -eq 1 ]; then
 		config_params+=' --with-uring'

test/common/config/vm_setup.sh

@@ -185,6 +185,7 @@ SPDK_TEST_ISCSI_INITIATOR=1
 SPDK_TEST_NVME=1
 SPDK_TEST_NVME_CLI=1
 SPDK_TEST_NVMF=1
+SPDK_TEST_VFIOUSER=1
 SPDK_TEST_RBD=1
 SPDK_TEST_BLOCKDEV=1
 SPDK_TEST_BLOBFS=1

test/make/check_so_deps.sh

@@ -49,17 +49,6 @@ function confirm_abi_deps() {
 	fi
 
 	cat << EOF > ${suppression_file}
-[suppress_type]
-	name = spdk_nvme_ctrlr_data
-[suppress_type]
-	type_kind = enum
-	changed_enumerators = SPDK_BDEV_NUM_IO_TYPES
-[suppress_type]
-	name = spdk_env_opts
-[suppress_type]
-	name = spdk_app_opts
-[suppress_type]
-	name = spdk_thread
 EOF
 
 	for object in "$libdir"/libspdk_*.so; do

test/nvmf/nvmf.sh

@@ -41,7 +41,7 @@ run_test "nvmf_invalid" test/nvmf/target/invalid.sh "${TEST_ARGS[@]}"
 run_test "nvmf_abort" test/nvmf/target/abort.sh "${TEST_ARGS[@]}"
 run_test "nvmf_ns_hotplug_stress" test/nvmf/target/ns_hotplug_stress.sh "${TEST_ARGS[@]}"
 
-if grep -q '#define SPDK_CONFIG_VFIO_USER 1' $rootdir/include/spdk/config.h; then
+if [ $SPDK_TEST_VFIOUSER -eq 1 ]; then
 	run_test "nvmf_vfio_user" test/nvmf/target/nvmf_vfio_user.sh "${TEST_ARGS[@]}"
 fi
 

test/spdkcli/match_files/spdkcli_details_vhost_ctrl.test.match

@@ -14,7 +14,7 @@
       }
     ]
   },
-  "cpumask": "$(S)",
+  "cpumask": "0x3",
   "ctrlr": "vhost_scsi1",
   "delay_base_us": 20,
   "iops_threshold": 1000000,

test/unit/lib/iscsi/iscsi.c/iscsi_ut.c

@@ -1993,6 +1993,43 @@ pdu_hdr_op_data_test(void)
 	g_task_pool_is_empty = false;
 }
 
+/* Test an ISCSI_OP_TEXT PDU with CONTINUE bit set but
+ * no data.
+ */
+static void
+empty_text_with_cbit_test(void)
+{
+	struct spdk_iscsi_sess sess = {};
+	struct spdk_iscsi_conn conn = {};
+	struct spdk_scsi_dev dev = {};
+	struct spdk_iscsi_pdu *req_pdu;
+	int rc;
+
+	req_pdu = iscsi_get_pdu(&conn);
+
+	sess.ExpCmdSN = 0;
+	sess.MaxCmdSN = 64;
+	sess.session_type = SESSION_TYPE_NORMAL;
+	sess.MaxBurstLength = 1024;
+
+	conn.full_feature = 1;
+	conn.sess = &sess;
+	conn.dev = &dev;
+	conn.state = ISCSI_CONN_STATE_RUNNING;
+
+	memset(&req_pdu->bhs, 0, sizeof(req_pdu->bhs));
+	req_pdu->bhs.opcode = ISCSI_OP_TEXT;
+	req_pdu->bhs.flags = ISCSI_TEXT_CONTINUE;
+
+	rc = iscsi_pdu_hdr_handle(&conn, req_pdu);
+	CU_ASSERT(rc == 0);
+	CU_ASSERT(!req_pdu->is_rejected);
+	rc = iscsi_pdu_payload_handle(&conn, req_pdu);
+	CU_ASSERT(rc == 0);
+
+	iscsi_put_pdu(req_pdu);
+}
+
 int
 main(int argc, char **argv)
 {
@@ -2024,6 +2061,7 @@ main(int argc, char **argv)
 	CU_ADD_TEST(suite, pdu_hdr_op_task_mgmt_test);
 	CU_ADD_TEST(suite, pdu_hdr_op_nopout_test);
 	CU_ADD_TEST(suite, pdu_hdr_op_data_test);
+	CU_ADD_TEST(suite, empty_text_with_cbit_test);
 
 	CU_basic_set_mode(CU_BRM_VERBOSE);
 	CU_basic_run_tests();

test/unit/lib/iscsi/param.c/param_ut.c

@@ -264,6 +264,14 @@ parse_valid_test(void)
 	EXPECT_VAL("F", "IIII");
 	CU_ASSERT_PTR_NULL(partial_parameter);
 
+	/* partial parameter: NULL data */
+	/* It is technically allowed to have a TEXT PDU with no data, yet
+	 * CONTINUE bit is enabled - make sure we handle that case correctly.
+	 */
+	rc = iscsi_parse_params(&params, NULL, 0, true, &partial_parameter);
+	CU_ASSERT(rc == 0);
+	CU_ASSERT_PTR_NULL(partial_parameter);
+
 	/* Second partial parameter is the only parameter */
 	PARSE("OOOO", true, &partial_parameter);
 	CU_ASSERT_STRING_EQUAL(partial_parameter, "OOOO");

test/unit/lib/nvme/nvme_ctrlr.c/nvme_ctrlr_ut.c

@@ -2162,6 +2162,27 @@ test_nvme_ctrlr_init_set_keep_alive_timeout(void)
 	nvme_ctrlr_destruct(&ctrlr);
 }
 
+static void
+test_alloc_io_qpair_fail(void)
+{
+	struct spdk_nvme_ctrlr ctrlr = {};
+	struct spdk_nvme_qpair *q0;
+
+	setup_qpairs(&ctrlr, 1);
+
+	/* Modify the connect_qpair return code to inject a failure */
+	g_connect_qpair_return_code = 1;
+
+	/* Attempt to allocate a qpair, this should fail */
+	q0 = spdk_nvme_ctrlr_alloc_io_qpair(&ctrlr, NULL, 0);
+	SPDK_CU_ASSERT_FATAL(q0 == NULL);
+
+	/* Verify that the qpair is removed from the lists */
+	SPDK_CU_ASSERT_FATAL(TAILQ_EMPTY(&ctrlr.active_io_qpairs));
+
+	cleanup_qpairs(&ctrlr);
+}
+
 int main(int argc, char **argv)
 {
 	CU_pSuite	suite = NULL;
@@ -2200,6 +2221,7 @@ int main(int argc, char **argv)
 	CU_ADD_TEST(suite, test_nvme_ctrlr_init_set_nvmf_ioccsz);
 	CU_ADD_TEST(suite, test_nvme_ctrlr_init_set_num_queues);
 	CU_ADD_TEST(suite, test_nvme_ctrlr_init_set_keep_alive_timeout);
+	CU_ADD_TEST(suite, test_alloc_io_qpair_fail);
 
 	CU_basic_set_mode(CU_BRM_VERBOSE);
 	CU_basic_run_tests();

test/unit/lib/vhost/vhost.c/vhost_ut.c

@@ -267,28 +267,56 @@ create_controller_test(void)
 	int ret;
 	char long_name[PATH_MAX];
 
-	spdk_cpuset_set_cpu(&g_vhost_core_mask, 0, true);
+	spdk_cpuset_parse(&g_vhost_core_mask, "0xf");
+
+	/* Create device with cpumask implcitly matching whole application */
+	ret = alloc_vdev(&vdev, "vdev_name_0", NULL);
+	SPDK_CU_ASSERT_FATAL(ret == 0 && vdev);
+	SPDK_CU_ASSERT_FATAL(!strcmp(spdk_cpuset_fmt(spdk_thread_get_cpumask(vdev->thread)), "f"));
+	cleanup_vdev(vdev);
+
+	/* Create device with cpumask matching whole application */
+	ret = alloc_vdev(&vdev, "vdev_name_0", "0xf");
+	SPDK_CU_ASSERT_FATAL(ret == 0 && vdev);
+	SPDK_CU_ASSERT_FATAL(!strcmp(spdk_cpuset_fmt(spdk_thread_get_cpumask(vdev->thread)), "f"));
+	cleanup_vdev(vdev);
+
+	/* Create device with single core in cpumask */
+	ret = alloc_vdev(&vdev, "vdev_name_0", "0x2");
+	SPDK_CU_ASSERT_FATAL(ret == 0 && vdev);
+	SPDK_CU_ASSERT_FATAL(!strcmp(spdk_cpuset_fmt(spdk_thread_get_cpumask(vdev->thread)), "2"));
+	cleanup_vdev(vdev);
+
+	/* Create device with cpumask spanning two cores */
+	ret = alloc_vdev(&vdev, "vdev_name_0", "0x3");
+	SPDK_CU_ASSERT_FATAL(ret == 0 && vdev);
+	SPDK_CU_ASSERT_FATAL(!strcmp(spdk_cpuset_fmt(spdk_thread_get_cpumask(vdev->thread)), "3"));
+	cleanup_vdev(vdev);
+
+	/* Create device with incorrect cpumask outside of application cpumask */
+	ret = alloc_vdev(&vdev, "vdev_name_0", "0xf0");
+	SPDK_CU_ASSERT_FATAL(ret != 0);
+
+	/* Create device with incorrect cpumask partially outside of application cpumask */
+	ret = alloc_vdev(&vdev, "vdev_name_0", "0xff");
+	SPDK_CU_ASSERT_FATAL(ret != 0);
 
 	/* Create device with no name */
-	ret = alloc_vdev(&vdev, NULL, "0x1");
+	ret = alloc_vdev(&vdev, NULL, NULL);
-	CU_ASSERT(ret != 0);
-
-	/* Create device with incorrect cpumask */
-	ret = alloc_vdev(&vdev, "vdev_name_0", "0x2");
 	CU_ASSERT(ret != 0);
 
 	/* Create device with too long name and path */
 	memset(long_name, 'x', sizeof(long_name));
 	long_name[PATH_MAX - 1] = 0;
 	snprintf(dev_dirname, sizeof(dev_dirname), "some_path/");
-	ret = alloc_vdev(&vdev, long_name, "0x1");
+	ret = alloc_vdev(&vdev, long_name, NULL);
 	CU_ASSERT(ret != 0);
 	dev_dirname[0] = 0;
 
 	/* Create device when device name is already taken */
-	ret = alloc_vdev(&vdev, "vdev_name_0", "0x1");
+	ret = alloc_vdev(&vdev, "vdev_name_0", NULL);
 	SPDK_CU_ASSERT_FATAL(ret == 0 && vdev);
-	ret = alloc_vdev(&vdev2, "vdev_name_0", "0x1");
+	ret = alloc_vdev(&vdev2, "vdev_name_0", NULL);
 	CU_ASSERT(ret != 0);
 	cleanup_vdev(vdev);
 }

test/vhost/initiator/blockdev.sh

@@ -19,7 +19,7 @@ function err_cleanup() {
 
 # start vhost and configure it
 trap 'err_cleanup; exit 1' SIGINT SIGTERM EXIT
-$SPDK_BIN_DIR/vhost &
+$SPDK_BIN_DIR/vhost -m 0xf &
 vhost_pid=$!
 waitforlisten $vhost_pid
 
@@ -35,17 +35,22 @@ rpc_cmd vhost_scsi_controller_add_target naa.Nvme0n1_scsi0.0 0 Nvme0n1p0
 rpc_cmd vhost_scsi_controller_add_target naa.Nvme0n1_scsi0.0 1 Nvme0n1p1
 rpc_cmd vhost_scsi_controller_add_target naa.Nvme0n1_scsi0.0 2 Nvme0n1p2
 rpc_cmd vhost_scsi_controller_add_target naa.Nvme0n1_scsi0.0 3 Nvme0n1p3
+[[ "$(rpc_cmd vhost_get_controllers -n naa.Nvme0n1_scsi0.0 | jq -r '.[].cpumask')" == "0xf" ]]
 
-rpc_cmd vhost_create_blk_controller naa.Nvme0n1_blk0.0 Nvme0n1p4
+rpc_cmd vhost_create_blk_controller naa.Nvme0n1_blk0.0 Nvme0n1p4 --cpumask 0xf
-rpc_cmd vhost_create_blk_controller naa.Nvme0n1_blk1.0 Nvme0n1p5
+[[ "$(rpc_cmd vhost_get_controllers -n naa.Nvme0n1_blk0.0 | jq -r '.[].cpumask')" == "0xf" ]]
+rpc_cmd vhost_create_blk_controller naa.Nvme0n1_blk1.0 Nvme0n1p5 --cpumask 0x1
+[[ "$(rpc_cmd vhost_get_controllers -n naa.Nvme0n1_blk1.0 | jq -r '.[].cpumask')" == "0x1" ]]
 
 rpc_cmd bdev_malloc_create 128 512 --name Malloc0
-rpc_cmd vhost_create_scsi_controller naa.Malloc0.0
+rpc_cmd vhost_create_scsi_controller naa.Malloc0.0 --cpumask 0x2
 rpc_cmd vhost_scsi_controller_add_target naa.Malloc0.0 0 Malloc0
+[[ "$(rpc_cmd vhost_get_controllers -n naa.Malloc0.0 | jq -r '.[].cpumask')" == "0x2" ]]
 
 rpc_cmd bdev_malloc_create 128 4096 --name Malloc1
-rpc_cmd vhost_create_scsi_controller naa.Malloc1.0
+rpc_cmd vhost_create_scsi_controller naa.Malloc1.0 --cpumask 0xc
 rpc_cmd vhost_scsi_controller_add_target naa.Malloc1.0 0 Malloc1
+[[ "$(rpc_cmd vhost_get_controllers -n naa.Malloc1.0 | jq -r '.[].cpumask')" == "0xc" ]]
 
 # start a dummy app, create vhost bdevs in it, then dump the config for FIO
 $SPDK_BIN_DIR/spdk_tgt -r /tmp/spdk2.sock -g &

test/vhost/other/negative.sh

@@ -63,7 +63,7 @@ notice "==============="
 notice ""
 notice "running SPDK"
 notice ""
-vhost_run 0
+vhost_run 0 "-m 0xf"
 notice ""
 rpc_py="$rootdir/scripts/rpc.py -s $(get_vhost_dir 0)/rpc.sock"
 $rpc_py bdev_malloc_create -b Malloc0 128 4096
@@ -88,8 +88,13 @@ if $rpc_py vhost_delete_controller unk0 > /dev/null; then
 fi
 
 # SCSI
-notice "Trying to create scsi controller with incorrect cpumask"
+notice "Trying to create scsi controller with incorrect cpumask outside of application cpumask"
-if $rpc_py vhost_create_scsi_controller vhost.invalid.cpumask --cpumask 0x2; then
+if $rpc_py vhost_create_scsi_controller vhost.invalid.cpumask --cpumask 0xf0; then
+	error "Creating scsi controller with incorrect cpumask succeeded, but it shouldn't"
+fi
+
+notice "Trying to create scsi controller with incorrect cpumask partially outside of application cpumask"
+if $rpc_py vhost_create_scsi_controller vhost.invalid.cpumask --cpumask 0xff; then
 	error "Creating scsi controller with incorrect cpumask succeeded, but it shouldn't"
 fi
 
@@ -167,8 +172,13 @@ notice "Re-adding device 0 to naa.0"
 $rpc_py vhost_scsi_controller_add_target naa.0 0 Malloc0
 
 # BLK
-notice "Trying to create block controller with incorrect cpumask"
+notice "Trying to create block controller with incorrect cpumask outside of application cpumask"
-if $rpc_py vhost_create_blk_controller vhost.invalid.cpumask Malloc0 --cpumask 0x2; then
+if $rpc_py vhost_create_blk_controller vhost.invalid.cpumask Malloc0 --cpumask 0xf0; then
+	error "Creating block controller with incorrect cpumask succeeded, but it shouldn't"
+fi
+
+notice "Trying to create block controller with incorrect cpumask partially outside of application cpumask"
+if $rpc_py vhost_create_blk_controller vhost.invalid.cpumask Malloc0 --cpumask 0xff; then
 	error "Creating block controller with incorrect cpumask succeeded, but it shouldn't"
 fi