Merge tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-nextgrafted

Pull networking updates from Jakub Kicinski:
 "Core:

   - Add dedicated kmem_cache for typical/small skb->head, avoid having
     to access struct page at kfree time, and improve memory use.

   - Introduce sysctl to set default RPS configuration for new netdevs.

   - Define Netlink protocol specification format which can be used to
     describe messages used by each family and auto-generate parsers.
     Add tools for generating kernel data structures and uAPI headers.

   - Expose all net/core sysctls inside netns.

   - Remove 4s sleep in netpoll if carrier is instantly detected on
     boot.

   - Add configurable limit of MDB entries per port, and port-vlan.

   - Continue populating drop reasons throughout the stack.

   - Retire a handful of legacy Qdiscs and classifiers.

  Protocols:

   - Support IPv4 big TCP (TSO frames larger than 64kB).

   - Add IP_LOCAL_PORT_RANGE socket option, to control local port range
     on socket by socket basis.

   - Track and report in procfs number of MPTCP sockets used.

   - Support mixing IPv4 and IPv6 flows in the in-kernel MPTCP path
     manager.

   - IPv6: don't check net.ipv6.route.max_size and rely on garbage
     collection to free memory (similarly to IPv4).

   - Support Penultimate Segment Pop (PSP) flavor in SRv6 (RFC8986).

   - ICMP: add per-rate limit counters.

   - Add support for user scanning requests in ieee802154.

   - Remove static WEP support.

   - Support minimal Wi-Fi 7 Extremely High Throughput (EHT) rate
     reporting.

   - WiFi 7 EHT channel puncturing support (client & AP).

  BPF:

   - Add a rbtree data structure following the "next-gen data structure"
     precedent set by recently added linked list, that is, by using
     kfunc + kptr instead of adding a new BPF map type.

   - Expose XDP hints via kfuncs with initial support for RX hash and
     timestamp metadata.

   - Add BPF_F_NO_TUNNEL_KEY extension to bpf_skb_set_tunnel_key to
     better support decap on GRE tunnel devices not operating in collect
     metadata.

   - Improve x86 JIT's codegen for PROBE_MEM runtime error checks.

   - Remove the need for trace_printk_lock for bpf_trace_printk and
     bpf_trace_vprintk helpers.

   - Extend libbpf's bpf_tracing.h support for tracing arguments of
     kprobes/uprobes and syscall as a special case.

   - Significantly reduce the search time for module symbols by
     livepatch and BPF.

   - Enable cpumasks to be used as kptrs, which is useful for tracing
     programs tracking which tasks end up running on which CPUs in
     different time intervals.

   - Add support for BPF trampoline on s390x and riscv64.

   - Add capability to export the XDP features supported by the NIC.

   - Add __bpf_kfunc tag for marking kernel functions as kfuncs.

   - Add cgroup.memory=nobpf kernel parameter option to disable BPF
     memory accounting for container environments.

  Netfilter:

   - Remove the CLUSTERIP target. It has been marked as obsolete for
     years, and we still have WARN splats wrt races of the out-of-band
     /proc interface installed by this target.

   - Add 'destroy' commands to nf_tables. They are identical to the
     existing 'delete' commands, but do not return an error if the
     referenced object (set, chain, rule...) did not exist.

  Driver API:

   - Improve cpumask_local_spread() locality to help NICs set the right
     IRQ affinity on AMD platforms.

   - Separate C22 and C45 MDIO bus transactions more clearly.

   - Introduce new DCB table to control DSCP rewrite on egress.

   - Support configuration of Physical Layer Collision Avoidance (PLCA)
     Reconciliation Sublayer (RS) (802.3cg-2019). Modern version of
     shared medium Ethernet.

   - Support for MAC Merge layer (IEEE 802.3-2018 clause 99). Allowing
     preemption of low priority frames by high priority frames.

   - Add support for controlling MACSec offload using netlink SET.

   - Rework devlink instance refcounts to allow registration and
     de-registration under the instance lock. Split the code into
     multiple files, drop some of the unnecessarily granular locks and
     factor out common parts of netlink operation handling.

   - Add TX frame aggregation parameters (for USB drivers).

   - Add a new attr TCA_EXT_WARN_MSG to report TC (offload) warning
     messages with notifications for debug.

   - Allow offloading of UDP NEW connections via act_ct.

   - Add support for per action HW stats in TC.

   - Support hardware miss to TC action (continue processing in SW from
     a specific point in the action chain).

   - Warn if old Wireless Extension user space interface is used with
     modern cfg80211/mac80211 drivers. Do not support Wireless
     Extensions for Wi-Fi 7 devices at all. Everyone should switch to
     using nl80211 interface instead.

   - Improve the CAN bit timing configuration. Use extack to return
     error messages directly to user space, update the SJW handling,
     including the definition of a new default value that will benefit
     CAN-FD controllers, by increasing their oscillator tolerance.

  New hardware / drivers:

   - Ethernet:
      - nVidia BlueField-3 support (control traffic driver)
      - Ethernet support for imx93 SoCs
      - Motorcomm yt8531 gigabit Ethernet PHY
      - onsemi NCN26000 10BASE-T1S PHY (with support for PLCA)
      - Microchip LAN8841 PHY (incl. cable diagnostics and PTP)
      - Amlogic gxl MDIO mux

   - WiFi:
      - RealTek RTL8188EU (rtl8xxxu)
      - Qualcomm Wi-Fi 7 devices (ath12k)

   - CAN:
      - Renesas R-Car V4H

  Drivers:

   - Bluetooth:
      - Set Per Platform Antenna Gain (PPAG) for Intel controllers.

   - Ethernet NICs:
      - Intel (1G, igc):
         - support TSN / Qbv / packet scheduling features of i226 model
      - Intel (100G, ice):
         - use GNSS subsystem instead of TTY
         - multi-buffer XDP support
         - extend support for GPIO pins to E823 devices
      - nVidia/Mellanox:
         - update the shared buffer configuration on PFC commands
         - implement PTP adjphase function for HW offset control
         - TC support for Geneve and GRE with VF tunnel offload
         - more efficient crypto key management method
         - multi-port eswitch support
      - Netronome/Corigine:
         - add DCB IEEE support
         - support IPsec offloading for NFP3800
      - Freescale/NXP (enetc):
         - support XDP_REDIRECT for XDP non-linear buffers
         - improve reconfig, avoid link flap and waiting for idle
         - support MAC Merge layer
      - Other NICs:
         - sfc/ef100: add basic devlink support for ef100
         - ionic: rx_push mode operation (writing descriptors via MMIO)
         - bnxt: use the auxiliary bus abstraction for RDMA
         - r8169: disable ASPM and reset bus in case of tx timeout
         - cpsw: support QSGMII mode for J721e CPSW9G
         - cpts: support pulse-per-second output
         - ngbe: add an mdio bus driver
         - usbnet: optimize usbnet_bh() by avoiding unnecessary queuing
         - r8152: handle devices with FW with NCM support
         - amd-xgbe: support 10Mbps, 2.5GbE speeds and rx-adaptation
         - virtio-net: support multi buffer XDP
         - virtio/vsock: replace virtio_vsock_pkt with sk_buff
         - tsnep: XDP support

   - Ethernet high-speed switches:
      - nVidia/Mellanox (mlxsw):
         - add support for latency TLV (in FW control messages)
      - Microchip (sparx5):
         - separate explicit and implicit traffic forwarding rules, make
           the implicit rules always active
         - add support for egress DSCP rewrite
         - IS0 VCAP support (Ingress Classification)
         - IS2 VCAP filters (protos, L3 addrs, L4 ports, flags, ToS
           etc.)
         - ES2 VCAP support (Egress Access Control)
         - support for Per-Stream Filtering and Policing (802.1Q,
           8.6.5.1)

   - Ethernet embedded switches:
      - Marvell (mv88e6xxx):
         - add MAB (port auth) offload support
         - enable PTP receive for mv88e6390
      - NXP (ocelot):
         - support MAC Merge layer
         - support for the the vsc7512 internal copper phys
      - Microchip:
         - lan9303: convert to PHYLINK
         - lan966x: support TC flower filter statistics
         - lan937x: PTP support for KSZ9563/KSZ8563 and LAN937x
         - lan937x: support Credit Based Shaper configuration
         - ksz9477: support Energy Efficient Ethernet
      - other:
         - qca8k: convert to regmap read/write API, use bulk operations
         - rswitch: Improve TX timestamp accuracy

   - Intel WiFi (iwlwifi):
      - EHT (Wi-Fi 7) rate reporting
      - STEP equalizer support: transfer some STEP (connection to radio
        on platforms with integrated wifi) related parameters from the
        BIOS to the firmware.

   - Qualcomm 802.11ax WiFi (ath11k):
      - IPQ5018 support
      - Fine Timing Measurement (FTM) responder role support
      - channel 177 support

   - MediaTek WiFi (mt76):
      - per-PHY LED support
      - mt7996: EHT (Wi-Fi 7) support
      - Wireless Ethernet Dispatch (WED) reset support
      - switch to using page pool allocator

   - RealTek WiFi (rtw89):
      - support new version of Bluetooth co-existance

   - Mobile:
      - rmnet: support TX aggregation"

* tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next: (1872 commits)
  page_pool: add a comment explaining the fragment counter usage
  net: ethtool: fix __ethtool_dev_mm_supported() implementation
  ethtool: pse-pd: Fix double word in comments
  xsk: add linux/vmalloc.h to xsk.c
  sefltests: netdevsim: wait for devlink instance after netns removal
  selftest: fib_tests: Always cleanup before exit
  net/mlx5e: Align IPsec ASO result memory to be as required by hardware
  net/mlx5e: TC, Set CT miss to the specific ct action instance
  net/mlx5e: Rename CHAIN_TO_REG to MAPPED_OBJ_TO_REG
  net/mlx5: Refactor tc miss handling to a single function
  net/mlx5: Kconfig: Make tc offload depend on tc skb extension
  net/sched: flower: Support hardware miss to tc action
  net/sched: flower: Move filter handle initialization earlier
  net/sched: cls_api: Support hardware miss to tc action
  net/sched: Rename user cookie and act cookie
  sfc: fix builds without CONFIG_RTC_LIB
  sfc: clean up some inconsistent indentings
  net/mlx4_en: Introduce flexible array to silence overflow warning
  net: lan966x: Fix possible deadlock inside PTP
  net/ulp: Remove redundant ->clone() test in inet_clone_ulp().
  ...

1 files changed, 556 insertions, 0 deletions

diff --git a/tools/perf/bench/epoll-wait.c b/tools/perf/bench/epoll-wait.c
new file mode 100644
index 000000000..c1cdf03c0
--- /dev/null
+++ b/tools/perf/bench/epoll-wait.c
@@ -0,0 +1,556 @@
+// SPDX-License-Identifier: GPL-2.0
+#ifdef HAVE_EVENTFD_SUPPORT
+/*
+ * Copyright (C) 2018 Davidlohr Bueso.
+ *
+ * This program benchmarks concurrent epoll_wait(2) monitoring multiple
+ * file descriptors under one or two load balancing models. The first,
+ * and default, is the single/combined queueing (which refers to a single
+ * epoll instance for N worker threads):
+ *
+ *                          |---> [worker A]
+ *                          |---> [worker B]
+ *        [combined queue]  .---> [worker C]
+ *                          |---> [worker D]
+ *                          |---> [worker E]
+ *
+ * While the second model, enabled via --multiq option, uses multiple
+ * queueing (which refers to one epoll instance per worker). For example,
+ * short lived tcp connections in a high throughput httpd server will
+ * distribute the accept()'ing  connections across CPUs. In this case each
+ * worker does a limited  amount of processing.
+ *
+ *             [queue A]  ---> [worker]
+ *             [queue B]  ---> [worker]
+ *             [queue C]  ---> [worker]
+ *             [queue D]  ---> [worker]
+ *             [queue E]  ---> [worker]
+ *
+ * Naturally, the single queue will enforce more concurrency on the epoll
+ * instance, and can therefore scale poorly compared to multiple queues.
+ * However, this is a benchmark raw data and must be taken with a grain of
+ * salt when choosing how to make use of sys_epoll.
+
+ * Each thread has a number of private, nonblocking file descriptors,
+ * referred to as fdmap. A writer thread will constantly be writing to
+ * the fdmaps of all threads, minimizing each threads's chances of
+ * epoll_wait not finding any ready read events and blocking as this
+ * is not what we want to stress. The size of the fdmap can be adjusted
+ * by the user; enlarging the value will increase the chances of
+ * epoll_wait(2) blocking as the lineal writer thread will take "longer",
+ * at least at a high level.
+ *
+ * Note that because fds are private to each thread, this workload does
+ * not stress scenarios where multiple tasks are awoken per ready IO; ie:
+ * EPOLLEXCLUSIVE semantics.
+ *
+ * The end result/metric is throughput: number of ops/second where an
+ * operation consists of:
+ *
+ *   epoll_wait(2) + [others]
+ *
+ *        ... where [others] is the cost of re-adding the fd (EPOLLET),
+ *            or rearming it (EPOLLONESHOT).
+ *
+ *
+ * The purpose of this is program is that it be useful for measuring
+ * kernel related changes to the sys_epoll, and not comparing different
+ * IO polling methods, for example. Hence everything is very adhoc and
+ * outputs raw microbenchmark numbers. Also this uses eventfd, similar
+ * tools tend to use pipes or sockets, but the result is the same.
+ */
+
+/* For the CLR_() macros */
+#include <string.h>
+#include <pthread.h>
+#include <unistd.h>
+
+#include <errno.h>
+#include <inttypes.h>
+#include <signal.h>
+#include <stdlib.h>
+#include <linux/compiler.h>
+#include <linux/kernel.h>
+#include <sys/time.h>
+#include <sys/resource.h>
+#include <sys/epoll.h>
+#include <sys/eventfd.h>
+#include <sys/types.h>
+#include <perf/cpumap.h>
+
+#include "../util/stat.h"
+#include "../util/mutex.h"
+#include <subcmd/parse-options.h>
+#include "bench.h"
+
+#include <err.h>
+
+#define printinfo(fmt, arg...) \
+	do { if (__verbose) { printf(fmt, ## arg); fflush(stdout); } } while (0)
+
+static unsigned int nthreads = 0;
+static unsigned int nsecs    = 8;
+static bool wdone, done, __verbose, randomize, nonblocking;
+
+/*
+ * epoll related shared variables.
+ */
+
+/* Maximum number of nesting allowed inside epoll sets */
+#define EPOLL_MAXNESTS 4
+
+static int epollfd;
+static int *epollfdp;
+static bool noaffinity;
+static unsigned int nested = 0;
+static bool et; /* edge-trigger */
+static bool oneshot;
+static bool multiq; /* use an epoll instance per thread */
+
+/* amount of fds to monitor, per thread */
+static unsigned int nfds = 64;
+
+static struct mutex thread_lock;
+static unsigned int threads_starting;
+static struct stats throughput_stats;
+static struct cond thread_parent, thread_worker;
+
+struct worker {
+	int tid;
+	int epollfd; /* for --multiq */
+	pthread_t thread;
+	unsigned long ops;
+	int *fdmap;
+};
+
+static const struct option options[] = {
+	/* general benchmark options */
+	OPT_UINTEGER('t', "threads", &nthreads, "Specify amount of threads"),
+	OPT_UINTEGER('r', "runtime", &nsecs, "Specify runtime (in seconds)"),
+	OPT_UINTEGER('f', "nfds",    &nfds,  "Specify amount of file descriptors to monitor for each thread"),
+	OPT_BOOLEAN( 'n', "noaffinity",  &noaffinity,   "Disables CPU affinity"),
+	OPT_BOOLEAN('R', "randomize", &randomize,   "Enable random write behaviour (default is lineal)"),
+	OPT_BOOLEAN( 'v', "verbose", &__verbose, "Verbose mode"),
+
+	/* epoll specific options */
+	OPT_BOOLEAN( 'm', "multiq",  &multiq,   "Use multiple epoll instances (one per thread)"),
+	OPT_BOOLEAN( 'B', "nonblocking", &nonblocking, "Nonblocking epoll_wait(2) behaviour"),
+	OPT_UINTEGER( 'N', "nested",  &nested,   "Nesting level epoll hierarchy (default is 0, no nesting)"),
+	OPT_BOOLEAN( 'S', "oneshot",  &oneshot,   "Use EPOLLONESHOT semantics"),
+	OPT_BOOLEAN( 'E', "edge",  &et,   "Use Edge-triggered interface (default is LT)"),
+
+	OPT_END()
+};
+
+static const char * const bench_epoll_wait_usage[] = {
+	"perf bench epoll wait <options>",
+	NULL
+};
+
+
+/*
+ * Arrange the N elements of ARRAY in random order.
+ * Only effective if N is much smaller than RAND_MAX;
+ * if this may not be the case, use a better random
+ * number generator. -- Ben Pfaff.
+ */
+static void shuffle(void *array, size_t n, size_t size)
+{
+	char *carray = array;
+	void *aux;
+	size_t i;
+
+	if (n <= 1)
+		return;
+
+	aux = calloc(1, size);
+	if (!aux)
+		err(EXIT_FAILURE, "calloc");
+
+	for (i = 1; i < n; ++i) {
+		size_t j =   i + rand() / (RAND_MAX / (n - i) + 1);
+		j *= size;
+
+		memcpy(aux, &carray[j], size);
+		memcpy(&carray[j], &carray[i*size], size);
+		memcpy(&carray[i*size], aux, size);
+	}
+
+	free(aux);
+}
+
+
+static void *workerfn(void *arg)
+{
+	int fd, ret, r;
+	struct worker *w = (struct worker *) arg;
+	unsigned long ops = w->ops;
+	struct epoll_event ev;
+	uint64_t val;
+	int to = nonblocking? 0 : -1;
+	int efd = multiq ? w->epollfd : epollfd;
+
+	mutex_lock(&thread_lock);
+	threads_starting--;
+	if (!threads_starting)
+		cond_signal(&thread_parent);
+	cond_wait(&thread_worker, &thread_lock);
+	mutex_unlock(&thread_lock);
+
+	do {
+		/*
+		 * Block indefinitely waiting for the IN event.
+		 * In order to stress the epoll_wait(2) syscall,
+		 * call it event per event, instead of a larger
+		 * batch (max)limit.
+		 */
+		do {
+			ret = epoll_wait(efd, &ev, 1, to);
+		} while (ret < 0 && errno == EINTR);
+		if (ret < 0)
+			err(EXIT_FAILURE, "epoll_wait");
+
+		fd = ev.data.fd;
+
+		do {
+			r = read(fd, &val, sizeof(val));
+		} while (!done && (r < 0 && errno == EAGAIN));
+
+		if (et) {
+			ev.events = EPOLLIN | EPOLLET;
+			ret = epoll_ctl(efd, EPOLL_CTL_ADD, fd, &ev);
+		}
+
+		if (oneshot) {
+			/* rearm the file descriptor with a new event mask */
+			ev.events |= EPOLLIN | EPOLLONESHOT;
+			ret = epoll_ctl(efd, EPOLL_CTL_MOD, fd, &ev);
+		}
+
+		ops++;
+	}  while (!done);
+
+	if (multiq)
+		close(w->epollfd);
+
+	w->ops = ops;
+	return NULL;
+}
+
+static void nest_epollfd(struct worker *w)
+{
+	unsigned int i;
+	struct epoll_event ev;
+	int efd = multiq ? w->epollfd : epollfd;
+
+	if (nested > EPOLL_MAXNESTS)
+		nested = EPOLL_MAXNESTS;
+
+	epollfdp = calloc(nested, sizeof(*epollfdp));
+	if (!epollfdp)
+		err(EXIT_FAILURE, "calloc");
+
+	for (i = 0; i < nested; i++) {
+		epollfdp[i] = epoll_create(1);
+		if (epollfdp[i] < 0)
+			err(EXIT_FAILURE, "epoll_create");
+	}
+
+	ev.events = EPOLLHUP; /* anything */
+	ev.data.u64 = i; /* any number */
+
+	for (i = nested - 1; i; i--) {
+		if (epoll_ctl(epollfdp[i - 1], EPOLL_CTL_ADD,
+			      epollfdp[i], &ev) < 0)
+			err(EXIT_FAILURE, "epoll_ctl");
+	}
+
+	if (epoll_ctl(efd, EPOLL_CTL_ADD, *epollfdp, &ev) < 0)
+		err(EXIT_FAILURE, "epoll_ctl");
+}
+
+static void toggle_done(int sig __maybe_unused,
+			siginfo_t *info __maybe_unused,
+			void *uc __maybe_unused)
+{
+	/* inform all threads that we're done for the day */
+	done = true;
+	gettimeofday(&bench__end, NULL);
+	timersub(&bench__end, &bench__start, &bench__runtime);
+}
+
+static void print_summary(void)
+{
+	unsigned long avg = avg_stats(&throughput_stats);
+	double stddev = stddev_stats(&throughput_stats);
+
+	printf("\nAveraged %ld operations/sec (+- %.2f%%), total secs = %d\n",
+	       avg, rel_stddev_stats(stddev, avg),
+	       (int)bench__runtime.tv_sec);
+}
+
+static int do_threads(struct worker *worker, struct perf_cpu_map *cpu)
+{
+	pthread_attr_t thread_attr, *attrp = NULL;
+	cpu_set_t *cpuset;
+	unsigned int i, j;
+	int ret = 0, events = EPOLLIN;
+	int nrcpus;
+	size_t size;
+
+	if (oneshot)
+		events |= EPOLLONESHOT;
+	if (et)
+		events |= EPOLLET;
+
+	printinfo("starting worker/consumer %sthreads%s\n",
+		  noaffinity ?  "":"CPU affinity ",
+		  nonblocking ? " (nonblocking)":"");
+	if (!noaffinity)
+		pthread_attr_init(&thread_attr);
+
+	nrcpus = perf_cpu_map__nr(cpu);
+	cpuset = CPU_ALLOC(nrcpus);
+	BUG_ON(!cpuset);
+	size = CPU_ALLOC_SIZE(nrcpus);
+
+	for (i = 0; i < nthreads; i++) {
+		struct worker *w = &worker[i];
+
+		if (multiq) {
+			w->epollfd = epoll_create(1);
+			if (w->epollfd < 0)
+				err(EXIT_FAILURE, "epoll_create");
+
+			if (nested)
+				nest_epollfd(w);
+		}
+
+		w->tid = i;
+		w->fdmap = calloc(nfds, sizeof(int));
+		if (!w->fdmap)
+			return 1;
+
+		for (j = 0; j < nfds; j++) {
+			int efd = multiq ? w->epollfd : epollfd;
+			struct epoll_event ev;
+
+			w->fdmap[j] = eventfd(0, EFD_NONBLOCK);
+			if (w->fdmap[j] < 0)
+				err(EXIT_FAILURE, "eventfd");
+
+			ev.data.fd = w->fdmap[j];
+			ev.events = events;
+
+			ret = epoll_ctl(efd, EPOLL_CTL_ADD,
+					w->fdmap[j], &ev);
+			if (ret < 0)
+				err(EXIT_FAILURE, "epoll_ctl");
+		}
+
+		if (!noaffinity) {
+			CPU_ZERO_S(size, cpuset);
+			CPU_SET_S(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu,
+					size, cpuset);
+
+			ret = pthread_attr_setaffinity_np(&thread_attr, size, cpuset);
+			if (ret) {
+				CPU_FREE(cpuset);
+				err(EXIT_FAILURE, "pthread_attr_setaffinity_np");
+			}
+
+			attrp = &thread_attr;
+		}
+
+		ret = pthread_create(&w->thread, attrp, workerfn,
+				     (void *)(struct worker *) w);
+		if (ret) {
+			CPU_FREE(cpuset);
+			err(EXIT_FAILURE, "pthread_create");
+		}
+	}
+
+	CPU_FREE(cpuset);
+	if (!noaffinity)
+		pthread_attr_destroy(&thread_attr);
+
+	return ret;
+}
+
+static void *writerfn(void *p)
+{
+	struct worker *worker = p;
+	size_t i, j, iter;
+	const uint64_t val = 1;
+	ssize_t sz;
+	struct timespec ts = { .tv_sec = 0,
+			       .tv_nsec = 500 };
+
+	printinfo("starting writer-thread: doing %s writes ...\n",
+		  randomize? "random":"lineal");
+
+	for (iter = 0; !wdone; iter++) {
+		if (randomize) {
+			shuffle((void *)worker, nthreads, sizeof(*worker));
+		}
+
+		for (i = 0; i < nthreads; i++) {
+			struct worker *w = &worker[i];
+
+			if (randomize) {
+				shuffle((void *)w->fdmap, nfds, sizeof(int));
+			}
+
+			for (j = 0; j < nfds; j++) {
+				do {
+					sz = write(w->fdmap[j], &val, sizeof(val));
+				} while (!wdone && (sz < 0 && errno == EAGAIN));
+			}
+		}
+
+		nanosleep(&ts, NULL);
+	}
+
+	printinfo("exiting writer-thread (total full-loops: %zd)\n", iter);
+	return NULL;
+}
+
+static int cmpworker(const void *p1, const void *p2)
+{
+
+	struct worker *w1 = (struct worker *) p1;
+	struct worker *w2 = (struct worker *) p2;
+	return w1->tid > w2->tid;
+}
+
+int bench_epoll_wait(int argc, const char **argv)
+{
+	int ret = 0;
+	struct sigaction act;
+	unsigned int i;
+	struct worker *worker = NULL;
+	struct perf_cpu_map *cpu;
+	pthread_t wthread;
+	struct rlimit rl, prevrl;
+
+	argc = parse_options(argc, argv, options, bench_epoll_wait_usage, 0);
+	if (argc) {
+		usage_with_options(bench_epoll_wait_usage, options);
+		exit(EXIT_FAILURE);
+	}
+
+	memset(&act, 0, sizeof(act));
+	sigfillset(&act.sa_mask);
+	act.sa_sigaction = toggle_done;
+	sigaction(SIGINT, &act, NULL);
+
+	cpu = perf_cpu_map__new(NULL);
+	if (!cpu)
+		goto errmem;
+
+	/* a single, main epoll instance */
+	if (!multiq) {
+		epollfd = epoll_create(1);
+		if (epollfd < 0)
+			err(EXIT_FAILURE, "epoll_create");
+
+		/*
+		 * Deal with nested epolls, if any.
+		 */
+		if (nested)
+			nest_epollfd(NULL);
+	}
+
+	printinfo("Using %s queue model\n", multiq ? "multi" : "single");
+	printinfo("Nesting level(s): %d\n", nested);
+
+	/* default to the number of CPUs and leave one for the writer pthread */
+	if (!nthreads)
+		nthreads = perf_cpu_map__nr(cpu) - 1;
+
+	worker = calloc(nthreads, sizeof(*worker));
+	if (!worker) {
+		goto errmem;
+	}
+
+	if (getrlimit(RLIMIT_NOFILE, &prevrl))
+		err(EXIT_FAILURE, "getrlimit");
+	rl.rlim_cur = rl.rlim_max = nfds * nthreads * 2 + 50;
+	printinfo("Setting RLIMIT_NOFILE rlimit from %" PRIu64 " to: %" PRIu64 "\n",
+		  (uint64_t)prevrl.rlim_max, (uint64_t)rl.rlim_max);
+	if (setrlimit(RLIMIT_NOFILE, &rl) < 0)
+		err(EXIT_FAILURE, "setrlimit");
+
+	printf("Run summary [PID %d]: %d threads monitoring%s on "
+	       "%d file-descriptors for %d secs.\n\n",
+	       getpid(), nthreads, oneshot ? " (EPOLLONESHOT semantics)": "", nfds, nsecs);
+
+	init_stats(&throughput_stats);
+	mutex_init(&thread_lock);
+	cond_init(&thread_parent);
+	cond_init(&thread_worker);
+
+	threads_starting = nthreads;
+
+	gettimeofday(&bench__start, NULL);
+
+	do_threads(worker, cpu);
+
+	mutex_lock(&thread_lock);
+	while (threads_starting)
+		cond_wait(&thread_parent, &thread_lock);
+	cond_broadcast(&thread_worker);
+	mutex_unlock(&thread_lock);
+
+	/*
+	 * At this point the workers should be blocked waiting for read events
+	 * to become ready. Launch the writer which will constantly be writing
+	 * to each thread's fdmap.
+	 */
+	ret = pthread_create(&wthread, NULL, writerfn,
+			     (void *)(struct worker *) worker);
+	if (ret)
+		err(EXIT_FAILURE, "pthread_create");
+
+	sleep(nsecs);
+	toggle_done(0, NULL, NULL);
+	printinfo("main thread: toggling done\n");
+
+	sleep(1); /* meh */
+	wdone = true;
+	ret = pthread_join(wthread, NULL);
+	if (ret)
+		err(EXIT_FAILURE, "pthread_join");
+
+	/* cleanup & report results */
+	cond_destroy(&thread_parent);
+	cond_destroy(&thread_worker);
+	mutex_destroy(&thread_lock);
+
+	/* sort the array back before reporting */
+	if (randomize)
+		qsort(worker, nthreads, sizeof(struct worker), cmpworker);
+
+	for (i = 0; i < nthreads; i++) {
+		unsigned long t = bench__runtime.tv_sec > 0 ?
+			worker[i].ops / bench__runtime.tv_sec : 0;
+
+		update_stats(&throughput_stats, t);
+
+		if (nfds == 1)
+			printf("[thread %2d] fdmap: %p [ %04ld ops/sec ]\n",
+			       worker[i].tid, &worker[i].fdmap[0], t);
+		else
+			printf("[thread %2d] fdmap: %p ... %p [ %04ld ops/sec ]\n",
+			       worker[i].tid, &worker[i].fdmap[0],
+			       &worker[i].fdmap[nfds-1], t);
+	}
+
+	print_summary();
+
+	close(epollfd);
+	return ret;
+errmem:
+	err(EXIT_FAILURE, "calloc");
+}
+#endif // HAVE_EVENTFD_SUPPORT