Update Linux to v5.10.109
Sourced from [1]
[1] https://cdn.kernel.org/pub/linux/kernel/v5.x/linux-5.10.109.tar.xz
Change-Id: I19bca9fc6762d4e63bcf3e4cba88bbe560d9c76c
Signed-off-by: Olivier Deprez <olivier.deprez@arm.com>
diff --git a/samples/bpf/xdpsock_user.c b/samples/bpf/xdpsock_user.c
index 3b604c1..2e4508a 100644
--- a/samples/bpf/xdpsock_user.c
+++ b/samples/bpf/xdpsock_user.c
@@ -10,6 +10,10 @@
#include <linux/if_link.h>
#include <linux/if_xdp.h>
#include <linux/if_ether.h>
+#include <linux/ip.h>
+#include <linux/limits.h>
+#include <linux/udp.h>
+#include <arpa/inet.h>
#include <locale.h>
#include <net/ethernet.h>
#include <net/if.h>
@@ -27,9 +31,10 @@
#include <time.h>
#include <unistd.h>
-#include "libbpf.h"
-#include "xsk.h"
+#include <bpf/libbpf.h>
+#include <bpf/xsk.h>
#include <bpf/bpf.h>
+#include "xdpsock.h"
#ifndef SOL_XDP
#define SOL_XDP 283
@@ -44,13 +49,14 @@
#endif
#define NUM_FRAMES (4 * 1024)
-#define BATCH_SIZE 64
+#define MIN_PKT_SIZE 64
#define DEBUG_HEXDUMP 0
-#define MAX_SOCKS 8
typedef __u64 u64;
typedef __u32 u32;
+typedef __u16 u16;
+typedef __u8 u8;
static unsigned long prev_time;
@@ -65,17 +71,67 @@
static const char *opt_if = "";
static int opt_ifindex;
static int opt_queue;
+static unsigned long opt_duration;
+static unsigned long start_time;
+static bool benchmark_done;
+static u32 opt_batch_size = 64;
+static int opt_pkt_count;
+static u16 opt_pkt_size = MIN_PKT_SIZE;
+static u32 opt_pkt_fill_pattern = 0x12345678;
+static bool opt_extra_stats;
+static bool opt_quiet;
+static bool opt_app_stats;
+static const char *opt_irq_str = "";
+static u32 irq_no;
+static int irqs_at_init = -1;
static int opt_poll;
static int opt_interval = 1;
static u32 opt_xdp_bind_flags = XDP_USE_NEED_WAKEUP;
static u32 opt_umem_flags;
static int opt_unaligned_chunks;
static int opt_mmap_flags;
-static u32 opt_xdp_bind_flags;
static int opt_xsk_frame_size = XSK_UMEM__DEFAULT_FRAME_SIZE;
static int opt_timeout = 1000;
static bool opt_need_wakeup = true;
-static __u32 prog_id;
+static u32 opt_num_xsks = 1;
+static u32 prog_id;
+
+struct xsk_ring_stats {
+ unsigned long rx_npkts;
+ unsigned long tx_npkts;
+ unsigned long rx_dropped_npkts;
+ unsigned long rx_invalid_npkts;
+ unsigned long tx_invalid_npkts;
+ unsigned long rx_full_npkts;
+ unsigned long rx_fill_empty_npkts;
+ unsigned long tx_empty_npkts;
+ unsigned long prev_rx_npkts;
+ unsigned long prev_tx_npkts;
+ unsigned long prev_rx_dropped_npkts;
+ unsigned long prev_rx_invalid_npkts;
+ unsigned long prev_tx_invalid_npkts;
+ unsigned long prev_rx_full_npkts;
+ unsigned long prev_rx_fill_empty_npkts;
+ unsigned long prev_tx_empty_npkts;
+};
+
+struct xsk_driver_stats {
+ unsigned long intrs;
+ unsigned long prev_intrs;
+};
+
+struct xsk_app_stats {
+ unsigned long rx_empty_polls;
+ unsigned long fill_fail_polls;
+ unsigned long copy_tx_sendtos;
+ unsigned long tx_wakeup_sendtos;
+ unsigned long opt_polls;
+ unsigned long prev_rx_empty_polls;
+ unsigned long prev_fill_fail_polls;
+ unsigned long prev_copy_tx_sendtos;
+ unsigned long prev_tx_wakeup_sendtos;
+ unsigned long prev_opt_polls;
+};
struct xsk_umem_info {
struct xsk_ring_prod fq;
@@ -89,10 +145,9 @@
struct xsk_ring_prod tx;
struct xsk_umem_info *umem;
struct xsk_socket *xsk;
- unsigned long rx_npkts;
- unsigned long tx_npkts;
- unsigned long prev_rx_npkts;
- unsigned long prev_tx_npkts;
+ struct xsk_ring_stats ring_stats;
+ struct xsk_app_stats app_stats;
+ struct xsk_driver_stats drv_stats;
u32 outstanding_tx;
};
@@ -135,6 +190,163 @@
}
}
+static int xsk_get_xdp_stats(int fd, struct xsk_socket_info *xsk)
+{
+ struct xdp_statistics stats;
+ socklen_t optlen;
+ int err;
+
+ optlen = sizeof(stats);
+ err = getsockopt(fd, SOL_XDP, XDP_STATISTICS, &stats, &optlen);
+ if (err)
+ return err;
+
+ if (optlen == sizeof(struct xdp_statistics)) {
+ xsk->ring_stats.rx_dropped_npkts = stats.rx_dropped;
+ xsk->ring_stats.rx_invalid_npkts = stats.rx_invalid_descs;
+ xsk->ring_stats.tx_invalid_npkts = stats.tx_invalid_descs;
+ xsk->ring_stats.rx_full_npkts = stats.rx_ring_full;
+ xsk->ring_stats.rx_fill_empty_npkts = stats.rx_fill_ring_empty_descs;
+ xsk->ring_stats.tx_empty_npkts = stats.tx_ring_empty_descs;
+ return 0;
+ }
+
+ return -EINVAL;
+}
+
+static void dump_app_stats(long dt)
+{
+ int i;
+
+ for (i = 0; i < num_socks && xsks[i]; i++) {
+ char *fmt = "%-18s %'-14.0f %'-14lu\n";
+ double rx_empty_polls_ps, fill_fail_polls_ps, copy_tx_sendtos_ps,
+ tx_wakeup_sendtos_ps, opt_polls_ps;
+
+ rx_empty_polls_ps = (xsks[i]->app_stats.rx_empty_polls -
+ xsks[i]->app_stats.prev_rx_empty_polls) * 1000000000. / dt;
+ fill_fail_polls_ps = (xsks[i]->app_stats.fill_fail_polls -
+ xsks[i]->app_stats.prev_fill_fail_polls) * 1000000000. / dt;
+ copy_tx_sendtos_ps = (xsks[i]->app_stats.copy_tx_sendtos -
+ xsks[i]->app_stats.prev_copy_tx_sendtos) * 1000000000. / dt;
+ tx_wakeup_sendtos_ps = (xsks[i]->app_stats.tx_wakeup_sendtos -
+ xsks[i]->app_stats.prev_tx_wakeup_sendtos)
+ * 1000000000. / dt;
+ opt_polls_ps = (xsks[i]->app_stats.opt_polls -
+ xsks[i]->app_stats.prev_opt_polls) * 1000000000. / dt;
+
+ printf("\n%-18s %-14s %-14s\n", "", "calls/s", "count");
+ printf(fmt, "rx empty polls", rx_empty_polls_ps, xsks[i]->app_stats.rx_empty_polls);
+ printf(fmt, "fill fail polls", fill_fail_polls_ps,
+ xsks[i]->app_stats.fill_fail_polls);
+ printf(fmt, "copy tx sendtos", copy_tx_sendtos_ps,
+ xsks[i]->app_stats.copy_tx_sendtos);
+ printf(fmt, "tx wakeup sendtos", tx_wakeup_sendtos_ps,
+ xsks[i]->app_stats.tx_wakeup_sendtos);
+ printf(fmt, "opt polls", opt_polls_ps, xsks[i]->app_stats.opt_polls);
+
+ xsks[i]->app_stats.prev_rx_empty_polls = xsks[i]->app_stats.rx_empty_polls;
+ xsks[i]->app_stats.prev_fill_fail_polls = xsks[i]->app_stats.fill_fail_polls;
+ xsks[i]->app_stats.prev_copy_tx_sendtos = xsks[i]->app_stats.copy_tx_sendtos;
+ xsks[i]->app_stats.prev_tx_wakeup_sendtos = xsks[i]->app_stats.tx_wakeup_sendtos;
+ xsks[i]->app_stats.prev_opt_polls = xsks[i]->app_stats.opt_polls;
+ }
+}
+
+static bool get_interrupt_number(void)
+{
+ FILE *f_int_proc;
+ char line[4096];
+ bool found = false;
+
+ f_int_proc = fopen("/proc/interrupts", "r");
+ if (f_int_proc == NULL) {
+ printf("Failed to open /proc/interrupts.\n");
+ return found;
+ }
+
+ while (!feof(f_int_proc) && !found) {
+ /* Make sure to read a full line at a time */
+ if (fgets(line, sizeof(line), f_int_proc) == NULL ||
+ line[strlen(line) - 1] != '\n') {
+ printf("Error reading from interrupts file\n");
+ break;
+ }
+
+ /* Extract interrupt number from line */
+ if (strstr(line, opt_irq_str) != NULL) {
+ irq_no = atoi(line);
+ found = true;
+ break;
+ }
+ }
+
+ fclose(f_int_proc);
+
+ return found;
+}
+
+static int get_irqs(void)
+{
+ char count_path[PATH_MAX];
+ int total_intrs = -1;
+ FILE *f_count_proc;
+ char line[4096];
+
+ snprintf(count_path, sizeof(count_path),
+ "/sys/kernel/irq/%i/per_cpu_count", irq_no);
+ f_count_proc = fopen(count_path, "r");
+ if (f_count_proc == NULL) {
+ printf("Failed to open %s\n", count_path);
+ return total_intrs;
+ }
+
+ if (fgets(line, sizeof(line), f_count_proc) == NULL ||
+ line[strlen(line) - 1] != '\n') {
+ printf("Error reading from %s\n", count_path);
+ } else {
+ static const char com[2] = ",";
+ char *token;
+
+ total_intrs = 0;
+ token = strtok(line, com);
+ while (token != NULL) {
+ /* sum up interrupts across all cores */
+ total_intrs += atoi(token);
+ token = strtok(NULL, com);
+ }
+ }
+
+ fclose(f_count_proc);
+
+ return total_intrs;
+}
+
+static void dump_driver_stats(long dt)
+{
+ int i;
+
+ for (i = 0; i < num_socks && xsks[i]; i++) {
+ char *fmt = "%-18s %'-14.0f %'-14lu\n";
+ double intrs_ps;
+ int n_ints = get_irqs();
+
+ if (n_ints < 0) {
+ printf("error getting intr info for intr %i\n", irq_no);
+ return;
+ }
+ xsks[i]->drv_stats.intrs = n_ints - irqs_at_init;
+
+ intrs_ps = (xsks[i]->drv_stats.intrs - xsks[i]->drv_stats.prev_intrs) *
+ 1000000000. / dt;
+
+ printf("\n%-18s %-14s %-14s\n", "", "intrs/s", "count");
+ printf(fmt, "irqs", intrs_ps, xsks[i]->drv_stats.intrs);
+
+ xsks[i]->drv_stats.prev_intrs = xsks[i]->drv_stats.intrs;
+ }
+}
+
static void dump_stats(void)
{
unsigned long now = get_nsecs();
@@ -144,32 +356,100 @@
prev_time = now;
for (i = 0; i < num_socks && xsks[i]; i++) {
- char *fmt = "%-15s %'-11.0f %'-11lu\n";
- double rx_pps, tx_pps;
+ char *fmt = "%-18s %'-14.0f %'-14lu\n";
+ double rx_pps, tx_pps, dropped_pps, rx_invalid_pps, full_pps, fill_empty_pps,
+ tx_invalid_pps, tx_empty_pps;
- rx_pps = (xsks[i]->rx_npkts - xsks[i]->prev_rx_npkts) *
+ rx_pps = (xsks[i]->ring_stats.rx_npkts - xsks[i]->ring_stats.prev_rx_npkts) *
1000000000. / dt;
- tx_pps = (xsks[i]->tx_npkts - xsks[i]->prev_tx_npkts) *
+ tx_pps = (xsks[i]->ring_stats.tx_npkts - xsks[i]->ring_stats.prev_tx_npkts) *
1000000000. / dt;
printf("\n sock%d@", i);
print_benchmark(false);
printf("\n");
- printf("%-15s %-11s %-11s %-11.2f\n", "", "pps", "pkts",
+ printf("%-18s %-14s %-14s %-14.2f\n", "", "pps", "pkts",
dt / 1000000000.);
- printf(fmt, "rx", rx_pps, xsks[i]->rx_npkts);
- printf(fmt, "tx", tx_pps, xsks[i]->tx_npkts);
+ printf(fmt, "rx", rx_pps, xsks[i]->ring_stats.rx_npkts);
+ printf(fmt, "tx", tx_pps, xsks[i]->ring_stats.tx_npkts);
- xsks[i]->prev_rx_npkts = xsks[i]->rx_npkts;
- xsks[i]->prev_tx_npkts = xsks[i]->tx_npkts;
+ xsks[i]->ring_stats.prev_rx_npkts = xsks[i]->ring_stats.rx_npkts;
+ xsks[i]->ring_stats.prev_tx_npkts = xsks[i]->ring_stats.tx_npkts;
+
+ if (opt_extra_stats) {
+ if (!xsk_get_xdp_stats(xsk_socket__fd(xsks[i]->xsk), xsks[i])) {
+ dropped_pps = (xsks[i]->ring_stats.rx_dropped_npkts -
+ xsks[i]->ring_stats.prev_rx_dropped_npkts) *
+ 1000000000. / dt;
+ rx_invalid_pps = (xsks[i]->ring_stats.rx_invalid_npkts -
+ xsks[i]->ring_stats.prev_rx_invalid_npkts) *
+ 1000000000. / dt;
+ tx_invalid_pps = (xsks[i]->ring_stats.tx_invalid_npkts -
+ xsks[i]->ring_stats.prev_tx_invalid_npkts) *
+ 1000000000. / dt;
+ full_pps = (xsks[i]->ring_stats.rx_full_npkts -
+ xsks[i]->ring_stats.prev_rx_full_npkts) *
+ 1000000000. / dt;
+ fill_empty_pps = (xsks[i]->ring_stats.rx_fill_empty_npkts -
+ xsks[i]->ring_stats.prev_rx_fill_empty_npkts) *
+ 1000000000. / dt;
+ tx_empty_pps = (xsks[i]->ring_stats.tx_empty_npkts -
+ xsks[i]->ring_stats.prev_tx_empty_npkts) *
+ 1000000000. / dt;
+
+ printf(fmt, "rx dropped", dropped_pps,
+ xsks[i]->ring_stats.rx_dropped_npkts);
+ printf(fmt, "rx invalid", rx_invalid_pps,
+ xsks[i]->ring_stats.rx_invalid_npkts);
+ printf(fmt, "tx invalid", tx_invalid_pps,
+ xsks[i]->ring_stats.tx_invalid_npkts);
+ printf(fmt, "rx queue full", full_pps,
+ xsks[i]->ring_stats.rx_full_npkts);
+ printf(fmt, "fill ring empty", fill_empty_pps,
+ xsks[i]->ring_stats.rx_fill_empty_npkts);
+ printf(fmt, "tx ring empty", tx_empty_pps,
+ xsks[i]->ring_stats.tx_empty_npkts);
+
+ xsks[i]->ring_stats.prev_rx_dropped_npkts =
+ xsks[i]->ring_stats.rx_dropped_npkts;
+ xsks[i]->ring_stats.prev_rx_invalid_npkts =
+ xsks[i]->ring_stats.rx_invalid_npkts;
+ xsks[i]->ring_stats.prev_tx_invalid_npkts =
+ xsks[i]->ring_stats.tx_invalid_npkts;
+ xsks[i]->ring_stats.prev_rx_full_npkts =
+ xsks[i]->ring_stats.rx_full_npkts;
+ xsks[i]->ring_stats.prev_rx_fill_empty_npkts =
+ xsks[i]->ring_stats.rx_fill_empty_npkts;
+ xsks[i]->ring_stats.prev_tx_empty_npkts =
+ xsks[i]->ring_stats.tx_empty_npkts;
+ } else {
+ printf("%-15s\n", "Error retrieving extra stats");
+ }
+ }
}
+
+ if (opt_app_stats)
+ dump_app_stats(dt);
+ if (irq_no)
+ dump_driver_stats(dt);
+}
+
+static bool is_benchmark_done(void)
+{
+ if (opt_duration > 0) {
+ unsigned long dt = (get_nsecs() - start_time);
+
+ if (dt >= opt_duration)
+ benchmark_done = true;
+ }
+ return benchmark_done;
}
static void *poller(void *arg)
{
(void)arg;
- for (;;) {
+ while (!is_benchmark_done()) {
sleep(opt_interval);
dump_stats();
}
@@ -179,7 +459,7 @@
static void remove_xdp_program(void)
{
- __u32 curr_prog_id = 0;
+ u32 curr_prog_id = 0;
if (bpf_get_link_xdp_id(opt_ifindex, &curr_prog_id, opt_xdp_flags)) {
printf("bpf_get_link_xdp_id failed\n");
@@ -195,16 +475,19 @@
static void int_exit(int sig)
{
- struct xsk_umem *umem = xsks[0]->umem->umem;
+ benchmark_done = true;
+}
- (void)sig;
+static void xdpsock_cleanup(void)
+{
+ struct xsk_umem *umem = xsks[0]->umem->umem;
+ int i;
dump_stats();
- xsk_socket__delete(xsks[0]->xsk);
+ for (i = 0; i < num_socks; i++)
+ xsk_socket__delete(xsks[i]->xsk);
(void)xsk_umem__delete(umem);
remove_xdp_program();
-
- exit(EXIT_SUCCESS);
}
static void __exit_with_error(int error, const char *file, const char *func,
@@ -219,13 +502,6 @@
#define exit_with_error(error) __exit_with_error(error, __FILE__, __func__, \
__LINE__)
-
-static const char pkt_data[] =
- "\x3c\xfd\xfe\x9e\x7f\x71\xec\xb1\xd7\x98\x3a\xc0\x08\x00\x45\x00"
- "\x00\x2e\x00\x00\x00\x00\x40\x11\x88\x97\x05\x08\x07\x08\xc8\x14"
- "\x1e\x04\x10\x92\x10\x92\x00\x1a\x6d\xa3\x34\x33\x1f\x69\x40\x6b"
- "\x54\x59\xb6\x14\x2d\x11\x44\xbf\xaf\xd9\xbe\xaa";
-
static void swap_mac_addresses(void *data)
{
struct ether_header *eth = (struct ether_header *)data;
@@ -273,24 +549,264 @@
printf("\n");
}
-static size_t gen_eth_frame(struct xsk_umem_info *umem, u64 addr)
+static void *memset32_htonl(void *dest, u32 val, u32 size)
+{
+ u32 *ptr = (u32 *)dest;
+ int i;
+
+ val = htonl(val);
+
+ for (i = 0; i < (size & (~0x3)); i += 4)
+ ptr[i >> 2] = val;
+
+ for (; i < size; i++)
+ ((char *)dest)[i] = ((char *)&val)[i & 3];
+
+ return dest;
+}
+
+/*
+ * This function code has been taken from
+ * Linux kernel lib/checksum.c
+ */
+static inline unsigned short from32to16(unsigned int x)
+{
+ /* add up 16-bit and 16-bit for 16+c bit */
+ x = (x & 0xffff) + (x >> 16);
+ /* add up carry.. */
+ x = (x & 0xffff) + (x >> 16);
+ return x;
+}
+
+/*
+ * This function code has been taken from
+ * Linux kernel lib/checksum.c
+ */
+static unsigned int do_csum(const unsigned char *buff, int len)
+{
+ unsigned int result = 0;
+ int odd;
+
+ if (len <= 0)
+ goto out;
+ odd = 1 & (unsigned long)buff;
+ if (odd) {
+#ifdef __LITTLE_ENDIAN
+ result += (*buff << 8);
+#else
+ result = *buff;
+#endif
+ len--;
+ buff++;
+ }
+ if (len >= 2) {
+ if (2 & (unsigned long)buff) {
+ result += *(unsigned short *)buff;
+ len -= 2;
+ buff += 2;
+ }
+ if (len >= 4) {
+ const unsigned char *end = buff +
+ ((unsigned int)len & ~3);
+ unsigned int carry = 0;
+
+ do {
+ unsigned int w = *(unsigned int *)buff;
+
+ buff += 4;
+ result += carry;
+ result += w;
+ carry = (w > result);
+ } while (buff < end);
+ result += carry;
+ result = (result & 0xffff) + (result >> 16);
+ }
+ if (len & 2) {
+ result += *(unsigned short *)buff;
+ buff += 2;
+ }
+ }
+ if (len & 1)
+#ifdef __LITTLE_ENDIAN
+ result += *buff;
+#else
+ result += (*buff << 8);
+#endif
+ result = from32to16(result);
+ if (odd)
+ result = ((result >> 8) & 0xff) | ((result & 0xff) << 8);
+out:
+ return result;
+}
+
+__sum16 ip_fast_csum(const void *iph, unsigned int ihl);
+
+/*
+ * This is a version of ip_compute_csum() optimized for IP headers,
+ * which always checksum on 4 octet boundaries.
+ * This function code has been taken from
+ * Linux kernel lib/checksum.c
+ */
+__sum16 ip_fast_csum(const void *iph, unsigned int ihl)
+{
+ return (__force __sum16)~do_csum(iph, ihl * 4);
+}
+
+/*
+ * Fold a partial checksum
+ * This function code has been taken from
+ * Linux kernel include/asm-generic/checksum.h
+ */
+static inline __sum16 csum_fold(__wsum csum)
+{
+ u32 sum = (__force u32)csum;
+
+ sum = (sum & 0xffff) + (sum >> 16);
+ sum = (sum & 0xffff) + (sum >> 16);
+ return (__force __sum16)~sum;
+}
+
+/*
+ * This function code has been taken from
+ * Linux kernel lib/checksum.c
+ */
+static inline u32 from64to32(u64 x)
+{
+ /* add up 32-bit and 32-bit for 32+c bit */
+ x = (x & 0xffffffff) + (x >> 32);
+ /* add up carry.. */
+ x = (x & 0xffffffff) + (x >> 32);
+ return (u32)x;
+}
+
+__wsum csum_tcpudp_nofold(__be32 saddr, __be32 daddr,
+ __u32 len, __u8 proto, __wsum sum);
+
+/*
+ * This function code has been taken from
+ * Linux kernel lib/checksum.c
+ */
+__wsum csum_tcpudp_nofold(__be32 saddr, __be32 daddr,
+ __u32 len, __u8 proto, __wsum sum)
+{
+ unsigned long long s = (__force u32)sum;
+
+ s += (__force u32)saddr;
+ s += (__force u32)daddr;
+#ifdef __BIG_ENDIAN__
+ s += proto + len;
+#else
+ s += (proto + len) << 8;
+#endif
+ return (__force __wsum)from64to32(s);
+}
+
+/*
+ * This function has been taken from
+ * Linux kernel include/asm-generic/checksum.h
+ */
+static inline __sum16
+csum_tcpudp_magic(__be32 saddr, __be32 daddr, __u32 len,
+ __u8 proto, __wsum sum)
+{
+ return csum_fold(csum_tcpudp_nofold(saddr, daddr, len, proto, sum));
+}
+
+static inline u16 udp_csum(u32 saddr, u32 daddr, u32 len,
+ u8 proto, u16 *udp_pkt)
+{
+ u32 csum = 0;
+ u32 cnt = 0;
+
+ /* udp hdr and data */
+ for (; cnt < len; cnt += 2)
+ csum += udp_pkt[cnt >> 1];
+
+ return csum_tcpudp_magic(saddr, daddr, len, proto, csum);
+}
+
+#define ETH_FCS_SIZE 4
+
+#define PKT_HDR_SIZE (sizeof(struct ethhdr) + sizeof(struct iphdr) + \
+ sizeof(struct udphdr))
+
+#define PKT_SIZE (opt_pkt_size - ETH_FCS_SIZE)
+#define IP_PKT_SIZE (PKT_SIZE - sizeof(struct ethhdr))
+#define UDP_PKT_SIZE (IP_PKT_SIZE - sizeof(struct iphdr))
+#define UDP_PKT_DATA_SIZE (UDP_PKT_SIZE - sizeof(struct udphdr))
+
+static u8 pkt_data[XSK_UMEM__DEFAULT_FRAME_SIZE];
+
+static void gen_eth_hdr_data(void)
+{
+ struct udphdr *udp_hdr = (struct udphdr *)(pkt_data +
+ sizeof(struct ethhdr) +
+ sizeof(struct iphdr));
+ struct iphdr *ip_hdr = (struct iphdr *)(pkt_data +
+ sizeof(struct ethhdr));
+ struct ethhdr *eth_hdr = (struct ethhdr *)pkt_data;
+
+ /* ethernet header */
+ memcpy(eth_hdr->h_dest, "\x3c\xfd\xfe\x9e\x7f\x71", ETH_ALEN);
+ memcpy(eth_hdr->h_source, "\xec\xb1\xd7\x98\x3a\xc0", ETH_ALEN);
+ eth_hdr->h_proto = htons(ETH_P_IP);
+
+ /* IP header */
+ ip_hdr->version = IPVERSION;
+ ip_hdr->ihl = 0x5; /* 20 byte header */
+ ip_hdr->tos = 0x0;
+ ip_hdr->tot_len = htons(IP_PKT_SIZE);
+ ip_hdr->id = 0;
+ ip_hdr->frag_off = 0;
+ ip_hdr->ttl = IPDEFTTL;
+ ip_hdr->protocol = IPPROTO_UDP;
+ ip_hdr->saddr = htonl(0x0a0a0a10);
+ ip_hdr->daddr = htonl(0x0a0a0a20);
+
+ /* IP header checksum */
+ ip_hdr->check = 0;
+ ip_hdr->check = ip_fast_csum((const void *)ip_hdr, ip_hdr->ihl);
+
+ /* UDP header */
+ udp_hdr->source = htons(0x1000);
+ udp_hdr->dest = htons(0x1000);
+ udp_hdr->len = htons(UDP_PKT_SIZE);
+
+ /* UDP data */
+ memset32_htonl(pkt_data + PKT_HDR_SIZE, opt_pkt_fill_pattern,
+ UDP_PKT_DATA_SIZE);
+
+ /* UDP header checksum */
+ udp_hdr->check = 0;
+ udp_hdr->check = udp_csum(ip_hdr->saddr, ip_hdr->daddr, UDP_PKT_SIZE,
+ IPPROTO_UDP, (u16 *)udp_hdr);
+}
+
+static void gen_eth_frame(struct xsk_umem_info *umem, u64 addr)
{
memcpy(xsk_umem__get_data(umem->buffer, addr), pkt_data,
- sizeof(pkt_data) - 1);
- return sizeof(pkt_data) - 1;
+ PKT_SIZE);
}
static struct xsk_umem_info *xsk_configure_umem(void *buffer, u64 size)
{
struct xsk_umem_info *umem;
struct xsk_umem_config cfg = {
- .fill_size = XSK_RING_PROD__DEFAULT_NUM_DESCS,
+ /* We recommend that you set the fill ring size >= HW RX ring size +
+ * AF_XDP RX ring size. Make sure you fill up the fill ring
+ * with buffers at regular intervals, and you will with this setting
+ * avoid allocation failures in the driver. These are usually quite
+ * expensive since drivers have not been written to assume that
+ * allocation failures are common. For regular sockets, kernel
+ * allocated memory is used that only runs out in OOM situations
+ * that should be rare.
+ */
+ .fill_size = XSK_RING_PROD__DEFAULT_NUM_DESCS * 2,
.comp_size = XSK_RING_CONS__DEFAULT_NUM_DESCS,
.frame_size = opt_xsk_frame_size,
.frame_headroom = XSK_UMEM__DEFAULT_FRAME_HEADROOM,
.flags = opt_umem_flags
};
-
int ret;
umem = calloc(1, sizeof(*umem));
@@ -299,7 +815,6 @@
ret = xsk_umem__create(&umem->umem, buffer, size, &umem->fq, &umem->cq,
&cfg);
-
if (ret)
exit_with_error(-ret);
@@ -307,13 +822,29 @@
return umem;
}
-static struct xsk_socket_info *xsk_configure_socket(struct xsk_umem_info *umem)
+static void xsk_populate_fill_ring(struct xsk_umem_info *umem)
+{
+ int ret, i;
+ u32 idx;
+
+ ret = xsk_ring_prod__reserve(&umem->fq,
+ XSK_RING_PROD__DEFAULT_NUM_DESCS * 2, &idx);
+ if (ret != XSK_RING_PROD__DEFAULT_NUM_DESCS * 2)
+ exit_with_error(-ret);
+ for (i = 0; i < XSK_RING_PROD__DEFAULT_NUM_DESCS * 2; i++)
+ *xsk_ring_prod__fill_addr(&umem->fq, idx++) =
+ i * opt_xsk_frame_size;
+ xsk_ring_prod__submit(&umem->fq, XSK_RING_PROD__DEFAULT_NUM_DESCS * 2);
+}
+
+static struct xsk_socket_info *xsk_configure_socket(struct xsk_umem_info *umem,
+ bool rx, bool tx)
{
struct xsk_socket_config cfg;
struct xsk_socket_info *xsk;
+ struct xsk_ring_cons *rxr;
+ struct xsk_ring_prod *txr;
int ret;
- u32 idx;
- int i;
xsk = calloc(1, sizeof(*xsk));
if (!xsk)
@@ -322,11 +853,17 @@
xsk->umem = umem;
cfg.rx_size = XSK_RING_CONS__DEFAULT_NUM_DESCS;
cfg.tx_size = XSK_RING_PROD__DEFAULT_NUM_DESCS;
- cfg.libbpf_flags = 0;
+ if (opt_num_xsks > 1)
+ cfg.libbpf_flags = XSK_LIBBPF_FLAGS__INHIBIT_PROG_LOAD;
+ else
+ cfg.libbpf_flags = 0;
cfg.xdp_flags = opt_xdp_flags;
cfg.bind_flags = opt_xdp_bind_flags;
+
+ rxr = rx ? &xsk->rx : NULL;
+ txr = tx ? &xsk->tx : NULL;
ret = xsk_socket__create(&xsk->xsk, opt_if, opt_queue, umem->umem,
- &xsk->rx, &xsk->tx, &cfg);
+ rxr, txr, &cfg);
if (ret)
exit_with_error(-ret);
@@ -334,16 +871,16 @@
if (ret)
exit_with_error(-ret);
- ret = xsk_ring_prod__reserve(&xsk->umem->fq,
- XSK_RING_PROD__DEFAULT_NUM_DESCS,
- &idx);
- if (ret != XSK_RING_PROD__DEFAULT_NUM_DESCS)
- exit_with_error(-ret);
- for (i = 0; i < XSK_RING_PROD__DEFAULT_NUM_DESCS; i++)
- *xsk_ring_prod__fill_addr(&xsk->umem->fq, idx++) =
- i * opt_xsk_frame_size;
- xsk_ring_prod__submit(&xsk->umem->fq,
- XSK_RING_PROD__DEFAULT_NUM_DESCS);
+ xsk->app_stats.rx_empty_polls = 0;
+ xsk->app_stats.fill_fail_polls = 0;
+ xsk->app_stats.copy_tx_sendtos = 0;
+ xsk->app_stats.tx_wakeup_sendtos = 0;
+ xsk->app_stats.opt_polls = 0;
+ xsk->app_stats.prev_rx_empty_polls = 0;
+ xsk->app_stats.prev_fill_fail_polls = 0;
+ xsk->app_stats.prev_copy_tx_sendtos = 0;
+ xsk->app_stats.prev_tx_wakeup_sendtos = 0;
+ xsk->app_stats.prev_opt_polls = 0;
return xsk;
}
@@ -363,6 +900,17 @@
{"frame-size", required_argument, 0, 'f'},
{"no-need-wakeup", no_argument, 0, 'm'},
{"unaligned", no_argument, 0, 'u'},
+ {"shared-umem", no_argument, 0, 'M'},
+ {"force", no_argument, 0, 'F'},
+ {"duration", required_argument, 0, 'd'},
+ {"batch-size", required_argument, 0, 'b'},
+ {"tx-pkt-count", required_argument, 0, 'C'},
+ {"tx-pkt-size", required_argument, 0, 's'},
+ {"tx-pkt-pattern", required_argument, 0, 'P'},
+ {"extra-stats", no_argument, 0, 'x'},
+ {"quiet", no_argument, 0, 'Q'},
+ {"app-stats", no_argument, 0, 'a'},
+ {"irq-string", no_argument, 0, 'I'},
{0, 0, 0, 0}
};
@@ -378,16 +926,34 @@
" -q, --queue=n Use queue n (default 0)\n"
" -p, --poll Use poll syscall\n"
" -S, --xdp-skb=n Use XDP skb-mod\n"
- " -N, --xdp-native=n Enfore XDP native mode\n"
+ " -N, --xdp-native=n Enforce XDP native mode\n"
" -n, --interval=n Specify statistics update interval (default 1 sec).\n"
" -z, --zero-copy Force zero-copy mode.\n"
" -c, --copy Force copy mode.\n"
- " -f, --frame-size=n Set the frame size (must be a power of two, default is %d).\n"
" -m, --no-need-wakeup Turn off use of driver need wakeup flag.\n"
" -f, --frame-size=n Set the frame size (must be a power of two in aligned mode, default is %d).\n"
" -u, --unaligned Enable unaligned chunk placement\n"
+ " -M, --shared-umem Enable XDP_SHARED_UMEM\n"
+ " -F, --force Force loading the XDP prog\n"
+ " -d, --duration=n Duration in secs to run command.\n"
+ " Default: forever.\n"
+ " -b, --batch-size=n Batch size for sending or receiving\n"
+ " packets. Default: %d\n"
+ " -C, --tx-pkt-count=n Number of packets to send.\n"
+ " Default: Continuous packets.\n"
+ " -s, --tx-pkt-size=n Transmit packet size.\n"
+ " (Default: %d bytes)\n"
+ " Min size: %d, Max size %d.\n"
+ " -P, --tx-pkt-pattern=nPacket fill pattern. Default: 0x%x\n"
+ " -x, --extra-stats Display extra statistics.\n"
+ " -Q, --quiet Do not display any stats.\n"
+ " -a, --app-stats Display application (syscall) statistics.\n"
+ " -I, --irq-string Display driver interrupt statistics for interface associated with irq-string.\n"
"\n";
- fprintf(stderr, str, prog, XSK_UMEM__DEFAULT_FRAME_SIZE);
+ fprintf(stderr, str, prog, XSK_UMEM__DEFAULT_FRAME_SIZE,
+ opt_batch_size, MIN_PKT_SIZE, MIN_PKT_SIZE,
+ XSK_UMEM__DEFAULT_FRAME_SIZE, opt_pkt_fill_pattern);
+
exit(EXIT_FAILURE);
}
@@ -398,7 +964,7 @@
opterr = 0;
for (;;) {
- c = getopt_long(argc, argv, "Frtli:q:psSNn:czf:mu",
+ c = getopt_long(argc, argv, "Frtli:q:pSNn:czf:muMd:b:C:s:P:xQaI:",
long_options, &option_index);
if (c == -1)
break;
@@ -427,7 +993,7 @@
opt_xdp_bind_flags |= XDP_COPY;
break;
case 'N':
- opt_xdp_flags |= XDP_FLAGS_DRV_MODE;
+ /* default, set below */
break;
case 'n':
opt_interval = atoi(optarg);
@@ -448,16 +1014,64 @@
break;
case 'f':
opt_xsk_frame_size = atoi(optarg);
+ break;
case 'm':
opt_need_wakeup = false;
opt_xdp_bind_flags &= ~XDP_USE_NEED_WAKEUP;
break;
+ case 'M':
+ opt_num_xsks = MAX_SOCKS;
+ break;
+ case 'd':
+ opt_duration = atoi(optarg);
+ opt_duration *= 1000000000;
+ break;
+ case 'b':
+ opt_batch_size = atoi(optarg);
+ break;
+ case 'C':
+ opt_pkt_count = atoi(optarg);
+ break;
+ case 's':
+ opt_pkt_size = atoi(optarg);
+ if (opt_pkt_size > (XSK_UMEM__DEFAULT_FRAME_SIZE) ||
+ opt_pkt_size < MIN_PKT_SIZE) {
+ fprintf(stderr,
+ "ERROR: Invalid frame size %d\n",
+ opt_pkt_size);
+ usage(basename(argv[0]));
+ }
+ break;
+ case 'P':
+ opt_pkt_fill_pattern = strtol(optarg, NULL, 16);
+ break;
+ case 'x':
+ opt_extra_stats = 1;
+ break;
+ case 'Q':
+ opt_quiet = 1;
+ break;
+ case 'a':
+ opt_app_stats = 1;
+ break;
+ case 'I':
+ opt_irq_str = optarg;
+ if (get_interrupt_number())
+ irqs_at_init = get_irqs();
+ if (irqs_at_init < 0) {
+ fprintf(stderr, "ERROR: Failed to get irqs for %s\n", opt_irq_str);
+ usage(basename(argv[0]));
+ }
+ break;
default:
usage(basename(argv[0]));
}
}
+ if (!(opt_xdp_flags & XDP_FLAGS_SKB_MODE))
+ opt_xdp_flags |= XDP_FLAGS_DRV_MODE;
+
opt_ifindex = if_nametoindex(opt_if);
if (!opt_ifindex) {
fprintf(stderr, "ERROR: interface \"%s\" does not exist\n",
@@ -478,7 +1092,8 @@
int ret;
ret = sendto(xsk_socket__fd(xsk->xsk), NULL, 0, MSG_DONTWAIT, NULL, 0);
- if (ret >= 0 || errno == ENOBUFS || errno == EAGAIN || errno == EBUSY)
+ if (ret >= 0 || errno == ENOBUFS || errno == EAGAIN ||
+ errno == EBUSY || errno == ENETDOWN)
return;
exit_with_error(errno);
}
@@ -494,10 +1109,17 @@
if (!xsk->outstanding_tx)
return;
- if (!opt_need_wakeup || xsk_ring_prod__needs_wakeup(&xsk->tx))
+ /* In copy mode, Tx is driven by a syscall so we need to use e.g. sendto() to
+ * really send the packets. In zero-copy mode we do not have to do this, since Tx
+ * is driven by the NAPI loop. So as an optimization, we do not have to call
+ * sendto() all the time in zero-copy mode for l2fwd.
+ */
+ if (opt_xdp_bind_flags & XDP_COPY) {
+ xsk->app_stats.copy_tx_sendtos++;
kick_tx(xsk);
+ }
- ndescs = (xsk->outstanding_tx > BATCH_SIZE) ? BATCH_SIZE :
+ ndescs = (xsk->outstanding_tx > opt_batch_size) ? opt_batch_size :
xsk->outstanding_tx;
/* re-add completed Tx buffers */
@@ -510,8 +1132,10 @@
while (ret != rcvd) {
if (ret < 0)
exit_with_error(-ret);
- if (xsk_ring_prod__needs_wakeup(&umem->fq))
+ if (xsk_ring_prod__needs_wakeup(&umem->fq)) {
+ xsk->app_stats.fill_fail_polls++;
ret = poll(fds, num_socks, opt_timeout);
+ }
ret = xsk_ring_prod__reserve(&umem->fq, rcvd, &idx_fq);
}
@@ -522,11 +1146,12 @@
xsk_ring_prod__submit(&xsk->umem->fq, rcvd);
xsk_ring_cons__release(&xsk->umem->cq, rcvd);
xsk->outstanding_tx -= rcvd;
- xsk->tx_npkts += rcvd;
+ xsk->ring_stats.tx_npkts += rcvd;
}
}
-static inline void complete_tx_only(struct xsk_socket_info *xsk)
+static inline void complete_tx_only(struct xsk_socket_info *xsk,
+ int batch_size)
{
unsigned int rcvd;
u32 idx;
@@ -534,14 +1159,16 @@
if (!xsk->outstanding_tx)
return;
- if (!opt_need_wakeup || xsk_ring_prod__needs_wakeup(&xsk->tx))
+ if (!opt_need_wakeup || xsk_ring_prod__needs_wakeup(&xsk->tx)) {
+ xsk->app_stats.tx_wakeup_sendtos++;
kick_tx(xsk);
+ }
- rcvd = xsk_ring_cons__peek(&xsk->umem->cq, BATCH_SIZE, &idx);
+ rcvd = xsk_ring_cons__peek(&xsk->umem->cq, batch_size, &idx);
if (rcvd > 0) {
xsk_ring_cons__release(&xsk->umem->cq, rcvd);
xsk->outstanding_tx -= rcvd;
- xsk->tx_npkts += rcvd;
+ xsk->ring_stats.tx_npkts += rcvd;
}
}
@@ -551,10 +1178,12 @@
u32 idx_rx = 0, idx_fq = 0;
int ret;
- rcvd = xsk_ring_cons__peek(&xsk->rx, BATCH_SIZE, &idx_rx);
+ rcvd = xsk_ring_cons__peek(&xsk->rx, opt_batch_size, &idx_rx);
if (!rcvd) {
- if (xsk_ring_prod__needs_wakeup(&xsk->umem->fq))
+ if (xsk_ring_prod__needs_wakeup(&xsk->umem->fq)) {
+ xsk->app_stats.rx_empty_polls++;
ret = poll(fds, num_socks, opt_timeout);
+ }
return;
}
@@ -562,8 +1191,10 @@
while (ret != rcvd) {
if (ret < 0)
exit_with_error(-ret);
- if (xsk_ring_prod__needs_wakeup(&xsk->umem->fq))
+ if (xsk_ring_prod__needs_wakeup(&xsk->umem->fq)) {
+ xsk->app_stats.fill_fail_polls++;
ret = poll(fds, num_socks, opt_timeout);
+ }
ret = xsk_ring_prod__reserve(&xsk->umem->fq, rcvd, &idx_fq);
}
@@ -581,16 +1212,14 @@
xsk_ring_prod__submit(&xsk->umem->fq, rcvd);
xsk_ring_cons__release(&xsk->rx, rcvd);
- xsk->rx_npkts += rcvd;
+ xsk->ring_stats.rx_npkts += rcvd;
}
static void rx_drop_all(void)
{
- struct pollfd fds[MAX_SOCKS + 1];
+ struct pollfd fds[MAX_SOCKS] = {};
int i, ret;
- memset(fds, 0, sizeof(fds));
-
for (i = 0; i < num_socks; i++) {
fds[i].fd = xsk_socket__fd(xsks[i]->xsk);
fds[i].events = POLLIN;
@@ -598,6 +1227,8 @@
for (;;) {
if (opt_poll) {
+ for (i = 0; i < num_socks; i++)
+ xsks[i]->app_stats.opt_polls++;
ret = poll(fds, num_socks, opt_timeout);
if (ret <= 0)
continue;
@@ -605,46 +1236,83 @@
for (i = 0; i < num_socks; i++)
rx_drop(xsks[i], fds);
+
+ if (benchmark_done)
+ break;
}
}
-static void tx_only(struct xsk_socket_info *xsk, u32 frame_nb)
+static void tx_only(struct xsk_socket_info *xsk, u32 *frame_nb, int batch_size)
{
u32 idx;
+ unsigned int i;
- if (xsk_ring_prod__reserve(&xsk->tx, BATCH_SIZE, &idx) == BATCH_SIZE) {
- unsigned int i;
-
- for (i = 0; i < BATCH_SIZE; i++) {
- xsk_ring_prod__tx_desc(&xsk->tx, idx + i)->addr =
- (frame_nb + i) << XSK_UMEM__DEFAULT_FRAME_SHIFT;
- xsk_ring_prod__tx_desc(&xsk->tx, idx + i)->len =
- sizeof(pkt_data) - 1;
- }
-
- xsk_ring_prod__submit(&xsk->tx, BATCH_SIZE);
- xsk->outstanding_tx += BATCH_SIZE;
- frame_nb += BATCH_SIZE;
- frame_nb %= NUM_FRAMES;
+ while (xsk_ring_prod__reserve(&xsk->tx, batch_size, &idx) <
+ batch_size) {
+ complete_tx_only(xsk, batch_size);
+ if (benchmark_done)
+ return;
}
- complete_tx_only(xsk);
+ for (i = 0; i < batch_size; i++) {
+ struct xdp_desc *tx_desc = xsk_ring_prod__tx_desc(&xsk->tx,
+ idx + i);
+ tx_desc->addr = (*frame_nb + i) * opt_xsk_frame_size;
+ tx_desc->len = PKT_SIZE;
+ }
+
+ xsk_ring_prod__submit(&xsk->tx, batch_size);
+ xsk->outstanding_tx += batch_size;
+ *frame_nb += batch_size;
+ *frame_nb %= NUM_FRAMES;
+ complete_tx_only(xsk, batch_size);
+}
+
+static inline int get_batch_size(int pkt_cnt)
+{
+ if (!opt_pkt_count)
+ return opt_batch_size;
+
+ if (pkt_cnt + opt_batch_size <= opt_pkt_count)
+ return opt_batch_size;
+
+ return opt_pkt_count - pkt_cnt;
+}
+
+static void complete_tx_only_all(void)
+{
+ bool pending;
+ int i;
+
+ do {
+ pending = false;
+ for (i = 0; i < num_socks; i++) {
+ if (xsks[i]->outstanding_tx) {
+ complete_tx_only(xsks[i], opt_batch_size);
+ pending = !!xsks[i]->outstanding_tx;
+ }
+ }
+ } while (pending);
}
static void tx_only_all(void)
{
- struct pollfd fds[MAX_SOCKS];
+ struct pollfd fds[MAX_SOCKS] = {};
u32 frame_nb[MAX_SOCKS] = {};
+ int pkt_cnt = 0;
int i, ret;
- memset(fds, 0, sizeof(fds));
for (i = 0; i < num_socks; i++) {
fds[0].fd = xsk_socket__fd(xsks[i]->xsk);
fds[0].events = POLLOUT;
}
- for (;;) {
+ while ((opt_pkt_count && pkt_cnt < opt_pkt_count) || !opt_pkt_count) {
+ int batch_size = get_batch_size(pkt_cnt);
+
if (opt_poll) {
+ for (i = 0; i < num_socks; i++)
+ xsks[i]->app_stats.opt_polls++;
ret = poll(fds, num_socks, opt_timeout);
if (ret <= 0)
continue;
@@ -654,8 +1322,16 @@
}
for (i = 0; i < num_socks; i++)
- tx_only(xsks[i], frame_nb[i]);
+ tx_only(xsks[i], &frame_nb[i], batch_size);
+
+ pkt_cnt += batch_size;
+
+ if (benchmark_done)
+ break;
}
+
+ if (opt_pkt_count)
+ complete_tx_only_all();
}
static void l2fwd(struct xsk_socket_info *xsk, struct pollfd *fds)
@@ -666,10 +1342,12 @@
complete_tx_l2fwd(xsk, fds);
- rcvd = xsk_ring_cons__peek(&xsk->rx, BATCH_SIZE, &idx_rx);
+ rcvd = xsk_ring_cons__peek(&xsk->rx, opt_batch_size, &idx_rx);
if (!rcvd) {
- if (xsk_ring_prod__needs_wakeup(&xsk->umem->fq))
+ if (xsk_ring_prod__needs_wakeup(&xsk->umem->fq)) {
+ xsk->app_stats.rx_empty_polls++;
ret = poll(fds, num_socks, opt_timeout);
+ }
return;
}
@@ -678,8 +1356,10 @@
if (ret < 0)
exit_with_error(-ret);
complete_tx_l2fwd(xsk, fds);
- if (xsk_ring_prod__needs_wakeup(&xsk->tx))
+ if (xsk_ring_prod__needs_wakeup(&xsk->tx)) {
+ xsk->app_stats.tx_wakeup_sendtos++;
kick_tx(xsk);
+ }
ret = xsk_ring_prod__reserve(&xsk->tx, rcvd, &idx_tx);
}
@@ -701,17 +1381,15 @@
xsk_ring_prod__submit(&xsk->tx, rcvd);
xsk_ring_cons__release(&xsk->rx, rcvd);
- xsk->rx_npkts += rcvd;
+ xsk->ring_stats.rx_npkts += rcvd;
xsk->outstanding_tx += rcvd;
}
static void l2fwd_all(void)
{
- struct pollfd fds[MAX_SOCKS];
+ struct pollfd fds[MAX_SOCKS] = {};
int i, ret;
- memset(fds, 0, sizeof(fds));
-
for (i = 0; i < num_socks; i++) {
fds[i].fd = xsk_socket__fd(xsks[i]->xsk);
fds[i].events = POLLOUT | POLLIN;
@@ -719,6 +1397,8 @@
for (;;) {
if (opt_poll) {
+ for (i = 0; i < num_socks; i++)
+ xsks[i]->app_stats.opt_polls++;
ret = poll(fds, num_socks, opt_timeout);
if (ret <= 0)
continue;
@@ -726,16 +1406,72 @@
for (i = 0; i < num_socks; i++)
l2fwd(xsks[i], fds);
+
+ if (benchmark_done)
+ break;
+ }
+}
+
+static void load_xdp_program(char **argv, struct bpf_object **obj)
+{
+ struct bpf_prog_load_attr prog_load_attr = {
+ .prog_type = BPF_PROG_TYPE_XDP,
+ };
+ char xdp_filename[256];
+ int prog_fd;
+
+ snprintf(xdp_filename, sizeof(xdp_filename), "%s_kern.o", argv[0]);
+ prog_load_attr.file = xdp_filename;
+
+ if (bpf_prog_load_xattr(&prog_load_attr, obj, &prog_fd))
+ exit(EXIT_FAILURE);
+ if (prog_fd < 0) {
+ fprintf(stderr, "ERROR: no program found: %s\n",
+ strerror(prog_fd));
+ exit(EXIT_FAILURE);
+ }
+
+ if (bpf_set_link_xdp_fd(opt_ifindex, prog_fd, opt_xdp_flags) < 0) {
+ fprintf(stderr, "ERROR: link set xdp fd failed\n");
+ exit(EXIT_FAILURE);
+ }
+}
+
+static void enter_xsks_into_map(struct bpf_object *obj)
+{
+ struct bpf_map *map;
+ int i, xsks_map;
+
+ map = bpf_object__find_map_by_name(obj, "xsks_map");
+ xsks_map = bpf_map__fd(map);
+ if (xsks_map < 0) {
+ fprintf(stderr, "ERROR: no xsks map found: %s\n",
+ strerror(xsks_map));
+ exit(EXIT_FAILURE);
+ }
+
+ for (i = 0; i < num_socks; i++) {
+ int fd = xsk_socket__fd(xsks[i]->xsk);
+ int key, ret;
+
+ key = i;
+ ret = bpf_map_update_elem(xsks_map, &key, &fd, 0);
+ if (ret) {
+ fprintf(stderr, "ERROR: bpf_map_update_elem %d\n", i);
+ exit(EXIT_FAILURE);
+ }
}
}
int main(int argc, char **argv)
{
struct rlimit r = {RLIM_INFINITY, RLIM_INFINITY};
+ bool rx = false, tx = false;
struct xsk_umem_info *umem;
+ struct bpf_object *obj;
pthread_t pt;
+ int i, ret;
void *bufs;
- int ret;
parse_command_line(argc, argv);
@@ -745,6 +1481,9 @@
exit(EXIT_FAILURE);
}
+ if (opt_num_xsks > 1)
+ load_xdp_program(argv, &obj);
+
/* Reserve memory for the umem. Use hugepages if unaligned chunk mode */
bufs = mmap(NULL, NUM_FRAMES * opt_xsk_frame_size,
PROT_READ | PROT_WRITE,
@@ -753,28 +1492,42 @@
printf("ERROR: mmap failed\n");
exit(EXIT_FAILURE);
}
- /* Create sockets... */
+
+ /* Create sockets... */
umem = xsk_configure_umem(bufs, NUM_FRAMES * opt_xsk_frame_size);
- xsks[num_socks++] = xsk_configure_socket(umem);
+ if (opt_bench == BENCH_RXDROP || opt_bench == BENCH_L2FWD) {
+ rx = true;
+ xsk_populate_fill_ring(umem);
+ }
+ if (opt_bench == BENCH_L2FWD || opt_bench == BENCH_TXONLY)
+ tx = true;
+ for (i = 0; i < opt_num_xsks; i++)
+ xsks[num_socks++] = xsk_configure_socket(umem, rx, tx);
if (opt_bench == BENCH_TXONLY) {
- int i;
+ gen_eth_hdr_data();
for (i = 0; i < NUM_FRAMES; i++)
- (void)gen_eth_frame(umem, i * opt_xsk_frame_size);
+ gen_eth_frame(umem, i * opt_xsk_frame_size);
}
+ if (opt_num_xsks > 1 && opt_bench != BENCH_TXONLY)
+ enter_xsks_into_map(obj);
+
signal(SIGINT, int_exit);
signal(SIGTERM, int_exit);
signal(SIGABRT, int_exit);
setlocale(LC_ALL, "");
- ret = pthread_create(&pt, NULL, poller, NULL);
- if (ret)
- exit_with_error(ret);
+ if (!opt_quiet) {
+ ret = pthread_create(&pt, NULL, poller, NULL);
+ if (ret)
+ exit_with_error(ret);
+ }
prev_time = get_nsecs();
+ start_time = prev_time;
if (opt_bench == BENCH_RXDROP)
rx_drop_all();
@@ -783,6 +1536,13 @@
else
l2fwd_all();
+ benchmark_done = true;
+
+ if (!opt_quiet)
+ pthread_join(pt, NULL);
+
+ xdpsock_cleanup();
+
munmap(bufs, NUM_FRAMES * opt_xsk_frame_size);
return 0;