server: remove leftover debug statement
[fio.git] / server.c
index 9537c967626784e5fee2fbb78a946f14d564ae7a..1d374fc69f5c7b37d500cb53e65057dd94a21880 100644 (file)
--- a/server.c
+++ b/server.c
@@ -4,35 +4,77 @@
 #include <unistd.h>
 #include <limits.h>
 #include <errno.h>
-#include <fcntl.h>
 #include <sys/poll.h>
 #include <sys/types.h>
 #include <sys/wait.h>
 #include <sys/socket.h>
 #include <sys/stat.h>
 #include <sys/un.h>
+#include <sys/uio.h>
 #include <netinet/in.h>
 #include <arpa/inet.h>
 #include <netdb.h>
 #include <syslog.h>
 #include <signal.h>
+#ifdef CONFIG_ZLIB
 #include <zlib.h>
+#endif
 
 #include "fio.h"
+#include "options.h"
 #include "server.h"
 #include "crc/crc16.h"
 #include "lib/ieee754.h"
+#include "verify.h"
+#include "smalloc.h"
 
 int fio_net_port = FIO_NET_PORT;
 
 int exit_backend = 0;
 
-static int server_fd = -1;
+enum {
+       SK_F_FREE       = 1,
+       SK_F_COPY       = 2,
+       SK_F_SIMPLE     = 4,
+       SK_F_VEC        = 8,
+       SK_F_INLINE     = 16,
+};
+
+struct sk_entry {
+       struct flist_head list; /* link on sk_out->list */
+       int flags;              /* SK_F_* */
+       int opcode;             /* Actual command fields */
+       void *buf;
+       off_t size;
+       uint64_t *tagptr;
+       struct flist_head next; /* Other sk_entry's, if linked command */
+};
+
+struct sk_out {
+       unsigned int refs;      /* frees sk_out when it drops to zero.
+                                * protected by below ->lock */
+
+       int sk;                 /* socket fd to talk to client */
+       struct fio_mutex lock;  /* protects ref and below list */
+       struct flist_head list; /* list of pending transmit work */
+       struct fio_mutex wait;  /* wake backend when items added to list */
+       struct fio_mutex xmit;  /* held while sending data */
+};
+
 static char *fio_server_arg;
 static char *bind_sock;
 static struct sockaddr_in saddr_in;
 static struct sockaddr_in6 saddr_in6;
 static int use_ipv6;
+#ifdef CONFIG_ZLIB
+static unsigned int has_zlib = 1;
+#else
+static unsigned int has_zlib = 0;
+#endif
+static unsigned int use_zlib;
+static char me[128];
+
+static pthread_key_t sk_out_key;
 
 struct fio_fork_item {
        struct flist_head list;
@@ -42,11 +84,12 @@ struct fio_fork_item {
        pid_t pid;
 };
 
-/* Created on fork on new connection */
-static FLIST_HEAD(conn_list);
-
-/* Created on job fork from connection */
-static FLIST_HEAD(job_list);
+struct cmd_reply {
+       struct fio_mutex lock;
+       void *data;
+       size_t size;
+       int error;
+};
 
 static const char *fio_server_ops[FIO_NET_CMD_NR] = {
        "",
@@ -65,10 +108,91 @@ static const char *fio_server_ops[FIO_NET_CMD_NR] = {
        "DISK_UTIL",
        "SERVER_START",
        "ADD_JOB",
-       "CMD_RUN"
-       "CMD_IOLOG",
+       "RUN",
+       "IOLOG",
+       "UPDATE_JOB",
+       "LOAD_FILE",
+       "VTRIGGER",
+       "SENDFILE",
 };
 
+static void sk_lock(struct sk_out *sk_out)
+{
+       fio_mutex_down(&sk_out->lock);
+}
+
+static void sk_unlock(struct sk_out *sk_out)
+{
+       fio_mutex_up(&sk_out->lock);
+}
+
+void sk_out_assign(struct sk_out *sk_out)
+{
+       if (!sk_out)
+               return;
+
+       sk_lock(sk_out);
+       sk_out->refs++;
+       sk_unlock(sk_out);
+       pthread_setspecific(sk_out_key, sk_out);
+}
+
+static void sk_out_free(struct sk_out *sk_out)
+{
+       __fio_mutex_remove(&sk_out->lock);
+       __fio_mutex_remove(&sk_out->wait);
+       __fio_mutex_remove(&sk_out->xmit);
+       sfree(sk_out);
+}
+
+static int __sk_out_drop(struct sk_out *sk_out)
+{
+       if (sk_out) {
+               int refs;
+
+               sk_lock(sk_out);
+               refs = --sk_out->refs;
+               sk_unlock(sk_out);
+
+               if (!refs) {
+                       sk_out_free(sk_out);
+                       return 0;
+               }
+       }
+
+       return 1;
+}
+
+void sk_out_drop(void)
+{
+       struct sk_out *sk_out;
+
+       sk_out = pthread_getspecific(sk_out_key);
+       if (!__sk_out_drop(sk_out))
+               pthread_setspecific(sk_out_key, NULL);
+}
+
+static void __fio_init_net_cmd(struct fio_net_cmd *cmd, uint16_t opcode,
+                              uint32_t pdu_len, uint64_t tag)
+{
+       memset(cmd, 0, sizeof(*cmd));
+
+       cmd->version    = __cpu_to_le16(FIO_SERVER_VER);
+       cmd->opcode     = cpu_to_le16(opcode);
+       cmd->tag        = cpu_to_le64(tag);
+       cmd->pdu_len    = cpu_to_le32(pdu_len);
+}
+
+
+static void fio_init_net_cmd(struct fio_net_cmd *cmd, uint16_t opcode,
+                            const void *pdu, uint32_t pdu_len, uint64_t tag)
+{
+       __fio_init_net_cmd(cmd, opcode, pdu_len, tag);
+
+       if (pdu)
+               memcpy(&cmd->payload, pdu, pdu_len);
+}
+
 const char *fio_server_op(unsigned int op)
 {
        static char buf[32];
@@ -125,13 +249,10 @@ static int fio_sendv_data(int sk, struct iovec *iov, int count)
        if (!total_len)
                return 0;
 
-       if (errno)
-               return -errno;
-
        return 1;
 }
 
-int fio_send_data(int sk, const void *p, unsigned int len)
+static int fio_send_data(int sk, const void *p, unsigned int len)
 {
        struct iovec iov = { .iov_base = (void *) p, .iov_len = len };
 
@@ -140,7 +261,7 @@ int fio_send_data(int sk, const void *p, unsigned int len)
        return fio_sendv_data(sk, &iov, 1);
 }
 
-int fio_recv_data(int sk, void *p, unsigned int len)
+static int fio_recv_data(int sk, void *p, unsigned int len)
 {
        do {
                int ret = recv(sk, p, len, MSG_WAITALL);
@@ -206,7 +327,7 @@ static int verify_convert_cmd(struct fio_net_cmd *cmd)
  */
 struct fio_net_cmd *fio_net_recv_cmd(int sk)
 {
-       struct fio_net_cmd cmd, *cmdret = NULL;
+       struct fio_net_cmd cmd, *tmp, *cmdret = NULL;
        size_t cmd_size = 0, pdu_offset = 0;
        uint16_t crc;
        int ret, first = 1;
@@ -229,7 +350,19 @@ struct fio_net_cmd *fio_net_recv_cmd(int sk)
                } else
                        cmd_size += cmd.pdu_len;
 
-               cmdret = realloc(cmdret, cmd_size);
+               if (cmd_size / 1024 > FIO_SERVER_MAX_CMD_MB * 1024) {
+                       log_err("fio: cmd+pdu too large (%llu)\n", (unsigned long long) cmd_size);
+                       ret = 1;
+                       break;
+               }
+
+               tmp = realloc(cmdret, cmd_size);
+               if (!tmp) {
+                       log_err("fio: server failed allocating cmd\n");
+                       ret = 1;
+                       break;
+               }
+               cmdret = tmp;
 
                if (first)
                        memcpy(cmdret, &cmd, sizeof(cmd));
@@ -271,14 +404,14 @@ struct fio_net_cmd *fio_net_recv_cmd(int sk)
                /* zero-terminate text input */
                if (cmdret->pdu_len) {
                        if (cmdret->opcode == FIO_NET_CMD_TEXT) {
-                               struct cmd_text_pdu *pdu = (struct cmd_text_pdu *) cmdret->payload;
-                               char *buf = (char *) pdu->buf;
+                               struct cmd_text_pdu *__pdu = (struct cmd_text_pdu *) cmdret->payload;
+                               char *buf = (char *) __pdu->buf;
 
-                               buf[pdu->buf_len] = '\0';
+                               buf[__pdu->buf_len] = '\0';
                        } else if (cmdret->opcode == FIO_NET_CMD_JOB) {
-                               struct cmd_job_pdu *pdu = (struct cmd_job_pdu *) cmdret->payload;
-                               char *buf = (char *) pdu->buf;
-                               int len = le32_to_cpu(pdu->buf_len);
+                               struct cmd_job_pdu *__pdu = (struct cmd_job_pdu *) cmdret->payload;
+                               char *buf = (char *) __pdu->buf;
+                               int len = le32_to_cpu(__pdu->buf_len);
 
                                buf[len] = '\0';
                        }
@@ -293,8 +426,9 @@ struct fio_net_cmd *fio_net_recv_cmd(int sk)
 
 static void add_reply(uint64_t tag, struct flist_head *list)
 {
-       struct fio_net_cmd_reply *reply = (struct fio_net_cmd_reply *) tag;
+       struct fio_net_cmd_reply *reply;
 
+       reply = (struct fio_net_cmd_reply *) (uintptr_t) tag;
        flist_add_tail(&reply->list, list);
 }
 
@@ -304,7 +438,7 @@ static uint64_t alloc_reply(uint64_t tag, uint16_t opcode)
 
        reply = calloc(1, sizeof(*reply));
        INIT_FLIST_HEAD(&reply->list);
-       gettimeofday(&reply->tv, NULL);
+       fio_gettime(&reply->tv, NULL);
        reply->saved_tag = tag;
        reply->opcode = opcode;
 
@@ -313,12 +447,13 @@ static uint64_t alloc_reply(uint64_t tag, uint16_t opcode)
 
 static void free_reply(uint64_t tag)
 {
-       struct fio_net_cmd_reply *reply = (struct fio_net_cmd_reply *) tag;
+       struct fio_net_cmd_reply *reply;
 
+       reply = (struct fio_net_cmd_reply *) (uintptr_t) tag;
        free(reply);
 }
 
-void fio_net_cmd_crc_pdu(struct fio_net_cmd *cmd, const void *pdu)
+static void fio_net_cmd_crc_pdu(struct fio_net_cmd *cmd, const void *pdu)
 {
        uint32_t pdu_len;
 
@@ -328,7 +463,7 @@ void fio_net_cmd_crc_pdu(struct fio_net_cmd *cmd, const void *pdu)
        cmd->pdu_crc16 = __cpu_to_le16(fio_crc16(pdu, pdu_len));
 }
 
-void fio_net_cmd_crc(struct fio_net_cmd *cmd)
+static void fio_net_cmd_crc(struct fio_net_cmd *cmd)
 {
        fio_net_cmd_crc_pdu(cmd, cmd->payload);
 }
@@ -385,6 +520,54 @@ int fio_net_send_cmd(int fd, uint16_t opcode, const void *buf, off_t size,
        return ret;
 }
 
+static struct sk_entry *fio_net_prep_cmd(uint16_t opcode, void *buf,
+                                        size_t size, uint64_t *tagptr,
+                                        int flags)
+{
+       struct sk_entry *entry;
+
+       entry = smalloc(sizeof(*entry));
+       INIT_FLIST_HEAD(&entry->next);
+       entry->opcode = opcode;
+       if (flags & SK_F_COPY) {
+               entry->buf = smalloc(size);
+               memcpy(entry->buf, buf, size);
+       } else
+               entry->buf = buf;
+
+       entry->size = size;
+       entry->tagptr = tagptr;
+       entry->flags = flags;
+       return entry;
+}
+
+static int handle_sk_entry(struct sk_out *sk_out, struct sk_entry *entry);
+
+static void fio_net_queue_entry(struct sk_entry *entry)
+{
+       struct sk_out *sk_out = pthread_getspecific(sk_out_key);
+
+       if (entry->flags & SK_F_INLINE)
+               handle_sk_entry(sk_out, entry);
+       else {
+               sk_lock(sk_out);
+               flist_add_tail(&entry->list, &sk_out->list);
+               sk_unlock(sk_out);
+
+               fio_mutex_up(&sk_out->wait);
+       }
+}
+
+static int fio_net_queue_cmd(uint16_t opcode, void *buf, off_t size,
+                            uint64_t *tagptr, int flags)
+{
+       struct sk_entry *entry;
+
+       entry = fio_net_prep_cmd(opcode, buf, size, tagptr, flags);
+       fio_net_queue_entry(entry);
+       return 0;
+}
+
 static int fio_net_send_simple_stack_cmd(int sk, uint16_t opcode, uint64_t tag)
 {
        struct fio_net_cmd cmd;
@@ -421,6 +604,13 @@ int fio_net_send_simple_cmd(int sk, uint16_t opcode, uint64_t tag,
        return 0;
 }
 
+static int fio_net_queue_quit(void)
+{
+       dprint(FD_NET, "server: sending quit\n");
+
+       return fio_net_queue_cmd(FIO_NET_CMD_QUIT, NULL, 0, 0, SK_F_SIMPLE);
+}
+
 int fio_net_send_quit(int sk)
 {
        dprint(FD_NET, "server: sending quit\n");
@@ -428,8 +618,7 @@ int fio_net_send_quit(int sk)
        return fio_net_send_simple_cmd(sk, FIO_NET_CMD_QUIT, 0, NULL);
 }
 
-static int fio_net_send_ack(int sk, struct fio_net_cmd *cmd, int error,
-                           int signal)
+static int fio_net_send_ack(struct fio_net_cmd *cmd, int error, int signal)
 {
        struct cmd_end_pdu epdu;
        uint64_t tag = 0;
@@ -439,13 +628,13 @@ static int fio_net_send_ack(int sk, struct fio_net_cmd *cmd, int error,
 
        epdu.error = __cpu_to_le32(error);
        epdu.signal = __cpu_to_le32(signal);
-       return fio_net_send_cmd(sk, FIO_NET_CMD_STOP, &epdu, sizeof(epdu), &tag, NULL);
+       return fio_net_queue_cmd(FIO_NET_CMD_STOP, &epdu, sizeof(epdu), &tag, SK_F_COPY);
 }
 
-int fio_net_send_stop(int sk, int error, int signal)
+static int fio_net_queue_stop(int error, int signal)
 {
        dprint(FD_NET, "server: sending stop (%d, %d)\n", error, signal);
-       return fio_net_send_ack(sk, NULL, error, signal);
+       return fio_net_send_ack(NULL, error, signal);
 }
 
 static void fio_server_add_fork_item(pid_t pid, struct flist_head *list)
@@ -460,16 +649,16 @@ static void fio_server_add_fork_item(pid_t pid, struct flist_head *list)
        flist_add_tail(&ffi->list, list);
 }
 
-static void fio_server_add_conn_pid(pid_t pid)
+static void fio_server_add_conn_pid(struct flist_head *conn_list, pid_t pid)
 {
-       dprint(FD_NET, "server: forked off connection job (pid=%u)\n", pid);
-       fio_server_add_fork_item(pid, &conn_list);
+       dprint(FD_NET, "server: forked off connection job (pid=%u)\n", (int) pid);
+       fio_server_add_fork_item(pid, conn_list);
 }
 
-static void fio_server_add_job_pid(pid_t pid)
+static void fio_server_add_job_pid(struct flist_head *job_list, pid_t pid)
 {
-       dprint(FD_NET, "server: forked off job job (pid=%u)\n", pid);
-       fio_server_add_fork_item(pid, &job_list);
+       dprint(FD_NET, "server: forked off job job (pid=%u)\n", (int) pid);
+       fio_server_add_fork_item(pid, job_list);
 }
 
 static void fio_server_check_fork_item(struct fio_fork_item *ffi)
@@ -479,7 +668,7 @@ static void fio_server_check_fork_item(struct fio_fork_item *ffi)
        ret = waitpid(ffi->pid, &status, WNOHANG);
        if (ret < 0) {
                if (errno == ECHILD) {
-                       log_err("fio: connection pid %u disappeared\n", ffi->pid);
+                       log_err("fio: connection pid %u disappeared\n", (int) ffi->pid);
                        ffi->exited = 1;
                } else
                        log_err("fio: waitpid: %s\n", strerror(errno));
@@ -496,20 +685,23 @@ static void fio_server_check_fork_item(struct fio_fork_item *ffi)
        }
 }
 
-static void fio_server_fork_item_done(struct fio_fork_item *ffi)
+static void fio_server_fork_item_done(struct fio_fork_item *ffi, bool stop)
 {
-       dprint(FD_NET, "pid %u exited, sig=%u, exitval=%d\n", ffi->pid, ffi->signal, ffi->exitval);
+       dprint(FD_NET, "pid %u exited, sig=%u, exitval=%d\n", (int) ffi->pid, ffi->signal, ffi->exitval);
 
        /*
         * Fold STOP and QUIT...
         */
-       fio_net_send_stop(server_fd, ffi->exitval, ffi->signal);
-       fio_net_send_quit(server_fd);
+       if (stop) {
+               fio_net_queue_stop(ffi->exitval, ffi->signal);
+               fio_net_queue_quit();
+       }
+
        flist_del(&ffi->list);
        free(ffi);
 }
 
-static void fio_server_check_fork_items(struct flist_head *list)
+static void fio_server_check_fork_items(struct flist_head *list, bool stop)
 {
        struct flist_head *entry, *tmp;
        struct fio_fork_item *ffi;
@@ -520,34 +712,58 @@ static void fio_server_check_fork_items(struct flist_head *list)
                fio_server_check_fork_item(ffi);
 
                if (ffi->exited)
-                       fio_server_fork_item_done(ffi);
+                       fio_server_fork_item_done(ffi, stop);
        }
 }
 
-static void fio_server_check_jobs(void)
+static void fio_server_check_jobs(struct flist_head *job_list)
+{
+       fio_server_check_fork_items(job_list, true);
+}
+
+static void fio_server_check_conns(struct flist_head *conn_list)
 {
-       fio_server_check_fork_items(&job_list);
+       fio_server_check_fork_items(conn_list, false);
 }
 
-static void fio_server_check_conns(void)
+static int handle_load_file_cmd(struct fio_net_cmd *cmd)
 {
-       fio_server_check_fork_items(&conn_list);
+       struct cmd_load_file_pdu *pdu = (struct cmd_load_file_pdu *) cmd->payload;
+       void *file_name = pdu->file;
+       struct cmd_start_pdu spdu;
+
+       dprint(FD_NET, "server: loading local file %s\n", (char *) file_name);
+
+       pdu->name_len = le16_to_cpu(pdu->name_len);
+       pdu->client_type = le16_to_cpu(pdu->client_type);
+
+       if (parse_jobs_ini(file_name, 0, 0, pdu->client_type)) {
+               fio_net_queue_quit();
+               return -1;
+       }
+
+       spdu.jobs = cpu_to_le32(thread_number);
+       spdu.stat_outputs = cpu_to_le32(stat_number);
+       fio_net_queue_cmd(FIO_NET_CMD_START, &spdu, sizeof(spdu), NULL, SK_F_COPY);
+       return 0;
 }
 
-static int handle_run_cmd(struct fio_net_cmd *cmd)
+static int handle_run_cmd(struct sk_out *sk_out, struct flist_head *job_list,
+                         struct fio_net_cmd *cmd)
 {
        pid_t pid;
        int ret;
 
+       fio_time_init();
        set_genesis_time();
 
        pid = fork();
        if (pid) {
-               fio_server_add_job_pid(pid);
+               fio_server_add_job_pid(job_list, pid);
                return 0;
        }
 
-       ret = fio_backend();
+       ret = fio_backend(sk_out);
        free_threads_shm();
        _exit(ret);
 }
@@ -562,13 +778,14 @@ static int handle_job_cmd(struct fio_net_cmd *cmd)
        pdu->client_type = le32_to_cpu(pdu->client_type);
 
        if (parse_jobs_ini(buf, 1, 0, pdu->client_type)) {
-               fio_net_send_quit(server_fd);
+               fio_net_queue_quit();
                return -1;
        }
 
        spdu.jobs = cpu_to_le32(thread_number);
        spdu.stat_outputs = cpu_to_le32(stat_number);
-       fio_net_send_cmd(server_fd, FIO_NET_CMD_START, &spdu, sizeof(spdu), NULL, NULL);
+
+       fio_net_queue_cmd(FIO_NET_CMD_START, &spdu, sizeof(spdu), NULL, SK_F_COPY);
        return 0;
 }
 
@@ -599,7 +816,7 @@ static int handle_jobline_cmd(struct fio_net_cmd *cmd)
        }
 
        if (parse_cmd_line(clp->lines, argv, clp->client_type)) {
-               fio_net_send_quit(server_fd);
+               fio_net_queue_quit();
                free(argv);
                return -1;
        }
@@ -608,17 +825,21 @@ static int handle_jobline_cmd(struct fio_net_cmd *cmd)
 
        spdu.jobs = cpu_to_le32(thread_number);
        spdu.stat_outputs = cpu_to_le32(stat_number);
-       fio_net_send_cmd(server_fd, FIO_NET_CMD_START, &spdu, sizeof(spdu), NULL, NULL);
+
+       fio_net_queue_cmd(FIO_NET_CMD_START, &spdu, sizeof(spdu), NULL, SK_F_COPY);
        return 0;
 }
 
 static int handle_probe_cmd(struct fio_net_cmd *cmd)
 {
-       struct cmd_probe_pdu probe;
+       struct cmd_client_probe_pdu *pdu = (struct cmd_client_probe_pdu *) cmd->payload;
+       struct cmd_probe_reply_pdu probe;
        uint64_t tag = cmd->tag;
 
        dprint(FD_NET, "server: sending probe reply\n");
 
+       strcpy(me, (char *) pdu->server);
+
        memset(&probe, 0, sizeof(probe));
        gethostname((char *) probe.hostname, sizeof(probe.hostname));
 #ifdef CONFIG_BIG_ENDIAN
@@ -630,64 +851,72 @@ static int handle_probe_cmd(struct fio_net_cmd *cmd)
        probe.arch      = FIO_ARCH;
        probe.bpp       = sizeof(void *);
        probe.cpus      = __cpu_to_le32(cpus_online());
-       probe.flags     = 0;
 
-       return fio_net_send_cmd(server_fd, FIO_NET_CMD_PROBE, &probe, sizeof(probe), &tag, NULL);
+       /*
+        * If the client supports compression and we do too, then enable it
+        */
+       if (has_zlib && le64_to_cpu(pdu->flags) & FIO_PROBE_FLAG_ZLIB) {
+               probe.flags = __cpu_to_le64(FIO_PROBE_FLAG_ZLIB);
+               use_zlib = 1;
+       } else {
+               probe.flags = 0;
+               use_zlib = 0;
+       }
+
+       return fio_net_queue_cmd(FIO_NET_CMD_PROBE, &probe, sizeof(probe), &tag, SK_F_COPY);
 }
 
 static int handle_send_eta_cmd(struct fio_net_cmd *cmd)
 {
        struct jobs_eta *je;
-       size_t size;
        uint64_t tag = cmd->tag;
+       size_t size;
        int i;
 
-       if (!thread_number)
-               return 0;
+       dprint(FD_NET, "server sending status\n");
 
-       size = sizeof(*je) + thread_number * sizeof(char) + 1;
-       je = malloc(size);
-       memset(je, 0, size);
+       /*
+        * Fake ETA return if we don't have a local one, otherwise the client
+        * will end up timing out waiting for a response to the ETA request
+        */
+       je = get_jobs_eta(true, &size);
+       if (!je) {
+               size = sizeof(*je);
+               je = calloc(1, size);
+       } else {
+               je->nr_running          = cpu_to_le32(je->nr_running);
+               je->nr_ramp             = cpu_to_le32(je->nr_ramp);
+               je->nr_pending          = cpu_to_le32(je->nr_pending);
+               je->nr_setting_up       = cpu_to_le32(je->nr_setting_up);
+               je->files_open          = cpu_to_le32(je->files_open);
+
+               for (i = 0; i < DDIR_RWDIR_CNT; i++) {
+                       je->m_rate[i]   = cpu_to_le32(je->m_rate[i]);
+                       je->t_rate[i]   = cpu_to_le32(je->t_rate[i]);
+                       je->m_iops[i]   = cpu_to_le32(je->m_iops[i]);
+                       je->t_iops[i]   = cpu_to_le32(je->t_iops[i]);
+                       je->rate[i]     = cpu_to_le32(je->rate[i]);
+                       je->iops[i]     = cpu_to_le32(je->iops[i]);
+               }
 
-       if (!calc_thread_status(je, 1)) {
-               free(je);
-               return 0;
+               je->elapsed_sec         = cpu_to_le64(je->elapsed_sec);
+               je->eta_sec             = cpu_to_le64(je->eta_sec);
+               je->nr_threads          = cpu_to_le32(je->nr_threads);
+               je->is_pow2             = cpu_to_le32(je->is_pow2);
+               je->unit_base           = cpu_to_le32(je->unit_base);
        }
 
-       dprint(FD_NET, "server sending status\n");
-
-       je->nr_running          = cpu_to_le32(je->nr_running);
-       je->nr_ramp             = cpu_to_le32(je->nr_ramp);
-       je->nr_pending          = cpu_to_le32(je->nr_pending);
-       je->files_open          = cpu_to_le32(je->files_open);
-
-       for (i = 0; i < DDIR_RWDIR_CNT; i++) {
-               je->m_rate[i]   = cpu_to_le32(je->m_rate[i]);
-               je->t_rate[i]   = cpu_to_le32(je->t_rate[i]);
-               je->m_iops[i]   = cpu_to_le32(je->m_iops[i]);
-               je->t_iops[i]   = cpu_to_le32(je->t_iops[i]);
-               je->rate[i]     = cpu_to_le32(je->rate[i]);
-               je->iops[i]     = cpu_to_le32(je->iops[i]);
-       }
-
-       je->elapsed_sec         = cpu_to_le64(je->elapsed_sec);
-       je->eta_sec             = cpu_to_le64(je->eta_sec);
-       je->nr_threads          = cpu_to_le32(je->nr_threads);
-       je->is_pow2             = cpu_to_le32(je->is_pow2);
-       je->unit_base           = cpu_to_le32(je->unit_base);
-
-       fio_net_send_cmd(server_fd, FIO_NET_CMD_ETA, je, size, &tag, NULL);
-       free(je);
+       fio_net_queue_cmd(FIO_NET_CMD_ETA, je, size, &tag, SK_F_FREE);
        return 0;
 }
 
-static int send_update_job_reply(int fd, uint64_t __tag, int error)
+static int send_update_job_reply(uint64_t __tag, int error)
 {
        uint64_t tag = __tag;
        uint32_t pdu_error;
 
        pdu_error = __cpu_to_le32(error);
-       return fio_net_send_cmd(fd, FIO_NET_CMD_UPDATE_JOB, &pdu_error, sizeof(pdu_error), &tag, NULL);
+       return fio_net_queue_cmd(FIO_NET_CMD_UPDATE_JOB, &pdu_error, sizeof(pdu_error), &tag, SK_F_COPY);
 }
 
 static int handle_update_job_cmd(struct fio_net_cmd *cmd)
@@ -701,17 +930,41 @@ static int handle_update_job_cmd(struct fio_net_cmd *cmd)
        dprint(FD_NET, "server: updating options for job %u\n", tnumber);
 
        if (!tnumber || tnumber > thread_number) {
-               send_update_job_reply(server_fd, cmd->tag, ENODEV);
+               send_update_job_reply(cmd->tag, ENODEV);
                return 0;
        }
 
        td = &threads[tnumber - 1];
        convert_thread_options_to_cpu(&td->o, &pdu->top);
-       send_update_job_reply(server_fd, cmd->tag, 0);
+       send_update_job_reply(cmd->tag, 0);
        return 0;
 }
 
-static int handle_command(struct fio_net_cmd *cmd)
+static int handle_trigger_cmd(struct fio_net_cmd *cmd)
+{
+       struct cmd_vtrigger_pdu *pdu = (struct cmd_vtrigger_pdu *) cmd->payload;
+       char *buf = (char *) pdu->cmd;
+       struct all_io_list *rep;
+       size_t sz;
+
+       pdu->len = le16_to_cpu(pdu->len);
+       buf[pdu->len] = '\0';
+
+       rep = get_all_io_list(IO_LIST_ALL, &sz);
+       if (!rep) {
+               struct all_io_list state;
+
+               state.threads = cpu_to_le64((uint64_t) 0);
+               fio_net_queue_cmd(FIO_NET_CMD_VTRIGGER, &state, sizeof(state), NULL, SK_F_COPY);
+       } else
+               fio_net_queue_cmd(FIO_NET_CMD_VTRIGGER, rep, sz, NULL, SK_F_FREE);
+
+       exec_trigger(buf);
+       return 0;
+}
+
+static int handle_command(struct sk_out *sk_out, struct flist_head *job_list,
+                         struct fio_net_cmd *cmd)
 {
        int ret;
 
@@ -722,10 +975,14 @@ static int handle_command(struct fio_net_cmd *cmd)
        switch (cmd->opcode) {
        case FIO_NET_CMD_QUIT:
                fio_terminate_threads(TERMINATE_ALL);
-               return -1;
+               ret = 0;
+               break;
        case FIO_NET_CMD_EXIT:
                exit_backend = 1;
                return -1;
+       case FIO_NET_CMD_LOAD_FILE:
+               ret = handle_load_file_cmd(cmd);
+               break;
        case FIO_NET_CMD_JOB:
                ret = handle_job_cmd(cmd);
                break;
@@ -739,11 +996,40 @@ static int handle_command(struct fio_net_cmd *cmd)
                ret = handle_send_eta_cmd(cmd);
                break;
        case FIO_NET_CMD_RUN:
-               ret = handle_run_cmd(cmd);
+               ret = handle_run_cmd(sk_out, job_list, cmd);
                break;
        case FIO_NET_CMD_UPDATE_JOB:
                ret = handle_update_job_cmd(cmd);
                break;
+       case FIO_NET_CMD_VTRIGGER:
+               ret = handle_trigger_cmd(cmd);
+               break;
+       case FIO_NET_CMD_SENDFILE: {
+               struct cmd_sendfile_reply *in;
+               struct cmd_reply *rep;
+
+               rep = (struct cmd_reply *) (uintptr_t) cmd->tag;
+
+               in = (struct cmd_sendfile_reply *) cmd->payload;
+               in->size = le32_to_cpu(in->size);
+               in->error = le32_to_cpu(in->error);
+               if (in->error) {
+                       ret = 1;
+                       rep->error = in->error;
+               } else {
+                       ret = 0;
+                       rep->data = smalloc(in->size);
+                       if (!rep->data) {
+                               ret = 1;
+                               rep->error = ENOMEM;
+                       } else {
+                               rep->size = in->size;
+                               memcpy(rep->data, in->data, in->size);
+                       }
+               }
+               fio_mutex_up(&rep->lock);
+               break;
+               }
        default:
                log_err("fio: unknown opcode: %s\n", fio_server_op(cmd->opcode));
                ret = 1;
@@ -752,19 +1038,139 @@ static int handle_command(struct fio_net_cmd *cmd)
        return ret;
 }
 
-static int handle_connection(int sk)
+/*
+ * Send a command with a separate PDU, not inlined in the command
+ */
+static int fio_send_cmd_ext_pdu(int sk, uint16_t opcode, const void *buf,
+                               off_t size, uint64_t tag, uint32_t flags)
+{
+       struct fio_net_cmd cmd;
+       struct iovec iov[2];
+
+       iov[0].iov_base = (void *) &cmd;
+       iov[0].iov_len = sizeof(cmd);
+       iov[1].iov_base = (void *) buf;
+       iov[1].iov_len = size;
+
+       __fio_init_net_cmd(&cmd, opcode, size, tag);
+       cmd.flags = __cpu_to_le32(flags);
+       fio_net_cmd_crc_pdu(&cmd, buf);
+
+       return fio_sendv_data(sk, iov, 2);
+}
+
+static void finish_entry(struct sk_entry *entry)
+{
+       if (entry->flags & SK_F_FREE)
+               free(entry->buf);
+       else if (entry->flags & SK_F_COPY)
+               sfree(entry->buf);
+
+       sfree(entry);
+}
+
+static void entry_set_flags_tag(struct sk_entry *entry, struct flist_head *list,
+                               unsigned int *flags, uint64_t *tag)
+{
+       if (!flist_empty(list))
+               *flags = FIO_NET_CMD_F_MORE;
+       else
+               *flags = 0;
+
+       if (entry->tagptr)
+               *tag = *entry->tagptr;
+       else
+               *tag = 0;
+}
+
+static int send_vec_entry(struct sk_out *sk_out, struct sk_entry *first)
+{
+       unsigned int flags;
+       uint64_t tag;
+       int ret;
+
+       entry_set_flags_tag(first, &first->next, &flags, &tag);
+
+       ret = fio_send_cmd_ext_pdu(sk_out->sk, first->opcode, first->buf, first->size, tag, flags);
+
+       while (!flist_empty(&first->next)) {
+               struct sk_entry *next;
+
+               next = flist_first_entry(&first->next, struct sk_entry, list);
+               flist_del_init(&next->list);
+
+               entry_set_flags_tag(next, &first->next, &flags, &tag);
+
+               ret += fio_send_cmd_ext_pdu(sk_out->sk, next->opcode, next->buf, next->size, tag, flags);
+               finish_entry(next);
+       }
+
+       return ret;
+}
+
+static int handle_sk_entry(struct sk_out *sk_out, struct sk_entry *entry)
+{
+       int ret;
+
+       fio_mutex_down(&sk_out->xmit);
+
+       if (entry->flags & SK_F_VEC)
+               ret = send_vec_entry(sk_out, entry);
+       else if (entry->flags & SK_F_SIMPLE) {
+               uint64_t tag = 0;
+
+               if (entry->tagptr)
+                       tag = *entry->tagptr;
+
+               ret = fio_net_send_simple_cmd(sk_out->sk, entry->opcode, tag, NULL);
+       } else
+               ret = fio_net_send_cmd(sk_out->sk, entry->opcode, entry->buf, entry->size, entry->tagptr, NULL);
+
+       fio_mutex_up(&sk_out->xmit);
+
+       if (ret)
+               log_err("fio: failed handling cmd %s\n", fio_server_op(entry->opcode));
+
+       finish_entry(entry);
+       return ret;
+}
+
+static int handle_xmits(struct sk_out *sk_out)
+{
+       struct sk_entry *entry;
+       FLIST_HEAD(list);
+       int ret = 0;
+
+       sk_lock(sk_out);
+       if (flist_empty(&sk_out->list)) {
+               sk_unlock(sk_out);
+               return 0;
+       }
+
+       flist_splice_init(&sk_out->list, &list);
+       sk_unlock(sk_out);
+
+       while (!flist_empty(&list)) {
+               entry = flist_entry(list.next, struct sk_entry, list);
+               flist_del(&entry->list);
+               ret += handle_sk_entry(sk_out, entry);
+       }
+
+       return ret;
+}
+
+static int handle_connection(struct sk_out *sk_out)
 {
        struct fio_net_cmd *cmd = NULL;
+       FLIST_HEAD(job_list);
        int ret = 0;
 
        reset_fio_state();
-       INIT_FLIST_HEAD(&job_list);
-       server_fd = sk;
 
        /* read forever */
        while (!exit_backend) {
                struct pollfd pfd = {
-                       .fd     = sk,
+                       .fd     = sk_out->sk,
                        .events = POLLIN,
                };
 
@@ -775,14 +1181,17 @@ static int handle_connection(int sk)
                        if (!flist_empty(&job_list))
                                timeout = 100;
 
-                       ret = poll(&pfd, 1, timeout);
+                       handle_xmits(sk_out);
+
+                       ret = poll(&pfd, 1, 0);
                        if (ret < 0) {
                                if (errno == EINTR)
                                        break;
                                log_err("fio: poll: %s\n", strerror(errno));
                                break;
                        } else if (!ret) {
-                               fio_server_check_jobs();
+                               fio_server_check_jobs(&job_list);
+                               fio_mutex_down_timeout(&sk_out->wait, timeout);
                                continue;
                        }
 
@@ -794,18 +1203,18 @@ static int handle_connection(int sk)
                        }
                } while (!exit_backend);
 
-               fio_server_check_jobs();
+               fio_server_check_jobs(&job_list);
 
                if (ret < 0)
                        break;
 
-               cmd = fio_net_recv_cmd(sk);
+               cmd = fio_net_recv_cmd(sk_out->sk);
                if (!cmd) {
                        ret = -1;
                        break;
                }
 
-               ret = handle_command(cmd);
+               ret = handle_command(sk_out, &job_list, cmd);
                if (ret)
                        break;
 
@@ -816,24 +1225,68 @@ static int handle_connection(int sk)
        if (cmd)
                free(cmd);
 
-       close(sk);
+       handle_xmits(sk_out);
+
+       close(sk_out->sk);
+       sk_out->sk = -1;
+       __sk_out_drop(sk_out);
        _exit(ret);
 }
 
+/* get the address on this host bound by the input socket, 
+ * whether it is ipv6 or ipv4 */
+
+static int get_my_addr_str(int sk)
+{
+       struct sockaddr_in6 myaddr6 = { 0, };
+       struct sockaddr_in myaddr4 = { 0, };
+       struct sockaddr *sockaddr_p;
+       char *net_addr;
+       socklen_t len;
+       int ret;
+
+       if (use_ipv6) {
+               len = sizeof(myaddr6);
+               sockaddr_p = (struct sockaddr * )&myaddr6;
+               net_addr = (char * )&myaddr6.sin6_addr;
+       } else {
+               len = sizeof(myaddr4);
+               sockaddr_p = (struct sockaddr * )&myaddr4;
+               net_addr = (char * )&myaddr4.sin_addr;
+       }
+
+       ret = getsockname(sk, sockaddr_p, &len);
+       if (ret) {
+               log_err("fio: getsockaddr: %s\n", strerror(errno));
+               return -1;
+       }
+
+       if (!inet_ntop(use_ipv6?AF_INET6:AF_INET, net_addr, client_sockaddr_str, INET6_ADDRSTRLEN - 1)) {
+               log_err("inet_ntop: failed to convert addr to string\n");
+               return -1;
+       }
+
+       dprint(FD_NET, "fio server bound to addr %s\n", client_sockaddr_str);
+       return 0;
+}
+
 static int accept_loop(int listen_sk)
 {
        struct sockaddr_in addr;
-       socklen_t len = sizeof(addr);
+       struct sockaddr_in6 addr6;
+       socklen_t len = use_ipv6 ? sizeof(addr6) : sizeof(addr);
        struct pollfd pfd;
-       int ret = 0, sk, flags, exitval = 0;
+       int ret = 0, sk, exitval = 0;
+       FLIST_HEAD(conn_list);
 
        dprint(FD_NET, "server enter accept loop\n");
 
-       flags = fcntl(listen_sk, F_GETFL);
-       flags |= O_NONBLOCK;
-       fcntl(listen_sk, F_SETFL, flags);
+       fio_set_fd_nonblocking(listen_sk, "server");
 
        while (!exit_backend) {
+               struct sk_out *sk_out;
+               const char *from;
+               char buf[64];
                pid_t pid;
 
                pfd.fd = listen_sk;
@@ -851,7 +1304,7 @@ static int accept_loop(int listen_sk)
                                log_err("fio: poll: %s\n", strerror(errno));
                                break;
                        } else if (!ret) {
-                               fio_server_check_conns();
+                               fio_server_check_conns(&conn_list);
                                continue;
                        }
 
@@ -859,28 +1312,51 @@ static int accept_loop(int listen_sk)
                                break;
                } while (!exit_backend);
 
-               fio_server_check_conns();
+               fio_server_check_conns(&conn_list);
 
                if (exit_backend || ret < 0)
                        break;
 
-               sk = accept(listen_sk, (struct sockaddr *) &addr, &len);
+               if (use_ipv6)
+                       sk = accept(listen_sk, (struct sockaddr *) &addr6, &len);
+               else
+                       sk = accept(listen_sk, (struct sockaddr *) &addr, &len);
+
                if (sk < 0) {
                        log_err("fio: accept: %s\n", strerror(errno));
                        return -1;
                }
 
-               dprint(FD_NET, "server: connect from %s\n", inet_ntoa(addr.sin_addr));
+               if (use_ipv6)
+                       from = inet_ntop(AF_INET6, (struct sockaddr *) &addr6.sin6_addr, buf, sizeof(buf));
+               else
+                       from = inet_ntop(AF_INET, (struct sockaddr *) &addr.sin_addr, buf, sizeof(buf));
+
+               dprint(FD_NET, "server: connect from %s\n", from);
+
+               sk_out = smalloc(sizeof(*sk_out));
+               sk_out->sk = sk;
+               INIT_FLIST_HEAD(&sk_out->list);
+               __fio_mutex_init(&sk_out->lock, FIO_MUTEX_UNLOCKED);
+               __fio_mutex_init(&sk_out->wait, FIO_MUTEX_LOCKED);
+               __fio_mutex_init(&sk_out->xmit, FIO_MUTEX_UNLOCKED);
 
                pid = fork();
                if (pid) {
                        close(sk);
-                       fio_server_add_conn_pid(pid);
+                       fio_server_add_conn_pid(&conn_list, pid);
                        continue;
                }
 
-               /* exits */
-               handle_connection(sk);
+               /* if error, it's already logged, non-fatal */
+               get_my_addr_str(sk);
+
+               /*
+                * Assign sk_out here, it'll be dropped in handle_connection()
+                * since that function calls _exit() when done
+                */
+               sk_out_assign(sk_out);
+               handle_connection(sk_out);
        }
 
        return exitval;
@@ -888,12 +1364,13 @@ static int accept_loop(int listen_sk)
 
 int fio_server_text_output(int level, const char *buf, size_t len)
 {
+       struct sk_out *sk_out = pthread_getspecific(sk_out_key);
        struct cmd_text_pdu *pdu;
        unsigned int tlen;
        struct timeval tv;
 
-       if (server_fd == -1)
-               return log_local_buf(buf, len);
+       if (!sk_out || sk_out->sk == -1)
+               return -1;
 
        tlen = sizeof(*pdu) + len;
        pdu = malloc(tlen);
@@ -907,7 +1384,7 @@ int fio_server_text_output(int level, const char *buf, size_t len)
 
        memcpy(pdu->buf, buf, len);
 
-       fio_net_send_cmd(server_fd, FIO_NET_CMD_TEXT, pdu, tlen, NULL, NULL);
+       fio_net_queue_cmd(FIO_NET_CMD_TEXT, pdu, tlen, NULL, SK_F_COPY);
        free(pdu);
        return len;
 }
@@ -921,8 +1398,8 @@ static void convert_io_stat(struct io_stat *dst, struct io_stat *src)
        /*
         * Encode to IEEE 754 for network transfer
         */
-       dst->mean.u.i   = __cpu_to_le64(fio_double_to_uint64(src->mean.u.f));
-       dst->S.u.i      = __cpu_to_le64(fio_double_to_uint64(src->S.u.f));
+       dst->mean.u.i   = cpu_to_le64(fio_double_to_uint64(src->mean.u.f));
+       dst->S.u.i      = cpu_to_le64(fio_double_to_uint64(src->S.u.f));
 }
 
 static void convert_gs(struct group_run_stats *dst, struct group_run_stats *src)
@@ -957,9 +1434,9 @@ void fio_server_send_ts(struct thread_stat *ts, struct group_run_stats *rs)
 
        memset(&p, 0, sizeof(p));
 
-       strcpy(p.ts.name, ts->name);
-       strcpy(p.ts.verror, ts->verror);
-       strcpy(p.ts.description, ts->description);
+       strncpy(p.ts.name, ts->name, FIO_JOBNAME_SIZE - 1);
+       strncpy(p.ts.verror, ts->verror, FIO_VERROR_SIZE - 1);
+       strncpy(p.ts.description, ts->description, FIO_JOBDESC_SIZE - 1);
 
        p.ts.error              = cpu_to_le32(ts->error);
        p.ts.thread_number      = cpu_to_le32(ts->thread_number);
@@ -981,12 +1458,13 @@ void fio_server_send_ts(struct thread_stat *ts, struct group_run_stats *rs)
        p.ts.minf               = cpu_to_le64(ts->minf);
        p.ts.majf               = cpu_to_le64(ts->majf);
        p.ts.clat_percentiles   = cpu_to_le64(ts->clat_percentiles);
+       p.ts.percentile_precision = cpu_to_le64(ts->percentile_precision);
 
        for (i = 0; i < FIO_IO_U_LIST_MAX_LEN; i++) {
                fio_fp64_t *src = &ts->percentile_list[i];
                fio_fp64_t *dst = &p.ts.percentile_list[i];
 
-               dst->u.i = __cpu_to_le64(fio_double_to_uint64(src->u.f));
+               dst->u.i = cpu_to_le64(fio_double_to_uint64(src->u.f));
        }
 
        for (i = 0; i < FIO_IO_U_MAP_NR; i++) {
@@ -1007,6 +1485,7 @@ void fio_server_send_ts(struct thread_stat *ts, struct group_run_stats *rs)
        for (i = 0; i < DDIR_RWDIR_CNT; i++) {
                p.ts.total_io_u[i]      = cpu_to_le64(ts->total_io_u[i]);
                p.ts.short_io_u[i]      = cpu_to_le64(ts->short_io_u[i]);
+               p.ts.drop_io_u[i]       = cpu_to_le64(ts->drop_io_u[i]);
        }
 
        p.ts.total_submit       = cpu_to_le64(ts->total_submit);
@@ -1024,9 +1503,18 @@ void fio_server_send_ts(struct thread_stat *ts, struct group_run_stats *rs)
        p.ts.kb_base            = cpu_to_le32(ts->kb_base);
        p.ts.unit_base          = cpu_to_le32(ts->unit_base);
 
+       p.ts.latency_depth      = cpu_to_le32(ts->latency_depth);
+       p.ts.latency_target     = cpu_to_le64(ts->latency_target);
+       p.ts.latency_window     = cpu_to_le64(ts->latency_window);
+       p.ts.latency_percentile.u.i = cpu_to_le64(fio_double_to_uint64(ts->latency_percentile.u.f));
+
+       p.ts.nr_block_infos     = le64_to_cpu(ts->nr_block_infos);
+       for (i = 0; i < p.ts.nr_block_infos; i++)
+               p.ts.block_infos[i] = le32_to_cpu(ts->block_infos[i]);
+
        convert_gs(&p.rs, rs);
 
-       fio_net_send_cmd(server_fd, FIO_NET_CMD_TS, &p, sizeof(p), NULL, NULL);
+       fio_net_queue_cmd(FIO_NET_CMD_TS, &p, sizeof(p), NULL, SK_F_COPY);
 }
 
 void fio_server_send_gs(struct group_run_stats *rs)
@@ -1036,7 +1524,48 @@ void fio_server_send_gs(struct group_run_stats *rs)
        dprint(FD_NET, "server sending group run stats\n");
 
        convert_gs(&gs, rs);
-       fio_net_send_cmd(server_fd, FIO_NET_CMD_GS, &gs, sizeof(gs), NULL, NULL);
+       fio_net_queue_cmd(FIO_NET_CMD_GS, &gs, sizeof(gs), NULL, SK_F_COPY);
+}
+
+void fio_server_send_job_options(struct flist_head *opt_list,
+                                unsigned int groupid)
+{
+       struct cmd_job_option pdu;
+       struct flist_head *entry;
+
+       if (flist_empty(opt_list))
+               return;
+
+       flist_for_each(entry, opt_list) {
+               struct print_option *p;
+               size_t len;
+
+               p = flist_entry(entry, struct print_option, list);
+               memset(&pdu, 0, sizeof(pdu));
+
+               if (groupid == -1U) {
+                       pdu.global = __cpu_to_le16(1);
+                       pdu.groupid = 0;
+               } else {
+                       pdu.global = 0;
+                       pdu.groupid = cpu_to_le32(groupid);
+               }
+               len = strlen(p->name);
+               if (len >= sizeof(pdu.name)) {
+                       len = sizeof(pdu.name) - 1;
+                       pdu.truncated = __cpu_to_le16(1);
+               }
+               memcpy(pdu.name, p->name, len);
+               if (p->value) {
+                       len = strlen(p->value);
+                       if (len >= sizeof(pdu.value)) {
+                               len = sizeof(pdu.value) - 1;
+                               pdu.truncated = __cpu_to_le16(1);
+                       }
+                       memcpy(pdu.value, p->value, len);
+               }
+               fio_net_queue_cmd(FIO_NET_CMD_JOB_OPT, &pdu, sizeof(pdu), NULL, SK_F_COPY);
+       }
 }
 
 static void convert_agg(struct disk_util_agg *dst, struct disk_util_agg *src)
@@ -1044,34 +1573,35 @@ static void convert_agg(struct disk_util_agg *dst, struct disk_util_agg *src)
        int i;
 
        for (i = 0; i < 2; i++) {
-               dst->ios[i]     = cpu_to_le32(src->ios[i]);
-               dst->merges[i]  = cpu_to_le32(src->merges[i]);
+               dst->ios[i]     = cpu_to_le64(src->ios[i]);
+               dst->merges[i]  = cpu_to_le64(src->merges[i]);
                dst->sectors[i] = cpu_to_le64(src->sectors[i]);
-               dst->ticks[i]   = cpu_to_le32(src->ticks[i]);
+               dst->ticks[i]   = cpu_to_le64(src->ticks[i]);
        }
 
-       dst->io_ticks           = cpu_to_le32(src->io_ticks);
-       dst->time_in_queue      = cpu_to_le32(src->time_in_queue);
+       dst->io_ticks           = cpu_to_le64(src->io_ticks);
+       dst->time_in_queue      = cpu_to_le64(src->time_in_queue);
        dst->slavecount         = cpu_to_le32(src->slavecount);
-       dst->max_util.u.i       = __cpu_to_le64(fio_double_to_uint64(src->max_util.u.f));
+       dst->max_util.u.i       = cpu_to_le64(fio_double_to_uint64(src->max_util.u.f));
 }
 
 static void convert_dus(struct disk_util_stat *dst, struct disk_util_stat *src)
 {
        int i;
 
-       strcpy((char *) dst->name, (char *) src->name);
+       dst->name[FIO_DU_NAME_SZ - 1] = '\0';
+       strncpy((char *) dst->name, (char *) src->name, FIO_DU_NAME_SZ - 1);
 
        for (i = 0; i < 2; i++) {
-               dst->ios[i]     = cpu_to_le32(src->ios[i]);
-               dst->merges[i]  = cpu_to_le32(src->merges[i]);
-               dst->sectors[i] = cpu_to_le64(src->sectors[i]);
-               dst->ticks[i]   = cpu_to_le32(src->ticks[i]);
+               dst->s.ios[i]           = cpu_to_le64(src->s.ios[i]);
+               dst->s.merges[i]        = cpu_to_le64(src->s.merges[i]);
+               dst->s.sectors[i]       = cpu_to_le64(src->s.sectors[i]);
+               dst->s.ticks[i]         = cpu_to_le64(src->s.ticks[i]);
        }
 
-       dst->io_ticks           = cpu_to_le32(src->io_ticks);
-       dst->time_in_queue      = cpu_to_le32(src->time_in_queue);
-       dst->msec               = cpu_to_le64(src->msec);
+       dst->s.io_ticks         = cpu_to_le64(src->s.io_ticks);
+       dst->s.time_in_queue    = cpu_to_le64(src->s.time_in_queue);
+       dst->s.msec             = cpu_to_le64(src->s.msec);
 }
 
 void fio_server_send_du(void)
@@ -1090,51 +1620,17 @@ void fio_server_send_du(void)
                convert_dus(&pdu.dus, &du->dus);
                convert_agg(&pdu.agg, &du->agg);
 
-               fio_net_send_cmd(server_fd, FIO_NET_CMD_DU, &pdu, sizeof(pdu), NULL, NULL);
+               fio_net_queue_cmd(FIO_NET_CMD_DU, &pdu, sizeof(pdu), NULL, SK_F_COPY);
        }
 }
 
-/*
- * Send a command with a separate PDU, not inlined in the command
- */
-static int fio_send_cmd_ext_pdu(int sk, uint16_t opcode, const void *buf,
-                               off_t size, uint64_t tag, uint32_t flags)
+static int fio_send_iolog_gz(struct sk_entry *first, struct io_log *log)
 {
-       struct fio_net_cmd cmd;
-       struct iovec iov[2];
-
-       iov[0].iov_base = &cmd;
-       iov[0].iov_len = sizeof(cmd);
-       iov[1].iov_base = (void *) buf;
-       iov[1].iov_len = size;
-
-       __fio_init_net_cmd(&cmd, opcode, size, tag);
-       cmd.flags = __cpu_to_le32(flags);
-       fio_net_cmd_crc_pdu(&cmd, buf);
-
-       return fio_sendv_data(sk, iov, 2);
-}
-
-int fio_send_iolog(struct thread_data *td, struct io_log *log, const char *name)
-{
-       struct cmd_iolog_pdu pdu;
+       int ret = 0;
+#ifdef CONFIG_ZLIB
+       struct sk_entry *entry;
        z_stream stream;
        void *out_pdu;
-       int i, ret = 0;
-
-       pdu.thread_number = cpu_to_le32(td->thread_number);
-       pdu.nr_samples = __cpu_to_le32(log->nr_samples);
-       pdu.log_type = cpu_to_le32(log->log_type);
-       strcpy((char *) pdu.name, name);
-
-       for (i = 0; i < log->nr_samples; i++) {
-               struct io_sample *s = &log->log[i];
-
-               s->time = cpu_to_le64(s->time);
-               s->val  = cpu_to_le64(s->val);
-               s->ddir = cpu_to_le32(s->ddir);
-               s->bs   = cpu_to_le32(s->bs);
-       }
 
        /*
         * Dirty - since the log is potentially huge, compress it into
@@ -1152,20 +1648,11 @@ int fio_send_iolog(struct thread_data *td, struct io_log *log, const char *name)
                goto err;
        }
 
-       /*
-        * Send header first, it's not compressed.
-        */
-       ret = fio_send_cmd_ext_pdu(server_fd, FIO_NET_CMD_IOLOG, &pdu,
-                                       sizeof(pdu), 0, FIO_NET_CMD_F_MORE);
-       if (ret)
-               goto err_zlib;
-
        stream.next_in = (void *) log->log;
-       stream.avail_in = log->nr_samples * sizeof(struct io_sample);
+       stream.avail_in = log->nr_samples * log_entry_sz(log);
 
        do {
-               unsigned int this_len, flags = 0;
-               int ret;
+               unsigned int this_len;
 
                stream.avail_out = FIO_SERVER_MAX_FRAGMENT_PDU;
                stream.next_out = out_pdu;
@@ -1176,19 +1663,101 @@ int fio_send_iolog(struct thread_data *td, struct io_log *log, const char *name)
 
                this_len = FIO_SERVER_MAX_FRAGMENT_PDU - stream.avail_out;
 
-               if (stream.avail_in)
-                       flags = FIO_NET_CMD_F_MORE;
-
-               ret = fio_send_cmd_ext_pdu(server_fd, FIO_NET_CMD_IOLOG,
-                                          out_pdu, this_len, 0, flags);
-               if (ret)
-                       goto err_zlib;
+               entry = fio_net_prep_cmd(FIO_NET_CMD_IOLOG, out_pdu, this_len,
+                                               NULL, SK_F_VEC | SK_F_INLINE | SK_F_FREE);
+               out_pdu = NULL;
+               flist_add_tail(&entry->list, &first->next);
        } while (stream.avail_in);
 
 err_zlib:
        deflateEnd(&stream);
 err:
        free(out_pdu);
+#endif
+       return ret;
+}
+
+static int fio_send_gz_chunks(struct sk_entry *first, struct io_log *log)
+{
+       struct sk_entry *entry;
+       struct flist_head *node;
+
+       pthread_mutex_lock(&log->chunk_lock);
+       flist_for_each(node, &log->chunk_list) {
+               struct iolog_compress *c;
+
+               c = flist_entry(node, struct iolog_compress, list);
+               entry = fio_net_prep_cmd(FIO_NET_CMD_IOLOG, c->buf, c->len,
+                                               NULL, SK_F_VEC | SK_F_INLINE);
+               flist_add_tail(&entry->list, &first->next);
+       }
+       pthread_mutex_unlock(&log->chunk_lock);
+
+       return 0;
+}
+
+int fio_send_iolog(struct thread_data *td, struct io_log *log, const char *name)
+{
+       struct cmd_iolog_pdu pdu;
+       struct sk_entry *first;
+       int i, ret = 0;
+
+       pdu.nr_samples = cpu_to_le64(log->nr_samples);
+       pdu.thread_number = cpu_to_le32(td->thread_number);
+       pdu.log_type = cpu_to_le32(log->log_type);
+
+       if (!flist_empty(&log->chunk_list))
+               pdu.compressed = __cpu_to_le32(STORE_COMPRESSED);
+       else if (use_zlib)
+               pdu.compressed = __cpu_to_le32(XMIT_COMPRESSED);
+       else
+               pdu.compressed = 0;
+
+       strncpy((char *) pdu.name, name, FIO_NET_NAME_MAX);
+       pdu.name[FIO_NET_NAME_MAX - 1] = '\0';
+
+       /*
+        * We can't do this for a pre-compressed log, but for that case,
+        * log->nr_samples is zero anyway.
+        */
+       for (i = 0; i < log->nr_samples; i++) {
+               struct io_sample *s = get_sample(log, i);
+
+               s->time         = cpu_to_le64(s->time);
+               s->val          = cpu_to_le64(s->val);
+               s->__ddir       = cpu_to_le32(s->__ddir);
+               s->bs           = cpu_to_le32(s->bs);
+
+               if (log->log_offset) {
+                       struct io_sample_offset *so = (void *) s;
+
+                       so->offset = cpu_to_le64(so->offset);
+               }
+       }
+
+       /*
+        * Assemble header entry first
+        */
+       first = fio_net_prep_cmd(FIO_NET_CMD_IOLOG, &pdu, sizeof(pdu), NULL, SK_F_VEC | SK_F_INLINE | SK_F_COPY);
+
+       /*
+        * Now append actual log entries. Compress if we can, otherwise just
+        * plain text output.
+        */
+       if (!flist_empty(&log->chunk_list))
+               ret = fio_send_gz_chunks(first, log);
+       else if (use_zlib)
+               ret = fio_send_iolog_gz(first, log);
+       else {
+               struct sk_entry *entry;
+               size_t size = log->nr_samples * log_entry_sz(log);
+
+               entry = fio_net_prep_cmd(FIO_NET_CMD_IOLOG, log->log, size,
+                                               NULL, SK_F_VEC | SK_F_INLINE);
+               flist_add_tail(&entry->list, &first->next);
+       }
+
+       fio_net_queue_entry(first);
        return ret;
 }
 
@@ -1201,20 +1770,89 @@ void fio_server_send_add_job(struct thread_data *td)
        pdu.groupid = cpu_to_le32(td->groupid);
        convert_thread_options_to_net(&pdu.top, &td->o);
 
-       fio_net_send_cmd(server_fd, FIO_NET_CMD_ADD_JOB, &pdu, sizeof(pdu), NULL, NULL);
+       fio_net_queue_cmd(FIO_NET_CMD_ADD_JOB, &pdu, sizeof(pdu), NULL, SK_F_COPY);
 }
 
 void fio_server_send_start(struct thread_data *td)
 {
-       assert(server_fd != -1);
+       struct sk_out *sk_out = pthread_getspecific(sk_out_key);
+
+       assert(sk_out->sk != -1);
+
+       fio_net_queue_cmd(FIO_NET_CMD_SERVER_START, NULL, 0, 0, SK_F_SIMPLE);
+}
+
+int fio_server_get_verify_state(const char *name, int threadnumber,
+                               void **datap, int *version)
+{
+       struct thread_io_list *s;
+       struct cmd_sendfile out;
+       struct cmd_reply *rep;
+       uint64_t tag;
+       void *data;
 
-       fio_net_send_simple_cmd(server_fd, FIO_NET_CMD_SERVER_START, 0, NULL);
+       dprint(FD_NET, "server: request verify state\n");
+
+       rep = smalloc(sizeof(*rep));
+       if (!rep) {
+               log_err("fio: smalloc pool too small\n");
+               return 1;
+       }
+
+       __fio_mutex_init(&rep->lock, FIO_MUTEX_LOCKED);
+       rep->data = NULL;
+       rep->error = 0;
+
+       verify_state_gen_name((char *) out.path, sizeof(out.path), name, me,
+                               threadnumber);
+       tag = (uint64_t) (uintptr_t) rep;
+       fio_net_queue_cmd(FIO_NET_CMD_SENDFILE, &out, sizeof(out), &tag, SK_F_COPY);
+
+       /*
+        * Wait for the backend to receive the reply
+        */
+       if (fio_mutex_down_timeout(&rep->lock, 10000)) {
+               log_err("fio: timed out waiting for reply\n");
+               goto fail;
+       }
+
+       if (rep->error) {
+               log_err("fio: failure on receiving state file: %s\n", strerror(rep->error));
+fail:
+               *datap = NULL;
+               sfree(rep);
+               fio_net_queue_quit();
+               return 1;
+       }
+
+       /*
+        * The format is verify_state_hdr, then thread_io_list. Verify
+        * the header, and the thread_io_list checksum
+        */
+       s = rep->data + sizeof(struct verify_state_hdr);
+       if (verify_state_hdr(rep->data, s, version))
+               goto fail;
+
+       /*
+        * Don't need the header from now, copy just the thread_io_list
+        */
+       rep->size -= sizeof(struct verify_state_hdr);
+       data = malloc(rep->size);
+       memcpy(data, s, rep->size);
+       *datap = data;
+
+       sfree(rep->data);
+       __fio_mutex_remove(&rep->lock);
+       sfree(rep);
+       return 0;
 }
 
 static int fio_init_server_ip(void)
 {
        struct sockaddr *addr;
        socklen_t socklen;
+       char buf[80];
+       const char *str;
        int sk, opt;
 
        if (use_ipv6)
@@ -1229,30 +1867,37 @@ static int fio_init_server_ip(void)
 
        opt = 1;
        if (setsockopt(sk, SOL_SOCKET, SO_REUSEADDR, (void *)&opt, sizeof(opt)) < 0) {
-               log_err("fio: setsockopt: %s\n", strerror(errno));
+               log_err("fio: setsockopt(REUSEADDR): %s\n", strerror(errno));
                close(sk);
                return -1;
        }
 #ifdef SO_REUSEPORT
        if (setsockopt(sk, SOL_SOCKET, SO_REUSEPORT, &opt, sizeof(opt)) < 0) {
-               log_err("fio: setsockopt: %s\n", strerror(errno));
+               log_err("fio: setsockopt(REUSEPORT): %s\n", strerror(errno));
                close(sk);
                return -1;
        }
 #endif
 
        if (use_ipv6) {
+               const void *src = &saddr_in6.sin6_addr;
+
                addr = (struct sockaddr *) &saddr_in6;
                socklen = sizeof(saddr_in6);
                saddr_in6.sin6_family = AF_INET6;
+               str = inet_ntop(AF_INET6, src, buf, sizeof(buf));
        } else {
+               const void *src = &saddr_in.sin_addr;
+
                addr = (struct sockaddr *) &saddr_in;
                socklen = sizeof(saddr_in);
                saddr_in.sin_family = AF_INET;
+               str = inet_ntop(AF_INET, src, buf, sizeof(buf));
        }
 
        if (bind(sk, addr, socklen) < 0) {
                log_err("fio: bind: %s\n", strerror(errno));
+               log_info("fio: failed with IPv%c %s\n", use_ipv6 ? '6' : '4', str);
                close(sk);
                return -1;
        }
@@ -1277,8 +1922,7 @@ static int fio_init_server_sock(void)
 
        memset(&addr, 0, sizeof(addr));
        addr.sun_family = AF_UNIX;
-       strcpy(addr.sun_path, bind_sock);
-       unlink(bind_sock);
+       strncpy(addr.sun_path, bind_sock, sizeof(addr.sun_path) - 1);
 
        len = sizeof(addr.sun_family) + strlen(bind_sock) + 1;
 
@@ -1307,6 +1951,8 @@ static int fio_init_server_connection(void)
        if (sk < 0)
                return sk;
 
+       memset(bind_str, 0, sizeof(bind_str));
+
        if (!bind_sock) {
                char *p, port[16];
                const void *src;
@@ -1326,55 +1972,53 @@ static int fio_init_server_connection(void)
                if (p)
                        strcat(p, port);
                else
-                       strcpy(bind_str, port);
+                       strncpy(bind_str, port, sizeof(bind_str) - 1);
        } else
-               strcpy(bind_str, bind_sock);
+               strncpy(bind_str, bind_sock, sizeof(bind_str) - 1);
 
        log_info("fio: server listening on %s\n", bind_str);
 
-       if (listen(sk, 0) < 0) {
+       if (listen(sk, 4) < 0) {
                log_err("fio: listen: %s\n", strerror(errno));
+               close(sk);
                return -1;
        }
 
        return sk;
 }
 
-int fio_server_parse_host(const char *host, int *ipv6, struct in_addr *inp,
+int fio_server_parse_host(const char *host, int ipv6, struct in_addr *inp,
                          struct in6_addr *inp6)
 
 {
        int ret = 0;
 
-       if (*ipv6)
+       if (ipv6)
                ret = inet_pton(AF_INET6, host, inp6);
        else
                ret = inet_pton(AF_INET, host, inp);
 
        if (ret != 1) {
-               struct hostent *hent;
+               struct addrinfo hints, *res;
 
-               hent = gethostbyname(host);
-               if (!hent) {
-                       log_err("fio: failed to resolve <%s>\n", host);
-                       return 0;
-               }
+               memset(&hints, 0, sizeof(hints));
+               hints.ai_family = ipv6 ? AF_INET6 : AF_INET;
+               hints.ai_socktype = SOCK_STREAM;
 
-               if (*ipv6) {
-                       if (hent->h_addrtype != AF_INET6) {
-                               log_info("fio: falling back to IPv4\n");
-                               *ipv6 = 0;
-                       } else
-                               memcpy(inp6, hent->h_addr_list[0], 16);
-               }
-               if (!*ipv6) {
-                       if (hent->h_addrtype != AF_INET) {
-                               log_err("fio: lookup type mismatch\n");
-                               return 0;
-                       }
-                       memcpy(inp, hent->h_addr_list[0], 4);
+               ret = getaddrinfo(host, NULL, &hints, &res);
+               if (ret) {
+                       log_err("fio: failed to resolve <%s> (%s)\n", host,
+                                       gai_strerror(ret));
+                       return 1;
                }
+
+               if (ipv6)
+                       memcpy(inp6, &((struct sockaddr_in6 *) res->ai_addr)->sin6_addr, sizeof(*inp6));
+               else
+                       memcpy(inp, &((struct sockaddr_in *) res->ai_addr)->sin_addr, sizeof(*inp));
+
                ret = 1;
+               freeaddrinfo(res);
        }
 
        return !(ret == 1);
@@ -1427,7 +2071,7 @@ int fio_server_parse_string(const char *str, char **ptr, int *is_sock,
                host++;
                lport = atoi(host);
                if (!lport || lport > 65535) {
-                       log_err("fio: bad server port %u\n", port);
+                       log_err("fio: bad server port %u\n", lport);
                        return 1;
                }
                /* no hostname given, we are done */
@@ -1436,7 +2080,7 @@ int fio_server_parse_string(const char *str, char **ptr, int *is_sock,
        }
 
        /*
-        * If no port seen yet, check if there's a last ':' at the end
+        * If no port seen yet, check if there's a last ',' at the end
         */
        if (!lport) {
                portp = strchr(host, ',');
@@ -1445,7 +2089,7 @@ int fio_server_parse_string(const char *str, char **ptr, int *is_sock,
                        portp++;
                        lport = atoi(portp);
                        if (!lport || lport > 65535) {
-                               log_err("fio: bad server port %u\n", port);
+                               log_err("fio: bad server port %u\n", lport);
                                return 1;
                        }
                }
@@ -1459,7 +2103,7 @@ int fio_server_parse_string(const char *str, char **ptr, int *is_sock,
 
        *ptr = strdup(host);
 
-       if (fio_server_parse_host(*ptr, ipv6, inp, inp6)) {
+       if (fio_server_parse_host(*ptr, *ipv6, inp, inp6)) {
                free(*ptr);
                *ptr = NULL;
                return 1;
@@ -1509,10 +2153,33 @@ out:
        return ret;
 }
 
+static void sig_int(int sig)
+{
+       if (bind_sock)
+               unlink(bind_sock);
+}
+
+static void set_sig_handlers(void)
+{
+       struct sigaction act;
+
+       memset(&act, 0, sizeof(act));
+       act.sa_handler = sig_int;
+       act.sa_flags = SA_RESTART;
+       sigaction(SIGINT, &act, NULL);
+}
+
 static int fio_server(void)
 {
        int sk, ret;
 
+       if (pthread_key_create(&sk_out_key, NULL)) {
+               log_err("fio: can't create sk_out backend key\n");
+               return -1;
+       }
+
+       pthread_setspecific(sk_out_key, NULL);
+
        dprint(FD_NET, "starting server\n");
 
        if (fio_handle_server_arg())
@@ -1522,6 +2189,8 @@ static int fio_server(void)
        if (sk < 0)
                return -1;
 
+       set_sig_handlers();
+
        ret = accept_loop(sk);
 
        close(sk);
@@ -1538,8 +2207,12 @@ static int fio_server(void)
 
 void fio_server_got_signal(int signal)
 {
+       struct sk_out *sk_out = pthread_getspecific(sk_out_key);
+
+       assert(sk_out);
+
        if (signal == SIGPIPE)
-               server_fd = -1;
+               sk_out->sk = -1;
        else {
                log_info("\nfio: terminating on signal %d\n", signal);
                exit_backend = 1;
@@ -1607,6 +2280,7 @@ int fio_start_server(char *pidfile)
        if (check_existing_pidfile(pidfile)) {
                log_err("fio: pidfile %s exists and server appears alive\n",
                                                                pidfile);
+               free(pidfile);
                return -1;
        }
 
@@ -1616,9 +2290,9 @@ int fio_start_server(char *pidfile)
                free(pidfile);
                return -1;
        } else if (pid) {
-               int ret = write_pid(pid, pidfile);
-
-               exit(ret);
+               ret = write_pid(pid, pidfile);
+               free(pidfile);
+               _exit(ret);
        }
 
        setsid();