lib: Add per-user packet pools

The IRMd will now check the user UID and GID for privileged access,
avoiding unprivileged users being able to disrupt all IPC (e.g. by
shm_open the single pool and corrupting its metadata).

Non-privileged users are now limited to a PUP (per-user pool) for
sending/receiving packets. It is still created by the IRMd, but owned
by the user (uid) with 600 permissions. It does not add additional
copies for local IPC between their own processes (i.e. over the local
IPCP), but packets between processes owned by a different user or
destined over the network (other IPCPs) will incur a copy when
crossing the PUP / PUP or the PUP / GSPP boundary.

Privileged users and users in the ouroboros group still have direct
access to the GSPP (globally shared private pool) for packet transfer
that will avoid additional copies when processing packets between
processes owned by different users and to the network.

This aligns the security model with UNIX trust domains defined by UID
and GID by leveraging file permission on the pools in shared memory.

┌─────────────────────────────────────────────────────────────┐
│ Source Pool    │ Dest Pool      │ Operation    │ Copies     │
├─────────────────────────────────────────────────────────────┤
│ GSPP           │ GSPP           │ Zero-copy    │ 0          │
│ PUP.uid        │ PUP.uid        │ Zero-copy    │ 0          │
│ PUP.uid1       │ PUP.uid2       │ memcpy()     │ 1          │
│ PUP.uid        │ GSPP           │ memcpy()     │ 1          │
│ GSPP           │ PUP.uid        │ memcpy()     │ 1          │
└─────────────────────────────────────────────────────────────┘

This also renames the struct ai ("application instance") in dev.c to
struct proc (process).

Signed-off-by: Dimitri Staessens <dimitri@ouroboros.rocks>
Signed-off-by: Sander Vrijders <sander@ouroboros.rocks>

3 files changed, 127 insertions, 38 deletions

diff --git a/src/tools/oping/oping.c b/src/tools/oping/oping.c
index 87c1ee18..47b86c68 100644
--- a/src/tools/oping/oping.c
+++ b/src/tools/oping/oping.c
@@ -75,12 +75,14 @@
 "\n"                                                                         \
 "  -c, --count             Number of packets\n"                              \
 "  -d, --duration          Duration of the test (default 1s)\n"              \
+"  -f, --flood             Send back-to-back without waiting\n"              \
 "  -i, --interval          Interval (default 1000ms)\n"                      \
 "  -n, --server-name       Name of the oping server\n"                       \
 "  -q, --qos               QoS (raw, best, video, voice, data)\n" \
 "  -s, --size              Payload size (B, default 64)\n"                   \
 "  -Q, --quiet             Only print final statistics\n"                    \
 "  -D, --timeofday         Print time of day before each line\n"             \
+"      --poll              Server uses polling (lower latency)\n"            \
 "\n"                                                                         \
 "      --help              Display this help text and exit\n"                \
 
@@ -90,6 +92,7 @@ struct {
         uint32_t  count;
         int       size;
         bool      timestamp;
+        bool      flood;
         qosspec_t qs;
 
         /* stats */
@@ -114,6 +117,7 @@ struct {
         pthread_mutex_t lock;
 
         bool quiet;
+        bool poll;
 
         pthread_t cleaner_pt;
         pthread_t accept_pt;
@@ -172,9 +176,11 @@ int main(int     argc,
         client.size      = 64;
         client.count     = INT_MAX;
         client.timestamp = false;
+        client.flood     = false;
         client.qs        = qos_raw;
         client.quiet     = false;
         server.quiet     = false;
+        server.poll      = false;
 
         while (argc > 0) {
                 if ((strcmp(*argv, "-i") == 0 ||
@@ -212,6 +218,9 @@ int main(int     argc,
                 } else if (strcmp(*argv, "-l") == 0 ||
                            strcmp(*argv, "--listen") == 0) {
                         serv = true;
+                } else if (strcmp(*argv, "-f") == 0 ||
+                           strcmp(*argv, "--flood") == 0) {
+                        client.flood = true;
                 } else if (strcmp(*argv, "-D") == 0 ||
                            strcmp(*argv, "--timeofday") == 0) {
                         client.timestamp = true;
@@ -219,6 +228,8 @@ int main(int     argc,
                            strcmp(*argv, "--quiet") == 0) {
                         client.quiet = true;
                         server.quiet = true;
+                } else if (strcmp(*argv, "--poll") == 0) {
+                        server.poll = true;
                 } else {
                         goto fail;
                 }
diff --git a/src/tools/oping/oping_client.c b/src/tools/oping/oping_client.c
index 5a9e03dc..1513eac8 100644
--- a/src/tools/oping/oping_client.c
+++ b/src/tools/oping/oping_client.c
@@ -53,6 +53,20 @@ void shutdown_client(int signo, siginfo_t * info, void * c)
         }
 }
 
+static void update_rtt_stats(double ms)
+{
+        double d;
+
+        if (ms < client.rtt_min)
+                client.rtt_min = ms;
+        if (ms > client.rtt_max)
+                client.rtt_max = ms;
+
+        d = (ms - client.rtt_avg);
+        client.rtt_avg += d / client.rcvd;
+        client.rtt_m2 += d * (ms - client.rtt_avg);
+}
+
 void * reader(void * o)
 {
         struct timespec timeout = {client.interval / 1000 + 2, 0};
@@ -64,7 +78,6 @@ void * reader(void * o)
         int                fd      = *((int *) o);
         int                msg_len = 0;
         double             ms      = 0;
-        double             d       = 0;
         uint32_t           exp_id  = 0;
 
         fccntl(fd, FLOWSRCVTIMEO, &timeout);
@@ -122,14 +135,7 @@ void * reader(void * o)
                                id < exp_id ? " [out-of-order]" : "");
                 }
 
-                if (ms < client.rtt_min)
-                        client.rtt_min = ms;
-                if (ms > client.rtt_max)
-                        client.rtt_max = ms;
-
-                d = (ms - client.rtt_avg);
-                client.rtt_avg += d / client.rcvd;
-                client.rtt_m2 += d * (ms - client.rtt_avg);
+                update_rtt_stats(ms);
 
                 if (id >= exp_id)
                         exp_id = id + 1;
@@ -204,13 +210,103 @@ static void client_fini(void)
         return;
 }
 
+static void print_stats(struct timespec * tic,
+                        struct timespec * toc)
+{
+        printf("\n");
+        printf("--- %s ping statistics ---\n", client.s_apn);
+        printf("%d packets transmitted, ", client.sent);
+        printf("%d received, ", client.rcvd);
+        printf("%zd out-of-order, ", client.ooo);
+        printf("%.0lf%% packet loss, ", client.sent == 0 ? 0 :
+               ceil(100 - (100 * (client.rcvd / (float) client.sent))));
+        printf("time: %.3f ms\n", ts_diff_us(toc, tic) / 1000.0);
+
+        if (client.rcvd > 0) {
+                printf("rtt min/avg/max/mdev = %.3f/%.3f/%.3f/",
+                       client.rtt_min,
+                       client.rtt_avg,
+                       client.rtt_max);
+                if (client.rcvd > 1)
+                        printf("%.3f ms\n",
+                               sqrt(client.rtt_m2 / (client.rcvd - 1)));
+                else
+                        printf("NaN ms\n");
+        }
+}
+
+static int flood_ping(int fd)
+{
+        char buf[OPING_BUF_SIZE];
+        struct oping_msg * msg = (struct oping_msg *) buf;
+        struct timespec sent;
+        struct timespec rcvd;
+        double ms;
+
+        memset(buf, 0, client.size);
+
+        if (!client.quiet)
+                printf("Pinging %s with %d bytes of data (%u packets):\n\n",
+                       client.s_apn, client.size, client.count);
+
+        while (!stop && client.sent < client.count) {
+                clock_gettime(CLOCK_MONOTONIC, &sent);
+
+                msg->type = htonl(ECHO_REQUEST);
+                msg->id = htonl(client.sent);
+                msg->tv_sec = sent.tv_sec;
+                msg->tv_nsec = sent.tv_nsec;
+
+                if (flow_write(fd, buf, client.size) < 0) {
+                        printf("Failed to send packet.\n");
+                        break;
+                }
+
+                ++client.sent;
+
+                if (flow_read(fd, buf, OPING_BUF_SIZE) < 0) {
+                        printf("Failed to read packet.\n");
+                        break;
+                }
+
+                clock_gettime(CLOCK_MONOTONIC, &rcvd);
+
+                if (ntohl(msg->type) != ECHO_REPLY)
+                        continue;
+
+                ++client.rcvd;
+
+                sent.tv_sec = msg->tv_sec;
+                sent.tv_nsec = msg->tv_nsec;
+                ms = ts_diff_us(&rcvd, &sent) / 1000.0;
+
+                update_rtt_stats(ms);
+
+                if (!client.quiet)
+                        printf("%d bytes from %s: seq=%d time=%.3f ms\n",
+                               client.size, client.s_apn,
+                               ntohl(msg->id), ms);
+        }
+
+        return 0;
+}
+
+static int threaded_ping(int fd)
+{
+        pthread_create(&client.reader_pt, NULL, reader, &fd);
+        pthread_create(&client.writer_pt, NULL, writer, &fd);
+
+        pthread_join(client.writer_pt, NULL);
+        pthread_join(client.reader_pt, NULL);
+
+        return 0;
+}
+
 static int client_main(void)
 {
         struct sigaction sig_act;
-
         struct timespec tic;
         struct timespec toc;
-
         int fd;
 
         memset(&sig_act, 0, sizeof sig_act);
@@ -241,37 +337,16 @@ static int client_main(void)
 
         clock_gettime(CLOCK_REALTIME, &tic);
 
-        pthread_create(&client.reader_pt, NULL, reader, &fd);
-        pthread_create(&client.writer_pt, NULL, writer, &fd);
-
-        pthread_join(client.writer_pt, NULL);
-        pthread_join(client.reader_pt, NULL);
+        if (client.flood)
+                flood_ping(fd);
+        else
+                threaded_ping(fd);
 
         clock_gettime(CLOCK_REALTIME, &toc);
 
-        printf("\n");
-        printf("--- %s ping statistics ---\n", client.s_apn);
-        printf("%d packets transmitted, ", client.sent);
-        printf("%d received, ", client.rcvd);
-        printf("%zd out-of-order, ", client.ooo);
-        printf("%.0lf%% packet loss, ", client.sent == 0 ? 0 :
-               ceil(100 - (100 * (client.rcvd / (float) client.sent))));
-        printf("time: %.3f ms\n", ts_diff_us(&toc, &tic) / 1000.0);
-
-        if (client.rcvd > 0) {
-                printf("rtt min/avg/max/mdev = %.3f/%.3f/%.3f/",
-                       client.rtt_min,
-                       client.rtt_avg,
-                       client.rtt_max);
-                if (client.rcvd > 1)
-                        printf("%.3f ms\n",
-                               sqrt(client.rtt_m2 / (client.rcvd - 1)));
-                else
-                        printf("NaN ms\n");
-        }
+        print_stats(&tic, &toc);
 
         flow_dealloc(fd);
-
         client_fini();
 
         return 0;
diff --git a/src/tools/oping/oping_server.c b/src/tools/oping/oping_server.c
index c1d5e6e5..13ab8f47 100644
--- a/src/tools/oping/oping_server.c
+++ b/src/tools/oping/oping_server.c
@@ -89,12 +89,15 @@ void * server_thread(void *o)
         struct oping_msg * msg = (struct oping_msg *) buf;
         struct timespec now = {0, 0};
         struct timespec timeout = {0, 100 * MILLION};
+        struct timespec poll_timeout = {0, 0};
         int fd;
 
         (void) o;
 
         while (true) {
-                if (fevent(server.flows, server.fq, &timeout) == -ETIMEDOUT)
+                if (fevent(server.flows, server.fq,
+                           server.poll ? &poll_timeout : &timeout)
+                    == -ETIMEDOUT)
                         continue;
 
                 while ((fd = fqueue_next(server.fq)) >= 0) {