Add --virt option for record mode for a virtualization environment. If we use this option on a guest, we can send trace data in low overhead. This is because guests can send trace data to a host without copying the data by using splice(2).
The format is: trace-cmd record --virt -e sched* <Restriction> This feature can use from kernel-3.6 which supports splice_read for ftrace and splice_write for virtio-serial. Signed-off-by: Yoshihiro YUNOMAE <[email protected]> --- Documentation/trace-cmd-record.1.txt | 11 +++++- trace-cmd.h | 3 +- trace-msg.c | 35 ++++++++++++++++-- trace-msg.h | 4 ++ trace-record.c | 66 ++++++++++++++++++++++++++++++++-- 5 files changed, 110 insertions(+), 9 deletions(-) diff --git a/Documentation/trace-cmd-record.1.txt b/Documentation/trace-cmd-record.1.txt index 832a257..7eb8ac9 100644 --- a/Documentation/trace-cmd-record.1.txt +++ b/Documentation/trace-cmd-record.1.txt @@ -240,6 +240,15 @@ OPTIONS timestamp to gettimeofday which will allow wall time output from the timestamps reading the created 'trace.dat' file. +*--virt*:: + This option is usded on a guest in a virtualization environment. If a host + is running "trace-cmd virt-server", this option is used to have the data + sent to the host with virtio-serial like *-N* option. (see also + trace-cmd-virt-server(1)) + + Note: This option is not supported with latency tracer plugins: + wakeup, wakeup_rt, irqsoff, preemptoff and preemptirqsoff + EXAMPLES -------- @@ -302,7 +311,7 @@ SEE ALSO -------- trace-cmd(1), trace-cmd-report(1), trace-cmd-start(1), trace-cmd-stop(1), trace-cmd-extract(1), trace-cmd-reset(1), trace-cmd-split(1), -trace-cmd-list(1), trace-cmd-listen(1) +trace-cmd-list(1), trace-cmd-listen(1), trace-cmd-virt-server(1) AUTHOR ------ diff --git a/trace-cmd.h b/trace-cmd.h index b4c2267..5d895ff 100644 --- a/trace-cmd.h +++ b/trace-cmd.h @@ -251,7 +251,8 @@ void tracecmd_stat_cpu(struct trace_seq *s, int cpu); long tracecmd_flush_recording(struct tracecmd_recorder *recorder); /* for clients */ -int tracecmd_msg_connect_to_server(int fd); +int tracecmd_msg_connect_to_server_nw(int fd); +int tracecmd_msg_connect_to_server_virt(int fd); int tracecmd_msg_metadata_send(int fd, char *buf, int size); int tracecmd_msg_finish_sending_metadata(int fd); void tracecmd_msg_send_close_msg(); diff --git a/trace-msg.c b/trace-msg.c index 251e99c..cdeeed3 100644 --- a/trace-msg.c +++ b/trace-msg.c @@ -30,6 +30,7 @@ #include <stdio.h> #include <unistd.h> #include <arpa/inet.h> +#include <sys/stat.h> #include <sys/types.h> #include <linux/types.h> @@ -503,11 +504,12 @@ static int tracecmd_msg_wait_for_msg(int fd, struct tracecmd_msg **msg) return 0; } -int tracecmd_msg_connect_to_server(int fd) +static int tracecmd_msg_connect_to_server(int fd, bool nw) { struct tracecmd_msg *msg; u32 reply_cmd, cmd; int i, ret, cpus; + char buf[PATH_MAX]; /* connect to a server */ cmd = MSG_TCONNECT; @@ -529,9 +531,24 @@ int tracecmd_msg_connect_to_server(int fd) } while (reply_cmd != MSG_RINIT); cpus = ntohl(msg->data.rinit.cpus); - client_ports = malloc_or_die(sizeof(int) * cpus); - for (i = 0; i < cpus; i++) - client_ports[i] = ntohl(msg->data.rinit.port_array[i]); + if (nw) { + client_ports = malloc_or_die(sizeof(int) * cpus); + for (i = 0; i < cpus; i++) + client_ports[i] = + ntohl(msg->data.rinit.port_array[i]); + } else { + virt_sfds = malloc_or_die(sizeof(int) * cpus); + + /* Open data paths of virtio-serial */ + for (i = 0; i < cpus; i++) { + snprintf(buf, PATH_MAX, TRACE_PATH_CPU, i); + virt_sfds[i] = open(buf, O_WRONLY); + if (virt_sfds[i] < 0) { + warning("Cannot open %s", TRACE_PATH_CPU, i); + return -errno; + } + } + } /* Next, send meta data */ send_metadata = true; @@ -543,6 +560,16 @@ error: return ret; } +int tracecmd_msg_connect_to_server_nw(int fd) +{ + return tracecmd_msg_connect_to_server(fd, true); +} + +int tracecmd_msg_connect_to_server_virt(int fd) +{ + return tracecmd_msg_connect_to_server(fd, false); +} + static bool process_option(struct tracecmd_msg_opt *opt) { /* currently the only option we have is to us TCP */ diff --git a/trace-msg.h b/trace-msg.h index 9ec95e5..d1f7321 100644 --- a/trace-msg.h +++ b/trace-msg.h @@ -2,6 +2,9 @@ #define _TRACE_MSG_H_ #include <stdbool.h> +#define VIRTIO_PORTS "/dev/virtio-ports/" +#define AGENT_CTL_PATH VIRTIO_PORTS "agent-ctl-path" +#define TRACE_PATH_CPU VIRTIO_PORTS "trace-path-cpu%d" /* for both clients and server */ bool use_tcp; @@ -11,6 +14,7 @@ int cpu_count; unsigned int page_size; int *client_ports; bool send_metadata; +int *virt_sfds; /* for server */ bool done; diff --git a/trace-record.c b/trace-record.c index ef30e31..df9e786 100644 --- a/trace-record.c +++ b/trace-record.c @@ -80,6 +80,9 @@ static int sfd; /* Max size to let a per cpu file get */ static int max_kb; +struct tracecmd_output *virt_handle; +static bool virt; + static int do_ptrace; static int filter_task; @@ -1576,6 +1579,9 @@ static int create_recorder(struct buffer_instance *instance, int cpu, int extrac if (client_ports) { connect_port(cpu); recorder = tracecmd_create_recorder_fd(client_ports[cpu], cpu, recorder_flags); + } else if (virt_sfds) { + recorder = tracecmd_create_recorder_fd(virt_sfds[cpu], cpu, + recorder_flags); } else { file = get_temp_file(instance, cpu); recorder = create_recorder_instance(instance, file, cpu); @@ -1600,9 +1606,15 @@ static int create_recorder(struct buffer_instance *instance, int cpu, int extrac exit(0); } -static void communicate_with_listener(int fd) +static void communicate_with_listener_nw(int fd) +{ + if (tracecmd_msg_connect_to_server_nw(fd) < 0) + die("Cannot communicate with server"); +} + +static void communicate_with_listener_virt(int fd) { - if (tracecmd_msg_connect_to_server(fd) < 0) + if (tracecmd_msg_connect_to_server_virt(fd) < 0) die("Cannot communicate with server"); } @@ -1654,7 +1666,7 @@ static void setup_network(void) freeaddrinfo(result); - communicate_with_listener(sfd); + communicate_with_listener_nw(sfd); /* Now create the handle through this socket */ handle = tracecmd_create_init_fd_glob(sfd, listed_events); @@ -1663,6 +1675,21 @@ static void setup_network(void) /* OK, we are all set, let'r rip! */ } +static void setup_virtio(void) +{ + int fd; + + fd = open(AGENT_CTL_PATH, O_RDWR); + if (fd < 0) + die("Cannot open %s", AGENT_CTL_PATH); + + communicate_with_listener_virt(fd); + + /* Now create the handle through this socket */ + virt_handle = tracecmd_create_init_fd_glob(fd, listed_events); + tracecmd_msg_finish_sending_metadata(fd); +} + static void finish_network(void) { tracecmd_msg_send_close_msg(); @@ -1670,6 +1697,13 @@ static void finish_network(void) free(host); } +static void finish_virt(void) +{ + tracecmd_msg_send_close_msg(); + free(virt_handle); + free(virt_sfds); +} + static void start_threads(void) { struct buffer_instance *instance; @@ -1677,6 +1711,8 @@ static void start_threads(void) if (host) setup_network(); + else if (virt) + setup_virtio(); /* make a thread for every CPU we have */ pids = malloc_or_die(sizeof(*pids) * cpu_count * (buffers + 1)); @@ -1721,6 +1757,9 @@ static void record_data(char *date2ts, struct trace_seq *s) if (host) { finish_network(); return; + } else if (virt) { + finish_virt(); + return; } if (latency) @@ -2207,6 +2246,7 @@ static void record_all_events(void) } enum { + OPT_virt = 252, OPT_nosplice = 253, OPT_funcstack = 254, OPT_date = 255, @@ -2278,6 +2318,7 @@ void trace_record (int argc, char **argv) {"date", no_argument, NULL, OPT_date}, {"func-stack", no_argument, NULL, OPT_funcstack}, {"nosplice", no_argument, NULL, OPT_nosplice}, + {"virt", no_argument, NULL, OPT_virt}, {"help", no_argument, NULL, '?'}, {NULL, 0, NULL, 0} }; @@ -2389,6 +2430,8 @@ void trace_record (int argc, char **argv) case 'o': if (host) die("-o incompatible with -N"); + if (virt) + die("-o incompatible with --virt"); if (!record && !extract) die("start does not take output\n" "Did you mean 'record'?"); @@ -2420,6 +2463,8 @@ void trace_record (int argc, char **argv) case 'N': if (!record) die("-N only available with record"); + if (virt) + die("-N incompatible with --virt"); if (output) die("-N incompatible with -o"); host = optarg; @@ -2432,6 +2477,8 @@ void trace_record (int argc, char **argv) max_kb = atoi(optarg); break; case 't': + if (virt) + die("-t incompatible with --virt"); use_tcp = 1; break; case 'b': @@ -2458,6 +2505,17 @@ void trace_record (int argc, char **argv) case OPT_nosplice: recorder_flags |= TRACECMD_RECORD_NOSPLICE; break; + case OPT_virt: + if (!record) + die("--virt only available with record"); + if (host) + die("--virt incompatible with -N"); + if (output) + die("--virt incompatible with -o"); + if (use_tcp) + die("--virt incompatible with -t"); + virt = true; + break; default: usage(argv); } @@ -2533,6 +2591,8 @@ void trace_record (int argc, char **argv) latency = 1; if (host) die("Network tracing not available with latency tracer plugins"); + if (virt) + die("Virtio-trace not available with latency tracer plugins"); } if (fset < 0 && (strcmp(plugin, "function") == 0 || strcmp(plugin, "function_graph") == 0)) -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
