* [PATCH v3 1/2] trace-cmd: Find PID of host-guest task from tracing vsock connection
2021-05-13 20:43 [PATCH v3 0/2] trace-cmd: Use vsock tracing to find cids and threads Steven Rostedt
@ 2021-05-13 20:43 ` Steven Rostedt
2021-05-13 20:43 ` [PATCH v3 2/2] trace-cmd: Trace timesync to find pids that map vCPUs Steven Rostedt
1 sibling, 0 replies; 7+ messages in thread
From: Steven Rostedt @ 2021-05-13 20:43 UTC (permalink / raw)
To: linux-trace-devel; +Cc: Steven Rostedt (VMware)
From: "Steven Rostedt (VMware)" <rostedt@goodmis.org>
When a CID is specified on the command line to connect to an agent running
in a KVM guest, in order to use the KVM information for the guest clock,
the main PID of that guest needs to be found.
The problem is that there's no interface that maps CIDs to tasks on the
host. Currently, the code has a hack to search all tasks, looking for
"qemu" tasks and then parsing its command line to see if it can find the
CID that it used. This is not robust and does not work for non qemu guests
running in a KVM environment.
Instead, trace the flow of "wake ups" from the trace-cmd task to a task
that does a "kvm_exit" when connecting to the given CID.
That is, when a connect() call is done on the give CID, tracing the
following flow:
trace-cmd wakes up "vhost-<pid>"
vhost-<pid> wakes up "CPUX/KVM"
CPUX/KVM calls "kvm_exit"
Looking at only wake up events that happen outside of interrupt context
(interrupt wake ups can be for other tasks), finding the task that does
the "kvm_exit" is the task that is running a guest vCPU.
The CPUX/KVM is the task that runs the vCPU of the guest, but we still
need the task group leader of this task to find the task that KVM has
for timestamp offsets and multipliers. To do that, look at the proc
file system for the PID of the CPUX/KVM and read its status file.
This holds "Tgid: <pid>" where "<pid>" is the PID of the task that
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
---
Changes since v2:
- rewrote the change log.
- Used some of the new APIs of libtracefs 1.2
tracecmd/trace-vm.c | 227 +++++++++++++++++++++++++++++++++++++++++++-
1 file changed, 226 insertions(+), 1 deletion(-)
diff --git a/tracecmd/trace-vm.c b/tracecmd/trace-vm.c
index c0333b67..163536ca 100644
--- a/tracecmd/trace-vm.c
+++ b/tracecmd/trace-vm.c
@@ -9,6 +9,7 @@
#include <sys/types.h>
#include <dirent.h>
#include <limits.h>
+#include <unistd.h>
#include "trace-local.h"
#include "trace-msg.h"
@@ -87,6 +88,227 @@ static struct trace_guest *add_guest(unsigned int cid, const char *name)
return &guests[guests_len - 1];
}
+static struct tracefs_instance *start_trace_connect(void)
+{
+ struct tracefs_instance *open_instance;
+
+ open_instance = tracefs_instance_create("vsock_find_pid");
+ if (!open_instance)
+ return NULL;
+
+ tracefs_event_enable(open_instance, "sched", "sched_waking");
+ tracefs_event_enable(open_instance, "kvm", "kvm_exit");
+ tracefs_trace_on(open_instance);
+ return open_instance;
+}
+
+struct pids {
+ struct pids *next;
+ int pid;
+};
+
+struct trace_fields {
+ struct tep_event *sched_waking;
+ struct tep_event *kvm_exit;
+ struct tep_format_field *common_pid;
+ struct tep_format_field *sched_next;
+ struct pids *pids;
+ int found_pid;
+};
+
+static void free_pids(struct pids *pids)
+{
+ struct pids *next;
+
+ while (pids) {
+ next = pids;
+ pids = pids->next;
+ free(next);
+ }
+}
+
+static void add_pid(struct pids **pids, int pid)
+{
+ struct pids *new_pid;
+
+ new_pid = malloc(sizeof(*new_pid));
+ if (!new_pid)
+ return;
+
+ new_pid->pid = pid;
+ new_pid->next = *pids;
+ *pids = new_pid;
+}
+
+static bool match_pid(struct pids *pids, int pid)
+{
+ while (pids) {
+ if (pids->pid == pid)
+ return true;
+ pids = pids->next;
+ }
+ return false;
+}
+
+static int callback(struct tep_event *event, struct tep_record *record, int cpu,
+ void *data)
+{
+ struct trace_fields *fields = data;
+ struct tep_handle *tep = event->tep;
+ unsigned long long val;
+ int flags;
+ int type;
+ int pid;
+ int ret;
+
+ ret = tep_read_number_field(fields->common_pid, record->data, &val);
+ if (ret < 0)
+ return 0;
+
+ flags = tep_data_flags(tep, record);
+
+ /* Ignore events in interrupts */
+ if (flags & (TRACE_FLAG_HARDIRQ | TRACE_FLAG_SOFTIRQ))
+ return 0;
+
+ /*
+ * First make sure that this event comes from a PID from
+ * this task (or a task woken by this task)
+ */
+ pid = val;
+ if (!match_pid(fields->pids, pid))
+ return 0;
+
+ type = tep_data_type(tep, record);
+
+ /*
+ * If this event is a kvm_exit, we have our PID
+ * and we can stop processing.
+ */
+ if (type == fields->kvm_exit->id) {
+ fields->found_pid = pid;
+ return -1;
+ }
+
+ if (type != fields->sched_waking->id)
+ return 0;
+
+ ret = tep_read_number_field(fields->sched_next, record->data, &val);
+ if (ret < 0)
+ return 0;
+
+ /* This is a task woken by our task or a chain of wake ups */
+ add_pid(&fields->pids, (int)val);
+ return 0;
+}
+
+static int find_tgid(int pid)
+{
+ FILE *fp;
+ char *path;
+ char *buf = NULL;
+ char *save;
+ size_t l = 0;
+ int tgid = -1;
+
+ if (asprintf(&path, "/proc/%d/status", pid) < 0)
+ return -1;
+
+ fp = fopen(path, "r");
+ free(path);
+ if (!fp)
+ return -1;
+
+ while (getline(&buf, &l, fp) > 0) {
+ char *tok;
+
+ if (strncmp(buf, "Tgid:", 5) != 0)
+ continue;
+ tok = strtok_r(buf, ":", &save);
+ if (!tok)
+ continue;
+ tok = strtok_r(NULL, ":", &save);
+ if (!tok)
+ continue;
+ while (isspace(*tok))
+ tok++;
+ tgid = strtol(tok, NULL, 0);
+ break;
+ }
+ free(buf);
+ fclose(fp);
+
+ return tgid;
+}
+
+static int stop_trace_connect(struct tracefs_instance *open_instance)
+{
+ const char *systems[] = { "kvm", "sched", NULL};
+ struct tep_handle *tep;
+ struct trace_fields trace_fields;
+ int tgid = -1;
+
+ if (!open_instance)
+ return -1;
+
+ /* The connection is finished, stop tracing, we have what we want */
+ tracefs_trace_off(open_instance);
+ tracefs_event_disable(open_instance, NULL, NULL);
+
+ tep = tracefs_local_events_system(NULL, systems);
+
+ trace_fields.sched_waking = tep_find_event_by_name(tep, "sched", "sched_waking");
+ if (!trace_fields.sched_waking)
+ goto out;
+ trace_fields.kvm_exit = tep_find_event_by_name(tep, "kvm", "kvm_exit");
+ if (!trace_fields.kvm_exit)
+ goto out;
+ trace_fields.common_pid = tep_find_common_field(trace_fields.sched_waking,
+ "common_pid");
+ if (!trace_fields.common_pid)
+ goto out;
+ trace_fields.sched_next = tep_find_any_field(trace_fields.sched_waking,
+ "pid");
+ if (!trace_fields.sched_next)
+ goto out;
+
+ trace_fields.found_pid = -1;
+ trace_fields.pids = NULL;
+ add_pid(&trace_fields.pids, getpid());
+ tracefs_iterate_raw_events(tep, open_instance, NULL, 0, callback, &trace_fields);
+ free_pids(trace_fields.pids);
+ out:
+ tracefs_instance_destroy(open_instance);
+ tracefs_instance_free(open_instance);
+
+ if (trace_fields.found_pid > 0)
+ tgid = find_tgid(trace_fields.found_pid);
+
+ return tgid;
+}
+
+/*
+ * In order to find the guest that is associated to the given cid,
+ * trace the sched_waking and kvm_exit events, connect to the cid
+ * (doesn't matter what port, use -1 to not connect to anything)
+ * and find what task gets woken up from this code and calls kvm_exit,
+ * then that is the task that is running the guest.
+ * Then look at the /proc/<guest-pid>/status file to find the task group
+ * id (Tgid), and this is the PID of the task running all the threads.
+ */
+static void find_pid_by_cid(struct trace_guest *guest)
+{
+ struct tracefs_instance *instance;
+ int fd;
+
+ instance = start_trace_connect();
+ fd = trace_open_vsock(guest->cid, -1);
+ guest->pid = stop_trace_connect(instance);
+ /* Just in case! */
+ if (fd >= 0)
+ close(fd);
+}
+
struct trace_guest *trace_get_guest(unsigned int cid, const char *name)
{
struct trace_guest *guest = NULL;
@@ -99,8 +321,11 @@ struct trace_guest *trace_get_guest(unsigned int cid, const char *name)
if (cid > 0) {
guest = get_guest_by_cid(cid);
- if (!guest && name)
+ if (!guest && name) {
guest = add_guest(cid, name);
+ if (guest)
+ find_pid_by_cid(guest);
+ }
}
return guest;
}
--
2.29.2
^ permalink raw reply related [flat|nested] 7+ messages in thread* [PATCH v3 2/2] trace-cmd: Trace timesync to find pids that map vCPUs
2021-05-13 20:43 [PATCH v3 0/2] trace-cmd: Use vsock tracing to find cids and threads Steven Rostedt
2021-05-13 20:43 ` [PATCH v3 1/2] trace-cmd: Find PID of host-guest task from tracing vsock connection Steven Rostedt
@ 2021-05-13 20:43 ` Steven Rostedt
2021-05-14 4:10 ` Tzvetomir Stoyanov
1 sibling, 1 reply; 7+ messages in thread
From: Steven Rostedt @ 2021-05-13 20:43 UTC (permalink / raw)
To: linux-trace-devel; +Cc: Steven Rostedt (VMware)
From: "Steven Rostedt (VMware)" <rostedt@goodmis.org>
If qemu is not found, then there's no mapping between the vCPUs of the
guest and the threads that run them. As the time sync has the agent run on
all the guest's CPUs, trace it, looking for all kvm_entry, which state
which vCPU the host thread is entering on the guest, and then use that to
create the mapping for the data files.
Only parse for qemu if no CID is given on the command line.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
---
Changes since v1:
- Skipped v2 (added to v3 of this patch series)
- Used some of the new APIs of libtracefs 1.2
- Call read_qemu_guests is CID is not supplied on the command line.
tracecmd/include/trace-local.h | 2 +
tracecmd/trace-record.c | 202 ++++++++++++++++++++++++++++++++-
2 files changed, 203 insertions(+), 1 deletion(-)
diff --git a/tracecmd/include/trace-local.h b/tracecmd/include/trace-local.h
index b3997d00..4c5669c9 100644
--- a/tracecmd/include/trace-local.h
+++ b/tracecmd/include/trace-local.h
@@ -298,11 +298,13 @@ void update_first_instance(struct buffer_instance *instance, int topt);
void show_instance_file(struct buffer_instance *instance, const char *name);
struct trace_guest {
+ struct tracefs_instance *instance;
char *name;
int cid;
int pid;
int cpu_max;
int *cpu_pid;
+ int *task_pids;
};
struct trace_guest *trace_get_guest(unsigned int cid, const char *name);
bool trace_have_guests_pid(void);
diff --git a/tracecmd/trace-record.c b/tracecmd/trace-record.c
index 5dd8be4a..5219d60b 100644
--- a/tracecmd/trace-record.c
+++ b/tracecmd/trace-record.c
@@ -3230,6 +3230,38 @@ static int do_accept(int sd)
return -1;
}
+/* Find all the tasks associated with the guest pid */
+static void find_tasks(struct trace_guest *guest)
+{
+ struct dirent *dent;
+ char *path;
+ DIR *dir;
+ int ret;
+ int tasks = 0;
+
+ ret = asprintf(&path, "/proc/%d/task", guest->pid);
+ if (ret < 0)
+ return;
+
+ dir = opendir(path);
+ free(path);
+ if (!dir)
+ return;
+
+ while ((dent = readdir(dir))) {
+ int *pids;
+ if (!(dent->d_type == DT_DIR && is_digits(dent->d_name)))
+ continue;
+ pids = realloc(guest->task_pids, sizeof(int) * (tasks + 2));
+ if (!pids)
+ break;
+ pids[tasks++] = strtol(dent->d_name, NULL, 0);
+ pids[tasks] = -1;
+ guest->task_pids = pids;
+ }
+ closedir(dir);
+}
+
static char *parse_guest_name(char *gname, int *cid, int *port)
{
struct trace_guest *guest;
@@ -3250,10 +3282,18 @@ static char *parse_guest_name(char *gname, int *cid, int *port)
} else if (is_digits(gname))
*cid = atoi(gname);
- read_qemu_guests();
+ if (*cid < 0)
+ read_qemu_guests();
+
+ if (*cid < 0)
+ return NULL;
+
guest = trace_get_guest(*cid, gname);
if (guest) {
*cid = guest->cid;
+ /* Mapping not found, search for them */
+ if (!guest->cpu_pid)
+ find_tasks(guest);
return guest->name;
}
@@ -3712,6 +3752,162 @@ static int open_guest_fifos(const char *guest, int **fds)
return i;
}
+struct trace_mapping {
+ struct tep_event *kvm_entry;
+ struct tep_format_field *vcpu_id;
+ struct tep_format_field *common_pid;
+ int *pids;
+ int *map;
+ int max_cpus;
+};
+
+static void start_mapping_vcpus(struct trace_guest *guest)
+{
+ char *pids = NULL;
+ char *t;
+ int len = 0;
+ int s;
+ int i;
+
+ if (!guest->task_pids)
+ return;
+
+ guest->instance = tracefs_instance_create("map_guest_pids");
+ if (!guest->instance)
+ return;
+
+ for (i = 0; guest->task_pids[i] >= 0; i++) {
+ s = snprintf(NULL, 0, "%d ", guest->task_pids[i]);
+ t = realloc(pids, len + s + 1);
+ if (!t) {
+ free(pids);
+ pids = NULL;
+ break;
+ }
+ pids = t;
+ sprintf(pids + len, "%d ", guest->task_pids[i]);
+ len += s;
+ }
+ if (pids) {
+ tracefs_instance_file_write(guest->instance, "set_event_pid", pids);
+ free(pids);
+ }
+ tracefs_instance_file_write(guest->instance, "events/kvm/kvm_entry/enable", "1");
+}
+
+static int map_vcpus(struct tep_event *event, struct tep_record *record,
+ int cpu, void *context)
+{
+ struct trace_mapping *tmap = context;
+ unsigned long long val;
+ int type;
+ int pid;
+ int ret;
+ int i;
+
+ /* Do we have junk in the buffer? */
+ type = tep_data_type(event->tep, record);
+ if (type != tmap->kvm_entry->id)
+ return 0;
+
+ ret = tep_read_number_field(tmap->common_pid, record->data, &val);
+ if (ret < 0)
+ return 0;
+ pid = (int)val;
+
+ for (i = 0; tmap->pids[i] >= 0; i++) {
+ if (pid == tmap->pids[i])
+ break;
+ }
+ /* Is this thread one we care about ? */
+ if (tmap->pids[i] < 0)
+ return 0;
+
+ ret = tep_read_number_field(tmap->vcpu_id, record->data, &val);
+ if (ret < 0)
+ return 0;
+
+ cpu = (int)val;
+
+ /* Sanity check, warn? */
+ if (cpu >= tmap->max_cpus)
+ return 0;
+
+ /* Already have this one? Should we check if it is the same? */
+ if (tmap->map[cpu] >= 0)
+ return 0;
+
+ tmap->map[cpu] = pid;
+
+ /* Did we get them all */
+ for (i = 0; i < tmap->max_cpus; i++) {
+ if (tmap->map[i] < 0)
+ break;
+ }
+
+ return i == tmap->max_cpus;
+}
+
+static void stop_mapping_vcpus(struct buffer_instance *instance,
+ struct trace_guest *guest)
+{
+ struct trace_mapping tmap = { };
+ struct tep_handle *tep;
+ const char *systems[] = { "kvm", NULL };
+ int i;
+
+ if (!guest->instance)
+ return;
+
+ tmap.pids = guest->task_pids;
+ tmap.max_cpus = instance->cpu_count;
+
+ tmap.map = malloc(sizeof(*tmap.map) * tmap.max_cpus);
+ if (!tmap.map)
+ return;
+
+ for (i = 0; i < tmap.max_cpus; i++)
+ tmap.map[i] = -1;
+
+ tracefs_instance_file_write(guest->instance, "events/kvm/kvm_entry/enable", "0");
+
+ tep = tracefs_local_events_system(NULL, systems);
+ if (!tep)
+ goto out;
+
+ tmap.kvm_entry = tep_find_event_by_name(tep, "kvm", "kvm_entry");
+ if (!tmap.kvm_entry)
+ goto out_free;
+
+ tmap.vcpu_id = tep_find_field(tmap.kvm_entry, "vcpu_id");
+ if (!tmap.vcpu_id)
+ goto out_free;
+
+ tmap.common_pid = tep_find_any_field(tmap.kvm_entry, "common_pid");
+ if (!tmap.common_pid)
+ goto out_free;
+
+ tracefs_iterate_raw_events(tep, guest->instance, NULL, 0, map_vcpus, &tmap);
+
+ for (i = 0; i < tmap.max_cpus; i++) {
+ if (tmap.map[i] < 0)
+ break;
+ }
+ /* We found all the mapped CPUs */
+ if (i == tmap.max_cpus) {
+ guest->cpu_pid = tmap.map;
+ guest->cpu_max = tmap.max_cpus;
+ tmap.map = NULL;
+ }
+
+ out_free:
+ tep_free(tep);
+ out:
+ free(tmap.map);
+ tracefs_instance_destroy(guest->instance);
+ tracefs_instance_free(guest->instance);
+}
+
static int host_tsync(struct common_record_context *ctx,
struct buffer_instance *instance,
unsigned int tsync_port, char *proto)
@@ -3724,11 +3920,15 @@ static int host_tsync(struct common_record_context *ctx,
if (guest == NULL)
return -1;
+ start_mapping_vcpus(guest);
+
instance->tsync = tracecmd_tsync_with_guest(top_instance.trace_id,
instance->tsync_loop_interval,
instance->cid, tsync_port,
guest->pid, instance->cpu_count,
proto, ctx->clock);
+ stop_mapping_vcpus(instance, guest);
+
if (!instance->tsync)
return -1;
--
2.29.2
^ permalink raw reply related [flat|nested] 7+ messages in thread