Re: [PATCH v5 24/37] tracing: Add support for 'synthetic' events

From: Tom Zanussi
Date: Tue Nov 14 2017 - 14:34:42 EST


Hi Namhyung,

On Tue, 2017-11-14 at 19:06 +0900, Namhyung Kim wrote:
> On Thu, Nov 09, 2017 at 02:33:55PM -0600, Tom Zanussi wrote:
> > Synthetic events are user-defined events generated from hist trigger
> > variables saved from one or more other events.
> >
> > To define a synthetic event, the user writes a simple specification
> > consisting of the name of the new event along with one or more
> > variables and their type(s), to the tracing/synthetic_events file.
> >
> > For instance, the following creates a new event named 'wakeup_latency'
> > with 3 fields: lat, pid, and prio:
> >
> > # echo 'wakeup_latency u64 lat; pid_t pid; int prio' >> \
> > /sys/kernel/debug/tracing/synthetic_events
> >
> > Reading the tracing/synthetic_events file lists all the
> > currently-defined synthetic events, in this case the event we defined
> > above:
> >
> > # cat /sys/kernel/debug/tracing/synthetic_events
> > wakeup_latency u64 lat; pid_t pid; int prio
> >
> > At this point, the synthetic event is ready to use, and a histogram
> > can be defined using it:
> >
> > # echo 'hist:keys=pid,prio,lat.log2:sort=pid,lat' >> \
> > /sys/kernel/debug/tracing/events/synthetic/wakeup_latency/trigger
> >
> > The new event is created under the tracing/events/synthetic/ directory
> > and looks and behaves just like any other event:
> >
> > # ls /sys/kernel/debug/tracing/events/synthetic/wakeup_latency
> > enable filter format hist id trigger
> >
> > Although a histogram can be defined for it, nothing will happen until
> > an action tracing that event via the trace_synth() function occurs.
> > The trace_synth() function is very similar to all the other trace_*
> > invocations spread throughout the kernel, except in this case the
> > trace_ function and its corresponding tracepoint isn't statically
> > generated but defined by the user at run-time.
> >
> > How this can be automatically hooked up via a hist trigger 'action' is
> > discussed in a subsequent patch.
> >
> > Signed-off-by: Tom Zanussi <tom.zanussi@xxxxxxxxxxxxxxx>
> > ---
> > kernel/trace/trace_events_hist.c | 908 ++++++++++++++++++++++++++++++++++++++-
> > 1 file changed, 906 insertions(+), 2 deletions(-)
> >
> > diff --git a/kernel/trace/trace_events_hist.c b/kernel/trace/trace_events_hist.c
> > index 3504aa8..510b10d 100644
> > --- a/kernel/trace/trace_events_hist.c
> > +++ b/kernel/trace/trace_events_hist.c
> > @@ -20,10 +20,16 @@
> > #include <linux/slab.h>
> > #include <linux/stacktrace.h>
> > #include <linux/rculist.h>
> > +#include <linux/tracefs.h>
> >
> > #include "tracing_map.h"
> > #include "trace.h"
> >
> > +#define SYNTH_SYSTEM "synthetic"
> > +#define SYNTH_FIELDS_MAX 16
> > +
> > +#define STR_VAR_LEN_MAX 32 /* must be multiple of sizeof(u64) */
> > +
> > struct hist_field;
> >
> > typedef u64 (*hist_field_fn_t) (struct hist_field *field,
> > @@ -270,6 +276,26 @@ struct hist_trigger_data {
> > unsigned int n_actions;
> > };
> >
> > +struct synth_field {
> > + char *type;
> > + char *name;
> > + size_t size;
> > + bool is_signed;
> > + bool is_string;
> > +};
> > +
> > +struct synth_event {
> > + struct list_head list;
> > + int ref;
> > + char *name;
> > + struct synth_field **fields;
> > + unsigned int n_fields;
> > + unsigned int n_u64;
> > + struct trace_event_class class;
> > + struct trace_event_call call;
> > + struct tracepoint *tp;
> > +};
> > +
> > struct action_data;
> >
> > typedef void (*action_fn_t) (struct hist_trigger_data *hist_data,
> > @@ -282,6 +308,803 @@ struct action_data {
> > unsigned int var_ref_idx;
> > };
> >
> > +static LIST_HEAD(synth_event_list);
> > +static DEFINE_MUTEX(synth_event_mutex);
> > +
> > +struct synth_trace_event {
> > + struct trace_entry ent;
> > + u64 fields[];
> > +};
> > +
> > +static int synth_event_define_fields(struct trace_event_call *call)
> > +{
> > + struct synth_trace_event trace;
> > + int offset = offsetof(typeof(trace), fields);
> > + struct synth_event *event = call->data;
> > + unsigned int i, size, n_u64;
> > + char *name, *type;
> > + bool is_signed;
> > + int ret = 0;
> > +
> > + for (i = 0, n_u64 = 0; i < event->n_fields; i++) {
> > + size = event->fields[i]->size;
> > + is_signed = event->fields[i]->is_signed;
> > + type = event->fields[i]->type;
> > + name = event->fields[i]->name;
> > + ret = trace_define_field(call, type, name, offset, size,
> > + is_signed, FILTER_OTHER);
> > + if (ret)
> > + break;
> > +
> > + if (event->fields[i]->is_string) {
> > + offset += STR_VAR_LEN_MAX;
> > + n_u64 += STR_VAR_LEN_MAX / sizeof(u64);
>
> Did you use fixed size array for strings?
>

Yes, for a string field, I allocate enough for the maximum string size,
regardless of how long it may be. Not the most efficient, but strings
are relatively rare, and it simplifies the code overall.

>
> > + } else {
> > + offset += sizeof(u64);
> > + n_u64++;
> > + }
> > + }
> > +
> > + event->n_u64 = n_u64;
> > +
> > + return ret;
> > +}
> > +
> > +static bool synth_field_signed(char *type)
> > +{
> > + if (strncmp(type, "u", 1) == 0)
> > + return false;
> > +
> > + return true;
> > +}
> > +
> > +static int synth_field_is_string(char *type)
> > +{
> > + if (strstr(type, "char[") != NULL)
> > + return true;
> > +
> > + return false;
> > +}
> > +
> > +static int synth_field_string_size(char *type)
> > +{
> > + char buf[4], *end, *start;
> > + unsigned int len;
> > + int size, err;
> > +
> > + start = strstr(type, "char[");
> > + if (start == NULL)
> > + return -EINVAL;
> > + start += strlen("char[");
> > +
> > + end = strchr(type, ']');
> > + if (!end || end < start)
> > + return -EINVAL;
> > +
> > + len = end - start;
> > + if (len > 3)
> > + return -EINVAL;
> > +
> > + strncpy(buf, start, len);
> > + buf[len] = '\0';
> > +
> > + err = kstrtouint(buf, 0, &size);
> > + if (err)
> > + return err;
> > +
> > + if (size > STR_VAR_LEN_MAX)
> > + return -EINVAL;
> > +
> > + return size;
> > +}
>
> But this seems to use the actual array size for string..
>

Yeah, we still track the actual size for the string in the event field,
regardless of the 'container' reserved for it.

> [SNIP]
> > +
> > +static notrace void trace_event_raw_event_synth(void *__data,
> > + u64 *var_ref_vals,
> > + unsigned int var_ref_idx)
> > +{
> > + struct trace_event_file *trace_file = __data;
> > + struct synth_trace_event *entry;
> > + struct trace_event_buffer fbuffer;
> > + struct synth_event *event;
> > + unsigned int i, n_u64;
> > + int fields_size = 0;
> > +
> > + event = trace_file->event_call->data;
> > +
> > + if (trace_trigger_soft_disabled(trace_file))
> > + return;
> > +
> > + fields_size = event->n_u64 * sizeof(u64);
> > +
> > + entry = trace_event_buffer_reserve(&fbuffer, trace_file,
> > + sizeof(*entry) + fields_size);
> > + if (!entry)
> > + return;
> > +
> > + for (i = 0, n_u64 = 0; i < event->n_fields; i++) {
> > + if (event->fields[i]->is_string) {
> > + char *str_val = (char *)(long)var_ref_vals[var_ref_idx + i];
> > + char *str_field = (char *)&entry->fields[n_u64];
> > +
> > + strncpy(str_field, str_val, STR_VAR_LEN_MAX);
> > + n_u64 += STR_VAR_LEN_MAX / sizeof(u64);
>
> Here it uses the fixed size again..
>

Right, that's as designed - we always reserve the maximum string size.
It's not as efficient and will waste some buffer space, but it
simplifies things overall by not having to keep track of variable-length
strings in events. I could change this or submit a more efficient
scheme later, but I'm not sure it's really worth it...

Thanks,

Tom

> Thanks,
> Namhyung
>
>
> > + } else {
> > + entry->fields[i] = var_ref_vals[var_ref_idx + i];
> > + n_u64++;
> > + }
> > + }
> > +
> > + trace_event_buffer_commit(&fbuffer);
> > +}