perf_counter: Update userspace callchain sampling uses
authorPeter Zijlstra <a.p.zijlstra@chello.nl>
Thu, 18 Jun 2009 20:20:45 +0000 (22:20 +0200)
committerIngo Molnar <mingo@elte.hu>
Fri, 19 Jun 2009 11:42:35 +0000 (13:42 +0200)
Update the tools to reflect the new callchain sampling format.

LKML-Reference: <new-submission>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
tools/perf/builtin-report.c

index 86981bd08f6510f6851bdc0bdbed19c63e7b97ce..7a6577bf9a41dcb811ea45cbb1b24cb48132279a 100644 (file)
@@ -59,6 +59,11 @@ struct ip_event {
        unsigned char __more_data[];
 };
 
+struct ip_callchain {
+       __u64 nr;
+       __u64 ips[0];
+};
+
 struct mmap_event {
        struct perf_event_header header;
        __u32 pid, tid;
@@ -833,15 +838,12 @@ got_dso:
        return dso->find_symbol(dso, ip);
 }
 
-static struct symbol *call__match(struct symbol *sym)
+static int call__match(struct symbol *sym)
 {
-       if (!sym)
-               return NULL;
-
        if (sym->name && !regexec(&parent_regex, sym->name, 0, NULL, 0))
-               return sym;
+               return 1;
 
-       return NULL;
+       return 0;
 }
 
 /*
@@ -850,7 +852,7 @@ static struct symbol *call__match(struct symbol *sym)
 
 static int
 hist_entry__add(struct thread *thread, struct map *map, struct dso *dso,
-               struct symbol *sym, __u64 ip, struct perf_callchain_entry *chain,
+               struct symbol *sym, __u64 ip, struct ip_callchain *chain,
                char level, __u64 count)
 {
        struct rb_node **p = &hist.rb_node;
@@ -869,31 +871,35 @@ hist_entry__add(struct thread *thread, struct map *map, struct dso *dso,
        int cmp;
 
        if (sort__has_parent && chain) {
-               int i, nr = chain->hv;
-               struct symbol *sym;
-               struct dso *dso;
-               __u64 ip;
-
-               for (i = 0; i < chain->kernel; i++) {
-                       ip = chain->ip[nr + i];
-                       dso = kernel_dso;
+               __u64 context = PERF_CONTEXT_MAX;
+               int i;
+
+               for (i = 0; i < chain->nr; i++) {
+                       __u64 ip = chain->ips[i];
+                       struct dso *dso = NULL;
+                       struct symbol *sym;
+
+                       if (ip >= PERF_CONTEXT_MAX) {
+                               context = ip;
+                               continue;
+                       }
+
+                       switch (context) {
+                       case PERF_CONTEXT_KERNEL:
+                               dso = kernel_dso;
+                               break;
+                       default:
+                               break;
+                       }
+
                        sym = resolve_symbol(thread, NULL, &dso, &ip);
-                       entry.parent = call__match(sym);
-                       if (entry.parent)
-                               goto got_parent;
-               }
-               nr += i;
-
-               for (i = 0; i < chain->user; i++) {
-                       ip = chain->ip[nr + i];
-                       sym = resolve_symbol(thread, NULL, NULL, &ip);
-                       entry.parent = call__match(sym);
-                       if (entry.parent)
-                               goto got_parent;
+
+                       if (sym && call__match(sym)) {
+                               entry.parent = sym;
+                               break;
+                       }
                }
-               nr += i;
        }
-got_parent:
 
        while (*p != NULL) {
                parent = *p;
@@ -1095,21 +1101,10 @@ static unsigned long total = 0,
                     total_unknown = 0,
                     total_lost = 0;
 
-static int validate_chain(struct perf_callchain_entry *chain, event_t *event)
+static int validate_chain(struct ip_callchain *chain, event_t *event)
 {
        unsigned int chain_size;
 
-       if (chain->nr > MAX_STACK_DEPTH)
-               return -1;
-       if (chain->hv > MAX_STACK_DEPTH)
-               return -1;
-       if (chain->kernel > MAX_STACK_DEPTH)
-               return -1;
-       if (chain->user > MAX_STACK_DEPTH)
-               return -1;
-       if (chain->hv + chain->kernel + chain->user != chain->nr)
-               return -1;
-
        chain_size = event->header.size;
        chain_size -= (unsigned long)&event->ip.__more_data - (unsigned long)event;
 
@@ -1130,7 +1125,7 @@ process_overflow_event(event_t *event, unsigned long offset, unsigned long head)
        __u64 period = 1;
        struct map *map = NULL;
        void *more_data = event->ip.__more_data;
-       struct perf_callchain_entry *chain = NULL;
+       struct ip_callchain *chain = NULL;
 
        if (event->header.type & PERF_SAMPLE_PERIOD) {
                period = *(__u64 *)more_data;
@@ -1150,10 +1145,7 @@ process_overflow_event(event_t *event, unsigned long offset, unsigned long head)
 
                chain = (void *)more_data;
 
-               dprintf("... chain: u:%d, k:%d, nr:%d\n",
-                       chain->user,
-                       chain->kernel,
-                       chain->nr);
+               dprintf("... chain: nr:%Lu\n", chain->nr);
 
                if (validate_chain(chain, event) < 0) {
                        eprintf("call-chain problem with event, skipping it.\n");
@@ -1162,7 +1154,7 @@ process_overflow_event(event_t *event, unsigned long offset, unsigned long head)
 
                if (dump_trace) {
                        for (i = 0; i < chain->nr; i++)
-                               dprintf("..... %2d: %016Lx\n", i, chain->ip[i]);
+                               dprintf("..... %2d: %016Lx\n", i, chain->ips[i]);
                }
        }