perf_counter tools: use mmap() output
authorPeter Zijlstra <a.p.zijlstra@chello.nl>
Mon, 23 Mar 2009 17:22:12 +0000 (18:22 +0100)
committerIngo Molnar <mingo@elte.hu>
Mon, 6 Apr 2009 07:30:28 +0000 (09:30 +0200)
update kerneltop to use the mmap() output to gather overflow information

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Paul Mackerras <paulus@samba.org>
Orig-LKML-Reference: <20090323172417.677932499@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Documentation/perf_counter/kerneltop.c

index a72c9bd28071b3913b37e34039be1329a88a3e66..80b790553ec81bb8775a90cce5cb76ff557c8048 100644 (file)
@@ -84,6 +84,7 @@
 #include <sys/prctl.h>
 #include <sys/wait.h>
 #include <sys/uio.h>
+#include <sys/mman.h>
 
 #include <linux/unistd.h>
 
@@ -119,17 +120,25 @@ typedef long long               __s64;
 
 
 #ifdef __x86_64__
-# define __NR_perf_counter_open 295
+#define __NR_perf_counter_open 295
+#define rmb()          asm volatile("lfence" ::: "memory")
+#define cpu_relax()    asm volatile("rep; nop" ::: "memory");
 #endif
 
 #ifdef __i386__
-# define __NR_perf_counter_open 333
+#define __NR_perf_counter_open 333
+#define rmb()          asm volatile("lfence" ::: "memory")
+#define cpu_relax()    asm volatile("rep; nop" ::: "memory");
 #endif
 
 #ifdef __powerpc__
 #define __NR_perf_counter_open 319
+#define rmb()          asm volatile ("sync" ::: "memory")
+#define cpu_relax()    asm volatile ("" ::: "memory");
 #endif
 
+#define unlikely(x)    __builtin_expect(!!(x), 0)
+
 asmlinkage int sys_perf_counter_open(
         struct perf_counter_hw_event    *hw_event_uptr          __user,
         pid_t                           pid,
@@ -181,6 +190,7 @@ static int                  profile_cpu                     = -1;
 static int                     nr_cpus                         =  0;
 static int                     nmi                             =  1;
 static int                     group                           =  0;
+static unsigned int            page_size;
 
 static char                    *vmlinux;
 
@@ -1117,16 +1127,68 @@ static void process_options(int argc, char *argv[])
        }
 }
 
+struct mmap_data {
+       int counter;
+       void *base;
+       unsigned int mask;
+       unsigned int prev;
+};
+
+static unsigned int mmap_read_head(struct mmap_data *md)
+{
+       struct perf_counter_mmap_page *pc = md->base;
+       unsigned int seq, head;
+
+repeat:
+       rmb();
+       seq = pc->lock;
+
+       if (unlikely(seq & 1)) {
+               cpu_relax();
+               goto repeat;
+       }
+
+       head = pc->data_head;
+
+       rmb();
+       if (pc->lock != seq)
+               goto repeat;
+
+       return head;
+}
+
+static void mmap_read(struct mmap_data *md)
+{
+       unsigned int head = mmap_read_head(md);
+       unsigned int old = md->prev;
+       unsigned char *data = md->base + page_size;
+
+       if (head - old > md->mask) {
+               printf("ERROR: failed to keep up with mmap data\n");
+               exit(-1);
+       }
+
+       for (; old != head;) {
+               __u64 *ptr = (__u64 *)&data[old & md->mask];
+               old += sizeof(__u64);
+
+               process_event(*ptr, md->counter);
+       }
+
+       md->prev = old;
+}
+
 int main(int argc, char *argv[])
 {
        struct pollfd event_array[MAX_NR_CPUS][MAX_COUNTERS];
+       struct mmap_data mmap_array[MAX_NR_CPUS][MAX_COUNTERS];
        struct perf_counter_hw_event hw_event;
        int i, counter, group_fd;
        unsigned int cpu;
-       uint64_t ip;
-       ssize_t res;
        int ret;
 
+       page_size = sysconf(_SC_PAGE_SIZE);
+
        process_options(argc, argv);
 
        nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
@@ -1153,8 +1215,6 @@ int main(int argc, char *argv[])
                        hw_event.record_type    = PERF_RECORD_IRQ;
                        hw_event.nmi            = nmi;
 
-                       printf("FOO: %d %llx %llx\n", counter, event_id[counter], event_count[counter]);
-
                        fd[i][counter] = sys_perf_counter_open(&hw_event, tid, cpu, group_fd, 0);
                        fcntl(fd[i][counter], F_SETFL, O_NONBLOCK);
                        if (fd[i][counter] < 0) {
@@ -1174,6 +1234,17 @@ int main(int argc, char *argv[])
 
                        event_array[i][counter].fd = fd[i][counter];
                        event_array[i][counter].events = POLLIN;
+
+                       mmap_array[i][counter].counter = counter;
+                       mmap_array[i][counter].prev = 0;
+                       mmap_array[i][counter].mask = 2*page_size - 1;
+                       mmap_array[i][counter].base = mmap(NULL, 3*page_size,
+                                       PROT_READ, MAP_SHARED, fd[i][counter], 0);
+                       if (mmap_array[i][counter].base == MAP_FAILED) {
+                               printf("kerneltop error: failed to mmap with %d (%s)\n",
+                                               errno, strerror(errno));
+                               exit(-1);
+                       }
                }
        }
 
@@ -1188,14 +1259,8 @@ int main(int argc, char *argv[])
                int hits = events;
 
                for (i = 0; i < nr_cpus; i++) {
-                       for (counter = 0; counter < nr_counters; counter++) {
-                               res = read(fd[i][counter], (char *) &ip, sizeof(ip));
-                               if (res > 0) {
-                                       assert(res == sizeof(ip));
-
-                                       process_event(ip, counter);
-                               }
-                       }
+                       for (counter = 0; counter < nr_counters; counter++)
+                               mmap_read(&mmap_array[i][counter]);
                }
 
                if (time(NULL) >= last_refresh + delay_secs) {