bring perf_event_open() syscall support up to Linux 3.10

[Vince Weaver]

The following patch implements the new features from Linux 3.10 as well
as properly sets up the new features from 3.7 that were only 
half-implemented before.

Signed-off-by: Vince Weaver <vincent.wea...@maine.edu>

diff --git a/syscalls/perf_event_open.c b/syscalls/perf_event_open.c
index 5127231..d1d59ec 100644
--- a/syscalls/perf_event_open.c
+++ b/syscalls/perf_event_open.c
@@ -785,6 +785,10 @@ static long long random_sample_type(void)
                sample_type |= PERF_SAMPLE_REGS_USER;
        if (rand_bool())
                sample_type |= PERF_SAMPLE_STACK_USER;
+       if (rand_bool())
+               sample_type |= PERF_SAMPLE_WEIGHT;
+       if (rand_bool())
+               sample_type |= PERF_SAMPLE_DATA_SRC;
 
        return sample_type;
 }
@@ -837,6 +841,34 @@ static int random_attr_size(void) {
        return size;
 }
 
+static long long random_branch_sample_type(void)
+{
+
+       long long branch_sample = 0;
+
+       if (rand() % 2)
+               return rand64();
+
+       if (rand_bool())
+               branch_sample |= PERF_SAMPLE_BRANCH_USER;
+       if (rand_bool())
+               branch_sample |= PERF_SAMPLE_BRANCH_KERNEL;
+       if (rand_bool())
+               branch_sample |= PERF_SAMPLE_BRANCH_HV;
+
+       if (rand_bool())
+               branch_sample |= PERF_SAMPLE_BRANCH_ANY;
+       if (rand_bool())
+               branch_sample |= PERF_SAMPLE_BRANCH_ANY_CALL;
+       if (rand_bool())
+               branch_sample |= PERF_SAMPLE_BRANCH_ANY_RETURN;
+       if (rand_bool())
+               branch_sample |= PERF_SAMPLE_BRANCH_IND_CALL;
+
+       return branch_sample;
+}
+
+
 static void create_mostly_valid_counting_event(struct perf_event_attr *attr,
                                                int group_leader)
 {
@@ -886,6 +918,11 @@ static void create_mostly_valid_counting_event(struct 
perf_event_attr *attr,
                /* leave config2 alone for now */
        }
 
+       /* branch_sample_type not relevant if not sampling */
+
+       /* sample_regs_user not relevant if not sampling */
+
+       /* sample_stack_user not relevant if not sampling */
 }
 
 static void create_mostly_valid_sampling_event(struct perf_event_attr *attr,
@@ -940,6 +977,26 @@ static void create_mostly_valid_sampling_event(struct 
perf_event_attr *attr,
                /* config1 set earlier */
        }
 
+       attr->branch_sample_type = random_branch_sample_type();
+
+       /* sample_regs_user is a bitmask of CPU registers to record.     */
+       /* The values come from arch/ARCH/include/uapi/asm/perf_regs.h   */
+       /* Most architectures have fewer than 64 registers...            */
+       switch(rand()%3) {
+               case 0:         attr->sample_regs_user = rand()%16;
+                               break;
+               case 1:         attr->sample_regs_user = rand()%64;
+                               break;
+               case 2:         attr->sample_regs_user = rand64();
+                               break;
+               default:
+                               break;
+       }
+
+       /* sample_stack_user is the size of user stack backtrace we want  */
+       /* if we pick too large of a value the kernel in theory truncates */
+       attr->sample_stack_user = rand32();
+
 }
 
 /* Creates a completely random event, unlikely to be valid */
--
To unsubscribe from this list: send the line "unsubscribe trinity" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html