powertop segfault

perf.patch

commit e0f684d20bf22126842a4d7d69a87be73135426b
gpg: Signature made Tue 23 Apr 2019 05:42:37 AM PDT
gpg:                using RSA key F6CC1F4DF325EACBBE2532481F299543498470BA
gpg:                issuer "justincapella@gmail.com"
gpg: Good signature from "Justin Capella <justincapella@gmail.com>" [ultimate]
Author: b1tninja <devnull@localhost>
Date:   Tue Apr 23 05:42:37 2019 -0700

    Call clear when unable to properly allocate the perf event

diff --git a/src/perf/perf.cpp b/src/perf/perf.cpp
index 9ed0ba8..93c938c 100644
--- a/src/perf/perf.cpp
+++ b/src/perf/perf.cpp
@@ -132,13 +132,16 @@ void perf_event::create_perf_event(char *eventname, int _cpu)
 				PROT_READ | PROT_WRITE, MAP_SHARED, perf_fd, 0);
 	if (perf_mmap == MAP_FAILED) {
 		fprintf(stderr, "failed to mmap with %d (%s)\n", errno, strerror(errno));
+		clear();
 		return;
 	}
 
 	ret = ioctl(perf_fd, PERF_EVENT_IOC_ENABLE, 0);
 
 	if (ret < 0) {
-		fprintf(stderr, "failed to enable perf \n");
+		fprintf(stderr, "failed to enable perf\n");
+		clear();
+		return;
 	}
 
 	pc = (perf_event_mmap_page *)perf_mmap;

For debugging reasons, figured it best to just check the other variables are not set-- instead of relying on the fd being negative by clearing many... may help with debugging, may reveal other locations making false assumptions based around perf_fd