Skip to contentSkip to conversation

fzsync: revoke thread_b if parent hits accidental break

PATCH New Li Wang 2 participants 3 messages Sep 26, 2019 – Oct 9, 2019

lore
Share & tools
lore

Li Wang

author
#0 lore raw

PATCH 1/1 fzsync: revoke thread_b if parent hits accidental break

We shouldn't rely entirely on the pair->exit flag in tst_fzsync_pair_cleanup() since there is possible to call tst_brk() at anyplace of thread_a, that will lead to timeout eventually because of thread_b(tst_fzsync_wait_b) fall into an infinite(no explicit condition to exit) loop.

Thread_a path trace:
tst_brk()

    cleanup()
      tst_fzsync_pair_cleanup()
        SAFE_PTHREAD_JOIN(pair->thread_b, NULL)
        #Or pthread_cancel(pair->thread_b)

We fix the problem via a way to kill thread_b with pthread_cancel. With new thread wrapper involves enabling thread cancel at the start of the thread B, then do asynchronous cancellation in tst_fzsync_pair_cleanup if pair->exit is not being set to 1.

Workaround: [commit 2e74d996: Check recvmmsg exists before entering fuzzy loop]

---
 include/tst_fuzzy_sync.h | 33 ++++++++++++++++++++++++++++++---
 1 file changed, 30 insertions(+), 3 deletions(-)
include/tst_fuzzy_sync.h +30 -3
index f9a1947c7..c1d0b00f9 100644
@@ -63,6 +63,7 @@
 #include <time.h>
 #include <math.h>
 #include <stdlib.h>
+#include <pthread.h>
 #include "tst_atomic.h"
 #include "tst_timer.h"
 #include "tst_safe_pthread.h"
@@ -218,12 +219,36 @@ static void tst_fzsync_pair_init(struct tst_fzsync_pair *pair)
 static void tst_fzsync_pair_cleanup(struct tst_fzsync_pair *pair)
 {
 	if (pair->thread_b) {
-		tst_atomic_store(1, &pair->exit);
+		/* Revoke thread B if parent hits accidental break */
+		if (!pair->exit) {
+			tst_atomic_store(1, &pair->exit);
+			usleep(100000);
+			pthread_cancel(pair->thread_b);
+		}
 		SAFE_PTHREAD_JOIN(pair->thread_b, NULL);
 		pair->thread_b = 0;
 	}
 }
 
+/** To store the run_b pointer and pass to tst_fzsync_thread_wrapper */
+struct tst_fzsync_run_thread {
+	void *(*func)(void *);
+	void *arg;
+};
+
+/**
+ * Wrap run_b for tst_fzsync_pair_reset to enable pthread cancel
+ * at the start of the thread B.
+ */
+static void *tst_fzsync_thread_wrapper(void *run_thread)
+{
+       struct tst_fzsync_run_thread t = *(struct tst_fzsync_run_thread *)run_thread;
+
+       pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, NULL);
+       pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, NULL);
+       return t.func(t.arg);
+}
+
 /**
  * Zero some stat fields
  *
@@ -271,8 +296,10 @@ static void tst_fzsync_pair_reset(struct tst_fzsync_pair *pair,
 	pair->a_cntr = 0;
 	pair->b_cntr = 0;
 	pair->exit = 0;
-	if (run_b)
-		SAFE_PTHREAD_CREATE(&pair->thread_b, 0, run_b, 0);
+	if (run_b) {
+		struct tst_fzsync_run_thread wrap_run_b = {.func = run_b, .arg = NULL};
+		SAFE_PTHREAD_CREATE(&pair->thread_b, 0, tst_fzsync_thread_wrapper, &wrap_run_b);
+	}
 
 	pair->exec_time_start = (float)tst_timeout_remaining();
 }
Signature · 1 line
2.20.1
#1 lore raw

PATCH 1/1 fzsync: revoke thread_b if parent hits accidental break

Hi LI,

Quoted 17 lines

We shouldn't rely entirely on the pair->exit flag in tst_fzsync_pair_cleanup() since there is possible to call tst_brk() at anyplace of thread_a, that will lead to timeout eventually because of thread_b(tst_fzsync_wait_b) fall into an infinite(no explicit condition to exit) loop.

Thread_a path trace:
tst_brk()

    cleanup()
      tst_fzsync_pair_cleanup()
        SAFE_PTHREAD_JOIN(pair->thread_b, NULL)
        #Or pthread_cancel(pair->thread_b)

We fix the problem via a way to kill thread_b with pthread_cancel. With new thread wrapper involves enabling thread cancel at the start of the thread B, then do asynchronous cancellation in tst_fzsync_pair_cleanup if pair->exit is not being set to 1.

Quoted 5 lines

Workaround: [commit 2e74d996: Check recvmmsg exists before entering fuzzy loop]

Kind regards,
Petr

author
#2 lore raw

PATCH 1/1 fzsync: revoke thread_b if parent hits accidental break

Quoted 25 lines · Petr Vorel

Hi LI,

Quoted 1 line

We shouldn't rely entirely on the pair->exit flag in

tst_fzsync_pair_cleanup()

Quoted 1 line

since there is possible to call tst_brk() at anyplace of thread_a, that

will

Quoted 1 line

lead to timeout eventually because of thread_b(tst_fzsync_wait_b) fall

into

Quoted 10 lines

an infinite(no explicit condition to exit) loop.

Thread_a path trace:
tst_brk()

    cleanup()
      tst_fzsync_pair_cleanup()
        SAFE_PTHREAD_JOIN(pair->thread_b, NULL)
        #Or pthread_cancel(pair->thread_b)

We fix the problem via a way to kill thread_b with pthread_cancel. With

new

Quoted 1 line

thread wrapper involves enabling thread cancel at the start of the

thread B,

Quoted 1 line

then do asynchronous cancellation in tst_fzsync_pair_cleanup if

pair->exit

Quoted 1 line

is not being set to 1.

Quoted 8 lines
Quoted 1 line

Workaround: [commit 2e74d996: Check recvmmsg exists before entering

fuzzy loop]

Quoted 4 lines

Pushed. Thanks for the review.

Signature · 5 lines
Regards,
Li Wang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20191009/9fb0a542/attachment.htm>