From mboxrd@z Thu Jan 1 00:00:00 1970 From: Richard Palethorpe Date: Wed, 25 Mar 2020 14:52:50 +0100 Subject: [LTP] [PATCH] Fix use after stack unwind in fzsync lib In-Reply-To: <20200325103734.31295-1-mdoucha@suse.cz> References: <20200325103734.31295-1-mdoucha@suse.cz> Message-ID: <875zeszglp.fsf@our.domain.is.not.set> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: ltp@lists.linux.it Hello, Martin Doucha writes: > tst_fzsync_pair_reset() passes a local variable to thread B which may be > already unwinded by the time the thread wrapper function executes. If new > variables get allocated and initialized on stack between pthread_create() > and thread wrapper execution, thread B will segfault. > > Signed-off-by: Martin Doucha > --- > include/tst_fuzzy_sync.h | 15 +++++++++++---- > 1 file changed, 11 insertions(+), 4 deletions(-) > > diff --git a/include/tst_fuzzy_sync.h b/include/tst_fuzzy_sync.h > index c1d0b00f9..32b9859a0 100644 > --- a/include/tst_fuzzy_sync.h > +++ b/include/tst_fuzzy_sync.h > @@ -242,11 +242,14 @@ struct tst_fzsync_run_thread { > */ > static void *tst_fzsync_thread_wrapper(void *run_thread) > { > - struct tst_fzsync_run_thread t = *(struct tst_fzsync_run_thread *)run_thread; > + struct tst_fzsync_run_thread *t = run_thread; > + void *ret; > > pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, NULL); > pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, NULL); > - return t.func(t.arg); > + ret = t->func(t->arg); > + free(t); Why not use SAFE_FREE? > + return ret; > } > > /** > @@ -297,8 +300,12 @@ static void tst_fzsync_pair_reset(struct tst_fzsync_pair *pair, > pair->b_cntr = 0; > pair->exit = 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); > + struct tst_fzsync_run_thread *wrap_run_b; > + > + wrap_run_b = SAFE_MALLOC(sizeof(struct tst_fzsync_run_thread)); > + wrap_run_b->func = run_b; > + wrap_run_b->arg = NULL; > + SAFE_PTHREAD_CREATE(&pair->thread_b, 0, > tst_fzsync_thread_wrapper, wrap_run_b); I suppose there is a memory leak here if thread create fails, but we probably don't care because the test will exit shortly afterwards. > } > > pair->exec_time_start = (float)tst_timeout_remaining(); -- Thank you, Richard.