summaryrefslogtreecommitdiff
path: root/src/sp/protocol
diff options
context:
space:
mode:
authorGarrett D'Amore <garrett@damore.org>2023-11-25 18:56:29 -0800
committerGarrett D'Amore <garrett@damore.org>2023-11-25 18:56:29 -0800
commitfe692071ed2121ef693eaaaeacc1feb596c3e23b (patch)
treebbd9d5ce56cec1d8c10e835b19c56ee37e4e7082 /src/sp/protocol
parentc2b3b332b65bc9a0f59d9a5e0695b320e35d3408 (diff)
downloadnng-fe692071ed2121ef693eaaaeacc1feb596c3e23b.tar.gz
nng-fe692071ed2121ef693eaaaeacc1feb596c3e23b.tar.bz2
nng-fe692071ed2121ef693eaaaeacc1feb596c3e23b.zip
fixes #1702 segfault canceling req receive while sending
Diffstat (limited to 'src/sp/protocol')
-rw-r--r--src/sp/protocol/reqrep0/req.c18
-rw-r--r--src/sp/protocol/reqrep0/req_test.c39
2 files changed, 57 insertions, 0 deletions
diff --git a/src/sp/protocol/reqrep0/req.c b/src/sp/protocol/reqrep0/req.c
index 5cb691d3..fdac7d5c 100644
--- a/src/sp/protocol/reqrep0/req.c
+++ b/src/sp/protocol/reqrep0/req.c
@@ -537,6 +537,23 @@ req0_ctx_cancel_recv(nni_aio *aio, void *arg, int rv)
req0_sock *s = ctx->sock;
nni_mtx_lock(&s->mtx);
+
+ // So it turns out that some users start receiving before waiting
+ // for the send notification. In this case if receiving is
+ // canceled before sending completes, we need to restore the
+ // message for the user. It's probably a mis-design if the user
+ // is trying to receive without waiting for sending to complete, but
+ // it was reported in the field. Users who want to avoid this mess
+ // should just start receiving from the send completion callback.
+ if (ctx->send_aio != NULL) {
+ nni_aio_set_msg(ctx->send_aio, ctx->req_msg);
+ nni_msg_header_clear(ctx->req_msg);
+ ctx->req_msg = NULL;
+ nni_aio_finish_error(ctx->send_aio, NNG_ECANCELED);
+ ctx->send_aio = NULL;
+ nni_list_remove(&s->send_queue, ctx);
+ }
+
if (ctx->recv_aio == aio) {
ctx->recv_aio = NULL;
@@ -551,6 +568,7 @@ req0_ctx_cancel_recv(nni_aio *aio, void *arg, int rv)
nni_aio_finish_error(aio, rv);
}
+
nni_mtx_unlock(&s->mtx);
}
diff --git a/src/sp/protocol/reqrep0/req_test.c b/src/sp/protocol/reqrep0/req_test.c
index 347ba61e..40a6fe6b 100644
--- a/src/sp/protocol/reqrep0/req_test.c
+++ b/src/sp/protocol/reqrep0/req_test.c
@@ -823,6 +823,44 @@ test_req_ctx_send_twice(void)
}
}
+// this tests for the case where sending and receiving are started in parallel,
+// and then the receiving aio is canceled. The sending should be errored,
+// and we should get back the original message. This test covers bug #1702.
+static void
+test_req_ctx_send_recv_abort(void)
+{
+ nng_socket req;
+ nng_ctx ctx;
+ nng_aio * aio[2];
+ nng_msg * msg;
+
+ NUTS_PASS(nng_req0_open(&req));
+ NUTS_PASS(nng_socket_set_ms(req, NNG_OPT_RECVTIMEO, 100));
+
+ NUTS_PASS(nng_ctx_open(&ctx, req));
+ NUTS_PASS(nng_aio_alloc(&aio[0], NULL, NULL));
+ NUTS_PASS(nng_aio_alloc(&aio[1], NULL, NULL));
+ NUTS_PASS(nng_msg_alloc(&msg, 0));
+
+ nng_aio_set_msg(aio[0], msg);
+ nng_ctx_send(ctx, aio[0]);
+ nng_msleep(1); // just to make sure the socket has it.
+ nng_ctx_recv(ctx, aio[1]);
+ msg = NULL;
+ nng_aio_wait(aio[0]);
+ nng_aio_wait(aio[1]);
+ msg = nng_aio_get_msg(aio[0]); // sent message
+ NUTS_TRUE(msg != NULL);
+ NUTS_TRUE(nng_msg_len(msg) == 0);
+ NUTS_FAIL(nng_aio_result(aio[0]), NNG_ECANCELED);
+ NUTS_FAIL(nng_aio_result(aio[1]), NNG_ETIMEDOUT);
+ NUTS_TRUE(nng_aio_get_msg(aio[0]) != NULL);
+ nng_msg_free(msg);
+ nng_aio_free(aio[0]);
+ nng_aio_free(aio[1]);
+ NUTS_CLOSE(req);
+}
+
static void
test_req_ctx_recv_nonblock(void)
{
@@ -959,6 +997,7 @@ NUTS_TESTS = {
{ "req context send close", test_req_ctx_send_close },
{ "req context send abort", test_req_ctx_send_abort },
{ "req context send twice", test_req_ctx_send_twice },
+ { "req context send recv abort", test_req_ctx_send_recv_abort },
{ "req context does not poll", test_req_ctx_no_poll },
{ "req context recv close socket", test_req_ctx_recv_close_socket },
{ "req context recv nonblock", test_req_ctx_recv_nonblock },