Fix wasi-sockets tests #2389
Fix wasi-sockets tests #2389
Conversation
loganek
force-pushed
the
loganek/wasi-sockets-tests-fix
branch
from
fa7ad7b
to
4aa9954
Compare
loganek
force-pushed
the
loganek/wasi-sockets-tests-fix
branch
7 times, most recently
from
e7338ee
to
53d28d1
Compare
| struct sockaddr_in6 addr_ipv6; | ||
| } addr; | ||
| typedef struct { | ||
| struct sockaddr_storage addr; |
There was a problem hiding this comment.
Where is it coming from? Is it still a union?
There was a problem hiding this comment.
The spec doesn't define whether it's an union or not - usually it's a large enough array + common fields (address family) + some padding. sockaddr_storage is a structure that's guaranteed to be big enough to contain all the address types supported (see https://man7.org/linux/man-pages/man7/ipv6.7.html).
| assert(pthread_join(server_thread, NULL) == 0); | ||
| assert(pthread_join(client_thread, NULL) == 0); | ||
|
|
||
| assert(pthread_mutex_destroy(&mut) == 0); | ||
| assert(pthread_cond_destroy(&cond) == 0); |
There was a problem hiding this comment.
Is it destroyed somewhere else? Looks like a leak
There was a problem hiding this comment.
Calling pthread_mutex_destroy on statically initialized mutex is not needed; from the The Linux Programming Interface book:
It is not necessary to call pthread_mutex_destroy() on a mutex that was statically initialized using PTHREAD_MUTEX_INITIALIZER.
I think the same applies to pthread_cond_destroy
| pthread_mutex_unlock(&mut); | ||
| pthread_cond_signal(&cond); |
There was a problem hiding this comment.
How about put pthread_cond_signal before pthread_mutex_unlock? I remember it is generally used by that way.
There was a problem hiding this comment.
Sure, can move it, although I don't think that really matters.
Tests were failing because the right permissions were not provided to iwasm. Also, test failures didn't trigger build failure due to typo - also fixed in this change In addition to that, I fixed a few issues with the test itself: * the `server_init_complete` was not reset early enough causing the client to occasionally assume the server started even though it didn't yet. * set `SO_REUSEADDR` on the server socket so the port can be reused shortly after closing the previous socket * defined receive-send-receive sequence from server to make sure server is alive at the time of sending messages from client
loganek
force-pushed
the
loganek/wasi-sockets-tests-fix
branch
from
53d28d1
to
56e9594
Compare
|
The PR #2364 is waiting for this PR, let's merge this PR first. If there is other issue found, |
Tests were failing because the right permissions were not provided to iwasm. Also, test failures didn't trigger build failure due to typo - also fixed in this change. In addition to that, this PR fixes a few issues with the test itself: * the `server_init_complete` was not reset early enough causing the client to occasionally assume the server started even though it didn't yet * set `SO_REUSEADDR` on the server socket so the port can be reused shortly after closing the previous socket * defined receive-send-receive sequence from server to make sure server is alive at the time of sending message
Tests were failing because the right permissions were not provided to iwasm.
Also, test failures didn't trigger build failure due to typo - also fixed in this change
In addition to that, I fixed a few issues with the test itself:
server_init_completewas not reset early enough causing the client to occasionally assume the server started even though it didn't yet.SO_REUSEADDRon the server socket so the port can be reused shortly after closing the previous socket