On Wed, Jun 22, 2022 at 10:42 PM -07, John Fastabend wrote:
> Jakub Sitnicki wrote:
>> Cover the scenario when we cannot insert a socket into the sockmap, because
>> it has it is using ULP. Failed insert should not have any effect on the ULP
>> state. This is a regression test.
>>
>> Signed-off-by: Jakub Sitnicki <jakub@xxxxxxxxxxxxxx>
>> ---
>
> Thanks, looks good. One small nit.
>
>>
>> +#include <netinet/tcp.h>
>> #include "test_progs.h"
>>
>> #define MAX_TEST_NAME 80
>> @@ -92,9 +93,78 @@ static void test_sockmap_ktls_disconnect_after_delete(int family, int map)
>> close(srv);
>> }
>>
>> +static void test_sockmap_ktls_update_fails_when_sock_has_ulp(int family, int map)
>> +{
>> + struct sockaddr_storage addr = {};
>> + socklen_t len = sizeof(addr);
>> + struct sockaddr_in6 *v6;
>> + struct sockaddr_in *v4;
>> + int err, s, zero = 0;
>> +
>> + s = socket(family, SOCK_STREAM, 0);
>> + if (!ASSERT_GE(s, 0, "socket"))
>> + return;
>> +
>> + switch (family) {
>> + case AF_INET:
>> + v4 = (struct sockaddr_in *)&addr;
>> + v4->sin_family = AF_INET;
>> + break;
>> + case AF_INET6:
>> + v6 = (struct sockaddr_in6 *)&addr;
>> + v6->sin6_family = AF_INET6;
>>k+ break;
>> + default:
>> + PRINT_FAIL("unsupported socket family %d", family);
>
> Probably want goto close here right?
Ah, thanks. Sent v2. I hope we can borrow a trick from systemd's book
and adapt __attribute__((cleanup(f))) in the future.
[...]
