[LTP] [PATCH v1] Add epoll_wait07 test

[Andrea Cervesato]

Hi,

On 10/20/22 11:27, Richard Palethorpe wrote:
> Hello,
>
> Andrea Cervesato via ltp <ltp@lists.linux.it> writes:
>
>> This test verifies EPOLLONESHOT functionality.
>>
>> Signed-off-by: Andrea Cervesato <andrea.cervesato@suse.com>
>> ---
>>   .../kernel/syscalls/epoll_wait/.gitignore     |  1 +
>>   .../kernel/syscalls/epoll_wait/epoll_wait07.c | 86 +++++++++++++++++++
>>   2 files changed, 87 insertions(+)
>>   create mode 100644 testcases/kernel/syscalls/epoll_wait/epoll_wait07.c
>>
>> diff --git a/testcases/kernel/syscalls/epoll_wait/.gitignore b/testcases/kernel/syscalls/epoll_wait/.gitignore
>> index 8c5ed7c5c..66ac18ae2 100644
>> --- a/testcases/kernel/syscalls/epoll_wait/.gitignore
>> +++ b/testcases/kernel/syscalls/epoll_wait/.gitignore
>> @@ -4,3 +4,4 @@ epoll_wait03
>>   epoll_wait04
>>   epoll_wait05
>>   epoll_wait06
>> +epoll_wait07
>> diff --git a/testcases/kernel/syscalls/epoll_wait/epoll_wait07.c b/testcases/kernel/syscalls/epoll_wait/epoll_wait07.c
>> new file mode 100644
>> index 000000000..760ab6c10
>> --- /dev/null
>> +++ b/testcases/kernel/syscalls/epoll_wait/epoll_wait07.c
>> @@ -0,0 +1,86 @@
>> +// SPDX-License-Identifier: GPL-2.0-or-later
>> +/*
>> + * Copyright (C) 2022 SUSE LLC Andrea Cervesato <andrea.cervesato@suse.com>
>> + */
>> +
>> +/*\
>> + * [Description]
>> + *
>> + * Verify that EPOLLONESHOT is correctly handled by epoll_wait.
>> + * We open a channel, write on it multiple times and we verify that EPOLLOUT
>> + * has been received only once.
> Without EPOLLONESHOT or EPOLLET, EPOLLOUT will fire constantly on a pipe
> who's buffer is not full. Writing to it will have no effect unless you
> listen on the other end for EPOLLIN.
>
> So in this test you don't need to write to the pipe unless I am
> mistaken.
>
I see there's not a real difference between EPOLLIN or EPOLLOUT in this 
case. We can listen to EPOLLIN, write to pipe stdin buffer once and call 
epoll_wait two times. The result will be the same, since EPOLLIN event 
has been triggered only once. We have many options, I choose the one 
which forces two EPOLLOUT events, but the same can be done with EPOLLIN 
in this case.

To check, you can remove EPOLLONESHOT flag and run test without it.

>> + */
>> +
>> +#include <poll.h>
>> +#include <sys/epoll.h>
>> +#include "tst_test.h"
>> +
>> +#define WRITE_SIZE 2048
>> +
>> +static int fds[2];
>> +static int epfd;
>> +
>> +static void cleanup(void)
>> +{
>> +	if (epfd > 0)
>> +		SAFE_CLOSE(epfd);
>> +
>> +	if (fds[0] > 0)
>> +		SAFE_CLOSE(fds[0]);
>> +
>> +	if (fds[1] > 0)
>> +		SAFE_CLOSE(fds[1]);
>> +}
>> +
>> +static void run(void)
>> +{
>> +	int res;
>> +	char buff[WRITE_SIZE];
>> +	struct epoll_event evt_receive;
>> +	struct epoll_event evt_request;
>> +
>> +	SAFE_PIPE(fds);
>> +
>> +	evt_request.events = EPOLLOUT | EPOLLONESHOT;
>> +	evt_request.data.fd = fds[1];
>> +
>> +	epfd = epoll_create(2);
>> +	if (epfd == -1)
>> +		tst_brk(TBROK | TERRNO, "fail to create epoll
>> instance");
> Same as other test, please use the SAFE_EPOLL_* functions I just merged.
>
Andrea