[LTP] [PATCH] syscalls/read05: Test read with zero count

[liaohj.jy@fujitsu.com]

Hi Richard


> Hello,
> 
> Liao Huangjie <liaohj.jy@fujitsu.com> writes:
> 
>> From: Huangjie Liao <liaohj.jy@fujitsu.com>
>>
>> Signed-off-by: Huangjie Liao <liaohj.jy@fujitsu.com>
>> ---
>>   runtest/syscalls                          |  1 +
>>   testcases/kernel/syscalls/read/.gitignore |  1 +
>>   testcases/kernel/syscalls/read/read05.c   | 44 +++++++++++++++++++++++++++++++
>>   3 files changed, 46 insertions(+)
>>   create mode 100644 testcases/kernel/syscalls/read/read05.c
>>
>> diff --git a/runtest/syscalls b/runtest/syscalls
>> index 3847e8a..0bd47fc 100644
>> --- a/runtest/syscalls
>> +++ b/runtest/syscalls
>> @@ -1094,6 +1094,7 @@ read01 read01
>>   read02 read02
>>   read03 read03
>>   read04 read04
>> +read05 read05
>>   
>>   readahead01 readahead01
>>   readahead02 readahead02
>> diff --git a/testcases/kernel/syscalls/read/.gitignore b/testcases/kernel/syscalls/read/.gitignore
>> index e89c008..69d97b3 100644
>> --- a/testcases/kernel/syscalls/read/.gitignore
>> +++ b/testcases/kernel/syscalls/read/.gitignore
>> @@ -2,3 +2,4 @@
>>   /read02
>>   /read03
>>   /read04
>> +/read05
>> diff --git a/testcases/kernel/syscalls/read/read05.c b/testcases/kernel/syscalls/read/read05.c
>> new file mode 100644
>> index 0000000..83ac763
>> --- /dev/null
>> +++ b/testcases/kernel/syscalls/read/read05.c
>> @@ -0,0 +1,44 @@
>> +// SPDX-License-Identifier: GPL-2.0-or-later
>> +/*
>> + * Copyright (c) 2022 FUJITSU LIMITED. All Rights Reserved.
>> + * Author: Huangjie Liao <liaohj.jy@fujitsu.com>
>> + */
>> +
>> +/*\
>> + * [Description]
>> + *
>> + * In the absence of any errors, a read() with a count of 0 returns
>>   zero.
> 
> This test is quite similar to read04. You could modify that test to read
> multiple times from 0 bytes to the total number of bytes available.
OK
> 
>> + */
>> +
>> +#include <errno.h>
>> +#include "tst_test.h"
>> +
>> +#define SIZE 512
> 
> This could be set to 2*getpagesize() at runtime. Then you can try
> reading every length between 0 and 2*getpagesize().

Then we can increase 2kb length every time, ie  0 2k 4k 6k 8kb.

But I have a question that used which patten to fill the 8KB space.

Maybe write  "A" in the [0-4k] and "B" in the [4-8k]. Or use "A" to
fill the whole 8k space?

> 
> Errors tend to be made on page boundaries and similar.
> 
>> +
>> +static int fd;
>> +static char buf[SIZE];
>> +
>> +static void verify_read(void)
>> +{
>> +	TST_EXP_VAL(read(fd, buf, 0), 0);
>> +}
>> +
>> +static void setup(void)
>> +{
>> +	memset(buf, '*', SIZE);
>> +	fd = SAFE_OPEN("testfile", O_RDWR | O_CREAT, 0700);
>> +	SAFE_WRITE(1, fd, buf, SIZE);
>> +}
>> +
>> +static void cleanup(void)
>> +{
>> +	if (fd > 0)
>> +		SAFE_CLOSE(fd);
>> +}
>> +
>> +static struct tst_test test = {
>> +	.test_all = verify_read,
>> +	.setup = setup,
>> +	.cleanup = cleanup,
>> +	.needs_tmpdir = 1,
> 
> It would increase test coverage greatly to run this on all filesystems
> (i.e. set .all_filesystems = 1). For an example see rename10.

Yes.

Best Regards
> 
>> +};
>> -- 
>> 1.8.3.1
> 
>