All of lore.kernel.org
 help / color / mirror / Atom feed
From: Yang Xu <xuyang2018.jy@cn.fujitsu.com>
To: ltp@lists.linux.it
Subject: [LTP] [PATCH v2] syscalls/copy_file_range02: skip if cross-fs isn't supported
Date: Tue, 22 Oct 2019 14:37:30 +0800	[thread overview]
Message-ID: <92e994a3-321f-1ef0-8b81-4262f317a461@cn.fujitsu.com> (raw)
In-Reply-To: <20191022061328.GA9267@dell5510>


on 2019/10/22 14:14, Petr Vorel wrote:

> Hi Xu, others,
>
> [ Cc Amir and linux-fsdevel ].
>
>> on 2019/09/30 17:36, Petr Vorel wrote:
>>> copy_file_range02 was written to verify copy_file_range() v5.3 changes.
>>> Detect it via cross-filesystem copy_file_range() functionality, so that we
>>> cover also backports to stable/enterprise distro kernels (if backported,
>>> it should be with all those API fixes).
>>> Missing these API fixes is detected by errno changes introduced by
>>> This fixes errors caused by commits from v5.3-rc1:
>>> 5dae222a5ff0 ("vfs: allow copy_file_range to copy across devices")
>>> 96e6e8f4a68d ("vfs: add missing checks to copy_file_range")
>>> This check requires to put back into copy_file_range02 .mount_device = 1
>>> and .mntpoint = MNTPOINT (but .all_filesystems = 1 is obviously not needed).
>> Hi Petr
>>     Why we must put back .mount_device and .mntpoint = MNTPOINT?
>> copy_file_range02 was not only written to verify copy_file_range() v5.3 changes
>> and it also tests other errors before v5.3.
> This fix was based on Amir's suggestion [1], he states the opposite:
>
> IIUC, copy_file_range02 was written after v5.3 changes to verify that
> copy_file_range
> stays unbroken.
> As such, I would suggest that you check if kernel supports cross-fs copy, like
> copy_file_range01 does and if it doesn't, skip the test entirely.
> If some one ever backports cross-fs copy to any distro stable kernel, then one
> would better also backkport all of those API fixes, otherwise test will fail.

Hi Petr

In copy_file_range01, I split it into two tests(not cross-filesytem test and cross-filesystem test)
 ??????? {FILE_DEST_PATH,? 0, "non cross-device"},
 ??????? {FILE_MNTED_PATH, 1, "cross-device"},

if kernel doesn't support cross-fs copy, the second test skips and the first still can run.

So for copy_file_range02, it should be same but not for entire test.

>
>
>> I think cross-filesystem copy_file_range is a kernel action change and then I
>> put it into copy_file_range01.c. So copy_file_range02.c doesn't test EXDEV error .
>> Also since commit v5.3-rc1 two commit, immutable file(EPERM)?swap file(ETXTBSY)?
>> overlaping range(EINVAL), max length lenght(EOVERFLOW),max file size(EFBIG) these
>> check have been add. But other errors still existed before this two commits such as:
>> copy contents to file open as readonly *    -> EBADF
>> Now, before v5.3-rc1, copy_file_range02.c  is notrun that we don't do error check.
>> It is unreasonable.
> So, do you suggest to test EBADF for all versions? Or something else?

Yes.? I think we can use cross_sup flag to decide whether test immutable file(EPERM)?swap file(ETXTBSY)?
overlaping range(EINVAL), max length lenght(EOVERFLOW),max file size(EFBIG). If? cross_sup
is equal to 1, test them. If not, don't test them as your patch do but not for all errors.

But for other errors, I think we can still test them for all versions such as pipe, char,and block dev.

>
>> ps:
>> copy_file_range newest man-pages
>> https://github.com/mkerrisk/man-pages/commit/88e75e2c56a68eaf8fcf662a63b802fdf77a4017
> Yep, Amir planned to fix it :).
>
>> Thanks
>> Yang Xu
> [1] http://lists.linux.it/pipermail/ltp/2019-September/013697.html
>
>>> + Remove few unused imports.
>>> Suggested-by: Amir Goldstein <amir73il@gmail.com>
>>> Signed-off-by: Petr Vorel <pvorel@suse.cz>
>>> ---
>>> Changes v1->v2:
>>> pass the source and destination as parameters to
>>> verify_cross_fs_copy_support(), remove bogus setup checks
>>> (Suggested by Cyril).
>>> Kind regards,
>>> Petr
>>>    .../copy_file_range/copy_file_range.h         | 23 ++++++++++++++++---
>>>    .../copy_file_range/copy_file_range01.c       | 22 ++----------------
>>>    .../copy_file_range/copy_file_range02.c       | 11 ++++++++-
>>>    3 files changed, 32 insertions(+), 24 deletions(-)
>>> diff --git a/testcases/kernel/syscalls/copy_file_range/copy_file_range.h b/testcases/kernel/syscalls/copy_file_range/copy_file_range.h
>>> index 40d05d653..1d80ab0f7 100644
>>> --- a/testcases/kernel/syscalls/copy_file_range/copy_file_range.h
>>> +++ b/testcases/kernel/syscalls/copy_file_range/copy_file_range.h
>>> @@ -7,9 +7,7 @@
>>>    #ifndef __COPY_FILE_RANGE_H__
>>>    #define __COPY_FILE_RANGE_H__
>>> -#include <stdbool.h>
>>> -#include <unistd.h>
>>> -#include <sys/sysmacros.h>
>>> +#include <stdio.h>
>>>    #include "lapi/syscalls.h"
>>>    #include "lapi/fs.h"
>>> @@ -62,4 +60,23 @@ static int sys_copy_file_range(int fd_in, loff_t *off_in,
>>>    	return -1;
>>>    }
>>> +static inline int verify_cross_fs_copy_support(const char *path_in, const char *path_out)
>>> +{
>>> +	int i, fd, fd_test;
>>> +
>>> +	fd = SAFE_OPEN(path_in, O_RDWR | O_CREAT, 0664);
>>> +	/* Writing page_size * 4 of data into test file */
>>> +	for (i = 0; i < (int)(getpagesize() * 4); i++)
>>> +		SAFE_WRITE(1, fd, CONTENT, CONTSIZE);
>>> +
>>> +	fd_test = SAFE_OPEN(path_out, O_RDWR | O_CREAT, 0664);
>>> +	TEST(sys_copy_file_range(fd, 0, fd_test, 0, CONTSIZE, 0));
>>> +
>>> +	SAFE_CLOSE(fd_test);
>>> +	remove(FILE_MNTED_PATH);
>>> +	SAFE_CLOSE(fd);
>>> +
>>> +	return TST_ERR == EXDEV ? 0 : 1;
>>> +}
>>> +
>>>    #endif /* __COPY_FILE_RANGE_H__ */
>>> diff --git a/testcases/kernel/syscalls/copy_file_range/copy_file_range01.c b/testcases/kernel/syscalls/copy_file_range/copy_file_range01.c
>>> index ec55e5da1..6097c85b3 100644
>>> --- a/testcases/kernel/syscalls/copy_file_range/copy_file_range01.c
>>> +++ b/testcases/kernel/syscalls/copy_file_range/copy_file_range01.c
>>> @@ -16,8 +16,6 @@
>>>    #define _GNU_SOURCE
>>> -#include <stdio.h>
>>> -#include <stdlib.h>
>>>    #include "tst_test.h"
>>>    #include "tst_safe_stdio.h"
>>>    #include "copy_file_range.h"
>>> @@ -179,7 +177,7 @@ static void copy_file_range_verify(unsigned int n)
>>>    	if (tc->flags && !cross_sup) {
>>>    		tst_res(TCONF,
>>> -			"copy_file_range doesn't support cross-device, skip it");
>>> +			"copy_file_range() doesn't support cross-device, skip it");
>>>    		return;
>>>    	}
>>> @@ -215,25 +213,9 @@ static void copy_file_range_verify(unsigned int n)
>>>    static void setup(void)
>>>    {
>>> -	int i, fd, fd_test;
>>> -
>>>    	syscall_info();
>>> -
>>>    	page_size = getpagesize();
>>> -	cross_sup = 1;
>>> -	fd = SAFE_OPEN(FILE_SRC_PATH, O_RDWR | O_CREAT, 0664);
>>> -	/* Writing page_size * 4 of data into test file */
>>> -	for (i = 0; i < (int)(page_size * 4); i++)
>>> -		SAFE_WRITE(1, fd, CONTENT, CONTSIZE);
>>> -
>>> -	fd_test = SAFE_OPEN(FILE_MNTED_PATH, O_RDWR | O_CREAT, 0664);
>>> -	TEST(sys_copy_file_range(fd, 0, fd_test, 0, CONTSIZE, 0));
>>> -	if (TST_ERR == EXDEV)
>>> -		cross_sup = 0;
>>> -
>>> -	SAFE_CLOSE(fd_test);
>>> -	remove(FILE_MNTED_PATH);
>>> -	SAFE_CLOSE(fd);
>>> +	cross_sup = verify_cross_fs_copy_support(FILE_SRC_PATH, FILE_MNTED_PATH);
>>>    }
>>>    static void cleanup(void)
>>> diff --git a/testcases/kernel/syscalls/copy_file_range/copy_file_range02.c b/testcases/kernel/syscalls/copy_file_range/copy_file_range02.c
>>> index d6e843ee4..6e385adbd 100644
>>> --- a/testcases/kernel/syscalls/copy_file_range/copy_file_range02.c
>>> +++ b/testcases/kernel/syscalls/copy_file_range/copy_file_range02.c
>>> @@ -49,6 +49,7 @@ static int fd_blkdev;
>>>    static int fd_chrdev;
>>>    static int fd_fifo;
>>>    static int fd_copy;
>>> +static int need_unlink;
>>>    static int chattr_i_nsup;
>>>    static int swap_nsup;
>>> @@ -160,7 +161,8 @@ static void cleanup(void)
>>>    		SAFE_CLOSE(fd_dup);
>>>    	if (fd_copy > 0)
>>>    		SAFE_CLOSE(fd_copy);
>>> -	SAFE_UNLINK(FILE_FIFO);
>>> +	if (need_unlink > 0)
>>> +		SAFE_UNLINK(FILE_FIFO);
>>>    }
>>>    static void setup(void)
>>> @@ -168,6 +170,10 @@ static void setup(void)
>>>    	syscall_info();
>>>    	char dev_path[1024];
>>> +	if (!verify_cross_fs_copy_support(FILE_SRC_PATH, FILE_MNTED_PATH))
>>> +		tst_brk(TCONF,
>>> +			"copy_file_range() doesn't support cross-device, skip it");
>>> +
>>>    	if (access(FILE_DIR_PATH, F_OK) == -1)
>>>    		SAFE_MKDIR(FILE_DIR_PATH, 0777);
>>>    	/*
>>> @@ -177,6 +183,7 @@ static void setup(void)
>>>    	loop_devn = tst_find_free_loopdev(dev_path, sizeof(dev_path));
>>>    	SAFE_MKNOD(FILE_FIFO, S_IFIFO | 0777, 0);
>>> +	need_unlink = 1;
>>>    	fd_src    = SAFE_OPEN(FILE_SRC_PATH, O_RDWR | O_CREAT, 0664);
>>>    	fd_dest   = SAFE_OPEN(FILE_DEST_PATH, O_RDWR | O_CREAT, 0664);
>>> @@ -223,6 +230,8 @@ static struct tst_test test = {
>>>    	.tcnt = ARRAY_SIZE(tcases),
>>>    	.setup = setup,
>>>    	.cleanup = cleanup,
>>> +	.mount_device = 1,
>>> +	.mntpoint = MNTPOINT,
>>>    	.needs_root = 1,
>>>    	.needs_tmpdir = 1,
>>>    	.test_variants = TEST_VARIANTS,
>
>
>


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20191022/b876ed27/attachment-0001.htm>

      reply	other threads:[~2019-10-22  6:37 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-09-30  9:36 [LTP] [PATCH v2] syscalls/copy_file_range02: skip if cross-fs isn't supported Petr Vorel
2019-10-22  2:50 ` Yang Xu
2019-10-22  6:13   ` Petr Vorel
2019-10-22  6:13     ` Petr Vorel
2019-10-22  6:37     ` Yang Xu [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=92e994a3-321f-1ef0-8b81-4262f317a461@cn.fujitsu.com \
    --to=xuyang2018.jy@cn.fujitsu.com \
    --cc=ltp@lists.linux.it \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.