[v1] fsconfig04.c: Check FSCONFIG_SET_PATH

Fixes: 1169

The fsconfig01.c does not test if FSCONFIG_SET_PATH have any effect,
most of the calls there just set dummy "sync" parameter. This test
case aims to verify if the FSCONFIG_SET_PATH operation can be used
to dynamically change the external journal device of ext3 or ext4
filesystem.

Signed-off-by: Wei Gao <wegao@suse.com>
---
 runtest/syscalls                              |   1 +
 testcases/kernel/syscalls/fsconfig/.gitignore |   1 +
 .../kernel/syscalls/fsconfig/fsconfig04.c     | 169 ++++++++++++++++++
 3 files changed, 171 insertions(+)
 create mode 100644 testcases/kernel/syscalls/fsconfig/fsconfig04.c

Hi Wei,

I'm sorry not a real review, just point out obvious API errors.

> Fixes: 1169
Fixes: #1169

> The fsconfig01.c does not test if FSCONFIG_SET_PATH have any effect,
> most of the calls there just set dummy "sync" parameter. This test
> case aims to verify if the FSCONFIG_SET_PATH operation can be used
> to dynamically change the external journal device of ext3 or ext4
> filesystem.

...
> diff --git a/testcases/kernel/syscalls/fsconfig/fsconfig04.c b/testcases/kernel/syscalls/fsconfig/fsconfig04.c
> new file mode 100644
> index 000000000..119b8b175
> --- /dev/null
> +++ b/testcases/kernel/syscalls/fsconfig/fsconfig04.c
> @@ -0,0 +1,169 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +/*
> + * Copyright (c) 2025 Wei Gao <wegao@suse.com>
> + *
This '*' should be '/*\' to be docparse.

> + * This test aims to validate the functionality of fsconfig(FSCONFIG_SET_PATH) in
> + * dynamically altering the external journal device of a ext3 or ext4 filesystem.
> + * Case acquires three loop devices (dev0, dev1, dev2), it formats dev1 and dev2
> + * as external journal devices using the -O journal_dev option and assigns them
> + * the same UUID. Then formats dev0 (the main filesystem) multiple times, first
> + * associating it with dev1, then change to dev2, finally back to dev1 again as
> + * an external journal using the -J device= option.
> + *
> + * 2 workarounds in this case need mention:
And here should be new line, right?

> + * - To avoid "journal UUID does not match" error when switch external journal device
> + *   we have to assign same UUID to dev1/dev2
> + * - Before fsconfig test we have to formats dev0 associating to dev1->dev2->dev1,
> + *   this will make sure both dev1/2's supper block contain correct content. Otherwise
> + *   you will encounter error such as "EXT4-fs (loop0): External journal has more than
> + *   one user (unsupported) - 0" when switch external journal device using fsconfig.
> + */
> +
> +#include "tst_test.h"
> +#include "lapi/fsmount.h"

> +#include "old/old_device.h"
Why this?

> +
> +#define MNTPOINT	"mntpoint"
> +#define LOOP_DEV_SIZE 10
> +#define UUID "d73c9e5e-97e4-4a9c-b17e-75a931b02660"
> +
> +static int fd;
> +static const char *dev;
> +static char dev0[NAME_MAX];
> +static char dev1[NAME_MAX];
> +static char dev2[NAME_MAX];
> +
> +static void cleanup(void)
> +{
> +	if (fd != -1)
> +		SAFE_CLOSE(fd);
> +
> +	TST_EXP_PASS_SILENT(tst_detach_device(dev1));
> +	TST_EXP_PASS_SILENT(tst_detach_device(dev2));
> +}
> +
> +static void setup(void)
> +{
> +	int ret = tst_system("tune2fs");
> +
> +	if (WEXITSTATUS(ret) == 127)
> +		tst_brk(TCONF, "tune2fs does not exist");

We have .needs_cmds

> +
> +	strcpy(dev0, tst_device->dev);
> +	dev = tst_acquire_loop_device(LOOP_DEV_SIZE, "dev1_file");
> +	if (!dev)
> +		tst_brk(TBROK, "acquire loop dev1 failed");
> +
> +	strcpy(dev1, dev);
> +	dev = NULL;
> +
> +	dev = tst_acquire_loop_device(LOOP_DEV_SIZE, "dev2_file");
> +	if (!dev)
> +		tst_brk(TBROK, "acquire loop dev2 failed");
> +	strcpy(dev2, dev);
> +
> +	const char *const *mkfs_opts_set_UUID;
> +	const char *const *mkfs_opts_set_journal_dev1;
> +	const char *const *mkfs_opts_set_journal_dev2;
> +
> +	mkfs_opts_set_UUID = (const char *const []) {"-F",
> +		"-U",
> +		UUID,
> +		"-O journal_dev",
> +		NULL};
very nit: why the args cannot be on a single line?

> +
> +	char device_option_dev1[PATH_MAX];
> +	char device_option_dev2[PATH_MAX];
> +
> +	sprintf(device_option_dev1, "device=%s", dev1);
> +	sprintf(device_option_dev2, "device=%s", dev2);
> +
> +	mkfs_opts_set_journal_dev1 = (const char *const []) {"-F",
> +		"-J",
> +		device_option_dev1,
> +		NULL};
> +
> +	mkfs_opts_set_journal_dev2 = (const char *const []) {"-F",
> +		"-J",
> +		device_option_dev2,
> +		NULL};
> +
> +	SAFE_MKFS(dev1, tst_device->fs_type, mkfs_opts_set_UUID, NULL);
> +	SAFE_MKFS(dev2, tst_device->fs_type, mkfs_opts_set_UUID, NULL);
> +	SAFE_MKFS(dev0, tst_device->fs_type, mkfs_opts_set_journal_dev1, NULL);
> +	SAFE_MKFS(dev0, tst_device->fs_type, mkfs_opts_set_journal_dev2, NULL);
> +	SAFE_MKFS(dev0, tst_device->fs_type, mkfs_opts_set_journal_dev1, NULL);
> +
> +}
> +
> +static void run(void)
> +{
> +	TEST(fd = fsopen(tst_device->fs_type, 0));
> +	if (fd == -1)
> +		tst_brk(TBROK | TTERRNO, "fsopen() failed");
> +
> +	TEST(fsconfig(fd, FSCONFIG_SET_STRING, "source", dev0, 0));
> +	if (TST_RET == -1)
> +		tst_brk(TFAIL | TTERRNO, "fsconfig(FSCONFIG_SET_STRING) failed");
> +
> +	TEST(fsconfig(fd, FSCONFIG_SET_PATH, "journal_path", dev2, 0));
> +	if (TST_RET == -1) {
> +		if (TST_ERR == EOPNOTSUPP)
> +			tst_brk(TCONF, "fsconfig(FSCONFIG_SET_PATH) not supported");
> +		else
> +			tst_brk(TFAIL | TTERRNO, "fsconfig(FSCONFIG_SET_PATH) failed");
> +	}
> +
> +	TEST(fsconfig(fd, FSCONFIG_CMD_CREATE, NULL, NULL, 0));
> +	if (TST_RET == -1)
> +		tst_brk(TFAIL | TTERRNO, "fsconfig(FSCONFIG_CMD_CREATE) failed");
> +
> +	char loop_name[NAME_MAX];
> +	char path[PATH_MAX];
> +	char device_str[NAME_MAX];
> +	unsigned int major, minor, device_num;
> +	unsigned int found = 0;
> +
> +	int ret = sscanf(dev2, "/dev/%s", loop_name);
> +
> +	if (ret != 1)
> +		tst_brk(TFAIL | TTERRNO, "can not parse loop_name");
> +
> +	sprintf(path, "/sys/block/%s/dev", loop_name);
> +	SAFE_FILE_SCANF(path, "%u:%u", &major, &minor);
> +	device_num = (major << 8) | minor;
> +	sprintf(device_str, "0x%04X", device_num);
> +
> +	char line[PATH_MAX];
> +	FILE *tune2fs;
> +
> +	sprintf(path, "tune2fs -l %s 2>&1", dev0);
> +	tune2fs = popen(path, "r");
> +
> +	while (fgets(line, PATH_MAX, tune2fs)) {
> +		if (*line && strstr(line, "Journal device:") && strstr(line, device_str)) {
> +			found = 1;
> +			break;
> +		}
> +	}
> +
> +	if (found == 1)
> +		tst_res(TPASS, "Journal device set pass");
Maybe "Device found in journal" ?
> +	else
> +		tst_res(TFAIL, "Journal device set failed");
Maybe "Device not found in journal" ?

> +
> +	pclose(tune2fs);
> +	SAFE_CLOSE(fd);
> +}
> +
> +static struct tst_test test = {
> +	.test_all = run,
> +	.setup = setup,
> +	.cleanup = cleanup,
> +	.needs_root = 1,
> +	.format_device = 1,
> +	.mntpoint = MNTPOINT,
> +	.all_filesystems = 1,
> +	.skip_filesystems = (const char *const []){"ntfs", "vfat", "exfat",
> +		"ext2", "tmpfs", "xfs", "btrfs", NULL},
So you skip nearly everything, right?
Why not use .mount_device = 1 ? That would mount the default filesystem.

Kind regards,
Petr

> +};

Hi Wei,

> On Thu, May 22, 2025 at 10:56:37PM +0200, Petr Vorel wrote:
> > Hi Wei,

> > I'm sorry not a real review, just point out obvious API errors.
> Thanks for pointing that out :)


> > > +#include "lapi/fsmount.h"

> > > +#include "old/old_device.h"
> > Why this?
> tst_acquire_loop_device is needed in this case, and i saw following test
> case using old_device.h, so i just copy it into current case.
> testcases/lib/tst_device.c

Indeed it also uses it, maybe there is really no other alternative for new API.

> > > +
> > > +
> > > +	.all_filesystems = 1,
> > > +	.skip_filesystems = (const char *const []){"ntfs", "vfat", "exfat",
> > > +		"ext2", "tmpfs", "xfs", "btrfs", NULL},
> > So you skip nearly everything, right?
> Yes, ONLY ext3 and ext4 support journal setup

> > Why not use .mount_device = 1 ? That would mount the default filesystem.
> This case need format default device with ext3 opts='-F -J device=/dev/loop1' firstly,
> so the device can not be mounted before format.

Thanks for info.

Kind regards,
Petr

> > Kind regards,
> > Petr

On Thu, May 22, 2025 at 10:56:37PM +0200, Petr Vorel wrote:
> Hi Wei,
> 
> I'm sorry not a real review, just point out obvious API errors.
Thanks for pointing that out :)

> 
> > +#include "lapi/fsmount.h"
> 
> > +#include "old/old_device.h"
> Why this?
tst_acquire_loop_device is needed in this case, and i saw following test
case using old_device.h, so i just copy it into current case.
testcases/lib/tst_device.c
> 
> > +
> > +
> > +	.all_filesystems = 1,
> > +	.skip_filesystems = (const char *const []){"ntfs", "vfat", "exfat",
> > +		"ext2", "tmpfs", "xfs", "btrfs", NULL},
> So you skip nearly everything, right?
Yes, ONLY ext3 and ext4 support journal setup

> Why not use .mount_device = 1 ? That would mount the default filesystem.
This case need format default device with ext3 opts='-F -J device=/dev/loop1' firstly,
so the device can not be mounted before format.
> 
> Kind regards,
> Petr
> 
> > +};