All of lore.kernel.org
 help / color / mirror / Atom feed
From: chrubis@suse.cz
To: Matus Marhefka <mmarhefk@redhat.com>
Cc: ltp-list@lists.sourceforge.net
Subject: Re: [LTP] [PATCH 1/2 v2] containers: added mountns/mountns03.c
Date: Wed, 24 Sep 2014 09:56:05 +0200	[thread overview]
Message-ID: <20140924075604.GB10825@rei> (raw)
In-Reply-To: <1409319967-12730-1-git-send-email-mmarhefk@redhat.com>

Hi!
> +static void setup(void)
> +{
> +	tst_require_root(NULL);
> +	check_newns();	/* from mountns_helper.h */

Please remove this comment.

> +	tst_tmpdir();
> +	SAFE_MKDIR(cleanup, DIRA, 0777);
> +	SAFE_MKDIR(cleanup, DIRB, 0777);
> +	SAFE_MKDIR(cleanup, DIRC, 0777);
> +	SAFE_MKDIR(cleanup, DIRD, 0777);
> +	SAFE_TOUCH(cleanup, DIRA"/A", 0, NULL);
> +	SAFE_TOUCH(cleanup, DIRB"/B", 0, NULL);
> +	SAFE_TOUCH(cleanup, DIRC"/C", 0, NULL);
> +	SAFE_TOUCH(cleanup, DIRD"/D", 0, NULL);
> +}
> +
> +int child_func(void *arg)
> +{
> +	char buf;
> +
> +	/* bind mounts DIRA to DIRB making contents of DIRA visible
> +	 * in DIRB */
> +	if (mount(DIRA, DIRB, "none", MS_BIND, NULL) == -1) {
> +		perror("mount");
> +		write(pipefd1[1], "1", 1);
> +		return 1;
> +	}
> +
> +	/* makes mount DIRB a slave of DIRA (all slave mounts have
> +	 * a master mount which is a shared mount) */
> +	if (mount("none", DIRB, "none", MS_SLAVE, NULL) == -1) {
> +		perror("mount");
> +		umount(DIRB);
> +		write(pipefd1[1], "1", 1);
> +		return 1;
> +	}
> +
> +	/* bind mounts DIRC to DIRA making contents of DIRC visible
> +	 * in DIRA */
> +	if (mount(DIRC, DIRA, "none", MS_BIND, NULL) == -1) {
> +		perror("mount");
> +		umount(DIRB);
> +		write(pipefd1[1], "1", 1);
> +		return 1;
> +	}
> +
> +	/* as DIRB is a slave of DIRA, previous mount (DIRC to DIRA)
> +	 * should be also propageted into DIRB */
> +	if (access(DIRB"/C", F_OK) != -1)
> +		tst_resm(TPASS, "propagation to slave mount passed");
> +	else
> +		tst_resm(TFAIL, "propagation to slave mount failed");
> +
> +	/* bind mounts DIRD to DIRB making contents of DIRD visible
> +	 * in DIRB (should not propagate to DIRA as DIRB is its slave) */
> +	if (mount(DIRD, DIRB, "none", MS_BIND, NULL) == -1) {
> +		perror("mount");
> +		umount(DIRA);
> +		umount(DIRB);
> +		write(pipefd1[1], "1", 1);
> +		return 1;
> +	}
> +
> +	/* tells parent to stop waiting and continue */
> +	write(pipefd1[1], "0", 1);
> +
> +	/* waits for parent approval to terminate */
> +	read(pipefd2[0], &buf, 1);

We have a child-parent synchronization functions in the LTP test
library. Have look at include/tst_checkpoint.h and
lib/tests/tst_checkpoint* and make use of it.

In short these two statements would became:

TST_CHECKPOINT_SIGNAL_PARENT(&checkpoint);
TST_CHECKPOINT_CHILD_WAIT(&checkpoint);

And you would have to declare checkpoint structure and initialize the
checkpoint in setup with:

struct tst_checkpoint checkpoint;

...

static void setup(void)
{
	...
	TST_CHECKPOINT_INIT(&checkpoint);
	...
}

The difference between doing write() and checkpoint interfaces is that
checkpoint checks return values from write(), suppport timeouts, etc.

> +	umount(DIRB);
> +	umount(DIRA);
> +	umount(DIRB);
> +	return 0;
> +}
> +
> +static void test(void)
> +{
> +	int status, pass;
> +	char buf;
> +
> +	/* creates a pipe for synchronization between parent and child */
> +	SAFE_PIPE(cleanup, pipefd1);
> +	SAFE_PIPE(cleanup, pipefd2);
> +
> +	/* unshares the mount ns */
> +	if (unshare(CLONE_NEWNS) == -1)
> +		tst_brkm(TBROK | TERRNO, cleanup, "unshare failed");
> +	/* makes sure parent mounts/umounts have no effect on a real system */
> +	SAFE_MOUNT(cleanup, "none", "/", "none", MS_REC|MS_PRIVATE, NULL);
> +
> +	/* bind mounts DIRA to itself */
> +	SAFE_MOUNT(cleanup, DIRA, DIRA, "none", MS_BIND, NULL);
> +	/* bind mounts DIRB to itself */
> +	SAFE_MOUNT(cleanup, DIRB, DIRB, "none", MS_BIND, NULL);
> +
> +	/* makes mount DIRA shared */
> +	SAFE_MOUNT(cleanup, "none", DIRA, "none", MS_SHARED, NULL);
> +	/* makes mount DIRB shared */
> +	SAFE_MOUNT(cleanup, "none", DIRB, "none", MS_SHARED, NULL);
> +
> +	if (do_clone_tests(CLONE_NEWNS, child_func, NULL, NULL, NULL) == -1)
> +		tst_brkm(TBROK | TERRNO, cleanup, "clone failed");
> +
> +	/* waits for child bind mount */
> +	SAFE_READ(cleanup, 0, pipefd1[0], &buf, 1);

Here you should do TST_CHECKPOINT_PARENT_WAIT(cleanup, &checkpoint);

> +	/* checks that mount in slave (DIRB) doesn't propagate to shared
> +	 * mount (DIRA) */
> +	if (access(DIRA"/D", F_OK) == -1)
> +		pass = 1;
> +	else
> +		pass = 0;
> +
> +	/* tells child to terminate (if no error occured in child) */
> +	SAFE_WRITE(cleanup, 0, pipefd2[1], "0", 1);

Here you should do TST_CHECKPOINT_SIGNAL_CHILD(cleanup, &checkpoint);

> +	SAFE_WAIT(cleanup, &status);
> +
> +	if (WIFEXITED(status) && WEXITSTATUS(status) != 0)
> +		tst_brkm(TBROK, cleanup, "child failed");
> +
> +	if (WIFSIGNALED(status)) {
> +		tst_resm(TFAIL, "child was killed with signal %s",
> +			 tst_strsig(WTERMSIG(status)));
> +		return;
> +	}
> +
> +	close(pipefd1[0]);
> +	close(pipefd1[1]);
> +	close(pipefd2[0]);
> +	close(pipefd2[1]);
> +	SAFE_UMOUNT(cleanup, DIRA);
> +	SAFE_UMOUNT(cleanup, DIRB);
> +
> +	if (pass)
> +		tst_resm(TPASS, "propagation from slave mount passed");
> +	else
> +		tst_resm(TFAIL, "propagation form slave mount failed");
> +}
> +
> +int main(int argc, char *argv[])
> +{
> +	const char *msg;
> +	int lc;
> +
> +	msg = parse_opts(argc, argv, NULL, NULL);
> +	if (msg != NULL)
> +		tst_brkm(TBROK, NULL, "OPTION PARSING ERROR - %s", msg);
> +
> +	setup();
> +
> +	for (lc = 0; TEST_LOOPING(lc); lc++)
> +		test();
> +
> +	cleanup();
> +	tst_exit();
> +}
> +
> +#else /* MS_SHARED && MS_PRIVATE && MS_REC && MS_SLAVE */

Have we settled if we should go with ifdefs or with defining the
constatns in include/lapi/?

I remeber that we disscussed it in one of the latter patches.

> +int main(void)
> +{
> +	tst_brkm(TCONF, NULL, "needed mountflags are not defined");
> +}
> +#endif

-- 
Cyril Hrubis
chrubis@suse.cz

------------------------------------------------------------------------------
Meet PCI DSS 3.0 Compliance Requirements with EventLog Analyzer
Achieve PCI DSS 3.0 Compliant Status with Out-of-the-box PCI DSS Reports
Are you Audit-Ready for PCI DSS 3.0 Compliance? Download White paper
Comply to PCI DSS 3.0 Requirement 10 and 11.5 with EventLog Analyzer
http://pubads.g.doubleclick.net/gampad/clk?id=154622311&iu=/4140/ostg.clktrk
_______________________________________________
Ltp-list mailing list
Ltp-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ltp-list

  parent reply	other threads:[~2014-09-24  7:56 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-08-26  9:59 [LTP] [PATCH 1/2] containers: added mountns/mountns03.c Matus Marhefka
2014-08-26  9:59 ` [LTP] [PATCH 2/2] containers: added mountns/mountns04.c Matus Marhefka
2014-08-29 13:47   ` [LTP] [PATCH 2/2 v2] " Matus Marhefka
2014-09-04 14:21     ` Jan Stancek
2014-10-02 13:47       ` Cyril Hrubis
     [not found]         ` <1080778288.46330076.1412258911538.JavaMail.zimbra@redhat.com>
2014-10-15 12:39           ` Cyril Hrubis
2014-09-24  8:14     ` chrubis
2014-10-02 12:29   ` [LTP] [PATCH 2/2 v3] " Matus Marhefka
2014-08-29 13:46 ` [LTP] [PATCH 1/2 v2] containers: added mountns/mountns03.c Matus Marhefka
2014-09-04 14:13   ` Jan Stancek
2014-09-11 15:09     ` Jiri Jaburek
2014-09-12 12:59       ` Jan Stancek
2014-09-12 13:41         ` Jiri Jaburek
2014-09-12 14:19           ` Jan Stancek
2014-09-24  7:56   ` chrubis [this message]
     [not found]     ` <273164774.30346037.1411649870605.JavaMail.zimbra@redhat.com>
     [not found]       ` <461070711.43439951.1411650253035.JavaMail.zimbra@redhat.com>
2014-09-29  8:14         ` chrubis
2014-10-02 12:28 ` [LTP] [PATCH 1/2 v3] " Matus Marhefka

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=20140924075604.GB10825@rei \
    --to=chrubis@suse.cz \
    --cc=ltp-list@lists.sourceforge.net \
    --cc=mmarhefk@redhat.com \
    /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.