[LTP] [PATCH] New core test of cgroups and extended attributes

[LTP] [PATCH] New core test of cgroups and extended attributes

[Alexey Kodanev]

Support of extended attributes in cgroups was added in Linux 3.7.

Signed-off-by: Alexey Kodanev <alexey.kodanev@oracle.com>
---
 runtest/controllers                                |    1 +
 .../kernel/controllers/cgroup_xattr/.gitignore     |    1 +
 testcases/kernel/controllers/cgroup_xattr/Makefile |   19 +
 testcases/kernel/controllers/cgroup_xattr/README   |   25 +
 .../kernel/controllers/cgroup_xattr/cgroup_xattr.c |  460 ++++++++++++++++++++
 5 files changed, 506 insertions(+), 0 deletions(-)
 create mode 100644 testcases/kernel/controllers/cgroup_xattr/.gitignore
 create mode 100644 testcases/kernel/controllers/cgroup_xattr/Makefile
 create mode 100644 testcases/kernel/controllers/cgroup_xattr/README
 create mode 100644 testcases/kernel/controllers/cgroup_xattr/cgroup_xattr.c

diff --git a/runtest/controllers b/runtest/controllers
index b78d828..94fc185 100644
--- a/runtest/controllers
+++ b/runtest/controllers
@@ -12,3 +12,4 @@ memcg_stress		memcg_stress_test.sh
 memcg_control		PAGESIZE=$(mem_process -p);memcg_control_test.sh $PAGESIZE $PAGESIZE $((PAGESIZE * 2))
 cgroup_fj	run_cgroup_test_fj.sh
 controllers	test_controllers.sh
+cgroup_xattr	cgroup_xattr
diff --git a/testcases/kernel/controllers/cgroup_xattr/.gitignore b/testcases/kernel/controllers/cgroup_xattr/.gitignore
new file mode 100644
index 0000000..3664cc9
--- /dev/null
+++ b/testcases/kernel/controllers/cgroup_xattr/.gitignore
@@ -0,0 +1 @@
+/cgroup_xattr
diff --git a/testcases/kernel/controllers/cgroup_xattr/Makefile b/testcases/kernel/controllers/cgroup_xattr/Makefile
new file mode 100644
index 0000000..cd49588
--- /dev/null
+++ b/testcases/kernel/controllers/cgroup_xattr/Makefile
@@ -0,0 +1,19 @@
+# Copyright (c) 2013 Oracle and/or its affiliates. All Rights Reserved.
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public License as
+# published by the Free Software Foundation.
+#
+# This program is distributed in the hope that it would be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write the Free Software Foundation,
+# Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA 
+
+top_srcdir		?= ../../../..
+
+include $(top_srcdir)/include/mk/testcases.mk
+include $(top_srcdir)/include/mk/generic_leaf_target.mk
diff --git a/testcases/kernel/controllers/cgroup_xattr/README b/testcases/kernel/controllers/cgroup_xattr/README
new file mode 100644
index 0000000..42e294d
--- /dev/null
+++ b/testcases/kernel/controllers/cgroup_xattr/README
@@ -0,0 +1,25 @@
+TEST SUITE:
+
+The directory cgroup_xattr contains the tests related to extended
+attributes in cgroup filesystem.
+
+WARNING:
+
+This test can cause a kernel panic due to a bug in kernels prior to 3.8.
+It was fixed by kernel upstream commit
+712317ad97f41e738e1a19aa0a6392a78a84094e:
+
+"We should store file xattrs in struct cfent instead of struct cftype,
+because cftype is a type while cfent is object instance of cftype."
+
+TESTS AIM:
+
+The aim of the tests is to check the extended attributes in cgroup
+filesystem. This feature was added in Linux 3.7 to allow attaching runtime
+meta information to cgroups and everything they model (services, apps, vms)
+and can easily be shared among applications.
+
+Test creates as many subsystems as possible (cpu, cpuset, memory, ...) in
+the cgroup directory and in one more created hierarchy. Then sets extended
+attributes to all files in cgroup fs and subsequently reads the file's
+extended attributes back, checking values during the process.
diff --git a/testcases/kernel/controllers/cgroup_xattr/cgroup_xattr.c b/testcases/kernel/controllers/cgroup_xattr/cgroup_xattr.c
new file mode 100644
index 0000000..ebf6ca4
--- /dev/null
+++ b/testcases/kernel/controllers/cgroup_xattr/cgroup_xattr.c
@@ -0,0 +1,460 @@
+/*
+ * Copyright (c) 2013 Oracle and/or its affiliates. All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ *
+ * Author:
+ * Alexey Kodanev <alexey.kodanev@oracle.com>
+ *
+ * Test checks following preconditions:
+ * since Linux kernel 3.7 it is possible to set extended attributes
+ * to cgroup files.
+ */
+
+#include <sys/stat.h>
+#include <sys/mount.h>
+#include <sys/types.h>
+#include <sys/xattr.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <fcntl.h>
+#include <dirent.h>
+#include <unistd.h>
+#include <string.h>
+#include <errno.h>
+
+#include "test.h"
+#include "usctest.h"
+#include "safe_macros.h"
+
+char *TCID = "cgroup_xattr";
+
+static const char subdir_name[]	= "test";
+
+#define MAX_SUBSYS		16
+#define MAX_OPTIONS_LEN		256
+#define MAX_DIR_NAME		64
+
+/* struct to store available mount options */
+struct cgrp_option {
+	char opt[MAX_OPTIONS_LEN];
+	char dir[MAX_DIR_NAME];
+	int hier;
+	int mounted;
+	int subdir;
+};
+static struct cgrp_option cgrp_opt[MAX_SUBSYS];
+static int cgrp_opt_num;
+
+struct tst_key {
+	const char *name;
+	int good;
+};
+
+/* only security.* & trusted.* are valid key names */
+static const struct tst_key tkeys[] = {
+	{ .name = "trusted.test",	.good = 1,	},
+	{ .name = "security.",		.good = 1,	},
+	{ .name = "user.",		.good = 0,	},
+	{ .name = "system.",		.good = 0,	},
+};
+
+#define DEFAULT_VALUE_SIZE	8
+
+/* struct to store key's value */
+struct tst_val {
+	char *buf;
+	size_t size;
+};
+static struct tst_val val;
+
+/* it fills value's buffer */
+static char id;
+
+/*
+ * When test breaks, all open dirs should be closed
+ * otherwise umount won't succeed
+ */
+#define MAX_OPEN_DIR		32
+static DIR *odir[MAX_OPEN_DIR];
+static int odir_num;
+
+/* test options */
+static char *narg;
+static int nflag;
+static int skip_cleanup;
+static int verbose;
+static const option_t options[] = {
+	{"n:", &nflag, &narg},
+	{"s", &skip_cleanup, NULL},
+	{"v", &verbose, NULL},
+	{NULL, NULL, NULL}
+};
+
+static void help(void);
+static void setup(int argc, char *argv[]);
+static void test_run(void);
+static void cleanup(void);
+
+/*
+ * get info about mounted subsystems,
+ * make cgroup mount options
+ */
+static void make_cgroup_options(void);
+static int set_xattrs(const char *file);
+static int get_xattrs(const char *file);
+/*
+ * set or get xattr recursively
+ *
+ * @path: start directory
+ * @xattr_operation: can be set_xattrs() or get_xattrs()
+ */
+static int cgrp_files_walking(const char *path,
+	int (*xattr_operation)(const char *));
+
+int main(int argc, char *argv[])
+{
+	setup(argc, argv);
+
+	test_run();
+
+	cleanup();
+
+	tst_exit();
+}
+
+static void help(void)
+{
+	printf("  -n x    Write x bytes to xattr value, default is %d\n",
+		DEFAULT_VALUE_SIZE);
+	printf("  -s      Skip cleanup\n");
+	printf("  -v      Verbose\n");
+}
+
+void setup(int argc, char *argv[])
+{
+	char *msg;
+	msg = parse_opts(argc, argv, options, help);
+	if (msg != NULL)
+		tst_brkm(TBROK, NULL, "OPTION PARSING ERROR - %s", msg);
+
+	tst_require_root(NULL);
+
+	if (access("/proc/cgroups", F_OK) == -1)
+		tst_brkm(TCONF, NULL, "Kernel doesn't support for cgroups");
+
+	if (tst_kvercmp(3, 7, 0) < 0) {
+		tst_brkm(TCONF, NULL,
+			"Test must be run with kernel 3.7 or newer");
+	}
+
+	int value_size = DEFAULT_VALUE_SIZE;
+	if (nflag) {
+		if (sscanf(narg, "%i", &value_size) != 1)
+			tst_brkm(TBROK, NULL, "-n option arg is not a number");
+		if (value_size <= 0)
+			tst_brkm(TBROK, NULL, "-n option arg is less than 1");
+	}
+
+	/* initialize test value */
+	val.size = value_size;
+	val.buf = SAFE_MALLOC(NULL, value_size);
+
+	tst_sig(FORK, DEF_HANDLER, cleanup);
+
+	tst_tmpdir();
+
+	make_cgroup_options();
+
+	int any_mounted = 0;
+	int i;
+	for (i = 0; i < cgrp_opt_num; ++i) {
+		char dir[MAX_DIR_NAME];
+		snprintf(cgrp_opt[i].dir, MAX_DIR_NAME, "cgx_%d",
+			cgrp_opt[i].hier);
+		SAFE_MKDIR(cleanup, cgrp_opt[i].dir, 0755);
+		if (mount(cgrp_opt[i].dir, cgrp_opt[i].dir, "cgroup", 0,
+			cgrp_opt[i].opt) == -1) {
+			tst_resm(TINFO, "Can't mount: %s", dir);
+			continue;
+		}
+
+		any_mounted = 1;
+		cgrp_opt[i].mounted = 1;
+
+		/* create new hierarchy */
+		SAFE_CHDIR(cleanup, cgrp_opt[i].dir);
+		SAFE_MKDIR(cleanup, subdir_name, 0755);
+		cgrp_opt[i].subdir = 1;
+		SAFE_CHDIR(cleanup, "..");
+	}
+
+	if (!any_mounted)
+		tst_brkm(TBROK, cleanup, "Mounted nothing");
+}
+
+static void test_run(void)
+{
+	int	set_res = 0,
+		get_res = 0,
+		i;
+	for (i = 0; i < cgrp_opt_num; ++i) {
+		if (!cgrp_opt[i].mounted)
+			continue;
+
+		SAFE_CHDIR(cleanup, cgrp_opt[i].dir);
+		/* reset value */
+		id = 0;
+		/* set xattr to each file in cgroup fs */
+		set_res |= cgrp_files_walking(".", set_xattrs);
+		/* reset value */
+		id = 0;
+		/* get & check xattr */
+		get_res |= cgrp_files_walking(".", get_xattrs);
+		SAFE_CHDIR(cleanup, "..");
+	}
+
+	/* final results */
+	tst_resm(TINFO, "All test-cases have been completed, summary:");
+	tst_resm(TINFO, "Set tests result: %s", (set_res) ? "FAIL" : "PASS");
+	tst_resm(TINFO, "Get tests result: %s", (get_res) ? "FAIL" : "PASS");
+}
+
+static void cleanup(void)
+{
+	if (val.buf != NULL)
+		free(val.buf);
+
+	if (skip_cleanup)
+		return;
+
+	/*
+	 * Kernels 3.7 can crash while unmounting cgroups with xattr,
+	 * call tst_flush() to make sure all buffered data written
+	 * before it happens
+	 */
+	tst_flush();
+
+	int i;
+	for (i = 0; i < odir_num; ++i) {
+		if (closedir(odir[i]) == -1)
+			tst_brkm(TBROK, NULL, "Failed to close dir\n");
+	}
+
+	char *cwd = get_tst_tmpdir();
+	SAFE_CHDIR(NULL, cwd);
+	free(cwd);
+
+	for (i = 0; i < cgrp_opt_num; ++i) {
+		if (cgrp_opt[i].subdir) {
+			SAFE_CHDIR(NULL, cgrp_opt[i].dir);
+			if (rmdir(subdir_name) == -1) {
+				tst_brkm(TBROK | TERRNO, NULL,
+					"Can't remove dir");
+			}
+			SAFE_CHDIR(NULL, "..");
+		}
+		if (cgrp_opt[i].mounted) {
+			if (umount(cgrp_opt[i].dir) == -1) {
+				tst_brkm(TBROK | TERRNO, NULL,
+					"Can't unmount: %s", cgrp_opt[i].dir);
+			}
+		}
+	}
+
+	tst_rmdir();
+	TEST_CLEANUP;
+}
+
+void make_cgroup_options(void)
+{
+	/* detect subsystems */
+	FILE *fd = fopen("/proc/cgroups", "r");
+	if (fd == NULL)
+		tst_brkm(TBROK, cleanup, "Failed to read /proc/cgroups");
+
+	char str[MAX_DIR_NAME];
+	char name[MAX_DIR_NAME];
+	int	hier	= 0,
+		num	= 0,
+		enabled	= 0,
+		first	= 1;
+
+	while ((fgets(str, MAX_DIR_NAME, fd)) != NULL) {
+		/* skip first line */
+		if (first) {
+			first = 0;
+			continue;
+		}
+		if (sscanf(str, "%s\t%d\t%d\t%d",
+			name, &hier, &num, &enabled) != 4)
+			tst_brkm(TBROK, cleanup, "Can't parse /proc/cgroups");
+
+		if (!enabled)
+			continue;
+
+		/* BUG WORKAROUND
+		 * Only mount those subsystems, which are not mounted yet.
+		 * It's a workaround to a bug when mount doesn't
+		 * return any err code while mounting already mounted
+		 * subsystems, but with additional "xattr" option.
+		 * In that case, mount will succeed, but xattr won't be
+		 * supported in the new mount anyway.
+		 * Should be removed as soon as a fix committed to upstream.
+		 */
+		if (hier != 0)
+			continue;
+		/* end of workaround */
+
+		int found = 0;
+		int i;
+		for (i = 0; i < cgrp_opt_num; ++i) {
+			if (cgrp_opt[i].hier == hier) {
+				found = 1;
+				break;
+			}
+		}
+
+		if (!found) {
+			i = cgrp_opt_num++;
+			cgrp_opt[i].hier = hier;
+		}
+
+		int len = strlen(cgrp_opt[i].opt);
+		if (len == 0) {
+			strcpy(cgrp_opt[i].opt, "xattr");
+			len = strlen(cgrp_opt[i].opt);
+		}
+
+		sprintf(cgrp_opt[i].opt + len, ",%s", name);
+	}
+	fclose(fd);
+
+	int i;
+	for (i = 0; i < cgrp_opt_num; ++i) {
+		tst_resm(TINFO, "mount options %d: %s (hier = %d)",
+			i, cgrp_opt[i].opt, cgrp_opt[i].hier);
+	}
+}
+
+static int set_xattrs(const char *file)
+{
+	int i, err, fail, res;
+	res = 0;
+	for (i = 0; i < ARRAY_SIZE(tkeys); ++i) {
+		err = setxattr(file, tkeys[i].name,
+			(const void *)val.buf, val.size, 0) == -1;
+
+		fail = err && tkeys[i].good;
+		res |= fail;
+
+		tst_resm((fail) ? TFAIL : TPASS,
+			"Expect: %s set xattr key '%s' to file '%s'",
+			(tkeys[i].good) ? "can" : "can't",
+			tkeys[i].name, file);
+
+		if (verbose && tkeys[i].good)
+			tst_resm_hexd(TINFO, val.buf, val.size, "value:");
+	}
+	return res;
+}
+
+static int get_xattrs(const char *file)
+{
+	int i, fail, res;
+	res = 0;
+	for (i = 0; i < ARRAY_SIZE(tkeys); ++i) {
+		/* get value size */
+		ssize_t size = getxattr(file, tkeys[i].name, NULL, 0);
+		fail = (size == -1 && tkeys[i].good);
+		res |= fail;
+
+		tst_resm((fail) ? TFAIL : TPASS,
+			"Expect: %s read xattr %s of file '%s'",
+			(tkeys[i].good) ? "can" : "can't",
+			tkeys[i].name, file);
+
+		if (fail || size == -1)
+			continue;
+
+		/* get xattr value */
+		char xval[size];
+		if (getxattr(file, tkeys[i].name, xval, size) == -1) {
+			tst_brkm(TBROK, cleanup,
+				"Can't get buffer of key %s",
+				tkeys[i].name);
+		}
+		fail = val.size != size ||
+			strncmp(val.buf, xval, val.size) != 0;
+		res |= fail;
+
+		tst_resm((fail) ? TFAIL : TPASS, "Expect: values equal");
+
+		if (verbose && fail) {
+			tst_resm_hexd(TINFO, xval, size,
+				"Read  xattr  value:");
+			tst_resm_hexd(TINFO, val.buf, val.size,
+				"Expect xattr value:");
+		}
+	}
+	return res;
+}
+
+static int cgrp_files_walking(const char *path,
+	int (*xattr_operation)(const char *))
+{
+	int res = 0;
+	struct dirent *entry;
+	DIR *dir = opendir(path);
+
+	odir[odir_num] = dir;
+	if (++odir_num >= MAX_OPEN_DIR) {
+		tst_brkm(TBROK, cleanup,
+			"Unexpected num of open dirs, max: %d", MAX_OPEN_DIR);
+	}
+
+	SAFE_CHDIR(cleanup, path);
+
+	tst_resm(TINFO, "In dir %s", path);
+
+	errno = 0;
+	while ((entry = readdir(dir)) != NULL) {
+		const char *file = entry->d_name;
+		/* skip current and up directories */
+		if (!strcmp(file, "..") || !strcmp(file, "."))
+			continue;
+		struct stat stat_buf;
+		TEST(lstat(file, &stat_buf));
+		if (TEST_RETURN != -1 && S_ISDIR(stat_buf.st_mode)) {
+			/* proceed to subdir */
+			res |= cgrp_files_walking(file, xattr_operation);
+			tst_resm(TINFO, "In dir %s", path);
+		}
+		memset(val.buf, id++, val.size);
+		res |= xattr_operation(file);
+		errno = 0;
+	}
+	if (errno && !entry) {
+		tst_brkm(TWARN | TERRNO, cleanup,
+			"Error while reading dir '%s'", path);
+	}
+	if (closedir(dir) == -1)
+		tst_brkm(TWARN, cleanup, "Failed to close dir '%s'", path);
+	else
+		odir[--odir_num] = NULL;
+
+	if (strcmp(path, "."))
+		SAFE_CHDIR(cleanup, "..");
+	return res;
+}
-- 
1.7.1


------------------------------------------------------------------------------
Try New Relic Now & We'll Send You this Cool Shirt
New Relic is the only SaaS-based application performance monitoring service 
that delivers powerful full stack analytics. Optimize and monitor your
browser, app, & servers with just a few lines of code. Try New Relic
and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_may
_______________________________________________
Ltp-list mailing list
Ltp-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ltp-list

Re: [LTP] [PATCH] New core test of cgroups and extended attributes

[chrubis]

Hi!
> +void setup(int argc, char *argv[])
> +{
> +	char *msg;
> +	msg = parse_opts(argc, argv, options, help);
> +	if (msg != NULL)
> +		tst_brkm(TBROK, NULL, "OPTION PARSING ERROR - %s", msg);
> +
> +	tst_require_root(NULL);
> +
> +	if (access("/proc/cgroups", F_OK) == -1)
> +		tst_brkm(TCONF, NULL, "Kernel doesn't support for cgroups");
                                                     ^         ^
				               Either add have here or
					       remove the for

> +	/* initialize test value */
> +	val.size = value_size;
> +	val.buf = SAFE_MALLOC(NULL, value_size);
> +
> +	tst_sig(FORK, DEF_HANDLER, cleanup);
> +
> +	tst_tmpdir();
> +
> +	make_cgroup_options();
> +
> +	int any_mounted = 0;
> +	int i;
> +	for (i = 0; i < cgrp_opt_num; ++i) {
> +		char dir[MAX_DIR_NAME];
> +		snprintf(cgrp_opt[i].dir, MAX_DIR_NAME, "cgx_%d",
> +			cgrp_opt[i].hier);
> +		SAFE_MKDIR(cleanup, cgrp_opt[i].dir, 0755);
> +		if (mount(cgrp_opt[i].dir, cgrp_opt[i].dir, "cgroup", 0,
> +			cgrp_opt[i].opt) == -1) {
> +			tst_resm(TINFO, "Can't mount: %s", dir);
> +			continue;
> +		}
> +
> +		any_mounted = 1;
> +		cgrp_opt[i].mounted = 1;
> +
> +		/* create new hierarchy */
> +		SAFE_CHDIR(cleanup, cgrp_opt[i].dir);
> +		SAFE_MKDIR(cleanup, subdir_name, 0755);
> +		cgrp_opt[i].subdir = 1;
> +		SAFE_CHDIR(cleanup, "..");
> +	}

I would keep the mounting code in the function that parses the cgroups
proc file, that way we can have simple:

	if (!mount_cgroups())
		tst_brkm(TBROK, cleanup, "Nothing mounted");

in the setup and keep all the details in the function.

> +	if (!any_mounted)
> +		tst_brkm(TBROK, cleanup, "Mounted nothing");

This should be TCONF.

> +}
> +
> +void make_cgroup_options(void)
> +{
> +	/* detect subsystems */
> +	FILE *fd = fopen("/proc/cgroups", "r");
> +	if (fd == NULL)
> +		tst_brkm(TBROK, cleanup, "Failed to read /proc/cgroups");
> +
> +	char str[MAX_DIR_NAME];
> +	char name[MAX_DIR_NAME];
> +	int	hier	= 0,
> +		num	= 0,
> +		enabled	= 0,
> +		first	= 1;

I would keep these at one line, which would save vertical space, but
that is purely cosmetic change.

> +	while ((fgets(str, MAX_DIR_NAME, fd)) != NULL) {
> +		/* skip first line */
> +		if (first) {
> +			first = 0;
> +			continue;
> +		}
> +		if (sscanf(str, "%s\t%d\t%d\t%d",
> +			name, &hier, &num, &enabled) != 4)
> +			tst_brkm(TBROK, cleanup, "Can't parse /proc/cgroups");
> +
> +		if (!enabled)
> +			continue;
> +
> +		/* BUG WORKAROUND
> +		 * Only mount those subsystems, which are not mounted yet.
> +		 * It's a workaround to a bug when mount doesn't
> +		 * return any err code while mounting already mounted
> +		 * subsystems, but with additional "xattr" option.
> +		 * In that case, mount will succeed, but xattr won't be
> +		 * supported in the new mount anyway.
> +		 * Should be removed as soon as a fix committed to upstream.
> +		 */
> +		if (hier != 0)
> +			continue;
> +		/* end of workaround */

Remove the /* end of workaround */ please.

> +		int found = 0;
> +		int i;
> +		for (i = 0; i < cgrp_opt_num; ++i) {
> +			if (cgrp_opt[i].hier == hier) {
> +				found = 1;
> +				break;
> +			}
> +		}
> +
> +		if (!found) {
> +			i = cgrp_opt_num++;
> +			cgrp_opt[i].hier = hier;
> +		}
> +
> +		int len = strlen(cgrp_opt[i].opt);
> +		if (len == 0) {

                Could also be if (cgrp_opt[i].opt[0] == '\0') {
		(or something similar)

> +			strcpy(cgrp_opt[i].opt, "xattr");
> +			len = strlen(cgrp_opt[i].opt);

                        len == 5 in this case ;)

> +		}
> +
> +		sprintf(cgrp_opt[i].opt + len, ",%s", name);

So we create a list of subsystem by hierarchy here, and try to mount
them with xattr later?

Ah but due to bug in kernel (hier == 0 when we got to this part) we end
up only with these that are not mounted at the moment, right?

That looks OK. Hopefully the kernel gets fixed. Will the kernel support
mouting the subsystems with additional xattr flags or will the mount
return an error (on my machine all mounted subsystems are mounted
without xattr)?

> +	}
> +	fclose(fd);
> +
> +	int i;
> +	for (i = 0; i < cgrp_opt_num; ++i) {
> +		tst_resm(TINFO, "mount options %d: %s (hier = %d)",
> +			i, cgrp_opt[i].opt, cgrp_opt[i].hier);
> +	}
> +}
> +


-- 
Cyril Hrubis
chrubis@suse.cz

------------------------------------------------------------------------------
Try New Relic Now & We'll Send You this Cool Shirt
New Relic is the only SaaS-based application performance monitoring service 
that delivers powerful full stack analytics. Optimize and monitor your
browser, app, & servers with just a few lines of code. Try New Relic
and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_may
_______________________________________________
Ltp-list mailing list
Ltp-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ltp-list

Re: [LTP] [PATCH] New core test of cgroups and extended attributes

[chrubis]

Hi!
> > So we create a list of subsystem by hierarchy here, and try to mount
> > them with xattr later?
> That's right
> > Ah but due to bug in kernel (hier == 0 when we got to this part) we end
> > up only with these that are not mounted at the moment, right?
> Yes
> > That looks OK. Hopefully the kernel gets fixed. Will the kernel support
> > mouting the subsystems with additional xattr flags or will the mount
> > return an error (on my machine all mounted subsystems are mounted
> > without xattr)?
> >
> It's difficult to say, for now, we have a recent upstream commit 
> <http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=873fe09ea5df6ccf6bb34811d8c9992aacb67598> 
> in which new cgroup mount option introduced (__DEVEL__sane_behavior). It 
> can be used to get mount error. Developer says that this bug is a part 
> of largerbreakage - cgroup silently ignores mount option changes 
> whenremounted. He thinks that fixing only this part is not meaningful, 
> suggesting to add warning message if the mount options mismatch.

OK. Lets get keep the test as it is now. We will fix it later, once the
kernel devs figure out their solution.

-- 
Cyril Hrubis
chrubis@suse.cz

------------------------------------------------------------------------------
Try New Relic Now & We'll Send You this Cool Shirt
New Relic is the only SaaS-based application performance monitoring service 
that delivers powerful full stack analytics. Optimize and monitor your
browser, app, & servers with just a few lines of code. Try New Relic
and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_may
_______________________________________________
Ltp-list mailing list
Ltp-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ltp-list

Re: [LTP] Device firmware loading test

[chrubis]

Hi!
> I'm developing a test of device firmware loading (after 3.7 it can be 
> loaded directly or as usual), but I'm not sure under which ltp directory 
> it should be placed. Is it OK if I use two locations as described below?
> 
> .../testcases/kernel/device-drivers/firmware/fw_load_kernel/fw_load.ko - 
> kernel space part (calls request_firmware with specified parameters), and
>
> .../testcases/kernel/device-drivers/firmware/fw_load_user/fw_load - user 
> space part (creates firmware files, replaces udev's firmware searched 
> paths, loads the module, prints results...).

The problem with device-drivers directory is that the code there is
broken (unmaintained since 2009 or so) and the make (from the top
directory) doesn't go there. It should be cleaned/removed/fixed but
nobody had time to look at the tests and decide what to do with them.

Otherwise the destination fits well enough.

And there is another problem with building kernel modules, I think that
we don't have any configure checks to find out if kernel-devel is
installed (i.e. /lib/modules/$(version)/build/ exists) and which kernel
version to use, etc. But this part shouldn't be that hard.

-- 
Cyril Hrubis
chrubis@suse.cz

------------------------------------------------------------------------------
Introducing AppDynamics Lite, a free troubleshooting tool for Java/.NET
Get 100% visibility into your production application - at no cost.
Code-level diagnostics for performance bottlenecks with <2% overhead
Download for free and get started troubleshooting in minutes.
http://p.sf.net/sfu/appdyn_d2d_ap1
_______________________________________________
Ltp-list mailing list
Ltp-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ltp-list

Re: [LTP] Device firmware loading test

[chrubis]

Hi!
> > The problem with device-drivers directory is that the code there is
> > broken (unmaintained since 2009 or so) and the make (from the top
> > directory) doesn't go there. It should be cleaned/removed/fixed but
> > nobody had time to look at the tests and decide what to do with them.
> >
> > Otherwise the destination fits well enough.
> As I understand from README file in that directory, these tests should 
> not be run automatically. It explains why make doesn't go there. But the 
> firmware loading test can be (preferably should be) run automatically. 
> That way, may be change directory to .../testcases/kernel/firmware?

Hmm, so the test tests userspace API for firmware loading but needs
helper kernel module, am I right?. In that case kernel/firmware/ should
be ok as well.

-- 
Cyril Hrubis
chrubis@suse.cz

------------------------------------------------------------------------------
Introducing AppDynamics Lite, a free troubleshooting tool for Java/.NET
Get 100% visibility into your production application - at no cost.
Code-level diagnostics for performance bottlenecks with <2% overhead
Download for free and get started troubleshooting in minutes.
http://p.sf.net/sfu/appdyn_d2d_ap1
_______________________________________________
Ltp-list mailing list
Ltp-list@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ltp-list