* [LTP] [PATCH 1/2] SAFE_MACROS: Add SAFE_UNAME() @ 2017-05-09 10:05 Guangwen Feng 2017-05-09 10:05 ` [LTP] [PATCH 2/2] controllers/memcg_test_3: Add new regression test Guangwen Feng 0 siblings, 1 reply; 12+ messages in thread From: Guangwen Feng @ 2017-05-09 10:05 UTC (permalink / raw) To: ltp Signed-off-by: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> --- include/tst_safe_macros.h | 4 ++++ lib/safe_macros.c | 15 +++++++++++++++ 2 files changed, 19 insertions(+) diff --git a/include/tst_safe_macros.h b/include/tst_safe_macros.h index a25a4f0..c6530e0 100644 --- a/include/tst_safe_macros.h +++ b/include/tst_safe_macros.h @@ -24,6 +24,7 @@ #include <sys/time.h> #include <sys/resource.h> #include <sys/stat.h> +#include <sys/utsname.h> #include <fcntl.h> #include <libgen.h> #include <signal.h> @@ -430,4 +431,7 @@ int safe_removexattr(const char *file, const int lineno, const char *path, int safe_fsync(const char *file, const int lineno, int fd); #define SAFE_FSYNC(fd) safe_fsync(__FILE__, __LINE__, (fd)) +int safe_uname(const char *file, const int lineno, struct utsname *buf); +#define SAFE_UNAME(buf) safe_uname(__FILE__, __LINE__, (buf)) + #endif /* SAFE_MACROS_H__ */ diff --git a/lib/safe_macros.c b/lib/safe_macros.c index bffc5a1..aec0195 100644 --- a/lib/safe_macros.c +++ b/lib/safe_macros.c @@ -6,6 +6,7 @@ #include <sys/wait.h> #include <sys/mount.h> #include <sys/xattr.h> +#include <sys/utsname.h> #include <errno.h> #include <fcntl.h> #include <libgen.h> @@ -888,3 +889,17 @@ int safe_fsync(const char *file, const int lineno, int fd) return rval; } + +int safe_uname(const char *file, const int lineno, struct utsname *buf) +{ + int rval; + + rval = uname(buf); + + if (rval) { + tst_brkm(TBROK | TERRNO, NULL, + "%s:%d: uname(%p) failed", file, lineno, buf); + } + + return rval; +} -- 1.8.4.2 ^ permalink raw reply related [flat|nested] 12+ messages in thread
* [LTP] [PATCH 2/2] controllers/memcg_test_3: Add new regression test 2017-05-09 10:05 [LTP] [PATCH 1/2] SAFE_MACROS: Add SAFE_UNAME() Guangwen Feng @ 2017-05-09 10:05 ` Guangwen Feng 2017-05-09 11:15 ` Guangwen Feng 0 siblings, 1 reply; 12+ messages in thread From: Guangwen Feng @ 2017-05-09 10:05 UTC (permalink / raw) To: ltp This kernel bug is about RHEL6 BZ1168185 and has been fixed in: [PATCH RHEL6] [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() Signed-off-by: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> --- runtest/controllers | 1 + testcases/kernel/controllers/memcg/.gitignore | 1 + .../controllers/memcg/regression/memcg_test_3.c | 124 +++++++++++++++++++++ 3 files changed, 126 insertions(+) create mode 100644 testcases/kernel/controllers/memcg/regression/memcg_test_3.c diff --git a/runtest/controllers b/runtest/controllers index b904e05..c51be61 100644 --- a/runtest/controllers +++ b/runtest/controllers @@ -1,6 +1,7 @@ #DESCRIPTION:Resource Management testing cgroup cgroup_regression_test.sh memcg_regression memcg_regression_test.sh +memcg_test_3 memcg_test_3 memcg_failcnt memcg_failcnt.sh memcg_force_empty memcg_force_empty.sh memcg_limit_in_bytes memcg_limit_in_bytes.sh diff --git a/testcases/kernel/controllers/memcg/.gitignore b/testcases/kernel/controllers/memcg/.gitignore index 8730251..c0b6d07 100644 --- a/testcases/kernel/controllers/memcg/.gitignore +++ b/testcases/kernel/controllers/memcg/.gitignore @@ -2,5 +2,6 @@ /functional/memcg_process /regression/memcg_test_1 /regression/memcg_test_2 +/regression/memcg_test_3 /regression/memcg_test_4 /stress/memcg_process_stress diff --git a/testcases/kernel/controllers/memcg/regression/memcg_test_3.c b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c new file mode 100644 index 0000000..8d6d325 --- /dev/null +++ b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c @@ -0,0 +1,124 @@ +/* + * Copyright (c) 2017 Fujitsu Ltd. + * Author: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> + * + * 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, either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will 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, see <http://www.gnu.org/licenses/>. + */ + +/* + * This is a regression test for a crash caused by memcg function + * reentrant on RHEL6. When doing rmdir(), a pending signal can + * interrupt the execution and lead to cgroup_clear_css_refs() + * being entered repeatedly, this results in a BUG_ON(). + * + * This bug is introduced by following RHEL6 patch on 2.6.32-488.el6: + * + * [mm] memcg: fix race condition between memcg teardown and swapin + * Bugzilla: 1001197 + * + * This test can crash the buggy kernel on RHEL6.6GA, and the bug + * has been fixed by following patch on 2.6.32-536.el6: + * + * [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() + * Bugzilla: 1168185 + */ + +#include <errno.h> +#include <unistd.h> +#include <stdlib.h> +#include <sys/types.h> +#include <sys/wait.h> +#include "tst_test.h" + +#define MNTPOINT "memcg" +#define SUBDIR "memcg/testdir" + +static int mount_flag; + +static struct tst_kern_exv kvers[] = { + {"RHEL6", "2.6.32-488"}, + {NULL, NULL} +}; + +static void sighandler(int sig LTP_ATTRIBUTE_UNUSED) +{ + +} + +static void do_child(void) +{ + while (1) + SAFE_KILL(getppid(), SIGUSR1); + + exit(0); +} + +static void do_test(void) +{ + int i; + pid_t cpid = -1; + + SAFE_SIGNAL(SIGUSR1, sighandler); + + cpid = SAFE_FORK(); + if (cpid == 0) + do_child(); + + for (i = 0; i < 10; i++) { + if (access(SUBDIR, F_OK)) + SAFE_MKDIR(SUBDIR, 0644); + rmdir(SUBDIR); + } + + SAFE_KILL(cpid, SIGKILL); + SAFE_WAIT(NULL); + + tst_res(TPASS, "Bug not reproduced"); +} + +static void setup(void) +{ + struct utsname buf; + + SAFE_UNAME(&buf); + if (!strstr(buf.release, ".el6")) + tst_brk(TCONF, "This test can only run on RHEL6"); + + if (tst_kvercmp2(2, 6, 24, kvers) < 0) + tst_brk(TCONF, "This test requires kernel >= 2.6.32-488.el6"); + + SAFE_MKDIR(MNTPOINT, 0644); + + SAFE_MOUNT("memcg", MNTPOINT, "cgroup", 0, "memory"); + mount_flag = 1; +} + +static void cleanup(void) +{ + if (!access(SUBDIR, F_OK)) + SAFE_RMDIR(SUBDIR); + + if (mount_flag) + tst_umount(MNTPOINT); +} + +static struct tst_test test = { + .tid = "memcg_test_3", + .needs_root = 1, + .needs_tmpdir = 1, + .forks_child = 1, + .setup = setup, + .cleanup = cleanup, + .test_all = do_test, +}; -- 1.8.4.2 ^ permalink raw reply related [flat|nested] 12+ messages in thread
* [LTP] [PATCH 2/2] controllers/memcg_test_3: Add new regression test 2017-05-09 10:05 ` [LTP] [PATCH 2/2] controllers/memcg_test_3: Add new regression test Guangwen Feng @ 2017-05-09 11:15 ` Guangwen Feng 2017-05-09 11:21 ` [LTP] [PATCH v2 " Guangwen Feng 0 siblings, 1 reply; 12+ messages in thread From: Guangwen Feng @ 2017-05-09 11:15 UTC (permalink / raw) To: ltp Hi! Sorry, there are some typos, I will resend the patch, please ignore this one. Best Regards, Guangwen Feng On 05/09/2017 06:05 PM, Guangwen Feng wrote: > This kernel bug is about RHEL6 BZ1168185 and has been fixed in: > > [PATCH RHEL6] [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() > > Signed-off-by: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> > --- > runtest/controllers | 1 + > testcases/kernel/controllers/memcg/.gitignore | 1 + > .../controllers/memcg/regression/memcg_test_3.c | 124 +++++++++++++++++++++ > 3 files changed, 126 insertions(+) > create mode 100644 testcases/kernel/controllers/memcg/regression/memcg_test_3.c > > diff --git a/runtest/controllers b/runtest/controllers > index b904e05..c51be61 100644 > --- a/runtest/controllers > +++ b/runtest/controllers > @@ -1,6 +1,7 @@ > #DESCRIPTION:Resource Management testing > cgroup cgroup_regression_test.sh > memcg_regression memcg_regression_test.sh > +memcg_test_3 memcg_test_3 > memcg_failcnt memcg_failcnt.sh > memcg_force_empty memcg_force_empty.sh > memcg_limit_in_bytes memcg_limit_in_bytes.sh > diff --git a/testcases/kernel/controllers/memcg/.gitignore b/testcases/kernel/controllers/memcg/.gitignore > index 8730251..c0b6d07 100644 > --- a/testcases/kernel/controllers/memcg/.gitignore > +++ b/testcases/kernel/controllers/memcg/.gitignore > @@ -2,5 +2,6 @@ > /functional/memcg_process > /regression/memcg_test_1 > /regression/memcg_test_2 > +/regression/memcg_test_3 > /regression/memcg_test_4 > /stress/memcg_process_stress > diff --git a/testcases/kernel/controllers/memcg/regression/memcg_test_3.c b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c > new file mode 100644 > index 0000000..8d6d325 > --- /dev/null > +++ b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c > @@ -0,0 +1,124 @@ > +/* > + * Copyright (c) 2017 Fujitsu Ltd. > + * Author: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> > + * > + * 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, either version 2 of the License, or > + * (at your option) any later version. > + * > + * This program is distributed in the hope that it will 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, see <http://www.gnu.org/licenses/>. > + */ > + > +/* > + * This is a regression test for a crash caused by memcg function > + * reentrant on RHEL6. When doing rmdir(), a pending signal can > + * interrupt the execution and lead to cgroup_clear_css_refs() > + * being entered repeatedly, this results in a BUG_ON(). > + * > + * This bug is introduced by following RHEL6 patch on 2.6.32-488.el6: > + * > + * [mm] memcg: fix race condition between memcg teardown and swapin > + * Bugzilla: 1001197 > + * > + * This test can crash the buggy kernel on RHEL6.6GA, and the bug > + * has been fixed by following patch on 2.6.32-536.el6: > + * > + * [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() > + * Bugzilla: 1168185 > + */ > + > +#include <errno.h> > +#include <unistd.h> > +#include <stdlib.h> > +#include <sys/types.h> > +#include <sys/wait.h> > +#include "tst_test.h" > + > +#define MNTPOINT "memcg" > +#define SUBDIR "memcg/testdir" > + > +static int mount_flag; > + > +static struct tst_kern_exv kvers[] = { > + {"RHEL6", "2.6.32-488"}, > + {NULL, NULL} > +}; > + > +static void sighandler(int sig LTP_ATTRIBUTE_UNUSED) > +{ > + > +} > + > +static void do_child(void) > +{ > + while (1) > + SAFE_KILL(getppid(), SIGUSR1); > + > + exit(0); > +} > + > +static void do_test(void) > +{ > + int i; > + pid_t cpid = -1; > + > + SAFE_SIGNAL(SIGUSR1, sighandler); > + > + cpid = SAFE_FORK(); > + if (cpid == 0) > + do_child(); > + > + for (i = 0; i < 10; i++) { > + if (access(SUBDIR, F_OK)) > + SAFE_MKDIR(SUBDIR, 0644); > + rmdir(SUBDIR); > + } > + > + SAFE_KILL(cpid, SIGKILL); > + SAFE_WAIT(NULL); > + > + tst_res(TPASS, "Bug not reproduced"); > +} > + > +static void setup(void) > +{ > + struct utsname buf; > + > + SAFE_UNAME(&buf); > + if (!strstr(buf.release, ".el6")) > + tst_brk(TCONF, "This test can only run on RHEL6"); > + > + if (tst_kvercmp2(2, 6, 24, kvers) < 0) > + tst_brk(TCONF, "This test requires kernel >= 2.6.32-488.el6"); > + > + SAFE_MKDIR(MNTPOINT, 0644); > + > + SAFE_MOUNT("memcg", MNTPOINT, "cgroup", 0, "memory"); > + mount_flag = 1; > +} > + > +static void cleanup(void) > +{ > + if (!access(SUBDIR, F_OK)) > + SAFE_RMDIR(SUBDIR); > + > + if (mount_flag) > + tst_umount(MNTPOINT); > +} > + > +static struct tst_test test = { > + .tid = "memcg_test_3", > + .needs_root = 1, > + .needs_tmpdir = 1, > + .forks_child = 1, > + .setup = setup, > + .cleanup = cleanup, > + .test_all = do_test, > +}; > ^ permalink raw reply [flat|nested] 12+ messages in thread
* [LTP] [PATCH v2 2/2] controllers/memcg_test_3: Add new regression test 2017-05-09 11:15 ` Guangwen Feng @ 2017-05-09 11:21 ` Guangwen Feng 2017-05-30 13:02 ` Cyril Hrubis 0 siblings, 1 reply; 12+ messages in thread From: Guangwen Feng @ 2017-05-09 11:21 UTC (permalink / raw) To: ltp This kernel bug is about RHEL6 BZ1168185 and has been fixed in: [PATCH RHEL6] [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() Signed-off-by: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> --- runtest/controllers | 1 + testcases/kernel/controllers/memcg/.gitignore | 1 + .../controllers/memcg/regression/memcg_test_3.c | 124 +++++++++++++++++++++ 3 files changed, 126 insertions(+) create mode 100644 testcases/kernel/controllers/memcg/regression/memcg_test_3.c diff --git a/runtest/controllers b/runtest/controllers index b904e05..c51be61 100644 --- a/runtest/controllers +++ b/runtest/controllers @@ -1,6 +1,7 @@ #DESCRIPTION:Resource Management testing cgroup cgroup_regression_test.sh memcg_regression memcg_regression_test.sh +memcg_test_3 memcg_test_3 memcg_failcnt memcg_failcnt.sh memcg_force_empty memcg_force_empty.sh memcg_limit_in_bytes memcg_limit_in_bytes.sh diff --git a/testcases/kernel/controllers/memcg/.gitignore b/testcases/kernel/controllers/memcg/.gitignore index 8730251..c0b6d07 100644 --- a/testcases/kernel/controllers/memcg/.gitignore +++ b/testcases/kernel/controllers/memcg/.gitignore @@ -2,5 +2,6 @@ /functional/memcg_process /regression/memcg_test_1 /regression/memcg_test_2 +/regression/memcg_test_3 /regression/memcg_test_4 /stress/memcg_process_stress diff --git a/testcases/kernel/controllers/memcg/regression/memcg_test_3.c b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c new file mode 100644 index 0000000..15edca9 --- /dev/null +++ b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c @@ -0,0 +1,124 @@ +/* + * Copyright (c) 2017 Fujitsu Ltd. + * Author: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> + * + * 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, either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will 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, see <http://www.gnu.org/licenses/>. + */ + +/* + * This is a regression test for a crash caused by memcg function + * reentrant on RHEL6. When doing rmdir(), a pending signal can + * interrupt the execution and lead to cgroup_clear_css_refs() + * being entered repeatedly, this results in a BUG_ON(). + * + * This bug was introduced by following RHEL6 patch on 2.6.32-488.el6: + * + * [mm] memcg: fix race condition between memcg teardown and swapin + * Bugzilla: 1001197 + * + * This test can crash the buggy kernel on RHEL6.6GA, and the bug + * was fixed by following patch on 2.6.32-536.el6: + * + * [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() + * Bugzilla: 1168185 + */ + +#include <errno.h> +#include <unistd.h> +#include <stdlib.h> +#include <sys/types.h> +#include <sys/wait.h> +#include "tst_test.h" + +#define MNTPOINT "memcg" +#define SUBDIR "memcg/testdir" + +static int mount_flag; + +static struct tst_kern_exv kvers[] = { + {"RHEL6", "2.6.32-488"}, + {NULL, NULL} +}; + +static void sighandler(int sig LTP_ATTRIBUTE_UNUSED) +{ + +} + +static void do_child(void) +{ + while (1) + SAFE_KILL(getppid(), SIGUSR1); + + exit(0); +} + +static void do_test(void) +{ + int i; + pid_t cpid = -1; + + SAFE_SIGNAL(SIGUSR1, sighandler); + + cpid = SAFE_FORK(); + if (cpid == 0) + do_child(); + + for (i = 0; i < 10; i++) { + if (access(SUBDIR, F_OK)) + SAFE_MKDIR(SUBDIR, 0644); + rmdir(SUBDIR); + } + + SAFE_KILL(cpid, SIGKILL); + SAFE_WAIT(NULL); + + tst_res(TPASS, "Bug not reproduced"); +} + +static void setup(void) +{ + struct utsname buf; + + SAFE_UNAME(&buf); + if (!strstr(buf.release, ".el6")) + tst_brk(TCONF, "This test can only run on RHEL6"); + + if (tst_kvercmp2(2, 6, 24, kvers) < 0) + tst_brk(TCONF, "This test requires kernel >= 2.6.32-488.el6"); + + SAFE_MKDIR(MNTPOINT, 0644); + + SAFE_MOUNT("memcg", MNTPOINT, "cgroup", 0, "memory"); + mount_flag = 1; +} + +static void cleanup(void) +{ + if (!access(SUBDIR, F_OK)) + SAFE_RMDIR(SUBDIR); + + if (mount_flag) + tst_umount(MNTPOINT); +} + +static struct tst_test test = { + .tid = "memcg_test_3", + .needs_root = 1, + .needs_tmpdir = 1, + .forks_child = 1, + .setup = setup, + .cleanup = cleanup, + .test_all = do_test, +}; -- 1.8.4.2 ^ permalink raw reply related [flat|nested] 12+ messages in thread
* [LTP] [PATCH v2 2/2] controllers/memcg_test_3: Add new regression test 2017-05-09 11:21 ` [LTP] [PATCH v2 " Guangwen Feng @ 2017-05-30 13:02 ` Cyril Hrubis 2017-06-05 9:20 ` Guangwen Feng 0 siblings, 1 reply; 12+ messages in thread From: Cyril Hrubis @ 2017-05-30 13:02 UTC (permalink / raw) To: ltp Hi! > +/* > + * This is a regression test for a crash caused by memcg function > + * reentrant on RHEL6. When doing rmdir(), a pending signal can > + * interrupt the execution and lead to cgroup_clear_css_refs() > + * being entered repeatedly, this results in a BUG_ON(). > + * > + * This bug was introduced by following RHEL6 patch on 2.6.32-488.el6: > + * > + * [mm] memcg: fix race condition between memcg teardown and swapin > + * Bugzilla: 1001197 Can you rather add a link here instead? Just "Bugzilla:" is too vague. > + * This test can crash the buggy kernel on RHEL6.6GA, and the bug > + * was fixed by following patch on 2.6.32-536.el6: > + * > + * [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() > + * Bugzilla: 1168185 > + */ > + > +#include <errno.h> > +#include <unistd.h> > +#include <stdlib.h> > +#include <sys/types.h> > +#include <sys/wait.h> > +#include "tst_test.h" > + > +#define MNTPOINT "memcg" > +#define SUBDIR "memcg/testdir" > + > +static int mount_flag; > + > +static struct tst_kern_exv kvers[] = { > + {"RHEL6", "2.6.32-488"}, > + {NULL, NULL} > +}; > + > +static void sighandler(int sig LTP_ATTRIBUTE_UNUSED) > +{ > + > +} > + > +static void do_child(void) > +{ > + while (1) > + SAFE_KILL(getppid(), SIGUSR1); > + > + exit(0); > +} > + > +static void do_test(void) > +{ > + int i; > + pid_t cpid = -1; > + > + SAFE_SIGNAL(SIGUSR1, sighandler); > + > + cpid = SAFE_FORK(); > + if (cpid == 0) > + do_child(); Shouldn't we wait here untill the child is actually running? I think that with just 10 iteration in the code below we may as well finish the loop too fast. So what about incrementing a counter in the signal handler and loop until it reaches some small value (50 or something)? > + for (i = 0; i < 10; i++) { > + if (access(SUBDIR, F_OK)) > + SAFE_MKDIR(SUBDIR, 0644); > + rmdir(SUBDIR); > + } > + > + SAFE_KILL(cpid, SIGKILL); > + SAFE_WAIT(NULL); > + > + tst_res(TPASS, "Bug not reproduced"); > +} > + > +static void setup(void) > +{ > + struct utsname buf; > + > + SAFE_UNAME(&buf); > + if (!strstr(buf.release, ".el6")) > + tst_brk(TCONF, "This test can only run on RHEL6"); > + > + if (tst_kvercmp2(2, 6, 24, kvers) < 0) > + tst_brk(TCONF, "This test requires kernel >= 2.6.32-488.el6"); Why do we skip this test on anything but RHEL6? It does not seem to me that the test actually does something that couldn't be tested on any other Linux distribution. We only have to check for memcg support here instead. > + SAFE_MKDIR(MNTPOINT, 0644); > + > + SAFE_MOUNT("memcg", MNTPOINT, "cgroup", 0, "memory"); > + mount_flag = 1; > +} > + > +static void cleanup(void) > +{ > + if (!access(SUBDIR, F_OK)) > + SAFE_RMDIR(SUBDIR); > + > + if (mount_flag) > + tst_umount(MNTPOINT); > +} > + > +static struct tst_test test = { > + .tid = "memcg_test_3", > + .needs_root = 1, > + .needs_tmpdir = 1, > + .forks_child = 1, > + .setup = setup, > + .cleanup = cleanup, > + .test_all = do_test, > +}; > -- > 1.8.4.2 > > > > > -- > Mailing list info: https://lists.linux.it/listinfo/ltp -- Cyril Hrubis chrubis@suse.cz ^ permalink raw reply [flat|nested] 12+ messages in thread
* [LTP] [PATCH v2 2/2] controllers/memcg_test_3: Add new regression test 2017-05-30 13:02 ` Cyril Hrubis @ 2017-06-05 9:20 ` Guangwen Feng 2017-06-05 9:23 ` [LTP] [PATCH v3] " Guangwen Feng 0 siblings, 1 reply; 12+ messages in thread From: Guangwen Feng @ 2017-06-05 9:20 UTC (permalink / raw) To: ltp Hi! Thanks for your review, and sorry for the late reply. On 05/30/2017 09:02 PM, Cyril Hrubis wrote: > Hi! >> +/* >> + * This is a regression test for a crash caused by memcg function >> + * reentrant on RHEL6. When doing rmdir(), a pending signal can >> + * interrupt the execution and lead to cgroup_clear_css_refs() >> + * being entered repeatedly, this results in a BUG_ON(). >> + * >> + * This bug was introduced by following RHEL6 patch on 2.6.32-488.el6: >> + * >> + * [mm] memcg: fix race condition between memcg teardown and swapin >> + * Bugzilla: 1001197 > > Can you rather add a link here instead? Just "Bugzilla:" is too vague. > OK, I will add a Bugzilla link instead, and add the patch url of ftp as well. >> + * This test can crash the buggy kernel on RHEL6.6GA, and the bug >> + * was fixed by following patch on 2.6.32-536.el6: >> + * >> + * [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() >> + * Bugzilla: 1168185 >> + */ >> + >> +#include <errno.h> >> +#include <unistd.h> >> +#include <stdlib.h> >> +#include <sys/types.h> >> +#include <sys/wait.h> >> +#include "tst_test.h" >> + >> +#define MNTPOINT "memcg" >> +#define SUBDIR "memcg/testdir" >> + >> +static int mount_flag; >> + >> +static struct tst_kern_exv kvers[] = { >> + {"RHEL6", "2.6.32-488"}, >> + {NULL, NULL} >> +}; >> + >> +static void sighandler(int sig LTP_ATTRIBUTE_UNUSED) >> +{ >> + >> +} >> + >> +static void do_child(void) >> +{ >> + while (1) >> + SAFE_KILL(getppid(), SIGUSR1); >> + >> + exit(0); >> +} >> + >> +static void do_test(void) >> +{ >> + int i; >> + pid_t cpid = -1; >> + >> + SAFE_SIGNAL(SIGUSR1, sighandler); >> + >> + cpid = SAFE_FORK(); >> + if (cpid == 0) >> + do_child(); > > Shouldn't we wait here untill the child is actually running? > > I think that with just 10 iteration in the code below we may as well > finish the loop too fast. > > So what about incrementing a counter in the signal handler and loop > until it reaches some small value (50 or something)? > Yes, we need to ensure the synchronization. It is a good idea to use a counter in signal handler, but 50 or hundreds is too small, since I tested and found that the signal is triggered way much faster than the loop in parent. I have tested that the value can be set to 50000, which makes sure 100% reproducible in buggy kernel and that the test can be done within 1 second when there is no bug. >> + for (i = 0; i < 10; i++) { >> + if (access(SUBDIR, F_OK)) >> + SAFE_MKDIR(SUBDIR, 0644); >> + rmdir(SUBDIR); >> + } >> + >> + SAFE_KILL(cpid, SIGKILL); >> + SAFE_WAIT(NULL); >> + >> + tst_res(TPASS, "Bug not reproduced"); >> +} >> + >> +static void setup(void) >> +{ >> + struct utsname buf; >> + >> + SAFE_UNAME(&buf); >> + if (!strstr(buf.release, ".el6")) >> + tst_brk(TCONF, "This test can only run on RHEL6"); >> + >> + if (tst_kvercmp2(2, 6, 24, kvers) < 0) >> + tst_brk(TCONF, "This test requires kernel >= 2.6.32-488.el6"); > > Why do we skip this test on anything but RHEL6? It does not seem to me > that the test actually does something that couldn't be tested on any > other Linux distribution. We only have to check for memcg support here > instead. > OK, I got it. Best Regards, Guangwen Feng >> + SAFE_MKDIR(MNTPOINT, 0644); >> + >> + SAFE_MOUNT("memcg", MNTPOINT, "cgroup", 0, "memory"); >> + mount_flag = 1; >> +} >> + >> +static void cleanup(void) >> +{ >> + if (!access(SUBDIR, F_OK)) >> + SAFE_RMDIR(SUBDIR); >> + >> + if (mount_flag) >> + tst_umount(MNTPOINT); >> +} >> + >> +static struct tst_test test = { >> + .tid = "memcg_test_3", >> + .needs_root = 1, >> + .needs_tmpdir = 1, >> + .forks_child = 1, >> + .setup = setup, >> + .cleanup = cleanup, >> + .test_all = do_test, >> +}; >> -- >> 1.8.4.2 >> >> >> >> >> -- >> Mailing list info: https://lists.linux.it/listinfo/ltp > ^ permalink raw reply [flat|nested] 12+ messages in thread
* [LTP] [PATCH v3] controllers/memcg_test_3: Add new regression test 2017-06-05 9:20 ` Guangwen Feng @ 2017-06-05 9:23 ` Guangwen Feng 2017-06-14 6:10 ` Guangwen Feng 0 siblings, 1 reply; 12+ messages in thread From: Guangwen Feng @ 2017-06-05 9:23 UTC (permalink / raw) To: ltp This kernel bug is about RHEL6 BZ1168185 and has been fixed in: [PATCH RHEL6] [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() Link: https://bugzilla.redhat.com/show_bug.cgi?id=1168185 Patch: ftp://partners.redhat.com/1c5d859a/de6aafa8185ed8fd934f2debc72b79eb/kernel-individual-patch/rhel6/v2.6.32-to-kernel-2.6.32-536.el6.tar.bz2 35944-mm-memcg-fix-crash-in-re-entrant-cgroup_clear_css_r.patch Signed-off-by: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> --- runtest/controllers | 1 + testcases/kernel/controllers/memcg/.gitignore | 1 + .../controllers/memcg/regression/memcg_test_3.c | 117 +++++++++++++++++++++ 3 files changed, 119 insertions(+) create mode 100644 testcases/kernel/controllers/memcg/regression/memcg_test_3.c diff --git a/runtest/controllers b/runtest/controllers index b904e05..2c43d0b 100644 --- a/runtest/controllers +++ b/runtest/controllers @@ -1,6 +1,7 @@ #DESCRIPTION:Resource Management testing cgroup cgroup_regression_test.sh memcg_regression memcg_regression_test.sh +memcg_test_3 memcg_test_3 memcg_failcnt memcg_failcnt.sh memcg_force_empty memcg_force_empty.sh memcg_limit_in_bytes memcg_limit_in_bytes.sh diff --git a/testcases/kernel/controllers/memcg/.gitignore b/testcases/kernel/controllers/memcg/.gitignore index 8730251..c0b6d07 100644 --- a/testcases/kernel/controllers/memcg/.gitignore +++ b/testcases/kernel/controllers/memcg/.gitignore @@ -2,5 +2,6 @@ /functional/memcg_process /regression/memcg_test_1 /regression/memcg_test_2 +/regression/memcg_test_3 /regression/memcg_test_4 /stress/memcg_process_stress diff --git a/testcases/kernel/controllers/memcg/regression/memcg_test_3.c b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c new file mode 100644 index 0000000..c698c12 --- /dev/null +++ b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c @@ -0,0 +1,117 @@ +/* + * Copyright (c) 2017 Fujitsu Ltd. + * Author: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> + * + * 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, either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will 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, see <http://www.gnu.org/licenses/>. + */ + +/* + * This is a regression test for a crash caused by memcg function + * reentrant on RHEL6. When doing rmdir(), a pending signal can + * interrupt the execution and lead to cgroup_clear_css_refs() + * being entered repeatedly, this results in a BUG_ON(). + * + * This bug was introduced by following RHEL6 patch on 2.6.32-488.el6: + * + * [mm] memcg: fix race condition between memcg teardown and swapin + * Link: https://bugzilla.redhat.com/show_bug.cgi?id=1001197 + * Patch: ftp://partners.redhat.com/1c5d859a/de6aafa8185ed8fd934f2debc72b79eb/kernel-individual-patch/rhel6/v2.6.32-to-kernel-2.6.32-488.el6.tar.bz2 + * 31675-mm-memcg-fix-race-condition-between-memcg-teardown-.patch + * + * This test can crash the buggy kernel on RHEL6.6GA, and the bug + * was fixed by following patch on 2.6.32-536.el6: + * + * [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() + * Link: https://bugzilla.redhat.com/show_bug.cgi?id=1168185 + * Patch: ftp://partners.redhat.com/1c5d859a/de6aafa8185ed8fd934f2debc72b79eb/kernel-individual-patch/rhel6/v2.6.32-to-kernel-2.6.32-536.el6.tar.bz2 + * 35944-mm-memcg-fix-crash-in-re-entrant-cgroup_clear_css_r.patch + */ + +#include <errno.h> +#include <unistd.h> +#include <stdlib.h> +#include <sys/types.h> +#include "tst_test.h" + +#define MNTPOINT "memcg" +#define SUBDIR "memcg/testdir" + +static int mount_flag; +static volatile int sigcounter; + +static void sighandler(int sig LTP_ATTRIBUTE_UNUSED) +{ + sigcounter++; +} + +static void do_child(void) +{ + while (1) + SAFE_KILL(getppid(), SIGUSR1); + + exit(0); +} + +static void do_test(void) +{ + pid_t cpid = -1; + + SAFE_SIGNAL(SIGUSR1, sighandler); + + cpid = SAFE_FORK(); + if (cpid == 0) + do_child(); + + while (sigcounter < 50000) { + if (access(SUBDIR, F_OK)) + SAFE_MKDIR(SUBDIR, 0644); + rmdir(SUBDIR); + } + + SAFE_KILL(cpid, SIGKILL); + SAFE_WAIT(NULL); + + tst_res(TPASS, "Bug not reproduced"); +} + +static void setup(void) +{ + if (tst_kvercmp(2, 6, 24) < 0) + tst_brk(TCONF, "This test can only run on kernels " + "that are 2.6.24 or higher"); + + SAFE_MKDIR(MNTPOINT, 0644); + + SAFE_MOUNT("memcg", MNTPOINT, "cgroup", 0, "memory"); + mount_flag = 1; +} + +static void cleanup(void) +{ + if (!access(SUBDIR, F_OK)) + SAFE_RMDIR(SUBDIR); + + if (mount_flag) + tst_umount(MNTPOINT); +} + +static struct tst_test test = { + .tid = "memcg_test_3", + .needs_root = 1, + .needs_tmpdir = 1, + .forks_child = 1, + .setup = setup, + .cleanup = cleanup, + .test_all = do_test, +}; -- 1.8.4.2 ^ permalink raw reply related [flat|nested] 12+ messages in thread
* [LTP] [PATCH v3] controllers/memcg_test_3: Add new regression test 2017-06-05 9:23 ` [LTP] [PATCH v3] " Guangwen Feng @ 2017-06-14 6:10 ` Guangwen Feng 2017-06-14 6:12 ` [LTP] [PATCH v4] " Guangwen Feng 0 siblings, 1 reply; 12+ messages in thread From: Guangwen Feng @ 2017-06-14 6:10 UTC (permalink / raw) To: ltp Hi! I just remembered that if we only check for memcg support, we should use the min_kver in tst_test struct instead of tst_kvercmp(), sorry, please ignore this one, I will send a v4. Best Regards, Guangwen Feng On 06/05/2017 05:23 PM, Guangwen Feng wrote: > This kernel bug is about RHEL6 BZ1168185 and has been fixed in: > > [PATCH RHEL6] [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() > > Link: https://bugzilla.redhat.com/show_bug.cgi?id=1168185 > Patch: ftp://partners.redhat.com/1c5d859a/de6aafa8185ed8fd934f2debc72b79eb/kernel-individual-patch/rhel6/v2.6.32-to-kernel-2.6.32-536.el6.tar.bz2 > 35944-mm-memcg-fix-crash-in-re-entrant-cgroup_clear_css_r.patch > > Signed-off-by: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> > --- > runtest/controllers | 1 + > testcases/kernel/controllers/memcg/.gitignore | 1 + > .../controllers/memcg/regression/memcg_test_3.c | 117 +++++++++++++++++++++ > 3 files changed, 119 insertions(+) > create mode 100644 testcases/kernel/controllers/memcg/regression/memcg_test_3.c > > diff --git a/runtest/controllers b/runtest/controllers > index b904e05..2c43d0b 100644 > --- a/runtest/controllers > +++ b/runtest/controllers > @@ -1,6 +1,7 @@ > #DESCRIPTION:Resource Management testing > cgroup cgroup_regression_test.sh > memcg_regression memcg_regression_test.sh > +memcg_test_3 memcg_test_3 > memcg_failcnt memcg_failcnt.sh > memcg_force_empty memcg_force_empty.sh > memcg_limit_in_bytes memcg_limit_in_bytes.sh > diff --git a/testcases/kernel/controllers/memcg/.gitignore b/testcases/kernel/controllers/memcg/.gitignore > index 8730251..c0b6d07 100644 > --- a/testcases/kernel/controllers/memcg/.gitignore > +++ b/testcases/kernel/controllers/memcg/.gitignore > @@ -2,5 +2,6 @@ > /functional/memcg_process > /regression/memcg_test_1 > /regression/memcg_test_2 > +/regression/memcg_test_3 > /regression/memcg_test_4 > /stress/memcg_process_stress > diff --git a/testcases/kernel/controllers/memcg/regression/memcg_test_3.c b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c > new file mode 100644 > index 0000000..c698c12 > --- /dev/null > +++ b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c > @@ -0,0 +1,117 @@ > +/* > + * Copyright (c) 2017 Fujitsu Ltd. > + * Author: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> > + * > + * 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, either version 2 of the License, or > + * (at your option) any later version. > + * > + * This program is distributed in the hope that it will 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, see <http://www.gnu.org/licenses/>. > + */ > + > +/* > + * This is a regression test for a crash caused by memcg function > + * reentrant on RHEL6. When doing rmdir(), a pending signal can > + * interrupt the execution and lead to cgroup_clear_css_refs() > + * being entered repeatedly, this results in a BUG_ON(). > + * > + * This bug was introduced by following RHEL6 patch on 2.6.32-488.el6: > + * > + * [mm] memcg: fix race condition between memcg teardown and swapin > + * Link: https://bugzilla.redhat.com/show_bug.cgi?id=1001197 > + * Patch: ftp://partners.redhat.com/1c5d859a/de6aafa8185ed8fd934f2debc72b79eb/kernel-individual-patch/rhel6/v2.6.32-to-kernel-2.6.32-488.el6.tar.bz2 > + * 31675-mm-memcg-fix-race-condition-between-memcg-teardown-.patch > + * > + * This test can crash the buggy kernel on RHEL6.6GA, and the bug > + * was fixed by following patch on 2.6.32-536.el6: > + * > + * [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() > + * Link: https://bugzilla.redhat.com/show_bug.cgi?id=1168185 > + * Patch: ftp://partners.redhat.com/1c5d859a/de6aafa8185ed8fd934f2debc72b79eb/kernel-individual-patch/rhel6/v2.6.32-to-kernel-2.6.32-536.el6.tar.bz2 > + * 35944-mm-memcg-fix-crash-in-re-entrant-cgroup_clear_css_r.patch > + */ > + > +#include <errno.h> > +#include <unistd.h> > +#include <stdlib.h> > +#include <sys/types.h> > +#include "tst_test.h" > + > +#define MNTPOINT "memcg" > +#define SUBDIR "memcg/testdir" > + > +static int mount_flag; > +static volatile int sigcounter; > + > +static void sighandler(int sig LTP_ATTRIBUTE_UNUSED) > +{ > + sigcounter++; > +} > + > +static void do_child(void) > +{ > + while (1) > + SAFE_KILL(getppid(), SIGUSR1); > + > + exit(0); > +} > + > +static void do_test(void) > +{ > + pid_t cpid = -1; > + > + SAFE_SIGNAL(SIGUSR1, sighandler); > + > + cpid = SAFE_FORK(); > + if (cpid == 0) > + do_child(); > + > + while (sigcounter < 50000) { > + if (access(SUBDIR, F_OK)) > + SAFE_MKDIR(SUBDIR, 0644); > + rmdir(SUBDIR); > + } > + > + SAFE_KILL(cpid, SIGKILL); > + SAFE_WAIT(NULL); > + > + tst_res(TPASS, "Bug not reproduced"); > +} > + > +static void setup(void) > +{ > + if (tst_kvercmp(2, 6, 24) < 0) > + tst_brk(TCONF, "This test can only run on kernels " > + "that are 2.6.24 or higher"); > + > + SAFE_MKDIR(MNTPOINT, 0644); > + > + SAFE_MOUNT("memcg", MNTPOINT, "cgroup", 0, "memory"); > + mount_flag = 1; > +} > + > +static void cleanup(void) > +{ > + if (!access(SUBDIR, F_OK)) > + SAFE_RMDIR(SUBDIR); > + > + if (mount_flag) > + tst_umount(MNTPOINT); > +} > + > +static struct tst_test test = { > + .tid = "memcg_test_3", > + .needs_root = 1, > + .needs_tmpdir = 1, > + .forks_child = 1, > + .setup = setup, > + .cleanup = cleanup, > + .test_all = do_test, > +}; > ^ permalink raw reply [flat|nested] 12+ messages in thread
* [LTP] [PATCH v4] controllers/memcg_test_3: Add new regression test 2017-06-14 6:10 ` Guangwen Feng @ 2017-06-14 6:12 ` Guangwen Feng 2017-06-21 10:44 ` Guangwen Feng 2017-06-22 9:27 ` Cyril Hrubis 0 siblings, 2 replies; 12+ messages in thread From: Guangwen Feng @ 2017-06-14 6:12 UTC (permalink / raw) To: ltp This kernel bug is about RHEL6 BZ1168185 and has been fixed in: [PATCH RHEL6] [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() Link: https://bugzilla.redhat.com/show_bug.cgi?id=1168185 Patch: ftp://partners.redhat.com/1c5d859a/de6aafa8185ed8fd934f2debc72b79eb/kernel-individual-patch/rhel6/v2.6.32-to-kernel-2.6.32-536.el6.tar.bz2 35944-mm-memcg-fix-crash-in-re-entrant-cgroup_clear_css_r.patch Signed-off-by: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> --- runtest/controllers | 1 + testcases/kernel/controllers/memcg/.gitignore | 1 + .../controllers/memcg/regression/memcg_test_3.c | 114 +++++++++++++++++++++ 3 files changed, 116 insertions(+) create mode 100644 testcases/kernel/controllers/memcg/regression/memcg_test_3.c diff --git a/runtest/controllers b/runtest/controllers index b904e05..2c43d0b 100644 --- a/runtest/controllers +++ b/runtest/controllers @@ -1,6 +1,7 @@ #DESCRIPTION:Resource Management testing cgroup cgroup_regression_test.sh memcg_regression memcg_regression_test.sh +memcg_test_3 memcg_test_3 memcg_failcnt memcg_failcnt.sh memcg_force_empty memcg_force_empty.sh memcg_limit_in_bytes memcg_limit_in_bytes.sh diff --git a/testcases/kernel/controllers/memcg/.gitignore b/testcases/kernel/controllers/memcg/.gitignore index 8730251..c0b6d07 100644 --- a/testcases/kernel/controllers/memcg/.gitignore +++ b/testcases/kernel/controllers/memcg/.gitignore @@ -2,5 +2,6 @@ /functional/memcg_process /regression/memcg_test_1 /regression/memcg_test_2 +/regression/memcg_test_3 /regression/memcg_test_4 /stress/memcg_process_stress diff --git a/testcases/kernel/controllers/memcg/regression/memcg_test_3.c b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c new file mode 100644 index 0000000..e9de903 --- /dev/null +++ b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c @@ -0,0 +1,114 @@ +/* + * Copyright (c) 2017 Fujitsu Ltd. + * Author: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> + * + * 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, either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will 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, see <http://www.gnu.org/licenses/>. + */ + +/* + * This is a regression test for a crash caused by memcg function + * reentrant on RHEL6. When doing rmdir(), a pending signal can + * interrupt the execution and lead to cgroup_clear_css_refs() + * being entered repeatedly, this results in a BUG_ON(). + * + * This bug was introduced by following RHEL6 patch on 2.6.32-488.el6: + * + * [mm] memcg: fix race condition between memcg teardown and swapin + * Link: https://bugzilla.redhat.com/show_bug.cgi?id=1001197 + * Patch: ftp://partners.redhat.com/1c5d859a/de6aafa8185ed8fd934f2debc72b79eb/kernel-individual-patch/rhel6/v2.6.32-to-kernel-2.6.32-488.el6.tar.bz2 + * 31675-mm-memcg-fix-race-condition-between-memcg-teardown-.patch + * + * This test can crash the buggy kernel on RHEL6.6GA, and the bug + * was fixed by following patch on 2.6.32-536.el6: + * + * [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() + * Link: https://bugzilla.redhat.com/show_bug.cgi?id=1168185 + * Patch: ftp://partners.redhat.com/1c5d859a/de6aafa8185ed8fd934f2debc72b79eb/kernel-individual-patch/rhel6/v2.6.32-to-kernel-2.6.32-536.el6.tar.bz2 + * 35944-mm-memcg-fix-crash-in-re-entrant-cgroup_clear_css_r.patch + */ + +#include <errno.h> +#include <unistd.h> +#include <stdlib.h> +#include <sys/types.h> +#include "tst_test.h" + +#define MNTPOINT "memcg" +#define SUBDIR "memcg/testdir" + +static int mount_flag; +static volatile int sigcounter; + +static void sighandler(int sig LTP_ATTRIBUTE_UNUSED) +{ + sigcounter++; +} + +static void do_child(void) +{ + while (1) + SAFE_KILL(getppid(), SIGUSR1); + + exit(0); +} + +static void do_test(void) +{ + pid_t cpid = -1; + + SAFE_SIGNAL(SIGUSR1, sighandler); + + cpid = SAFE_FORK(); + if (cpid == 0) + do_child(); + + while (sigcounter < 50000) { + if (access(SUBDIR, F_OK)) + SAFE_MKDIR(SUBDIR, 0644); + rmdir(SUBDIR); + } + + SAFE_KILL(cpid, SIGKILL); + SAFE_WAIT(NULL); + + tst_res(TPASS, "Bug not reproduced"); +} + +static void setup(void) +{ + SAFE_MKDIR(MNTPOINT, 0644); + + SAFE_MOUNT("memcg", MNTPOINT, "cgroup", 0, "memory"); + mount_flag = 1; +} + +static void cleanup(void) +{ + if (!access(SUBDIR, F_OK)) + SAFE_RMDIR(SUBDIR); + + if (mount_flag) + tst_umount(MNTPOINT); +} + +static struct tst_test test = { + .tid = "memcg_test_3", + .needs_root = 1, + .needs_tmpdir = 1, + .forks_child = 1, + .min_kver = "2.6.24", + .setup = setup, + .cleanup = cleanup, + .test_all = do_test, +}; -- 1.8.4.2 ^ permalink raw reply related [flat|nested] 12+ messages in thread
* [LTP] [PATCH v4] controllers/memcg_test_3: Add new regression test 2017-06-14 6:12 ` [LTP] [PATCH v4] " Guangwen Feng @ 2017-06-21 10:44 ` Guangwen Feng 2017-06-22 9:27 ` Cyril Hrubis 1 sibling, 0 replies; 12+ messages in thread From: Guangwen Feng @ 2017-06-21 10:44 UTC (permalink / raw) To: ltp Hi! Ping, thanks! Best Regards, Guangwen Feng On 06/14/2017 02:12 PM, Guangwen Feng wrote: > This kernel bug is about RHEL6 BZ1168185 and has been fixed in: > > [PATCH RHEL6] [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() > > Link: https://bugzilla.redhat.com/show_bug.cgi?id=1168185 > Patch: ftp://partners.redhat.com/1c5d859a/de6aafa8185ed8fd934f2debc72b79eb/kernel-individual-patch/rhel6/v2.6.32-to-kernel-2.6.32-536.el6.tar.bz2 > 35944-mm-memcg-fix-crash-in-re-entrant-cgroup_clear_css_r.patch > > Signed-off-by: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> > --- > runtest/controllers | 1 + > testcases/kernel/controllers/memcg/.gitignore | 1 + > .../controllers/memcg/regression/memcg_test_3.c | 114 +++++++++++++++++++++ > 3 files changed, 116 insertions(+) > create mode 100644 testcases/kernel/controllers/memcg/regression/memcg_test_3.c > > diff --git a/runtest/controllers b/runtest/controllers > index b904e05..2c43d0b 100644 > --- a/runtest/controllers > +++ b/runtest/controllers > @@ -1,6 +1,7 @@ > #DESCRIPTION:Resource Management testing > cgroup cgroup_regression_test.sh > memcg_regression memcg_regression_test.sh > +memcg_test_3 memcg_test_3 > memcg_failcnt memcg_failcnt.sh > memcg_force_empty memcg_force_empty.sh > memcg_limit_in_bytes memcg_limit_in_bytes.sh > diff --git a/testcases/kernel/controllers/memcg/.gitignore b/testcases/kernel/controllers/memcg/.gitignore > index 8730251..c0b6d07 100644 > --- a/testcases/kernel/controllers/memcg/.gitignore > +++ b/testcases/kernel/controllers/memcg/.gitignore > @@ -2,5 +2,6 @@ > /functional/memcg_process > /regression/memcg_test_1 > /regression/memcg_test_2 > +/regression/memcg_test_3 > /regression/memcg_test_4 > /stress/memcg_process_stress > diff --git a/testcases/kernel/controllers/memcg/regression/memcg_test_3.c b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c > new file mode 100644 > index 0000000..e9de903 > --- /dev/null > +++ b/testcases/kernel/controllers/memcg/regression/memcg_test_3.c > @@ -0,0 +1,114 @@ > +/* > + * Copyright (c) 2017 Fujitsu Ltd. > + * Author: Guangwen Feng <fenggw-fnst@cn.fujitsu.com> > + * > + * 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, either version 2 of the License, or > + * (at your option) any later version. > + * > + * This program is distributed in the hope that it will 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, see <http://www.gnu.org/licenses/>. > + */ > + > +/* > + * This is a regression test for a crash caused by memcg function > + * reentrant on RHEL6. When doing rmdir(), a pending signal can > + * interrupt the execution and lead to cgroup_clear_css_refs() > + * being entered repeatedly, this results in a BUG_ON(). > + * > + * This bug was introduced by following RHEL6 patch on 2.6.32-488.el6: > + * > + * [mm] memcg: fix race condition between memcg teardown and swapin > + * Link: https://bugzilla.redhat.com/show_bug.cgi?id=1001197 > + * Patch: ftp://partners.redhat.com/1c5d859a/de6aafa8185ed8fd934f2debc72b79eb/kernel-individual-patch/rhel6/v2.6.32-to-kernel-2.6.32-488.el6.tar.bz2 > + * 31675-mm-memcg-fix-race-condition-between-memcg-teardown-.patch > + * > + * This test can crash the buggy kernel on RHEL6.6GA, and the bug > + * was fixed by following patch on 2.6.32-536.el6: > + * > + * [mm] memcg: fix crash in re-entrant cgroup_clear_css_refs() > + * Link: https://bugzilla.redhat.com/show_bug.cgi?id=1168185 > + * Patch: ftp://partners.redhat.com/1c5d859a/de6aafa8185ed8fd934f2debc72b79eb/kernel-individual-patch/rhel6/v2.6.32-to-kernel-2.6.32-536.el6.tar.bz2 > + * 35944-mm-memcg-fix-crash-in-re-entrant-cgroup_clear_css_r.patch > + */ > + > +#include <errno.h> > +#include <unistd.h> > +#include <stdlib.h> > +#include <sys/types.h> > +#include "tst_test.h" > + > +#define MNTPOINT "memcg" > +#define SUBDIR "memcg/testdir" > + > +static int mount_flag; > +static volatile int sigcounter; > + > +static void sighandler(int sig LTP_ATTRIBUTE_UNUSED) > +{ > + sigcounter++; > +} > + > +static void do_child(void) > +{ > + while (1) > + SAFE_KILL(getppid(), SIGUSR1); > + > + exit(0); > +} > + > +static void do_test(void) > +{ > + pid_t cpid = -1; > + > + SAFE_SIGNAL(SIGUSR1, sighandler); > + > + cpid = SAFE_FORK(); > + if (cpid == 0) > + do_child(); > + > + while (sigcounter < 50000) { > + if (access(SUBDIR, F_OK)) > + SAFE_MKDIR(SUBDIR, 0644); > + rmdir(SUBDIR); > + } > + > + SAFE_KILL(cpid, SIGKILL); > + SAFE_WAIT(NULL); > + > + tst_res(TPASS, "Bug not reproduced"); > +} > + > +static void setup(void) > +{ > + SAFE_MKDIR(MNTPOINT, 0644); > + > + SAFE_MOUNT("memcg", MNTPOINT, "cgroup", 0, "memory"); > + mount_flag = 1; > +} > + > +static void cleanup(void) > +{ > + if (!access(SUBDIR, F_OK)) > + SAFE_RMDIR(SUBDIR); > + > + if (mount_flag) > + tst_umount(MNTPOINT); > +} > + > +static struct tst_test test = { > + .tid = "memcg_test_3", > + .needs_root = 1, > + .needs_tmpdir = 1, > + .forks_child = 1, > + .min_kver = "2.6.24", > + .setup = setup, > + .cleanup = cleanup, > + .test_all = do_test, > +}; > ^ permalink raw reply [flat|nested] 12+ messages in thread
* [LTP] [PATCH v4] controllers/memcg_test_3: Add new regression test 2017-06-14 6:12 ` [LTP] [PATCH v4] " Guangwen Feng 2017-06-21 10:44 ` Guangwen Feng @ 2017-06-22 9:27 ` Cyril Hrubis 2017-06-22 9:43 ` Guangwen Feng 1 sibling, 1 reply; 12+ messages in thread From: Cyril Hrubis @ 2017-06-22 9:27 UTC (permalink / raw) To: ltp Hi! I've changed the setup to produce TCONF if mount returns with ENOENT (which means that cgroup support is not compiled in kenrnel) and pushed, thanks. -- Cyril Hrubis chrubis@suse.cz ^ permalink raw reply [flat|nested] 12+ messages in thread
* [LTP] [PATCH v4] controllers/memcg_test_3: Add new regression test 2017-06-22 9:27 ` Cyril Hrubis @ 2017-06-22 9:43 ` Guangwen Feng 0 siblings, 0 replies; 12+ messages in thread From: Guangwen Feng @ 2017-06-22 9:43 UTC (permalink / raw) To: ltp Hi! On 06/22/2017 05:27 PM, Cyril Hrubis wrote: > Hi! > I've changed the setup to produce TCONF if mount returns with ENOENT > (which means that cgroup support is not compiled in kenrnel) and pushed, Ah, I forgot this! Thanks very much for the review! Best Regards, Guangwen Feng > thanks. > ^ permalink raw reply [flat|nested] 12+ messages in thread
end of thread, other threads:[~2017-06-22 9:43 UTC | newest] Thread overview: 12+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2017-05-09 10:05 [LTP] [PATCH 1/2] SAFE_MACROS: Add SAFE_UNAME() Guangwen Feng 2017-05-09 10:05 ` [LTP] [PATCH 2/2] controllers/memcg_test_3: Add new regression test Guangwen Feng 2017-05-09 11:15 ` Guangwen Feng 2017-05-09 11:21 ` [LTP] [PATCH v2 " Guangwen Feng 2017-05-30 13:02 ` Cyril Hrubis 2017-06-05 9:20 ` Guangwen Feng 2017-06-05 9:23 ` [LTP] [PATCH v3] " Guangwen Feng 2017-06-14 6:10 ` Guangwen Feng 2017-06-14 6:12 ` [LTP] [PATCH v4] " Guangwen Feng 2017-06-21 10:44 ` Guangwen Feng 2017-06-22 9:27 ` Cyril Hrubis 2017-06-22 9:43 ` Guangwen Feng
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.