[LTP] [PATCH v2 0/5] Add close_range01, SAFE_DUP2 and SAFE_CLONE

[LTP] [PATCH v2 0/5] Add close_range01, SAFE_DUP2 and SAFE_CLONE

[Richard Palethorpe]

Hi,

This series adds a test for the new system call close_range. The
kernel selftests already added some tests, so I based this on those.

Also of note, is the addition of tst_clone and SAFE_CLONE which mirror
the clone3 interface. This is not really necessary for this test as it
is unlikely close_range is backported, but clone3 is not. OTOH I think
it will be usefull for other tests especially as the libc clone
wrapper seems to force the user to allocate the stack even though the
kernel can handle this. Also the clone_args is growing rapidly.

The old API has ltp_clone* functions, but I think we will have an
easier time tracking the clone3 interface, than what we were doing
with the old API.

V2:
* Add docs for tst_clone
* Add lapi/close_range.h

Richard Palethorpe (5):
  API: Add close_range syscall wrapper
  API: Add SAFE_DUP2
  API: Fix clone.h
  API: Add tst_clone
  close_range: Add test

 configure.ac                                  |   2 +
 doc/test-writing-guidelines.txt               |  27 ++-
 include/lapi/clone.h                          |   5 +-
 include/lapi/close_range.h                    |  28 +++
 include/lapi/syscalls/aarch64.in              |   1 +
 include/lapi/syscalls/arc.in                  |   1 +
 include/lapi/syscalls/arm.in                  |   1 +
 include/lapi/syscalls/hppa.in                 |   1 +
 include/lapi/syscalls/i386.in                 |   1 +
 include/lapi/syscalls/ia64.in                 |   1 +
 include/lapi/syscalls/mips_n32.in             |   1 +
 include/lapi/syscalls/mips_n64.in             |   1 +
 include/lapi/syscalls/mips_o32.in             |   1 +
 include/lapi/syscalls/powerpc.in              |   1 +
 include/lapi/syscalls/powerpc64.in            |   1 +
 include/lapi/syscalls/s390.in                 |   1 +
 include/lapi/syscalls/s390x.in                |   1 +
 include/lapi/syscalls/sh.in                   |   1 +
 include/lapi/syscalls/sparc.in                |   1 +
 include/lapi/syscalls/sparc64.in              |   1 +
 include/lapi/syscalls/x86_64.in               |   1 +
 include/tst_clone.h                           |  28 +++
 include/tst_safe_macros.h                     |   5 +
 include/tst_test.h                            |   2 +-
 lib/tst_clone.c                               |  46 ++++
 lib/tst_safe_macros.c                         |  18 ++
 lib/tst_test.c                                |  25 +++
 .../kernel/syscalls/close_range/.gitignore    |   1 +
 .../kernel/syscalls/close_range/Makefile      |  10 +
 .../syscalls/close_range/close_range01.c      | 200 ++++++++++++++++++
 30 files changed, 406 insertions(+), 8 deletions(-)
 create mode 100644 include/lapi/close_range.h
 create mode 100644 lib/tst_clone.c
 create mode 100644 testcases/kernel/syscalls/close_range/.gitignore
 create mode 100644 testcases/kernel/syscalls/close_range/Makefile
 create mode 100644 testcases/kernel/syscalls/close_range/close_range01.c

-- 
2.30.0

[LTP] [PATCH v2 1/5] API: Add close_range syscall wrapper

[Richard Palethorpe]

Add the syscall number and a wrapper function. Presently there is no
wrapper function in libc.

Signed-off-by: Richard Palethorpe <rpalethorpe@suse.com>
---
 configure.ac                       |  2 ++
 include/lapi/close_range.h         | 28 ++++++++++++++++++++++++++++
 include/lapi/syscalls/aarch64.in   |  1 +
 include/lapi/syscalls/arc.in       |  1 +
 include/lapi/syscalls/arm.in       |  1 +
 include/lapi/syscalls/hppa.in      |  1 +
 include/lapi/syscalls/i386.in      |  1 +
 include/lapi/syscalls/ia64.in      |  1 +
 include/lapi/syscalls/mips_n32.in  |  1 +
 include/lapi/syscalls/mips_n64.in  |  1 +
 include/lapi/syscalls/mips_o32.in  |  1 +
 include/lapi/syscalls/powerpc.in   |  1 +
 include/lapi/syscalls/powerpc64.in |  1 +
 include/lapi/syscalls/s390.in      |  1 +
 include/lapi/syscalls/s390x.in     |  1 +
 include/lapi/syscalls/sh.in        |  1 +
 include/lapi/syscalls/sparc.in     |  1 +
 include/lapi/syscalls/sparc64.in   |  1 +
 include/lapi/syscalls/x86_64.in    |  1 +
 19 files changed, 47 insertions(+)
 create mode 100644 include/lapi/close_range.h

diff --git a/configure.ac b/configure.ac
index 9fb50c173..d4bef5e45 100644
--- a/configure.ac
+++ b/configure.ac
@@ -43,6 +43,7 @@ AC_CHECK_HEADERS_ONCE([ \
     linux/can.h \
     linux/cgroupstats.h \
     linux/cryptouser.h \
+    linux/close_range.h \
     linux/dccp.h \
     linux/fs.h \
     linux/genetlink.h \
@@ -77,6 +78,7 @@ AC_SUBST(HAVE_FTS_H, $have_fts)
 
 AC_CHECK_FUNCS_ONCE([ \
     clone3 \
+    close_range \
     copy_file_range \
     epoll_pwait \
     execveat \
diff --git a/include/lapi/close_range.h b/include/lapi/close_range.h
new file mode 100644
index 000000000..19db52d3d
--- /dev/null
+++ b/include/lapi/close_range.h
@@ -0,0 +1,28 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/* Copyright (c) 2021 SUSE LLC  */
+
+#ifndef LAPI_CLOSE_RANGE__
+# define LAPI_CLOSE_RANGE__
+
+# include "lapi/syscalls.h"
+
+# ifdef HAVE_LINUX_CLOSE_RANGE_H
+#  include <linux/close_range.h>
+# endif
+
+# ifndef CLOSE_RANGE_UNSHARE
+#  define CLOSE_RANGE_UNSHARE	(1U << 1)
+# endif
+
+# ifndef CLOSE_RANGE_CLOEXEC
+#  define CLOSE_RANGE_CLOEXEC	(1U << 2)
+# endif
+
+# ifndef HAVE_CLOSE_RANGE
+static inline int close_range(unsigned int fd, unsigned int max_fd,
+			      unsigned int flags)
+{
+	return tst_syscall(__NR_close_range, fd, max_fd, flags);
+}
+# endif
+#endif	/* LAPI_CLOSE_RANGE_H__ */
diff --git a/include/lapi/syscalls/aarch64.in b/include/lapi/syscalls/aarch64.in
index 07556933f..2def6ba3d 100644
--- a/include/lapi/syscalls/aarch64.in
+++ b/include/lapi/syscalls/aarch64.in
@@ -310,6 +310,7 @@ fsmount 432
 fspick 433
 pidfd_open 434
 clone3 435
+close_range 436
 openat2 437
 pidfd_getfd 438
 _sysctl 1078
diff --git a/include/lapi/syscalls/arc.in b/include/lapi/syscalls/arc.in
index 0cadb150c..9bcd84706 100644
--- a/include/lapi/syscalls/arc.in
+++ b/include/lapi/syscalls/arc.in
@@ -310,5 +310,6 @@ fsmount 432
 fspick 433
 pidfd_open 434
 clone3 435
+close_range 436
 openat2 437
 pidfd_getfd 438
diff --git a/include/lapi/syscalls/arm.in b/include/lapi/syscalls/arm.in
index 395d53a68..98c840cb8 100644
--- a/include/lapi/syscalls/arm.in
+++ b/include/lapi/syscalls/arm.in
@@ -388,5 +388,6 @@ fsmount (__NR_SYSCALL_BASE+432)
 fspick (__NR_SYSCALL_BASE+433)
 pidfd_open (__NR_SYSCALL_BASE+434)
 clone3 (__NR_SYSCALL_BASE+435)
+close_range (__NR_SYSCALL_BASE+436)
 openat2 (__NR_SYSCALL_BASE+437)
 pidfd_getfd (__NR_SYSCALL_BASE+438)
diff --git a/include/lapi/syscalls/hppa.in b/include/lapi/syscalls/hppa.in
index 7f9b5e0a0..e1628c4b1 100644
--- a/include/lapi/syscalls/hppa.in
+++ b/include/lapi/syscalls/hppa.in
@@ -39,3 +39,4 @@ fsconfig 431
 fsmount 432
 fspick 433
 pidfd_open 434
+close_range 436
\ No newline@end of file
diff --git a/include/lapi/syscalls/i386.in b/include/lapi/syscalls/i386.in
index d6773abcb..aaa02c7bf 100644
--- a/include/lapi/syscalls/i386.in
+++ b/include/lapi/syscalls/i386.in
@@ -424,5 +424,6 @@ fsmount 432
 fspick 433
 pidfd_open 434
 clone3 435
+close_range 436
 openat2 437
 pidfd_getfd 438
diff --git a/include/lapi/syscalls/ia64.in b/include/lapi/syscalls/ia64.in
index 427b711b1..5467f80f2 100644
--- a/include/lapi/syscalls/ia64.in
+++ b/include/lapi/syscalls/ia64.in
@@ -337,5 +337,6 @@ fsconfig 1455
 fsmount 1456
 fspick 1457
 pidfd_open 1458
+close_range 1460
 openat2 1461
 pidfd_getfd 1462
diff --git a/include/lapi/syscalls/mips_n32.in b/include/lapi/syscalls/mips_n32.in
index eb6140ebc..a509ca5c6 100644
--- a/include/lapi/syscalls/mips_n32.in
+++ b/include/lapi/syscalls/mips_n32.in
@@ -362,5 +362,6 @@ fsmount 432
 fspick 433
 pidfd_open 434
 clone3 435
+close_range 436
 openat2 437
 pidfd_getfd 438
diff --git a/include/lapi/syscalls/mips_n64.in b/include/lapi/syscalls/mips_n64.in
index 5480aa3c9..532c69dd8 100644
--- a/include/lapi/syscalls/mips_n64.in
+++ b/include/lapi/syscalls/mips_n64.in
@@ -338,5 +338,6 @@ fsmount 432
 fspick 433
 pidfd_open 434
 clone3 435
+close_range 436
 openat2 437
 pidfd_getfd 438
diff --git a/include/lapi/syscalls/mips_o32.in b/include/lapi/syscalls/mips_o32.in
index feee8fbf3..c71a62a35 100644
--- a/include/lapi/syscalls/mips_o32.in
+++ b/include/lapi/syscalls/mips_o32.in
@@ -408,5 +408,6 @@ fsmount 432
 fspick 433
 pidfd_open 434
 clone3 435
+close_range 436
 openat2 437
 pidfd_getfd 438
diff --git a/include/lapi/syscalls/powerpc.in b/include/lapi/syscalls/powerpc.in
index cdbebc62d..2d287a606 100644
--- a/include/lapi/syscalls/powerpc.in
+++ b/include/lapi/syscalls/powerpc.in
@@ -417,5 +417,6 @@ fsmount 432
 fspick 433
 pidfd_open 434
 clone3 435
+close_range 436
 openat2 437
 pidfd_getfd 438
diff --git a/include/lapi/syscalls/powerpc64.in b/include/lapi/syscalls/powerpc64.in
index cdbebc62d..2d287a606 100644
--- a/include/lapi/syscalls/powerpc64.in
+++ b/include/lapi/syscalls/powerpc64.in
@@ -417,5 +417,6 @@ fsmount 432
 fspick 433
 pidfd_open 434
 clone3 435
+close_range 436
 openat2 437
 pidfd_getfd 438
diff --git a/include/lapi/syscalls/s390.in b/include/lapi/syscalls/s390.in
index c3f249aed..c978b6660 100644
--- a/include/lapi/syscalls/s390.in
+++ b/include/lapi/syscalls/s390.in
@@ -404,5 +404,6 @@ fsmount 432
 fspick 433
 pidfd_open 434
 clone3 435
+close_range 436
 openat2 437
 pidfd_getfd 438
diff --git a/include/lapi/syscalls/s390x.in b/include/lapi/syscalls/s390x.in
index 88cc9b86b..d123db6cb 100644
--- a/include/lapi/syscalls/s390x.in
+++ b/include/lapi/syscalls/s390x.in
@@ -352,5 +352,6 @@ fsmount 432
 fspick 433
 pidfd_open 434
 clone3 435
+close_range 436
 openat2 437
 pidfd_getfd 438
diff --git a/include/lapi/syscalls/sh.in b/include/lapi/syscalls/sh.in
index 06055ed10..22da7d6ff 100644
--- a/include/lapi/syscalls/sh.in
+++ b/include/lapi/syscalls/sh.in
@@ -398,5 +398,6 @@ fsconfig 431
 fsmount 432
 fspick 433
 pidfd_open 434
+close_range 436
 openat2 437
 pidfd_getfd 438
diff --git a/include/lapi/syscalls/sparc.in b/include/lapi/syscalls/sparc.in
index 522e3c997..7324b4ac6 100644
--- a/include/lapi/syscalls/sparc.in
+++ b/include/lapi/syscalls/sparc.in
@@ -403,5 +403,6 @@ fsconfig 431
 fsmount 432
 fspick 433
 pidfd_open 434
+close_range 436
 openat2 437
 pidfd_getfd 438
diff --git a/include/lapi/syscalls/sparc64.in b/include/lapi/syscalls/sparc64.in
index 6f884a7bf..862d806fa 100644
--- a/include/lapi/syscalls/sparc64.in
+++ b/include/lapi/syscalls/sparc64.in
@@ -368,5 +368,6 @@ fsconfig 431
 fsmount 432
 fspick 433
 pidfd_open 434
+close_range 436
 openat2 437
 pidfd_getfd 438
diff --git a/include/lapi/syscalls/x86_64.in b/include/lapi/syscalls/x86_64.in
index c76328c93..1345002f5 100644
--- a/include/lapi/syscalls/x86_64.in
+++ b/include/lapi/syscalls/x86_64.in
@@ -345,6 +345,7 @@ fsmount 432
 fspick 433
 pidfd_open 434
 clone3 435
+close_range 436
 openat2 437
 pidfd_getfd 438
 rt_sigaction 512
-- 
2.30.0

[LTP] [PATCH v2 2/5] API: Add SAFE_DUP2

[Richard Palethorpe]

Signed-off-by: Richard Palethorpe <rpalethorpe@suse.com>
---
 include/tst_safe_macros.h |  5 +++++
 lib/tst_safe_macros.c     | 18 ++++++++++++++++++
 2 files changed, 23 insertions(+)

diff --git a/include/tst_safe_macros.h b/include/tst_safe_macros.h
index ee3df4142..2a2b0088a 100644
--- a/include/tst_safe_macros.h
+++ b/include/tst_safe_macros.h
@@ -50,6 +50,11 @@ int safe_dup(const char *file, const int lineno, int oldfd);
 #define SAFE_DUP(oldfd) \
 	safe_dup(__FILE__, __LINE__, (oldfd))
 
+int safe_dup2(const char *file, const int lineno, int oldfd, int newfd);
+
+#define SAFE_DUP2(oldfd, newfd)			\
+	safe_dup2(__FILE__, __LINE__, (oldfd), (newfd))
+
 #define SAFE_GETCWD(buf, size) \
 	safe_getcwd(__FILE__, __LINE__, NULL, (buf), (size))
 
diff --git a/lib/tst_safe_macros.c b/lib/tst_safe_macros.c
index aa03a6d5c..182b690bb 100644
--- a/lib/tst_safe_macros.c
+++ b/lib/tst_safe_macros.c
@@ -415,6 +415,24 @@ int safe_dup(const char *file, const int lineno, int oldfd)
 	return rval;
 }
 
+int safe_dup2(const char *file, const int lineno, int oldfd, int newfd)
+{
+	int rval;
+
+	rval = dup2(oldfd, newfd);
+
+	if (rval == -1) {
+		tst_brk_(file, lineno, TBROK | TERRNO,
+			 "dup2(%i, %i) failed", oldfd, newfd);
+	} else if (rval != newfd) {
+		tst_brk_(file, lineno, TBROK | TERRNO,
+			 "Invalid dup2(%i, %i) return value %d",
+			 oldfd, newfd, rval);
+	}
+
+	return rval;
+}
+
 sighandler_t safe_signal(const char *file, const int lineno,
 	int signum, sighandler_t handler)
 {
-- 
2.30.0

[LTP] [PATCH v2 3/5] API: Fix clone.h

[Richard Palethorpe]

Functions in headers should be static inline. Also add stdint.h
because it results in confusing error messages if it is missing.

Signed-off-by: Richard Palethorpe <rpalethorpe@suse.com>
---
 include/lapi/clone.h | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/include/lapi/clone.h b/include/lapi/clone.h
index 2b8cbdbe0..81db443c9 100644
--- a/include/lapi/clone.h
+++ b/include/lapi/clone.h
@@ -10,6 +10,7 @@
 #include <sys/syscall.h>
 #include <linux/types.h>
 #include <sched.h>
+#include <stdint.h>
 
 #include "config.h"
 #include "lapi/syscalls.h"
@@ -26,7 +27,7 @@ struct clone_args {
 	uint64_t __attribute__((aligned(8))) tls;
 };
 
-int clone3(struct clone_args *args, size_t size)
+static inline int clone3(struct clone_args *args, size_t size)
 {
 	return tst_syscall(__NR_clone3, args, size);
 }
@@ -36,7 +37,7 @@ int clone3(struct clone_args *args, size_t size)
 #define CLONE_PIDFD	0x00001000	/* set if a pidfd should be placed in parent */
 #endif
 
-void clone3_supported_by_kernel(void)
+static inline void clone3_supported_by_kernel(void)
 {
 	if ((tst_kvercmp(5, 3, 0)) < 0) {
 		/* Check if the syscall is backported on an older kernel */
-- 
2.30.0

[LTP] [PATCH v2 4/5] API: Add tst_clone

[Richard Palethorpe]

The raw clone system call and clone3 have relatively simple interfaces
if the stack pointer is set to NULL. The libc wrapper complicates
things hugely. So introduce an interface similar to clone3, but that
falls back to clone.

Not all features of clone3 are implemented in clone, but we could
either return TCONF or implement them in user land.

Signed-off-by: Richard Palethorpe <rpalethorpe@suse.com>
---
 doc/test-writing-guidelines.txt | 27 +++++++++++++++----
 include/tst_clone.h             | 28 ++++++++++++++++++++
 include/tst_test.h              |  2 +-
 lib/tst_clone.c                 | 46 +++++++++++++++++++++++++++++++++
 lib/tst_test.c                  | 25 ++++++++++++++++++
 5 files changed, 122 insertions(+), 6 deletions(-)
 create mode 100644 lib/tst_clone.c

diff --git a/doc/test-writing-guidelines.txt b/doc/test-writing-guidelines.txt
index f3a55cf26..dd1911ceb 100644
--- a/doc/test-writing-guidelines.txt
+++ b/doc/test-writing-guidelines.txt
@@ -547,9 +547,9 @@ void tst_flush(void);
 Flush output streams, handling errors appropriately.
 
 This function is rarely needed when you have to flush the output streams
-before calling 'fork()' or 'clone()'. Note that the 'SAFE_FORK()' calls this
-function automatically. See 3.4 FILE buffers and fork() for explanation why is
-this needed.
+before calling 'fork()' or 'clone()'. Note that the 'SAFE_FORK()' and 'SAFE_CLONE()'
+calls this function automatically. See 3.4 FILE buffers and fork() for explanation
+why is this needed.
 
 2.2.3 Test temporary directory
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
@@ -732,10 +732,10 @@ This happens when 'stdout' gets redirected to a file. In this case, the
 'stdout' is not line buffered, but block buffered. Hence after a fork content
 of the buffers will be printed by the parent and each of the children.
 
-To avoid that you should use 'SAFE_FORK()'.
+To avoid that you should use 'SAFE_FORK()', 'SAFE_CLONE()' or 'tst_clone()'.
 
 IMPORTANT: You have to set the '.forks_child' flag in the test structure
-           if your testcase forks.
+           if your testcase forks or calls 'SAFE_CLONE()'.
 
 2.2.8 Doing the test in the child process
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
@@ -769,6 +769,23 @@ The 'tst_reap_children()' function makes the process wait for all of its
 children and exits with 'tst_brk(TBROK, ...)' if any of them returned
 a non zero exit code.
 
+When using 'SAFE_CLONE' or 'tst_clone', this may not work depending on
+the parameters passed to clone. The following call to 'SAFE_CLONE' is
+identical to 'fork()', so will work as expected.
+
+[source,c]
+--------------------------------------------------------------------------------
+const struct tst_clone_args args = {
+	.exit_signal = SIGCHLD,
+};
+
+SAFE_CLONE(&args);
+--------------------------------------------------------------------------------
+
+If 'exit_signal' is set to something else, then this will break
+'tst_reap_children'. It's not expected that all parameters to clone will
+work with the LTP library unless specific action is taken by the test code.
+
 .Using 'tst_res()' from binaries started by 'exec()'
 [source,c]
 -------------------------------------------------------------------------------
diff --git a/include/tst_clone.h b/include/tst_clone.h
index 88188525d..9ffdc68d1 100644
--- a/include/tst_clone.h
+++ b/include/tst_clone.h
@@ -5,6 +5,34 @@
 #ifndef TST_CLONE_H__
 #define TST_CLONE_H__
 
+#ifdef TST_TEST_H__
+
+/* The parts of clone3's clone_args we support */
+struct tst_clone_args {
+	uint64_t flags;
+	uint64_t exit_signal;
+};
+
+/* clone3 with fallbacks to clone when possible. Be aware that it
+ * returns -1 if clone3 fails (except ENOSYS), but -2 if clone fails.
+ *
+ * Without CLONE_VM this acts like fork so you may want to set
+ * tst_test.forks_child (safe_clone requires this).
+ *
+ * You should set exit_signal to SIGCHLD for
+ * tst_reap_children. Otherwise you must call wait with the
+ * appropriate parameters.
+ */
+pid_t tst_clone(const struct tst_clone_args *args);
+
+pid_t safe_clone(const char *file, const int lineno,
+		 const struct tst_clone_args *args);
+
+/* "Safe" version of tst_clone */
+#define SAFE_CLONE(args) safe_clone(__FILE__, __LINE__, args)
+
+#endif	/* TST_TEST_H__ */
+
 /* Functions from lib/cloner.c */
 int ltp_clone(unsigned long flags, int (*fn)(void *arg), void *arg,
 		size_t stack_size, void *stack);
diff --git a/include/tst_test.h b/include/tst_test.h
index c87251870..7dab5f761 100644
--- a/include/tst_test.h
+++ b/include/tst_test.h
@@ -29,7 +29,6 @@
 #include "tst_process_state.h"
 #include "tst_atomic.h"
 #include "tst_kvercmp.h"
-#include "tst_clone.h"
 #include "tst_kernel.h"
 #include "tst_minmax.h"
 #include "tst_get_bad_addr.h"
@@ -94,6 +93,7 @@ pid_t safe_fork(const char *filename, unsigned int lineno);
 #include "tst_safe_macros.h"
 #include "tst_safe_file_ops.h"
 #include "tst_safe_net.h"
+#include "tst_clone.h"
 
 /*
  * Wait for all children and exit with TBROK if
diff --git a/lib/tst_clone.c b/lib/tst_clone.c
new file mode 100644
index 000000000..07e7f0767
--- /dev/null
+++ b/lib/tst_clone.c
@@ -0,0 +1,46 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/* Copyright (c) 2021 SUSE LLC
+ * Richard Palethorpe <rpalethorpe@suse.com>
+ */
+
+#define TST_NO_DEFAULT_MAIN
+
+#include <stddef.h>
+
+#include "tst_test.h"
+#include "lapi/clone.h"
+
+pid_t tst_clone(const struct tst_clone_args *tst_args)
+{
+	struct clone_args args = {
+		.flags = tst_args->flags,
+		.exit_signal = tst_args->exit_signal,
+	};
+	int flags;
+	pid_t pid = -1;
+
+	tst_flush();
+
+	errno = ENOSYS;
+	if (__NR_clone3 != __LTP__NR_INVALID_SYSCALL)
+		pid = syscall(__NR_clone3, &args, sizeof(args));
+
+	if (pid == -1 && errno != ENOSYS)
+		return -1;
+
+	if (pid != -1)
+		return pid;
+
+	flags = args.exit_signal | args.flags;
+
+#ifdef __s390x__
+	pid = syscall(__NR_clone, NULL, flags);
+#else
+	pid = syscall(__NR_clone, flags, NULL);
+#endif
+
+	if (pid == -1)
+		return -2;
+
+	return pid;
+}
diff --git a/lib/tst_test.c b/lib/tst_test.c
index 0714f0a0e..6bbee030b 100644
--- a/lib/tst_test.c
+++ b/lib/tst_test.c
@@ -424,6 +424,31 @@ pid_t safe_fork(const char *filename, unsigned int lineno)
 	return pid;
 }
 
+pid_t safe_clone(const char *file, const int lineno,
+		 const struct tst_clone_args *args)
+{
+	pid_t pid;
+
+	if (!tst_test->forks_child)
+		tst_brk(TBROK, "test.forks_child must be set!");
+
+	pid = tst_clone(args);
+
+	switch (pid) {
+	case -1:
+		tst_brk_(file, lineno, TBROK | TERRNO, "clone3 failed");
+		break;
+	case -2:
+		tst_brk_(file, lineno, TBROK | TERRNO, "clone failed");
+		return -1;
+	}
+
+	if (!pid)
+		atexit(tst_free_all);
+
+	return pid;
+}
+
 static struct option {
 	char *optstr;
 	char *help;
-- 
2.30.0

[LTP] [PATCH v2 5/5] close_range: Add test

[Richard Palethorpe]

Signed-off-by: Richard Palethorpe <rpalethorpe@suse.com>
---
 .../kernel/syscalls/close_range/.gitignore    |   1 +
 .../kernel/syscalls/close_range/Makefile      |  10 +
 .../syscalls/close_range/close_range01.c      | 200 ++++++++++++++++++
 3 files changed, 211 insertions(+)
 create mode 100644 testcases/kernel/syscalls/close_range/.gitignore
 create mode 100644 testcases/kernel/syscalls/close_range/Makefile
 create mode 100644 testcases/kernel/syscalls/close_range/close_range01.c

diff --git a/testcases/kernel/syscalls/close_range/.gitignore b/testcases/kernel/syscalls/close_range/.gitignore
new file mode 100644
index 000000000..291a0379c
--- /dev/null
+++ b/testcases/kernel/syscalls/close_range/.gitignore
@@ -0,0 +1 @@
+close_range01
\ No newline at end of file
diff --git a/testcases/kernel/syscalls/close_range/Makefile b/testcases/kernel/syscalls/close_range/Makefile
new file mode 100644
index 000000000..dc6413b10
--- /dev/null
+++ b/testcases/kernel/syscalls/close_range/Makefile
@@ -0,0 +1,10 @@
+# SPDX-License-Identifier: GPL-2.0-or-later
+# Copyright (c) 2019-2021 Linux Test Project
+
+top_srcdir		?= ../../../..
+
+include $(top_srcdir)/include/mk/testcases.mk
+
+CFLAGS			+= -D_GNU_SOURCE
+
+include $(top_srcdir)/include/mk/generic_leaf_target.mk
diff --git a/testcases/kernel/syscalls/close_range/close_range01.c b/testcases/kernel/syscalls/close_range/close_range01.c
new file mode 100644
index 000000000..ed3f6b416
--- /dev/null
+++ b/testcases/kernel/syscalls/close_range/close_range01.c
@@ -0,0 +1,200 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Taken from the kernel self tests, which in turn were based on
+ * a Syzkaller reproducer.
+ *
+ * Self test author and close_range author:
+ * Christian Brauner <christian.brauner@ubuntu.com>
+ *
+ * LTP Author: Richard Palethorpe <rpalethorpe@suse.com>
+ * Copyright (c) 2021 SUSE LLC, other copyrights may apply.
+ *
+ * - First we test if close_range closes some FDs.
+ * - Then we test if it UNSHARES some FDs before closing them.
+ * - Then we test if it sets CLOEXEC (in both cloned process and parent).
+ * - Finally we test a combination of CLOEXEC and UNSHARE.
+ *
+ * The final test is the actual bug reproducer. Note that we call
+ * clone directly to share the file table.
+ */
+
+#include <stdlib.h>
+
+#include "tst_test.h"
+#include "tst_clone.h"
+
+#include "lapi/clone.h"
+#include "lapi/close_range.h"
+
+static int fd[3];
+
+static inline void do_close_range(unsigned int fd, unsigned int max_fd,
+				  unsigned int flags)
+{
+	int ret = close_range(fd, max_fd, flags);
+
+	if (!ret)
+		return;
+
+	if (errno == EINVAL) {
+		if (flags & CLOSE_RANGE_UNSHARE)
+			tst_brk(TCONF | TERRNO, "No CLOSE_RANGE_UNSHARE");
+		if (flags & CLOSE_RANGE_CLOEXEC)
+			tst_brk(TCONF | TERRNO, "No CLOSE_RANGE_CLOEXEC");
+	}
+
+	tst_brk(TBROK | TERRNO, "close_range(%d, %d, %d)", fd, max_fd, flags);
+}
+
+static void setup(void)
+{
+	struct rlimit nfd;
+
+	SAFE_GETRLIMIT(RLIMIT_NOFILE, &nfd);
+
+	if (nfd.rlim_max < 1000) {
+		tst_brk(TCONF, "NOFILE limit max too low: %lu < 1000",
+			nfd.rlim_max);
+	}
+
+	nfd.rlim_cur = nfd.rlim_max;
+	SAFE_SETRLIMIT(RLIMIT_NOFILE, &nfd);
+}
+
+static void check_cloexec(int i, int expected)
+{
+	int present = SAFE_FCNTL(fd[i], F_GETFD) & FD_CLOEXEC;
+
+	if (expected && !present)
+		tst_res(TFAIL, "fd[%d] flags do not contain FD_CLOEXEC", i);
+
+	if (!expected && present)
+		tst_res(TFAIL, "fd[%d] flags contain FD_CLOEXEC", i);
+}
+
+static void check_closed(int min)
+{
+	int i;
+
+	for (i = min; i < 3; i++) {
+		if (fcntl(fd[i], F_GETFD) > -1)
+			tst_res(TFAIL, "fd[%d] is still open", i);
+	}
+}
+
+static void child(unsigned int n)
+{
+	switch (n) {
+	case 0:
+		SAFE_DUP2(fd[1], fd[2]);
+		do_close_range(3, ~0U, 0);
+		check_closed(0);
+		break;
+	case 1:
+		SAFE_DUP2(fd[1], fd[2]);
+		do_close_range(3, ~0U, CLOSE_RANGE_UNSHARE);
+		check_closed(0);
+		break;
+	case 2:
+		do_close_range(3, ~0U, CLOSE_RANGE_CLOEXEC);
+		check_cloexec(0, 1);
+		check_cloexec(1, 1);
+
+		SAFE_DUP2(fd[1], fd[2]);
+		check_cloexec(2, 0);
+		break;
+	case 3:
+		do_close_range(3, ~0U,
+			       CLOSE_RANGE_CLOEXEC | CLOSE_RANGE_UNSHARE);
+		check_cloexec(0, 1);
+		check_cloexec(1, 1);
+
+		SAFE_DUP2(fd[1], fd[2]);
+		check_cloexec(2, 0);
+		break;
+	}
+
+	exit(0);
+}
+
+static void run(unsigned int n)
+{
+	const struct tst_clone_args args = {
+		.flags = CLONE_FILES,
+		.exit_signal = SIGCHLD,
+	};
+
+	switch (n) {
+	case 0:
+		tst_res(TINFO, "Plain close range");
+		do_close_range(3, ~0U, 0);
+		break;
+	case 1:
+		tst_res(TINFO, "Set UNSHARE and close range");
+		do_close_range(3, ~0U, CLOSE_RANGE_UNSHARE);
+		break;
+	case 2:
+		tst_res(TINFO, "Set CLOEXEC on range");
+		do_close_range(3, ~0U, CLOSE_RANGE_CLOEXEC);
+		break;
+	case 3:
+		tst_res(TINFO, "Set UNSHARE and CLOEXEC on range");
+		do_close_range(3, ~0U,
+			       CLOSE_RANGE_CLOEXEC | CLOSE_RANGE_UNSHARE);
+		break;
+	}
+
+	fd[0] = SAFE_OPEN("mnt/tmpfile", O_RDWR | O_CREAT, 0644);
+	fd[1] = SAFE_DUP2(fd[0], 1000);
+	fd[2] = 42;
+
+	if (!SAFE_CLONE(&args))
+		child(n);
+
+	tst_reap_children();
+
+	switch (n) {
+	case 0:
+		check_closed(0);
+		break;
+	case 1:
+		check_cloexec(0, 0);
+		check_cloexec(1, 0);
+		check_cloexec(2, 0);
+		break;
+	case 2:
+		check_cloexec(0, 1);
+		check_cloexec(1, 1);
+		check_cloexec(2, 0);
+		break;
+	case 3:
+		check_cloexec(0, 0);
+		check_cloexec(1, 0);
+		check_closed(2);
+		break;
+	}
+
+	do_close_range(3, ~0U, 0);
+	check_closed(0);
+
+	if (tst_taint_check())
+		tst_res(TFAIL, "Kernel tainted");
+	else
+		tst_res(TPASS, "No kernel taints");
+}
+
+static struct tst_test test = {
+	.tcnt = 4,
+	.needs_tmpdir = 1,
+	.forks_child = 1,
+	.mount_device = 1,
+	.mntpoint = "mnt",
+	.all_filesystems = 1,
+	.test = run,
+	.taint_check = TST_TAINT_W | TST_TAINT_D,
+	.setup = setup,
+	.tags = (const struct tst_tag[]) {
+		{"linux-git", "fec8a6a691033f2538cd46848f17f337f0739923"},
+		{},
+	},
+};
-- 
2.30.0

[LTP] [PATCH v2 5/5] close_range: Add test

[Petr Vorel]

Hi Richie,

> Signed-off-by: Richard Palethorpe <rpalethorpe@suse.com>
> ---
>  .../kernel/syscalls/close_range/.gitignore    |   1 +
>  .../kernel/syscalls/close_range/Makefile      |  10 +
>  .../syscalls/close_range/close_range01.c      | 200 ++++++++++++++++++

It looks like you haven't added close_range01 to runtest file.
It should probably go to syscalls and Cyril can add it before merging.

Patchset LGTM and compiles ok.
Thanks!

Kind regards,
Petr

[LTP] [PATCH v2 5/5] close_range: Add test

[Cyril Hrubis]

Hi!
> >  .../kernel/syscalls/close_range/.gitignore    |   1 +
> >  .../kernel/syscalls/close_range/Makefile      |  10 +
> >  .../syscalls/close_range/close_range01.c      | 200 ++++++++++++++++++
> 
> It looks like you haven't added close_range01 to runtest file.
> It should probably go to syscalls and Cyril can add it before merging.

That's what I did along with a few more minor changes and pushed.

> Patchset LGTM and compiles ok.

Sorry I haven't included your review in the tags.

-- 
Cyril Hrubis
chrubis@suse.cz

[LTP] [PATCH v2 5/5] close_range: Add test

[Cyril Hrubis]

Hi!
Pushed with small changes, thanks.

* Added a runtest file entry

* Add .needs_root which is what we have for tests that utilize loop devices

* Reformatted the top level comment so it's picked by docparser

* Shortened the git hash to 12 characters, since that's what we have in
  the rest of the tests


Also we are missing simple test for a negative cases, that is:

* Check that EINVAL is returned if we pass ~0 to flags

* And check that EINVAL is returned when fd < fd_max as well


And one more a bit more complicated test:

We should also get EMFILE if we do CLOSE_RANGE_UNSHARE and the parent
rlimit is exhausted.


Will you write these tests as well, or should I have a look?

-- 
Cyril Hrubis
chrubis@suse.cz

[LTP] [PATCH v2 5/5] close_range: Add test

[Petr Vorel]

Hi Cyril,

> Hi!
> > >  .../kernel/syscalls/close_range/.gitignore    |   1 +
> > >  .../kernel/syscalls/close_range/Makefile      |  10 +
> > >  .../syscalls/close_range/close_range01.c      | 200 ++++++++++++++++++

> > It looks like you haven't added close_range01 to runtest file.
> > It should probably go to syscalls and Cyril can add it before merging.

> That's what I did along with a few more minor changes and pushed.
Great :).

> > Patchset LGTM and compiles ok.

> Sorry I haven't included your review in the tags.
np :). Thanks for handling this.

Kind regards,
Petr

[LTP] [PATCH v2 5/5] close_range: Add test

[Richard Palethorpe]

Hello,

Cyril Hrubis <chrubis@suse.cz> writes:

> Hi!
> Pushed with small changes, thanks.
>
> * Added a runtest file entry
>
> * Add .needs_root which is what we have for tests that utilize loop
> devices

Ah, of course. I guess in that case we may also want to drop
CAP_SYS_ADMIN in caps.

>
> * Reformatted the top level comment so it's picked by docparser
>
> * Shortened the git hash to 12 characters, since that's what we have in
>   the rest of the tests
>

Ah, OK.

>
> Also we are missing simple test for a negative cases, that is:
>
> * Check that EINVAL is returned if we pass ~0 to flags
>
> * And check that EINVAL is returned when fd < fd_max as well
>
>
> And one more a bit more complicated test:
>
> We should also get EMFILE if we do CLOSE_RANGE_UNSHARE and the parent
> rlimit is exhausted.

Indeed those may be missing in the self tests as well.

>
>
> Will you write these tests as well, or should I have a look?

If you want to do it today very quickly, then feel free. Otherwise I
will do it next week.

-- 
Thank you,
Richard.