[PATCH 0/4] Tests for readahead() and fadvise() on overlayfs

[PATCH 0/4] Tests for readahead() and fadvise() on overlayfs

[Amir Goldstein]

Cyril,

The following series adds test covergae for readahead() syscall
over overlayfs file and adds test coverage for a specific
posix_fadvise() syscall advice (POSIX_FADV_WILLNEED).

So far, the posix_fadvise syscall tests in LTP only test for error
conditions, but not functionality. The functionality of the advice
POSIX_FADV_WILLNEED is indentical to readahead(), so I decided not
to duplicate the readahead functional test and add test cases to
readahead02 that use the POSIX_FADV_WILLNEED implementation.

Before I decided not to write new posix_fadvise() tests I converted
the existing tests to newlib. That is available for you as github PR:
https://github.com/linux-test-project/ltp/pull/401

Thanks,
Amir.

Amir Goldstein (4):
  syscalls/readahead01: Convert to newlib
  syscalls/readahead02: Convert to newlib and cleanup
  syscalls/readahead02: test readahead() on an overlayfs file
  syscalls/readahead02: test readahead using posix_fadvise()

 .../kernel/syscalls/readahead/readahead01.c   | 119 ++-----
 .../kernel/syscalls/readahead/readahead02.c   | 332 +++++++++---------
 2 files changed, 191 insertions(+), 260 deletions(-)

-- 
2.17.1

[LTP] [PATCH 0/4] Tests for readahead() and fadvise() on overlayfs

[Amir Goldstein]

Cyril,

The following series adds test covergae for readahead() syscall
over overlayfs file and adds test coverage for a specific
posix_fadvise() syscall advice (POSIX_FADV_WILLNEED).

So far, the posix_fadvise syscall tests in LTP only test for error
conditions, but not functionality. The functionality of the advice
POSIX_FADV_WILLNEED is indentical to readahead(), so I decided not
to duplicate the readahead functional test and add test cases to
readahead02 that use the POSIX_FADV_WILLNEED implementation.

Before I decided not to write new posix_fadvise() tests I converted
the existing tests to newlib. That is available for you as github PR:
https://github.com/linux-test-project/ltp/pull/401

Thanks,
Amir.

Amir Goldstein (4):
  syscalls/readahead01: Convert to newlib
  syscalls/readahead02: Convert to newlib and cleanup
  syscalls/readahead02: test readahead() on an overlayfs file
  syscalls/readahead02: test readahead using posix_fadvise()

 .../kernel/syscalls/readahead/readahead01.c   | 119 ++-----
 .../kernel/syscalls/readahead/readahead02.c   | 332 +++++++++---------
 2 files changed, 191 insertions(+), 260 deletions(-)

-- 
2.17.1

[PATCH 1/4] syscalls/readahead01: Convert to newlib

[Amir Goldstein]

* Use SPDX-License-Identifier

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
---
 .../kernel/syscalls/readahead/readahead01.c   | 119 +++++-------------
 1 file changed, 34 insertions(+), 85 deletions(-)

diff --git a/testcases/kernel/syscalls/readahead/readahead01.c b/testcases/kernel/syscalls/readahead/readahead01.c
index f35019488..682b524b3 100644
--- a/testcases/kernel/syscalls/readahead/readahead01.c
+++ b/testcases/kernel/syscalls/readahead/readahead01.c
@@ -1,25 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2012 Linux Test Project, Inc.
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of version 2 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.
- *
- * Further, this software is distributed without any warranty that it
- * is free of the rightful claim of any third person regarding
- * infringement or the like.  Any license provided herein, whether
- * implied or otherwise, applies only to this software file.  Patent
- * licenses, if any, provided herein do not apply to combinations of
- * this program with other software, or any other product whatsoever.
- *
- * 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 Street, Fifth Floor, Boston, MA
- * 02110-1301, USA.
  */
 
 /*
@@ -36,47 +17,36 @@
 #include <sys/syscall.h>
 #include <sys/types.h>
 #include "config.h"
-#include "test.h"
-#include "safe_macros.h"
+#include "tst_test.h"
 #include "lapi/syscalls.h"
 
-char *TCID = "readahead01";
-int TST_TOTAL = 1;
-
-option_t options[] = {
-	{NULL, NULL, NULL}
-};
-
 #if defined(__NR_readahead)
-static void setup(void);
-static void cleanup(void);
-
 static int check_ret(long expected_ret)
 {
-	if (expected_ret == TEST_RETURN) {
-		tst_resm(TPASS, "expected ret success - "
-			 "returned value = %ld", TEST_RETURN);
+	if (expected_ret == TST_RET) {
+		tst_res(TPASS, "expected ret success - "
+			"returned value = %ld", TST_RET);
 		return 0;
 	}
-	tst_resm(TFAIL, "unexpected failure - "
-		 "returned value = %ld, expected: %ld",
-		 TEST_RETURN, expected_ret);
+	tst_res(TFAIL, "unexpected failure - "
+		"returned value = %ld, expected: %ld",
+		TST_RET, expected_ret);
 	return 1;
 }
 
 static int check_errno(long expected_errno)
 {
-	if (TEST_ERRNO == expected_errno) {
-		tst_resm(TPASS | TTERRNO, "expected failure");
+	if (TST_ERR == expected_errno) {
+		tst_res(TPASS | TTERRNO, "expected failure");
 		return 0;
 	}
 
-	if (TEST_ERRNO == 0)
-		tst_resm(TFAIL, "call succeeded unexpectedly");
+	if (TST_ERR == 0)
+		tst_res(TFAIL, "call succeeded unexpectedly");
 	else
-		tst_resm(TFAIL | TTERRNO, "unexpected failure - "
-			 "expected = %ld : %s, actual",
-			 expected_errno, strerror(expected_errno));
+		tst_res(TFAIL | TTERRNO, "unexpected failure - "
+			"expected = %ld : %s, actual",
+			expected_errno, strerror(expected_errno));
 	return 1;
 }
 
@@ -85,19 +55,17 @@ static void test_bad_fd(void)
 	char tempname[PATH_MAX] = "readahead01_XXXXXX";
 	int fd;
 
-	tst_resm(TINFO, "test_bad_fd -1");
+	tst_res(TINFO, "test_bad_fd -1");
 	TEST(readahead(-1, 0, getpagesize()));
 	check_ret(-1);
 	check_errno(EBADF);
 
-	tst_resm(TINFO, "test_bad_fd O_WRONLY");
+	tst_res(TINFO, "test_bad_fd O_WRONLY");
 	fd = mkstemp(tempname);
 	if (fd == -1)
-		tst_resm(TBROK | TERRNO, "mkstemp failed");
+		tst_res(TBROK | TERRNO, "mkstemp failed");
 	close(fd);
-	fd = open(tempname, O_WRONLY);
-	if (fd == -1)
-		tst_resm(TBROK | TERRNO, "Failed to open testfile");
+	fd = SAFE_OPEN(tempname, O_WRONLY);
 	TEST(readahead(fd, 0, getpagesize()));
 	check_ret(-1);
 	check_errno(EBADF);
@@ -109,60 +77,41 @@ static void test_invalid_fd(void)
 {
 	int fd[2];
 
-	tst_resm(TINFO, "test_invalid_fd pipe");
-	if (pipe(fd) < 0)
-		tst_resm(TBROK | TERRNO, "Failed to create pipe");
+	tst_res(TINFO, "test_invalid_fd pipe");
+	SAFE_PIPE(fd);
 	TEST(readahead(fd[0], 0, getpagesize()));
 	check_ret(-1);
 	check_errno(EINVAL);
 	close(fd[0]);
 	close(fd[1]);
 
-	tst_resm(TINFO, "test_invalid_fd socket");
-	fd[0] = socket(AF_INET, SOCK_STREAM, 0);
-	if (fd[0] < 0)
-		tst_resm(TBROK | TERRNO, "Failed to create socket");
+	tst_res(TINFO, "test_invalid_fd socket");
+	fd[0] = SAFE_SOCKET(AF_INET, SOCK_STREAM, 0);
 	TEST(readahead(fd[0], 0, getpagesize()));
 	check_ret(-1);
 	check_errno(EINVAL);
 	close(fd[0]);
 }
 
-int main(int argc, char *argv[])
+void test_readahead(void)
 {
-	int lc;
-
-	tst_parse_opts(argc, argv, options, NULL);
-
-	setup();
-	for (lc = 0; TEST_LOOPING(lc); lc++) {
-		tst_count = 0;
-		test_bad_fd();
-		test_invalid_fd();
-	}
-	cleanup();
-	tst_exit();
+	test_bad_fd();
+	test_invalid_fd();
 }
 
 static void setup(void)
 {
-	tst_require_root();
-	tst_tmpdir();
-
 	/* check if readahead syscall is supported */
-	ltp_syscall(__NR_readahead, 0, 0, 0);
-
-	TEST_PAUSE;
+	tst_syscall(__NR_readahead, 0, 0, 0);
 }
 
-static void cleanup(void)
-{
-	tst_rmdir();
-}
+static struct tst_test test = {
+	.needs_root = 1,
+	.needs_tmpdir = 1,
+	.setup = setup,
+	.test_all = test_readahead,
+};
 
 #else /* __NR_readahead */
-int main(void)
-{
-	tst_brkm(TCONF, NULL, "System doesn't support __NR_readahead");
-}
+	TST_TEST_TCONF("System doesn't support __NR_readahead");
 #endif
-- 
2.17.1

[LTP] [PATCH 1/4] syscalls/readahead01: Convert to newlib

[Amir Goldstein]

* Use SPDX-License-Identifier

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
---
 .../kernel/syscalls/readahead/readahead01.c   | 119 +++++-------------
 1 file changed, 34 insertions(+), 85 deletions(-)

diff --git a/testcases/kernel/syscalls/readahead/readahead01.c b/testcases/kernel/syscalls/readahead/readahead01.c
index f35019488..682b524b3 100644
--- a/testcases/kernel/syscalls/readahead/readahead01.c
+++ b/testcases/kernel/syscalls/readahead/readahead01.c
@@ -1,25 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2012 Linux Test Project, Inc.
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of version 2 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.
- *
- * Further, this software is distributed without any warranty that it
- * is free of the rightful claim of any third person regarding
- * infringement or the like.  Any license provided herein, whether
- * implied or otherwise, applies only to this software file.  Patent
- * licenses, if any, provided herein do not apply to combinations of
- * this program with other software, or any other product whatsoever.
- *
- * 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 Street, Fifth Floor, Boston, MA
- * 02110-1301, USA.
  */
 
 /*
@@ -36,47 +17,36 @@
 #include <sys/syscall.h>
 #include <sys/types.h>
 #include "config.h"
-#include "test.h"
-#include "safe_macros.h"
+#include "tst_test.h"
 #include "lapi/syscalls.h"
 
-char *TCID = "readahead01";
-int TST_TOTAL = 1;
-
-option_t options[] = {
-	{NULL, NULL, NULL}
-};
-
 #if defined(__NR_readahead)
-static void setup(void);
-static void cleanup(void);
-
 static int check_ret(long expected_ret)
 {
-	if (expected_ret == TEST_RETURN) {
-		tst_resm(TPASS, "expected ret success - "
-			 "returned value = %ld", TEST_RETURN);
+	if (expected_ret == TST_RET) {
+		tst_res(TPASS, "expected ret success - "
+			"returned value = %ld", TST_RET);
 		return 0;
 	}
-	tst_resm(TFAIL, "unexpected failure - "
-		 "returned value = %ld, expected: %ld",
-		 TEST_RETURN, expected_ret);
+	tst_res(TFAIL, "unexpected failure - "
+		"returned value = %ld, expected: %ld",
+		TST_RET, expected_ret);
 	return 1;
 }
 
 static int check_errno(long expected_errno)
 {
-	if (TEST_ERRNO == expected_errno) {
-		tst_resm(TPASS | TTERRNO, "expected failure");
+	if (TST_ERR == expected_errno) {
+		tst_res(TPASS | TTERRNO, "expected failure");
 		return 0;
 	}
 
-	if (TEST_ERRNO == 0)
-		tst_resm(TFAIL, "call succeeded unexpectedly");
+	if (TST_ERR == 0)
+		tst_res(TFAIL, "call succeeded unexpectedly");
 	else
-		tst_resm(TFAIL | TTERRNO, "unexpected failure - "
-			 "expected = %ld : %s, actual",
-			 expected_errno, strerror(expected_errno));
+		tst_res(TFAIL | TTERRNO, "unexpected failure - "
+			"expected = %ld : %s, actual",
+			expected_errno, strerror(expected_errno));
 	return 1;
 }
 
@@ -85,19 +55,17 @@ static void test_bad_fd(void)
 	char tempname[PATH_MAX] = "readahead01_XXXXXX";
 	int fd;
 
-	tst_resm(TINFO, "test_bad_fd -1");
+	tst_res(TINFO, "test_bad_fd -1");
 	TEST(readahead(-1, 0, getpagesize()));
 	check_ret(-1);
 	check_errno(EBADF);
 
-	tst_resm(TINFO, "test_bad_fd O_WRONLY");
+	tst_res(TINFO, "test_bad_fd O_WRONLY");
 	fd = mkstemp(tempname);
 	if (fd == -1)
-		tst_resm(TBROK | TERRNO, "mkstemp failed");
+		tst_res(TBROK | TERRNO, "mkstemp failed");
 	close(fd);
-	fd = open(tempname, O_WRONLY);
-	if (fd == -1)
-		tst_resm(TBROK | TERRNO, "Failed to open testfile");
+	fd = SAFE_OPEN(tempname, O_WRONLY);
 	TEST(readahead(fd, 0, getpagesize()));
 	check_ret(-1);
 	check_errno(EBADF);
@@ -109,60 +77,41 @@ static void test_invalid_fd(void)
 {
 	int fd[2];
 
-	tst_resm(TINFO, "test_invalid_fd pipe");
-	if (pipe(fd) < 0)
-		tst_resm(TBROK | TERRNO, "Failed to create pipe");
+	tst_res(TINFO, "test_invalid_fd pipe");
+	SAFE_PIPE(fd);
 	TEST(readahead(fd[0], 0, getpagesize()));
 	check_ret(-1);
 	check_errno(EINVAL);
 	close(fd[0]);
 	close(fd[1]);
 
-	tst_resm(TINFO, "test_invalid_fd socket");
-	fd[0] = socket(AF_INET, SOCK_STREAM, 0);
-	if (fd[0] < 0)
-		tst_resm(TBROK | TERRNO, "Failed to create socket");
+	tst_res(TINFO, "test_invalid_fd socket");
+	fd[0] = SAFE_SOCKET(AF_INET, SOCK_STREAM, 0);
 	TEST(readahead(fd[0], 0, getpagesize()));
 	check_ret(-1);
 	check_errno(EINVAL);
 	close(fd[0]);
 }
 
-int main(int argc, char *argv[])
+void test_readahead(void)
 {
-	int lc;
-
-	tst_parse_opts(argc, argv, options, NULL);
-
-	setup();
-	for (lc = 0; TEST_LOOPING(lc); lc++) {
-		tst_count = 0;
-		test_bad_fd();
-		test_invalid_fd();
-	}
-	cleanup();
-	tst_exit();
+	test_bad_fd();
+	test_invalid_fd();
 }
 
 static void setup(void)
 {
-	tst_require_root();
-	tst_tmpdir();
-
 	/* check if readahead syscall is supported */
-	ltp_syscall(__NR_readahead, 0, 0, 0);
-
-	TEST_PAUSE;
+	tst_syscall(__NR_readahead, 0, 0, 0);
 }
 
-static void cleanup(void)
-{
-	tst_rmdir();
-}
+static struct tst_test test = {
+	.needs_root = 1,
+	.needs_tmpdir = 1,
+	.setup = setup,
+	.test_all = test_readahead,
+};
 
 #else /* __NR_readahead */
-int main(void)
-{
-	tst_brkm(TCONF, NULL, "System doesn't support __NR_readahead");
-}
+	TST_TEST_TCONF("System doesn't support __NR_readahead");
 #endif
-- 
2.17.1

[PATCH 2/4] syscalls/readahead02: Convert to newlib and cleanup

[Amir Goldstein]

* Use SAFE macros

* Use SPDX-License-Identifier

* No need to cleanup test file from temp dir

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
---
 .../kernel/syscalls/readahead/readahead02.c   | 246 +++++-------------
 1 file changed, 70 insertions(+), 176 deletions(-)

diff --git a/testcases/kernel/syscalls/readahead/readahead02.c b/testcases/kernel/syscalls/readahead/readahead02.c
index c9d92a6a4..4132b4de1 100644
--- a/testcases/kernel/syscalls/readahead/readahead02.c
+++ b/testcases/kernel/syscalls/readahead/readahead02.c
@@ -1,25 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2012 Linux Test Project, Inc.
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of version 2 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.
- *
- * Further, this software is distributed without any warranty that it
- * is free of the rightful claim of any third person regarding
- * infringement or the like.  Any license provided herein, whether
- * implied or otherwise, applies only to this software file.  Patent
- * licenses, if any, provided herein do not apply to combinations of
- * this program with other software, or any other product whatsoever.
- *
- * 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 Street, Fifth Floor, Boston, MA
- * 02110-1301, USA.
  */
 
 /*
@@ -43,86 +24,55 @@
 #include <unistd.h>
 #include <fcntl.h>
 #include "config.h"
-#include "test.h"
-#include "safe_macros.h"
+#include "tst_test.h"
 #include "lapi/syscalls.h"
 
-char *TCID = "readahead02";
-int TST_TOTAL = 1;
-
 #if defined(__NR_readahead)
 static char testfile[PATH_MAX] = "testfile";
 static const char drop_caches_fname[] = "/proc/sys/vm/drop_caches";
 static const char meminfo_fname[] = "/proc/meminfo";
 static size_t testfile_size = 64 * 1024 * 1024;
-static int opt_fsize;
 static char *opt_fsizestr;
 static int pagesize;
-static const char *fs_type;
-static const char *device;
-static int mount_flag;
 
 #define MNTPOINT        "mntpoint"
+static const char mntpoint[] = "mntpoint";
 #define MIN_SANE_READAHEAD (4u * 1024u)
 
-option_t options[] = {
-	{"s:", &opt_fsize, &opt_fsizestr},
+static struct tst_option options[] = {
+	{"s:", &opt_fsizestr, "-s    testfile size (default 64MB)"},
 	{NULL, NULL, NULL}
 };
 
-static void setup(void);
-static void cleanup(void);
-
-static void help(void)
-{
-	printf("  -s x    testfile size (default 64MB)\n");
-}
-
 static int check_ret(long expected_ret)
 {
-	if (expected_ret == TEST_RETURN) {
-		tst_resm(TPASS, "expected ret success - "
-			 "returned value = %ld", TEST_RETURN);
+	if (expected_ret == TST_RET) {
+		tst_res(TPASS, "expected ret success - "
+			"returned value = %ld", TST_RET);
 		return 0;
 	}
-	tst_resm(TFAIL | TTERRNO, "unexpected failure - "
-		 "returned value = %ld, expected: %ld",
-		 TEST_RETURN, expected_ret);
+	tst_res(TFAIL | TTERRNO, "unexpected failure - "
+		"returned value = %ld, expected: %ld",
+		TST_RET, expected_ret);
 	return 1;
 }
 
 static int has_file(const char *fname, int required)
 {
-	int ret;
 	struct stat buf;
-	ret = stat(fname, &buf);
-	if (ret == -1) {
-		if (errno == ENOENT)
-			if (required)
-				tst_brkm(TCONF, cleanup, "%s not available",
-					 fname);
-			else
-				return 0;
-		else
-			tst_brkm(TBROK | TERRNO, cleanup, "stat %s", fname);
+
+	if (stat(fname, &buf) == -1) {
+		if (errno != ENOENT)
+			tst_brk(TBROK | TERRNO, "stat %s", fname);
+		if (required)
+			tst_brk(TCONF, "%s not available", fname);
 	}
 	return 1;
 }
 
 static void drop_caches(void)
 {
-	int ret;
-	FILE *f;
-
-	f = fopen(drop_caches_fname, "w");
-	if (f) {
-		ret = fprintf(f, "1");
-		fclose(f);
-		if (ret < 1)
-			tst_brkm(TBROK, cleanup, "Failed to drop caches");
-	} else {
-		tst_brkm(TBROK, cleanup, "Failed to open drop_caches");
-	}
+	SAFE_FILE_PRINTF(drop_caches_fname, "1");
 }
 
 static unsigned long parse_entry(const char *fname, const char *entry)
@@ -161,32 +111,21 @@ static unsigned long get_cached_size(void)
 
 static void create_testfile(void)
 {
-	FILE *f;
+	int fd;
 	char *tmp;
 	size_t i;
 
-	tst_resm(TINFO, "creating test file of size: %zu", testfile_size);
-	tmp = SAFE_MALLOC(cleanup, pagesize);
+	tst_res(TINFO, "creating test file of size: %zu", testfile_size);
+	tmp = SAFE_MALLOC(pagesize);
 
 	/* round to page size */
 	testfile_size = testfile_size & ~((long)pagesize - 1);
 
-	f = fopen(testfile, "w");
-	if (!f) {
-		free(tmp);
-		tst_brkm(TBROK | TERRNO, cleanup, "Failed to create %s",
-			 testfile);
-	}
-
+	fd = SAFE_CREAT(testfile, 0644);
 	for (i = 0; i < testfile_size; i += pagesize)
-		if (fwrite(tmp, pagesize, 1, f) < 1) {
-			free(tmp);
-			tst_brkm(TBROK, cleanup, "Failed to create %s",
-				 testfile);
-		}
-	fflush(f);
-	fsync(fileno(f));
-	fclose(f);
+		SAFE_WRITE(1, fd, tmp, pagesize);
+	SAFE_FSYNC(fd);
+	SAFE_CLOSE(fd);
 	free(tmp);
 }
 
@@ -215,13 +154,13 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 	off_t offset = 0;
 	struct timeval now;
 
-	fd = SAFE_OPEN(cleanup, fname, O_RDONLY);
+	fd = SAFE_OPEN(fname, O_RDONLY);
 
 	if (do_readahead) {
 		cached_start = get_cached_size();
 		do {
 			TEST(readahead(fd, offset, fsize - offset));
-			if (TEST_RETURN != 0) {
+			if (TST_RET != 0) {
 				check_ret(0);
 				break;
 			}
@@ -232,7 +171,7 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 				if (*cached > cached_start) {
 					max_ra_estimate = (1024 *
 						(*cached - cached_start));
-					tst_resm(TINFO, "max ra estimate: %lu",
+					tst_res(TINFO, "max ra estimate: %lu",
 						max_ra_estimate);
 				}
 				max_ra_estimate = MAX(max_ra_estimate,
@@ -242,27 +181,23 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 			i++;
 			offset += max_ra_estimate;
 		} while ((size_t)offset < fsize);
-		tst_resm(TINFO, "readahead calls made: %zu", i);
+		tst_res(TINFO, "readahead calls made: %zu", i);
 		*cached = get_cached_size();
 
 		/* offset of file shouldn't change after readahead */
-		offset = lseek(fd, 0, SEEK_CUR);
-		if (offset == (off_t) - 1)
-			tst_brkm(TBROK | TERRNO, cleanup, "lseek failed");
+		offset = SAFE_LSEEK(fd, 0, SEEK_CUR);
 		if (offset == 0)
-			tst_resm(TPASS, "offset is still at 0 as expected");
+			tst_res(TPASS, "offset is still at 0 as expected");
 		else
-			tst_resm(TFAIL, "offset has changed to: %lu", offset);
+			tst_res(TFAIL, "offset has changed to: %lu", offset);
 	}
 
 	if (gettimeofday(&now, NULL) == -1)
-		tst_brkm(TBROK | TERRNO, cleanup, "gettimeofday failed");
+		tst_brk(TBROK | TERRNO, "gettimeofday failed");
 	time_start_usec = now.tv_sec * 1000000 + now.tv_usec;
 	read_bytes_start = get_bytes_read();
 
-	p = mmap(NULL, fsize, PROT_READ, MAP_SHARED | MAP_POPULATE, fd, 0);
-	if (p == MAP_FAILED)
-		tst_brkm(TBROK | TERRNO, cleanup, "mmap failed");
+	p = SAFE_MMAP(NULL, fsize, PROT_READ, MAP_SHARED | MAP_POPULATE, fd, 0);
 
 	/* for old kernels, where MAP_POPULATE doesn't work, touch each page */
 	tmp = 0;
@@ -270,20 +205,20 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 		tmp = tmp ^ p[i];
 	/* prevent gcc from optimizing out loop above */
 	if (tmp != 0)
-		tst_brkm(TBROK, NULL, "This line should not be reached");
+		tst_brk(TBROK, "This line should not be reached");
 
 	if (!do_readahead)
 		*cached = get_cached_size();
 
-	SAFE_MUNMAP(cleanup, p, fsize);
+	SAFE_MUNMAP(p, fsize);
 
 	*read_bytes = get_bytes_read() - read_bytes_start;
 	if (gettimeofday(&now, NULL) == -1)
-		tst_brkm(TBROK | TERRNO, cleanup, "gettimeofday failed");
+		tst_brk(TBROK | TERRNO, "gettimeofday failed");
 	time_end_usec = now.tv_sec * 1000000 + now.tv_usec;
 	*usec = time_end_usec - time_start_usec;
 
-	SAFE_CLOSE(cleanup, fd);
+	SAFE_CLOSE(fd);
 }
 
 static void test_readahead(void)
@@ -302,7 +237,7 @@ static void test_readahead(void)
 	cached_low = get_cached_size();
 	cached_max = cached_max - cached_low;
 
-	tst_resm(TINFO, "read_testfile(0)");
+	tst_res(TINFO, "read_testfile(0)");
 	read_testfile(0, testfile, testfile_size, &read_bytes, &usec, &cached);
 	if (cached > cached_low)
 		cached = cached - cached_low;
@@ -312,7 +247,7 @@ static void test_readahead(void)
 	sync();
 	drop_caches();
 	cached_low = get_cached_size();
-	tst_resm(TINFO, "read_testfile(1)");
+	tst_res(TINFO, "read_testfile(1)");
 	read_testfile(1, testfile, testfile_size, &read_bytes_ra,
 		      &usec_ra, &cached_ra);
 	if (cached_ra > cached_low)
@@ -320,25 +255,25 @@ static void test_readahead(void)
 	else
 		cached_ra = 0;
 
-	tst_resm(TINFO, "read_testfile(0) took: %ld usec", usec);
-	tst_resm(TINFO, "read_testfile(1) took: %ld usec", usec_ra);
+	tst_res(TINFO, "read_testfile(0) took: %ld usec", usec);
+	tst_res(TINFO, "read_testfile(1) took: %ld usec", usec_ra);
 	if (has_file(proc_io_fname, 0)) {
-		tst_resm(TINFO, "read_testfile(0) read: %ld bytes", read_bytes);
-		tst_resm(TINFO, "read_testfile(1) read: %ld bytes",
-			 read_bytes_ra);
+		tst_res(TINFO, "read_testfile(0) read: %ld bytes", read_bytes);
+		tst_res(TINFO, "read_testfile(1) read: %ld bytes",
+			read_bytes_ra);
 		/* actual number of read bytes depends on total RAM */
 		if (read_bytes_ra < read_bytes)
-			tst_resm(TPASS, "readahead saved some I/O");
+			tst_res(TPASS, "readahead saved some I/O");
 		else
-			tst_resm(TFAIL, "readahead failed to save any I/O");
+			tst_res(TFAIL, "readahead failed to save any I/O");
 	} else {
-		tst_resm(TCONF, "Your system doesn't have /proc/pid/io,"
-			 " unable to determine read bytes during test");
+		tst_res(TCONF, "Your system doesn't have /proc/pid/io,"
+			" unable to determine read bytes during test");
 	}
 
-	tst_resm(TINFO, "cache can hold at least: %ld kB", cached_max);
-	tst_resm(TINFO, "read_testfile(0) used cache: %ld kB", cached);
-	tst_resm(TINFO, "read_testfile(1) used cache: %ld kB", cached_ra);
+	tst_res(TINFO, "cache can hold at least: %ld kB", cached_max);
+	tst_res(TINFO, "read_testfile(0) used cache: %ld kB", cached);
+	tst_res(TINFO, "read_testfile(1) used cache: %ld kB", cached_ra);
 
 	if (cached_max * 1024 >= testfile_size) {
 		/*
@@ -346,83 +281,42 @@ static void test_readahead(void)
 		 * for readahead should be at least testfile_size/2
 		 */
 		if (cached_ra * 1024 > testfile_size / 2)
-			tst_resm(TPASS, "using cache as expected");
+			tst_res(TPASS, "using cache as expected");
 		else
-			tst_resm(TWARN, "using less cache than expected");
+			tst_res(TWARN, "using less cache than expected");
 	} else {
-		tst_resm(TCONF, "Page cache on your system is too small "
-			 "to hold whole testfile.");
+		tst_res(TCONF, "Page cache on your system is too small "
+			"to hold whole testfile.");
 	}
 }
 
-int main(int argc, char *argv[])
-{
-	int lc;
-
-	tst_parse_opts(argc, argv, options, help);
-
-	if (opt_fsize)
-		testfile_size = atoi(opt_fsizestr);
-
-	setup();
-	for (lc = 0; TEST_LOOPING(lc); lc++) {
-		tst_count = 0;
-		test_readahead();
-	}
-	cleanup();
-	tst_exit();
-}
-
 static void setup(void)
 {
-	tst_require_root();
-	tst_tmpdir();
-	TEST_PAUSE;
+	if (opt_fsizestr)
+		testfile_size = SAFE_STRTOL(opt_fsizestr, 1, INT_MAX);
 
 	has_file(drop_caches_fname, 1);
 	has_file(meminfo_fname, 1);
 
 	/* check if readahead is supported */
-	ltp_syscall(__NR_readahead, 0, 0, 0);
+	tst_syscall(__NR_readahead, 0, 0, 0);
 
 	pagesize = getpagesize();
 
-	if (tst_fs_type(cleanup, ".") == TST_TMPFS_MAGIC) {
-		tst_resm(TINFO, "TMPFS detected, creating loop device");
-
-		fs_type = tst_dev_fs_type();
-		device = tst_acquire_device(cleanup);
-		if (!device) {
-			tst_brkm(TCONF, cleanup,
-				"Failed to obtain block device");
-		}
-
-		tst_mkfs(cleanup, device, fs_type, NULL, NULL);
-
-		SAFE_MKDIR(cleanup, MNTPOINT, 0755);
-		SAFE_MOUNT(cleanup, device, MNTPOINT, fs_type, 0, NULL);
-		mount_flag = 1;
-
-		sprintf(testfile, MNTPOINT"/testfile");
-	}
+	sprintf(testfile, "%s/testfile", mntpoint);
 	create_testfile();
 }
 
-static void cleanup(void)
-{
-	unlink(testfile);
-	if (mount_flag && tst_umount(MNTPOINT) < 0)
-		tst_resm(TWARN | TERRNO, "umount device:%s failed", device);
-
-	if (device)
-		tst_release_device(device);
-
-	tst_rmdir();
-}
+static struct tst_test test = {
+	.needs_root = 1,
+	.needs_tmpdir = 1,
+	.mount_device = 1,
+	.mntpoint = mntpoint,
+	.setup = setup,
+	.options = options,
+	.test_all = test_readahead,
+};
 
 #else /* __NR_readahead */
-int main(void)
-{
-	tst_brkm(TCONF, NULL, "System doesn't support __NR_readahead");
-}
+	TST_TEST_TCONF("System doesn't support __NR_readahead");
 #endif
-- 
2.17.1

[LTP] [PATCH 2/4] syscalls/readahead02: Convert to newlib and cleanup

[Amir Goldstein]

* Use SAFE macros

* Use SPDX-License-Identifier

* No need to cleanup test file from temp dir

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
---
 .../kernel/syscalls/readahead/readahead02.c   | 246 +++++-------------
 1 file changed, 70 insertions(+), 176 deletions(-)

diff --git a/testcases/kernel/syscalls/readahead/readahead02.c b/testcases/kernel/syscalls/readahead/readahead02.c
index c9d92a6a4..4132b4de1 100644
--- a/testcases/kernel/syscalls/readahead/readahead02.c
+++ b/testcases/kernel/syscalls/readahead/readahead02.c
@@ -1,25 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0
 /*
  * Copyright (C) 2012 Linux Test Project, Inc.
- *
- * This program is free software; you can redistribute it and/or
- * modify it under the terms of version 2 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.
- *
- * Further, this software is distributed without any warranty that it
- * is free of the rightful claim of any third person regarding
- * infringement or the like.  Any license provided herein, whether
- * implied or otherwise, applies only to this software file.  Patent
- * licenses, if any, provided herein do not apply to combinations of
- * this program with other software, or any other product whatsoever.
- *
- * 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 Street, Fifth Floor, Boston, MA
- * 02110-1301, USA.
  */
 
 /*
@@ -43,86 +24,55 @@
 #include <unistd.h>
 #include <fcntl.h>
 #include "config.h"
-#include "test.h"
-#include "safe_macros.h"
+#include "tst_test.h"
 #include "lapi/syscalls.h"
 
-char *TCID = "readahead02";
-int TST_TOTAL = 1;
-
 #if defined(__NR_readahead)
 static char testfile[PATH_MAX] = "testfile";
 static const char drop_caches_fname[] = "/proc/sys/vm/drop_caches";
 static const char meminfo_fname[] = "/proc/meminfo";
 static size_t testfile_size = 64 * 1024 * 1024;
-static int opt_fsize;
 static char *opt_fsizestr;
 static int pagesize;
-static const char *fs_type;
-static const char *device;
-static int mount_flag;
 
 #define MNTPOINT        "mntpoint"
+static const char mntpoint[] = "mntpoint";
 #define MIN_SANE_READAHEAD (4u * 1024u)
 
-option_t options[] = {
-	{"s:", &opt_fsize, &opt_fsizestr},
+static struct tst_option options[] = {
+	{"s:", &opt_fsizestr, "-s    testfile size (default 64MB)"},
 	{NULL, NULL, NULL}
 };
 
-static void setup(void);
-static void cleanup(void);
-
-static void help(void)
-{
-	printf("  -s x    testfile size (default 64MB)\n");
-}
-
 static int check_ret(long expected_ret)
 {
-	if (expected_ret == TEST_RETURN) {
-		tst_resm(TPASS, "expected ret success - "
-			 "returned value = %ld", TEST_RETURN);
+	if (expected_ret == TST_RET) {
+		tst_res(TPASS, "expected ret success - "
+			"returned value = %ld", TST_RET);
 		return 0;
 	}
-	tst_resm(TFAIL | TTERRNO, "unexpected failure - "
-		 "returned value = %ld, expected: %ld",
-		 TEST_RETURN, expected_ret);
+	tst_res(TFAIL | TTERRNO, "unexpected failure - "
+		"returned value = %ld, expected: %ld",
+		TST_RET, expected_ret);
 	return 1;
 }
 
 static int has_file(const char *fname, int required)
 {
-	int ret;
 	struct stat buf;
-	ret = stat(fname, &buf);
-	if (ret == -1) {
-		if (errno == ENOENT)
-			if (required)
-				tst_brkm(TCONF, cleanup, "%s not available",
-					 fname);
-			else
-				return 0;
-		else
-			tst_brkm(TBROK | TERRNO, cleanup, "stat %s", fname);
+
+	if (stat(fname, &buf) == -1) {
+		if (errno != ENOENT)
+			tst_brk(TBROK | TERRNO, "stat %s", fname);
+		if (required)
+			tst_brk(TCONF, "%s not available", fname);
 	}
 	return 1;
 }
 
 static void drop_caches(void)
 {
-	int ret;
-	FILE *f;
-
-	f = fopen(drop_caches_fname, "w");
-	if (f) {
-		ret = fprintf(f, "1");
-		fclose(f);
-		if (ret < 1)
-			tst_brkm(TBROK, cleanup, "Failed to drop caches");
-	} else {
-		tst_brkm(TBROK, cleanup, "Failed to open drop_caches");
-	}
+	SAFE_FILE_PRINTF(drop_caches_fname, "1");
 }
 
 static unsigned long parse_entry(const char *fname, const char *entry)
@@ -161,32 +111,21 @@ static unsigned long get_cached_size(void)
 
 static void create_testfile(void)
 {
-	FILE *f;
+	int fd;
 	char *tmp;
 	size_t i;
 
-	tst_resm(TINFO, "creating test file of size: %zu", testfile_size);
-	tmp = SAFE_MALLOC(cleanup, pagesize);
+	tst_res(TINFO, "creating test file of size: %zu", testfile_size);
+	tmp = SAFE_MALLOC(pagesize);
 
 	/* round to page size */
 	testfile_size = testfile_size & ~((long)pagesize - 1);
 
-	f = fopen(testfile, "w");
-	if (!f) {
-		free(tmp);
-		tst_brkm(TBROK | TERRNO, cleanup, "Failed to create %s",
-			 testfile);
-	}
-
+	fd = SAFE_CREAT(testfile, 0644);
 	for (i = 0; i < testfile_size; i += pagesize)
-		if (fwrite(tmp, pagesize, 1, f) < 1) {
-			free(tmp);
-			tst_brkm(TBROK, cleanup, "Failed to create %s",
-				 testfile);
-		}
-	fflush(f);
-	fsync(fileno(f));
-	fclose(f);
+		SAFE_WRITE(1, fd, tmp, pagesize);
+	SAFE_FSYNC(fd);
+	SAFE_CLOSE(fd);
 	free(tmp);
 }
 
@@ -215,13 +154,13 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 	off_t offset = 0;
 	struct timeval now;
 
-	fd = SAFE_OPEN(cleanup, fname, O_RDONLY);
+	fd = SAFE_OPEN(fname, O_RDONLY);
 
 	if (do_readahead) {
 		cached_start = get_cached_size();
 		do {
 			TEST(readahead(fd, offset, fsize - offset));
-			if (TEST_RETURN != 0) {
+			if (TST_RET != 0) {
 				check_ret(0);
 				break;
 			}
@@ -232,7 +171,7 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 				if (*cached > cached_start) {
 					max_ra_estimate = (1024 *
 						(*cached - cached_start));
-					tst_resm(TINFO, "max ra estimate: %lu",
+					tst_res(TINFO, "max ra estimate: %lu",
 						max_ra_estimate);
 				}
 				max_ra_estimate = MAX(max_ra_estimate,
@@ -242,27 +181,23 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 			i++;
 			offset += max_ra_estimate;
 		} while ((size_t)offset < fsize);
-		tst_resm(TINFO, "readahead calls made: %zu", i);
+		tst_res(TINFO, "readahead calls made: %zu", i);
 		*cached = get_cached_size();
 
 		/* offset of file shouldn't change after readahead */
-		offset = lseek(fd, 0, SEEK_CUR);
-		if (offset == (off_t) - 1)
-			tst_brkm(TBROK | TERRNO, cleanup, "lseek failed");
+		offset = SAFE_LSEEK(fd, 0, SEEK_CUR);
 		if (offset == 0)
-			tst_resm(TPASS, "offset is still at 0 as expected");
+			tst_res(TPASS, "offset is still@0 as expected");
 		else
-			tst_resm(TFAIL, "offset has changed to: %lu", offset);
+			tst_res(TFAIL, "offset has changed to: %lu", offset);
 	}
 
 	if (gettimeofday(&now, NULL) == -1)
-		tst_brkm(TBROK | TERRNO, cleanup, "gettimeofday failed");
+		tst_brk(TBROK | TERRNO, "gettimeofday failed");
 	time_start_usec = now.tv_sec * 1000000 + now.tv_usec;
 	read_bytes_start = get_bytes_read();
 
-	p = mmap(NULL, fsize, PROT_READ, MAP_SHARED | MAP_POPULATE, fd, 0);
-	if (p == MAP_FAILED)
-		tst_brkm(TBROK | TERRNO, cleanup, "mmap failed");
+	p = SAFE_MMAP(NULL, fsize, PROT_READ, MAP_SHARED | MAP_POPULATE, fd, 0);
 
 	/* for old kernels, where MAP_POPULATE doesn't work, touch each page */
 	tmp = 0;
@@ -270,20 +205,20 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 		tmp = tmp ^ p[i];
 	/* prevent gcc from optimizing out loop above */
 	if (tmp != 0)
-		tst_brkm(TBROK, NULL, "This line should not be reached");
+		tst_brk(TBROK, "This line should not be reached");
 
 	if (!do_readahead)
 		*cached = get_cached_size();
 
-	SAFE_MUNMAP(cleanup, p, fsize);
+	SAFE_MUNMAP(p, fsize);
 
 	*read_bytes = get_bytes_read() - read_bytes_start;
 	if (gettimeofday(&now, NULL) == -1)
-		tst_brkm(TBROK | TERRNO, cleanup, "gettimeofday failed");
+		tst_brk(TBROK | TERRNO, "gettimeofday failed");
 	time_end_usec = now.tv_sec * 1000000 + now.tv_usec;
 	*usec = time_end_usec - time_start_usec;
 
-	SAFE_CLOSE(cleanup, fd);
+	SAFE_CLOSE(fd);
 }
 
 static void test_readahead(void)
@@ -302,7 +237,7 @@ static void test_readahead(void)
 	cached_low = get_cached_size();
 	cached_max = cached_max - cached_low;
 
-	tst_resm(TINFO, "read_testfile(0)");
+	tst_res(TINFO, "read_testfile(0)");
 	read_testfile(0, testfile, testfile_size, &read_bytes, &usec, &cached);
 	if (cached > cached_low)
 		cached = cached - cached_low;
@@ -312,7 +247,7 @@ static void test_readahead(void)
 	sync();
 	drop_caches();
 	cached_low = get_cached_size();
-	tst_resm(TINFO, "read_testfile(1)");
+	tst_res(TINFO, "read_testfile(1)");
 	read_testfile(1, testfile, testfile_size, &read_bytes_ra,
 		      &usec_ra, &cached_ra);
 	if (cached_ra > cached_low)
@@ -320,25 +255,25 @@ static void test_readahead(void)
 	else
 		cached_ra = 0;
 
-	tst_resm(TINFO, "read_testfile(0) took: %ld usec", usec);
-	tst_resm(TINFO, "read_testfile(1) took: %ld usec", usec_ra);
+	tst_res(TINFO, "read_testfile(0) took: %ld usec", usec);
+	tst_res(TINFO, "read_testfile(1) took: %ld usec", usec_ra);
 	if (has_file(proc_io_fname, 0)) {
-		tst_resm(TINFO, "read_testfile(0) read: %ld bytes", read_bytes);
-		tst_resm(TINFO, "read_testfile(1) read: %ld bytes",
-			 read_bytes_ra);
+		tst_res(TINFO, "read_testfile(0) read: %ld bytes", read_bytes);
+		tst_res(TINFO, "read_testfile(1) read: %ld bytes",
+			read_bytes_ra);
 		/* actual number of read bytes depends on total RAM */
 		if (read_bytes_ra < read_bytes)
-			tst_resm(TPASS, "readahead saved some I/O");
+			tst_res(TPASS, "readahead saved some I/O");
 		else
-			tst_resm(TFAIL, "readahead failed to save any I/O");
+			tst_res(TFAIL, "readahead failed to save any I/O");
 	} else {
-		tst_resm(TCONF, "Your system doesn't have /proc/pid/io,"
-			 " unable to determine read bytes during test");
+		tst_res(TCONF, "Your system doesn't have /proc/pid/io,"
+			" unable to determine read bytes during test");
 	}
 
-	tst_resm(TINFO, "cache can hold at least: %ld kB", cached_max);
-	tst_resm(TINFO, "read_testfile(0) used cache: %ld kB", cached);
-	tst_resm(TINFO, "read_testfile(1) used cache: %ld kB", cached_ra);
+	tst_res(TINFO, "cache can hold@least: %ld kB", cached_max);
+	tst_res(TINFO, "read_testfile(0) used cache: %ld kB", cached);
+	tst_res(TINFO, "read_testfile(1) used cache: %ld kB", cached_ra);
 
 	if (cached_max * 1024 >= testfile_size) {
 		/*
@@ -346,83 +281,42 @@ static void test_readahead(void)
 		 * for readahead should be at least testfile_size/2
 		 */
 		if (cached_ra * 1024 > testfile_size / 2)
-			tst_resm(TPASS, "using cache as expected");
+			tst_res(TPASS, "using cache as expected");
 		else
-			tst_resm(TWARN, "using less cache than expected");
+			tst_res(TWARN, "using less cache than expected");
 	} else {
-		tst_resm(TCONF, "Page cache on your system is too small "
-			 "to hold whole testfile.");
+		tst_res(TCONF, "Page cache on your system is too small "
+			"to hold whole testfile.");
 	}
 }
 
-int main(int argc, char *argv[])
-{
-	int lc;
-
-	tst_parse_opts(argc, argv, options, help);
-
-	if (opt_fsize)
-		testfile_size = atoi(opt_fsizestr);
-
-	setup();
-	for (lc = 0; TEST_LOOPING(lc); lc++) {
-		tst_count = 0;
-		test_readahead();
-	}
-	cleanup();
-	tst_exit();
-}
-
 static void setup(void)
 {
-	tst_require_root();
-	tst_tmpdir();
-	TEST_PAUSE;
+	if (opt_fsizestr)
+		testfile_size = SAFE_STRTOL(opt_fsizestr, 1, INT_MAX);
 
 	has_file(drop_caches_fname, 1);
 	has_file(meminfo_fname, 1);
 
 	/* check if readahead is supported */
-	ltp_syscall(__NR_readahead, 0, 0, 0);
+	tst_syscall(__NR_readahead, 0, 0, 0);
 
 	pagesize = getpagesize();
 
-	if (tst_fs_type(cleanup, ".") == TST_TMPFS_MAGIC) {
-		tst_resm(TINFO, "TMPFS detected, creating loop device");
-
-		fs_type = tst_dev_fs_type();
-		device = tst_acquire_device(cleanup);
-		if (!device) {
-			tst_brkm(TCONF, cleanup,
-				"Failed to obtain block device");
-		}
-
-		tst_mkfs(cleanup, device, fs_type, NULL, NULL);
-
-		SAFE_MKDIR(cleanup, MNTPOINT, 0755);
-		SAFE_MOUNT(cleanup, device, MNTPOINT, fs_type, 0, NULL);
-		mount_flag = 1;
-
-		sprintf(testfile, MNTPOINT"/testfile");
-	}
+	sprintf(testfile, "%s/testfile", mntpoint);
 	create_testfile();
 }
 
-static void cleanup(void)
-{
-	unlink(testfile);
-	if (mount_flag && tst_umount(MNTPOINT) < 0)
-		tst_resm(TWARN | TERRNO, "umount device:%s failed", device);
-
-	if (device)
-		tst_release_device(device);
-
-	tst_rmdir();
-}
+static struct tst_test test = {
+	.needs_root = 1,
+	.needs_tmpdir = 1,
+	.mount_device = 1,
+	.mntpoint = mntpoint,
+	.setup = setup,
+	.options = options,
+	.test_all = test_readahead,
+};
 
 #else /* __NR_readahead */
-int main(void)
-{
-	tst_brkm(TCONF, NULL, "System doesn't support __NR_readahead");
-}
+	TST_TEST_TCONF("System doesn't support __NR_readahead");
 #endif
-- 
2.17.1

[PATCH 3/4] syscalls/readahead02: test readahead() on an overlayfs file

[Amir Goldstein]

Repeat the test case on an overlayfs file.

The new test case is a regression test for kernel commit b833a3660394
("ovl: add ovl_fadvise()") which fixes a regression of readahead() on
an overlay file that was introduced by kernel commit 5b910bd615ba
("ovl: fix GPF in swapfile_activate of file from overlayfs over xfs").

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
---
 .../kernel/syscalls/readahead/readahead02.c   | 55 +++++++++++++++++--
 1 file changed, 49 insertions(+), 6 deletions(-)

diff --git a/testcases/kernel/syscalls/readahead/readahead02.c b/testcases/kernel/syscalls/readahead/readahead02.c
index 4132b4de1..191116f62 100644
--- a/testcases/kernel/syscalls/readahead/readahead02.c
+++ b/testcases/kernel/syscalls/readahead/readahead02.c
@@ -14,6 +14,7 @@
 #include <sys/types.h>
 #include <sys/syscall.h>
 #include <sys/mman.h>
+#include <sys/mount.h>
 #include <sys/stat.h>
 #include <sys/types.h>
 #include <sys/time.h>
@@ -34,9 +35,15 @@ static const char meminfo_fname[] = "/proc/meminfo";
 static size_t testfile_size = 64 * 1024 * 1024;
 static char *opt_fsizestr;
 static int pagesize;
+static int ovl_mounted;
 
 #define MNTPOINT        "mntpoint"
-static const char mntpoint[] = "mntpoint";
+#define OVL_LOWER	MNTPOINT"/lower"
+#define OVL_UPPER	MNTPOINT"/upper"
+#define OVL_WORK	MNTPOINT"/work"
+#define OVL_MNT		MNTPOINT"/ovl"
+
+static const char mntpoint[] = MNTPOINT;
 #define MIN_SANE_READAHEAD (4u * 1024u)
 
 static struct tst_option options[] = {
@@ -109,12 +116,39 @@ static unsigned long get_cached_size(void)
 	return parse_entry(meminfo_fname, entry);
 }
 
-static void create_testfile(void)
+static int setup_overlay(void)
+{
+	int ret;
+
+	/* Setup an overlay mount with lower dir and file */
+	SAFE_MKDIR(OVL_LOWER, 0755);
+	SAFE_MKDIR(OVL_UPPER, 0755);
+	SAFE_MKDIR(OVL_WORK, 0755);
+	SAFE_MKDIR(OVL_MNT, 0755);
+	ret = mount("overlay", OVL_MNT, "overlay", 0, "lowerdir="OVL_LOWER
+		    ",upperdir="OVL_UPPER",workdir="OVL_WORK);
+	if (ret < 0) {
+		if (errno == ENODEV) {
+			tst_res(TCONF,
+				"overlayfs is not configured in this kernel.");
+			return ret;
+		}
+		tst_brk(TBROK | TERRNO, "overlayfs mount failed");
+	}
+	ovl_mounted = 1;
+	return 0;
+}
+
+static int create_testfile(unsigned int use_overlay)
 {
 	int fd;
 	char *tmp;
 	size_t i;
 
+	if (use_overlay && setup_overlay() != 0)
+		return -1;
+
+	sprintf(testfile, "%s/testfile", use_overlay ? OVL_MNT : MNTPOINT);
 	tst_res(TINFO, "creating test file of size: %zu", testfile_size);
 	tmp = SAFE_MALLOC(pagesize);
 
@@ -127,6 +161,7 @@ static void create_testfile(void)
 	SAFE_FSYNC(fd);
 	SAFE_CLOSE(fd);
 	free(tmp);
+	return 0;
 }
 
 
@@ -221,7 +256,7 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 	SAFE_CLOSE(fd);
 }
 
-static void test_readahead(void)
+static void test_readahead(unsigned int use_overlay)
 {
 	unsigned long read_bytes, read_bytes_ra;
 	long usec, usec_ra;
@@ -229,6 +264,9 @@ static void test_readahead(void)
 	char proc_io_fname[128];
 	sprintf(proc_io_fname, "/proc/%u/io", getpid());
 
+	if (create_testfile(use_overlay) != 0)
+		return;
+
 	/* find out how much can cache hold if we read whole file */
 	read_testfile(0, testfile, testfile_size, &read_bytes, &usec, &cached);
 	cached_max = get_cached_size();
@@ -302,9 +340,12 @@ static void setup(void)
 	tst_syscall(__NR_readahead, 0, 0, 0);
 
 	pagesize = getpagesize();
+}
 
-	sprintf(testfile, "%s/testfile", mntpoint);
-	create_testfile();
+static void cleanup(void)
+{
+	if (ovl_mounted)
+		SAFE_UMOUNT(OVL_MNT);
 }
 
 static struct tst_test test = {
@@ -313,8 +354,10 @@ static struct tst_test test = {
 	.mount_device = 1,
 	.mntpoint = mntpoint,
 	.setup = setup,
+	.cleanup = cleanup,
 	.options = options,
-	.test_all = test_readahead,
+	.test = test_readahead,
+	.tcnt = 2, /* Repeat with overlayfs */
 };
 
 #else /* __NR_readahead */
-- 
2.17.1

[LTP] [PATCH 3/4] syscalls/readahead02: test readahead() on an overlayfs file

[Amir Goldstein]

Repeat the test case on an overlayfs file.

The new test case is a regression test for kernel commit b833a3660394
("ovl: add ovl_fadvise()") which fixes a regression of readahead() on
an overlay file that was introduced by kernel commit 5b910bd615ba
("ovl: fix GPF in swapfile_activate of file from overlayfs over xfs").

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
---
 .../kernel/syscalls/readahead/readahead02.c   | 55 +++++++++++++++++--
 1 file changed, 49 insertions(+), 6 deletions(-)

diff --git a/testcases/kernel/syscalls/readahead/readahead02.c b/testcases/kernel/syscalls/readahead/readahead02.c
index 4132b4de1..191116f62 100644
--- a/testcases/kernel/syscalls/readahead/readahead02.c
+++ b/testcases/kernel/syscalls/readahead/readahead02.c
@@ -14,6 +14,7 @@
 #include <sys/types.h>
 #include <sys/syscall.h>
 #include <sys/mman.h>
+#include <sys/mount.h>
 #include <sys/stat.h>
 #include <sys/types.h>
 #include <sys/time.h>
@@ -34,9 +35,15 @@ static const char meminfo_fname[] = "/proc/meminfo";
 static size_t testfile_size = 64 * 1024 * 1024;
 static char *opt_fsizestr;
 static int pagesize;
+static int ovl_mounted;
 
 #define MNTPOINT        "mntpoint"
-static const char mntpoint[] = "mntpoint";
+#define OVL_LOWER	MNTPOINT"/lower"
+#define OVL_UPPER	MNTPOINT"/upper"
+#define OVL_WORK	MNTPOINT"/work"
+#define OVL_MNT		MNTPOINT"/ovl"
+
+static const char mntpoint[] = MNTPOINT;
 #define MIN_SANE_READAHEAD (4u * 1024u)
 
 static struct tst_option options[] = {
@@ -109,12 +116,39 @@ static unsigned long get_cached_size(void)
 	return parse_entry(meminfo_fname, entry);
 }
 
-static void create_testfile(void)
+static int setup_overlay(void)
+{
+	int ret;
+
+	/* Setup an overlay mount with lower dir and file */
+	SAFE_MKDIR(OVL_LOWER, 0755);
+	SAFE_MKDIR(OVL_UPPER, 0755);
+	SAFE_MKDIR(OVL_WORK, 0755);
+	SAFE_MKDIR(OVL_MNT, 0755);
+	ret = mount("overlay", OVL_MNT, "overlay", 0, "lowerdir="OVL_LOWER
+		    ",upperdir="OVL_UPPER",workdir="OVL_WORK);
+	if (ret < 0) {
+		if (errno == ENODEV) {
+			tst_res(TCONF,
+				"overlayfs is not configured in this kernel.");
+			return ret;
+		}
+		tst_brk(TBROK | TERRNO, "overlayfs mount failed");
+	}
+	ovl_mounted = 1;
+	return 0;
+}
+
+static int create_testfile(unsigned int use_overlay)
 {
 	int fd;
 	char *tmp;
 	size_t i;
 
+	if (use_overlay && setup_overlay() != 0)
+		return -1;
+
+	sprintf(testfile, "%s/testfile", use_overlay ? OVL_MNT : MNTPOINT);
 	tst_res(TINFO, "creating test file of size: %zu", testfile_size);
 	tmp = SAFE_MALLOC(pagesize);
 
@@ -127,6 +161,7 @@ static void create_testfile(void)
 	SAFE_FSYNC(fd);
 	SAFE_CLOSE(fd);
 	free(tmp);
+	return 0;
 }
 
 
@@ -221,7 +256,7 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 	SAFE_CLOSE(fd);
 }
 
-static void test_readahead(void)
+static void test_readahead(unsigned int use_overlay)
 {
 	unsigned long read_bytes, read_bytes_ra;
 	long usec, usec_ra;
@@ -229,6 +264,9 @@ static void test_readahead(void)
 	char proc_io_fname[128];
 	sprintf(proc_io_fname, "/proc/%u/io", getpid());
 
+	if (create_testfile(use_overlay) != 0)
+		return;
+
 	/* find out how much can cache hold if we read whole file */
 	read_testfile(0, testfile, testfile_size, &read_bytes, &usec, &cached);
 	cached_max = get_cached_size();
@@ -302,9 +340,12 @@ static void setup(void)
 	tst_syscall(__NR_readahead, 0, 0, 0);
 
 	pagesize = getpagesize();
+}
 
-	sprintf(testfile, "%s/testfile", mntpoint);
-	create_testfile();
+static void cleanup(void)
+{
+	if (ovl_mounted)
+		SAFE_UMOUNT(OVL_MNT);
 }
 
 static struct tst_test test = {
@@ -313,8 +354,10 @@ static struct tst_test test = {
 	.mount_device = 1,
 	.mntpoint = mntpoint,
 	.setup = setup,
+	.cleanup = cleanup,
 	.options = options,
-	.test_all = test_readahead,
+	.test = test_readahead,
+	.tcnt = 2, /* Repeat with overlayfs */
 };
 
 #else /* __NR_readahead */
-- 
2.17.1

[PATCH 4/4] syscalls/readahead02: test readahead using posix_fadvise()

[Amir Goldstein]

The call to posix_fadvise(POSIX_FADV_WILLNEED) should have the same
effect as the call to readahead() syscall.

Repeat the test cases for local file and overlayfs file with
posix_fadvise().

The new test case is a regression test for kernel commit b833a3660394
("ovl: add ovl_fadvise()") which fixes a regression of fadvise() on
an overlay file that was introduced by kernel commit 5b910bd615ba
("ovl: fix GPF in swapfile_activate of file from overlayfs over xfs").

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
---
 .../kernel/syscalls/readahead/readahead02.c   | 57 +++++++++++++++++--
 1 file changed, 51 insertions(+), 6 deletions(-)

diff --git a/testcases/kernel/syscalls/readahead/readahead02.c b/testcases/kernel/syscalls/readahead/readahead02.c
index 191116f62..9ebed359d 100644
--- a/testcases/kernel/syscalls/readahead/readahead02.c
+++ b/testcases/kernel/syscalls/readahead/readahead02.c
@@ -51,6 +51,42 @@ static struct tst_option options[] = {
 	{NULL, NULL, NULL}
 };
 
+#ifndef _FILE_OFFSET_BITS
+#define _FILE_OFFSET_BITS 32
+#endif
+
+#ifndef __NR_fadvise64
+#define __NR_fadvise64 0
+#endif
+
+static struct tcase {
+	const char *tname;
+	int use_overlay;
+	int use_fadvise;
+} tcases[] = {
+	{ "readahead on file", 0, 0 },
+	{ "readahead on overlayfs file", 1, 0 },
+/* Check this system has fadvise64 system which is used in posix_fadvise. */
+#if ((_FILE_OFFSET_BITS == 64) || (__NR_fadvise64 != 0))
+	{ "POSIX_FADV_WILLNEED on file", 0, 1 },
+	{ "POSIX_FADV_WILLNEED on overlayfs file", 1, 1 },
+#endif
+};
+
+static int fadvise_willneed(int fd, off_t offset, size_t len)
+{
+	/* Should have the same effect as readahead() syscall */
+	return posix_fadvise(fd, offset, len, POSIX_FADV_WILLNEED);
+}
+
+static int libc_readahead(int fd, off_t offset, size_t len)
+{
+	return readahead(fd, offset, len);
+}
+
+typedef int (*readahead_func_t)(int, off_t, size_t);
+static readahead_func_t readahead_func = libc_readahead;
+
 static int check_ret(long expected_ret)
 {
 	if (expected_ret == TST_RET) {
@@ -120,6 +156,9 @@ static int setup_overlay(void)
 {
 	int ret;
 
+	if (ovl_mounted)
+		return 0;
+
 	/* Setup an overlay mount with lower dir and file */
 	SAFE_MKDIR(OVL_LOWER, 0755);
 	SAFE_MKDIR(OVL_UPPER, 0755);
@@ -139,7 +178,7 @@ static int setup_overlay(void)
 	return 0;
 }
 
-static int create_testfile(unsigned int use_overlay)
+static int create_testfile(int use_overlay)
 {
 	int fd;
 	char *tmp;
@@ -164,7 +203,6 @@ static int create_testfile(unsigned int use_overlay)
 	return 0;
 }
 
-
 /* read_testfile - mmap testfile and read every page.
  * This functions measures how many I/O and time it takes to fully
  * read contents of test file.
@@ -194,7 +232,7 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 	if (do_readahead) {
 		cached_start = get_cached_size();
 		do {
-			TEST(readahead(fd, offset, fsize - offset));
+			TEST(readahead_func(fd, offset, fsize - offset));
 			if (TST_RET != 0) {
 				check_ret(0);
 				break;
@@ -256,17 +294,24 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 	SAFE_CLOSE(fd);
 }
 
-static void test_readahead(unsigned int use_overlay)
+static void test_readahead(unsigned int n)
 {
 	unsigned long read_bytes, read_bytes_ra;
 	long usec, usec_ra;
 	unsigned long cached_max, cached_low, cached, cached_ra;
 	char proc_io_fname[128];
+	struct tcase *tc = &tcases[n];
+
+	tst_res(TINFO, "Test #%d: %s", n, tc->tname);
+
 	sprintf(proc_io_fname, "/proc/%u/io", getpid());
 
-	if (create_testfile(use_overlay) != 0)
+	if (create_testfile(tc->use_overlay) != 0)
 		return;
 
+	/* Use either readahead() syscall or POSIX_FADV_WILLNEED */
+	readahead_func = tc->use_fadvise ? fadvise_willneed : libc_readahead;
+
 	/* find out how much can cache hold if we read whole file */
 	read_testfile(0, testfile, testfile_size, &read_bytes, &usec, &cached);
 	cached_max = get_cached_size();
@@ -357,7 +402,7 @@ static struct tst_test test = {
 	.cleanup = cleanup,
 	.options = options,
 	.test = test_readahead,
-	.tcnt = 2, /* Repeat with overlayfs */
+	.tcnt = ARRAY_SIZE(tcases),
 };
 
 #else /* __NR_readahead */
-- 
2.17.1

[LTP] [PATCH 4/4] syscalls/readahead02: test readahead using posix_fadvise()

[Amir Goldstein]

The call to posix_fadvise(POSIX_FADV_WILLNEED) should have the same
effect as the call to readahead() syscall.

Repeat the test cases for local file and overlayfs file with
posix_fadvise().

The new test case is a regression test for kernel commit b833a3660394
("ovl: add ovl_fadvise()") which fixes a regression of fadvise() on
an overlay file that was introduced by kernel commit 5b910bd615ba
("ovl: fix GPF in swapfile_activate of file from overlayfs over xfs").

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
---
 .../kernel/syscalls/readahead/readahead02.c   | 57 +++++++++++++++++--
 1 file changed, 51 insertions(+), 6 deletions(-)

diff --git a/testcases/kernel/syscalls/readahead/readahead02.c b/testcases/kernel/syscalls/readahead/readahead02.c
index 191116f62..9ebed359d 100644
--- a/testcases/kernel/syscalls/readahead/readahead02.c
+++ b/testcases/kernel/syscalls/readahead/readahead02.c
@@ -51,6 +51,42 @@ static struct tst_option options[] = {
 	{NULL, NULL, NULL}
 };
 
+#ifndef _FILE_OFFSET_BITS
+#define _FILE_OFFSET_BITS 32
+#endif
+
+#ifndef __NR_fadvise64
+#define __NR_fadvise64 0
+#endif
+
+static struct tcase {
+	const char *tname;
+	int use_overlay;
+	int use_fadvise;
+} tcases[] = {
+	{ "readahead on file", 0, 0 },
+	{ "readahead on overlayfs file", 1, 0 },
+/* Check this system has fadvise64 system which is used in posix_fadvise. */
+#if ((_FILE_OFFSET_BITS == 64) || (__NR_fadvise64 != 0))
+	{ "POSIX_FADV_WILLNEED on file", 0, 1 },
+	{ "POSIX_FADV_WILLNEED on overlayfs file", 1, 1 },
+#endif
+};
+
+static int fadvise_willneed(int fd, off_t offset, size_t len)
+{
+	/* Should have the same effect as readahead() syscall */
+	return posix_fadvise(fd, offset, len, POSIX_FADV_WILLNEED);
+}
+
+static int libc_readahead(int fd, off_t offset, size_t len)
+{
+	return readahead(fd, offset, len);
+}
+
+typedef int (*readahead_func_t)(int, off_t, size_t);
+static readahead_func_t readahead_func = libc_readahead;
+
 static int check_ret(long expected_ret)
 {
 	if (expected_ret == TST_RET) {
@@ -120,6 +156,9 @@ static int setup_overlay(void)
 {
 	int ret;
 
+	if (ovl_mounted)
+		return 0;
+
 	/* Setup an overlay mount with lower dir and file */
 	SAFE_MKDIR(OVL_LOWER, 0755);
 	SAFE_MKDIR(OVL_UPPER, 0755);
@@ -139,7 +178,7 @@ static int setup_overlay(void)
 	return 0;
 }
 
-static int create_testfile(unsigned int use_overlay)
+static int create_testfile(int use_overlay)
 {
 	int fd;
 	char *tmp;
@@ -164,7 +203,6 @@ static int create_testfile(unsigned int use_overlay)
 	return 0;
 }
 
-
 /* read_testfile - mmap testfile and read every page.
  * This functions measures how many I/O and time it takes to fully
  * read contents of test file.
@@ -194,7 +232,7 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 	if (do_readahead) {
 		cached_start = get_cached_size();
 		do {
-			TEST(readahead(fd, offset, fsize - offset));
+			TEST(readahead_func(fd, offset, fsize - offset));
 			if (TST_RET != 0) {
 				check_ret(0);
 				break;
@@ -256,17 +294,24 @@ static void read_testfile(int do_readahead, const char *fname, size_t fsize,
 	SAFE_CLOSE(fd);
 }
 
-static void test_readahead(unsigned int use_overlay)
+static void test_readahead(unsigned int n)
 {
 	unsigned long read_bytes, read_bytes_ra;
 	long usec, usec_ra;
 	unsigned long cached_max, cached_low, cached, cached_ra;
 	char proc_io_fname[128];
+	struct tcase *tc = &tcases[n];
+
+	tst_res(TINFO, "Test #%d: %s", n, tc->tname);
+
 	sprintf(proc_io_fname, "/proc/%u/io", getpid());
 
-	if (create_testfile(use_overlay) != 0)
+	if (create_testfile(tc->use_overlay) != 0)
 		return;
 
+	/* Use either readahead() syscall or POSIX_FADV_WILLNEED */
+	readahead_func = tc->use_fadvise ? fadvise_willneed : libc_readahead;
+
 	/* find out how much can cache hold if we read whole file */
 	read_testfile(0, testfile, testfile_size, &read_bytes, &usec, &cached);
 	cached_max = get_cached_size();
@@ -357,7 +402,7 @@ static struct tst_test test = {
 	.cleanup = cleanup,
 	.options = options,
 	.test = test_readahead,
-	.tcnt = 2, /* Repeat with overlayfs */
+	.tcnt = ARRAY_SIZE(tcases),
 };
 
 #else /* __NR_readahead */
-- 
2.17.1

Re: [LTP] [PATCH 2/4] syscalls/readahead02: Convert to newlib and cleanup

[Jan Stancek]

----- Original Message -----
> * Use SAFE macros
> 
> * Use SPDX-License-Identifier
> 
> * No need to cleanup test file from temp dir
> 
> Signed-off-by: Amir Goldstein <amir73il@gmail.com>

Hi,

ack to 1/4

>  
>  static int has_file(const char *fname, int required)
>  {
> -	int ret;
>  	struct stat buf;
> -	ret = stat(fname, &buf);
> -	if (ret == -1) {
> -		if (errno == ENOENT)
> -			if (required)
> -				tst_brkm(TCONF, cleanup, "%s not available",
> -					 fname);
> -			else
> -				return 0;
> -		else
> -			tst_brkm(TBROK | TERRNO, cleanup, "stat %s", fname);
> +
> +	if (stat(fname, &buf) == -1) {
> +		if (errno != ENOENT)
> +			tst_brk(TBROK | TERRNO, "stat %s", fname);
> +		if (required)
> +			tst_brk(TCONF, "%s not available", fname);
>  	}
>  	return 1;
>  }

This will return 1 even when file doesn't exist.
(Not that it makes big difference for test)

<snip>

> +static struct tst_test test = {
> +	.needs_root = 1,
> +	.needs_tmpdir = 1,
> +	.mount_device = 1,
> +	.mntpoint = mntpoint,
> +	.setup = setup,
> +	.options = options,
> +	.test_all = test_readahead,
> +};

Would it make sense to enable this for 'all_filesystems = 1'?
Previously we used whatever fs /tmp was, now we seem to default
always to ext2.

Regards,
Jan

>  
>  #else /* __NR_readahead */
> -int main(void)
> -{
> -	tst_brkm(TCONF, NULL, "System doesn't support __NR_readahead");
> -}
> +	TST_TEST_TCONF("System doesn't support __NR_readahead");
>  #endif
> --
> 2.17.1
> 
> 
> --
> Mailing list info: https://lists.linux.it/listinfo/ltp
> 

[LTP] [PATCH 2/4] syscalls/readahead02: Convert to newlib and cleanup

[Jan Stancek]

----- Original Message -----
> * Use SAFE macros
> 
> * Use SPDX-License-Identifier
> 
> * No need to cleanup test file from temp dir
> 
> Signed-off-by: Amir Goldstein <amir73il@gmail.com>

Hi,

ack to 1/4

>  
>  static int has_file(const char *fname, int required)
>  {
> -	int ret;
>  	struct stat buf;
> -	ret = stat(fname, &buf);
> -	if (ret == -1) {
> -		if (errno == ENOENT)
> -			if (required)
> -				tst_brkm(TCONF, cleanup, "%s not available",
> -					 fname);
> -			else
> -				return 0;
> -		else
> -			tst_brkm(TBROK | TERRNO, cleanup, "stat %s", fname);
> +
> +	if (stat(fname, &buf) == -1) {
> +		if (errno != ENOENT)
> +			tst_brk(TBROK | TERRNO, "stat %s", fname);
> +		if (required)
> +			tst_brk(TCONF, "%s not available", fname);
>  	}
>  	return 1;
>  }

This will return 1 even when file doesn't exist.
(Not that it makes big difference for test)

<snip>

> +static struct tst_test test = {
> +	.needs_root = 1,
> +	.needs_tmpdir = 1,
> +	.mount_device = 1,
> +	.mntpoint = mntpoint,
> +	.setup = setup,
> +	.options = options,
> +	.test_all = test_readahead,
> +};

Would it make sense to enable this for 'all_filesystems = 1'?
Previously we used whatever fs /tmp was, now we seem to default
always to ext2.

Regards,
Jan

>  
>  #else /* __NR_readahead */
> -int main(void)
> -{
> -	tst_brkm(TCONF, NULL, "System doesn't support __NR_readahead");
> -}
> +	TST_TEST_TCONF("System doesn't support __NR_readahead");
>  #endif
> --
> 2.17.1
> 
> 
> --
> Mailing list info: https://lists.linux.it/listinfo/ltp
> 

Re: [LTP] [PATCH 3/4] syscalls/readahead02: test readahead() on an overlayfs file

[Jan Stancek]

----- Original Message -----
> Repeat the test case on an overlayfs file.
> 
> The new test case is a regression test for kernel commit b833a3660394
> ("ovl: add ovl_fadvise()") which fixes a regression of readahead() on
> an overlay file that was introduced by kernel commit 5b910bd615ba
> ("ovl: fix GPF in swapfile_activate of file from overlayfs over xfs").
> 
> Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> ---
>  .../kernel/syscalls/readahead/readahead02.c   | 55 +++++++++++++++++--
>  1 file changed, 49 insertions(+), 6 deletions(-)
> 
> diff --git a/testcases/kernel/syscalls/readahead/readahead02.c
> b/testcases/kernel/syscalls/readahead/readahead02.c
> index 4132b4de1..191116f62 100644
> --- a/testcases/kernel/syscalls/readahead/readahead02.c
> +++ b/testcases/kernel/syscalls/readahead/readahead02.c
> @@ -14,6 +14,7 @@
>  #include <sys/types.h>
>  #include <sys/syscall.h>
>  #include <sys/mman.h>
> +#include <sys/mount.h>
>  #include <sys/stat.h>
>  #include <sys/types.h>
>  #include <sys/time.h>
> @@ -34,9 +35,15 @@ static const char meminfo_fname[] = "/proc/meminfo";
>  static size_t testfile_size = 64 * 1024 * 1024;
>  static char *opt_fsizestr;
>  static int pagesize;
> +static int ovl_mounted;
>  
>  #define MNTPOINT        "mntpoint"
> -static const char mntpoint[] = "mntpoint";
> +#define OVL_LOWER	MNTPOINT"/lower"
> +#define OVL_UPPER	MNTPOINT"/upper"
> +#define OVL_WORK	MNTPOINT"/work"
> +#define OVL_MNT		MNTPOINT"/ovl"
> +
> +static const char mntpoint[] = MNTPOINT;
>  #define MIN_SANE_READAHEAD (4u * 1024u)
>  
>  static struct tst_option options[] = {
> @@ -109,12 +116,39 @@ static unsigned long get_cached_size(void)
>  	return parse_entry(meminfo_fname, entry);
>  }
>  
> -static void create_testfile(void)
> +static int setup_overlay(void)
> +{
> +	int ret;
> +
> +	/* Setup an overlay mount with lower dir and file */
> +	SAFE_MKDIR(OVL_LOWER, 0755);
> +	SAFE_MKDIR(OVL_UPPER, 0755);
> +	SAFE_MKDIR(OVL_WORK, 0755);
> +	SAFE_MKDIR(OVL_MNT, 0755);
> +	ret = mount("overlay", OVL_MNT, "overlay", 0, "lowerdir="OVL_LOWER
> +		    ",upperdir="OVL_UPPER",workdir="OVL_WORK);
> +	if (ret < 0) {
> +		if (errno == ENODEV) {
> +			tst_res(TCONF,
> +				"overlayfs is not configured in this kernel.");
> +			return ret;
> +		}
> +		tst_brk(TBROK | TERRNO, "overlayfs mount failed");
> +	}
> +	ovl_mounted = 1;
> +	return 0;
> +}
> +
> +static int create_testfile(unsigned int use_overlay)
>  {
>  	int fd;
>  	char *tmp;
>  	size_t i;
>  
> +	if (use_overlay && setup_overlay() != 0)
> +		return -1;
> +
> +	sprintf(testfile, "%s/testfile", use_overlay ? OVL_MNT : MNTPOINT);
>  	tst_res(TINFO, "creating test file of size: %zu", testfile_size);
>  	tmp = SAFE_MALLOC(pagesize);
>  
> @@ -127,6 +161,7 @@ static void create_testfile(void)
>  	SAFE_FSYNC(fd);
>  	SAFE_CLOSE(fd);
>  	free(tmp);
> +	return 0;
>  }
>  
>  
> @@ -221,7 +256,7 @@ static void read_testfile(int do_readahead, const char
> *fname, size_t fsize,
>  	SAFE_CLOSE(fd);
>  }
>  
> -static void test_readahead(void)
> +static void test_readahead(unsigned int use_overlay)
>  {
>  	unsigned long read_bytes, read_bytes_ra;
>  	long usec, usec_ra;
> @@ -229,6 +264,9 @@ static void test_readahead(void)
>  	char proc_io_fname[128];
>  	sprintf(proc_io_fname, "/proc/%u/io", getpid());
>  
> +	if (create_testfile(use_overlay) != 0)
> +		return;
> +

This could use some TINFO message, what is being tested.
.. and I see you added it in 4/4.

>  	/* find out how much can cache hold if we read whole file */
>  	read_testfile(0, testfile, testfile_size, &read_bytes, &usec, &cached);
>  	cached_max = get_cached_size();
> @@ -302,9 +340,12 @@ static void setup(void)
>  	tst_syscall(__NR_readahead, 0, 0, 0);
>  
>  	pagesize = getpagesize();
> +}
>  
> -	sprintf(testfile, "%s/testfile", mntpoint);
> -	create_testfile();
> +static void cleanup(void)
> +{
> +	if (ovl_mounted)
> +		SAFE_UMOUNT(OVL_MNT);
>  }

This creates a small conflict, because setup and cleanup don't match.
If you run this with multiple iterations (parameter -i), it's going
to fail:
  safe_macros.c:169: BROK: readahead02.c:124: mkdir(mntpoint/lower,0755) failed: EEXIST

>  
>  static struct tst_test test = {
> @@ -313,8 +354,10 @@ static struct tst_test test = {
>  	.mount_device = 1,
>  	.mntpoint = mntpoint,
>  	.setup = setup,
> +	.cleanup = cleanup,
>  	.options = options,
> -	.test_all = test_readahead,
> +	.test = test_readahead,
> +	.tcnt = 2, /* Repeat with overlayfs */

I'd rather add a function to call old and new test,
single int will make it harder to add additional tests in future.
.. and I see you changed this in 4/4 already.

Regards,
Jan

>  };
>  
>  #else /* __NR_readahead */
> --
> 2.17.1
> 
> 
> --
> Mailing list info: https://lists.linux.it/listinfo/ltp
> 

[LTP] [PATCH 3/4] syscalls/readahead02: test readahead() on an overlayfs file

[Jan Stancek]

----- Original Message -----
> Repeat the test case on an overlayfs file.
> 
> The new test case is a regression test for kernel commit b833a3660394
> ("ovl: add ovl_fadvise()") which fixes a regression of readahead() on
> an overlay file that was introduced by kernel commit 5b910bd615ba
> ("ovl: fix GPF in swapfile_activate of file from overlayfs over xfs").
> 
> Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> ---
>  .../kernel/syscalls/readahead/readahead02.c   | 55 +++++++++++++++++--
>  1 file changed, 49 insertions(+), 6 deletions(-)
> 
> diff --git a/testcases/kernel/syscalls/readahead/readahead02.c
> b/testcases/kernel/syscalls/readahead/readahead02.c
> index 4132b4de1..191116f62 100644
> --- a/testcases/kernel/syscalls/readahead/readahead02.c
> +++ b/testcases/kernel/syscalls/readahead/readahead02.c
> @@ -14,6 +14,7 @@
>  #include <sys/types.h>
>  #include <sys/syscall.h>
>  #include <sys/mman.h>
> +#include <sys/mount.h>
>  #include <sys/stat.h>
>  #include <sys/types.h>
>  #include <sys/time.h>
> @@ -34,9 +35,15 @@ static const char meminfo_fname[] = "/proc/meminfo";
>  static size_t testfile_size = 64 * 1024 * 1024;
>  static char *opt_fsizestr;
>  static int pagesize;
> +static int ovl_mounted;
>  
>  #define MNTPOINT        "mntpoint"
> -static const char mntpoint[] = "mntpoint";
> +#define OVL_LOWER	MNTPOINT"/lower"
> +#define OVL_UPPER	MNTPOINT"/upper"
> +#define OVL_WORK	MNTPOINT"/work"
> +#define OVL_MNT		MNTPOINT"/ovl"
> +
> +static const char mntpoint[] = MNTPOINT;
>  #define MIN_SANE_READAHEAD (4u * 1024u)
>  
>  static struct tst_option options[] = {
> @@ -109,12 +116,39 @@ static unsigned long get_cached_size(void)
>  	return parse_entry(meminfo_fname, entry);
>  }
>  
> -static void create_testfile(void)
> +static int setup_overlay(void)
> +{
> +	int ret;
> +
> +	/* Setup an overlay mount with lower dir and file */
> +	SAFE_MKDIR(OVL_LOWER, 0755);
> +	SAFE_MKDIR(OVL_UPPER, 0755);
> +	SAFE_MKDIR(OVL_WORK, 0755);
> +	SAFE_MKDIR(OVL_MNT, 0755);
> +	ret = mount("overlay", OVL_MNT, "overlay", 0, "lowerdir="OVL_LOWER
> +		    ",upperdir="OVL_UPPER",workdir="OVL_WORK);
> +	if (ret < 0) {
> +		if (errno == ENODEV) {
> +			tst_res(TCONF,
> +				"overlayfs is not configured in this kernel.");
> +			return ret;
> +		}
> +		tst_brk(TBROK | TERRNO, "overlayfs mount failed");
> +	}
> +	ovl_mounted = 1;
> +	return 0;
> +}
> +
> +static int create_testfile(unsigned int use_overlay)
>  {
>  	int fd;
>  	char *tmp;
>  	size_t i;
>  
> +	if (use_overlay && setup_overlay() != 0)
> +		return -1;
> +
> +	sprintf(testfile, "%s/testfile", use_overlay ? OVL_MNT : MNTPOINT);
>  	tst_res(TINFO, "creating test file of size: %zu", testfile_size);
>  	tmp = SAFE_MALLOC(pagesize);
>  
> @@ -127,6 +161,7 @@ static void create_testfile(void)
>  	SAFE_FSYNC(fd);
>  	SAFE_CLOSE(fd);
>  	free(tmp);
> +	return 0;
>  }
>  
>  
> @@ -221,7 +256,7 @@ static void read_testfile(int do_readahead, const char
> *fname, size_t fsize,
>  	SAFE_CLOSE(fd);
>  }
>  
> -static void test_readahead(void)
> +static void test_readahead(unsigned int use_overlay)
>  {
>  	unsigned long read_bytes, read_bytes_ra;
>  	long usec, usec_ra;
> @@ -229,6 +264,9 @@ static void test_readahead(void)
>  	char proc_io_fname[128];
>  	sprintf(proc_io_fname, "/proc/%u/io", getpid());
>  
> +	if (create_testfile(use_overlay) != 0)
> +		return;
> +

This could use some TINFO message, what is being tested.
.. and I see you added it in 4/4.

>  	/* find out how much can cache hold if we read whole file */
>  	read_testfile(0, testfile, testfile_size, &read_bytes, &usec, &cached);
>  	cached_max = get_cached_size();
> @@ -302,9 +340,12 @@ static void setup(void)
>  	tst_syscall(__NR_readahead, 0, 0, 0);
>  
>  	pagesize = getpagesize();
> +}
>  
> -	sprintf(testfile, "%s/testfile", mntpoint);
> -	create_testfile();
> +static void cleanup(void)
> +{
> +	if (ovl_mounted)
> +		SAFE_UMOUNT(OVL_MNT);
>  }

This creates a small conflict, because setup and cleanup don't match.
If you run this with multiple iterations (parameter -i), it's going
to fail:
  safe_macros.c:169: BROK: readahead02.c:124: mkdir(mntpoint/lower,0755) failed: EEXIST

>  
>  static struct tst_test test = {
> @@ -313,8 +354,10 @@ static struct tst_test test = {
>  	.mount_device = 1,
>  	.mntpoint = mntpoint,
>  	.setup = setup,
> +	.cleanup = cleanup,
>  	.options = options,
> -	.test_all = test_readahead,
> +	.test = test_readahead,
> +	.tcnt = 2, /* Repeat with overlayfs */

I'd rather add a function to call old and new test,
single int will make it harder to add additional tests in future.
.. and I see you changed this in 4/4 already.

Regards,
Jan

>  };
>  
>  #else /* __NR_readahead */
> --
> 2.17.1
> 
> 
> --
> Mailing list info: https://lists.linux.it/listinfo/ltp
> 

Re: [LTP] [PATCH 4/4] syscalls/readahead02: test readahead using posix_fadvise()

[Jan Stancek]

----- Original Message -----
> The call to posix_fadvise(POSIX_FADV_WILLNEED) should have the same
> effect as the call to readahead() syscall.
> 
> Repeat the test cases for local file and overlayfs file with
> posix_fadvise().
> 
> The new test case is a regression test for kernel commit b833a3660394
> ("ovl: add ovl_fadvise()") which fixes a regression of fadvise() on
> an overlay file that was introduced by kernel commit 5b910bd615ba
> ("ovl: fix GPF in swapfile_activate of file from overlayfs over xfs").
> 
> Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> ---
>  .../kernel/syscalls/readahead/readahead02.c   | 57 +++++++++++++++++--
>  1 file changed, 51 insertions(+), 6 deletions(-)
> 
> diff --git a/testcases/kernel/syscalls/readahead/readahead02.c
> b/testcases/kernel/syscalls/readahead/readahead02.c
> index 191116f62..9ebed359d 100644
> --- a/testcases/kernel/syscalls/readahead/readahead02.c
> +++ b/testcases/kernel/syscalls/readahead/readahead02.c
> @@ -51,6 +51,42 @@ static struct tst_option options[] = {
>  	{NULL, NULL, NULL}
>  };
>  
> +#ifndef _FILE_OFFSET_BITS
> +#define _FILE_OFFSET_BITS 32
> +#endif

I don't think we should be touching this.

> +
> +#ifndef __NR_fadvise64
> +#define __NR_fadvise64 0
> +#endif
> +
> +static struct tcase {
> +	const char *tname;
> +	int use_overlay;
> +	int use_fadvise;
> +} tcases[] = {
> +	{ "readahead on file", 0, 0 },
> +	{ "readahead on overlayfs file", 1, 0 },
> +/* Check this system has fadvise64 system which is used in posix_fadvise. */
> +#if ((_FILE_OFFSET_BITS == 64) || (__NR_fadvise64 != 0))

Can you elaborate on this check? Why do we need to care
about _FILE_OFFSET_BITS?

> +	{ "POSIX_FADV_WILLNEED on file", 0, 1 },
> +	{ "POSIX_FADV_WILLNEED on overlayfs file", 1, 1 },
> +#endif
> +};
> +
> +static int fadvise_willneed(int fd, off_t offset, size_t len)
> +{
> +	/* Should have the same effect as readahead() syscall */
> +	return posix_fadvise(fd, offset, len, POSIX_FADV_WILLNEED);
> +}
> +
> +static int libc_readahead(int fd, off_t offset, size_t len)
> +{
> +	return readahead(fd, offset, len);
> +}
> +
> +typedef int (*readahead_func_t)(int, off_t, size_t);
> +static readahead_func_t readahead_func = libc_readahead;
> +
>  static int check_ret(long expected_ret)
>  {
>  	if (expected_ret == TST_RET) {
> @@ -120,6 +156,9 @@ static int setup_overlay(void)
>  {
>  	int ret;
>  
> +	if (ovl_mounted)
> +		return 0;
> +

You could call it once from setup() and drop check above
and other call from create_testfile().

Regards,
Jan

>  	/* Setup an overlay mount with lower dir and file */
>  	SAFE_MKDIR(OVL_LOWER, 0755);
>  	SAFE_MKDIR(OVL_UPPER, 0755);
> @@ -139,7 +178,7 @@ static int setup_overlay(void)
>  	return 0;
>  }
>  
> -static int create_testfile(unsigned int use_overlay)
> +static int create_testfile(int use_overlay)
>  {
>  	int fd;
>  	char *tmp;
> @@ -164,7 +203,6 @@ static int create_testfile(unsigned int use_overlay)
>  	return 0;
>  }
>  
> -
>  /* read_testfile - mmap testfile and read every page.
>   * This functions measures how many I/O and time it takes to fully
>   * read contents of test file.
> @@ -194,7 +232,7 @@ static void read_testfile(int do_readahead, const char
> *fname, size_t fsize,
>  	if (do_readahead) {
>  		cached_start = get_cached_size();
>  		do {
> -			TEST(readahead(fd, offset, fsize - offset));
> +			TEST(readahead_func(fd, offset, fsize - offset));
>  			if (TST_RET != 0) {
>  				check_ret(0);
>  				break;
> @@ -256,17 +294,24 @@ static void read_testfile(int do_readahead, const char
> *fname, size_t fsize,
>  	SAFE_CLOSE(fd);
>  }
>  
> -static void test_readahead(unsigned int use_overlay)
> +static void test_readahead(unsigned int n)
>  {
>  	unsigned long read_bytes, read_bytes_ra;
>  	long usec, usec_ra;
>  	unsigned long cached_max, cached_low, cached, cached_ra;
>  	char proc_io_fname[128];
> +	struct tcase *tc = &tcases[n];
> +
> +	tst_res(TINFO, "Test #%d: %s", n, tc->tname);
> +
>  	sprintf(proc_io_fname, "/proc/%u/io", getpid());
>  
> -	if (create_testfile(use_overlay) != 0)
> +	if (create_testfile(tc->use_overlay) != 0)
>  		return;
>  
> +	/* Use either readahead() syscall or POSIX_FADV_WILLNEED */
> +	readahead_func = tc->use_fadvise ? fadvise_willneed : libc_readahead;
> +
>  	/* find out how much can cache hold if we read whole file */
>  	read_testfile(0, testfile, testfile_size, &read_bytes, &usec, &cached);
>  	cached_max = get_cached_size();
> @@ -357,7 +402,7 @@ static struct tst_test test = {
>  	.cleanup = cleanup,
>  	.options = options,
>  	.test = test_readahead,
> -	.tcnt = 2, /* Repeat with overlayfs */
> +	.tcnt = ARRAY_SIZE(tcases),
>  };
>  
>  #else /* __NR_readahead */
> --
> 2.17.1
> 
> 
> --
> Mailing list info: https://lists.linux.it/listinfo/ltp
> 

[LTP] [PATCH 4/4] syscalls/readahead02: test readahead using posix_fadvise()

[Jan Stancek]

----- Original Message -----
> The call to posix_fadvise(POSIX_FADV_WILLNEED) should have the same
> effect as the call to readahead() syscall.
> 
> Repeat the test cases for local file and overlayfs file with
> posix_fadvise().
> 
> The new test case is a regression test for kernel commit b833a3660394
> ("ovl: add ovl_fadvise()") which fixes a regression of fadvise() on
> an overlay file that was introduced by kernel commit 5b910bd615ba
> ("ovl: fix GPF in swapfile_activate of file from overlayfs over xfs").
> 
> Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> ---
>  .../kernel/syscalls/readahead/readahead02.c   | 57 +++++++++++++++++--
>  1 file changed, 51 insertions(+), 6 deletions(-)
> 
> diff --git a/testcases/kernel/syscalls/readahead/readahead02.c
> b/testcases/kernel/syscalls/readahead/readahead02.c
> index 191116f62..9ebed359d 100644
> --- a/testcases/kernel/syscalls/readahead/readahead02.c
> +++ b/testcases/kernel/syscalls/readahead/readahead02.c
> @@ -51,6 +51,42 @@ static struct tst_option options[] = {
>  	{NULL, NULL, NULL}
>  };
>  
> +#ifndef _FILE_OFFSET_BITS
> +#define _FILE_OFFSET_BITS 32
> +#endif

I don't think we should be touching this.

> +
> +#ifndef __NR_fadvise64
> +#define __NR_fadvise64 0
> +#endif
> +
> +static struct tcase {
> +	const char *tname;
> +	int use_overlay;
> +	int use_fadvise;
> +} tcases[] = {
> +	{ "readahead on file", 0, 0 },
> +	{ "readahead on overlayfs file", 1, 0 },
> +/* Check this system has fadvise64 system which is used in posix_fadvise. */
> +#if ((_FILE_OFFSET_BITS == 64) || (__NR_fadvise64 != 0))

Can you elaborate on this check? Why do we need to care
about _FILE_OFFSET_BITS?

> +	{ "POSIX_FADV_WILLNEED on file", 0, 1 },
> +	{ "POSIX_FADV_WILLNEED on overlayfs file", 1, 1 },
> +#endif
> +};
> +
> +static int fadvise_willneed(int fd, off_t offset, size_t len)
> +{
> +	/* Should have the same effect as readahead() syscall */
> +	return posix_fadvise(fd, offset, len, POSIX_FADV_WILLNEED);
> +}
> +
> +static int libc_readahead(int fd, off_t offset, size_t len)
> +{
> +	return readahead(fd, offset, len);
> +}
> +
> +typedef int (*readahead_func_t)(int, off_t, size_t);
> +static readahead_func_t readahead_func = libc_readahead;
> +
>  static int check_ret(long expected_ret)
>  {
>  	if (expected_ret == TST_RET) {
> @@ -120,6 +156,9 @@ static int setup_overlay(void)
>  {
>  	int ret;
>  
> +	if (ovl_mounted)
> +		return 0;
> +

You could call it once from setup() and drop check above
and other call from create_testfile().

Regards,
Jan

>  	/* Setup an overlay mount with lower dir and file */
>  	SAFE_MKDIR(OVL_LOWER, 0755);
>  	SAFE_MKDIR(OVL_UPPER, 0755);
> @@ -139,7 +178,7 @@ static int setup_overlay(void)
>  	return 0;
>  }
>  
> -static int create_testfile(unsigned int use_overlay)
> +static int create_testfile(int use_overlay)
>  {
>  	int fd;
>  	char *tmp;
> @@ -164,7 +203,6 @@ static int create_testfile(unsigned int use_overlay)
>  	return 0;
>  }
>  
> -
>  /* read_testfile - mmap testfile and read every page.
>   * This functions measures how many I/O and time it takes to fully
>   * read contents of test file.
> @@ -194,7 +232,7 @@ static void read_testfile(int do_readahead, const char
> *fname, size_t fsize,
>  	if (do_readahead) {
>  		cached_start = get_cached_size();
>  		do {
> -			TEST(readahead(fd, offset, fsize - offset));
> +			TEST(readahead_func(fd, offset, fsize - offset));
>  			if (TST_RET != 0) {
>  				check_ret(0);
>  				break;
> @@ -256,17 +294,24 @@ static void read_testfile(int do_readahead, const char
> *fname, size_t fsize,
>  	SAFE_CLOSE(fd);
>  }
>  
> -static void test_readahead(unsigned int use_overlay)
> +static void test_readahead(unsigned int n)
>  {
>  	unsigned long read_bytes, read_bytes_ra;
>  	long usec, usec_ra;
>  	unsigned long cached_max, cached_low, cached, cached_ra;
>  	char proc_io_fname[128];
> +	struct tcase *tc = &tcases[n];
> +
> +	tst_res(TINFO, "Test #%d: %s", n, tc->tname);
> +
>  	sprintf(proc_io_fname, "/proc/%u/io", getpid());
>  
> -	if (create_testfile(use_overlay) != 0)
> +	if (create_testfile(tc->use_overlay) != 0)
>  		return;
>  
> +	/* Use either readahead() syscall or POSIX_FADV_WILLNEED */
> +	readahead_func = tc->use_fadvise ? fadvise_willneed : libc_readahead;
> +
>  	/* find out how much can cache hold if we read whole file */
>  	read_testfile(0, testfile, testfile_size, &read_bytes, &usec, &cached);
>  	cached_max = get_cached_size();
> @@ -357,7 +402,7 @@ static struct tst_test test = {
>  	.cleanup = cleanup,
>  	.options = options,
>  	.test = test_readahead,
> -	.tcnt = 2, /* Repeat with overlayfs */
> +	.tcnt = ARRAY_SIZE(tcases),
>  };
>  
>  #else /* __NR_readahead */
> --
> 2.17.1
> 
> 
> --
> Mailing list info: https://lists.linux.it/listinfo/ltp
> 

Re: [LTP] [PATCH 3/4] syscalls/readahead02: test readahead() on an overlayfs file

[Amir Goldstein]

On Wed, Oct 3, 2018 at 3:48 PM Jan Stancek <jstancek@redhat.com> wrote:
>
>
>
> ----- Original Message -----
> > Repeat the test case on an overlayfs file.
> >
> > The new test case is a regression test for kernel commit b833a3660394
> > ("ovl: add ovl_fadvise()") which fixes a regression of readahead() on
> > an overlay file that was introduced by kernel commit 5b910bd615ba
> > ("ovl: fix GPF in swapfile_activate of file from overlayfs over xfs").
> >
> > Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> > ---
[...]
> This could use some TINFO message, what is being tested.
> .. and I see you added it in 4/4.
>

Yap. I hope you don;t mind this temporary simplicity....

> >       /* find out how much can cache hold if we read whole file */
> >       read_testfile(0, testfile, testfile_size, &read_bytes, &usec, &cached);
> >       cached_max = get_cached_size();
> > @@ -302,9 +340,12 @@ static void setup(void)
> >       tst_syscall(__NR_readahead, 0, 0, 0);
> >
> >       pagesize = getpagesize();
> > +}
> >
> > -     sprintf(testfile, "%s/testfile", mntpoint);
> > -     create_testfile();
> > +static void cleanup(void)
> > +{
> > +     if (ovl_mounted)
> > +             SAFE_UMOUNT(OVL_MNT);
> >  }
>
> This creates a small conflict, because setup and cleanup don't match.
> If you run this with multiple iterations (parameter -i), it's going
> to fail:
>   safe_macros.c:169: BROK: readahead02.c:124: mkdir(mntpoint/lower,0755) failed: EEXIST
>

Right. I have noticed and fixes it in
https://github.com/amir73il/ltp/commits/overlayfs-devel
was waiting on more review before posting v2.

Thanks,
Amir.

[LTP] [PATCH 3/4] syscalls/readahead02: test readahead() on an overlayfs file

[Amir Goldstein]

On Wed, Oct 3, 2018 at 3:48 PM Jan Stancek <jstancek@redhat.com> wrote:
>
>
>
> ----- Original Message -----
> > Repeat the test case on an overlayfs file.
> >
> > The new test case is a regression test for kernel commit b833a3660394
> > ("ovl: add ovl_fadvise()") which fixes a regression of readahead() on
> > an overlay file that was introduced by kernel commit 5b910bd615ba
> > ("ovl: fix GPF in swapfile_activate of file from overlayfs over xfs").
> >
> > Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> > ---
[...]
> This could use some TINFO message, what is being tested.
> .. and I see you added it in 4/4.
>

Yap. I hope you don;t mind this temporary simplicity....

> >       /* find out how much can cache hold if we read whole file */
> >       read_testfile(0, testfile, testfile_size, &read_bytes, &usec, &cached);
> >       cached_max = get_cached_size();
> > @@ -302,9 +340,12 @@ static void setup(void)
> >       tst_syscall(__NR_readahead, 0, 0, 0);
> >
> >       pagesize = getpagesize();
> > +}
> >
> > -     sprintf(testfile, "%s/testfile", mntpoint);
> > -     create_testfile();
> > +static void cleanup(void)
> > +{
> > +     if (ovl_mounted)
> > +             SAFE_UMOUNT(OVL_MNT);
> >  }
>
> This creates a small conflict, because setup and cleanup don't match.
> If you run this with multiple iterations (parameter -i), it's going
> to fail:
>   safe_macros.c:169: BROK: readahead02.c:124: mkdir(mntpoint/lower,0755) failed: EEXIST
>

Right. I have noticed and fixes it in
https://github.com/amir73il/ltp/commits/overlayfs-devel
was waiting on more review before posting v2.

Thanks,
Amir.

Re: [LTP] [PATCH 4/4] syscalls/readahead02: test readahead using posix_fadvise()

[Amir Goldstein]

On Wed, Oct 3, 2018 at 3:48 PM Jan Stancek <jstancek@redhat.com> wrote:
>
>
>
> ----- Original Message -----
> > The call to posix_fadvise(POSIX_FADV_WILLNEED) should have the same
> > effect as the call to readahead() syscall.
> >
> > Repeat the test cases for local file and overlayfs file with
> > posix_fadvise().
> >
> > The new test case is a regression test for kernel commit b833a3660394
> > ("ovl: add ovl_fadvise()") which fixes a regression of fadvise() on
> > an overlay file that was introduced by kernel commit 5b910bd615ba
> > ("ovl: fix GPF in swapfile_activate of file from overlayfs over xfs").
> >
> > Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> > ---
> >  .../kernel/syscalls/readahead/readahead02.c   | 57 +++++++++++++++++--
> >  1 file changed, 51 insertions(+), 6 deletions(-)
> >
> > diff --git a/testcases/kernel/syscalls/readahead/readahead02.c
> > b/testcases/kernel/syscalls/readahead/readahead02.c
> > index 191116f62..9ebed359d 100644
> > --- a/testcases/kernel/syscalls/readahead/readahead02.c
> > +++ b/testcases/kernel/syscalls/readahead/readahead02.c
> > @@ -51,6 +51,42 @@ static struct tst_option options[] = {
> >       {NULL, NULL, NULL}
> >  };
> >
> > +#ifndef _FILE_OFFSET_BITS
> > +#define _FILE_OFFSET_BITS 32
> > +#endif
>
> I don't think we should be touching this.
>
> > +
> > +#ifndef __NR_fadvise64
> > +#define __NR_fadvise64 0
> > +#endif
> > +
> > +static struct tcase {
> > +     const char *tname;
> > +     int use_overlay;
> > +     int use_fadvise;
> > +} tcases[] = {
> > +     { "readahead on file", 0, 0 },
> > +     { "readahead on overlayfs file", 1, 0 },
> > +/* Check this system has fadvise64 system which is used in posix_fadvise. */
> > +#if ((_FILE_OFFSET_BITS == 64) || (__NR_fadvise64 != 0))
>
> Can you elaborate on this check? Why do we need to care
> about _FILE_OFFSET_BITS?

I am not completely sure, this is copied along with the comment and ifndefs
above from posix_fadvise tests.

>
> > +     { "POSIX_FADV_WILLNEED on file", 0, 1 },
> > +     { "POSIX_FADV_WILLNEED on overlayfs file", 1, 1 },
> > +#endif
> > +};
> > +
> > +static int fadvise_willneed(int fd, off_t offset, size_t len)
> > +{
> > +     /* Should have the same effect as readahead() syscall */
> > +     return posix_fadvise(fd, offset, len, POSIX_FADV_WILLNEED);
> > +}
> > +
> > +static int libc_readahead(int fd, off_t offset, size_t len)
> > +{
> > +     return readahead(fd, offset, len);
> > +}
> > +
> > +typedef int (*readahead_func_t)(int, off_t, size_t);
> > +static readahead_func_t readahead_func = libc_readahead;
> > +
> >  static int check_ret(long expected_ret)
> >  {
> >       if (expected_ret == TST_RET) {
> > @@ -120,6 +156,9 @@ static int setup_overlay(void)
> >  {
> >       int ret;
> >
> > +     if (ovl_mounted)
> > +             return 0;
> > +
>
> You could call it once from setup() and drop check above
> and other call from create_testfile().
>

Yap. Already fixed in my github overlayfs-devel branch :)

Thanks,
Amir.

[LTP] [PATCH 4/4] syscalls/readahead02: test readahead using posix_fadvise()

[Amir Goldstein]

On Wed, Oct 3, 2018 at 3:48 PM Jan Stancek <jstancek@redhat.com> wrote:
>
>
>
> ----- Original Message -----
> > The call to posix_fadvise(POSIX_FADV_WILLNEED) should have the same
> > effect as the call to readahead() syscall.
> >
> > Repeat the test cases for local file and overlayfs file with
> > posix_fadvise().
> >
> > The new test case is a regression test for kernel commit b833a3660394
> > ("ovl: add ovl_fadvise()") which fixes a regression of fadvise() on
> > an overlay file that was introduced by kernel commit 5b910bd615ba
> > ("ovl: fix GPF in swapfile_activate of file from overlayfs over xfs").
> >
> > Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> > ---
> >  .../kernel/syscalls/readahead/readahead02.c   | 57 +++++++++++++++++--
> >  1 file changed, 51 insertions(+), 6 deletions(-)
> >
> > diff --git a/testcases/kernel/syscalls/readahead/readahead02.c
> > b/testcases/kernel/syscalls/readahead/readahead02.c
> > index 191116f62..9ebed359d 100644
> > --- a/testcases/kernel/syscalls/readahead/readahead02.c
> > +++ b/testcases/kernel/syscalls/readahead/readahead02.c
> > @@ -51,6 +51,42 @@ static struct tst_option options[] = {
> >       {NULL, NULL, NULL}
> >  };
> >
> > +#ifndef _FILE_OFFSET_BITS
> > +#define _FILE_OFFSET_BITS 32
> > +#endif
>
> I don't think we should be touching this.
>
> > +
> > +#ifndef __NR_fadvise64
> > +#define __NR_fadvise64 0
> > +#endif
> > +
> > +static struct tcase {
> > +     const char *tname;
> > +     int use_overlay;
> > +     int use_fadvise;
> > +} tcases[] = {
> > +     { "readahead on file", 0, 0 },
> > +     { "readahead on overlayfs file", 1, 0 },
> > +/* Check this system has fadvise64 system which is used in posix_fadvise. */
> > +#if ((_FILE_OFFSET_BITS == 64) || (__NR_fadvise64 != 0))
>
> Can you elaborate on this check? Why do we need to care
> about _FILE_OFFSET_BITS?

I am not completely sure, this is copied along with the comment and ifndefs
above from posix_fadvise tests.

>
> > +     { "POSIX_FADV_WILLNEED on file", 0, 1 },
> > +     { "POSIX_FADV_WILLNEED on overlayfs file", 1, 1 },
> > +#endif
> > +};
> > +
> > +static int fadvise_willneed(int fd, off_t offset, size_t len)
> > +{
> > +     /* Should have the same effect as readahead() syscall */
> > +     return posix_fadvise(fd, offset, len, POSIX_FADV_WILLNEED);
> > +}
> > +
> > +static int libc_readahead(int fd, off_t offset, size_t len)
> > +{
> > +     return readahead(fd, offset, len);
> > +}
> > +
> > +typedef int (*readahead_func_t)(int, off_t, size_t);
> > +static readahead_func_t readahead_func = libc_readahead;
> > +
> >  static int check_ret(long expected_ret)
> >  {
> >       if (expected_ret == TST_RET) {
> > @@ -120,6 +156,9 @@ static int setup_overlay(void)
> >  {
> >       int ret;
> >
> > +     if (ovl_mounted)
> > +             return 0;
> > +
>
> You could call it once from setup() and drop check above
> and other call from create_testfile().
>

Yap. Already fixed in my github overlayfs-devel branch :)

Thanks,
Amir.

Re: [LTP] [PATCH 2/4] syscalls/readahead02: Convert to newlib and cleanup

[Amir Goldstein]

On Wed, Oct 3, 2018 at 3:47 PM Jan Stancek <jstancek@redhat.com> wrote:
>
>
> ----- Original Message -----
> > * Use SAFE macros
> >
> > * Use SPDX-License-Identifier
> >
> > * No need to cleanup test file from temp dir
> >
> > Signed-off-by: Amir Goldstein <amir73il@gmail.com>
>
> Hi,
>
> ack to 1/4
>
> >
> >  static int has_file(const char *fname, int required)
> >  {
> > -     int ret;
> >       struct stat buf;
> > -     ret = stat(fname, &buf);
> > -     if (ret == -1) {
> > -             if (errno == ENOENT)
> > -                     if (required)
> > -                             tst_brkm(TCONF, cleanup, "%s not available",
> > -                                      fname);
> > -                     else
> > -                             return 0;
> > -             else
> > -                     tst_brkm(TBROK | TERRNO, cleanup, "stat %s", fname);
> > +
> > +     if (stat(fname, &buf) == -1) {
> > +             if (errno != ENOENT)
> > +                     tst_brk(TBROK | TERRNO, "stat %s", fname);
> > +             if (required)
> > +                     tst_brk(TCONF, "%s not available", fname);
> >       }
> >       return 1;
> >  }
>
> This will return 1 even when file doesn't exist.
> (Not that it makes big difference for test)
>

Oops. better fix it anyway.
I think it matters if /proc/pid/io does not exist.

>
> > +static struct tst_test test = {
> > +     .needs_root = 1,
> > +     .needs_tmpdir = 1,
> > +     .mount_device = 1,
> > +     .mntpoint = mntpoint,
> > +     .setup = setup,
> > +     .options = options,
> > +     .test_all = test_readahead,
> > +};
>
> Would it make sense to enable this for 'all_filesystems = 1'?

I don't know. Not sure which tests are good candidates for that.
But anyway it's a different change.

> Previously we used whatever fs /tmp was, now we seem to default
> always to ext2.
>

Not exactly. Before cleanup we either used whatever fs /tmp was
or if it was tmpfs we created a loop backed default fs (ext2?).

Now we skip the conditional part and always format fs.
Do you think that matters for the test?

Thanks,
Amir.

[LTP] [PATCH 2/4] syscalls/readahead02: Convert to newlib and cleanup

[Amir Goldstein]

On Wed, Oct 3, 2018 at 3:47 PM Jan Stancek <jstancek@redhat.com> wrote:
>
>
> ----- Original Message -----
> > * Use SAFE macros
> >
> > * Use SPDX-License-Identifier
> >
> > * No need to cleanup test file from temp dir
> >
> > Signed-off-by: Amir Goldstein <amir73il@gmail.com>
>
> Hi,
>
> ack to 1/4
>
> >
> >  static int has_file(const char *fname, int required)
> >  {
> > -     int ret;
> >       struct stat buf;
> > -     ret = stat(fname, &buf);
> > -     if (ret == -1) {
> > -             if (errno == ENOENT)
> > -                     if (required)
> > -                             tst_brkm(TCONF, cleanup, "%s not available",
> > -                                      fname);
> > -                     else
> > -                             return 0;
> > -             else
> > -                     tst_brkm(TBROK | TERRNO, cleanup, "stat %s", fname);
> > +
> > +     if (stat(fname, &buf) == -1) {
> > +             if (errno != ENOENT)
> > +                     tst_brk(TBROK | TERRNO, "stat %s", fname);
> > +             if (required)
> > +                     tst_brk(TCONF, "%s not available", fname);
> >       }
> >       return 1;
> >  }
>
> This will return 1 even when file doesn't exist.
> (Not that it makes big difference for test)
>

Oops. better fix it anyway.
I think it matters if /proc/pid/io does not exist.

>
> > +static struct tst_test test = {
> > +     .needs_root = 1,
> > +     .needs_tmpdir = 1,
> > +     .mount_device = 1,
> > +     .mntpoint = mntpoint,
> > +     .setup = setup,
> > +     .options = options,
> > +     .test_all = test_readahead,
> > +};
>
> Would it make sense to enable this for 'all_filesystems = 1'?

I don't know. Not sure which tests are good candidates for that.
But anyway it's a different change.

> Previously we used whatever fs /tmp was, now we seem to default
> always to ext2.
>

Not exactly. Before cleanup we either used whatever fs /tmp was
or if it was tmpfs we created a loop backed default fs (ext2?).

Now we skip the conditional part and always format fs.
Do you think that matters for the test?

Thanks,
Amir.

Re: [LTP] [PATCH 2/4] syscalls/readahead02: Convert to newlib and cleanup

[Jan Stancek]

----- Original Message -----
> On Wed, Oct 3, 2018 at 3:47 PM Jan Stancek <jstancek@redhat.com> wrote:
> >
> >
> > ----- Original Message -----
> > > * Use SAFE macros
> > >
> > > * Use SPDX-License-Identifier
> > >
> > > * No need to cleanup test file from temp dir
> > >
> > > Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> >
> > Hi,
> >
> > ack to 1/4
> >
> > >
> > >  static int has_file(const char *fname, int required)
> > >  {
> > > -     int ret;
> > >       struct stat buf;
> > > -     ret = stat(fname, &buf);
> > > -     if (ret == -1) {
> > > -             if (errno == ENOENT)
> > > -                     if (required)
> > > -                             tst_brkm(TCONF, cleanup, "%s not
> > > available",
> > > -                                      fname);
> > > -                     else
> > > -                             return 0;
> > > -             else
> > > -                     tst_brkm(TBROK | TERRNO, cleanup, "stat %s",
> > > fname);
> > > +
> > > +     if (stat(fname, &buf) == -1) {
> > > +             if (errno != ENOENT)
> > > +                     tst_brk(TBROK | TERRNO, "stat %s", fname);
> > > +             if (required)
> > > +                     tst_brk(TCONF, "%s not available", fname);
> > >       }
> > >       return 1;
> > >  }
> >
> > This will return 1 even when file doesn't exist.
> > (Not that it makes big difference for test)
> >
> 
> Oops. better fix it anyway.
> I think it matters if /proc/pid/io does not exist.
> 
> >
> > > +static struct tst_test test = {
> > > +     .needs_root = 1,
> > > +     .needs_tmpdir = 1,
> > > +     .mount_device = 1,
> > > +     .mntpoint = mntpoint,
> > > +     .setup = setup,
> > > +     .options = options,
> > > +     .test_all = test_readahead,
> > > +};
> >
> > Would it make sense to enable this for 'all_filesystems = 1'?
> 
> I don't know. Not sure which tests are good candidates for that.
> But anyway it's a different change.
> 
> > Previously we used whatever fs /tmp was, now we seem to default
> > always to ext2.
> >
> 
> Not exactly. Before cleanup we either used whatever fs /tmp was
> or if it was tmpfs we created a loop backed default fs (ext2?).
> 
> Now we skip the conditional part and always format fs.
> Do you think that matters for the test?

Isn't the reason for your series to test on different fs as well? :-)

I don't have strong opinion on this, it's a test for "suggestion to
kernel", so it could also make it prone to report more false positives.
Feel free to ignore this comment for v2.

Regards,
Jan

> 
> Thanks,
> Amir.
> 

[LTP] [PATCH 2/4] syscalls/readahead02: Convert to newlib and cleanup

[Jan Stancek]

----- Original Message -----
> On Wed, Oct 3, 2018 at 3:47 PM Jan Stancek <jstancek@redhat.com> wrote:
> >
> >
> > ----- Original Message -----
> > > * Use SAFE macros
> > >
> > > * Use SPDX-License-Identifier
> > >
> > > * No need to cleanup test file from temp dir
> > >
> > > Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> >
> > Hi,
> >
> > ack to 1/4
> >
> > >
> > >  static int has_file(const char *fname, int required)
> > >  {
> > > -     int ret;
> > >       struct stat buf;
> > > -     ret = stat(fname, &buf);
> > > -     if (ret == -1) {
> > > -             if (errno == ENOENT)
> > > -                     if (required)
> > > -                             tst_brkm(TCONF, cleanup, "%s not
> > > available",
> > > -                                      fname);
> > > -                     else
> > > -                             return 0;
> > > -             else
> > > -                     tst_brkm(TBROK | TERRNO, cleanup, "stat %s",
> > > fname);
> > > +
> > > +     if (stat(fname, &buf) == -1) {
> > > +             if (errno != ENOENT)
> > > +                     tst_brk(TBROK | TERRNO, "stat %s", fname);
> > > +             if (required)
> > > +                     tst_brk(TCONF, "%s not available", fname);
> > >       }
> > >       return 1;
> > >  }
> >
> > This will return 1 even when file doesn't exist.
> > (Not that it makes big difference for test)
> >
> 
> Oops. better fix it anyway.
> I think it matters if /proc/pid/io does not exist.
> 
> >
> > > +static struct tst_test test = {
> > > +     .needs_root = 1,
> > > +     .needs_tmpdir = 1,
> > > +     .mount_device = 1,
> > > +     .mntpoint = mntpoint,
> > > +     .setup = setup,
> > > +     .options = options,
> > > +     .test_all = test_readahead,
> > > +};
> >
> > Would it make sense to enable this for 'all_filesystems = 1'?
> 
> I don't know. Not sure which tests are good candidates for that.
> But anyway it's a different change.
> 
> > Previously we used whatever fs /tmp was, now we seem to default
> > always to ext2.
> >
> 
> Not exactly. Before cleanup we either used whatever fs /tmp was
> or if it was tmpfs we created a loop backed default fs (ext2?).
> 
> Now we skip the conditional part and always format fs.
> Do you think that matters for the test?

Isn't the reason for your series to test on different fs as well? :-)

I don't have strong opinion on this, it's a test for "suggestion to
kernel", so it could also make it prone to report more false positives.
Feel free to ignore this comment for v2.

Regards,
Jan

> 
> Thanks,
> Amir.
> 

Re: [LTP] [PATCH 2/4] syscalls/readahead02: Convert to newlib and cleanup

[Amir Goldstein]

On Wed, Oct 3, 2018 at 7:17 PM Jan Stancek <jstancek@redhat.com> wrote:
>
[...]
> > > Would it make sense to enable this for 'all_filesystems = 1'?
> >
> > I don't know. Not sure which tests are good candidates for that.
> > But anyway it's a different change.
> >
[...]
>
> Isn't the reason for your series to test on different fs as well? :-)
>

Well, yes and no.

Yes, overlayfs is a filesystem driver.

No, I don't think overlayfs was what the author of all_filesystems had in mind,
because setting .dev_fs_type = "overlay" is not enough to describe the setup.
That sort of setup would need to be described something like this:
.dev_fs_type = "foo",
.mntpoint = MNTPOINT,
.overlay_mount = OVL_MNT,

So overlayfs does not belong on the all_filesystems axis IMO, it belongs
on a tangent axis of the test matrix, e.g.:
.all_filesystems = 1,
.needs_overlay = 1,
.mntpoint = MNTPOINT,
.overlay_mount = OVL_MNT,

I personally question the value of such feature, but putting something like
setup_overlay() in a library is probably not a bad idea.
A similar pattern already exists in inotify07/inotify08, however in
those tests as
well as in readahead02, a feature of .need_overlay = 1 if existed would not
have been sufficient.

Thanks,
Amir.

[LTP] [PATCH 2/4] syscalls/readahead02: Convert to newlib and cleanup

[Amir Goldstein]

On Wed, Oct 3, 2018 at 7:17 PM Jan Stancek <jstancek@redhat.com> wrote:
>
[...]
> > > Would it make sense to enable this for 'all_filesystems = 1'?
> >
> > I don't know. Not sure which tests are good candidates for that.
> > But anyway it's a different change.
> >
[...]
>
> Isn't the reason for your series to test on different fs as well? :-)
>

Well, yes and no.

Yes, overlayfs is a filesystem driver.

No, I don't think overlayfs was what the author of all_filesystems had in mind,
because setting .dev_fs_type = "overlay" is not enough to describe the setup.
That sort of setup would need to be described something like this:
.dev_fs_type = "foo",
.mntpoint = MNTPOINT,
.overlay_mount = OVL_MNT,

So overlayfs does not belong on the all_filesystems axis IMO, it belongs
on a tangent axis of the test matrix, e.g.:
.all_filesystems = 1,
.needs_overlay = 1,
.mntpoint = MNTPOINT,
.overlay_mount = OVL_MNT,

I personally question the value of such feature, but putting something like
setup_overlay() in a library is probably not a bad idea.
A similar pattern already exists in inotify07/inotify08, however in
those tests as
well as in readahead02, a feature of .need_overlay = 1 if existed would not
have been sufficient.

Thanks,
Amir.