syscalls/cma: test case should return TCONF if syscall doesn't exist

PATCH New Jason Xing 6 participants 23 messages Dec 3, 2018 – Feb 26, 2019

Share & tools

Jason Xing

author

2018-12-03 09:43 UTC #0 lore raw

PATCH 1/1 syscalls/cma: test case should return TCONF if syscall doesn't exist

The original code returns TFAIL when process_vm_readv/_writev syscalls do not exist, while we are expecting a TCONF. During investigation, we find it is due to '#if defined(__NR_process_vm_readv/_writev)' fails to work as expected. We fix this issue by using ltp_syscall() function in setup function. BTW, we don't use new tst_syscall() interface due to compatibility issue exists between old apis and new ones.

Signed-off-by jasonxing <[email protected]>

Signed-off-by Caspar Zhang <[email protected]>

---
 testcases/kernel/syscalls/cma/process_vm.h         | 62 ----------------------
 testcases/kernel/syscalls/cma/process_vm01.c       | 30 ++++-------
 testcases/kernel/syscalls/cma/process_vm_readv02.c | 12 ++---
 testcases/kernel/syscalls/cma/process_vm_readv03.c | 14 ++---
 .../kernel/syscalls/cma/process_vm_writev02.c      | 12 ++---
 5 files changed, 30 insertions(+), 100 deletions(-)
 delete mode 100644 testcases/kernel/syscalls/cma/process_vm.h

testcases/kernel/syscalls/cma/process_vm.h deleted +0 ~~-62~~

deleted file mode 100644
index 1ecf325..0000000
@@ -1,62 +0,0 @@
-/*
- * 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.
- */
-
-#ifndef _PROCESS_VM_H_
-#define _PROCESS_VM_H_
-
-#include <sys/types.h>
-#include <sys/ipc.h>
-#include <sys/sem.h>
-#include <sys/syscall.h>
-#include <sys/uio.h>
-#include <unistd.h>
-#include "lapi/semun.h"
-
-static inline ssize_t test_process_vm_readv(pid_t pid,
-		const struct iovec *lvec, unsigned long liovcnt,
-		const struct iovec *rvec, unsigned long riovcnt,
-		unsigned long flags)
-{
-#if defined(__NR_process_vm_readv)
-	return syscall(__NR_process_vm_readv, pid, lvec, liovcnt,
-			rvec, riovcnt, flags);
-#else
-	return syscall(-1);
-#endif
-}
-
-static inline ssize_t test_process_vm_writev(pid_t pid,
-		const struct iovec *lvec, unsigned long liovcnt,
-		const struct iovec *rvec, unsigned long riovcnt,
-		unsigned long flags)
-{
-#if defined(__NR_process_vm_writev)
-	return syscall(__NR_process_vm_writev, pid, lvec, liovcnt,
-			rvec, riovcnt, flags);
-#else
-	return syscall(-1);
-#endif
-}
-
-#endif /* _PROCESS_VM_H_ */

testcases/kernel/syscalls/cma/process_vm01.c +11 ~~-19~~

index bc6b04f..f9bd865 100644
@@ -42,7 +42,7 @@
 #include "config.h"
 #include "test.h"
 #include "safe_macros.h"
-#include "process_vm.h"
+#include "lapi/syscalls.h"
 
 struct process_vm_params {
 	int len;

@@ -103,20 +103,10 @@ static void setup(char *argv[])
 			 " at the same time.");
 	else if (rflag) {
 		TCID = TCID_readv;
-#if defined(__NR_process_vm_readv)
 		cma_test_params = cma_test_params_read;
-#else
-		tst_brkm(TCONF, NULL, "process_vm_readv does not"
-			 " exist on your system.");
-#endif
 	} else if (wflag) {
 		TCID = TCID_writev;
-#if defined(__NR_process_vm_writev)
 		cma_test_params = cma_test_params_write;
-#else
-		tst_brkm(TCONF, NULL, "process_vm_writev does not"
-			 " exist on your system.");
-#endif
 	} else
 		tst_brkm(TBROK, NULL, "Parameter missing, required -r or -w.");
 	TEST_PAUSE;

@@ -134,18 +124,20 @@ static void help(void)
 
 static void cma_test_params_read(struct process_vm_params *params)
 {
-	TEST(test_process_vm_readv(params->pid,
-				   params->lvec, params->liovcnt,
-				   params->rvec, params->riovcnt,
-				   params->flags));
+	TEST(ltp_syscall(__NR_process_vm_readv,
+			 params->pid,
+			 params->lvec, params->liovcnt,
+			 params->rvec, params->riovcnt,
+			 params->flags));
 }
 
 static void cma_test_params_write(struct process_vm_params *params)
 {
-	TEST(test_process_vm_writev(params->pid,
-				    params->lvec, params->liovcnt,
-				    params->rvec, params->riovcnt,
-				    params->flags));
+	TEST(ltp_syscall(__NR_process_vm_writev,
+			 params->pid,
+			 params->lvec, params->liovcnt,
+			 params->rvec, params->riovcnt,
+			 params->flags));
 }
 
 static int cma_check_ret(long expected_ret, long act_ret)

testcases/kernel/syscalls/cma/process_vm_readv02.c +6 -6

index ef9b7e1..e1fc32a 100644
@@ -29,7 +29,7 @@
 
 #include "test.h"
 #include "safe_macros.h"
-#include "process_vm.h"
+#include "lapi/syscalls.h"
 
 char *TCID = "process_vm_readv02";
 int TST_TOTAL = 1;

@@ -134,7 +134,8 @@ static void child_invoke(void)
 	remote.iov_len = len;
 
 	tst_resm(TINFO, "child 1: reading string from same memory location.");
-	TEST(test_process_vm_readv(pids[0], &local, 1, &remote, 1, 0));
+	TEST(ltp_syscall(__NR_process_vm_readv, pids[0],
+			 &local, 1, &remote, 1, 0));
 	if (TEST_RETURN != len)
 		tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");
 	if (strncmp(lp, tst_string, len) != 0)

@@ -148,10 +149,9 @@ static void setup(void)
 {
 	tst_require_root();
 
-#if !defined(__NR_process_vm_readv)
-	tst_brkm(TCONF, NULL, "process_vm_readv does not exist "
-		 "on your system");
-#endif
+	/* Just a sanity check of the existence of syscall */
+	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0, NULL, 0, 0);
+
 	tst_tmpdir();
 	TST_CHECKPOINT_INIT(cleanup);

testcases/kernel/syscalls/cma/process_vm_readv03.c +7 -7

index 8b8dfc3..45f7c92 100644
@@ -32,7 +32,7 @@
 
 #include "test.h"
 #include "safe_macros.h"
-#include "process_vm.h"
+#include "lapi/syscalls.h"
 
 char *TCID = "process_vm_readv03";
 int TST_TOTAL = 1;

@@ -180,7 +180,8 @@ static long *fetch_remote_addrs(void)
 	remote.iov_base = foo;
 	remote.iov_len = len;
 
-	TEST(test_process_vm_readv(pids[0], &local, 1, &remote, 1, 0));
+	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], &local,
+			 1, &remote, 1, 0));
 	if (TEST_RETURN != len)
 		tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");

@@ -212,7 +213,7 @@ static void child_invoke(int *bufsz_arr)
 	tst_resm(TINFO, "child 1: %d local iovecs initialized.",
 		 NUM_LOCAL_VECS);
 
-	TEST(test_process_vm_readv(pids[0], local, NUM_LOCAL_VECS,
+	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local, NUM_LOCAL_VECS,
 				   remote, nr_iovecs, 0));
 	if (TEST_RETURN != bufsz)
 		tst_brkm(TBROK | TERRNO, tst_exit, "process_vm_readv");

@@ -246,14 +247,13 @@ static void setup(void)
 {
 	tst_require_root();
 
+	/* Just a sanity check of the existence of syscall */
+	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0, NULL, 0, 0);
+
 	nr_iovecs = nflag ? SAFE_STRTOL(NULL, nr_opt, 1, IOV_MAX) : 10;
 	bufsz = sflag ? SAFE_STRTOL(NULL, sz_opt, NUM_LOCAL_VECS, LONG_MAX)
 	    : 100000;
 
-#if !defined(__NR_process_vm_readv)
-	tst_brkm(TCONF, NULL, "process_vm_readv does not exist "
-		 "on your system");
-#endif
 	tst_tmpdir();
 	TST_CHECKPOINT_INIT(cleanup);
 	srand(time(NULL));

testcases/kernel/syscalls/cma/process_vm_writev02.c +6 -6

index b707760..b1edba5 100644
@@ -30,7 +30,7 @@
 
 #include "test.h"
 #include "safe_macros.h"
-#include "process_vm.h"
+#include "lapi/syscalls.h"
 
 char *TCID = "process_vm_writev02";
 int TST_TOTAL = 1;

@@ -171,7 +171,8 @@ static void child_write(void)
 	remote.iov_len = bufsz;
 
 	tst_resm(TINFO, "child 2: write to the same memory location.");
-	TEST(test_process_vm_writev(pids[0], &local, 1, &remote, 1, 0));
+	TEST(ltp_syscall(__NR_process_vm_writev, pids[0], &local,
+			 1, &remote, 1, 0));
 	if (TEST_RETURN != bufsz)
 		tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");
 }

@@ -180,14 +181,13 @@ static void setup(void)
 {
 	tst_require_root();
 
+	/* Just a sanity check of the existence of syscall */
+	ltp_syscall(__NR_process_vm_writev, getpid(), NULL, 0, NULL, 0, 0);
+
 	bufsz =
 	    sflag ? SAFE_STRTOL(NULL, sz_opt, 1, LONG_MAX - PADDING_SIZE * 2)
 	    : 100000;
 
-#if !defined(__NR_process_vm_readv)
-	tst_brkm(TCONF, NULL, "process_vm_writev does not exist "
-		 "on your system");
-#endif
 	tst_tmpdir();
 	TST_CHECKPOINT_INIT(cleanup);

Signature · 1 line

1.8.3.1

Jan Stancek

2018-12-03 09:58 UTC #1 lore raw

PATCH 1/1 syscalls/cma: test case should return TCONF if syscall doesn't exist

----- Original Message -----

Quoted 6 lines

The original code returns TFAIL when process_vm_readv/_writev syscalls do not exist, while we are expecting a TCONF. During investigation, we find it is due to '#if defined(__NR_process_vm_readv/_writev)' fails to work as expected. We fix this issue by using ltp_syscall() function in setup function. BTW, we don't use new tst_syscall() interface due to compatibility issue exists between old apis and new ones.

Fine by me, but then you should also add syscall numbers to include/lapi/syscalls:

process_vm01.c: In function ‘setup’:
process_vm01.c:97: warning: unused parameter ‘argv’
process_vm01.c: In function ‘cma_test_params_read’:
process_vm01.c:127: error: ‘__NR_process_vm_readv’ undeclared (first use in this function) process_vm01.c:127: error: (Each undeclared identifier is reported only once process_vm01.c:127: error: for each function it appears in.)
process_vm01.c: In function ‘cma_test_params_write’:
process_vm01.c:136: error: ‘__NR_process_vm_writev’ undeclared (first use in this function)

Regards,
Jan

Jason Xing

author

2018-12-04 02:49 UTC #2 lore raw

PATCH 1/1 syscalls/cma: test case should return TCONF if syscall doesn't exist

Quoted 10 lines · Jan Stancek

----- Original Message -----
Quoted 6 lines
The original code returns TFAIL when process_vm_readv/_writev syscalls do not exist, while we are expecting a TCONF. During investigation, we find it is due to '#if defined(__NR_process_vm_readv/_writev)' fails to work as expected. We fix this issue by using ltp_syscall() function in setup function. BTW, we don't use new tst_syscall() interface due to compatibility issue exists between old apis and new ones.

Quoted 3 lines

Fine by me, but then you should also add syscall numbers to include/lapi/syscalls:

Thanks.
I have TWO questions:

1. If I use this command #./strip_syscall.awk /usr/include/asm/unistd_64.h > x86_64.in, it will update and add more new syscall numbers into x86_64.in file which includes many other syscall numbers not only process_vm_readv/_writev. To solve this problem, can I just add the process_vm_readv/_writev syscall numbers and let go of other numbers?

2. I don't have so many types of machines that support different architectures like hppa and s390x. So what should I do to know this syscall number on other architectures. I'm running the tests on x86_64 arch.

Jason

Quoted 12 lines

process_vm01.c: In function ‘setup’:
process_vm01.c:97: warning: unused parameter ‘argv’
process_vm01.c: In function ‘cma_test_params_read’:
process_vm01.c:127: error: ‘__NR_process_vm_readv’ undeclared (first use in this function) process_vm01.c:127: error: (Each undeclared identifier is reported only once process_vm01.c:127: error: for each function it appears in.)
process_vm01.c: In function ‘cma_test_params_write’:
process_vm01.c:136: error: ‘__NR_process_vm_writev’ undeclared (first use in this function)
Regards,
Jan

Jan Stancek

2018-12-04 08:54 UTC #3 lore raw

PATCH 1/1 syscalls/cma: test case should return TCONF if syscall doesn't exist

----- Original Message -----

Quoted 20 lines · Jan Stancek

Quoted 10 lines
----- Original Message -----
Quoted 6 lines
The original code returns TFAIL when process_vm_readv/_writev syscalls do not exist, while we are expecting a TCONF. During investigation, we find it is due to '#if defined(__NR_process_vm_readv/_writev)' fails to work as expected. We fix this issue by using ltp_syscall() function in setup function. BTW, we don't use new tst_syscall() interface due to compatibility issue exists between old apis and new ones.
Quoted 3 lines
Fine by me, but then you should also add syscall numbers to include/lapi/syscalls:
Thanks.
I have TWO questions:

Quoted 6 lines

1. If I use this command #./strip_syscall.awk /usr/include/asm/unistd_64.h > x86_64.in, it will update and add more new syscall numbers into x86_64.in file which includes many other syscall numbers not only process_vm_readv/_writev. To solve this problem, can I just add the process_vm_readv/_writev syscall numbers and let go of other numbers?

Yes, I think updating only syscalls your patch is touching would be preferred.

Quoted 4 lines

2. I don't have so many types of machines that support different architectures like hppa and s390x. So what should I do to know this syscall number on other architectures. I'm running the tests on x86_64 arch.

You can find it in kernel sources, e.g. grep for it, or usually one of first commits mentioning the syscall will also add number.

Regards,
Jan

Quoted 19 lines

Jason
Quoted 15 lines
process_vm01.c: In function ‘setup’:
process_vm01.c:97: warning: unused parameter ‘argv’ process_vm01.c: In function ‘cma_test_params_read’:
process_vm01.c:127: error: ‘__NR_process_vm_readv’ undeclared (first use in this function)
process_vm01.c:127: error: (Each undeclared identifier is reported only once
process_vm01.c:127: error: for each function it appears in.) process_vm01.c: In function ‘cma_test_params_write’:
process_vm01.c:136: error: ‘__NR_process_vm_writev’ undeclared (first use in this function)
Regards,
Jan

Jason Xing

author

2018-12-05 13:43 UTC #4 lore raw

PATCH v2 1/2 syscalls/cma: test case should return TCONF if syscall doesn't exist

Signed-off-by Jason Xing <[email protected]>

Signed-off-by Caspar Zhang <[email protected]>

---
 testcases/kernel/syscalls/cma/process_vm.h         | 62 ----------------------
 testcases/kernel/syscalls/cma/process_vm01.c       | 30 ++++-------
 testcases/kernel/syscalls/cma/process_vm_readv02.c | 12 ++---
 testcases/kernel/syscalls/cma/process_vm_readv03.c | 14 ++---
 .../kernel/syscalls/cma/process_vm_writev02.c      | 12 ++---
 5 files changed, 30 insertions(+), 100 deletions(-)
 delete mode 100644 testcases/kernel/syscalls/cma/process_vm.h

testcases/kernel/syscalls/cma/process_vm.h deleted +0 ~~-62~~

deleted file mode 100644
index 1ecf325..0000000
@@ -1,62 +0,0 @@
-/*
- * 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.
- */
-
-#ifndef _PROCESS_VM_H_
-#define _PROCESS_VM_H_
-
-#include <sys/types.h>
-#include <sys/ipc.h>
-#include <sys/sem.h>
-#include <sys/syscall.h>
-#include <sys/uio.h>
-#include <unistd.h>
-#include "lapi/semun.h"
-
-static inline ssize_t test_process_vm_readv(pid_t pid,
-		const struct iovec *lvec, unsigned long liovcnt,
-		const struct iovec *rvec, unsigned long riovcnt,
-		unsigned long flags)
-{
-#if defined(__NR_process_vm_readv)
-	return syscall(__NR_process_vm_readv, pid, lvec, liovcnt,
-			rvec, riovcnt, flags);
-#else
-	return syscall(-1);
-#endif
-}
-
-static inline ssize_t test_process_vm_writev(pid_t pid,
-		const struct iovec *lvec, unsigned long liovcnt,
-		const struct iovec *rvec, unsigned long riovcnt,
-		unsigned long flags)
-{
-#if defined(__NR_process_vm_writev)
-	return syscall(__NR_process_vm_writev, pid, lvec, liovcnt,
-			rvec, riovcnt, flags);
-#else
-	return syscall(-1);
-#endif
-}
-
-#endif /* _PROCESS_VM_H_ */

testcases/kernel/syscalls/cma/process_vm01.c +11 ~~-19~~

index bc6b04f..f9bd865 100644
@@ -42,7 +42,7 @@
 #include "config.h"
 #include "test.h"
 #include "safe_macros.h"
-#include "process_vm.h"
+#include "lapi/syscalls.h"
 
 struct process_vm_params {
 	int len;

@@ -103,20 +103,10 @@ static void setup(char *argv[])
 			 " at the same time.");
 	else if (rflag) {
 		TCID = TCID_readv;
-#if defined(__NR_process_vm_readv)
 		cma_test_params = cma_test_params_read;
-#else
-		tst_brkm(TCONF, NULL, "process_vm_readv does not"
-			 " exist on your system.");
-#endif
 	} else if (wflag) {
 		TCID = TCID_writev;
-#if defined(__NR_process_vm_writev)
 		cma_test_params = cma_test_params_write;
-#else
-		tst_brkm(TCONF, NULL, "process_vm_writev does not"
-			 " exist on your system.");
-#endif
 	} else
 		tst_brkm(TBROK, NULL, "Parameter missing, required -r or -w.");
 	TEST_PAUSE;

@@ -134,18 +124,20 @@ static void help(void)
 
 static void cma_test_params_read(struct process_vm_params *params)
 {
-	TEST(test_process_vm_readv(params->pid,
-				   params->lvec, params->liovcnt,
-				   params->rvec, params->riovcnt,
-				   params->flags));
+	TEST(ltp_syscall(__NR_process_vm_readv,
+			 params->pid,
+			 params->lvec, params->liovcnt,
+			 params->rvec, params->riovcnt,
+			 params->flags));
 }
 
 static void cma_test_params_write(struct process_vm_params *params)
 {
-	TEST(test_process_vm_writev(params->pid,
-				    params->lvec, params->liovcnt,
-				    params->rvec, params->riovcnt,
-				    params->flags));
+	TEST(ltp_syscall(__NR_process_vm_writev,
+			 params->pid,
+			 params->lvec, params->liovcnt,
+			 params->rvec, params->riovcnt,
+			 params->flags));
 }
 
 static int cma_check_ret(long expected_ret, long act_ret)

testcases/kernel/syscalls/cma/process_vm_readv02.c +6 -6

index ef9b7e1..e1fc32a 100644
@@ -29,7 +29,7 @@
 
 #include "test.h"
 #include "safe_macros.h"
-#include "process_vm.h"
+#include "lapi/syscalls.h"
 
 char *TCID = "process_vm_readv02";
 int TST_TOTAL = 1;

@@ -134,7 +134,8 @@ static void child_invoke(void)
 	remote.iov_len = len;
 
 	tst_resm(TINFO, "child 1: reading string from same memory location.");
-	TEST(test_process_vm_readv(pids[0], &local, 1, &remote, 1, 0));
+	TEST(ltp_syscall(__NR_process_vm_readv, pids[0],
+			 &local, 1, &remote, 1, 0));
 	if (TEST_RETURN != len)
 		tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");
 	if (strncmp(lp, tst_string, len) != 0)

@@ -148,10 +149,9 @@ static void setup(void)
 {
 	tst_require_root();
 
-#if !defined(__NR_process_vm_readv)
-	tst_brkm(TCONF, NULL, "process_vm_readv does not exist "
-		 "on your system");
-#endif
+	/* Just a sanity check of the existence of syscall */
+	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0, NULL, 0, 0);
+
 	tst_tmpdir();
 	TST_CHECKPOINT_INIT(cleanup);

testcases/kernel/syscalls/cma/process_vm_readv03.c +7 -7

index 8b8dfc3..45f7c92 100644
@@ -32,7 +32,7 @@
 
 #include "test.h"
 #include "safe_macros.h"
-#include "process_vm.h"
+#include "lapi/syscalls.h"
 
 char *TCID = "process_vm_readv03";
 int TST_TOTAL = 1;

@@ -180,7 +180,8 @@ static long *fetch_remote_addrs(void)
 	remote.iov_base = foo;
 	remote.iov_len = len;
 
-	TEST(test_process_vm_readv(pids[0], &local, 1, &remote, 1, 0));
+	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], &local,
+			 1, &remote, 1, 0));
 	if (TEST_RETURN != len)
 		tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");

@@ -212,7 +213,7 @@ static void child_invoke(int *bufsz_arr)
 	tst_resm(TINFO, "child 1: %d local iovecs initialized.",
 		 NUM_LOCAL_VECS);
 
-	TEST(test_process_vm_readv(pids[0], local, NUM_LOCAL_VECS,
+	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local, NUM_LOCAL_VECS,
 				   remote, nr_iovecs, 0));
 	if (TEST_RETURN != bufsz)
 		tst_brkm(TBROK | TERRNO, tst_exit, "process_vm_readv");

@@ -246,14 +247,13 @@ static void setup(void)
 {
 	tst_require_root();
 
+	/* Just a sanity check of the existence of syscall */
+	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0, NULL, 0, 0);
+
 	nr_iovecs = nflag ? SAFE_STRTOL(NULL, nr_opt, 1, IOV_MAX) : 10;
 	bufsz = sflag ? SAFE_STRTOL(NULL, sz_opt, NUM_LOCAL_VECS, LONG_MAX)
 	    : 100000;
 
-#if !defined(__NR_process_vm_readv)
-	tst_brkm(TCONF, NULL, "process_vm_readv does not exist "
-		 "on your system");
-#endif
 	tst_tmpdir();
 	TST_CHECKPOINT_INIT(cleanup);
 	srand(time(NULL));

testcases/kernel/syscalls/cma/process_vm_writev02.c +6 -6

index b707760..b1edba5 100644
@@ -30,7 +30,7 @@
 
 #include "test.h"
 #include "safe_macros.h"
-#include "process_vm.h"
+#include "lapi/syscalls.h"
 
 char *TCID = "process_vm_writev02";
 int TST_TOTAL = 1;

@@ -171,7 +171,8 @@ static void child_write(void)
 	remote.iov_len = bufsz;
 
 	tst_resm(TINFO, "child 2: write to the same memory location.");
-	TEST(test_process_vm_writev(pids[0], &local, 1, &remote, 1, 0));
+	TEST(ltp_syscall(__NR_process_vm_writev, pids[0], &local,
+			 1, &remote, 1, 0));
 	if (TEST_RETURN != bufsz)
 		tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");
 }

@@ -180,14 +181,13 @@ static void setup(void)
 {
 	tst_require_root();
 
+	/* Just a sanity check of the existence of syscall */
+	ltp_syscall(__NR_process_vm_writev, getpid(), NULL, 0, NULL, 0, 0);
+
 	bufsz =
 	    sflag ? SAFE_STRTOL(NULL, sz_opt, 1, LONG_MAX - PADDING_SIZE * 2)
 	    : 100000;
 
-#if !defined(__NR_process_vm_readv)
-	tst_brkm(TCONF, NULL, "process_vm_writev does not exist "
-		 "on your system");
-#endif
 	tst_tmpdir();
 	TST_CHECKPOINT_INIT(cleanup);

Signature · 1 line

1.8.3.1

Jason Xing

author

2018-12-05 13:43 UTC #5 lore raw

PATCH v2 2/2 lapi/syscalls: Add syscall numbers for process_vm_readv/_writev

Signed-off-by Jason Xing <[email protected]>

Signed-off-by Caspar Zhang <[email protected]>

---
 include/lapi/syscalls/aarch64.in   | 2 ++
 include/lapi/syscalls/arm.in       | 2 ++
 include/lapi/syscalls/hppa.in      | 2 ++
 include/lapi/syscalls/i386.in      | 2 ++
 include/lapi/syscalls/ia64.in      | 2 ++
 include/lapi/syscalls/powerpc.in   | 2 ++
 include/lapi/syscalls/powerpc64.in | 2 ++
 include/lapi/syscalls/s390.in      | 2 ++
 include/lapi/syscalls/s390x.in     | 2 ++
 include/lapi/syscalls/sh.in        | 2 ++
 include/lapi/syscalls/sparc.in     | 2 ++
 include/lapi/syscalls/sparc64.in   | 2 ++
 include/lapi/syscalls/x86_64.in    | 2 ++
 13 files changed, 26 insertions(+)

include/lapi/syscalls/aarch64.in +2 -0

index c7ed2fe..bf04895 100644
@@ -253,6 +253,8 @@ clock_adjtime 266
 syncfs 267
 setns 268
 sendmmsg 269
+process_vm_readv 270
+process_vm_writev 271
 kcmp 272
 getrandom 278
 memfd_create 279

include/lapi/syscalls/arm.in +2 -0

index e63d15b..38de1d5 100644
@@ -334,6 +334,8 @@ accept4 (__NR_SYSCALL_BASE+366)
 fanotify_init (__NR_SYSCALL_BASE+367)
 fanotify_mark (__NR_SYSCALL_BASE+368)
 prlimit64 (__NR_SYSCALL_BASE+369)
+process_vm_readv(__NR_SYSCALL_BASE+376)
+process_vm_writev(__NR_SYSCALL_BASE+377)
 kcmp (__NR_SYSCALL_BASE+378)
 sched_setattr (__NR_SYSCALL_BASE+380)
 sched_getattr (__NR_SYSCALL_BASE+381)

include/lapi/syscalls/hppa.in +2 -0

index 56e89ba..182d393 100644
@@ -16,6 +16,8 @@ faccessat (__NR_openat + 12)
 splice 291
 tee 293
 vmsplice 294
+process_vm_readv 330
+process_vm_writev 331
 memfd_create 340
 membarrier 343
 execveat 342

include/lapi/syscalls/i386.in +2 -0

index ad14b36..2fb0188 100644
@@ -332,6 +332,8 @@ recvmmsg 337
 fanotify_init 338
 fanotify_mark 339
 prlimit64 340
+process_vm_readv 347
+process_vm_writev 348
 kcmp 349
 sched_setattr 351
 sched_getattr 352

include/lapi/syscalls/ia64.in +2 -0

index 828765f..081c9ed 100644
@@ -293,6 +293,8 @@ recvmmsg 1322
 fanotify_init 1323
 fanotify_mark 1324
 prlimit64 1325
+process_vm_readv 1332
+process_vm_writev 1333
 renameat2 1338
 getrandom 1339
 memfd_create 1340

include/lapi/syscalls/powerpc.in +2 -0

index 6176116..d16f9f9 100644
@@ -339,6 +339,8 @@ recvmsg 342
 recvmmsg 343
 accept4 344
 syscalls 345
+process_vm_readv  351
+process_vm_writev 352
 kcmp 354
 sched_setattr 355
 sched_getattr 356

include/lapi/syscalls/powerpc64.in +2 -0

index 6176116..d16f9f9 100644
@@ -339,6 +339,8 @@ recvmsg 342
 recvmmsg 343
 accept4 344
 syscalls 345
+process_vm_readv  351
+process_vm_writev 352
 kcmp 354
 sched_setattr 355
 sched_getattr 356

include/lapi/syscalls/s390.in +2 -0

index fa70d73..4019438 100644
@@ -323,6 +323,8 @@ setgid 214
 setfsuid 215
 setfsgid 216
 newfstatat 293
+process_vm_readv 340
+process_vm_writev 341
 kcmp 343
 sched_setattr 345
 sched_getattr 346

include/lapi/syscalls/s390x.in +2 -0

index fa70d73..4019438 100644
@@ -323,6 +323,8 @@ setgid 214
 setfsuid 215
 setfsgid 216
 newfstatat 293
+process_vm_readv 340
+process_vm_writev 341
 kcmp 343
 sched_setattr 345
 sched_getattr 346

include/lapi/syscalls/sh.in +2 -0

index f8ae691..8520bb1 100644
@@ -360,6 +360,8 @@ accept4 366
 fanotify_init 367
 fanotify_mark 368
 prlimit64 369
+process_vm_readv 376
+process_vm_writev 377
 kcmp 378
 memfd_create 385
 execveat 387

include/lapi/syscalls/sparc.in +2 -0

index 7d05da2..b7bb135 100644
@@ -330,6 +330,8 @@ recvmmsg 328
 fanotify_init 329
 fanotify_mark 330
 prlimit64 331
+process_vm_readv 338
+process_vm_writev 339
 kcmp 341
 renameat2 345
 getrandom 347

include/lapi/syscalls/sparc64.in +2 -0

index fb305c3..a211df8 100644
@@ -306,6 +306,8 @@ recvmmsg 328
 fanotify_init 329
 fanotify_mark 330
 prlimit64 331
+process_vm_readv 338
+process_vm_writev 339
 kcmp 341
 renameat2 345
 getrandom 347

include/lapi/syscalls/x86_64.in +2 -0

index 9504dc5..dae5769 100644
@@ -299,6 +299,8 @@ recvmmsg 299
 fanotify_init 300
 fanotify_mark 301
 prlimit64 302
+process_vm_readv 310
+process_vm_writev 311
 kcmp 312
 sched_setattr 314
 sched_getattr 315

Signature · 1 line

1.8.3.1

Petr Vorel

2018-12-05 22:07 UTC #6 lore raw

PATCH v2 2/2 lapi/syscalls: Add syscall numbers for process_vm_readv/_writev

Hi Jason,

Quoted 3 lines

Signed-off-by Jason Xing <[email protected]>
Signed-off-by Caspar Zhang <[email protected]>
---

...

include/lapi/syscalls/arm.in accept4 (__NR_SYSCALL_BASE+366) · line 338

Quoted 12 lines

include/lapi/syscalls/aarch64.in +0 -0

index c7ed2fe..bf04895 100644

include/lapi/syscalls/arm.in +2 -0

@@ -334,6 +334,8 @@ accept4 (__NR_SYSCALL_BASE+366)
 fanotify_init (__NR_SYSCALL_BASE+367)
 fanotify_mark (__NR_SYSCALL_BASE+368)
 prlimit64 (__NR_SYSCALL_BASE+369)
+process_vm_readv(__NR_SYSCALL_BASE+376)
+process_vm_writev(__NR_SYSCALL_BASE+377)

This triggered error during generation headers:
invalid line found: process_vm_readv(__NR_SYSCALL_BASE+376) (reproducible by: cd $l/include/lapi/syscalls; ./regen.sh)

and thus build error:
../../../../include/lapi/syscalls.h:13: error: unterminated #ifndef
#ifndef __LAPI_SYSCALLS_H__

I fixed it by adding space:

-process_vm_readv(__NR_SYSCALL_BASE+376)
-process_vm_writev(__NR_SYSCALL_BASE+377)
+process_vm_readv (__NR_SYSCALL_BASE+376)
+process_vm_writev (__NR_SYSCALL_BASE+377)

+ put this commit first and pushed.

Thanks for your patches.

Kind regards,
Petr

Jason Xing

author

2018-12-06 03:37 UTC #7 lore raw

PATCH v2 2/2 lapi/syscalls: Add syscall numbers for process_vm_readv/_writev

Quoted 29 lines · Petr Vorel

Hi Jason,
Quoted 3 lines
Signed-off-by Jason Xing <[email protected]>
Signed-off-by Caspar Zhang <[email protected]>
---
...
include/lapi/syscalls/arm.in accept4 (__NR_SYSCALL_BASE+366) · line 338
Quoted 12 lines
include/lapi/syscalls/aarch64.in +0 -0
index c7ed2fe..bf04895 100644
include/lapi/syscalls/arm.in +2 -0
@@ -334,6 +334,8 @@ accept4 (__NR_SYSCALL_BASE+366)
  fanotify_init (__NR_SYSCALL_BASE+367)
  fanotify_mark (__NR_SYSCALL_BASE+368)
  prlimit64 (__NR_SYSCALL_BASE+369)
+process_vm_readv(__NR_SYSCALL_BASE+376)
+process_vm_writev(__NR_SYSCALL_BASE+377)
This triggered error during generation headers:
invalid line found: process_vm_readv(__NR_SYSCALL_BASE+376) (reproducible by: cd $l/include/lapi/syscalls; ./regen.sh)
and thus build error:
../../../../include/lapi/syscalls.h:13: error: unterminated #ifndef
#ifndef __LAPI_SYSCALLS_H__
I fixed it by adding space:

Review

Quoted 6 lines

(quoted diff) +3 -2

-process_vm_readv(__NR_SYSCALL_BASE+376)
-process_vm_writev(__NR_SYSCALL_BASE+377)
+process_vm_readv (__NR_SYSCALL_BASE+376)
+process_vm_writev (__NR_SYSCALL_BASE+377)
 
+ put this commit first and pushed.

Thanks for your instructions.

Quoted 6 lines

Thanks for your patches.
Kind regards,
Petr

Li Wang

2019-02-15 10:50 UTC #8 lore raw

PATCH v2 1/2 syscalls/cma: test case should return TCONF if syscall doesn't exist

In reply to Jason Xing

Hi Jason,

A new failure was occured after applying this new patch.

From our observation, the latest process_vm_readv03.c failed with process_vm_readv() Invalid argument error on rhel7.3 platfrom. We suspect that this problem maybe caused by glibc or compiler, because if compile the test program with -O0 it get pass, but with the ltp
default optimization -O2 it get fail. Also, if we revert this new patch, the test PASS whatever with -O2 and -O0, that's very strange.

--------------------------------

# gcc -g -O2 -g -O2   -fno-strict-aliasing -pipe -Wall -W
-Wold-style-definition -D_FORTIFY_SOURCE=2 -I../../../../include
-I../../../../include -I../../../../include/old/   -L../../../../lib
process_vm_readv03.c   -lltp -o process_vm_readv03_O2

# ./process_vm_readv03_O2
process_vm_readv03 0 TINFO : child 0: 10 iovecs allocated and
initialized.

process_vm_readv03    0  TINFO  :  child 1: 10 remote iovecs received.
process_vm_readv03    0  TINFO  :  child 1: 4 local iovecs initialized.
process_vm_readv03    1  TBROK  :  process_vm_readv03.c:220:
process_vm_readv: errno=EINVAL(22): Invalid argument
process_vm_readv03    2  TBROK  :  process_vm_readv03.c:220: Remaining

cases broken
process_vm_readv03 1 TFAIL : process_vm_readv03.c:103: child 1
returns 512

# strace -f ./process_vm_readv03_O2
...
[pid 14676] process_vm_readv(14675, 0x7ffe40232ea0, 4, 0x7640c0, 10,
140728898420736) = -1 EINVAL (Invalid argument)
[pid 14676] write(1, "process_vm_readv03    1  \33[1;31m"...,
125process_vm_readv03    1  TBROK  :  process_vm_readv03.c:220:
process_vm_readv: errno=EINVAL(22): Invalid argument
) = 125
...

--------------------------------------------

# gcc -g -O0 -g -O0   -fno-strict-aliasing -pipe -Wall -W
-Wold-style-definition -D_FORTIFY_SOURCE=2 -I../../../../include
-I../../../../include -I../../../../include/old/   -L../../../../lib
process_vm_readv03.c   -lltp -o process_vm_readv03_O0

# ./process_vm_readv03_O0
process_vm_readv03 0 TINFO : child 0: 10 iovecs allocated and
initialized.

process_vm_readv03    0  TINFO  :  child 1: 10 remote iovecs received.
process_vm_readv03    0  TINFO  :  child 1: 4 local iovecs initialized.
process_vm_readv03    1  TPASS  :  child 1: all bytes are correctly

received.

# strace -f ./process_vm_readv03_O0
...
[pid 14703] process_vm_readv(14702,
[{"\0\1\2\3\4\5\6\7\10\t\n\v\f\r\16\17\20\21\22\23\24\25\26\27\30\31\32\33\34\35\36\37"...,
43269},
{"\5\6\7\10\t\n\v\f\r\16\17\20\21\22\23\24\25\26\27\30\31\32\33\34\35\36\37
!\"#$"..., 18907},
{"\340\341\342\343\344\345\346\347\350\351\352\353\354\355\356\357\360\361\362\363\364\365\366\367\370\371\372\373\374\375\376\377"...,
3203}, {"cdefghijklmnopqrstuvwxyz{|}~\177\200\201\202"..., 34621}], 4,
[{0x9570c0, 24638}, {0x95d110, 35640}, {0x965c50, 19802}, {0x96a9c0, 490},
{0x96abc0, 8624}, {0x96cd80, 3571}, {0x96db80, 2114}, {0x96e3d0, 2374},
{0x96ed20, 1139}, {0x96f1a0, 1608}], 10, 0) = 100000
[pid 14703] write(1, "process_vm_readv03    1  \33[1;32m"...,
89process_vm_readv03    1  TPASS  :  child 1: all bytes are correctly
received.
) = 89
...

# uname  -r
3.10.0-514.55.4.el7.x86_64

# rpm -q glibc
glibc-2.17-157.el7_3.5.x86_64

Signature · 5 lines

Regards,
Li Wang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190215/02b113eb/attachment-0001.html>

Jan Stancek

2019-02-15 13:53 UTC #9 lore raw

PATCH v2 1/2 syscalls/cma: test case should return TCONF if syscall doesn't exist

----- Original Message -----

Quoted 10 lines

Hi Jason,
A new failure was occured after applying this new patch.
From our observation, the latest process_vm_readv03.c failed with process_vm_readv() Invalid argument error on rhel7.3 platfrom. We suspect that this problem maybe caused by glibc or compiler, because if compile the test program with -O0 it get pass, but with the ltp default optimization -O2 it get fail. Also, if we revert this new patch, the test PASS whatever with -O2 and -O0, that's very strange.

One difference is that the extra function promoted int to long.
Can you try with patch below? (Note that I dropped also root requirement, man page says that it has same requirements as ptrace - you should be able to read from processes you own)

Regards,
Jan

testcases/kernel/syscalls/cma/process_vm_readv03.c +0 -0

index 45f7c92e499c..1cec442b3456 100644 
@@ -181,7 +181,7 @@ static long *fetch_remote_addrs(void)

remote.iov_len = len;

TEST(ltp_syscall(__NR_process_vm_readv, pids[0], &local,
- 1, &remote, 1, 0));
+ 1, &remote, 1, 0L));
if (TEST_RETURN != len)
tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");

@@ -214,7 +214,7 @@ static void child_invoke(int *bufsz_arr) NUM_LOCAL_VECS);

TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local, NUM_LOCAL_VECS,
- remote, nr_iovecs, 0));
+ remote, nr_iovecs, 0L));
if (TEST_RETURN != bufsz)
tst_brkm(TBROK | TERRNO, tst_exit, "process_vm_readv");

@@ -245,10 +245,8 @@ static void child_invoke(int *bufsz_arr)

static void setup(void)
{
- tst_require_root();
-
/* Just a sanity check of the existence of syscall */
- ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0, NULL, 0, 0);
+ ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0, NULL, 0, 0L);

nr_iovecs = nflag ? SAFE_STRTOL(NULL, nr_opt, 1, IOV_MAX) : 10; bufsz = sflag ? SAFE_STRTOL(NULL, sz_opt, NUM_LOCAL_VECS, LONG_MAX)

Quoted 57 lines

--------------------------------
# gcc -g -O2 -g -O2 -fno-strict-aliasing -pipe -Wall -W
(quoted diff) +0 -2
-Wold-style-definition -D_FORTIFY_SOURCE=2 -I../../../../include
-I../../../../include -I../../../../include/old/ -L../../../../lib
process_vm_readv03.c -lltp -o process_vm_readv03_O2
# ./process_vm_readv03_O2
process_vm_readv03 0 TINFO : child 0: 10 iovecs allocated and initialized. process_vm_readv03 0 TINFO : child 1: 10 remote iovecs received. process_vm_readv03 0 TINFO : child 1: 4 local iovecs initialized. process_vm_readv03 1 TBROK : process_vm_readv03.c:220: process_vm_readv:
errno=EINVAL(22): Invalid argument
process_vm_readv03 2 TBROK : process_vm_readv03.c:220: Remaining cases broken process_vm_readv03 1 TFAIL : process_vm_readv03.c:103: child 1 returns 512
# strace -f ./process_vm_readv03_O2
...
[pid 14676] process_vm_readv(14675, 0x7ffe40232ea0, 4, 0x7640c0, 10,
140728898420736) = -1 EINVAL (Invalid argument)
[pid 14676] write(1, "process_vm_readv03 1 \33[1;31m"...,
125process_vm_readv03 1 TBROK : process_vm_readv03.c:220: process_vm_readv:
errno=EINVAL(22): Invalid argument
) = 125
...
--------------------------------------------
# gcc -g -O0 -g -O0 -fno-strict-aliasing -pipe -Wall -W
(quoted diff) +0 -2
-Wold-style-definition -D_FORTIFY_SOURCE=2 -I../../../../include
-I../../../../include -I../../../../include/old/ -L../../../../lib
process_vm_readv03.c -lltp -o process_vm_readv03_O0
# ./process_vm_readv03_O0
process_vm_readv03 0 TINFO : child 0: 10 iovecs allocated and initialized. process_vm_readv03 0 TINFO : child 1: 10 remote iovecs received. process_vm_readv03 0 TINFO : child 1: 4 local iovecs initialized. process_vm_readv03 1 TPASS : child 1: all bytes are correctly received.
# strace -f ./process_vm_readv03_O0
...
[pid 14703] process_vm_readv(14702,
[{"\0\1\2\3\4\5\6\7\10\t\n\v\f\r\16\17\20\21\22\23\24\25\26\27\30\31\32\33\34\35\36\37"...,
43269},
{"\5\6\7\10\t\n\v\f\r\16\17\20\21\22\23\24\25\26\27\30\31\32\33\34\35\36\37
!\"#$"..., 18907},
{"\340\341\342\343\344\345\346\347\350\351\352\353\354\355\356\357\360\361\362\363\364\365\366\367\370\371\372\373\374\375\376\377"...,
3203}, {"cdefghijklmnopqrstuvwxyz{|}~\177\200\201\202"..., 34621}], 4,
[{0x9570c0, 24638}, {0x95d110, 35640}, {0x965c50, 19802}, {0x96a9c0, 490},
{0x96abc0, 8624}, {0x96cd80, 3571}, {0x96db80, 2114}, {0x96e3d0, 2374},
{0x96ed20, 1139}, {0x96f1a0, 1608}], 10, 0) = 100000
[pid 14703] write(1, "process_vm_readv03 1 \33[1;32m"...,
89process_vm_readv03 1 TPASS : child 1: all bytes are correctly received.
) = 89
...
# uname -r
3.10.0-514.55.4.el7.x86_64

Quoted 6 lines

# rpm -q glibc
glibc-2.17-157.el7_3.5.x86_64

--
Regards,
Li Wang

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190215/21715672/attachment.html>

Li Wang

2019-02-16 07:18 UTC #10 lore raw

PATCH v2 1/2 syscalls/cma: test case should return TCONF if syscall doesn't exist

Quoted 16 lines · Jan Stancek

------------------------------
Hi Jason,
A new failure was occured after applying this new patch.
From our observation, the latest process_vm_readv03.c failed with process_vm_readv() Invalid argument error on rhel7.3 platfrom. We suspect that this problem maybe caused by glibc or compiler, because if compile the test program with -O0 it get pass, but with the ltp
default optimization -O2 it get fail. Also, if we revert this new patch, the test PASS whatever with -O2 and -O0, that's very strange.

Quoted 2 lines

One difference is that the extra function promoted int to long.

Good catch.

Quoted 6 lines

Can you try with patch below? (Note that I dropped also root requirement, man page
says that it has same requirements as ptrace - you should be able to read from processes you own)
This change works for me(one comment in below).

But there still have a tiny question, in the syscall tracing log, I noticed a process_vm_readv() calling does not get this failure in function fetch_remote_addrs, what makes it difference with other two calls is something I'm thinking now.

====Tracing log without '0L' change====
12211 process_vm_readv(12211, 0, 0, 0, 0, 140728898420736) = -1 EINVAL (Invalid argument)
...
12213 process_vm_readv(12212, [{"\300\334\0\0\0\0\0\2\335\0\0\0\0\0\340\26\335\0\0\0\0\0\220h\335\0\0\0\0\0"..., 80}], 1, [{0xdc6060, 80}], 1, 0) = 80
...
12213 process_vm_readv(12212, 0x7ffecef1e330, 4, 0xdc60c0, 10, 140728898420736) = -1 EINVAL (Invalid argument)

testcases/kernel/syscalls/cma/process_vm_readv03.c static long *fetch_remote_addrs(void) · line 185

Quoted 13 lines

testcases/kernel/syscalls/cma/process_vm_readv03.c +1 -1

index 45f7c92e499c..1cec442b3456 100644
@@ -181,7 +181,7 @@ static long *fetch_remote_addrs(void)
        remote.iov_len = len;
 
        TEST(ltp_syscall(__NR_process_vm_readv, pids[0], &local,
-                        1, &remote, 1, 0));
+                        1, &remote, 1, 0L));

To make things more safely, probably we'd better change all the integer parameter too?

- 1, &remote, 1, 0));
+ 1L, &remote, 1L, 0L));

Signature · 5 lines

Regards,
Li Wang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190216/ba233afa/attachment-0001.html>

Jan Stancek

2019-02-16 10:28 UTC #11 lore raw

PATCH v2 1/2 syscalls/cma: test case should return TCONF if syscall doesn't exist

----- Original Message -----

Quoted 28 lines · Jan Stancek

Quoted 17 lines
------------------------------
Hi Jason,
A new failure was occured after applying this new patch.
From our observation, the latest process_vm_readv03.c failed with process_vm_readv() Invalid argument error on rhel7.3 platfrom. We suspect that this problem maybe caused by glibc or compiler, because if compile the test program with -O0 it get pass, but with the ltp
default optimization -O2 it get fail. Also, if we revert this new patch, the test PASS whatever with -O2 and -O0, that's very strange.
Quoted 2 lines
One difference is that the extra function promoted int to long.
Good catch.
Quoted 5 lines
Can you try with patch below? (Note that I dropped also root requirement, man page
says that it has same requirements as ptrace - you should be able to read from processes you own)

Quoted 6 lines

Quoted 1 line
This change works for me(one comment in below).
But there still have a tiny question, in the syscall tracing log, I noticed a process_vm_readv() calling does not get this failure in function fetch_remote_addrs, what makes it difference with other two calls is something I'm thinking now.

"Garbage" on stack most likely. On x86_64, 6th parameter is passed via stack, glibc copies quadword:

ENTRY (syscall)
movq %rdi, %rax /* Syscall number -> rax. */
movq %rsi, %rdi /* shift arg1 - arg5. */
movq %rdx, %rsi
movq %rcx, %rdx
movq %r8, %r10
movq %r9, %r8
movq 8(%rsp),%r9 /* arg6 is on the stack. */
syscall /* Do the system call. */

Quoted 31 lines

====Tracing log without '0L' change====
12211 process_vm_readv(12211, 0, 0, 0, 0, 140728898420736) = -1 EINVAL (Invalid argument)
...
12213 process_vm_readv(12212, [{"\300\334\0\0\0\0\0\2\335\0\0\0\0\0\340\26\335\0\0\0\0\0\220h\335\0\0\0\0\0"..., 80}], 1, [{0xdc6060, 80}], 1, 0) = 80
...
12213 process_vm_readv(12212, 0x7ffecef1e330, 4, 0xdc60c0, 10, 140728898420736) = -1 EINVAL (Invalid argument)
testcases/kernel/syscalls/cma/process_vm_readv03.c static long *fetch_remote_addrs(void) · line 185
Quoted 13 lines
testcases/kernel/syscalls/cma/process_vm_readv03.c +1 -1
index 45f7c92e499c..1cec442b3456 100644
@@ -181,7 +181,7 @@ static long *fetch_remote_addrs(void)
        remote.iov_len = len;
 
        TEST(ltp_syscall(__NR_process_vm_readv, pids[0], &local,
-                        1, &remote, 1, 0));
+                        1, &remote, 1, 0L));
 
To make things more safely, probably we'd better change all the integer parameter too?

Review

Quoted 2 lines

(quoted diff) +1 -1

-                        1, &remote, 1, 0));
+                        1L, &remote, 1L, 0L));

Agreed, let's match all parameters types as before commit:
9b02cd465f70 ("syscalls/cma: test case should return TCONF if syscall doesn't exist")

Quoted 5 lines

--
Regards,
Li Wang

Jan Stancek

2019-02-22 18:29 UTC #12 lore raw

PATCH v2 1/2 syscalls/cma: test case should return TCONF if syscall doesn't exist

----- Original Message -----

Quoted 83 lines

----- Original Message -----
Quoted 29 lines · Jan Stancek
Quoted 18 lines
------------------------------
Hi Jason,
A new failure was occured after applying this new patch.
From our observation, the latest process_vm_readv03.c failed with process_vm_readv() Invalid argument error on rhel7.3 platfrom. We suspect that this problem maybe caused by glibc or compiler, because if compile the
test program with -O0 it get pass, but with the ltp
default optimization -O2 it get fail. Also, if we revert this new patch, the test PASS whatever with -O2 and -O0, that's very strange.
Quoted 2 lines
One difference is that the extra function promoted int to long.
Good catch.
Quoted 5 lines
Can you try with patch below? (Note that I dropped also root requirement, man page
says that it has same requirements as ptrace - you should be able to read from processes you own)
Quoted 6 lines
Quoted 1 line
This change works for me(one comment in below).
But there still have a tiny question, in the syscall tracing log, I noticed a process_vm_readv() calling does not get this failure in function fetch_remote_addrs, what makes it difference with other two calls is something I'm thinking now.
"Garbage" on stack most likely. On x86_64, 6th parameter is passed via stack, glibc copies quadword:
ENTRY (syscall)
movq %rdi, %rax /* Syscall number -> rax. */
movq %rsi, %rdi /* shift arg1 - arg5. */
movq %rdx, %rsi
movq %rcx, %rdx
movq %r8, %r10
movq %r9, %r8
movq 8(%rsp),%r9 /* arg6 is on the stack. */
syscall /* Do the system call. */
Quoted 31 lines
====Tracing log without '0L' change====
12211 process_vm_readv(12211, 0, 0, 0, 0, 140728898420736) = -1 EINVAL (Invalid argument)
...
12213 process_vm_readv(12212, [{"\300\334\0\0\0\0\0\2\335\0\0\0\0\0\340\26\335\0\0\0\0\0\220h\335\0\0\0\0\0"..., 80}], 1, [{0xdc6060, 80}], 1, 0) = 80
...
12213 process_vm_readv(12212, 0x7ffecef1e330, 4, 0xdc60c0, 10, 140728898420736) = -1 EINVAL (Invalid argument)
testcases/kernel/syscalls/cma/process_vm_readv03.c static long *fetch_remote_addrs(void) · line 185
Quoted 13 lines
testcases/kernel/syscalls/cma/process_vm_readv03.c +1 -1
index 45f7c92e499c..1cec442b3456 100644
@@ -181,7 +181,7 @@ static long *fetch_remote_addrs(void)
        remote.iov_len = len;
 
        TEST(ltp_syscall(__NR_process_vm_readv, pids[0], &local,
-                        1, &remote, 1, 0));
+                        1, &remote, 1, 0L));
 
To make things more safely, probably we'd better change all the integer parameter too?

Quoted 6 lines

Review
Quoted 2 lines
(quoted diff) +1 -1
-                        1, &remote, 1, 0));
+                        1L, &remote, 1L, 0L));
Agreed, let's match all parameters types as before commit:
9b02cd465f70 ("syscalls/cma: test case should return TCONF if syscall doesn't exist")

So, it looks like this thread has stalled. I'll look into posting a patch, unless I hear someone is already working on it.

Regards,
Jan

Quoted 7 lines

Quoted 6 lines
--
Regards,
Li Wang

Caspar Zhang

2019-02-23 06:21 UTC #13 lore raw

PATCH v2 1/2 syscalls/cma: test case should return TCONF if syscall doesn't exist

Quoted 3 lines · Jan Stancek

----- Original Message -----

[snip]

Quoted 19 lines

Quoted 19 lines
Quoted 18 lines
testcases/kernel/syscalls/cma/process_vm_readv03.c static long *fetch_remote_addrs(void) · line 185
Quoted 12 lines
testcases/kernel/syscalls/cma/process_vm_readv03.c +1 -1
index 45f7c92e499c..1cec442b3456 100644
@@ -181,7 +181,7 @@ static long *fetch_remote_addrs(void)
        remote.iov_len = len;
 
        TEST(ltp_syscall(__NR_process_vm_readv, pids[0], &local,
-                        1, &remote, 1, 0));
+                        1, &remote, 1, 0L));
 
To make things more safely, probably we'd better change all the integer parameter too?
(quoted diff) +1 -1
-                        1, &remote, 1, 0));
+                        1L, &remote, 1L, 0L));

Quoted 6 lines

Quoted 3 lines
Agreed, let's match all parameters types as before commit:
9b02cd465f70 ("syscalls/cma: test case should return TCONF if syscall doesn't exist")
So, it looks like this thread has stalled. I'll look into posting a patch, unless I hear someone is already working on it.

Hi Jan,

Jason told me he was looking into it, I'll double confirm with him about the progress. Will update ASAP.

Thanks,
Caspar

Jason Xing

author

2019-02-25 08:22 UTC #14 lore raw

PATCH 1/1 syscalls/cma: fix the failure of compiling with O2 option

This issue is triggered by the commit 9b02cd465f70 which I wrote. If compile the process_vm_readv03.c with O2 option and run, it will return TFAIL. By casting the int type to long type in the ltp_syscall calling functions can we avoid this failure.

Signed-off-by Jason Xing <[email protected]>

---
 testcases/kernel/syscalls/cma/process_vm_readv03.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

testcases/kernel/syscalls/cma/process_vm_readv03.c +4 -4

index 45f7c92..e15ffee 100644
@@ -181,7 +181,7 @@ static long *fetch_remote_addrs(void)
 	remote.iov_len = len;
 
 	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], &local,
-			 1, &remote, 1, 0));
+			 1L, &remote, 1L, 0L));
 	if (TEST_RETURN != len)
 		tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");

@@ -213,8 +213,8 @@ static void child_invoke(int *bufsz_arr)
 	tst_resm(TINFO, "child 1: %d local iovecs initialized.",
 		 NUM_LOCAL_VECS);
 
-	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local, NUM_LOCAL_VECS,
-				   remote, nr_iovecs, 0));
+	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local, (long)NUM_LOCAL_VECS,
+				   remote, (long)nr_iovecs, 0L));
 	if (TEST_RETURN != bufsz)
 		tst_brkm(TBROK | TERRNO, tst_exit, "process_vm_readv");

@@ -248,7 +248,7 @@ static void setup(void)
 	tst_require_root();
 
 	/* Just a sanity check of the existence of syscall */
-	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0, NULL, 0, 0);
+	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0L, NULL, 0L, 0L);
 
 	nr_iovecs = nflag ? SAFE_STRTOL(NULL, nr_opt, 1, IOV_MAX) : 10;
 	bufsz = sflag ? SAFE_STRTOL(NULL, sz_opt, NUM_LOCAL_VECS, LONG_MAX)

Signature · 1 line

1.8.3.1

Li Wang

2019-02-25 10:34 UTC #15 lore raw

PATCH 1/1 syscalls/cma: fix the failure of compiling with O2 option

Quoted 25 lines · Jason Xing

This issue is triggered by the commit 9b02cd465f70 which I wrote. If compile the process_vm_readv03.c with O2 option and run, it will return TFAIL. By casting the int type to long type in the ltp_syscall calling functions can we avoid this failure.
Signed-off-by Jason Xing <[email protected]>
---
 testcases/kernel/syscalls/cma/process_vm_readv03.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
testcases/kernel/syscalls/cma/process_vm_readv03.c +1 -1
index 45f7c92..e15ffee 100644
@@ -181,7 +181,7 @@ static long *fetch_remote_addrs(void)
        remote.iov_len = len;
 
        TEST(ltp_syscall(__NR_process_vm_readv, pids[0], &local,
-                        1, &remote, 1, 0));
+                        1L, &remote, 1L, 0L));
        if (TEST_RETURN != len)
                tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");
 

testcases/kernel/syscalls/cma/process_vm_readv03.c static void child_invoke(int *bufsz_arr) · line 220

Quoted 10 lines

testcases/kernel/syscalls/cma/process_vm_readv03.c +2 -2

@@ -213,8 +213,8 @@ static void child_invoke(int *bufsz_arr)
        tst_resm(TINFO, "child 1: %d local iovecs initialized.",
                 NUM_LOCAL_VECS);
 
-       TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local,
 NUM_LOCAL_VECS,
-                                  remote, nr_iovecs, 0));
+       TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local,
 (long)NUM_LOCAL_VECS,
+                                  remote, (long)nr_iovecs, 0L));

Maybe cast them into 'unsigned long' type is better? otherwise this patch LGTM.

Cut from linux manual page:
------------------
ssize_t process_vm_readv(pid_t pid,
const struct iovec *local_iov, unsigned long liovcnt, const struct iovec *remote_iov, unsigned long riovcnt, unsigned long flags);

Signature · 5 lines

Regards,
Li Wang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190225/b534ae24/attachment.html>

Jan Stancek

2019-02-25 10:47 UTC #16 lore raw

PATCH 1/1 syscalls/cma: fix the failure of compiling with O2 option

In reply to Jason Xing

----- Original Message -----

Quoted 8 lines

This issue is triggered by the commit 9b02cd465f70 which I wrote. If compile the process_vm_readv03.c with O2 option and run, it will return TFAIL. By casting the int type to long type in the ltp_syscall calling functions can we avoid this failure.
Signed-off-by Jason Xing <[email protected]>
---
 testcases/kernel/syscalls/cma/process_vm_readv03.c | 8 ++++----

What about other tests that were changed by commit 9b02cd465f70 ?

Quoted 41 lines

 1 file changed, 4 insertions(+), 4 deletions(-)

testcases/kernel/syscalls/cma/process_vm_readv03.c +4 -5

index 45f7c92..e15ffee 100644
@@ -181,7 +181,7 @@ static long *fetch_remote_addrs(void)
 	remote.iov_len = len;
 
 	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], &local,
-			 1, &remote, 1, 0));
+			 1L, &remote, 1L, 0L));
 	if (TEST_RETURN != len)
 		tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");

@@ -213,8 +213,8 @@ static void child_invoke(int *bufsz_arr)
 	tst_resm(TINFO, "child 1: %d local iovecs initialized.",
 		 NUM_LOCAL_VECS);
 
-	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local, NUM_LOCAL_VECS,
-				   remote, nr_iovecs, 0));
+	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local,
 (long)NUM_LOCAL_VECS,
+				   remote, (long)nr_iovecs, 0L));
 	if (TEST_RETURN != bufsz)
 		tst_brkm(TBROK | TERRNO, tst_exit, "process_vm_readv");

@@ -248,7 +248,7 @@ static void setup(void)
 	tst_require_root();
 
 	/* Just a sanity check of the existence of syscall */
-	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0, NULL, 0, 0);
+	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0L, NULL, 0L, 0L);
 
 	nr_iovecs = nflag ? SAFE_STRTOL(NULL, nr_opt, 1, IOV_MAX) : 10;
 	bufsz = sflag ? SAFE_STRTOL(NULL, sz_opt, NUM_LOCAL_VECS, LONG_MAX)
--

1.8.3.1

Caspar Zhang

2019-02-25 13:16 UTC #17 lore raw

PATCH 1/1 syscalls/cma: fix the failure of compiling with O2 option

Quoted 8 lines · Jan Stancek

----- Original Message -----
Quoted 5 lines
This issue is triggered by the commit 9b02cd465f70 which I wrote. If compile the process_vm_readv03.c with O2 option and run, it will return TFAIL. By casting the int type to long type in the ltp_syscall calling functions can we avoid this failure.

Quoted 5 lines

Quoted 3 lines
Signed-off-by Jason Xing <[email protected]>
---
 testcases/kernel/syscalls/cma/process_vm_readv03.c | 8 ++++----
What about other tests that were changed by commit 9b02cd465f70 ?

I downgraded my glibc from 2.17-196 (a centos 7.4 version) to 2.17-157 was able to reproduce this issue. I also tried modifying the params in all TEST(ltp_syscall(__NR_process_vm... calls, seemd that this is the only one causes (and solves) the failure:

TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local, NUM_LOCAL_VECS,
- remote, nr_iovecs, 0));
+ remote, nr_iovecs, 0UL));

If we have enough time we might want to do a bisect between glibc-2.17-157 and 196 to dig into the real problem behind the simple fix, but for now I'm agree to convert all int params to correct unsigned long changed by this commit, so that we can get rid of any potential failures.

Jason, can you make an updated patch to include Li Wang's and Jan's advices?

Thanks,
Caspar

Jason Xing

author

2019-02-26 01:42 UTC #18 lore raw

PATCH v2 1/1 syscalls/cma: fix the failure of compiling with O2 option

This issue is triggered by the commit 9b02cd465f70 which I wrote. If compile the process_vm_readv03.c with O2 option and run, it will return TFAIL. By casting the int type to unsigned long type in the ltp_syscall calling functions can we avoid this failure.

Signed-off-by Jason Xing <[email protected]>

---
 testcases/kernel/syscalls/cma/process_vm_readv03.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

testcases/kernel/syscalls/cma/process_vm_readv03.c +5 -4

index 45f7c92..f09dfb0 100644
@@ -181,7 +181,7 @@ static long *fetch_remote_addrs(void)
 	remote.iov_len = len;
 
 	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], &local,
-			 1, &remote, 1, 0));
+			 1UL, &remote, 1UL, 0UL));
 	if (TEST_RETURN != len)
 		tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");

@@ -213,8 +213,9 @@ static void child_invoke(int *bufsz_arr)
 	tst_resm(TINFO, "child 1: %d local iovecs initialized.",
 		 NUM_LOCAL_VECS);
 
-	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local, NUM_LOCAL_VECS,
-				   remote, nr_iovecs, 0));
+	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local,
+			    (unsigned long)NUM_LOCAL_VECS, remote,
+			    (unsigned long)nr_iovecs, 0UL));
 	if (TEST_RETURN != bufsz)
 		tst_brkm(TBROK | TERRNO, tst_exit, "process_vm_readv");

@@ -248,7 +249,7 @@ static void setup(void)
 	tst_require_root();
 
 	/* Just a sanity check of the existence of syscall */
-	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0, NULL, 0, 0);
+	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0UL, NULL, 0UL, 0UL);
 
 	nr_iovecs = nflag ? SAFE_STRTOL(NULL, nr_opt, 1, IOV_MAX) : 10;
 	bufsz = sflag ? SAFE_STRTOL(NULL, sz_opt, NUM_LOCAL_VECS, LONG_MAX)

Signature · 1 line

1.8.3.1

Jason Xing

author

2019-02-26 02:22 UTC #19 lore raw

PATCH v3 1/1 syscalls/cma: fix the failure of compiling with O2 option

Signed-off-by Jason Xing <[email protected]>

---
 testcases/kernel/syscalls/cma/process_vm_readv02.c  | 4 ++--
 testcases/kernel/syscalls/cma/process_vm_readv03.c  | 9 +++++----
 testcases/kernel/syscalls/cma/process_vm_writev02.c | 4 ++--
 3 files changed, 9 insertions(+), 8 deletions(-)

testcases/kernel/syscalls/cma/process_vm_readv02.c +2 -2

index e1fc32a..68e1bea 100644
@@ -135,7 +135,7 @@ static void child_invoke(void)
 
 	tst_resm(TINFO, "child 1: reading string from same memory location.");
 	TEST(ltp_syscall(__NR_process_vm_readv, pids[0],
-			 &local, 1, &remote, 1, 0));
+			 &local, 1UL, &remote, 1UL, 0UL));
 	if (TEST_RETURN != len)
 		tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");
 	if (strncmp(lp, tst_string, len) != 0)

@@ -150,7 +150,7 @@ static void setup(void)
 	tst_require_root();
 
 	/* Just a sanity check of the existence of syscall */
-	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0, NULL, 0, 0);
+	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0UL, NULL, 0UL, 0UL);
 
 	tst_tmpdir();
 	TST_CHECKPOINT_INIT(cleanup);

testcases/kernel/syscalls/cma/process_vm_readv03.c +5 -4

index 45f7c92..f09dfb0 100644
@@ -181,7 +181,7 @@ static long *fetch_remote_addrs(void)
 	remote.iov_len = len;
 
 	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], &local,
-			 1, &remote, 1, 0));
+			 1UL, &remote, 1UL, 0UL));
 	if (TEST_RETURN != len)
 		tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");

@@ -213,8 +213,9 @@ static void child_invoke(int *bufsz_arr)
 	tst_resm(TINFO, "child 1: %d local iovecs initialized.",
 		 NUM_LOCAL_VECS);
 
-	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local, NUM_LOCAL_VECS,
-				   remote, nr_iovecs, 0));
+	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local,
+			    (unsigned long)NUM_LOCAL_VECS, remote,
+			    (unsigned long)nr_iovecs, 0UL));
 	if (TEST_RETURN != bufsz)
 		tst_brkm(TBROK | TERRNO, tst_exit, "process_vm_readv");

@@ -248,7 +249,7 @@ static void setup(void)
 	tst_require_root();
 
 	/* Just a sanity check of the existence of syscall */
-	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0, NULL, 0, 0);
+	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0UL, NULL, 0UL, 0UL);
 
 	nr_iovecs = nflag ? SAFE_STRTOL(NULL, nr_opt, 1, IOV_MAX) : 10;
 	bufsz = sflag ? SAFE_STRTOL(NULL, sz_opt, NUM_LOCAL_VECS, LONG_MAX)

testcases/kernel/syscalls/cma/process_vm_writev02.c +2 -2

index b1edba5..2e7a8d6 100644
@@ -172,7 +172,7 @@ static void child_write(void)
 
 	tst_resm(TINFO, "child 2: write to the same memory location.");
 	TEST(ltp_syscall(__NR_process_vm_writev, pids[0], &local,
-			 1, &remote, 1, 0));
+			 1UL, &remote, 1UL, 0UL));
 	if (TEST_RETURN != bufsz)
 		tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");
 }

@@ -182,7 +182,7 @@ static void setup(void)
 	tst_require_root();
 
 	/* Just a sanity check of the existence of syscall */
-	ltp_syscall(__NR_process_vm_writev, getpid(), NULL, 0, NULL, 0, 0);
+	ltp_syscall(__NR_process_vm_writev, getpid(), NULL, 0UL, NULL, 0UL, 0UL);
 
 	bufsz =
 	    sflag ? SAFE_STRTOL(NULL, sz_opt, 1, LONG_MAX - PADDING_SIZE * 2)

Signature · 1 line

1.8.3.1

Li Wang

2019-02-26 03:17 UTC #20 lore raw

PATCH v3 1/1 syscalls/cma: fix the failure of compiling with O2 option

Quoted 8 lines · Jason Xing

This issue is triggered by the commit 9b02cd465f70 which I wrote. If compile the process_vm_readv03.c with O2 option and run, it will return TFAIL. By casting the int type to unsigned long type in the ltp_syscall calling functions can we avoid this failure. Thus I modify all the testcases which might be affected by the previous commit.
Signed-off-by Jason Xing <[email protected]>

Reviewed-by Li Wang <[email protected]>

Signature · 5 lines

Regards,
Li Wang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linux.it/pipermail/ltp/attachments/20190226/a7ad216d/attachment.html>

Jan Stancek

2019-02-26 16:03 UTC #21 lore raw

PATCH v3 1/1 syscalls/cma: fix the failure of compiling with O2 option

In reply to Jason Xing

----- Original Message -----

Quoted 7 lines

This issue is triggered by the commit 9b02cd465f70 which I wrote. If compile the process_vm_readv03.c with O2 option and run, it will return TFAIL. By casting the int type to unsigned long type in the ltp_syscall calling functions can we avoid this failure. Thus I modify all the testcases which might be affected by the previous commit.
Signed-off-by Jason Xing <[email protected]>

Pushed.

Thanks,
Jan

Jason Xing

author

2019-02-26 02:00 UTC #22 lore raw

PATCH 1/1 syscalls/cma: fix the failure of compiling with O2 option

In reply to Jan Stancek

Quoted 8 lines · Jan Stancek

----- Original Message -----
Quoted 5 lines
This issue is triggered by the commit 9b02cd465f70 which I wrote. If compile the process_vm_readv03.c with O2 option and run, it will return TFAIL. By casting the int type to long type in the ltp_syscall calling functions can we avoid this failure.

Quoted 6 lines

Quoted 3 lines
Signed-off-by Jason Xing <[email protected]>
---
 testcases/kernel/syscalls/cma/process_vm_readv03.c | 8 ++++----
What about other tests that were changed by commit 9b02cd465f70 ?

After testing all the process_vm_* cases with O0 and O2 options, we find that only process_vm_readv03 encounters this problem.

Quoted 41 lines

 1 file changed, 4 insertions(+), 4 deletions(-)

testcases/kernel/syscalls/cma/process_vm_readv03.c +4 -5

index 45f7c92..e15ffee 100644
@@ -181,7 +181,7 @@ static long *fetch_remote_addrs(void)
 	remote.iov_len = len;
 
 	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], &local,
-			 1, &remote, 1, 0));
+			 1L, &remote, 1L, 0L));
 	if (TEST_RETURN != len)
 		tst_brkm(TFAIL | TERRNO, tst_exit, "process_vm_readv");

@@ -213,8 +213,8 @@ static void child_invoke(int *bufsz_arr)
 	tst_resm(TINFO, "child 1: %d local iovecs initialized.",
 		 NUM_LOCAL_VECS);
 
-	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local, NUM_LOCAL_VECS,
-				   remote, nr_iovecs, 0));
+	TEST(ltp_syscall(__NR_process_vm_readv, pids[0], local,
 (long)NUM_LOCAL_VECS,
+				   remote, (long)nr_iovecs, 0L));
 	if (TEST_RETURN != bufsz)
 		tst_brkm(TBROK | TERRNO, tst_exit, "process_vm_readv");

@@ -248,7 +248,7 @@ static void setup(void)
 	tst_require_root();
 
 	/* Just a sanity check of the existence of syscall */
-	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0, NULL, 0, 0);
+	ltp_syscall(__NR_process_vm_readv, getpid(), NULL, 0L, NULL, 0L, 0L);
 
 	nr_iovecs = nflag ? SAFE_STRTOL(NULL, nr_opt, 1, IOV_MAX) : 10;
 	bufsz = sflag ? SAFE_STRTOL(NULL, sz_opt, NUM_LOCAL_VECS, LONG_MAX)
--

1.8.3.1