Skip to contentSkip to conversation

mbind01, get_mempolicy01: dont use hardcoded node 0

PATCH Superseded Jan Stancek 2 participants 2 messages May 24, 2012 – May 25, 2012

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Jan Stancek

author
#0 lore raw

PATCH v2 1/1 mbind01, get_mempolicy01: dont use hardcoded node 0

These tests are failing on following setup:

  # numactl -H
  available: 2 nodes (0-1)
  node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
  23 24 25 26 27 28 29 30 31
  node 0 size: 0 MB
  node 0 free: 0 MB
  node 1 cpus:
  node 1 size: 12288 MB
  node 1 free: 9998 MB
  node distances:
  node   0   1
    0:  10  40
    1:  40  10

Avoid hardcoded node 0 by picking first available node obtained using get_mempolicy(..., MPOL_F_MEMS_ALLOWED);

Using whole nodemask doesn't work with MPOL_PREFERRED, because set_mempolicy will use only first node and subsequent check for bitmask equality may fail.

Signed-off-by Jan Stancek <[email protected]>
---
 .../syscalls/get_mempolicy/get_mempolicy01.c       |   19 +++++++++++++++++--
 testcases/kernel/syscalls/mbind/mbind01.c          |   19 +++++++++++++++++--
 2 files changed, 34 insertions(+), 4 deletions(-)
testcases/kernel/syscalls/get_mempolicy/get_mempolicy01.c +17 -2
index 0565d6f..8069df1 100644
@@ -231,15 +231,30 @@ static int do_test(struct test_case *tc)
 #endif
 	char *p = NULL;
 	unsigned long len = MEM_LENGTH;
+	struct bitmask *allowed = NULL;
+	int i, test_node = -1;
+
+	allowed = numa_allocate_nodemask();
+	// avoid numa_get_mems_allowed(), because of bug in getpol() utility function
+	// in older versions: http://www.spinics.net/lists/linux-numa/msg00849.html
+	TEST(syscall(__NR_get_mempolicy, NULL, allowed->maskp,
+		allowed->size, 0, MPOL_F_MEMS_ALLOWED));
+	if (TEST_RETURN < 0)
+		tst_resm(TFAIL|TERRNO, "get_mempolicy: MPOL_F_MEMS_ALLOWED");
+	for (i = 0; i < allowed->size; i++)
+		if (numa_bitmask_isbitset(allowed, i)) {
+			test_node = i;
+			break;
+		}
 
 #if !defined(LIBNUMA_API_VERSION) || LIBNUMA_API_VERSION < 2
 	nodemask = malloc(sizeof(nodemask_t));
 	nodemask_zero(nodemask);
-	nodemask_set(nodemask, 0);
+	nodemask_set(nodemask, test_node);
 	getnodemask = malloc(sizeof(nodemask_t));
 	nodemask_zero(getnodemask);
 #else
-	numa_bitmask_setbit(nodemask, 0);
+	numa_bitmask_setbit(nodemask, test_node);
 #endif
 	switch (tc->ttype) {
 	case DEFAULT:
testcases/kernel/syscalls/mbind/mbind01.c +17 -2
index 4b25db0..c3170ec 100644
@@ -217,15 +217,30 @@ static int do_test(struct test_case *tc)
 	unsigned long maxnode = NUMA_NUM_NODES;
 	unsigned long len = MEM_LENGTH;
 	unsigned long *invalid_nodemask;
+	struct bitmask *allowed = NULL;
+	int i, test_node = -1;
+
+	allowed = numa_allocate_nodemask();
+	// avoid numa_get_mems_allowed(), because of bug in getpol() utility function
+	// in older versions: http://www.spinics.net/lists/linux-numa/msg00849.html
+	TEST(syscall(__NR_get_mempolicy, NULL, allowed->maskp,
+		allowed->size, 0, MPOL_F_MEMS_ALLOWED));
+	if (TEST_RETURN < 0)
+		tst_resm(TFAIL|TERRNO, "get_mempolicy: MPOL_F_MEMS_ALLOWED");
+	for (i = 0; i < allowed->size; i++)
+		if (numa_bitmask_isbitset(allowed, i)) {
+			test_node = i;
+			break;
+		}
 
 #if !defined(LIBNUMA_API_VERSION) || LIBNUMA_API_VERSION < 2
 	nodemask = malloc(sizeof(nodemask_t));
 	nodemask_zero(nodemask);
-	nodemask_set(nodemask, 0);
+	nodemask_set(nodemask, test_node);
 	getnodemask = malloc(sizeof(nodemask_t));
 	nodemask_zero(getnodemask);
 #else
-	numa_bitmask_setbit(nodemask, 0);
+	numa_bitmask_setbit(nodemask, test_node);
 #endif
 	p = mmap(NULL, len, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS,
 		    0, 0);

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/

_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

#1 lore raw
Quoted 38 lines · Jan Stancek

These tests are failing on following setup:

  # numactl -H
  available: 2 nodes (0-1)
  node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
  23 24 25 26 27 28 29 30 31
  node 0 size: 0 MB
  node 0 free: 0 MB
  node 1 cpus:
  node 1 size: 12288 MB
  node 1 free: 9998 MB
  node distances:
  node   0   1
    0:  10  40
    1:  40  10

Avoid hardcoded node 0 by picking first available node obtained using get_mempolicy(..., MPOL_F_MEMS_ALLOWED);

Using whole nodemask doesn't work with MPOL_PREFERRED, because set_mempolicy will use only first node and subsequent check for bitmask equality may fail.

Signed-off-by Jan Stancek <[email protected]>
---
 .../syscalls/get_mempolicy/get_mempolicy01.c       |   19 +++++++++++++++++--
 testcases/kernel/syscalls/mbind/mbind01.c          |   19 +++++++++++++++++--
 2 files changed, 34 insertions(+), 4 deletions(-)

0001-mbind01-get_mempolicy01-dont-use-hardcoded-node-0.patch

testcases/kernel/syscalls/get_mempolicy/get_mempolicy01.c +0 -0
index 0565d6f..8069df1 100644

testcases/kernel/syscalls/get_mempolicy/get_mempolicy01.c static int do_test(struct test_case *tc) · line 239

Quoted 9 lines
testcases/kernel/syscalls/get_mempolicy/get_mempolicy01.c +6 -0
@@ -231,15 +231,30 @@ static int do_test(struct test_case *tc)
 #endif
 	char *p = NULL;
 	unsigned long len = MEM_LENGTH;
+	struct bitmask *allowed = NULL;
+	int i, test_node = -1;
+
+	allowed = numa_allocate_nodemask();
+	// avoid numa_get_mems_allowed(), because of bug in getpol() utility function
+	// in older versions: http://www.spinics.net/lists/linux-numa/msg00849.html

OK, thanks for point out this, but here, please use comment style like /*
* xxx
*/

Quoted 28 lines
testcases/kernel/syscalls/get_mempolicy/get_mempolicy01.c +11 -2
+	TEST(syscall(__NR_get_mempolicy, NULL, allowed->maskp,
+		allowed->size, 0, MPOL_F_MEMS_ALLOWED));
+	if (TEST_RETURN < 0)
+		tst_resm(TFAIL|TERRNO, "get_mempolicy: MPOL_F_MEMS_ALLOWED");
+	for (i = 0; i < allowed->size; i++)
+		if (numa_bitmask_isbitset(allowed, i)) {
+			test_node = i;
+			break;
+		}
 
 #if !defined(LIBNUMA_API_VERSION) || LIBNUMA_API_VERSION < 2
 	nodemask = malloc(sizeof(nodemask_t));
 	nodemask_zero(nodemask);
-	nodemask_set(nodemask, 0);
+	nodemask_set(nodemask, test_node);
 	getnodemask = malloc(sizeof(nodemask_t));
 	nodemask_zero(getnodemask);
 #else
-	numa_bitmask_setbit(nodemask, 0);
+	numa_bitmask_setbit(nodemask, test_node);
 #endif
 	switch (tc->ttype) {
 	case DEFAULT:
testcases/kernel/syscalls/mbind/mbind01.c +0 -0
index 4b25db0..c3170ec 100644

testcases/kernel/syscalls/mbind/mbind01.c static int do_test(struct test_case *tc) · line 225

Quoted 9 lines
testcases/kernel/syscalls/mbind/mbind01.c +6 -0
@@ -217,15 +217,30 @@ static int do_test(struct test_case *tc)
 	unsigned long maxnode = NUMA_NUM_NODES;
 	unsigned long len = MEM_LENGTH;
 	unsigned long *invalid_nodemask;
+	struct bitmask *allowed = NULL;
+	int i, test_node = -1;
+
+	allowed = numa_allocate_nodemask();
+	// avoid numa_get_mems_allowed(), because of bug in getpol() utility function
+	// in older versions: http://www.spinics.net/lists/linux-numa/msg00849.html

And ditto

Thanks,
Wanlong Gao

Quoted 36 lines
testcases/kernel/syscalls/mbind/mbind01.c +11 -3
+	TEST(syscall(__NR_get_mempolicy, NULL, allowed->maskp,
+		allowed->size, 0, MPOL_F_MEMS_ALLOWED));
+	if (TEST_RETURN < 0)
+		tst_resm(TFAIL|TERRNO, "get_mempolicy: MPOL_F_MEMS_ALLOWED");
+	for (i = 0; i < allowed->size; i++)
+		if (numa_bitmask_isbitset(allowed, i)) {
+			test_node = i;
+			break;
+		}
 
 #if !defined(LIBNUMA_API_VERSION) || LIBNUMA_API_VERSION < 2
 	nodemask = malloc(sizeof(nodemask_t));
 	nodemask_zero(nodemask);
-	nodemask_set(nodemask, 0);
+	nodemask_set(nodemask, test_node);
 	getnodemask = malloc(sizeof(nodemask_t));
 	nodemask_zero(getnodemask);
 #else
-	numa_bitmask_setbit(nodemask, 0);
+	numa_bitmask_setbit(nodemask, test_node);
 #endif
 	p = mmap(NULL, len, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS,
 		    0, 0);
 
 
 
 
------------------------------------------------------------------------------

Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/

Quoted 5 lines

_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list