mbind01, get_mempolicy01: dont use hardcoded node 0

PATCH In review Jan Stancek 2 participants 2 messages May 25, 2012

lore

Share & tools

lore

Jan Stancek

author

2012-05-25 07:05 UTC #0 lore raw

PATCH v3 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 fails.

Signed-off-by Jan Stancek <[email protected]>

---
 .../syscalls/get_mempolicy/get_mempolicy01.c       |   22 ++++++++++++++++++-
 testcases/kernel/syscalls/mbind/mbind01.c          |   22 ++++++++++++++++++-
 2 files changed, 40 insertions(+), 4 deletions(-)

testcases/kernel/syscalls/get_mempolicy/get_mempolicy01.c +20 -2

index 0565d6f..d2de92b 100644
@@ -231,15 +231,33 @@ 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 +20 -2

index 4b25db0..cdc1721 100644
@@ -217,15 +217,33 @@ 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

Wanlong Gao

Acked-by on v3 1/1

2012-05-25 07:15 UTC #1 lore raw

Quoted 16 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

Quoted 6 lines

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 fails.

Thank you Jan,

Acked-by Wanlong Gao <[email protected]>

Quoted 19 lines

Signed-off-by Jan Stancek <[email protected]>
---
 .../syscalls/get_mempolicy/get_mempolicy01.c       |   22 ++++++++++++++++++-
 testcases/kernel/syscalls/mbind/mbind01.c          |   22 ++++++++++++++++++-
 2 files changed, 40 insertions(+), 4 deletions(-)
------------------------------------------------------------------------------
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 4 lines

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