Skip to contentSkip to conversation

KVM: selftests: access_tracking_perf_test: bump number of NUMA nodes to 32

PATCH New Maxim Levitsky 2 participants 2 messages Jun 12

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 KVM: selftests: access_tracking_perf_test: bump number of NUMA nodes to 32

It's rare to find a system that has more than 4 sockets, but a system can have more than 4 NUMA nodes if each socket exposes its chiplets as separate NUMA nodes.

In particular, our CI caught a failure in this test on a system with two sockets, each containing an 'AMD EPYC 7601 32-Core Processor'.

Bump the limit to 32, just in case.

---
 tools/testing/selftests/kvm/include/lru_gen_util.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/testing/selftests/kvm/include/lru_gen_util.h +1 -1
index d32ff5d8ffd0..49c8139d398c 100644
@@ -14,7 +14,7 @@
 #include "test_util.h"
 
 #define MAX_NR_GENS 16 /* MAX_NR_GENS in include/linux/mmzone.h */
-#define MAX_NR_NODES 4 /* Maximum number of nodes supported by the test */
+#define MAX_NR_NODES 32 /* Maximum number of nodes supported by the test */
 
 #define LRU_GEN_DEBUGFS "/sys/kernel/debug/lru_gen"
 #define LRU_GEN_ENABLED_PATH "/sys/kernel/mm/lru_gen/enabled"
Signature · 1 line
2.49.0
#1 lore raw
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [Medium] Incorrect type for asprintf return value defeats error checking and causes undefined behavior/crash if allocation fails. --

commit c131ebf5598f598f5bdd7d212b93b7da852ad82f
Author: Maxim Levitsky <[email protected]>

KVM: selftests: access_tracking_perf_test: bump number of NUMA nodes to 32

It's rare to find a system that has more than 4 sockets, but a system can have more than 4 NUMA nodes if each socket exposes its chiplets as separate NUMA nodes. Bump the limit to 32, just in case.

tools/testing/selftests/kvm/include/lru_gen_util.h line 17

Quoted 10 lines
tools/testing/selftests/kvm/include/lru_gen_util.h +1 -1
index d32ff5d8ffd05..49c8139d398ca 100644
@@ -14,7 +14,7 @@
 #include "test_util.h"
 
 #define MAX_NR_GENS 16 /* MAX_NR_GENS in include/linux/mmzone.h */
-#define MAX_NR_NODES 4 /* Maximum number of nodes supported by the test */
+#define MAX_NR_NODES 32 /* Maximum number of nodes supported by the test */

[Severity: Medium]
This is a pre-existing issue, but does run_aging_impl() in tools/testing/selftests/kvm/lib/lru_gen_util.c mishandle the return value of asprintf() when creating the aging command?

tools/testing/selftests/kvm/lib/lru_gen_util.c:run_aging_impl() {
    ...
	FILE *f = fopen(LRU_GEN_DEBUGFS, "w");
	char *command;
	size_t sz;

	TEST_ASSERT(f, "fopen(%s) failed", LRU_GEN_DEBUGFS);
	sz = asprintf(&command, "+ %lu %d %d 1 %d\n",
		      memcg_id, node_id, max_gen, force_scan);
	TEST_ASSERT(sz > 0, "creating aging command failed");
    ...
}

asprintf() returns an int which is -1 on failure. Because it is assigned to the size_t sz variable, -1 wraps around to SIZE_MAX. The error check TEST_ASSERT(sz > 0, ...) will incorrectly pass, and the code will then call fwrite() with an undefined command pointer and SIZE_MAX size, resulting in a segmentation fault rather than a graceful test failure.

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1