Re: [PATCH v4 2/2] selftests: kvm: Add mem_slot_test test

From: Wainer dos Santos Moschetta
Date: Wed Apr 08 2020 - 23:01:47 EST


On 4/8/20 10:31 PM, Krish Sadhukhan wrote:

On 4/8/20 3:08 PM, Wainer dos Santos Moschetta wrote:
This patch introduces the mem_slot_test test which checks
an VM can have added memory slots up to the limit defined in
KVM_CAP_NR_MEMSLOTS. Then attempt to add one more slot to
verify it fails as expected.

Signed-off-by: Wainer dos Santos Moschetta <wainersm@xxxxxxxxxx>
---
 tools/testing/selftests/kvm/.gitignore | 1 +
 tools/testing/selftests/kvm/Makefile | 3 +
 tools/testing/selftests/kvm/mem_slot_test.c | 76 +++++++++++++++++++++
 3 files changed, 80 insertions(+)
 create mode 100644 tools/testing/selftests/kvm/mem_slot_test.c

diff --git a/tools/testing/selftests/kvm/.gitignore b/tools/testing/selftests/kvm/.gitignore
index 16877c3daabf..127d27188427 100644
--- a/tools/testing/selftests/kvm/.gitignore
+++ b/tools/testing/selftests/kvm/.gitignore
@@ -21,4 +21,5 @@
 /demand_paging_test
 /dirty_log_test
 /kvm_create_max_vcpus
+/mem_slot_test
 /steal_time
diff --git a/tools/testing/selftests/kvm/Makefile b/tools/testing/selftests/kvm/Makefile
index 712a2ddd2a27..338b6cdce1a0 100644
--- a/tools/testing/selftests/kvm/Makefile
+++ b/tools/testing/selftests/kvm/Makefile
@@ -32,12 +32,14 @@ TEST_GEN_PROGS_x86_64 += clear_dirty_log_test
 TEST_GEN_PROGS_x86_64 += demand_paging_test
 TEST_GEN_PROGS_x86_64 += dirty_log_test
 TEST_GEN_PROGS_x86_64 += kvm_create_max_vcpus
+TEST_GEN_PROGS_x86_64 += mem_slot_test
 TEST_GEN_PROGS_x86_64 += steal_time
  TEST_GEN_PROGS_aarch64 += clear_dirty_log_test
 TEST_GEN_PROGS_aarch64 += demand_paging_test
 TEST_GEN_PROGS_aarch64 += dirty_log_test
 TEST_GEN_PROGS_aarch64 += kvm_create_max_vcpus
+TEST_GEN_PROGS_aarch64 += mem_slot_test
 TEST_GEN_PROGS_aarch64 += steal_time
  TEST_GEN_PROGS_s390x = s390x/memop
@@ -46,6 +48,7 @@ TEST_GEN_PROGS_s390x += s390x/sync_regs_test
 TEST_GEN_PROGS_s390x += demand_paging_test
 TEST_GEN_PROGS_s390x += dirty_log_test
 TEST_GEN_PROGS_s390x += kvm_create_max_vcpus
+TEST_GEN_PROGS_s390x += mem_slot_test
  TEST_GEN_PROGS += $(TEST_GEN_PROGS_$(UNAME_M))
 LIBKVM += $(LIBKVM_$(UNAME_M))
diff --git a/tools/testing/selftests/kvm/mem_slot_test.c b/tools/testing/selftests/kvm/mem_slot_test.c
new file mode 100644
index 000000000000..7c1009f0bc07
--- /dev/null
+++ b/tools/testing/selftests/kvm/mem_slot_test.c
@@ -0,0 +1,76 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * mem_slot_test
+ *
+ * Copyright (C) 2020, Red Hat, Inc.
+ *
+ * Test suite for memory region operations.
+ */
+#define _GNU_SOURCE /* for program_invocation_short_name */
+#include <linux/kvm.h>
+#include <sys/mman.h>
+
+#include "test_util.h"
+#include "kvm_util.h"
+
+/*
+ * Test it can be added memory slots up to KVM_CAP_NR_MEMSLOTS, then any
+ * tentative to add further slots should fail.
+ */
+static void test_add_max_slots(void)
+{
+ÂÂÂ int ret;
+ÂÂÂ struct kvm_vm *vm;
+ÂÂÂ uint32_t max_mem_slots;
+ÂÂÂ uint32_t slot;
+ÂÂÂ uint64_t guest_addr;
+ÂÂÂ uint64_t mem_reg_npages;
+ÂÂÂ uint64_t mem_reg_size;
+ÂÂÂ void *mem;
+
+ÂÂÂ max_mem_slots = kvm_check_cap(KVM_CAP_NR_MEMSLOTS);
+ÂÂÂ TEST_ASSERT(max_mem_slots > 0,
+ÂÂÂÂÂÂÂÂÂÂÂ "KVM_CAP_NR_MEMSLOTS should be greater than 0");
+ÂÂÂ pr_info("Allowed number of memory slots: %i\n", max_mem_slots);
+
+ÂÂÂ vm = vm_create(VM_MODE_DEFAULT, 0, O_RDWR);
+
+ÂÂÂ /*
+ÂÂÂÂ * Uses 1MB sized/aligned memory region since this is the minimal
+ÂÂÂÂ * required on s390x.
+ÂÂÂÂ */
+ÂÂÂ mem_reg_size = 0x100000;
+ÂÂÂ mem_reg_npages = vm_calc_num_guest_pages(VM_MODE_DEFAULT, mem_reg_size);
+
+ÂÂÂ guest_addr = 0x0;


Nit: Can't this be initialized where it's defined above ?


I don't have a strong preference. Is it generally initialized on definition on kvm (selftests or not) code?




+
+ÂÂÂ /* Check it can be added memory slots up to the maximum allowed */
+ÂÂÂ pr_info("Adding slots 0..%i, each memory region with %ldK size\n",
+ÂÂÂÂÂÂÂ (max_mem_slots - 1), mem_reg_size >> 10);
+ÂÂÂ for (slot = 0; slot < max_mem_slots; slot++) {
+ÂÂÂÂÂÂÂ vm_userspace_mem_region_add(vm, VM_MEM_SRC_ANONYMOUS,
+ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ guest_addr, slot, mem_reg_npages,
+ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ 0);
+ÂÂÂÂÂÂÂ guest_addr += mem_reg_size;
+ÂÂÂ }
+
+ÂÂÂ /* Check it cannot be added memory slots beyond the limit */
+ÂÂÂ mem = mmap(NULL, mem_reg_size, PROT_READ | PROT_WRITE,
+ÂÂÂÂÂÂÂÂÂÂ MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+ÂÂÂ TEST_ASSERT(mem != MAP_FAILED, "Failed to mmap() host");
+
+ÂÂÂ ret = ioctl(vm_get_fd(vm), KVM_SET_USER_MEMORY_REGION,
+ÂÂÂÂÂÂÂÂÂÂÂ &(struct kvm_userspace_memory_region) {slot, 0, guest_addr,
+ÂÂÂÂÂÂÂÂÂÂÂ mem_reg_size, (uint64_t) mem});
+ÂÂÂ TEST_ASSERT(ret == -1 && errno == EINVAL,
+ÂÂÂÂÂÂÂÂÂÂÂ "Adding one more memory slot should fail with EINVAL");


Why not add a test here for adding memory at an existing slot ?

Good question.

I'm working on another test which should check it cannot be added overlapping slots. I will send it in a separate patch series, depending on the fate of this one. :)

More precisely, those are the cases I will cover on this new test:


ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ 0x100000Â 0x300000
ÂÂÂÂÂÂÂÂÂ 0x0ÂÂÂÂÂÂÂÂ 0x200000 0x400000
Âslot0 |ÂÂÂÂÂÂÂÂÂÂÂÂÂ |---2MB--|ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ (SUCCESS)
Âslot1ÂÂÂÂÂÂ |---2MB--|ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ (FAIL)
Âslot2 |---2MB--|ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ (SUCCESS)
Âslot3ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ |---2MB--|ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ (FAIL)
Âslot4ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ |---2MB--|ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ (FAIL)
Âslot5ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ |---2MB--|ÂÂÂÂÂÂÂÂÂ (SUCCESS)

Thanks!

Wainer



+
+ÂÂÂ munmap(mem, mem_reg_size);
+ÂÂÂ kvm_vm_free(vm);
+}
+
+int main(int argc, char *argv[])
+{
+ÂÂÂ test_add_max_slots();
+ÂÂÂ return 0;
+}