Re: [PATCH 1/5] KVM: selftests: allow different backing memory types for demand paging

On Thu, May 13, 2021 at 2:32 PM Ben Gardon bgardon@google.com wrote:

On Wed, May 12, 2021 at 2:45 PM Axel Rasmussen axelrasmussen@google.com wrote:

...

Add an argument which lets us specify a different backing memory type for the test. The default is just to use anonymous, matching existing behavior (if the argument is omitted).

This is in preparation for testing UFFD minor faults. For that, we need to use a new backing memory type which is setup with MAP_SHARED.

This notably requires one other change. Perhaps counter-intuitively, perf_test_args.host_page_size is the host's *native* page size, not the size of the pages the host is using to back the guest. This means, if we try to run the test with e.g. VM_MEM_SRC_ANONYMOUS_HUGETLB, we'll try to do demand paging with 4k pages instead of 2M hugepages.

Would it make sense to factor this change out into another commit preceding this one? Perhaps only worth it if you send a v2.

When you say "we'll try to do demand paging with 4k pages instead of 2M hugepages," what would that mean? Would we only copy 4k worth of the contents of the 2M page in, leading to the guest seeing bad memory? Do we have the capability to do demand paging at a smaller granularity than the backing page size with UFFD?

Basically I think the existing behavior is to always use 4k - so we UFFDIO_COPY that size at a time.

This is totally fine for anonymous, or even for THPs. But, once we're using hugetlbfs, the UFFDIO_COPY ioctl will just fail unless we do 2M (or some multiple thereof) at a time. If memory serves it returns -EINVAL in this case.

It wouldn't be so much trouble to factor it out, so just let me know what is preferred. For now, I'll do it if a v2 is needed for other reasons.

Otherwise this patch looks reasonable to me. I'll try to review the rest of your patches today / Monday.

...

So, convert everything to use a new demand_paging_size, computed based on the backing memory type.

Signed-off-by: Axel Rasmussen axelrasmussen@google.com

.../selftests/kvm/demand_paging_test.c | 24 +++++++++++++------ 1 file changed, 17 insertions(+), 7 deletions(-)

diff --git a/tools/testing/selftests/kvm/demand_paging_test.c b/tools/testing/selftests/kvm/demand_paging_test.c index 5f7a229c3af1..10c7ba76a9c6 100644 --- a/tools/testing/selftests/kvm/demand_paging_test.c +++ b/tools/testing/selftests/kvm/demand_paging_test.c @@ -38,6 +38,7 @@

static int nr_vcpus = 1; static uint64_t guest_percpu_mem_size = DEFAULT_PER_VCPU_MEM_SIZE; +static size_t demand_paging_size; static char *guest_data_prototype;

static void *vcpu_worker(void *data) @@ -83,7 +84,7 @@ static int handle_uffd_page_request(int uffd, uint64_t addr)

    copy.src = (uint64_t)guest_data_prototype;
    copy.dst = addr;

•   copy.len = perf_test_args.host_page_size;
  

•   copy.len = demand_paging_size;
    copy.mode = 0;
  
    clock_gettime(CLOCK_MONOTONIC, &start);
  

@@ -100,7 +101,7 @@ static int handle_uffd_page_request(int uffd, uint64_t addr) PER_PAGE_DEBUG("UFFDIO_COPY %d \t%ld ns\n", tid, timespec_to_ns(ts_diff)); PER_PAGE_DEBUG("Paged in %ld bytes at 0x%lx from thread %d\n",

•                  perf_test_args.host_page_size, addr, tid);
  

•                  demand_paging_size, addr, tid);
  
    return 0;
  

} @@ -250,6 +251,7 @@ static int setup_demand_paging(struct kvm_vm *vm, struct test_params { bool use_uffd; useconds_t uffd_delay;

•   enum vm_mem_backing_src_type src_type;
    bool partition_vcpu_memory_access;
  

};

@@ -267,14 +269,16 @@ static void run_test(enum vm_guest_mode mode, void *arg) int r;

    vm = perf_test_create_vm(mode, nr_vcpus, guest_percpu_mem_size,

•                            VM_MEM_SRC_ANONYMOUS);
  

•                            p->src_type);
  
    perf_test_args.wr_fract = 1;
  
  

•   guest_data_prototype = malloc(perf_test_args.host_page_size);
  

•   demand_paging_size = get_backing_src_pagesz(p->src_type);
  

•   guest_data_prototype = malloc(demand_paging_size);
    TEST_ASSERT(guest_data_prototype,
                "Failed to allocate buffer for guest data pattern");
  

•   memset(guest_data_prototype, 0xAB, perf_test_args.host_page_size);
  

•   memset(guest_data_prototype, 0xAB, demand_paging_size);
  
    vcpu_threads = malloc(nr_vcpus * sizeof(*vcpu_threads));
    TEST_ASSERT(vcpu_threads, "Memory allocation failed");
  

@@ -388,7 +392,7 @@ static void help(char *name) { puts(""); printf("usage: %s [-h] [-m mode] [-u] [-d uffd_delay_usec]\n"

•          "          [-b memory] [-v vcpus] [-o]\n", name);
  

•          "          [-b memory] [-t type] [-v vcpus] [-o]\n", name);
    guest_modes_help();
    printf(" -u: use User Fault FD to handle vCPU page\n"
           "     faults.\n");
  

@@ -398,6 +402,8 @@ static void help(char *name) printf(" -b: specify the size of the memory region which should be\n" " demand paged by each vCPU. e.g. 10M or 3G.\n" " Default: 1G\n");

•   printf(" -t: The type of backing memory to use. Default: anonymous\n");
  

•   backing_src_help();
    printf(" -v: specify the number of vCPUs to run.\n");
    printf(" -o: Overlap guest memory accesses instead of partitioning\n"
           "     them into a separate region of memory for each vCPU.\n");
  

@@ -409,13 +415,14 @@ int main(int argc, char *argv[]) { int max_vcpus = kvm_check_cap(KVM_CAP_MAX_VCPUS); struct test_params p = {

•           .src_type = VM_MEM_SRC_ANONYMOUS,
            .partition_vcpu_memory_access = true,
    };
    int opt;
  
    guest_modes_append_default();
  
  

•   while ((opt = getopt(argc, argv, "hm:ud:b:v:o")) != -1) {
  

•   while ((opt = getopt(argc, argv, "hm:ud:b:t:v:o")) != -1) {
            switch (opt) {
            case 'm':
                    guest_modes_cmdline(optarg);
  

@@ -430,6 +437,9 @@ int main(int argc, char *argv[]) case 'b': guest_percpu_mem_size = parse_size(optarg); break;

•           case 't':
  

•                   p.src_type = parse_backing_src_type(optarg);
  

•                   break;
            case 'v':
                    nr_vcpus = atoi(optarg);
                    TEST_ASSERT(nr_vcpus > 0 && nr_vcpus <= max_vcpus,
  

-- 2.31.1.607.g51e8a6a459-goog