On Mon, Apr 11, 2022 at 3:15 PM David Matlack <dmatlack@xxxxxxxxxx> wrote:
>
> On Mon, Apr 11, 2022 at 2:10 PM Ben Gardon <bgardon@xxxxxxxxxx> wrote:
> >
> > Move the code to read the binary stats data to the KVM selftests
> > library. It will be re-used by other tests to check KVM behavior.
> >
> > No functional change intended.
> >
> > Signed-off-by: Ben Gardon <bgardon@xxxxxxxxxx>
> > ---
> > .../selftests/kvm/include/kvm_util_base.h | 3 +++
> > .../selftests/kvm/kvm_binary_stats_test.c | 20 +++++-------------
> > tools/testing/selftests/kvm/lib/kvm_util.c | 21 +++++++++++++++++++
> > 3 files changed, 29 insertions(+), 15 deletions(-)
> >
> > diff --git a/tools/testing/selftests/kvm/include/kvm_util_base.h b/tools/testing/selftests/kvm/include/kvm_util_base.h
> > index c5f34551ff76..b2684cfc2cb1 100644
> > --- a/tools/testing/selftests/kvm/include/kvm_util_base.h
> > +++ b/tools/testing/selftests/kvm/include/kvm_util_base.h
> > @@ -405,6 +405,9 @@ struct kvm_stats_desc *alloc_vm_stats_desc(int stats_fd,
> > struct kvm_stats_header *header);
> > void read_vm_stats_desc(int stats_fd, struct kvm_stats_header *header,
> > struct kvm_stats_desc *stats_desc);
> > +int read_stat_data(int stats_fd, struct kvm_stats_header *header,
> > + struct kvm_stats_desc *desc, uint64_t *data,
> > + ssize_t max_elements);
> >
> > uint32_t guest_get_vcpuid(void);
> >
> > diff --git a/tools/testing/selftests/kvm/kvm_binary_stats_test.c b/tools/testing/selftests/kvm/kvm_binary_stats_test.c
> > index e4795bad7db6..97b180249ba0 100644
> > --- a/tools/testing/selftests/kvm/kvm_binary_stats_test.c
> > +++ b/tools/testing/selftests/kvm/kvm_binary_stats_test.c
> > @@ -20,6 +20,8 @@
> > #include "asm/kvm.h"
> > #include "linux/kvm.h"
> >
> > +#define STAT_MAX_ELEMENTS 1000
> > +
> > static void stats_test(int stats_fd)
> > {
> > ssize_t ret;
> > @@ -29,7 +31,7 @@ static void stats_test(int stats_fd)
> > struct kvm_stats_header header;
> > char *id;
> > struct kvm_stats_desc *stats_desc;
> > - u64 *stats_data;
> > + u64 stats_data[STAT_MAX_ELEMENTS];
>
> What is the benefit of changing stats_data to a stack allocation with
> a fixed limit?
There isn't really a benefit. Will remove.
>
> > struct kvm_stats_desc *pdesc;
> >
> > /* Read kvm stats header */
> > @@ -130,25 +132,13 @@ static void stats_test(int stats_fd)
> > pdesc->offset, pdesc->name);
> > }
> >
> > - /* Allocate memory for stats data */
> > - stats_data = malloc(size_data);
> > - TEST_ASSERT(stats_data, "Allocate memory for stats data");
> > - /* Read kvm stats data as a bulk */
> > - ret = pread(stats_fd, stats_data, size_data, header.data_offset);
> > - TEST_ASSERT(ret == size_data, "Read KVM stats data");
> > /* Read kvm stats data one by one */
> > - size_data = 0;
> > for (i = 0; i < header.num_desc; ++i) {
> > pdesc = (void *)stats_desc + i * size_desc;
> > - ret = pread(stats_fd, stats_data,
> > - pdesc->size * sizeof(*stats_data),
> > - header.data_offset + size_data);
> > - TEST_ASSERT(ret == pdesc->size * sizeof(*stats_data),
> > - "Read data of KVM stats: %s", pdesc->name);
> > - size_data += pdesc->size * sizeof(*stats_data);
> > + read_stat_data(stats_fd, &header, pdesc, stats_data,
> > + ARRAY_SIZE(stats_data));
> > }
> >
> > - free(stats_data);
> > free(stats_desc);
> > free(id);
> > }
> > diff --git a/tools/testing/selftests/kvm/lib/kvm_util.c b/tools/testing/selftests/kvm/lib/kvm_util.c
> > index e3ae26fbef03..64e2085f1129 100644
> > --- a/tools/testing/selftests/kvm/lib/kvm_util.c
> > +++ b/tools/testing/selftests/kvm/lib/kvm_util.c
> > @@ -2593,3 +2593,24 @@ void read_vm_stats_desc(int stats_fd, struct kvm_stats_header *header,
> > TEST_ASSERT(ret == stats_descs_size(header),
> > "Read KVM stats descriptors");
> > }
> > +
> > +int read_stat_data(int stats_fd, struct kvm_stats_header *header,
>
> I would like to keep up the practice of adding docstrings to functions
> in kvm_util. Can you add docstring comments for this function and the
> other kvm_util functions introduced by this series?
Will do.
>
> > + struct kvm_stats_desc *desc, uint64_t *data,
> > + ssize_t max_elements)
> > +{
> > + ssize_t ret;
> > +
> > + TEST_ASSERT(desc->size <= max_elements,
> > + "Max data elements should be at least as large as stat data");
>
> What is the reason for this assertion? Callers are required to read
> all the data elements of a given stat?
Yeah, that was the idea, but it doesn't seem very useful. I'll remove it.
>
> > +
> > + ret = pread(stats_fd, data, desc->size * sizeof(*data),
> > + header->data_offset + desc->offset);
> > +
> > + /* ret from pread is in bytes. */
> > + ret = ret / sizeof(*data);
> > +
> > + TEST_ASSERT(ret == desc->size,
> > + "Read data of KVM stats: %s", desc->name);
>
> Won't this assertion fail when called from kvm_binary_stats_test.c?
> kvm_binary_stats_test.c looks like it reads all the stat data at once,
> which means ret will be the total number of stat data points, and
> desc->size will be the number of stat data points in the first stat.
Hmmm it shouldn't. I think we're just reading one stat at at time.
>
> > +
> > + return ret;
> > +}
> > --
> > 2.35.1.1178.g4f1659d476-goog
> >
