| Message ID | 20210608233816.423958-1-zhenzhong.duan@intel.com |
|---|---|
| Headers | show |
| Series | Restore extra_mem_pages and add slot0_mem_pages | expand |
On Wed, Jun 09, 2021 at 07:38:16AM +0800, Zhenzhong Duan wrote: > Current implementatiion only create VM with slot0 memory size > of fixed 512 pages plus page tables and per-cpu memory. > > Add a new parameter slot0_mem_pages to vm_create_with_vcpus() and > some comments to clarify the meaning of slot0_mem_pages and > extra_mem_pages. > > Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com> > --- > .../testing/selftests/kvm/include/kvm_util.h | 7 +-- > .../selftests/kvm/kvm_page_table_test.c | 2 +- > tools/testing/selftests/kvm/lib/kvm_util.c | 47 +++++++++++++++---- > .../selftests/kvm/lib/perf_test_util.c | 2 +- > 4 files changed, 44 insertions(+), 14 deletions(-) > > diff --git a/tools/testing/selftests/kvm/include/kvm_util.h b/tools/testing/selftests/kvm/include/kvm_util.h > index fcd8e3855111..8bcadfa8cea5 100644 > --- a/tools/testing/selftests/kvm/include/kvm_util.h > +++ b/tools/testing/selftests/kvm/include/kvm_util.h > @@ -285,10 +285,11 @@ struct kvm_vm *vm_create_default_with_vcpus(uint32_t nr_vcpus, uint64_t extra_me > uint32_t num_percpu_pages, void *guest_code, > uint32_t vcpuids[]); > > -/* Like vm_create_default_with_vcpus, but accepts mode as a parameter */ > +/* Like vm_create_default_with_vcpus, but accepts mode and slot0 memory as a parameter */ > struct kvm_vm *vm_create_with_vcpus(enum vm_guest_mode mode, uint32_t nr_vcpus, > - uint64_t extra_mem_pages, uint32_t num_percpu_pages, > - void *guest_code, uint32_t vcpuids[]); > + uint64_t slot0_mem_pages, uint64_t extra_mem_pages, > + uint32_t num_percpu_pages, void *guest_code, > + uint32_t vcpuids[]); > > /* > * Adds a vCPU with reasonable defaults (e.g. a stack) > diff --git a/tools/testing/selftests/kvm/kvm_page_table_test.c b/tools/testing/selftests/kvm/kvm_page_table_test.c > index 1c4753fff19e..82171f17c1d7 100644 > --- a/tools/testing/selftests/kvm/kvm_page_table_test.c > +++ b/tools/testing/selftests/kvm/kvm_page_table_test.c > @@ -268,7 +268,7 @@ static struct kvm_vm *pre_init_before_test(enum vm_guest_mode mode, void *arg) > > /* Create a VM with enough guest pages */ > guest_num_pages = test_mem_size / guest_page_size; > - vm = vm_create_with_vcpus(mode, nr_vcpus, > + vm = vm_create_with_vcpus(mode, nr_vcpus, DEFAULT_GUEST_PHY_PAGES, > guest_num_pages, 0, guest_code, NULL); > > /* Align down GPA of the testing memslot */ > diff --git a/tools/testing/selftests/kvm/lib/kvm_util.c b/tools/testing/selftests/kvm/lib/kvm_util.c > index 63418df921f0..d86422a119fd 100644 > --- a/tools/testing/selftests/kvm/lib/kvm_util.c > +++ b/tools/testing/selftests/kvm/lib/kvm_util.c > @@ -308,21 +308,50 @@ struct kvm_vm *vm_create(enum vm_guest_mode mode, uint64_t phy_pages, int perm) > return vm; > } > > +/* > + * VM Create with customized parameters > + * > + * Input Args: > + * mode - VM Mode (e.g. VM_MODE_P52V48_4K) > + * nr_vcpus - VCPU count > + * slot0_mem_pages - Slot0 physical memory size > + * extra_mem_pages - Non-slot0 physical memory total size > + * num_percpu_pages - Per-cpu physical memory pages > + * guest_code - Guest entry point > + * vcpuids - VCPU IDs > + * > + * Output Args: None > + * > + * Return: > + * Pointer to opaque structure that describes the created VM. > + * > + * Creates a VM with the mode specified by mode (e.g. VM_MODE_P52V48_4K), > + * with customized slot0 memory size, at least 512 pages currently. > + * extra_mem_pages is only used to calculate the maximum page table size, > + * no real memory allocation for non-slot0 memory in this function. > + */ > struct kvm_vm *vm_create_with_vcpus(enum vm_guest_mode mode, uint32_t nr_vcpus, > - uint64_t extra_mem_pages, uint32_t num_percpu_pages, > - void *guest_code, uint32_t vcpuids[]) > + uint64_t slot0_mem_pages, uint64_t extra_mem_pages, > + uint32_t num_percpu_pages, void *guest_code, > + uint32_t vcpuids[]) > { > + uint64_t vcpu_pages, extra_pg_pages, pages; > + struct kvm_vm *vm; > + int i; > + > + /* Force slot0 memory size not small than DEFAULT_GUEST_PHY_PAGES */ > + if (slot0_mem_pages < DEFAULT_GUEST_PHY_PAGES) > + slot0_mem_pages = DEFAULT_GUEST_PHY_PAGES; > + > /* The maximum page table size for a memory region will be when the > * smallest pages are used. Considering each page contains x page > * table descriptors, the total extra size for page tables (for extra > * N pages) will be: N/x+N/x^2+N/x^3+... which is definitely smaller > * than N/x*2. > */ > - uint64_t vcpu_pages = (DEFAULT_STACK_PGS + num_percpu_pages) * nr_vcpus; > - uint64_t extra_pg_pages = (extra_mem_pages + vcpu_pages) / PTES_PER_MIN_PAGE * 2; > - uint64_t pages = DEFAULT_GUEST_PHY_PAGES + vcpu_pages + extra_pg_pages; > - struct kvm_vm *vm; > - int i; > + vcpu_pages = (DEFAULT_STACK_PGS + num_percpu_pages) * nr_vcpus; > + extra_pg_pages = (slot0_mem_pages + extra_mem_pages + vcpu_pages) / PTES_PER_MIN_PAGE * 2; We weren't including the slot0 memory (which was DEFAULT_GUEST_PHY_PAGES) in the extra_pg_pages calculation before, since they weren't considered "extra". It's probably a good idea to do this calculation for the caller, but it's making a change not called out in the commit message. > + pages = slot0_mem_pages + vcpu_pages + extra_pg_pages; > > TEST_ASSERT(nr_vcpus <= kvm_check_cap(KVM_CAP_MAX_VCPUS), > "nr_vcpus = %d too large for host, max-vcpus = %d", > @@ -354,8 +383,8 @@ struct kvm_vm *vm_create_default_with_vcpus(uint32_t nr_vcpus, uint64_t extra_me > uint32_t num_percpu_pages, void *guest_code, > uint32_t vcpuids[]) > { > - return vm_create_with_vcpus(VM_MODE_DEFAULT, nr_vcpus, extra_mem_pages, > - num_percpu_pages, guest_code, vcpuids); > + return vm_create_with_vcpus(VM_MODE_DEFAULT, nr_vcpus, DEFAULT_GUEST_PHY_PAGES, > + extra_mem_pages, num_percpu_pages, guest_code, vcpuids); > } > > struct kvm_vm *vm_create_default(uint32_t vcpuid, uint64_t extra_mem_pages, > diff --git a/tools/testing/selftests/kvm/lib/perf_test_util.c b/tools/testing/selftests/kvm/lib/perf_test_util.c > index abf381800a59..7397ca299835 100644 > --- a/tools/testing/selftests/kvm/lib/perf_test_util.c > +++ b/tools/testing/selftests/kvm/lib/perf_test_util.c > @@ -69,7 +69,7 @@ struct kvm_vm *perf_test_create_vm(enum vm_guest_mode mode, int vcpus, > TEST_ASSERT(vcpu_memory_bytes % perf_test_args.guest_page_size == 0, > "Guest memory size is not guest page size aligned."); > > - vm = vm_create_with_vcpus(mode, vcpus, > + vm = vm_create_with_vcpus(mode, vcpus, DEFAULT_GUEST_PHY_PAGES, > (vcpus * vcpu_memory_bytes) / perf_test_args.guest_page_size, > 0, guest_code, NULL); > > -- > 2.25.1 > Besides the extra_pg_pages comment Reviewed-by: Andrew Jones <drjones@redhat.com> Thanks, drew
> -----Original Message----- > From: Paolo Bonzini <pbonzini@redhat.com> > Sent: Wednesday, June 9, 2021 1:30 AM > To: Duan, Zhenzhong <zhenzhong.duan@intel.com>; linux- > kernel@vger.kernel.org > Cc: linux-kselftest@vger.kernel.org; kvm@vger.kernel.org; > maciej.szmigiero@oracle.com; drjones@redhat.com; shuah@kernel.org > Subject: Re: [PATCH 0/3] Restore extra_mem_pages and add > slot0_mem_pages > > On 09/06/21 01:38, Zhenzhong Duan wrote: > > (39fe2fc96694 "selftests: kvm: make allocation of extra memory take > > effect") changed the meaning of extra_mem_pages and treated it as slot0 > memory size. > > > > In fact extra_mem_pages is used for non-slot0 memory size, there is no > > custom > > slot0 memory size support. See discuss in > > https://lkml.org/lkml/2021/6/3/551 > > for more details. > > > > This patchset restores extra_mem_pages's original meaning and adds > > support for custom slot0 memory with a new parameter slot0_mem_pages. > > Because the two reverts are so small, I squashed everything in a single patch > with the following message: > > Until commit 39fe2fc96694 ("selftests: kvm: make allocation of extra > memory take effect", 2021-05-27), parameter extra_mem_pages was used > only to calculate the page table size for all the memory chunks, > because real memory allocation happened with calls of > vm_userspace_mem_region_add() after vm_create_default(). > > Commit 39fe2fc96694 however changed the meaning of > extra_mem_pages to > the size of memory slot 0. This makes the memory allocation more > flexible, but makes it harder to account for the number of > pages needed for the page tables. For example, memslot_perf_test > has a small amount of memory in slot 0 but a lot in other slots, > and adding that memory twice (both in slot 0 and with later > calls to vm_userspace_mem_region_add()) causes an error that > was fixed in commit 000ac4295339 ("selftests: kvm: fix overlapping > addresses in memslot_perf_test", 2021-05-29) > > Since both uses are sensible, add a new parameter slot0_mem_pages > to vm_create_with_vcpus() and some comments to clarify the meaning of > slot0_mem_pages and extra_mem_pages. With this change, > memslot_perf_test can go back to passing the number of memory > pages as extra_mem_pages. > This looks more clear, thanks for doing that. Regards Zhenzhong
(39fe2fc96694 "selftests: kvm: make allocation of extra memory take effect") changed the meaning of extra_mem_pages and treated it as slot0 memory size. In fact extra_mem_pages is used for non-slot0 memory size, there is no custom slot0 memory size support. See discuss in https://lkml.org/lkml/2021/6/3/551 for more details. This patchset restores extra_mem_pages's original meaning and adds support for custom slot0 memory with a new parameter slot0_mem_pages. Run below command, all 39 tests passed. # make -C tools/testing/selftests/ TARGETS=kvm run_tests Zhenzhong Duan (3): Revert "selftests: kvm: make allocation of extra memory take effect" Revert "selftests: kvm: fix overlapping addresses in memslot_perf_test" selftests: kvm: Add support for customized slot0 memory size .../testing/selftests/kvm/include/kvm_util.h | 7 +-- .../selftests/kvm/kvm_page_table_test.c | 2 +- tools/testing/selftests/kvm/lib/kvm_util.c | 47 +++++++++++++++---- .../selftests/kvm/lib/perf_test_util.c | 2 +- .../testing/selftests/kvm/memslot_perf_test.c | 2 +- 5 files changed, 45 insertions(+), 15 deletions(-)