KVM: selftests: Add simple SEV test
Subsystems (4)
v3 cover KVM: selftests: Add simple SEV test
This patch series combines the work Michael Roth has done in supporting SEV guests in selftests and the work Sean Christopherson suggested to allow ucalls from SEV guests. And the work Sean has sent to consolidate the ucall boilerplate code. Along with a very simple version of the SEV selftests Michael originally proposed.
V3
* Addressed more of andrew.jones@ in ucall patches.
* Fix build in non-x86 archs.
V2
* Dropped RFC tag
* Correctly separated Sean's ucall patches into 2 as originally intended.
* Addressed andrew.jones@ in ucall patches.
* Fixed ucall pool usage to work for other archs
V1
* https://lore.kernel.org/all/[email protected]/
Michael Roth (6):
KVM: selftests: move vm_phy_pages_alloc() earlier in file
KVM: selftests: sparsebit: add const where appropriate
KVM: selftests: add hooks for managing encrypted guest memory
KVM: selftests: handle encryption bits in page tables
KVM: selftests: add support for encrypted vm_vaddr_* allocations
KVM: selftests: add library for creating/interacting with SEV guests
Peter Gonda (3):
tools: Add atomic_test_and_set_bit()
KVM: selftests: Add ucall pool based implementation
KVM: selftests: Add simple sev vm testing
Sean Christopherson (2):
KVM: selftests: Consolidate common code for popuplating
KVM: selftests: Consolidate boilerplate code in get_ucall()
tools/arch/x86/include/asm/atomic.h | 7 + tools/include/asm-generic/atomic-gcc.h | 15 + tools/testing/selftests/kvm/.gitignore | 1 + tools/testing/selftests/kvm/Makefile | 3 + .../selftests/kvm/include/kvm_util_base.h | 25 ++ .../testing/selftests/kvm/include/sparsebit.h | 36 +-- .../selftests/kvm/include/ucall_common.h | 14 +- .../selftests/kvm/include/x86_64/sev.h | 47 +++ .../testing/selftests/kvm/lib/aarch64/ucall.c | 38 +-- tools/testing/selftests/kvm/lib/kvm_util.c | 267 +++++++++++++----- tools/testing/selftests/kvm/lib/riscv/ucall.c | 43 +-- tools/testing/selftests/kvm/lib/s390x/ucall.c | 40 +-- tools/testing/selftests/kvm/lib/sparsebit.c | 48 ++-- .../testing/selftests/kvm/lib/ucall_common.c | 139 +++++++++ .../selftests/kvm/lib/x86_64/processor.c | 15 +- tools/testing/selftests/kvm/lib/x86_64/sev.c | 249 ++++++++++++++++ .../testing/selftests/kvm/lib/x86_64/ucall.c | 39 +-- .../selftests/kvm/x86_64/sev_all_boot_test.c | 131 +++++++++ 18 files changed, 911 insertions(+), 246 deletions(-) create mode 100644 tools/testing/selftests/kvm/include/x86_64/sev.h create mode 100644 tools/testing/selftests/kvm/lib/ucall_common.c create mode 100644 tools/testing/selftests/kvm/lib/x86_64/sev.c create mode 100644 tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c
Signature · 1 line
2.37.1.559.g78731f0fdb-goog
PATCH v3 1/11 KVM: selftests: move vm_phy_pages_alloc() earlier in file
From: Michael Roth <[email protected]>
Subsequent patches will break some of this code out into file-local helper functions, which will be used by functions like vm_vaddr_alloc(), which currently are defined earlier in the file, so a forward declaration would be needed.
Instead, move it earlier in the file, just above vm_vaddr_alloc() and and friends, which are the main users.
--- tools/testing/selftests/kvm/lib/kvm_util.c | 145 ++++++++++----------- 1 file changed, 72 insertions(+), 73 deletions(-)
tools/testing/selftests/kvm/lib/kvm_util.c +72 -73
@@ -1089,6 +1089,78 @@ struct kvm_vcpu *__vm_vcpu_add(struct kvm_vm *vm, uint32_t vcpu_id)return vcpu; } +/* +* Physical Contiguous Page Allocator +* +* Input Args: +* vm - Virtual Machine +* num - number of pages +* paddr_min - Physical address minimum +* memslot - Memory region to allocate page from +* +* Output Args: None +* +* Return: +* Starting physical address +* +* Within the VM specified by vm, locates a range of available physical +* pages at or above paddr_min. If found, the pages are marked as in use +* and their base address is returned. A TEST_ASSERT failure occurs if +* not enough pages are available at or above paddr_min. +*/ +vm_paddr_t vm_phy_pages_alloc (struct kvm_vm *vm,size_t num, +vm_paddr_t paddr_min,uint32_t memslot) +{ +struct userspace_mem_region *region; +sparsebit_idx_t pg, base; + +TEST_ASSERT (num >0 ,"Must allocate at least one page" ); + +TEST_ASSERT ((paddr_min % vm->page_size) ==0 , +"Min physical address not divisible by page size.\n paddr_min: 0x%lx page_size: 0x%x" , + paddr_min, vm->page_size); + + region =memslot2region (vm, memslot); + base = pg = paddr_min >> vm->page_shift; + +do { +for (; pg < base + num; ++pg) { +if (!sparsebit_is_set (region->unused_phy_pages, pg)) { + base = pg =sparsebit_next_set (region->unused_phy_pages, pg); +break ; + } + } + }while (pg && pg != base + num); + +if (pg ==0 ) { +fprintf (stderr, +"No guest physical page available, paddr_min: 0x%lx page_size: 0x%x memslot: %u\n" , + paddr_min, vm->page_size, memslot); +fputs ("---- vm dump ----\n" , stderr); +vm_dump (stderr, vm,2 ); +abort (); + } + +for (pg = base; pg < base + num; ++pg) +sparsebit_clear (region->unused_phy_pages, pg); + +return base * vm->page_size; +} + +vm_paddr_t vm_phy_page_alloc (struct kvm_vm *vm,vm_paddr_t paddr_min, +uint32_t memslot) +{ +return vm_phy_pages_alloc (vm,1 , paddr_min, memslot); +} + +/* Arbitrary minimum physical address used for virtual translation tables. */ +#define KVM_GUEST_PAGE_TABLE_MIN_PADDR 0x180000 + +vm_paddr_t vm_alloc_page_table (struct kvm_vm *vm) +{ +return vm_phy_page_alloc (vm,KVM_GUEST_PAGE_TABLE_MIN_PADDR ,0 ); +} +/* * VM Virtual Address Unused Gap *
@@ -1735,79 +1807,6 @@ const char *exit_reason_str(unsigned int exit_reason)return "Unknown" ; } -/* -* Physical Contiguous Page Allocator -* -* Input Args: -* vm - Virtual Machine -* num - number of pages -* paddr_min - Physical address minimum -* memslot - Memory region to allocate page from -* -* Output Args: None -* -* Return: -* Starting physical address -* -* Within the VM specified by vm, locates a range of available physical -* pages at or above paddr_min. If found, the pages are marked as in use -* and their base address is returned. A TEST_ASSERT failure occurs if -* not enough pages are available at or above paddr_min. -*/ -vm_paddr_t vm_phy_pages_alloc (struct kvm_vm *vm,size_t num, -vm_paddr_t paddr_min,uint32_t memslot) -{ -struct userspace_mem_region *region; -sparsebit_idx_t pg, base; - -TEST_ASSERT (num >0 ,"Must allocate at least one page" ); - -TEST_ASSERT ((paddr_min % vm->page_size) ==0 ,"Min physical address " -"not divisible by page size.\n" -" paddr_min: 0x%lx page_size: 0x%x" , - paddr_min, vm->page_size); - - region =memslot2region (vm, memslot); - base = pg = paddr_min >> vm->page_shift; - -do { -for (; pg < base + num; ++pg) { -if (!sparsebit_is_set (region->unused_phy_pages, pg)) { - base = pg =sparsebit_next_set (region->unused_phy_pages, pg); -break ; - } - } - }while (pg && pg != base + num); - -if (pg ==0 ) { -fprintf (stderr,"No guest physical page available, " -"paddr_min: 0x%lx page_size: 0x%x memslot: %u\n" , - paddr_min, vm->page_size, memslot); -fputs ("---- vm dump ----\n" , stderr); -vm_dump (stderr, vm,2 ); -abort (); - } - -for (pg = base; pg < base + num; ++pg) -sparsebit_clear (region->unused_phy_pages, pg); - -return base * vm->page_size; -} - -vm_paddr_t vm_phy_page_alloc (struct kvm_vm *vm,vm_paddr_t paddr_min, -uint32_t memslot) -{ -return vm_phy_pages_alloc (vm,1 , paddr_min, memslot); -} - -/* Arbitrary minimum physical address used for virtual translation tables. */ -#define KVM_GUEST_PAGE_TABLE_MIN_PADDR 0x180000 - -vm_paddr_t vm_alloc_page_table (struct kvm_vm *vm) -{ -return vm_phy_page_alloc (vm,KVM_GUEST_PAGE_TABLE_MIN_PADDR ,0 ); -} -/* * Address Guest Virtual to Host Virtual *
Signature · 1 line
2.37.1.559.g78731f0fdb-goog
PATCH v3 2/11 KVM: selftests: sparsebit: add const where appropriate
In reply to Peter GondaFrom: Michael Roth <[email protected]>
Subsequent patches will introduce an encryption bitmap in kvm_util that would be useful to allow tests to access in read-only fashion. This will be done via a const sparsebit*. To avoid warnings or the need to add casts everywhere, add const to the various sparsebit functions that are applicable for read-only usage of sparsebit.
--- .../testing/selftests/kvm/include/sparsebit.h | 36 +++++++------- tools/testing/selftests/kvm/lib/sparsebit.c | 48 +++++++++---------- 2 files changed, 42 insertions(+), 42 deletions(-)
tools/testing/selftests/kvm/include/sparsebit.h +18 -18
@@ -30,26 +30,26 @@ typedef uint64_t sparsebit_num_t;struct sparsebit *sparsebit_alloc (void );void sparsebit_free (struct sparsebit **sbitp); -void sparsebit_copy (struct sparsebit *dstp,struct sparsebit *src); +void sparsebit_copy (struct sparsebit *dstp,const struct sparsebit *src); -bool sparsebit_is_set (struct sparsebit *sbit,sparsebit_idx_t idx); -bool sparsebit_is_set_num (struct sparsebit *sbit, +bool sparsebit_is_set (const struct sparsebit *sbit,sparsebit_idx_t idx); +bool sparsebit_is_set_num (const struct sparsebit *sbit,sparsebit_idx_t idx,sparsebit_num_t num); -bool sparsebit_is_clear (struct sparsebit *sbit,sparsebit_idx_t idx); -bool sparsebit_is_clear_num (struct sparsebit *sbit, +bool sparsebit_is_clear (const struct sparsebit *sbit,sparsebit_idx_t idx); +bool sparsebit_is_clear_num (const struct sparsebit *sbit,sparsebit_idx_t idx,sparsebit_num_t num); -sparsebit_num_t sparsebit_num_set (struct sparsebit *sbit); -bool sparsebit_any_set (struct sparsebit *sbit); -bool sparsebit_any_clear (struct sparsebit *sbit); -bool sparsebit_all_set (struct sparsebit *sbit); -bool sparsebit_all_clear (struct sparsebit *sbit); -sparsebit_idx_t sparsebit_first_set (struct sparsebit *sbit); -sparsebit_idx_t sparsebit_first_clear (struct sparsebit *sbit); -sparsebit_idx_t sparsebit_next_set (struct sparsebit *sbit,sparsebit_idx_t prev); -sparsebit_idx_t sparsebit_next_clear (struct sparsebit *sbit,sparsebit_idx_t prev); -sparsebit_idx_t sparsebit_next_set_num (struct sparsebit *sbit, +sparsebit_num_t sparsebit_num_set (const struct sparsebit *sbit); +bool sparsebit_any_set (const struct sparsebit *sbit); +bool sparsebit_any_clear (const struct sparsebit *sbit); +bool sparsebit_all_set (const struct sparsebit *sbit); +bool sparsebit_all_clear (const struct sparsebit *sbit); +sparsebit_idx_t sparsebit_first_set (const struct sparsebit *sbit); +sparsebit_idx_t sparsebit_first_clear (const struct sparsebit *sbit); +sparsebit_idx_t sparsebit_next_set (const struct sparsebit *sbit,sparsebit_idx_t prev); +sparsebit_idx_t sparsebit_next_clear (const struct sparsebit *sbit,sparsebit_idx_t prev); +sparsebit_idx_t sparsebit_next_set_num (const struct sparsebit *sbit,sparsebit_idx_t start,sparsebit_num_t num); -sparsebit_idx_t sparsebit_next_clear_num (struct sparsebit *sbit, +sparsebit_idx_t sparsebit_next_clear_num (const struct sparsebit *sbit,sparsebit_idx_t start,sparsebit_num_t num);void sparsebit_set (struct sparsebit *sbitp,sparsebit_idx_t idx);
@@ -62,9 +62,9 @@ void sparsebit_clear_num(struct sparsebit *sbitp, sparsebit_idx_t start, sparsebit_num_t num);void sparsebit_clear_all (struct sparsebit *sbitp); -void sparsebit_dump (FILE *stream,struct sparsebit *sbit, +void sparsebit_dump (FILE *stream,const struct sparsebit *sbit,unsigned int indent); -void sparsebit_validate_internal (struct sparsebit *sbit); +void sparsebit_validate_internal (const struct sparsebit *sbit); #ifdef __cplusplus }
tools/testing/selftests/kvm/lib/sparsebit.c +24 -24
190 lines not on this page · raw message
Signature · 1 line
2.37.1.559.g78731f0fdb-goog
PATCH v3 3/11 KVM: selftests: add hooks for managing encrypted guest memory
In reply to Peter GondaFrom: Michael Roth <[email protected]>
VM implementations that make use of encrypted memory need a way to configure things like the encryption/shared bit position for page table handling, the default encryption policy for internal allocations made by the core library, and a way to fetch the list/bitmap of encrypted pages to do the actual memory encryption. Add an interface to configure these parameters. Also introduce a sparsebit map to track allocations/mappings that should be treated as encrypted, and provide a way for VM implementations to retrieve it to handle operations related memory encryption.
--- .../selftests/kvm/include/kvm_util_base.h | 17 ++++++ tools/testing/selftests/kvm/lib/kvm_util.c | 52 +++++++++++++++++-- 2 files changed, 66 insertions(+), 3 deletions(-)
tools/testing/selftests/kvm/include/kvm_util_base.h +17 -0
@@ -32,6 +32,7 @@ typedef uint64_t vm_vaddr_t; /* Virtual Machine (Guest) virtual address */struct userspace_mem_region {struct kvm_userspace_memory_region region;struct sparsebit *unused_phy_pages; +struct sparsebit *encrypted_phy_pages;int fd;off_t offset;void *host_mem;
@@ -64,6 +65,14 @@ struct userspace_mem_regions {DECLARE_HASHTABLE (slot_hash,9 ); }; +/* Memory encryption policy/configuration. */ +struct vm_memcrypt { +bool enabled; +int8_t enc_by_default; +bool has_enc_bit; +int8_t enc_bit; +}; +struct kvm_vm {int mode;unsigned long type;
@@ -87,6 +96,7 @@ struct kvm_vm {vm_vaddr_t idt;vm_vaddr_t handlers;uint32_t dirty_ring_size; +struct vm_memcrypt memcrypt;/* Cache of information for binary stats interface */ int stats_fd;
@@ -834,4 +844,11 @@ static inline int __vm_disable_nx_huge_pages(struct kvm_vm *vm) return __vm_enable_cap(vm, KVM_CAP_VM_DISABLE_NX_HUGE_PAGES, 0); } +void vm_set_memory_encryption (struct kvm_vm *vm,bool enc_by_default,bool has_enc_bit, +uint8_t enc_bit); + +const struct sparsebit *vm_get_encrypted_phy_pages (struct kvm_vm *vm,int slot, +vm_paddr_t *gpa_start, +uint64_t *size); +#endif /* SELFTEST_KVM_UTIL_BASE_H */
tools/testing/selftests/kvm/lib/kvm_util.c +49 -3
96 lines not on this page · raw message
Signature · 1 line
2.37.1.559.g78731f0fdb-goog
PATCH v3 4/11 KVM: selftests: handle encryption bits in page tables
In reply to Peter GondaFrom: Michael Roth <[email protected]>
SEV guests rely on an encyption bit which resides within the range that current code treats as address bits. Guest code will expect these bits to be set appropriately in their page tables, whereas the rest of the kvm_util functions will generally expect these bits to not be present. Introduce addr_gpa2raw()/addr_raw2gpa() to add/remove these bits, then use them where appropriate.
--- .../selftests/kvm/include/kvm_util_base.h | 2 + tools/testing/selftests/kvm/lib/kvm_util.c | 55 ++++++++++++++++++- .../selftests/kvm/lib/x86_64/processor.c | 15 +++-- 3 files changed, 66 insertions(+), 6 deletions(-)
tools/testing/selftests/kvm/include/kvm_util_base.h +2 -0
@@ -399,6 +399,8 @@ void *addr_gpa2hva(struct kvm_vm *vm, vm_paddr_t gpa);void *addr_gva2hva (struct kvm_vm *vm,vm_vaddr_t gva);vm_paddr_t addr_hva2gpa (struct kvm_vm *vm,void *hva);void *addr_gpa2alias (struct kvm_vm *vm,vm_paddr_t gpa); +vm_paddr_t addr_raw2gpa (struct kvm_vm *vm,vm_vaddr_t gpa_raw); +vm_paddr_t addr_gpa2raw (struct kvm_vm *vm,vm_vaddr_t gpa);void vcpu_run (struct kvm_vcpu *vcpu);int _vcpu_run (struct kvm_vcpu *vcpu);
tools/testing/selftests/kvm/lib/kvm_util.c +54 -1
69 lines not on this page · raw message
tools/testing/selftests/kvm/lib/x86_64/processor.c +10 -5
@@ -118,7 +118,7 @@ void virt_arch_pgd_alloc(struct kvm_vm *vm)/* If needed, create page map l4 table. */ if (!vm->pgd_created) { - vm->pgd =vm_alloc_page_table (vm); + vm->pgd =addr_gpa2raw (vm,vm_alloc_page_table (vm)); vm->pgd_created = true; } }
@@ -140,13 +140,15 @@ static uint64_t *virt_create_upper_pte(struct kvm_vm *vm, int target_level) {uint64_t *pte =virt_get_pte (vm, pt_pfn, vaddr, current_level); +uint64_t paddr_raw =addr_gpa2raw (vm, paddr);if (!(*pte &PTE_PRESENT_MASK )) { *pte =PTE_PRESENT_MASK |PTE_WRITABLE_MASK ;if (current_level == target_level) - *pte |=PTE_LARGE_MASK | (paddr &PHYSICAL_PAGE_MASK ); + *pte |=PTE_LARGE_MASK | (paddr_raw &PHYSICAL_PAGE_MASK );else - *pte |=vm_alloc_page_table (vm) &PHYSICAL_PAGE_MASK ; + *pte |=addr_gpa2raw (vm,vm_alloc_page_table (vm)) &PHYSICAL_PAGE_MASK ; + }else {/* * Entry already present. Assert that the caller doesn't want
@@ -184,6 +186,8 @@ void __virt_pg_map(struct kvm_vm *vm, uint64_t vaddr, uint64_t paddr, int level)"Physical address beyond maximum supported,\n" " paddr: 0x%lx vm->max_gfn: 0x%lx vm->page_size: 0x%x" , paddr, vm->max_gfn, vm->page_size); +TEST_ASSERT (addr_raw2gpa (vm, paddr) == paddr, +"Unexpected bits in paddr: %lx" , paddr);/* * Allocate upper level page tables, if not already present. Return
@@ -206,7 +210,8 @@ void __virt_pg_map(struct kvm_vm *vm, uint64_t vaddr, uint64_t paddr, int level) pte =virt_get_pte (vm,PTE_GET_PFN (*pde), vaddr,PG_LEVEL_4K );TEST_ASSERT (!(*pte &PTE_PRESENT_MASK ),"PTE already present for 4k page at vaddr: 0x%lx\n" , vaddr); - *pte =PTE_PRESENT_MASK |PTE_WRITABLE_MASK | (paddr &PHYSICAL_PAGE_MASK ); + *pte =PTE_PRESENT_MASK |PTE_WRITABLE_MASK | + (addr_gpa2raw (vm, paddr) &PHYSICAL_PAGE_MASK ); }void virt_arch_pg_map (struct kvm_vm *vm,uint64_t vaddr,uint64_t paddr)
@@ -515,7 +520,7 @@ vm_paddr_t addr_arch_gva2gpa(struct kvm_vm *vm, vm_vaddr_t gva)if (!(pte[index[0 ]] &PTE_PRESENT_MASK )) goto unmapped_gva; -return (PTE_GET_PFN (pte[index[0 ]]) * vm->page_size) + (gva & ~PAGE_MASK ); +return addr_raw2gpa (vm,PTE_GET_PFN (pte[index[0 ]]) * vm->page_size) + (gva & ~PAGE_MASK ); unmapped_gva:TEST_FAIL ("No mapping for vm virtual address, gva: 0x%lx" , gva);
Signature · 1 line
2.37.1.559.g78731f0fdb-goog
PATCH v3 5/11 KVM: selftests: add support for encrypted vm_vaddr_* allocations
In reply to Peter GondaFrom: Michael Roth <[email protected]>
The default policy for whether to handle allocations as encrypted or shared pages is currently determined by vm_phy_pages_alloc(), which in turn uses the policy defined by vm->memcrypt.enc_by_default.
Test programs may wish to allocate shared vaddrs for things like sharing memory with the guest. Since enc_by_default will be true in the case of SEV guests (since it's required in order to have the initial ELF binary and page table become part of the initial guest payload), an interface is needed to explicitly request shared pages.
Implement this by splitting the common code out from vm_vaddr_alloc() and introducing a new vm_vaddr_alloc_shared().
--- .../selftests/kvm/include/kvm_util_base.h | 1 + tools/testing/selftests/kvm/lib/kvm_util.c | 21 +++++++++++++++---- 2 files changed, 18 insertions(+), 4 deletions(-)
tools/testing/selftests/kvm/include/kvm_util_base.h +1 -0
@@ -390,6 +390,7 @@ void vm_mem_region_move(struct kvm_vm *vm, uint32_t slot, uint64_t new_gpa);void vm_mem_region_delete (struct kvm_vm *vm,uint32_t slot);struct kvm_vcpu *__vm_vcpu_add (struct kvm_vm *vm,uint32_t vcpu_id);vm_vaddr_t vm_vaddr_alloc (struct kvm_vm *vm,size_t sz,vm_vaddr_t vaddr_min); +vm_vaddr_t vm_vaddr_alloc_shared (struct kvm_vm *vm,size_t sz,vm_vaddr_t vaddr_min);vm_vaddr_t vm_vaddr_alloc_pages (struct kvm_vm *vm,int nr_pages);vm_vaddr_t vm_vaddr_alloc_page (struct kvm_vm *vm);
tools/testing/selftests/kvm/lib/kvm_util.c +17 -4
@@ -1262,12 +1262,13 @@ static vm_vaddr_t vm_vaddr_unused_gap(struct kvm_vm *vm, size_t sz, }/* -* VM Virtual Address Allocate +* VM Virtual Address Allocate Shared/Encrypted * * Input Args: * vm - Virtual Machine * sz - Size in bytes * vaddr_min - Minimum starting virtual address +* encrypt - Whether the region should be handled as encrypted * * Output Args: None *
@@ -1280,13 +1281,15 @@ static vm_vaddr_t vm_vaddr_unused_gap(struct kvm_vm *vm, size_t sz,* a unique set of pages, with the minimum real allocation being at least * a page. */ -vm_vaddr_t vm_vaddr_alloc (struct kvm_vm *vm,size_t sz,vm_vaddr_t vaddr_min) +static vm_vaddr_t +_vm_vaddr_alloc (struct kvm_vm *vm,size_t sz,vm_vaddr_t vaddr_min,bool encrypt) {uint64_t pages = (sz >> vm->page_shift) + ((sz % vm->page_size) !=0 );virt_pgd_alloc (vm); -vm_paddr_t paddr =vm_phy_pages_alloc (vm, pages, -KVM_UTIL_MIN_PFN * vm->page_size,0 ); +vm_paddr_t paddr =_vm_phy_pages_alloc (vm, pages, +KVM_UTIL_MIN_PFN * vm->page_size, +0 , encrypt);/* * Find an unused range of virtual page addresses of at least
@@ -1307,6 +1310,16 @@ vm_vaddr_t vm_vaddr_alloc(struct kvm_vm *vm, size_t sz, vm_vaddr_t vaddr_min)return vaddr_start; } +vm_vaddr_t vm_vaddr_alloc (struct kvm_vm *vm,size_t sz,vm_vaddr_t vaddr_min) +{ +return _vm_vaddr_alloc (vm, sz, vaddr_min, vm->memcrypt.enc_by_default); +} + +vm_vaddr_t vm_vaddr_alloc_shared (struct kvm_vm *vm,size_t sz,vm_vaddr_t vaddr_min) +{ +return _vm_vaddr_alloc (vm, sz, vaddr_min, false); +} +/* * VM Virtual Address Allocate Pages *
Signature · 1 line
2.37.1.559.g78731f0fdb-goog
PATCH v3 6/11 KVM: selftests: Consolidate common code for popuplating
In reply to Peter GondaFrom: Sean Christopherson <[email protected]>
Make ucall() a common helper that populates struct ucall, and only calls into arch code to make the actually call out to userspace.
Rename all arch-specific helpers to make it clear they're arch-specific, and to avoid collisions with common helpers (one more on its way...)
Add WRITE_ONCE() to stores in ucall() code to as already done to aarch64 code in commit 9e2f6498efbb ("selftests: KVM: Handle compiler optimizations in ucall") to prevent clang optimizations breaking ucalls.
--- tools/testing/selftests/kvm/Makefile | 1 + .../selftests/kvm/include/ucall_common.h | 23 ++++++++++++++++--- .../testing/selftests/kvm/lib/aarch64/ucall.c | 20 ++++------------ tools/testing/selftests/kvm/lib/riscv/ucall.c | 23 ++++--------------- tools/testing/selftests/kvm/lib/s390x/ucall.c | 23 ++++--------------- .../testing/selftests/kvm/lib/ucall_common.c | 20 ++++++++++++++++ .../testing/selftests/kvm/lib/x86_64/ucall.c | 23 ++++--------------- 7 files changed, 60 insertions(+), 73 deletions(-) create mode 100644 tools/testing/selftests/kvm/lib/ucall_common.c
tools/testing/selftests/kvm/Makefile +1 -0
@@ -46,6 +46,7 @@ LIBKVM += lib/perf_test_util.c LIBKVM += lib/rbtree.c LIBKVM += lib/sparsebit.c LIBKVM += lib/test_util.c +LIBKVM += lib/ucall_common.c LIBKVM_x86_64 += lib/x86_64/apic.c LIBKVM_x86_64 += lib/x86_64/handlers.S
tools/testing/selftests/kvm/include/ucall_common.h +20 -3
@@ -24,10 +24,27 @@ struct ucall {uint64_t args[UCALL_MAX_ARGS ]; }; -void ucall_init (struct kvm_vm *vm,void *arg); -void ucall_uninit (struct kvm_vm *vm); +void ucall_arch_init (struct kvm_vm *vm,void *arg); +void ucall_arch_uninit (struct kvm_vm *vm); +void ucall_arch_do_ucall (vm_vaddr_t uc); +uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc); +void ucall (uint64_t cmd,int nargs, ...); -uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc); + +static inline void ucall_init (struct kvm_vm *vm,void *arg) +{ +ucall_arch_init (vm, arg); +} + +static inline void ucall_uninit (struct kvm_vm *vm) +{ +ucall_arch_uninit (vm); +} + +static inline uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +{ +return ucall_arch_get_ucall (vcpu, uc); +}#define GUEST_SYNC_ARGS (stage, arg1, arg2, arg3, arg4) \ ucall(UCALL_SYNC, 6, "hello", stage, arg1, arg2, arg3, arg4)
tools/testing/selftests/kvm/lib/aarch64/ucall.c +4 -16
@@ -21,7 +21,7 @@ static bool ucall_mmio_init(struct kvm_vm *vm, vm_paddr_t gpa)return true; } -void ucall_init (struct kvm_vm *vm,void *arg) +void ucall_arch_init (struct kvm_vm *vm,void *arg) {vm_paddr_t gpa, start, end, step, offset;unsigned int bits;
@@ -64,30 +64,18 @@ void ucall_init(struct kvm_vm *vm, void *arg)TEST_FAIL ("Can't find a ucall mmio address" ); } -void ucall_uninit (struct kvm_vm *vm) +void ucall_arch_uninit (struct kvm_vm *vm) { ucall_exit_mmio_addr =0 ;sync_global_to_guest (vm, ucall_exit_mmio_addr); } -void ucall (uint64_t cmd,int nargs, ...) +void ucall_arch_do_ucall (vm_vaddr_t uc) { -struct ucall uc = {}; -va_list va; -int i; - -WRITE_ONCE (uc.cmd, cmd); - nargs =min (nargs,UCALL_MAX_ARGS ); - -va_start (va, nargs); -for (i =0 ; i < nargs; ++i) -WRITE_ONCE (uc.args[i],va_arg (va, uint64_t)); -va_end (va); -WRITE_ONCE (*ucall_exit_mmio_addr, (vm_vaddr_t)&uc); } -uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) {struct kvm_run *run = vcpu->run;struct ucall ucall = {};
tools/testing/selftests/kvm/lib/riscv/ucall.c +5 -18
@@ -10,11 +10,11 @@#include "kvm_util.h" #include "processor.h" -void ucall_init (struct kvm_vm *vm,void *arg) +void ucall_arch_init (struct kvm_vm *vm,void *arg) { } -void ucall_uninit (struct kvm_vm *vm) +void ucall_arch_uninit (struct kvm_vm *vm) { }
@@ -44,27 +44,14 @@ struct sbiret sbi_ecall(int ext, int fid, unsigned long arg0,return ret; } -void ucall (uint64_t cmd,int nargs, ...) +void ucall_arch_do_ucall (vm_vaddr_t uc) { -struct ucall uc = { - .cmd = cmd, - }; -va_list va; -int i; - - nargs =min (nargs,UCALL_MAX_ARGS ); - -va_start (va, nargs); -for (i =0 ; i < nargs; ++i) - uc.args[i] =va_arg (va, uint64_t); -va_end (va); -sbi_ecall (KVM_RISCV_SELFTESTS_SBI_EXT ,KVM_RISCV_SELFTESTS_SBI_UCALL , - (vm_vaddr_t)&uc,0 ,0 ,0 ,0 ,0 ); + uc,0 ,0 ,0 ,0 ,0 ); } -uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) {struct kvm_run *run = vcpu->run;struct ucall ucall = {};
tools/testing/selftests/kvm/lib/s390x/ucall.c +5 -18
@@ -6,34 +6,21 @@*/ #include "kvm_util.h" -void ucall_init (struct kvm_vm *vm,void *arg) +void ucall_arch_init (struct kvm_vm *vm,void *arg) { } -void ucall_uninit (struct kvm_vm *vm) +void ucall_arch_uninit (struct kvm_vm *vm) { } -void ucall (uint64_t cmd,int nargs, ...) +void ucall_arch_do_ucall (vm_vaddr_t uc) { -struct ucall uc = { - .cmd = cmd, - }; -va_list va; -int i; - - nargs =min (nargs,UCALL_MAX_ARGS ); - -va_start (va, nargs); -for (i =0 ; i < nargs; ++i) - uc.args[i] =va_arg (va, uint64_t); -va_end (va); -/* Exit via DIAGNOSE 0x501 (normally used for breakpoints) */ - asmvolatile ("diag 0,%0,0x501" : :"a" (&uc) :"memory" ); + asmvolatile ("diag 0,%0,0x501" : :"a" (uc) :"memory" ); } -uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) {struct kvm_run *run = vcpu->run;struct ucall ucall = {};
tools/testing/selftests/kvm/lib/ucall_common.c new file +20 -0
@@ -0,0 +1,20 @@ +// SPDX-License-Identifier: GPL-2.0-only +#include "kvm_util.h" + +void ucall (uint64_t cmd,int nargs, ...) +{ +struct ucall uc = {}; +va_list va; +int i; + +WRITE_ONCE (uc.cmd, cmd); + + nargs =min (nargs,UCALL_MAX_ARGS ); + +va_start (va, nargs); +for (i =0 ; i < nargs; ++i) +WRITE_ONCE (uc.args[i],va_arg (va, uint64_t)); +va_end (va); + +ucall_arch_do_ucall ((vm_vaddr_t)&uc); +}
tools/testing/selftests/kvm/lib/x86_64/ucall.c +5 -18
@@ -8,34 +8,21 @@#define UCALL_PIO_PORT ((uint16_t)0x1000) -void ucall_init (struct kvm_vm *vm,void *arg) +void ucall_arch_init (struct kvm_vm *vm,void *arg) { } -void ucall_uninit (struct kvm_vm *vm) +void ucall_arch_uninit (struct kvm_vm *vm) { } -void ucall (uint64_t cmd,int nargs, ...) +void ucall_arch_do_ucall (vm_vaddr_t uc) { -struct ucall uc = { - .cmd = cmd, - }; -va_list va; -int i; - - nargs =min (nargs,UCALL_MAX_ARGS ); - -va_start (va, nargs); -for (i =0 ; i < nargs; ++i) - uc.args[i] =va_arg (va, uint64_t); -va_end (va); - asmvolatile ("in %[port], %%al" - : : [port]"d" (UCALL_PIO_PORT ),"D" (&uc) :"rax" ,"memory" ); + : : [port]"d" (UCALL_PIO_PORT ),"D" (uc) :"rax" ,"memory" ); } -uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) {struct kvm_run *run = vcpu->run;struct ucall ucall = {};
Signature · 1 line
2.37.1.559.g78731f0fdb-goog
Re: [V3 06/11] KVM: selftests: Consolidate common code for popuplating
Quoted 104 lines · Peter Gonda
From: Sean Christopherson <[email protected]>
Make
ucall()a common helper that populates struct ucall, and only calls into arch code to make the actually call out to userspace.Rename all arch-specific helpers to make it clear they're arch-specific, and to avoid collisions with common helpers (one more on its way...)
Add
WRITE_ONCE()to stores inucall()code to as already done to aarch64 code in commit9e2f6498efbb("selftests: KVM: Handle compiler optimizations in ucall") to prevent clang optimizations breaking ucalls.Cc Colton Lewis <[email protected]>Cc Andrew Jones <[email protected]>Signed-off-by Sean Christopherson <[email protected]>Signed-off-by Peter Gonda <[email protected]>--- tools/testing/selftests/kvm/Makefile | 1 + .../selftests/kvm/include/ucall_common.h | 23 ++++++++++++++++--- .../testing/selftests/kvm/lib/aarch64/ucall.c | 20 ++++------------ tools/testing/selftests/kvm/lib/riscv/ucall.c | 23 ++++--------------- tools/testing/selftests/kvm/lib/s390x/ucall.c | 23 ++++--------------- .../testing/selftests/kvm/lib/ucall_common.c | 20 ++++++++++++++++ .../testing/selftests/kvm/lib/x86_64/ucall.c | 23 ++++--------------- 7 files changed, 60 insertions(+), 73 deletions(-) create mode 100644 tools/testing/selftests/kvm/lib/ucall_common.ctools/testing/selftests/kvm/Makefile +1
-0@@ -46,6 +46,7 @@ LIBKVM += lib/perf_test_util.c LIBKVM += lib/rbtree.c LIBKVM += lib/sparsebit.c LIBKVM += lib/test_util.c +LIBKVM += lib/ucall_common.c LIBKVM_x86_64 += lib/x86_64/apic.c LIBKVM_x86_64 += lib/x86_64/handlers.Stools/testing/selftests/kvm/include/ucall_common.h +20
-3@@ -24,10 +24,27 @@ struct ucall {uint64_t args[UCALL_MAX_ARGS ]; }; -void ucall_init (struct kvm_vm *vm,void *arg); -void ucall_uninit (struct kvm_vm *vm); +void ucall_arch_init (struct kvm_vm *vm,void *arg); +void ucall_arch_uninit (struct kvm_vm *vm); +void ucall_arch_do_ucall (vm_vaddr_t uc); +uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc); +void ucall (uint64_t cmd,int nargs, ...); -uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc); + +static inline void ucall_init (struct kvm_vm *vm,void *arg) +{ +ucall_arch_init (vm, arg); +} + +static inline void ucall_uninit (struct kvm_vm *vm) +{ +ucall_arch_uninit (vm); +} + +static inline uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +{ +return ucall_arch_get_ucall (vcpu, uc); +}#define GUEST_SYNC_ARGS (stage, arg1, arg2, arg3, arg4) \ ucall(UCALL_SYNC, 6, "hello", stage, arg1, arg2, arg3, arg4)tools/testing/selftests/kvm/lib/aarch64/ucall.c +3
-6@@ -21,7 +21,7 @@ static bool ucall_mmio_init(struct kvm_vm *vm, vm_paddr_t gpa)return true; } -void ucall_init (struct kvm_vm *vm,void *arg) +void ucall_arch_init (struct kvm_vm *vm,void *arg) {vm_paddr_t gpa, start, end, step, offset;unsigned int bits;@@ -64,30 +64,18 @@ void ucall_init(struct kvm_vm *vm, void *arg)TEST_FAIL ("Can't find a ucall mmio address" ); } -void ucall_uninit (struct kvm_vm *vm) +void ucall_arch_uninit (struct kvm_vm *vm) { ucall_exit_mmio_addr =0 ;sync_global_to_guest (vm, ucall_exit_mmio_addr); } -void ucall (uint64_t cmd,int nargs, ...) +void ucall_arch_do_ucall (vm_vaddr_t uc) { -struct ucall uc = {}; -va_list va; -int i;
tools/testing/selftests/kvm/lib/aarch64/ucall.c void ucall_init(struct kvm_vm *vm, void *arg) · line 75
Quoted 10 lines
tools/testing/selftests/kvm/lib/aarch64/ucall.c +0
-9- -WRITE_ONCE (uc.cmd, cmd); - nargs =min (nargs,UCALL_MAX_ARGS ); - -va_start (va, nargs); -for (i =0 ; i < nargs; ++i) -WRITE_ONCE (uc.args[i],va_arg (va, uint64_t)); -va_end (va); -WRITE_ONCE (*ucall_exit_mmio_addr, (vm_vaddr_t)&uc);
^ should just be uc
Quoted 173 lines
tools/testing/selftests/kvm/lib/aarch64/ucall.c +1
-1} -uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) {struct kvm_run *run = vcpu->run;struct ucall ucall = {};tools/testing/selftests/kvm/lib/riscv/ucall.c +5
-18@@ -10,11 +10,11 @@#include "kvm_util.h" #include "processor.h" -void ucall_init (struct kvm_vm *vm,void *arg) +void ucall_arch_init (struct kvm_vm *vm,void *arg) { } -void ucall_uninit (struct kvm_vm *vm) +void ucall_arch_uninit (struct kvm_vm *vm) { }@@ -44,27 +44,14 @@ struct sbiret sbi_ecall(int ext, int fid, unsigned long arg0,return ret; } -void ucall (uint64_t cmd,int nargs, ...) +void ucall_arch_do_ucall (vm_vaddr_t uc) { -struct ucall uc = { - .cmd = cmd, - }; -va_list va; -int i; - - nargs =min (nargs,UCALL_MAX_ARGS ); - -va_start (va, nargs); -for (i =0 ; i < nargs; ++i) - uc.args[i] =va_arg (va, uint64_t); -va_end (va); -sbi_ecall (KVM_RISCV_SELFTESTS_SBI_EXT ,KVM_RISCV_SELFTESTS_SBI_UCALL , - (vm_vaddr_t)&uc,0 ,0 ,0 ,0 ,0 ); + uc,0 ,0 ,0 ,0 ,0 ); } -uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) {struct kvm_run *run = vcpu->run;struct ucall ucall = {};tools/testing/selftests/kvm/lib/s390x/ucall.c +5
-18@@ -6,34 +6,21 @@*/ #include "kvm_util.h" -void ucall_init (struct kvm_vm *vm,void *arg) +void ucall_arch_init (struct kvm_vm *vm,void *arg) { } -void ucall_uninit (struct kvm_vm *vm) +void ucall_arch_uninit (struct kvm_vm *vm) { } -void ucall (uint64_t cmd,int nargs, ...) +void ucall_arch_do_ucall (vm_vaddr_t uc) { -struct ucall uc = { - .cmd = cmd, - }; -va_list va; -int i; - - nargs =min (nargs,UCALL_MAX_ARGS ); - -va_start (va, nargs); -for (i =0 ; i < nargs; ++i) - uc.args[i] =va_arg (va, uint64_t); -va_end (va); -/* Exit via DIAGNOSE 0x501 (normally used for breakpoints) */ - asmvolatile ("diag 0,%0,0x501" : :"a" (&uc) :"memory" ); + asmvolatile ("diag 0,%0,0x501" : :"a" (uc) :"memory" ); } -uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) {struct kvm_run *run = vcpu->run;struct ucall ucall = {};tools/testing/selftests/kvm/lib/ucall_common.c new file +20
-0@@ -0,0 +1,20 @@ +// SPDX-License-Identifier: GPL-2.0-only +#include "kvm_util.h" + +void ucall (uint64_t cmd,int nargs, ...) +{ +struct ucall uc = {}; +va_list va; +int i; + +WRITE_ONCE (uc.cmd, cmd); + + nargs =min (nargs,UCALL_MAX_ARGS ); + +va_start (va, nargs); +for (i =0 ; i < nargs; ++i) +WRITE_ONCE (uc.args[i],va_arg (va, uint64_t)); +va_end (va); + +ucall_arch_do_ucall ((vm_vaddr_t)&uc); +}tools/testing/selftests/kvm/lib/x86_64/ucall.c +5
-18@@ -8,34 +8,21 @@#define UCALL_PIO_PORT ((uint16_t)0x1000) -void ucall_init (struct kvm_vm *vm,void *arg) +void ucall_arch_init (struct kvm_vm *vm,void *arg) { } -void ucall_uninit (struct kvm_vm *vm) +void ucall_arch_uninit (struct kvm_vm *vm) { } -void ucall (uint64_t cmd,int nargs, ...) +void ucall_arch_do_ucall (vm_vaddr_t uc) { -struct ucall uc = { - .cmd = cmd, - }; -va_list va; -int i; - - nargs =min (nargs,UCALL_MAX_ARGS ); - -va_start (va, nargs); -for (i =0 ; i < nargs; ++i) - uc.args[i] =va_arg (va, uint64_t); -va_end (va); - asmvolatile ("in %[port], %%al" - : : [port]"d" (UCALL_PIO_PORT ),"D" (&uc) :"rax" ,"memory" ); + : : [port]"d" (UCALL_PIO_PORT ),"D" (uc) :"rax" ,"memory" ); } -uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) {struct kvm_run *run = vcpu->run;struct ucall ucall = {};Signature · 1 line
2.37.1.559.g78731f0fdb-goog
Otherwise,
PATCH v3 7/11 KVM: selftests: Consolidate boilerplate code in get_ucall()
In reply to Peter GondaFrom: Sean Christopherson <[email protected]>
Consolidate the actual copying of a ucall struct from guest=>host into the common get_ucall(). Return a host virtual address instead of a guest virtual address even though the addr_gva2hva() part could be moved to get_ucall() too. Conceptually, get_ucall() is invoked from the host and should return a host virtual address (and returning NULL for "nothing to see here" is far superior to returning 0).
Use pointer shenanigans instead of an unnecessary bounce buffer when the caller of get_ucall() provides a valid pointer.
--- .../selftests/kvm/include/ucall_common.h | 8 ++------ .../testing/selftests/kvm/lib/aarch64/ucall.c | 14 +++----------- tools/testing/selftests/kvm/lib/riscv/ucall.c | 19 +++---------------- tools/testing/selftests/kvm/lib/s390x/ucall.c | 16 +++------------- .../testing/selftests/kvm/lib/ucall_common.c | 19 +++++++++++++++++++ .../testing/selftests/kvm/lib/x86_64/ucall.c | 16 +++------------- 6 files changed, 33 insertions(+), 59 deletions(-)
tools/testing/selftests/kvm/include/ucall_common.h +2 -6
@@ -27,9 +27,10 @@ struct ucall {void ucall_arch_init (struct kvm_vm *vm,void *arg);void ucall_arch_uninit (struct kvm_vm *vm);void ucall_arch_do_ucall (vm_vaddr_t uc); -uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc); +void *ucall_arch_get_ucall (struct kvm_vcpu *vcpu);void ucall (uint64_t cmd,int nargs, ...); +uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc);static inline void ucall_init (struct kvm_vm *vm,void *arg) {
@@ -41,11 +42,6 @@ static inline void ucall_uninit(struct kvm_vm *vm)ucall_arch_uninit (vm); } -static inline uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) -{ -return ucall_arch_get_ucall (vcpu, uc); -} -#define GUEST_SYNC_ARGS (stage, arg1, arg2, arg3, arg4) \ ucall(UCALL_SYNC, 6, "hello", stage, arg1, arg2, arg3, arg4)#define GUEST_SYNC (stage) ucall(UCALL_SYNC, 2, "hello", stage)
tools/testing/selftests/kvm/lib/aarch64/ucall.c +3 -11
@@ -75,13 +75,9 @@ void ucall_arch_do_ucall(vm_vaddr_t uc)WRITE_ONCE (*ucall_exit_mmio_addr, (vm_vaddr_t)&uc); } -uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +void *ucall_arch_get_ucall (struct kvm_vcpu *vcpu) {struct kvm_run *run = vcpu->run; -struct ucall ucall = {}; - -if (uc) -memset (uc,0 ,sizeof (*uc));if (run->exit_reason ==KVM_EXIT_MMIO && run->mmio.phys_addr == (uint64_t )ucall_exit_mmio_addr) {
@@ -90,12 +86,8 @@ uint64_t ucall_arch_get_ucall(struct kvm_vcpu *vcpu, struct ucall *uc)TEST_ASSERT (run->mmio.is_write && run->mmio.len ==8 ,"Unexpected ucall exit mmio address access" );memcpy (&gva, run->mmio.data,sizeof (gva)); -memcpy (&ucall,addr_gva2hva (vcpu->vm, gva),sizeof (ucall)); - -vcpu_run_complete_io (vcpu); -if (uc) -memcpy (uc, &ucall,sizeof (ucall)); +return addr_gva2hva (vcpu->vm, gva); } -return ucall.cmd; +return NULL ; }
tools/testing/selftests/kvm/lib/riscv/ucall.c +3 -16
@@ -51,27 +51,15 @@ void ucall_arch_do_ucall(vm_vaddr_t uc) uc, 0, 0, 0, 0, 0); } -uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +void *ucall_arch_get_ucall (struct kvm_vcpu *vcpu) {struct kvm_run *run = vcpu->run; -struct ucall ucall = {}; - -if (uc) -memset (uc,0 ,sizeof (*uc));if (run->exit_reason ==KVM_EXIT_RISCV_SBI && run->riscv_sbi.extension_id ==KVM_RISCV_SELFTESTS_SBI_EXT ) {switch (run->riscv_sbi.function_id) {case KVM_RISCV_SELFTESTS_SBI_UCALL : -memcpy (&ucall, -addr_gva2hva (vcpu->vm, run->riscv_sbi.args[0 ]), -sizeof (ucall)); - -vcpu_run_complete_io (vcpu); -if (uc) -memcpy (uc, &ucall,sizeof (ucall)); - -break ; +return addr_gva2hva (vcpu->vm, run->riscv_sbi.args[0 ]);case KVM_RISCV_SELFTESTS_SBI_UNEXP :vcpu_dump (stderr, vcpu,2 );TEST_ASSERT (0 ,"Unexpected trap taken by guest" );
@@ -80,6 +68,5 @@ uint64_t ucall_arch_get_ucall(struct kvm_vcpu *vcpu, struct ucall *uc)break ; } } - -return ucall.cmd; +return NULL ; }
tools/testing/selftests/kvm/lib/s390x/ucall.c +3 -13
@@ -20,13 +20,9 @@ void ucall_arch_do_ucall(vm_vaddr_t uc) asmvolatile ("diag 0,%0,0x501" : :"a" (uc) :"memory" ); } -uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +void *ucall_arch_get_ucall (struct kvm_vcpu *vcpu) {struct kvm_run *run = vcpu->run; -struct ucall ucall = {}; - -if (uc) -memset (uc,0 ,sizeof (*uc)); if (run->exit_reason == KVM_EXIT_S390_SIEIC && run->s390_sieic.icptcode == 4 &&
@@ -34,13 +30,7 @@ uint64_t ucall_arch_get_ucall(struct kvm_vcpu *vcpu, struct ucall *uc) (run->s390_sieic.ipb >> 16) == 0x501) {int reg = run->s390_sieic.ipa &0xf ; -memcpy (&ucall,addr_gva2hva (vcpu->vm, run->s.regs.gprs[reg]), -sizeof (ucall)); - -vcpu_run_complete_io (vcpu); -if (uc) -memcpy (uc, &ucall,sizeof (ucall)); +return addr_gva2hva (vcpu->vm, run->s.regs.gprs[reg]); } - -return ucall.cmd; +return NULL ; }
tools/testing/selftests/kvm/lib/ucall_common.c +19 -0
@@ -18,3 +18,22 @@ void ucall(uint64_t cmd, int nargs, ...)ucall_arch_do_ucall ((vm_vaddr_t)&uc); } + +uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +{ +struct ucall ucall; +void *addr; + +if (!uc) + uc = &ucall; + + addr =ucall_arch_get_ucall (vcpu); +if (addr) { +memcpy (uc, addr,sizeof (*uc)); +vcpu_run_complete_io (vcpu); + }else { +memset (uc,0 ,sizeof (*uc)); + } + +return uc->cmd; +}
tools/testing/selftests/kvm/lib/x86_64/ucall.c +3 -13
@@ -22,25 +22,15 @@ void ucall_arch_do_ucall(vm_vaddr_t uc) : : [port] "d" (UCALL_PIO_PORT), "D" (uc) : "rax", "memory"); } -uint64_t ucall_arch_get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc) +void *ucall_arch_get_ucall (struct kvm_vcpu *vcpu) {struct kvm_run *run = vcpu->run; -struct ucall ucall = {}; - -if (uc) -memset (uc,0 ,sizeof (*uc));if (run->exit_reason ==KVM_EXIT_IO && run->io.port ==UCALL_PIO_PORT ) {struct kvm_regs regs;vcpu_regs_get (vcpu, ®s); -memcpy (&ucall,addr_gva2hva (vcpu->vm, (vm_vaddr_t )regs.rdi), -sizeof (ucall)); - -vcpu_run_complete_io (vcpu); -if (uc) -memcpy (uc, &ucall,sizeof (ucall)); +return addr_gva2hva (vcpu->vm, regs.rdi); } - -return ucall.cmd; +return NULL ; }
Signature · 1 line
2.37.1.559.g78731f0fdb-goog
Re: [V3 07/11] KVM: selftests: Consolidate boilerplate code in get_ucall()
Quoted 14 lines · Peter Gonda
From: Sean Christopherson <[email protected]>
Consolidate the actual copying of a ucall struct from guest=>host into the common
get_ucall(). Return a host virtual address instead of a guest virtual address even though theaddr_gva2hva()part could be moved toget_ucall()too. Conceptually,get_ucall()is invoked from the host and should return a host virtual address (and returning NULL for "nothing to see here" is far superior to returning 0).Use pointer shenanigans instead of an unnecessary bounce buffer when the caller of
get_ucall()provides a valid pointer.Signed-off-by Sean Christopherson <[email protected]>Signed-off-by Peter Gonda <[email protected]>
Quoted 9 lines
--- .../selftests/kvm/include/ucall_common.h | 8 ++------ .../testing/selftests/kvm/lib/aarch64/ucall.c | 14 +++----------- tools/testing/selftests/kvm/lib/riscv/ucall.c | 19 +++---------------- tools/testing/selftests/kvm/lib/s390x/ucall.c | 16 +++------------- .../testing/selftests/kvm/lib/ucall_common.c | 19 +++++++++++++++++++ .../testing/selftests/kvm/lib/x86_64/ucall.c | 16 +++------------- 6 files changed, 33 insertions(+), 59 deletions(-)
PATCH v3 8/11 KVM: selftests: add library for creating/interacting with SEV guests
In reply to Peter GondaFrom: Michael Roth <[email protected]>
Add interfaces to allow tests to create/manage SEV guests. The additional state associated with these guests is encapsulated in a new struct sev_vm, which is a light wrapper around struct kvm_vm. These VMs will use vm_set_memory_encryption() and vm_get_encrypted_phy_pages() under the covers to configure and sync up with the core kvm_util library on what should/shouldn't be treated as encrypted memory.
--- tools/testing/selftests/kvm/Makefile | 1 + .../selftests/kvm/include/kvm_util_base.h | 3 + .../selftests/kvm/include/x86_64/sev.h | 44 +++ tools/testing/selftests/kvm/lib/x86_64/sev.c | 251 ++++++++++++++++++ 4 files changed, 299 insertions(+) create mode 100644 tools/testing/selftests/kvm/include/x86_64/sev.h create mode 100644 tools/testing/selftests/kvm/lib/x86_64/sev.c
tools/testing/selftests/kvm/Makefile +1 -0
@@ -55,6 +55,7 @@ LIBKVM_x86_64 += lib/x86_64/processor.c LIBKVM_x86_64 += lib/x86_64/svm.c LIBKVM_x86_64 += lib/x86_64/ucall.c LIBKVM_x86_64 += lib/x86_64/vmx.c +LIBKVM_x86_64 += lib/x86_64/sev.c LIBKVM_aarch64 += lib/aarch64/gic.c LIBKVM_aarch64 += lib/aarch64/gic_v3.c
tools/testing/selftests/kvm/include/kvm_util_base.h +3 -0
@@ -68,6 +68,7 @@ struct userspace_mem_regions {/* Memory encryption policy/configuration. */ struct vm_memcrypt {bool enabled; +bool encrypted;int8_t enc_by_default;bool has_enc_bit;int8_t enc_bit;
@@ -816,6 +817,8 @@ vm_paddr_t addr_arch_gva2gpa(struct kvm_vm *vm, vm_vaddr_t gva);static inline vm_paddr_t addr_gva2gpa (struct kvm_vm *vm,vm_vaddr_t gva) { +TEST_ASSERT (!vm->memcrypt.encrypted, +"Encrypted guests have their page tables encrypted so gva2gpa conversions are not possible." );return addr_arch_gva2gpa (vm, gva); }
tools/testing/selftests/kvm/include/x86_64/sev.h new file +44 -0
44 lines not on this page · raw message
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +251 -0
251 lines not on this page · raw message
Signature · 1 line
2.37.1.559.g78731f0fdb-goog
Sean Christopherson
reviewerRe: [V3 08/11] KVM: selftests: add library for creating/interacting with SEV guests
tools/testing/selftests/kvm/include/x86_64/sev.h line 31
Quoted 7 lines · Peter Gonda
tools/testing/selftests/kvm/include/x86_64/sev.h new file +7
-0+enum { +SEV_GSTATE_UNINIT =0 , +SEV_GSTATE_LUPDATE , +SEV_GSTATE_LSECRET , +SEV_GSTATE_RUNNING , +}; +
Name the enum, e.g. enum sev_guest_state?
And s/GSTATE/GUEST? Ugh, AMD's documentation uses GSTATE.
But looking at the docs, I only see GSTATE_LAUNCH? Or does SEV have different status codes than -ES and/or -SNP?
tools/testing/selftests/kvm/lib/x86_64/sev.c line 30
Quoted 3 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +3
-0+struct kvm_vm *sev_get_vm (struct sev_vm *sev) +{ +return sev->vm;
Why bother with a wrapper?
tools/testing/selftests/kvm/lib/x86_64/sev.c line 34
Quoted 4 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +4
-0+} + +uint8_t sev_get_enc_bit (struct sev_vm *sev) +{
Same here, IMO it just obfuscates code with no real benefit. ANd it's inconsistent, e.g. why have a wrapper for enc_bit but not sev->fd?
Quoted 24 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +24
-0+return sev->enc_bit; +} + +void sev_ioctl (int sev_fd,int cmd,void *data) +{ +int ret; +struct sev_issue_cmd arg; + + arg.cmd = cmd; + arg.data = (unsigned long )data; + ret =ioctl (sev_fd,SEV_ISSUE_CMD , &arg); +TEST_ASSERT (ret ==0 , +"SEV ioctl %d failed, error: %d, fw_error: %d" , + cmd, ret, arg.error); +} + +void kvm_sev_ioctl (struct sev_vm *sev,int cmd,void *data) +{ +struct kvm_sev_cmd arg = {0 }; +int ret; + + arg.id = cmd; + arg.sev_fd = sev->fd; + arg.data = (__u64 )data;
tools/testing/selftests/kvm/lib/x86_64/sev.c line 68
Quoted 10 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +10
-0+ + ret =ioctl (sev->vm->fd,KVM_MEMORY_ENCRYPT_OP , &arg); +TEST_ASSERT (ret ==0 , +"SEV KVM ioctl %d failed, rc: %i errno: %i (%s), fw_error: %d" , + cmd, ret, errno,strerror (errno), arg.error); +} + +/* Local helpers. */ + +static void
Don't split here, e.g. a grep/search for the function, should also show the return type and any attributes, e.g. "static" vs. something else is typically much more interesting than the parameters (and parameters is not a fully solvable problem).
Quoted 1 line
+sev_register_user_region(struct
sev_vm*sev, void *hva,uint64_tsize)
Align like so:
static void sev_register_user_region(struct sev_vm *sev, void *hva,
uint64_t size)
or maybe even let it poke out.
Quoted 5 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +5
-0+{ +struct kvm_enc_region range = {0 }; +int ret; + +pr_debug ("%s: hva: %p, size: %lu\n" , __func__, hva, size);
tools/testing/selftests/kvm/lib/x86_64/sev.c line 84
Quoted 10 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +10
-0+ + range.addr = (__u64 )hva; + range.size = size; + + ret =ioctl (sev->vm->fd,KVM_MEMORY_ENCRYPT_REG_REGION , &range); +TEST_ASSERT (ret ==0 ,"failed to register user range, errno: %i\n" , errno); +} + +static void +sev_encrypt_phy_range (struct sev_vm *sev,vm_paddr_t gpa,uint64_t size)
Same thing here.
Quoted 10 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +10
-0+{ +struct kvm_sev_launch_update_data ksev_update_data = {0 }; + +pr_debug ("%s: addr: 0x%lx, size: %lu\n" , __func__, gpa, size); + + ksev_update_data.uaddr = (__u64 )addr_gpa2hva (sev->vm, gpa); + ksev_update_data.len = size; + +kvm_sev_ioctl (sev,KVM_SEV_LAUNCH_UPDATE_DATA , &ksev_update_data); +}
tools/testing/selftests/kvm/lib/x86_64/sev.c line 104
Quoted 10 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +10
-0+ +static void sev_encrypt (struct sev_vm *sev) +{ +const struct sparsebit *enc_phy_pages; +struct kvm_vm *vm = sev->vm; +sparsebit_idx_t pg =0 ; +vm_paddr_t gpa_start; +uint64_t memory_size; + +/* Only memslot 0 supported for now. */
Eww. Haven't looked at this in depth, but is there a way to avoid hardcoding the memslot in this code?
tools/testing/selftests/kvm/lib/x86_64/sev.c line 205
Quoted 4 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +4
-0+void sev_vm_launch (struct sev_vm *sev) +{ +struct kvm_sev_launch_start ksev_launch_start = {0 }; +struct kvm_sev_guest_status ksev_status = {0 };
Doesn't " = {};" do the same thing? And for the status, and any other cases where userspace is reading, wouldn't it be better from a test coverage perspective to _not_ zero the data? Hmm, though I suppose false passes are possible in that case...
tools/testing/selftests/kvm/lib/x86_64/sev.c line 208
Quoted 2 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +2
-0+/* Need to use ucall_shared for synchronization. */ +//ucall_init_ops(sev_get_vm(sev), NULL, &ucall_ops_halt);
Can this be deleted? If not, what's up?
Re: [V3 08/11] KVM: selftests: add library for creating/interacting with SEV guests
Quoted 11 lines · Sean Christopherson
tools/testing/selftests/kvm/include/x86_64/sev.h line 31
Quoted 7 lines · Peter Gonda
tools/testing/selftests/kvm/include/x86_64/sev.h new file +7
-0+enum { +SEV_GSTATE_UNINIT =0 , +SEV_GSTATE_LUPDATE , +SEV_GSTATE_LSECRET , +SEV_GSTATE_RUNNING , +}; +Name the enum, e.g. enum
sev_guest_state?
Done
Quoted 5 lines
And s/GSTATE/GUEST? Ugh, AMD's documentation uses GSTATE.
But looking at the docs, I only see
GSTATE_LAUNCH? Or does SEV have different status codes than -ES and/or -SNP?
SEV and SEV-ES have the same states, SNP has a different set.
See "Table 43. GSTATE Finite State Machine":
https://www.amd.com/system/files/TechDocs/55766_SEV-KM_API_Specification.pdf
Quoted 7 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c line 30
Quoted 4 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +3
-0+struct kvm_vm *sev_get_vm (struct sev_vm *sev) +{ +return sev->vm;Why bother with a wrapper?
Quoted 8 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c line 34
Quoted 4 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +4
-0+} + +uint8_t sev_get_enc_bit (struct sev_vm *sev) +{Same here, IMO it just obfuscates code with no real benefit. ANd it's inconsistent, e.g. why have a wrapper for
enc_bitbut notsev->fd?
Removed.
Quoted 39 lines
Quoted 24 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +24
-0+return sev->enc_bit; +} + +void sev_ioctl (int sev_fd,int cmd,void *data) +{ +int ret; +struct sev_issue_cmd arg; + + arg.cmd = cmd; + arg.data = (unsigned long )data; + ret =ioctl (sev_fd,SEV_ISSUE_CMD , &arg); +TEST_ASSERT (ret ==0 , +"SEV ioctl %d failed, error: %d, fw_error: %d" , + cmd, ret, arg.error); +} + +void kvm_sev_ioctl (struct sev_vm *sev,int cmd,void *data) +{ +struct kvm_sev_cmd arg = {0 }; +int ret; + + arg.id = cmd; + arg.sev_fd = sev->fd; + arg.data = (__u64 )data;tools/testing/selftests/kvm/lib/x86_64/sev.c line 68
Quoted 10 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +10
-0+ + ret =ioctl (sev->vm->fd,KVM_MEMORY_ENCRYPT_OP , &arg); +TEST_ASSERT (ret ==0 , +"SEV KVM ioctl %d failed, rc: %i errno: %i (%s), fw_error: %d" , + cmd, ret, errno,strerror (errno), arg.error); +} + +/* Local helpers. */ + +static voidDon't split here, e.g. a grep/search for the function, should also show the return type and any attributes, e.g. "static" vs. something else is typically much more interesting than the parameters (and parameters is not a fully solvable problem).
Quoted 8 lines
Quoted 1 line
+sev_register_user_region(struct
sev_vm*sev, void *hva,uint64_tsize)Align like so:
static void sev_register_user_region(struct
sev_vm*sev, void *hva,uint64_t size)or maybe even let it poke out.
Ah never thought about that, sounds good. Done.
Quoted 34 lines
Quoted 6 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +5
-0+{ +struct kvm_enc_region range = {0 }; +int ret; + +pr_debug ("%s: hva: %p, size: %lu\n" , __func__, hva, size);tools/testing/selftests/kvm/lib/x86_64/sev.c line 84
Quoted 10 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +10
-0+ + range.addr = (__u64 )hva; + range.size = size; + + ret =ioctl (sev->vm->fd,KVM_MEMORY_ENCRYPT_REG_REGION , &range); +TEST_ASSERT (ret ==0 ,"failed to register user range, errno: %i\n" , errno); +} + +static void +sev_encrypt_phy_range (struct sev_vm *sev,vm_paddr_t gpa,uint64_t size)Same thing here.
Quoted 15 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +15
-0+{ +struct kvm_sev_launch_update_data ksev_update_data = {0 }; + +pr_debug ("%s: addr: 0x%lx, size: %lu\n" , __func__, gpa, size); + + ksev_update_data.uaddr = (__u64 )addr_gpa2hva (sev->vm, gpa); + ksev_update_data.len = size; + +kvm_sev_ioctl (sev,KVM_SEV_LAUNCH_UPDATE_DATA , &ksev_update_data); +} + +static void sev_encrypt (struct sev_vm *sev) +{ +const struct sparsebit *enc_phy_pages; +struct kvm_vm *vm = sev->vm;
Quoted 8 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c line 104
Quoted 5 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +5
-0+sparsebit_idx_t pg =0 ; +vm_paddr_t gpa_start; +uint64_t memory_size; + +/* Only memslot 0 supported for now. */ Eww. Haven't looked at this in depth, but is there a way to avoid hardcoding the memslot in this code?
Done.
Quoted 9 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c line 205
Quoted 5 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +4
-0+void sev_vm_launch (struct sev_vm *sev) +{ +struct kvm_sev_launch_start ksev_launch_start = {0 }; +struct kvm_sev_guest_status ksev_status = {0 };Doesn't " = {};" do the same thing? And for the status, and any other cases where userspace is reading, wouldn't it be better from a test coverage perspective to _not_ zero the data? Hmm, though I suppose false passes are possible in that case...
We want to zero out kvm_sev_launch_start because it has main fields we explicitly want zero: the dh_* and session_* fields need to be set up correctly or zeroed. We can add a test for these later if we want, but it would be mostly testing userspace + the PSP.
We can not zero status I agree.
Quoted 5 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c line 208
Quoted 3 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c new file +2
-0+/* Need to use ucall_shared for synchronization. */ +//ucall_init_ops(sev_get_vm(sev), NULL, &ucall_ops_halt); Can this be deleted? If not, what's up?
Whoops, removed.
PATCH v3 9/11 tools: Add atomic_test_and_set_bit()
In reply to Peter Gondaatomic_test_and_set_bit() allows for atomic bitmap usage from KVM selftests.
--- tools/arch/x86/include/asm/atomic.h | 7 +++++++ tools/include/asm-generic/atomic-gcc.h | 15 +++++++++++++++ 2 files changed, 22 insertions(+)
tools/arch/x86/include/asm/atomic.h +7 -0
@@ -8,6 +8,7 @@#define LOCK_PREFIX "\n\tlock; " +#include <asm/asm.h> #include <asm/cmpxchg.h> /*
@@ -70,4 +71,10 @@ static __always_inline int atomic_cmpxchg(atomic_t *v, int old, int new) return cmpxchg(&v->counter, old, new); } +static inline int atomic_test_and_set_bit(long nr, unsigned long *addr) +{ + GEN_BINARY_RMWcc(LOCK_PREFIX __ASM_SIZE(bts), *addr, "Ir", nr, "%0", "c"); + +} + #endif /* _TOOLS_LINUX_ASM_X86_ATOMIC_H */
tools/include/asm-generic/atomic-gcc.h +15 -0
@@ -4,6 +4,7 @@#include <linux/compiler.h> #include <linux/types.h> +#include <linux/bitops.h> /* * Atomic operations that C can't guarantee us. Useful for
@@ -69,4 +70,18 @@ static inline int atomic_cmpxchg(atomic_t *v, int oldval, int newval) return cmpxchg(&(v)->counter, oldval, newval); } +static inline int atomic_test_and_set_bit (long nr,unsigned long *addr) +{ +long old, val; +unsigned long mask =BIT_MASK (nr); + + addr +=BIT_WORD (nr); + val =READ_ONCE (*addr); +if (val & mask) +return 1 ; + + old =cmpxchg (addr, val, val & mask); +return !!(old & mask); +} +#endif /* __TOOLS_ASM_GENERIC_ATOMIC_H */
Signature · 1 line
2.37.1.559.g78731f0fdb-goog
Sean Christopherson
reviewerRe: [V3 09/11] tools: Add atomic_test_and_set_bit()
Quoted 23 lines · Peter Gonda
atomic_test_and_set_bit()allows for atomic bitmap usage from KVM selftests.Signed-off-by Peter Gonda <[email protected]>Suggested-by Sean Christopherson <[email protected]>--- tools/arch/x86/include/asm/atomic.h | 7 +++++++ tools/include/asm-generic/atomic-gcc.h | 15 +++++++++++++++ 2 files changed, 22 insertions(+)tools/arch/x86/include/asm/atomic.h +1
-0@@ -8,6 +8,7 @@#define LOCK_PREFIX "\n\tlock; " +#include <asm/asm.h> #include <asm/cmpxchg.h> /*
tools/arch/x86/include/asm/atomic.h static __always_inline int atomic_cmpxchg(atomic_t *v, int old, int new) · line 77
Quoted 7 lines
tools/arch/x86/include/asm/atomic.h +4
-0@@ -70,4 +71,10 @@ static __always_inline int atomic_cmpxchg(atomic_t *v, int old, int new)return cmpxchg (&v->counter, old, new); } +static inline int atomic_test_and_set_bit (long nr,unsigned long *addr) +{ + GEN_BINARY_RMWcc(LOCK_PREFIX __ASM_SIZE(bts), *addr, "Ir", nr, "%0", "c"); +
Unnecessary newline.
Quoted 17 lines
tools/arch/x86/include/asm/atomic.h +2
-0+} +#endif /* _TOOLS_LINUX_ASM_X86_ATOMIC_H */ tools/include/asm-generic/atomic-gcc.h +1
-0@@ -4,6 +4,7 @@#include <linux/compiler.h> #include <linux/types.h> +#include <linux/bitops.h> /* * Atomic operations that C can't guarantee us. Useful for @@ -69,4 +70,18 @@ static inline int atomic_cmpxchg(atomic_t *v, int oldval, int newval)return cmpxchg (&(v)->counter, oldval, newval);
tools/include/asm-generic/atomic-gcc.h static inline int atomic_cmpxchg(atomic_t *v, int oldval, int newval) · line 80
Quoted 10 lines
tools/include/asm-generic/atomic-gcc.h +9
-0} +static inline int atomic_test_and_set_bit (long nr,unsigned long *addr) +{ +long old, val; +unsigned long mask =BIT_MASK (nr); + + addr +=BIT_WORD (nr); + val =READ_ONCE (*addr); +if (val & mask) +return 1 ;
Probably should drop the READ_ONCE() shortcut to stay consistent with the kernel proper.
https://lore.kernel.org/all/CAHk-=wgSNiT5qJX53RHtWECsUiFq6d6VWYNAvu71ViOEan07yw@mail.gmail.com
tools/include/asm-generic/atomic-gcc.h static inline int atomic_cmpxchg(atomic_t *v, int oldval, int newval) · line 82
Quoted 2 lines
tools/include/asm-generic/atomic-gcc.h +2
-0+ + old =cmpxchg (addr, val, val & mask);
This is wrong on two fronts: 1) cmpxchg() writes the entire new value, and 2) it fails if the old value is not an exact match with what's in memory. Bug #1 means that setting a bit will clear all existing bits, and bug #2 means that this will fail to set the bit if another atomic_test_and_set_bit() sneaks in between reading into "val" and doing the cmpxchg.
And obviously dropping the READ_ONCE() above makes cmpxchg impossible (not a coincidence, it's simply the wrong operation to use).
I believe what we want is:
unsigned long mask = BIT_MASK(nr); long old; old = __sync_fetch_and_or(addr, mask); return !!(old & mask);
Quoted 7 lines
tools/include/asm-generic/atomic-gcc.h +3
-0+ return !!(old & mask); +} +#endif /* __TOOLS_ASM_GENERIC_ATOMIC_H */ Signature · 1 line
2.37.1.559.g78731f0fdb-goog
PATCH v3 10/11 KVM: selftests: Add ucall pool based implementation
In reply to Peter GondaTo add support for encrypted, SEV, guests in the ucall framework introduce a new "ucall pool" implementation. This was suggested in the thread on "[RFC PATCH 00/10] KVM: selftests: Add support for test-selectable ucall implementations". Using a listed as suggested there doesn't work well because the list is setup using HVAs not GVAs so use a bitmap + array solution instead to get the same pool of ucall structs result.
This allows for guests with encryption enabled set up a pool of ucall structs in the guest's shared memory region.
--- .../selftests/kvm/include/kvm_util_base.h | 2 + .../selftests/kvm/include/ucall_common.h | 13 +-- .../testing/selftests/kvm/lib/aarch64/ucall.c | 10 +- tools/testing/selftests/kvm/lib/riscv/ucall.c | 5 +- tools/testing/selftests/kvm/lib/s390x/ucall.c | 5 +- .../testing/selftests/kvm/lib/ucall_common.c | 108 +++++++++++++++++- .../testing/selftests/kvm/lib/x86_64/ucall.c | 6 +- 7 files changed, 131 insertions(+), 18 deletions(-)
tools/testing/selftests/kvm/include/kvm_util_base.h +2 -0
@@ -103,6 +103,8 @@ struct kvm_vm {int stats_fd;struct kvm_stats_header stats_header;struct kvm_stats_desc *stats_desc; + +bool use_ucall_pool; };
tools/testing/selftests/kvm/include/ucall_common.h +5 -8
@@ -22,6 +22,9 @@ enum {struct ucall {uint64_t cmd;uint64_t args[UCALL_MAX_ARGS ]; + +/* For ucall pool usage. */ +struct ucall *hva; };void ucall_arch_init (struct kvm_vm *vm,void *arg);
@@ -32,15 +35,9 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu);void ucall (uint64_t cmd,int nargs, ...);uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc); -static inline void ucall_init (struct kvm_vm *vm,void *arg) -{ -ucall_arch_init (vm, arg); -} +void ucall_init (struct kvm_vm *vm,void *arg); -static inline void ucall_uninit (struct kvm_vm *vm) -{ -ucall_arch_uninit (vm); -} +void ucall_uninit (struct kvm_vm *vm);#define GUEST_SYNC_ARGS (stage, arg1, arg2, arg3, arg4) \ ucall(UCALL_SYNC, 6, "hello", stage, arg1, arg2, arg3, arg4)
tools/testing/selftests/kvm/lib/aarch64/ucall.c +7 -3
@@ -81,12 +81,16 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu)if (run->exit_reason ==KVM_EXIT_MMIO && run->mmio.phys_addr == (uint64_t )ucall_exit_mmio_addr) { -vm_vaddr_t gva; +uint64_t ucall_addr;TEST_ASSERT (run->mmio.is_write && run->mmio.len ==8 ,"Unexpected ucall exit mmio address access" ); -memcpy (&gva, run->mmio.data,sizeof (gva)); -return addr_gva2hva (vcpu->vm, gva); +memcpy (&ucall_addr, run->mmio.data,sizeof (ucall_addr)); + +if (vcpu->vm->use_ucall_pool) +return (void *)ucall_addr; +else +return addr_gva2hva (vcpu->vm, ucall_addr); }return NULL ;
tools/testing/selftests/kvm/lib/riscv/ucall.c +4 -1
@@ -59,7 +59,10 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) run->riscv_sbi.extension_id == KVM_RISCV_SELFTESTS_SBI_EXT) {switch (run->riscv_sbi.function_id) {case KVM_RISCV_SELFTESTS_SBI_UCALL : -return addr_gva2hva (vcpu->vm, run->riscv_sbi.args[0 ]); +if (vcpu->vm->use_ucall_pool) +return (void *)run->riscv_sbi.args[0 ]; +else +return addr_gva2hva (vcpu->vm, run->riscv_sbi.args[0 ]);case KVM_RISCV_SELFTESTS_SBI_UNEXP :vcpu_dump (stderr, vcpu,2 );TEST_ASSERT (0 ,"Unexpected trap taken by guest" );
tools/testing/selftests/kvm/lib/s390x/ucall.c +4 -1
@@ -30,7 +30,10 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) (run->s390_sieic.ipb >> 16) == 0x501) {int reg = run->s390_sieic.ipa &0xf ; -return addr_gva2hva (vcpu->vm, run->s.regs.gprs[reg]); +if (vcpu->vm->use_ucall_pool) +return (void *)run->s.regs.gprs[reg]; +else +return addr_gva2hva (vcpu->vm, run->s.regs.gprs[reg]); }return NULL ; }
tools/testing/selftests/kvm/lib/ucall_common.c +104 -4
126 lines not on this page · raw message
tools/testing/selftests/kvm/lib/x86_64/ucall.c +5 -1
@@ -30,7 +30,11 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu)struct kvm_regs regs;vcpu_regs_get (vcpu, ®s); -return addr_gva2hva (vcpu->vm, regs.rdi); + +if (vcpu->vm->use_ucall_pool) +return (void *)regs.rdi; +else +return addr_gva2hva (vcpu->vm, regs.rdi); }return NULL ; }
Signature · 1 line
2.37.1.559.g78731f0fdb-goog
Re: [V3 10/11] KVM: selftests: Add ucall pool based implementation
Quoted 4 lines · Peter Gonda
To add support for encrypted, SEV, guests in the ucall framework introduce a new "ucall pool" implementation. This was suggested in the thread on "[RFC PATCH 00/10] KVM: selftests: Add support for test-selectable ucall implementations". Using a listed as suggested
s/listed/list/
Quoted 5 lines
there doesn't work well because the list is setup using HVAs not GVAs so use a bitmap + array solution instead to get the same pool of ucall structs result.
This allows for guests with encryption enabled set up a pool of ucall
to set up
Quoted 64 lines
structs in the guest's shared memory region.
Suggested-by Sean Christopherson <[email protected]>Signed-off-by Peter Gonda <[email protected]>--- .../selftests/kvm/include/kvm_util_base.h | 2 + .../selftests/kvm/include/ucall_common.h | 13 +-- .../testing/selftests/kvm/lib/aarch64/ucall.c | 10 +- tools/testing/selftests/kvm/lib/riscv/ucall.c | 5 +- tools/testing/selftests/kvm/lib/s390x/ucall.c | 5 +- .../testing/selftests/kvm/lib/ucall_common.c | 108 +++++++++++++++++- .../testing/selftests/kvm/lib/x86_64/ucall.c | 6 +- 7 files changed, 131 insertions(+), 18 deletions(-)tools/testing/selftests/kvm/include/kvm_util_base.h +2
-0@@ -103,6 +103,8 @@ struct kvm_vm {int stats_fd;struct kvm_stats_header stats_header;struct kvm_stats_desc *stats_desc; + +bool use_ucall_pool; };tools/testing/selftests/kvm/include/ucall_common.h +5
-8@@ -22,6 +22,9 @@ enum {struct ucall {uint64_t cmd;uint64_t args[UCALL_MAX_ARGS ]; + +/* For ucall pool usage. */ +struct ucall *hva; };void ucall_arch_init (struct kvm_vm *vm,void *arg);@@ -32,15 +35,9 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu);void ucall (uint64_t cmd,int nargs, ...);uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc); -static inline void ucall_init (struct kvm_vm *vm,void *arg) -{ -ucall_arch_init (vm, arg); -} +void ucall_init (struct kvm_vm *vm,void *arg); -static inline void ucall_uninit (struct kvm_vm *vm) -{ -ucall_arch_uninit (vm); -} +void ucall_uninit (struct kvm_vm *vm);#define GUEST_SYNC_ARGS (stage, arg1, arg2, arg3, arg4) \ ucall(UCALL_SYNC, 6, "hello", stage, arg1, arg2, arg3, arg4)tools/testing/selftests/kvm/lib/aarch64/ucall.c +0
-0
tools/testing/selftests/kvm/lib/aarch64/ucall.c void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) · line 84
Quoted 5 lines
tools/testing/selftests/kvm/lib/aarch64/ucall.c +1
-1@@ -81,12 +81,16 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu)if (run->exit_reason ==KVM_EXIT_MMIO && run->mmio.phys_addr == (uint64_t )ucall_exit_mmio_addr) { -vm_vaddr_t gva; +uint64_t ucall_addr;
Why change this vm_vaddr_t to a uint64_t? We shouldn't, because...
tools/testing/selftests/kvm/lib/aarch64/ucall.c void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) · line 88
Quoted 6 lines
tools/testing/selftests/kvm/lib/aarch64/ucall.c +1
-2TEST_ASSERT (run->mmio.is_write && run->mmio.len ==8 ,"Unexpected ucall exit mmio address access" ); -memcpy (&gva, run->mmio.data,sizeof (gva)); -return addr_gva2hva (vcpu->vm, gva); +memcpy (&ucall_addr, run->mmio.data,sizeof (ucall_addr));
...here we assume it's a vm_vaddr_t and only...
tools/testing/selftests/kvm/lib/aarch64/ucall.c void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) · line 91
Quoted 3 lines
tools/testing/selftests/kvm/lib/aarch64/ucall.c +3
-0+ +if (vcpu->vm->use_ucall_pool) +return (void *)ucall_addr;
...here do we know otherwise. So only here should be any casting.
Also, I think here and in all the ucall_arch_get_ucall() functions we should add a comment explaining a host-shared address is used, which is why we don't need addr_gva2hva()
Quoted 10 lines
tools/testing/selftests/kvm/lib/aarch64/ucall.c +2
-0+else +return addr_gva2hva (vcpu->vm, ucall_addr); }return NULL ;tools/testing/selftests/kvm/lib/riscv/ucall.c +0
-0
tools/testing/selftests/kvm/lib/riscv/ucall.c void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) · line 65
Quoted 8 lines
tools/testing/selftests/kvm/lib/riscv/ucall.c +4
-1@@ -59,7 +59,10 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) run->riscv_sbi.extension_id == KVM_RISCV_SELFTESTS_SBI_EXT) {switch (run->riscv_sbi.function_id) {case KVM_RISCV_SELFTESTS_SBI_UCALL : -return addr_gva2hva (vcpu->vm, run->riscv_sbi.args[0 ]); +if (vcpu->vm->use_ucall_pool) +return (void *)run->riscv_sbi.args[0 ]; +else +return addr_gva2hva (vcpu->vm, run->riscv_sbi.args[0 ]);
Using vm_vaddr_t gva variable for run->riscv_sbi.args[0] like aarch64 does for it's address would look a bit nicer.
Quoted 8 lines
tools/testing/selftests/kvm/lib/riscv/ucall.c +0
-0case KVM_RISCV_SELFTESTS_SBI_UNEXP :vcpu_dump (stderr, vcpu,2 );TEST_ASSERT (0 ,"Unexpected trap taken by guest" );tools/testing/selftests/kvm/lib/s390x/ucall.c +0
-0
tools/testing/selftests/kvm/lib/s390x/ucall.c void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) · line 36
Quoted 8 lines
tools/testing/selftests/kvm/lib/s390x/ucall.c +4
-1@@ -30,7 +30,10 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) (run->s390_sieic.ipb >> 16) == 0x501) {int reg = run->s390_sieic.ipa &0xf ; -return addr_gva2hva (vcpu->vm, run->s.regs.gprs[reg]); +if (vcpu->vm->use_ucall_pool) +return (void *)run->s.regs.gprs[reg]; +else +return addr_gva2hva (vcpu->vm, run->s.regs.gprs[reg]);
Same comment as for riscv.
Quoted 10 lines
tools/testing/selftests/kvm/lib/s390x/ucall.c +0
-0}return NULL ; }tools/testing/selftests/kvm/lib/ucall_common.c +0
-0@@ -1,22 +1,122 @@// SPDX-License-Identifier: GPL-2.0-only #include "kvm_util.h"
tools/testing/selftests/kvm/lib/ucall_common.c line 12
Quoted 10 lines
tools/testing/selftests/kvm/lib/ucall_common.c +10
-0+#include "linux/types.h" +#include "linux/bitmap.h" +#include "linux/atomic.h" + +struct ucall_header { + DECLARE_BITMAP(in_use, KVM_MAX_VCPUS); +struct ucall ucalls[KVM_MAX_VCPUS ]; +}; + +static bool use_ucall_pool;
I don't think we need this boolean. It will always be true when ucall_pool is non-null and always false with ucall_pool is null. So we can just test ucall_pool.
Quoted 5 lines
tools/testing/selftests/kvm/lib/ucall_common.c +5
-0+static struct ucall_header *ucall_pool; + +void ucall_init (struct kvm_vm *vm,void *arg) +{ +struct ucall *uc;
tools/testing/selftests/kvm/lib/ucall_common.c line 27
Quoted 10 lines
tools/testing/selftests/kvm/lib/ucall_common.c +10
-0+struct ucall_header *hdr; +vm_vaddr_t vaddr; +int i; + + use_ucall_pool = vm->use_ucall_pool; +sync_global_to_guest (vm, use_ucall_pool); +if (!use_ucall_pool) + goto out; + +TEST_ASSERT (!ucall_pool,"Only a single encrypted guest at a time for ucalls." );
"Only a single VM may use a ucall pool at a time"
Quoted 12 lines
tools/testing/selftests/kvm/lib/ucall_common.c +12
-0+ vaddr =vm_vaddr_alloc_shared (vm,sizeof (*hdr), vm->page_size); + hdr = (struct ucall_header *)addr_gva2hva (vm, vaddr); +memset (hdr,0 ,sizeof (*hdr)); + +for (i =0 ; i <KVM_MAX_VCPUS ; ++i) { + uc = &hdr->ucalls[i]; + uc->hva = uc; + } + + ucall_pool = (struct ucall_header *)vaddr; +sync_global_to_guest (vm, ucall_pool); +
tools/testing/selftests/kvm/lib/ucall_common.c line 49
Quoted 10 lines
tools/testing/selftests/kvm/lib/ucall_common.c +10
-0+out: +ucall_arch_init (vm, arg); +} + +void ucall_uninit (struct kvm_vm *vm) +{ + use_ucall_pool = false; + ucall_pool =NULL ; + +if (!vm->memcrypt.encrypted) {
Why is this condition here?
Quoted 9 lines
tools/testing/selftests/kvm/lib/ucall_common.c +9
-0+sync_global_to_guest (vm, use_ucall_pool); +sync_global_to_guest (vm, ucall_pool); + } + +ucall_arch_uninit (vm); +} + +static struct ucall *ucall_alloc (void ) +{
tools/testing/selftests/kvm/lib/ucall_common.c line 68
Quoted 10 lines
tools/testing/selftests/kvm/lib/ucall_common.c +10
-0+struct ucall *uc =NULL ; +int i; + +if (!use_ucall_pool) + goto out; + +for (i =0 ; i <KVM_MAX_VCPUS ; ++i) { +if (!atomic_test_and_set_bit (i, ucall_pool->in_use)) { + uc = &ucall_pool->ucalls[i]; +memset (uc->args,0 ,sizeof (uc->args));
Do we need to zero uc? If so, what about uc->cmd?
tools/testing/selftests/kvm/lib/ucall_common.c line 71
Quoted 3 lines
tools/testing/selftests/kvm/lib/ucall_common.c +3
-0+break ; + } + }
Need a blank line here
Quoted 75 lines
tools/testing/selftests/kvm/lib/ucall_common.c +35
-4+out: +return uc; +} + +static inline size_t uc_pool_idx (struct ucall *uc) +{ +return uc->hva - ucall_pool->ucalls; +} + +static void ucall_free (struct ucall *uc) +{ +if (!use_ucall_pool) +return ; + +clear_bit (uc_pool_idx (uc), ucall_pool->in_use); +}void ucall (uint64_t cmd,int nargs, ...) { -struct ucall uc = {}; +struct ucall *uc; +struct ucall tmp = {};va_list va;int i; -WRITE_ONCE (uc.cmd, cmd); + uc =ucall_alloc (); +if (!uc) + uc = &tmp; + +WRITE_ONCE (uc->cmd, cmd); nargs =min (nargs,UCALL_MAX_ARGS );va_start (va, nargs);for (i =0 ; i < nargs; ++i) -WRITE_ONCE (uc.args[i],va_arg (va, uint64_t)); +WRITE_ONCE (uc->args[i],va_arg (va, uint64_t));va_end (va); -ucall_arch_do_ucall ((vm_vaddr_t)&uc); +/* +* When using the ucall pool implementation the @hva member of the ucall +* structs in the pool has been initialized to the hva of the ucall +* object. +*/ +if (use_ucall_pool) +ucall_arch_do_ucall ((vm_vaddr_t )uc->hva); +else +ucall_arch_do_ucall ((vm_vaddr_t )uc); + +ucall_free (uc); }uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc)tools/testing/selftests/kvm/lib/x86_64/ucall.c +5
-1@@ -30,7 +30,11 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu)struct kvm_regs regs;vcpu_regs_get (vcpu, ®s); -return addr_gva2hva (vcpu->vm, regs.rdi); + +if (vcpu->vm->use_ucall_pool) +return (void *)regs.rdi; +else +return addr_gva2hva (vcpu->vm, regs.rdi); }return NULL ; }Signature · 1 line
2.37.1.559.g78731f0fdb-goog
Thanks,
drew
Sean Christopherson
reviewerRe: [V3 10/11] KVM: selftests: Add ucall pool based implementation
Quoted 23 lines · Andrew Jones
tools/testing/selftests/kvm/lib/aarch64/ucall.c void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) · line 84
Quoted 10 lines · Peter Gonda
tools/testing/selftests/kvm/lib/aarch64/ucall.c +1
-1@@ -81,12 +81,16 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu)if (run->exit_reason ==KVM_EXIT_MMIO && run->mmio.phys_addr == (uint64_t )ucall_exit_mmio_addr) { -vm_vaddr_t gva; +uint64_t ucall_addr;Why change this
vm_vaddr_tto auint64_t? We shouldn't, because...tools/testing/selftests/kvm/lib/aarch64/ucall.c void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) · line 88
Quoted 6 lines
tools/testing/selftests/kvm/lib/aarch64/ucall.c +1
-2TEST_ASSERT (run->mmio.is_write && run->mmio.len ==8 ,"Unexpected ucall exit mmio address access" ); -memcpy (&gva, run->mmio.data,sizeof (gva)); -return addr_gva2hva (vcpu->vm, gva); +memcpy (&ucall_addr, run->mmio.data,sizeof (ucall_addr));...here we assume it's a
vm_vaddr_tand only...
Quoted 5 lines
tools/testing/selftests/kvm/lib/aarch64/ucall.c void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) · line 91
Quoted 3 lines
tools/testing/selftests/kvm/lib/aarch64/ucall.c +3
-0+ +if (vcpu->vm->use_ucall_pool) +return (void *)ucall_addr;...here do we know otherwise. So only here should be any casting.
It technically should be a union, because if sizeof(vm_vaddr_t) < sizeof(void *) then declaring it as a vm_addr_t will do the wrong thing. But then it's possible that this could read too many bytes and inducs failure. So I guess what we really need is a "static_assert(sizeof(vm_vaddr_t) == sizeof(void *))".
But why is "use_ucall_pool" optional? Unless there's a use case that fundamentally conflicts with the pool approach, let's make the pool approach the _only_ approach. IIRC, ARM's implementation isn't thread safe, i.e. there's at least one other use case that _needs_ the pool implementation.
By supporting both, we are signing ourselves up for extra maintenance and pain, e.g. inevitably we'll break whichever option isn't the default and not notice for quite some time.
Re: [V3 10/11] KVM: selftests: Add ucall pool based implementation
Quoted 30 lines · Sean Christopherson
Quoted 23 lines · Andrew Jones
tools/testing/selftests/kvm/lib/aarch64/ucall.c void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) · line 84
Quoted 10 lines · Peter Gonda
tools/testing/selftests/kvm/lib/aarch64/ucall.c +1
-1@@ -81,12 +81,16 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu)if (run->exit_reason ==KVM_EXIT_MMIO && run->mmio.phys_addr == (uint64_t )ucall_exit_mmio_addr) { -vm_vaddr_t gva; +uint64_t ucall_addr;Why change this
vm_vaddr_tto auint64_t? We shouldn't, because...tools/testing/selftests/kvm/lib/aarch64/ucall.c void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) · line 88
Quoted 6 lines
tools/testing/selftests/kvm/lib/aarch64/ucall.c +1
-2TEST_ASSERT (run->mmio.is_write && run->mmio.len ==8 ,"Unexpected ucall exit mmio address access" ); -memcpy (&gva, run->mmio.data,sizeof (gva)); -return addr_gva2hva (vcpu->vm, gva); +memcpy (&ucall_addr, run->mmio.data,sizeof (ucall_addr));...here we assume it's a
vm_vaddr_tand only...Quoted 5 lines
tools/testing/selftests/kvm/lib/aarch64/ucall.c void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) · line 91
Quoted 3 lines
tools/testing/selftests/kvm/lib/aarch64/ucall.c +3
-0+ +if (vcpu->vm->use_ucall_pool) +return (void *)ucall_addr;...here do we know otherwise. So only here should be any casting.
Quoted 4 lines
It technically should be a union, because if
sizeof(vm_vaddr_t)<sizeof(void *)then declaring it as avm_addr_twill do the wrong thing. But then it's possible that this could read too many bytes and inducs failure. So I guess what we really need is a "static_assert(sizeof(vm_vaddr_t) == sizeof(void *))".
ack
Quoted 5 lines
But why is "
use_ucall_pool" optional? Unless there's a use case that fundamentally conflicts with the pool approach, let's make the pool approach the _only_ approach. IIRC, ARM's implementation isn't thread safe, i.e. there's at least one other use case that _needs_ the pool implementation.
Really? The ucall structure is on the vcpu's stack like the other architectures. Ah, you're probably thinking about the shared address used to exit to userspace. The address doesn't matter as long as no VM maps it, but, yes, a multi-VM test where the VMs have different maps could end up breaking ucalls for one or more VMs. It wouldn't be hard to make that address per-VM, though, if ever necessary.
Quoted 4 lines
By supporting both, we are signing ourselves up for extra maintenance and pain, e.g. inevitably we'll break whichever option isn't the default and not notice for quite some time.
uc pools are currently limited to a single VM. That could be changed, but at the expense of even more code to maintain. The simple uc implementation is, well, simple, and also supports multiple VMs. I'd prefer we keep that one and keep it as the default.
Thanks,
drew
Sean Christopherson
reviewerRe: [V3 10/11] KVM: selftests: Add ucall pool based implementation
Quoted 11 lines · Andrew Jones
Quoted 4 lines · Sean Christopherson
But why is "
use_ucall_pool" optional? Unless there's a use case that fundamentally conflicts with the pool approach, let's make the pool approach the _only_ approach. IIRC, ARM's implementation isn't thread safe, i.e. there's at least one other use case that _needs_ the pool implementation.Really? The ucall structure is on the vcpu's stack like the other architectures. Ah, you're probably thinking about the shared address used to exit to userspace. The address doesn't matter as long as no VM maps it, but, yes, a multi-VM test where the VMs have different maps could end up breaking ucalls for one or more VMs.
Yah, that's what I was thinking of.
Quoted 9 lines
It wouldn't be hard to make that address per-VM, though, if ever necessary.
Quoted 4 lines
By supporting both, we are signing ourselves up for extra maintenance and pain, e.g. inevitably we'll break whichever option isn't the default and not notice for quite some time.
uc pools are currently limited to a single VM. That could be changed, but at the expense of even more code to maintain.
Unless I'm missing something, it's actually less code. "globals" that are sync'd to the guest aren't truly global, each VM has a separate physical page that is a copy of the global, hence the need for sync_global_to_guest().
And FWIW, the code is actually "safe" in practice because I doubt any test spawns multiple VMs in parallel, i.e. the host might get confused over the value of ucall_pool, but guests won't stomp on each other so long as they're created serially. Ditto for ARM's ucall_exit_mmio_addr.
To make this truly thread safe, we just need a way to atomically sync the pointer to the guest, and that's easy enough to add.
With that out of the way, all of the conditional "use pool" code goes away, which IMO simplifies things overall. Using a pool versus stack memory isn't _that_ much more complicated, and we _know_ the stack approach doesn't work for all VM types.
Indeed, this partial conversion is:
7 files changed, 131 insertions(+), 18 deletions(-)
versus a full conversion:
6 files changed, 89 insertions(+), 20 deletions(-)
And simplification is only a secondary concern, what I'm really worried about is things bitrotting and then some poor sod having to wade through unrelated issues because somebody else broke the pool implementation and didn't notice.
Argh, case in point, none of the x86 (or s390 or RISC-V) tests do ucall_init(), which would have been a lurking bug if any of them tried to switch to the pool.
Argh + case in point #2, this code is broken, and that bug may have sat unnoticed due to only the esoteric SEV test using the pool.
-static inline size_t uc_pool_idx(struct ucall *uc) +static noinline void ucall_free(struct ucall *uc)
{
- return uc->hva - ucall_pool->ucalls;
-}
-
-static void ucall_free(struct ucall *uc) -{
- clear_bit(uc_pool_idx(uc), ucall_pool->in_use); + /* Beware, here be pointer arithmetic. */ + clear_bit(uc - ucall_pool->ucalls, ucall_pool->in_use);
}
So my very strong vote is to fully switch to a single implementation. That way our code either works or it doesn't, i.e. we notice very quickly if it breaks.
Peter, if I can convince Drew of One Pool To Rule Them All, can you slot in the attached patches and slightly rework the ordering so that all SEV patches are at the end? E.g. something like:
KVM: selftests: Automatically do init_ucall() for non-barebones VMs
KVM: selftests: move vm_phy_pages_alloc() earlier in file
KVM: selftests: sparsebit: add const where appropriate
KVM: selftests: add hooks for managing encrypted guest memory
KVM: selftests: handle encryption bits in page tables
KVM: selftests: add support for encrypted vm_vaddr_* allocations
KVM: selftests: Consolidate common code for popuplating
KVM: selftests: Consolidate boilerplate code in get_ucall()
tools: Add atomic_test_and_set_bit()
KVM: selftests: Add macro to atomically sync per-VM "global" pointers
KVM: selftests: Add ucall pool based implementation
KVM: selftests: add library for creating/interacting with SEV guests
KVM: selftests: Add simple sev vm testing
The patches are tested on x86 and compile tested on arm. I can provide more testing if/when necessary. I also haven't addressed any other feedback in the "ucall pool" patch, though I'm guessing much of it no longer applies.
From 9264f40a8cff865e5845d89e8e44e0c3a6fa482a Mon Sep 17 00:00:00 2001
From: Sean Christopherson <[email protected]>
Date: Thu, 18 Aug 2022 15:15:57 -0700
Subject: [PATCH 01/13] KVM: selftests: Automatically do init_ucall() for
non-barebones VMs
Do init_ucall() automatically during VM creation in preparation for making ucall initialization meaningful on all architectures. aarch64 is currently the only arch that needs to do any setup, but that will change in the future by switching to a pool-based implementation (instead of the current stack-based approach).
--- tools/testing/selftests/kvm/aarch64/arch_timer.c | 1 - tools/testing/selftests/kvm/aarch64/debug-exceptions.c | 1 - tools/testing/selftests/kvm/aarch64/hypercalls.c | 1 - tools/testing/selftests/kvm/aarch64/psci_test.c | 1 - tools/testing/selftests/kvm/aarch64/vgic_init.c | 2 -- tools/testing/selftests/kvm/aarch64/vgic_irq.c | 1 - tools/testing/selftests/kvm/dirty_log_test.c | 2 -- tools/testing/selftests/kvm/kvm_page_table_test.c | 1 - tools/testing/selftests/kvm/lib/kvm_util.c | 3 +++ tools/testing/selftests/kvm/lib/perf_test_util.c | 2 -- tools/testing/selftests/kvm/memslot_perf_test.c | 1 - tools/testing/selftests/kvm/rseq_test.c | 1 - tools/testing/selftests/kvm/steal_time.c | 1 - tools/testing/selftests/kvm/system_counter_offset_test.c | 1 - 14 files changed, 3 insertions(+), 16 deletions(-)
tools/testing/selftests/kvm/aarch64/arch_timer.c +0 -1
@@ -375,7 +375,6 @@ static struct kvm_vm *test_vm_create(void)for (i =0 ; i < nr_vcpus; i++)vcpu_init_descriptor_tables (vcpus[i]); -ucall_init (vm,NULL );test_init_timer_irq (vm); gic_fd =vgic_v3_setup (vm, nr_vcpus,64 ,GICD_BASE_GPA ,GICR_BASE_GPA );__TEST_REQUIRE (gic_fd >=0 ,"Failed to create vgic-v3" );
tools/testing/selftests/kvm/aarch64/debug-exceptions.c +0 -1
@@ -254,7 +254,6 @@ int main(int argc, char *argv[])int stage; vm =vm_create_with_one_vcpu (&vcpu, guest_code); -ucall_init (vm,NULL );vm_init_descriptor_tables (vm);vcpu_init_descriptor_tables (vcpu);
tools/testing/selftests/kvm/aarch64/hypercalls.c +0 -1
@@ -236,7 +236,6 @@ static struct kvm_vm *test_vm_create(struct kvm_vcpu **vcpu) vm =vm_create_with_one_vcpu (vcpu, guest_code); -ucall_init (vm,NULL );steal_time_init (*vcpu);return vm;
tools/testing/selftests/kvm/aarch64/psci_test.c +0 -1
@@ -77,7 +77,6 @@ static struct kvm_vm *setup_vm(void *guest_code, struct kvm_vcpu **source,struct kvm_vm *vm; vm =vm_create (2 ); -ucall_init (vm,NULL );vm_ioctl (vm,KVM_ARM_PREFERRED_TARGET , &init); init.features[0 ] |= (1 <<KVM_ARM_VCPU_PSCI_0_2 );
tools/testing/selftests/kvm/aarch64/vgic_init.c +0 -2
@@ -68,8 +68,6 @@ static void guest_code(void)/* we don't want to assert on run execution, hence that helper */ static int run_vcpu (struct kvm_vcpu *vcpu) { -ucall_init (vcpu->vm,NULL ); -return __vcpu_run (vcpu) ? -errno :0 ; }
tools/testing/selftests/kvm/aarch64/vgic_irq.c +0 -1
@@ -756,7 +756,6 @@ static void test_vgic(uint32_t nr_irqs, bool level_sensitive, bool eoi_split)print_args (&args); vm =vm_create_with_one_vcpu (&vcpu, guest_code); -ucall_init (vm,NULL );vm_init_descriptor_tables (vm);vcpu_init_descriptor_tables (vcpu);
tools/testing/selftests/kvm/dirty_log_test.c +0 -2
@@ -754,8 +754,6 @@ static void run_test(enum vm_guest_mode mode, void *arg)/* Cache the HVA pointer of the region */ host_test_mem =addr_gpa2hva (vm, (vm_paddr_t )guest_test_phys_mem); -ucall_init (vm,NULL ); -/* Export the shared variables to the guest */ sync_global_to_guest (vm, host_page_size);sync_global_to_guest (vm, guest_page_size);
tools/testing/selftests/kvm/kvm_page_table_test.c +0 -1
@@ -289,7 +289,6 @@ static struct kvm_vm *pre_init_before_test(enum vm_guest_mode mode, void *arg) host_test_mem =addr_gpa2hva (vm, (vm_paddr_t )guest_test_phys_mem);/* Export shared structure test_args to guest */ -ucall_init (vm,NULL );sync_global_to_guest (vm, test_args); ret =sem_init (&test_stage_updated,0 ,0 );
tools/testing/selftests/kvm/lib/kvm_util.c +3 -0
@@ -301,6 +301,9 @@ struct kvm_vm *__vm_create(enum vm_guest_mode mode, uint32_t nr_runnable_vcpus,#ifdef __x86_64__vm_create_irqchip (vm);#endif + +ucall_init (vm,NULL ); +return vm; }
tools/testing/selftests/kvm/lib/perf_test_util.c +0 -2
@@ -209,8 +209,6 @@ struct kvm_vm *perf_test_create_vm(enum vm_guest_mode mode, int nr_vcpus,perf_test_setup_nested (vm, nr_vcpus, vcpus); } -ucall_init (vm,NULL ); -/* Export the shared variables to the guest. */ sync_global_to_guest (vm, perf_test_args);
tools/testing/selftests/kvm/memslot_perf_test.c +0 -1
@@ -277,7 +277,6 @@ static bool prepare_vm(struct vm_data *data, int nslots, uint64_t *maxslots,TEST_ASSERT (data->hva_slots,"malloc() fail" ); data->vm =__vm_create_with_one_vcpu (&data->vcpu, mempages, guest_code); -ucall_init (data->vm,NULL );pr_info_v ("Adding slots 1..%i, each slot with %" PRIu64" pages + %" PRIu64" extra pages last\n" , max_mem_slots -1 , data->pages_per_slot, rempages);
tools/testing/selftests/kvm/rseq_test.c +0 -1
@@ -224,7 +224,6 @@ int main(int argc, char *argv[])* CPU affinity. */ vm =vm_create_with_one_vcpu (&vcpu, guest_code); -ucall_init (vm,NULL );pthread_create (&migration_thread,NULL , migration_worker, (void *)(unsigned long )gettid ());
tools/testing/selftests/kvm/steal_time.c +0 -1
@@ -266,7 +266,6 @@ int main(int ac, char **av) gpages =vm_calc_num_guest_pages (VM_MODE_DEFAULT ,STEAL_TIME_SIZE *NR_VCPUS );vm_userspace_mem_region_add (vm,VM_MEM_SRC_ANONYMOUS ,ST_GPA_BASE ,1 , gpages,0 );virt_map (vm,ST_GPA_BASE ,ST_GPA_BASE , gpages); -ucall_init (vm,NULL );TEST_REQUIRE (is_steal_time_supported (vcpus[0 ]));
tools/testing/selftests/kvm/system_counter_offset_test.c +0 -1
@@ -121,7 +121,6 @@ int main(void) vm =vm_create_with_one_vcpu (&vcpu, guest_main);check_preconditions (vcpu); -ucall_init (vm,NULL );enter_guest (vcpu);kvm_vm_free (vm);
Signature · 1 line
2.37.1.595.g718a3a8f04-goog
From bbc527a48fe0481495d6a0a562815e3d33fb655b Mon Sep 17 00:00:00 2001
From: Sean Christopherson <[email protected]>
Date: Thu, 18 Aug 2022 15:22:55 -0700
Subject: [PATCH 10/13] KVM: selftests: Add macro to atomically sync per-VM
"global" pointers
Add atomic_sync_global_pointer_to_guest() to allow sync'ing "global" pointers that hold per-VM values, i.e. technically need to be handled in a thread-safe manner.
Use the new macro to fix a mostly-theoretical bug where ARM's ucall MMIO setup could result in different VMs stomping on each other.
--- .../testing/selftests/kvm/include/kvm_util_base.h | 15 +++++++++++++++ tools/testing/selftests/kvm/lib/aarch64/ucall.c | 15 +++++++++++---- 2 files changed, 26 insertions(+), 4 deletions(-)
tools/testing/selftests/kvm/include/kvm_util_base.h +15 -0
@@ -16,6 +16,7 @@#include <linux/kvm.h> #include "linux/rbtree.h" +#include <asm/atomic.h> #include <sys/ioctl.h>
@@ -727,6 +728,20 @@ kvm_userspace_memory_region_find(struct kvm_vm *vm, uint64_t start,memcpy (&(g), _p,sizeof (g)); \ }) +/* +* Sync a global pointer to the guest that has a per-VM value, in which case +* writes to the host copy of the "global" must be serialized (in case a test +* is being truly crazy and spawning multiple VMs concurrently). +*/ +#define atomic_sync_global_pointer_to_guest (vm, g, val) ({ \ + typeof(g) *_p = addr_gva2hva(vm, (vm_vaddr_t)&(g)); \ + \ + while (cmpxchg(&g, NULL, val)) \ + ; \ + memcpy(_p, &(g), sizeof(g)); \ + WRITE_ONCE(g, NULL); \ +}) +void assert_on_unhandled_exception (struct kvm_vcpu *vcpu); void vcpu_arch_dump(FILE *stream, struct kvm_vcpu *vcpu,
tools/testing/selftests/kvm/lib/aarch64/ucall.c +11 -4
@@ -6,8 +6,17 @@*/ #include "kvm_util.h" +/* +* This "global" holds different per-VM values, it must not be accessed from +* host code except to sync the guest value, and that must be done atomically. +*/ static vm_vaddr_t *ucall_exit_mmio_addr; +static void ucall_set_mmio_addr (struct kvm_vm *vm,vm_vaddr_t *val) +{ +atomic_sync_global_pointer_to_guest (vm, ucall_exit_mmio_addr, val); +} +static bool ucall_mmio_init (struct kvm_vm *vm,vm_paddr_t gpa) {if (kvm_userspace_memory_region_find (vm, gpa, gpa +1 ))
@@ -15,8 +24,7 @@ static bool ucall_mmio_init(struct kvm_vm *vm, vm_paddr_t gpa)virt_pg_map (vm, gpa, gpa); - ucall_exit_mmio_addr = (vm_vaddr_t *)gpa; -sync_global_to_guest (vm, ucall_exit_mmio_addr); +ucall_set_mmio_addr (vm, (vm_vaddr_t *)gpa);return true; }
@@ -66,8 +74,7 @@ void ucall_arch_init(struct kvm_vm *vm, void *arg)void ucall_arch_uninit (struct kvm_vm *vm) { - ucall_exit_mmio_addr =0 ; -sync_global_to_guest (vm, ucall_exit_mmio_addr); +ucall_set_mmio_addr (vm, (vm_vaddr_t *)NULL ); }void ucall_arch_do_ucall (vm_vaddr_t uc)
Signature · 1 line
2.37.1.595.g718a3a8f04-goog
From ba28b353f632bcf8a3b6a8bc4e678c18a3dae7e9 Mon Sep 17 00:00:00 2001
From: Sean Christopherson <[email protected]>
Date: Wed, 10 Aug 2022 08:20:32 -0700
Subject: [PATCH 11/13] KVM: selftests: Add ucall pool based implementation
From: Peter Gonda <[email protected]>
To play nice with guests whose stack memory is encrypted, e.g. AMD SEV, introduce a new "ucall pool" implementation that passes the ucall struct via dedicated memory (which can be mapped shared, a.k.a. as plain text).
Because not all architectures have access to the vCPU index in the guest, use a bitmap with atomic accesses to track which entries in the pool are free/used. A list+lock could also work in theory, but synchronizing the individual pointers to the guest would be a mess.
Note, there's no need to rewalk the bitmap to ensure success. If all vCPUs are simply allocating, success is guaranteed because there are enough entries for all vCPUs. If one or more vCPUs are freeing and then reallocating, success is guaranteed because vCPUs _always_ walk the bitmap from 0=>N; if vCPU frees an entry and then wins a race to re-allocate, then either it will consume the entry it just freed (bit is the first free bit), or the losing vCPU is guaranteed to see the freed bit (winner consumes an earlier bit, which the loser hasn't yet visited).
--- .../selftests/kvm/include/ucall_common.h | 13 ++- .../testing/selftests/kvm/lib/aarch64/ucall.c | 7 +- tools/testing/selftests/kvm/lib/riscv/ucall.c | 2 +- tools/testing/selftests/kvm/lib/s390x/ucall.c | 2 +- .../testing/selftests/kvm/lib/ucall_common.c | 83 ++++++++++++++++++- .../testing/selftests/kvm/lib/x86_64/ucall.c | 2 +- 6 files changed, 89 insertions(+), 20 deletions(-)
tools/testing/selftests/kvm/include/ucall_common.h +5 -8
@@ -22,6 +22,9 @@ enum {struct ucall {uint64_t cmd;uint64_t args[UCALL_MAX_ARGS ]; + +/* For ucall pool usage. */ +struct ucall *hva; };void ucall_arch_init (struct kvm_vm *vm,void *arg);
@@ -32,15 +35,9 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu);void ucall (uint64_t cmd,int nargs, ...);uint64_t get_ucall (struct kvm_vcpu *vcpu,struct ucall *uc); -static inline void ucall_init (struct kvm_vm *vm,void *arg) -{ -ucall_arch_init (vm, arg); -} +void ucall_init (struct kvm_vm *vm,void *arg); -static inline void ucall_uninit (struct kvm_vm *vm) -{ -ucall_arch_uninit (vm); -} +void ucall_uninit (struct kvm_vm *vm);#define GUEST_SYNC_ARGS (stage, arg1, arg2, arg3, arg4) \ ucall(UCALL_SYNC, 6, "hello", stage, arg1, arg2, arg3, arg4)
tools/testing/selftests/kvm/lib/aarch64/ucall.c +2 -5
@@ -88,12 +88,9 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu)if (run->exit_reason ==KVM_EXIT_MMIO && run->mmio.phys_addr == (uint64_t )ucall_exit_mmio_addr) { -vm_vaddr_t gva; - -TEST_ASSERT (run->mmio.is_write && run->mmio.len ==8 , +TEST_ASSERT (run->mmio.is_write && run->mmio.len ==sizeof (uint64_t ),"Unexpected ucall exit mmio address access" ); -memcpy (&gva, run->mmio.data,sizeof (gva)); -return addr_gva2hva (vcpu->vm, gva); +return (void *)(*((uint64_t *)run->mmio.data)); }return NULL ;
tools/testing/selftests/kvm/lib/riscv/ucall.c +1 -1
@@ -59,7 +59,7 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) run->riscv_sbi.extension_id == KVM_RISCV_SELFTESTS_SBI_EXT) {switch (run->riscv_sbi.function_id) {case KVM_RISCV_SELFTESTS_SBI_UCALL : -return addr_gva2hva (vcpu->vm, run->riscv_sbi.args[0 ]); +return (void *)run->riscv_sbi.args[0 ];case KVM_RISCV_SELFTESTS_SBI_UNEXP :vcpu_dump (stderr, vcpu,2 );TEST_ASSERT (0 ,"Unexpected trap taken by guest" );
tools/testing/selftests/kvm/lib/s390x/ucall.c +1 -1
@@ -30,7 +30,7 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu) (run->s390_sieic.ipb >> 16) == 0x501) {int reg = run->s390_sieic.ipa &0xf ; -return addr_gva2hva (vcpu->vm, run->s.regs.gprs[reg]); +return (void *)run->s.regs.gprs[reg]; }return NULL ; }
tools/testing/selftests/kvm/lib/ucall_common.c +79 -4
101 lines not on this page · raw message
tools/testing/selftests/kvm/lib/x86_64/ucall.c +1 -1
@@ -30,7 +30,7 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu)struct kvm_regs regs;vcpu_regs_get (vcpu, ®s); -return addr_gva2hva (vcpu->vm, regs.rdi); +return (void *)regs.rdi; }return NULL ; }
Signature · 1 line
2.37.1.595.g718a3a8f04-goog
Re: [V3 10/11] KVM: selftests: Add ucall pool based implementation
Quoted 29 lines · Sean Christopherson
Quoted 11 lines · Andrew Jones
Quoted 4 lines · Sean Christopherson
But why is "
use_ucall_pool" optional? Unless there's a use case that fundamentally conflicts with the pool approach, let's make the pool approach the _only_ approach. IIRC, ARM's implementation isn't thread safe, i.e. there's at least one other use case that _needs_ the pool implementation.Really? The ucall structure is on the vcpu's stack like the other architectures. Ah, you're probably thinking about the shared address used to exit to userspace. The address doesn't matter as long as no VM maps it, but, yes, a multi-VM test where the VMs have different maps could end up breaking ucalls for one or more VMs.
Yah, that's what I was thinking of.
Quoted 9 lines
It wouldn't be hard to make that address per-VM, though, if ever necessary.
Quoted 4 lines
By supporting both, we are signing ourselves up for extra maintenance and pain, e.g. inevitably we'll break whichever option isn't the default and not notice for quite some time.
uc pools are currently limited to a single VM. That could be changed, but at the expense of even more code to maintain.
Unless I'm missing something, it's actually less code. "globals" that are sync'd to the guest aren't truly global, each VM has a separate physical page that is a copy of the global, hence the need for
sync_global_to_guest().
Quoted 7 lines
And FWIW, the code is actually "safe" in practice because I doubt any test spawns multiple VMs in parallel, i.e. the host might get confused over the value of
ucall_pool, but guests won't stomp on each other so long as they're created serially. Ditto for ARM'sucall_exit_mmio_addr.To make this truly thread safe, we just need a way to atomically sync the pointer to the guest, and that's easy enough to add.
I like that.
Quoted 25 lines
With that out of the way, all of the conditional "use pool" code goes away, which IMO simplifies things overall. Using a pool versus stack memory isn't _that_ much more complicated, and we _know_ the stack approach doesn't work for all VM types.
Indeed, this partial conversion is:
7 files changed, 131 insertions(+), 18 deletions(-)versus a full conversion:
6 files changed, 89 insertions(+), 20 deletions(-)And simplification is only a secondary concern, what I'm really worried about is things bitrotting and then some poor sod having to wade through unrelated issues because somebody else broke the pool implementation and didn't notice.
Argh, case in point, none of the x86 (or s390 or RISC-V) tests do
ucall_init(), which would have been a lurking bug if any of them tried to switch to the pool.Argh + case in point #2, this code is broken, and that bug may have sat unnoticed due to only the esoteric SEV test using the pool.
(quoted diff) +1
-1-static inline size_t uc_pool_idx(struct ucall *uc) +static noinline void ucall_free(struct ucall *uc)
Review
Quoted 10 lines
(quoted diff) +2
-6{ - return uc->hva - ucall_pool->ucalls; -} - -static void ucall_free(struct ucall *uc) -{ - clear_bit(uc_pool_idx(uc), ucall_pool->in_use); + /* Beware, here be pointer arithmetic. */ + clear_bit(uc - ucall_pool->ucalls, ucall_pool->in_use); }
Dropping the only once-used uc_pool_idx() and adding the comment looks better, but I don't think there was a bug because uc == uc->hva.
Quoted 6 lines
So my very strong vote is to fully switch to a single implementation. That way our code either works or it doesn't, i.e. we notice very quickly if it breaks.
Peter, if I can convince Drew of One Pool To Rule Them All, can you slot in the
There are other hills for me to stand on, so I won't insist on this one.
Quoted 17 lines
attached patches and slightly rework the ordering so that all SEV patches are at the end? E.g. something like:
KVM: selftests: Automatically do
init_ucall()for non-barebones VMs
KVM: selftests: movevm_phy_pages_alloc()earlier in file
KVM: selftests: sparsebit: add const where appropriate
KVM: selftests: add hooks for managing encrypted guest memory
KVM: selftests: handle encryption bits in page tables
KVM: selftests: add support for encrypted vm_vaddr_* allocations
KVM: selftests: Consolidate common code for popuplating
KVM: selftests: Consolidate boilerplate code inget_ucall()
tools: Addatomic_test_and_set_bit()
KVM: selftests: Add macro to atomically sync per-VM "global" pointers
KVM: selftests: Add ucall pool based implementation
KVM: selftests: add library for creating/interacting with SEV guests
KVM: selftests: Add simple sev vm testing
Quoted 3 lines
The patches are tested on x86 and compile tested on arm. I can provide more testing if/when necessary. I also haven't addressed any other feedback in the "ucall pool" patch, though I'm guessing much of it no longer applies.
I skimmed the patches but don't know how to comment on attachments, so I'll add the two things that popped to mind here.
1) Doing ucall_init() at VM create time may be too early for Arm. Arm probes for an unmapped address for the MMIO address it needs, so it's best to do that after all mapping.
2) We'll need to change the sanity checks in Arm's get_ucall() to not check that the MMIO address == ucall_exit_mmio_addr since there's no guarantee that the exiting guest's address matches whatever is lingering in the host's version. We can either drop the sanity check altogether or try to come up with something else.
Thanks,
drew
Sean Christopherson
reviewerRe: [V3 10/11] KVM: selftests: Add ucall pool based implementation
Quoted 4 lines · Andrew Jones
Quoted 1 line · Sean Christopherson
On Thu, Aug 18, 2022, Andrew Jones wrote:
Dropping the only once-used
uc_pool_idx()and adding the comment looks better, but I don't think there was a bug because uc ==uc->hva.
uc == uc->hva for the host, but not for the guest. From the guest's perspective, "hva" is an opaque pointer that has no direct relation to "uc".
Quoted 3 lines
1) Doing
ucall_init()at VM create time may be too early for Arm. Arm probes for an unmapped address for the MMIO address it needs, so it's best to do that after all mapping.
Argh. I really hate the MMIO probing. Is there really no arbitrary address that selftests can carve out and simply say "thou shalt not create a memslot here"?
Or can we just say that it's always immediate after memslot0? That would allow us to delete the searching code in ARM's ucall_arch_init().
Quoted 5 lines
2) We'll need to change the sanity checks in Arm's
get_ucall()to not check that the MMIO address ==ucall_exit_mmio_addrsince there's no guarantee that the exiting guest's address matches whatever is lingering in the host's version. We can either drop the sanity check altogether or try to come up with something else.
Ah, right. This one at least is easy to handle. If the address is per-VM, just track the address. If it's hardcoded (as a fix for #1) then, it's even simpler.
tools/testing/selftests/kvm/include/kvm_util_base.h +2 -1
@@ -94,7 +94,8 @@ struct kvm_vm {vm_paddr_t pgd;vm_vaddr_t gdt;vm_vaddr_t tss; -vm_vaddr_t idt; +vm_paddr_t pgd; +vm_vaddr_t ucall_addr;vm_vaddr_t handlers;uint32_t dirty_ring_size;struct vm_memcrypt memcrypt;
tools/testing/selftests/kvm/lib/aarch64/ucall.c +3 -1
@@ -14,6 +14,8 @@ static vm_vaddr_t *ucall_exit_mmio_addr;static void ucall_set_mmio_addr (struct kvm_vm *vm,vm_vaddr_t val) { + vm->ucall_addr = val; +atomic_sync_global_pointer_to_guest (vm, ucall_exit_mmio_addr, val); }
@@ -87,7 +89,7 @@ void *ucall_arch_get_ucall(struct kvm_vcpu *vcpu)struct kvm_run *run = vcpu->run;if (run->exit_reason ==KVM_EXIT_MMIO && - run->mmio.phys_addr == (uint64_t )ucall_exit_mmio_addr) { + run->mmio.phys_addr == vcpu->vm->ucall_addr) {TEST_ASSERT (run->mmio.is_write && run->mmio.len ==sizeof (uint64_t ),"Unexpected ucall exit mmio address access" );return (void *)(*((uint64_t *)run->mmio.data));
Sean Christopherson
reviewerRe: [V3 10/11] KVM: selftests: Add ucall pool based implementation
Quoted 2 lines · Sean Christopherson
Or can we just say that it's always immediate after memslot0? That would allow us to delete the searching code in ARM's
ucall_arch_init().
I have this coded up, will test on x86 and arm64 and send out a series (essentially all of the non-SEV bits in this series).
Prescribing an MMIO address from __vm_create() has a some nice side effects.
1) KVM treats writes to read-only memslots as MMIO, so a future cleanup would be to have __vm_create() create a memslot for the MMIO range to prevent silently clobbering the address. I'll leave this for later because selftests currently assumes they can use all memslots except memslot0.
2) It will simplify wwitching x86 and RISC-V to a common MMIO implementation, if we ever want to do that. I.e. have common code for everything except s390.
Re: [V3 10/11] KVM: selftests: Add ucall pool based implementation
In reply to Peter GondaQuoted 29 lines · Peter Gonda
...
tools/testing/selftests/kvm/lib/ucall_common.c +28
-0+ +static struct ucall *ucall_alloc (void ) +{ +struct ucall *uc =NULL ; +int i; + +if (!use_ucall_pool) + goto out; + +for (i =0 ; i <KVM_MAX_VCPUS ; ++i) { +if (!atomic_test_and_set_bit (i, ucall_pool->in_use)) { + uc = &ucall_pool->ucalls[i]; +memset (uc->args,0 ,sizeof (uc->args)); +break ; + } + } +out: +return uc; +} + +static inline size_t uc_pool_idx (struct ucall *uc) +{ +return uc->hva - ucall_pool->ucalls; +} + +static void ucall_free (struct ucall *uc) +{ +if (!use_ucall_pool)
tools/testing/selftests/kvm/lib/ucall_common.c line 92
Quoted 10 lines
tools/testing/selftests/kvm/lib/ucall_common.c +6
-1+return ; + +clear_bit (uc_pool_idx (uc), ucall_pool->in_use); +}void ucall (uint64_t cmd,int nargs, ...) { -struct ucall uc = {}; +struct ucall *uc; +struct ucall tmp = {};
This steps seems to result in generating instructions that need SSE support on x86:
struct ucall tmp = {};
movaps %xmm0,0x20(%rsp)
movaps %xmm0,0x30(%rsp)
movaps %xmm0,0x40(%rsp)
movaps %xmm0,0x50(%rsp)
This initialization will need proper compilation flags to generate instructions according to VM configuration.
Quoted 20 lines
va_list va; int i;tools/testing/selftests/kvm/lib/ucall_common.c +6
-3-WRITE_ONCE (uc.cmd, cmd); + uc =ucall_alloc (); +if (!uc) + uc = &tmp; + +WRITE_ONCE (uc->cmd, cmd); nargs =min (nargs,UCALL_MAX_ARGS );va_start (va, nargs);for (i =0 ; i < nargs; ++i) -WRITE_ONCE (uc.args[i],va_arg (va, uint64_t)); +WRITE_ONCE (uc->args[i],va_arg (va, uint64_t));va_end (va); -ucall_arch_do_ucall ((vm_vaddr_t)&uc);
Sean Christopherson
reviewerRe: [V3 10/11] KVM: selftests: Add ucall pool based implementation
Quoted 9 lines · Vishal Annapurve
tools/testing/selftests/kvm/lib/ucall_common.c line 92
Quoted 5 lines · Peter Gonda
void
ucall(uint64_t cmd, int nargs, ...)
{tools/testing/selftests/kvm/lib/ucall_common.c +2
-1-struct ucall uc = {}; +struct ucall *uc; +struct ucall tmp = {};This steps seems to result in generating instructions that need SSE support on x86:
Quoted 8 lines
struct ucall tmp = {}; movaps %xmm0,0x20(%rsp) movaps %xmm0,0x30(%rsp) movaps %xmm0,0x40(%rsp) movaps %xmm0,0x50(%rsp)This initialization will need proper compilation flags to generate instructions according to VM configuration.
Can you be more specific as to why generating SSE instructiions is problematic? The compiler emitting fancy instructions for struct initialization is not out of the ordinary.
Re: [V3 10/11] KVM: selftests: Add ucall pool based implementation
Quoted 17 lines · Sean Christopherson
Quoted 15 lines · Vishal Annapurve
tools/testing/selftests/kvm/lib/ucall_common.c line 92
Quoted 5 lines · Peter Gonda
void
ucall(uint64_t cmd, int nargs, ...)
{tools/testing/selftests/kvm/lib/ucall_common.c +2
-1-struct ucall uc = {}; +struct ucall *uc; +struct ucall tmp = {};This steps seems to result in generating instructions that need SSE support on x86:
struct ucall tmp = {}; movaps %xmm0,0x20(%rsp) movaps %xmm0,0x30(%rsp) movaps %xmm0,0x40(%rsp) movaps %xmm0,0x50(%rsp)
Quoted 6 lines
Quoted 2 lines
This initialization will need proper compilation flags to generate instructions according to VM configuration.
Can you be more specific as to why generating SSE instructiions is problematic? The compiler emitting fancy instructions for struct initialization is not out of the ordinary.
When executing these instructions, I was hitting #GP in the guest VM. CR4/CPUID settings seem to allow a VM to execute instructions that need SSE support enabled (which I was doubting earlier). After some more digging, it looks like the compiler is not able to ensure that operands for such instructions are 16 byte aligned as required by the hardware. There is a bug discussing this same issue:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=40838.
Adding -mstackrealign seems to help avoid running into #GP and things seem to work normally. Though I am not sure if SSE instruction execution is reliable from guest VM within selftests and so was suggesting usage of "-mno-sse" or equivalent flags.
PATCH v3 11/11 KVM: selftests: Add simple sev vm testing
In reply to Peter GondaA very simple of booting SEV guests that checks related CPUID bits. This is a stripped down version of "[PATCH v2 08/13] KVM: selftests: add SEV boot tests" from Michael but much simpler.
--- tools/testing/selftests/kvm/.gitignore | 1 + tools/testing/selftests/kvm/Makefile | 1 + .../selftests/kvm/include/x86_64/sev.h | 3 + tools/testing/selftests/kvm/lib/x86_64/sev.c | 2 - .../selftests/kvm/x86_64/sev_all_boot_test.c | 131 ++++++++++++++++++ 5 files changed, 136 insertions(+), 2 deletions(-) create mode 100644 tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c
tools/testing/selftests/kvm/.gitignore +1 -0
@@ -33,6 +33,7 @@ /x86_64/pmu_event_filter_test /x86_64/set_boot_cpu_id /x86_64/set_sregs_test +/x86_64/sev_all_boot_test /x86_64/sev_migrate_tests /x86_64/smm_test /x86_64/state_test
tools/testing/selftests/kvm/Makefile +1 -0
@@ -122,6 +122,7 @@ TEST_GEN_PROGS_x86_64 += x86_64/tsc_msrs_test TEST_GEN_PROGS_x86_64 += x86_64/vmx_pmu_caps_test TEST_GEN_PROGS_x86_64 += x86_64/xen_shinfo_test TEST_GEN_PROGS_x86_64 += x86_64/xen_vmcall_test +TEST_GEN_PROGS_x86_64 += x86_64/sev_all_boot_test TEST_GEN_PROGS_x86_64 += x86_64/sev_migrate_tests TEST_GEN_PROGS_x86_64 += x86_64/amx_test TEST_GEN_PROGS_x86_64 += x86_64/max_vcpuid_cap_test
tools/testing/selftests/kvm/include/x86_64/sev.h +3 -0
@@ -22,6 +22,9 @@#define SEV_POLICY_NO_DBG (1UL << 0)#define SEV_POLICY_ES (1UL << 2) +#define CPUID_MEM_ENC_LEAF 0x8000001f +#define CPUID_EBX_CBIT_MASK 0x3f +enum {SEV_GSTATE_UNINIT =0 ,SEV_GSTATE_LUPDATE ,
tools/testing/selftests/kvm/lib/x86_64/sev.c +0 -2
@@ -13,8 +13,6 @@#include "sev.h" #define PAGE_SHIFT 12 -#define CPUID_MEM_ENC_LEAF 0x8000001f -#define CPUID_EBX_CBIT_MASK 0x3fstruct sev_vm {struct kvm_vm *vm;
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +131 -0
131 lines not on this page · raw message
Signature · 1 line
2.37.1.559.g78731f0fdb-goog
Sean Christopherson
reviewerRe: [V3 11/11] KVM: selftests: Add simple sev vm testing
tools/testing/selftests/kvm/include/x86_64/sev.h line 26
Quoted 10 lines · Peter Gonda
tools/testing/selftests/kvm/include/x86_64/sev.h +2
-0@@ -22,6 +22,9 @@#define SEV_POLICY_NO_DBG (1UL << 0)#define SEV_POLICY_ES (1UL << 2) +#define CPUID_MEM_ENC_LEAF 0x8000001f +#define CPUID_EBX_CBIT_MASK 0x3f
Ha! I was going to say "put these in processor.h", but I have an even better idea. I'll try to a series posted tomorrow (compile tested only at this point), but what I'm hoping to do is to allow automagic retrieval of multi-bit CPUID properties, a la the existing this_cpu_has() stuff.
E.g.
#define X86_PROPERTY_CBIT_LOCATION KVM_X86_CPU_PROPERTY(0x8000001F, 0, EBX, 0, 5)
and then
sev->enc_bit = this_cpu_property(X86_PROPERTY_CBIT_LOCATION);
LOL, now I see that the defines in sev.c were introduced back in patch 08. That's probably fine for your submission so as not to take a dependency on the "property" idea. This patch doesn't need to move the CPUID_* defines because it can use this_cpu_has(X86_FEATURE_SEV) and avoid referencing CPUID_MEM_ENC_LEAF.
tools/testing/selftests/kvm/lib/x86_64/sev.c line 15
Quoted 11 lines
tools/testing/selftests/kvm/include/x86_64/sev.h +0
-0enum {SEV_GSTATE_UNINIT =0 ,SEV_GSTATE_LUPDATE ,tools/testing/selftests/kvm/lib/x86_64/sev.c +0
-0@@ -13,8 +13,6 @@#include "sev.h" #define PAGE_SHIFT 12
Already defined in processor.h
Quoted 23 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c +0
-2-#define CPUID_MEM_ENC_LEAF 0x8000001f -#define CPUID_EBX_CBIT_MASK 0x3fstruct sev_vm {struct kvm_vm *vm;tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +12
-0@@ -0,0 +1,131 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* +* Basic SEV boot tests. +* +* Copyright (C) 2021 Advanced Micro Devices +*/ +#define _GNU_SOURCE/* for program_invocation_short_name */ +#include <fcntl.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <sys/ioctl.h>
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c line 22
Quoted 10 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +10
-0+ +#include "test_util.h" + +#include "kvm_util.h" +#include "processor.h" +#include "svm_util.h" +#include "linux/psp-sev.h" +#include "sev.h" + +#define VCPU_ID 2
Nooooooo. Unless there is a really, REALLY good reason this needs to be '2', just pass '0' as a literal to vm_vcpu_add() and delete VCPU_ID.
Quoted 27 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +27
-0+#define PAGE_STRIDE 32 + +#define SHARED_PAGES 8192 +#define SHARED_VADDR_MIN 0x1000000 + +#define PRIVATE_PAGES 2048 +#define PRIVATE_VADDR_MIN (SHARED_VADDR_MIN + SHARED_PAGES * PAGE_SIZE) + +#define TOTAL_PAGES (512 + SHARED_PAGES + PRIVATE_PAGES) + +#define NR_SYNCS 1 + +static void guest_run_loop (struct kvm_vcpu *vcpu) +{ +struct ucall uc; +int i; + +for (i =0 ; i <=NR_SYNCS ; ++i) { +vcpu_run (vcpu); +switch (get_ucall (vcpu, &uc)) { +case UCALL_SYNC : +continue ; +case UCALL_DONE : +return ; +case UCALL_ABORT : + TEST_ASSERT(false, "%s at %s:%ld\n\tvalues: %#lx, %#lx", + (const char *)uc.args[0], __FILE__,
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c line 59
Quoted 10 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +10
-0+ uc.args[1], uc.args[2], uc.args[3]); +default : +TEST_ASSERT ( + false,"Unexpected exit: %s" , +exit_reason_str (vcpu->run->exit_reason)); + } + } +} + +static void __attribute__((__flatten__))guest_sev_code (void )
Is __flatten__ strictly necessary? I don't see this being copied over anything that would require it to be a contiguous chunk.
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c line 67
Quoted 8 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +8
-0+{ +uint32_t eax, ebx, ecx, edx; +uint64_t sev_status; + +GUEST_SYNC (1 ); + +cpuid (CPUID_MEM_ENC_LEAF , &eax, &ebx, &ecx, &edx); +GUEST_ASSERT (eax & (1 <<1 ));
GUEST_ASSERT(this_cpu_has(X86_FEATURE_SEV));
Quoted 6 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +6
-0+ + sev_status =rdmsr (MSR_AMD64_SEV ); +GUEST_ASSERT ((sev_status &0x1 ) ==1 ); + +GUEST_DONE (); +}
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c line 83
Quoted 10 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +10
-0+ +static struct sev_vm *setup_test_common (void *guest_code,uint64_t policy, +struct kvm_vcpu **vcpu) +{ +uint8_t measurement[512 ]; +struct sev_vm *sev; +struct kvm_vm *vm; +int i; + + sev =sev_vm_create (policy,TOTAL_PAGES );
TEST_ASSERT(sev, ...) so that this doesn't silently "pass"?
Quoted 21 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +21
-0+if (!sev) +return NULL ; + vm =sev_get_vm (sev); + +/* Set up VCPU and initial guest kernel. */ + *vcpu =vm_vcpu_add (vm,VCPU_ID , guest_code); +kvm_vm_elf_load (vm, program_invocation_name); + +/* Allocations/setup done. Encrypt initial guest payload. */ +sev_vm_launch (sev); + +/* Dump the initial measurement. A test to actually verify it would be nice. */ +sev_vm_launch_measure (sev, measurement); +pr_info ("guest measurement: " ); +for (i =0 ; i <32 ; ++i) +pr_info ("%02x" , measurement[i]); +pr_info ("\n" ); + +sev_vm_launch_finish (sev); + +return sev;
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c line 114
Quoted 10 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +10
-0+} + +static void test_sev (void *guest_code,uint64_t policy) +{ +struct sev_vm *sev; +struct kvm_vcpu *vcpu; + + sev =setup_test_common (guest_code, policy, &vcpu); +if (!sev) +return ;
And with an assert above, this return goes away. Or better yet, fold setup_test_common() into test_sev(), there's only the one user of the so called "common" function.
Quoted 20 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +17
-0+ +/* Guest is ready to run. Do the tests. */ +guest_run_loop (vcpu); + +pr_info ("guest ran successfully\n" ); + +sev_vm_free (sev); +} + +int main (int argc,char *argv[]) +{ +/* SEV tests */ +test_sev (guest_sev_code,SEV_POLICY_NO_DBG ); +test_sev (guest_sev_code,0 ); + +return 0 ; +}Signature · 1 line
2.37.1.559.g78731f0fdb-goog
Re: [V3 11/11] KVM: selftests: Add simple sev vm testing
Quoted 22 lines · Sean Christopherson
tools/testing/selftests/kvm/include/x86_64/sev.h line 26
Quoted 10 lines · Peter Gonda
tools/testing/selftests/kvm/include/x86_64/sev.h +2
-0@@ -22,6 +22,9 @@#define SEV_POLICY_NO_DBG (1UL << 0)#define SEV_POLICY_ES (1UL << 2) +#define CPUID_MEM_ENC_LEAF 0x8000001f +#define CPUID_EBX_CBIT_MASK 0x3fHa! I was going to say "put these in
processor.h", but I have an even better idea. I'll try to a series posted tomorrow (compile tested only at this point), but what I'm hoping to do is to allow automagic retrieval of multi-bit CPUID properties, a la the existingthis_cpu_has()stuff.E.g.
#define X86_PROPERTY_CBIT_LOCATION KVM_X86_CPU_PROPERTY(0x8000001F, 0, EBX, 0, 5)
Quoted 8 lines
and then
sev->enc_bit = this_cpu_property(X86_PROPERTY_CBIT_LOCATION);LOL, now I see that the defines in
sev.cwere introduced back in patch 08. That's probably fine for your submission so as not to take a dependency on the "property" idea. This patch doesn't need to move the CPUID_* defines because it can usethis_cpu_has(X86_FEATURE_SEV)and avoid referencingCPUID_MEM_ENC_LEAF.
OK I've put these into sev.c instead of the header. I can clean up after you property patch series goes in.
Quoted 11 lines
Quoted 11 lines
tools/testing/selftests/kvm/include/x86_64/sev.h +0
-0enum {SEV_GSTATE_UNINIT =0 ,SEV_GSTATE_LUPDATE ,tools/testing/selftests/kvm/lib/x86_64/sev.c +0
-0@@ -13,8 +13,6 @@#include "sev.h"
Quoted 3 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c line 15
Quoted 1 line
tools/testing/selftests/kvm/lib/x86_64/sev.c +0
-0#define PAGE_SHIFT 12Already defined in
processor.h
Removed
Quoted 29 lines
Quoted 29 lines
tools/testing/selftests/kvm/lib/x86_64/sev.c +0
-2-#define CPUID_MEM_ENC_LEAF 0x8000001f -#define CPUID_EBX_CBIT_MASK 0x3fstruct sev_vm {struct kvm_vm *vm;tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +17
-0@@ -0,0 +1,131 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* +* Basic SEV boot tests. +* +* Copyright (C) 2021 Advanced Micro Devices +*/ +#define _GNU_SOURCE/* for program_invocation_short_name */ +#include <fcntl.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <sys/ioctl.h> + +#include "test_util.h" + +#include "kvm_util.h" +#include "processor.h"
Quoted 8 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c line 22
Quoted 5 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +5
-0+#include "svm_util.h" +#include "linux/psp-sev.h" +#include "sev.h" + +#define VCPU_ID 2Nooooooo. Unless there is a really, REALLY good reason this needs to be '2', just pass '0' as a literal to
vm_vcpu_add()and deleteVCPU_ID.
Not needed, removed.
Quoted 33 lines
Quoted 33 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +32
-0+#define PAGE_STRIDE 32 + +#define SHARED_PAGES 8192 +#define SHARED_VADDR_MIN 0x1000000 + +#define PRIVATE_PAGES 2048 +#define PRIVATE_VADDR_MIN (SHARED_VADDR_MIN + SHARED_PAGES * PAGE_SIZE) + +#define TOTAL_PAGES (512 + SHARED_PAGES + PRIVATE_PAGES) + +#define NR_SYNCS 1 + +static void guest_run_loop (struct kvm_vcpu *vcpu) +{ +struct ucall uc; +int i; + +for (i =0 ; i <=NR_SYNCS ; ++i) { +vcpu_run (vcpu); +switch (get_ucall (vcpu, &uc)) { +case UCALL_SYNC : +continue ; +case UCALL_DONE : +return ; +case UCALL_ABORT : +TEST_ASSERT (false,"%s at %s:%ld\n\tvalues: %#lx, %#lx" , + (const char *)uc.args[0 ], __FILE__, + uc.args[1 ], uc.args[2 ], uc.args[3 ]); +default : +TEST_ASSERT ( + false,"Unexpected exit: %s" , +exit_reason_str (vcpu->run->exit_reason));
Quoted 8 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c line 59
Quoted 5 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +5
-0+ } + } +} + +static void __attribute__((__flatten__))guest_sev_code (void )Is
__flatten__strictly necessary? I don't see this being copied over anything that would require it to be a contiguous chunk.
Nope, removed.
Quoted 11 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c line 67
Quoted 9 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +8
-0+{ +uint32_t eax, ebx, ecx, edx; +uint64_t sev_status; + +GUEST_SYNC (1 ); + +cpuid (CPUID_MEM_ENC_LEAF , &eax, &ebx, &ecx, &edx); +GUEST_ASSERT (eax & (1 <<1 ));GUEST_ASSERT(this_cpu_has(X86_FEATURE_SEV));
Done.
Quoted 10 lines
Quoted 10 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +10
-0+ + sev_status =rdmsr (MSR_AMD64_SEV ); +GUEST_ASSERT ((sev_status &0x1 ) ==1 ); + +GUEST_DONE (); +} + +static struct sev_vm *setup_test_common (void *guest_code,uint64_t policy, +struct kvm_vcpu **vcpu) +{
Quoted 8 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c line 83
Quoted 6 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +6
-0+uint8_t measurement[512 ]; +struct sev_vm *sev; +struct kvm_vm *vm; +int i; + + sev =sev_vm_create (policy,TOTAL_PAGES );
TEST_ASSERT(sev, ...)so that this doesn't silently "pass"?
Done.
Quoted 27 lines
Quoted 27 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +26
-0+if (!sev) +return NULL ; + vm =sev_get_vm (sev); + +/* Set up VCPU and initial guest kernel. */ + *vcpu =vm_vcpu_add (vm,VCPU_ID , guest_code); +kvm_vm_elf_load (vm, program_invocation_name); + +/* Allocations/setup done. Encrypt initial guest payload. */ +sev_vm_launch (sev); + +/* Dump the initial measurement. A test to actually verify it would be nice. */ +sev_vm_launch_measure (sev, measurement); +pr_info ("guest measurement: " ); +for (i =0 ; i <32 ; ++i) +pr_info ("%02x" , measurement[i]); +pr_info ("\n" ); + +sev_vm_launch_finish (sev); + +return sev; +} + +static void test_sev (void *guest_code,uint64_t policy) +{ +struct sev_vm *sev;
Quoted 8 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c line 114
Quoted 5 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +5
-0+struct kvm_vcpu *vcpu; + + sev =setup_test_common (guest_code, policy, &vcpu); +if (!sev) +return ;And with an assert above, this return goes away. Or better yet, fold
setup_test_common()intotest_sev(), there's only the one user of the so called "common" function.
Done.
Quoted 21 lines
Quoted 21 lines
tools/testing/selftests/kvm/x86_64/sev_all_boot_test.c new file +17
-1+ +/* Guest is ready to run. Do the tests. */ +guest_run_loop (vcpu); + +pr_info ("guest ran successfully\n" ); + +sev_vm_free (sev); +} + +int main (int argc,char *argv[]) +{ +/* SEV tests */ +test_sev (guest_sev_code,SEV_POLICY_NO_DBG ); +test_sev (guest_sev_code,0 ); + +return 0 ; +} --2.37.1.559.g78731f0fdb-goog
Sean Christopherson
reviewerRe: [V3 11/11] KVM: selftests: Add simple sev vm testing
In reply to Peter GondaQuoted 5 lines · Peter Gonda
A very simple of booting SEV guests that checks related CPUID bits. This is a stripped down version of "[PATCH v2 08/13] KVM: selftests: add SEV boot tests" from Michael but much simpler.
Signed-off-by Michael Roth <[email protected]>
If you're including Mike's SOB, then he should either be credited as the author or with Co-developed-by. If the code has been significantly rewritten, i.e. carrying Mike's SOB would be disingenuous, then do a less "official"
or