Subject: [PATCH v4 0/4] KVM: cpuid: fix KVM_GET_EMULATED_CPUID implementation

This series aims to clarify the behavior of the KVM_GET_EMULATED_CPUID
ioctl, and fix a corner case where -E2BIG is returned when
the nent field of struct kvm_cpuid2 is matching the amount of
emulated entries that kvm returns.

Patch 1 proposes the nent field fix to cpuid.c,
patch 2 updates the ioctl documentation accordingly and
patches 3 and 4 extend the x86_64/get_cpuid_test.c selftest to check
the intended behavior of KVM_GET_EMULATED_CPUID.

Signed-off-by: Emanuele Giuseppe Esposito <[email protected]>
---
v4:
- Address nitpicks given in the mailing list

Emanuele Giuseppe Esposito (4):
KVM: x86: Fix a spurious -E2BIG in KVM_GET_EMULATED_CPUID
Documentation: KVM: update KVM_GET_EMULATED_CPUID ioctl description
selftests: add kvm_get_emulated_cpuid to processor.h
selftests: KVM: extend get_cpuid_test to include
KVM_GET_EMULATED_CPUID

Documentation/virt/kvm/api.rst | 10 +--
arch/x86/kvm/cpuid.c | 33 ++++---
.../selftests/kvm/include/x86_64/processor.h | 1 +
.../selftests/kvm/lib/x86_64/processor.c | 33 +++++++
.../selftests/kvm/x86_64/get_cpuid_test.c | 90 ++++++++++++++++++-
5 files changed, 142 insertions(+), 25 deletions(-)

--
2.30.2


Subject: [PATCH v4 2/4] Documentation: KVM: update KVM_GET_EMULATED_CPUID ioctl description

KVM_GET_EMULATED_CPUID returns -E2BIG if the nent field of
struct kvm_cpuid2 is smaller than the actual entries, while
it adjusts nent if the provided amount is bigger than the
actual amount.

Update documentation accordingly. ENOMEM is just returned if the
allocation fails, like all other calls.

Signed-off-by: Emanuele Giuseppe Esposito <[email protected]>
---
Documentation/virt/kvm/api.rst | 10 ++++------
1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/Documentation/virt/kvm/api.rst b/Documentation/virt/kvm/api.rst
index 307f2fcf1b02..8ba23bc2a625 100644
--- a/Documentation/virt/kvm/api.rst
+++ b/Documentation/virt/kvm/api.rst
@@ -3404,12 +3404,10 @@ which features are emulated by kvm instead of being present natively.

Userspace invokes KVM_GET_EMULATED_CPUID by passing a kvm_cpuid2
structure with the 'nent' field indicating the number of entries in
-the variable-size array 'entries'. If the number of entries is too low
-to describe the cpu capabilities, an error (E2BIG) is returned. If the
-number is too high, the 'nent' field is adjusted and an error (ENOMEM)
-is returned. If the number is just right, the 'nent' field is adjusted
-to the number of valid entries in the 'entries' array, which is then
-filled.
+the variable-size array 'entries'.
+If the number of entries is too low to describe the cpu
+capabilities, an error (E2BIG) is returned. If the number is too high,
+the 'nent' field is adjusted and the entries array is filled.

The entries returned are the set CPUID bits of the respective features
which kvm emulates, as returned by the CPUID instruction, with unknown
--
2.30.2

Subject: [PATCH v4 4/4] selftests: KVM: extend get_cpuid_test to include KVM_GET_EMULATED_CPUID

Extend the get_cpuid_test.c selftest to include the KVM_GET_EMULATED_CPUID
ioctl. Since the behavior and functionality is similar to
KVM_GET_SUPPORTED_CPUID, we only check additionally:

1) checks for corner case in the nent field of the struct kvm_cpuid2.
2) sets and gets it as cpuid from the guest VM

Signed-off-by: Emanuele Giuseppe Esposito <[email protected]>
---
.../selftests/kvm/x86_64/get_cpuid_test.c | 90 ++++++++++++++++++-
1 file changed, 88 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/kvm/x86_64/get_cpuid_test.c b/tools/testing/selftests/kvm/x86_64/get_cpuid_test.c
index 9b78e8889638..b9f0fba1b0ea 100644
--- a/tools/testing/selftests/kvm/x86_64/get_cpuid_test.c
+++ b/tools/testing/selftests/kvm/x86_64/get_cpuid_test.c
@@ -13,6 +13,7 @@
#include "processor.h"

#define VCPU_ID 0
+#define MAX_NENT 1000

/* CPUIDs known to differ */
struct {
@@ -137,7 +138,8 @@ static void run_vcpu(struct kvm_vm *vm, uint32_t vcpuid, int stage)
}
}

-struct kvm_cpuid2 *vcpu_alloc_cpuid(struct kvm_vm *vm, vm_vaddr_t *p_gva, struct kvm_cpuid2 *cpuid)
+static struct kvm_cpuid2 *vcpu_alloc_cpuid(struct kvm_vm *vm, vm_vaddr_t *p_gva,
+ struct kvm_cpuid2 *cpuid)
{
int size = sizeof(*cpuid) + cpuid->nent * sizeof(cpuid->entries[0]);
vm_vaddr_t gva = vm_vaddr_alloc(vm, size,
@@ -150,9 +152,84 @@ struct kvm_cpuid2 *vcpu_alloc_cpuid(struct kvm_vm *vm, vm_vaddr_t *p_gva, struct
return guest_cpuids;
}

+static struct kvm_cpuid2 *alloc_custom_kvm_cpuid2(int nent)
+{
+ struct kvm_cpuid2 *cpuid;
+ size_t size;
+
+ size = sizeof(*cpuid);
+ size += nent * sizeof(struct kvm_cpuid_entry2);
+ cpuid = calloc(1, size);
+ if (!cpuid) {
+ perror("malloc");
+ abort();
+ }
+
+ cpuid->nent = nent;
+
+ return cpuid;
+}
+
+static void clean_entries_kvm_cpuid2(struct kvm_cpuid2 *cpuid)
+{
+ size_t size;
+ int old_nent = cpuid->nent;
+
+ size = sizeof(*cpuid);
+ size += MAX_NENT * sizeof(struct kvm_cpuid_entry2);
+ memset(cpuid, 0, size);
+ cpuid->nent = old_nent;
+}
+
+static void test_emulated_entries(struct kvm_vm *vm)
+{
+ int res, right_nent;
+ struct kvm_cpuid2 *cpuid;
+
+ cpuid = alloc_custom_kvm_cpuid2(MAX_NENT);
+
+ /* 0 nent, return E2BIG */
+ cpuid->nent = 0;
+ res = _kvm_ioctl(vm, KVM_GET_EMULATED_CPUID, cpuid);
+ TEST_ASSERT(res == -1 && errno == E2BIG, "nent=0 should fail as E2BIG");
+ clean_entries_kvm_cpuid2(cpuid);
+
+ /* high nent, set the entries and adjust */
+ cpuid->nent = MAX_NENT;
+ res = _kvm_ioctl(vm, KVM_GET_EMULATED_CPUID, cpuid);
+ TEST_ASSERT(res == 0, "nent > actual nent should not fail");
+ right_nent = cpuid->nent;
+ clean_entries_kvm_cpuid2(cpuid);
+
+ /* high nent, set the entries and adjust */
+ cpuid->nent++;
+ res = _kvm_ioctl(vm, KVM_GET_EMULATED_CPUID, cpuid);
+ TEST_ASSERT(res == 0, "nent > actual nent should not fail");
+ TEST_ASSERT(right_nent == cpuid->nent, "nent should be always the same");
+ clean_entries_kvm_cpuid2(cpuid);
+
+ /* low nent, return E2BIG */
+ if (right_nent > 1) {
+ cpuid->nent = 1;
+ res = _kvm_ioctl(vm, KVM_GET_EMULATED_CPUID, cpuid);
+ TEST_ASSERT(res == -1 && errno == E2BIG, "nent=1 should fail");
+ clean_entries_kvm_cpuid2(cpuid);
+ }
+
+ /* exact nent */
+ cpuid->nent = right_nent;
+ res = _kvm_ioctl(vm, KVM_GET_EMULATED_CPUID, cpuid);
+ TEST_ASSERT(res == 0, "nent == actual nent should not fail");
+ TEST_ASSERT(cpuid->nent == right_nent,
+ "KVM_GET_EMULATED_CPUID should be invaried when nent is exact");
+ clean_entries_kvm_cpuid2(cpuid);
+
+ free(cpuid);
+}
+
int main(void)
{
- struct kvm_cpuid2 *supp_cpuid, *cpuid2;
+ struct kvm_cpuid2 *supp_cpuid, *emul_cpuid, *cpuid2;
vm_vaddr_t cpuid_gva;
struct kvm_vm *vm;
int stage;
@@ -171,5 +248,14 @@ int main(void)
for (stage = 0; stage < 3; stage++)
run_vcpu(vm, VCPU_ID, stage);

+ if (kvm_check_cap(KVM_CAP_EXT_EMUL_CPUID)) {
+ emul_cpuid = kvm_get_emulated_cpuid();
+ vcpu_set_cpuid(vm, VCPU_ID, emul_cpuid);
+ cpuid2 = vcpu_get_cpuid(vm, VCPU_ID);
+
+ test_emulated_entries(vm);
+ compare_cpuids(emul_cpuid, cpuid2);
+ }
+
kvm_vm_free(vm);
}
--
2.30.2

Subject: [PATCH v4 3/4] selftests: add kvm_get_emulated_cpuid to processor.h

As the similar kvm_get_supported_cpuid(),
kvm_get_emulated_cpuid allocates and gets
a struct kvm_cpuid2 filled with emulated features.

Signed-off-by: Emanuele Giuseppe Esposito <[email protected]>
---
.../selftests/kvm/include/x86_64/processor.h | 1 +
.../selftests/kvm/lib/x86_64/processor.c | 33 +++++++++++++++++++
2 files changed, 34 insertions(+)

diff --git a/tools/testing/selftests/kvm/include/x86_64/processor.h b/tools/testing/selftests/kvm/include/x86_64/processor.h
index 0b30b4e15c38..ae1b9530e187 100644
--- a/tools/testing/selftests/kvm/include/x86_64/processor.h
+++ b/tools/testing/selftests/kvm/include/x86_64/processor.h
@@ -353,6 +353,7 @@ void vcpu_load_state(struct kvm_vm *vm, uint32_t vcpuid,
struct kvm_msr_list *kvm_get_msr_index_list(void);
uint64_t kvm_get_feature_msr(uint64_t msr_index);
struct kvm_cpuid2 *kvm_get_supported_cpuid(void);
+struct kvm_cpuid2 *kvm_get_emulated_cpuid(void);

struct kvm_cpuid2 *vcpu_get_cpuid(struct kvm_vm *vm, uint32_t vcpuid);
void vcpu_set_cpuid(struct kvm_vm *vm, uint32_t vcpuid,
diff --git a/tools/testing/selftests/kvm/lib/x86_64/processor.c b/tools/testing/selftests/kvm/lib/x86_64/processor.c
index e676fe40bfe6..2ea14421bdfe 100644
--- a/tools/testing/selftests/kvm/lib/x86_64/processor.c
+++ b/tools/testing/selftests/kvm/lib/x86_64/processor.c
@@ -669,6 +669,39 @@ struct kvm_cpuid2 *kvm_get_supported_cpuid(void)
return cpuid;
}

+/*
+ * KVM Emulated CPUID Get
+ *
+ * Input Args: None
+ *
+ * Output Args:
+ *
+ * Return: The emulated KVM CPUID
+ *
+ * Get the guest CPUID emulated by KVM.
+ */
+struct kvm_cpuid2 *kvm_get_emulated_cpuid(void)
+{
+ static struct kvm_cpuid2 *cpuid;
+ int ret;
+ int kvm_fd;
+
+ if (cpuid)
+ return cpuid;
+
+ cpuid = allocate_kvm_cpuid2();
+ kvm_fd = open(KVM_DEV_PATH, O_RDONLY);
+ if (kvm_fd < 0)
+ exit(KSFT_SKIP);
+
+ ret = ioctl(kvm_fd, KVM_GET_EMULATED_CPUID, cpuid);
+ TEST_ASSERT(ret == 0, "KVM_GET_EMULATED_CPUID failed %d %d\n",
+ ret, errno);
+
+ close(kvm_fd);
+ return cpuid;
+}
+
/*
* KVM Get MSR
*
--
2.30.2

Subject: [PATCH v4 1/4] KVM: x86: Fix a spurious -E2BIG in KVM_GET_EMULATED_CPUID

When retrieving emulated CPUID entries, check for an insufficient array
size if and only if KVM is actually inserting an entry.
If userspace has a priori knowledge of the exact array size,
KVM_GET_EMULATED_CPUID will incorrectly fail due to effectively requiring
an extra, unused entry.

Fixes: 433f4ba19041 ("KVM: x86: fix out-of-bounds write in KVM_GET_EMULATED_CPUID (CVE-2019-19332)")
Signed-off-by: Emanuele Giuseppe Esposito <[email protected]>
---
arch/x86/kvm/cpuid.c | 33 ++++++++++++++++-----------------
1 file changed, 16 insertions(+), 17 deletions(-)

diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
index 6bd2f8b830e4..d30194081892 100644
--- a/arch/x86/kvm/cpuid.c
+++ b/arch/x86/kvm/cpuid.c
@@ -567,34 +567,33 @@ static struct kvm_cpuid_entry2 *do_host_cpuid(struct kvm_cpuid_array *array,

static int __do_cpuid_func_emulated(struct kvm_cpuid_array *array, u32 func)
{
- struct kvm_cpuid_entry2 *entry;
-
- if (array->nent >= array->maxnent)
- return -E2BIG;
+ struct kvm_cpuid_entry2 entry;

- entry = &array->entries[array->nent];
- entry->function = func;
- entry->index = 0;
- entry->flags = 0;
+ memset(&entry, 0, sizeof(entry));

switch (func) {
case 0:
- entry->eax = 7;
- ++array->nent;
+ entry.eax = 7;
break;
case 1:
- entry->ecx = F(MOVBE);
- ++array->nent;
+ entry.ecx = F(MOVBE);
break;
case 7:
- entry->flags |= KVM_CPUID_FLAG_SIGNIFCANT_INDEX;
- entry->eax = 0;
- entry->ecx = F(RDPID);
- ++array->nent;
- default:
+ entry.flags = KVM_CPUID_FLAG_SIGNIFCANT_INDEX;
+ entry.ecx = F(RDPID);
break;
+ default:
+ goto out;
}

+ /* This check is performed only when func is valid */
+ if (array->nent >= array->maxnent)
+ return -E2BIG;
+
+ entry.function = func;
+ memcpy(&array->entries[array->nent++], &entry, sizeof(entry));
+
+out:
return 0;
}

--
2.30.2

2021-04-08 20:33:25

by Sean Christopherson

[permalink] [raw]
Subject: Re: [PATCH v4 1/4] KVM: x86: Fix a spurious -E2BIG in KVM_GET_EMULATED_CPUID

On Thu, Apr 08, 2021, Emanuele Giuseppe Esposito wrote:
> When retrieving emulated CPUID entries, check for an insufficient array
> size if and only if KVM is actually inserting an entry.
> If userspace has a priori knowledge of the exact array size,
> KVM_GET_EMULATED_CPUID will incorrectly fail due to effectively requiring
> an extra, unused entry.
>
> Fixes: 433f4ba19041 ("KVM: x86: fix out-of-bounds write in KVM_GET_EMULATED_CPUID (CVE-2019-19332)")
> Signed-off-by: Emanuele Giuseppe Esposito <[email protected]>
> ---
> arch/x86/kvm/cpuid.c | 33 ++++++++++++++++-----------------
> 1 file changed, 16 insertions(+), 17 deletions(-)
>
> diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
> index 6bd2f8b830e4..d30194081892 100644
> --- a/arch/x86/kvm/cpuid.c
> +++ b/arch/x86/kvm/cpuid.c
> @@ -567,34 +567,33 @@ static struct kvm_cpuid_entry2 *do_host_cpuid(struct kvm_cpuid_array *array,
>
> static int __do_cpuid_func_emulated(struct kvm_cpuid_array *array, u32 func)
> {
> - struct kvm_cpuid_entry2 *entry;
> -
> - if (array->nent >= array->maxnent)
> - return -E2BIG;
> + struct kvm_cpuid_entry2 entry;
>
> - entry = &array->entries[array->nent];
> - entry->function = func;
> - entry->index = 0;
> - entry->flags = 0;
> + memset(&entry, 0, sizeof(entry));
>
> switch (func) {
> case 0:
> - entry->eax = 7;
> - ++array->nent;
> + entry.eax = 7;
> break;
> case 1:
> - entry->ecx = F(MOVBE);
> - ++array->nent;
> + entry.ecx = F(MOVBE);
> break;
> case 7:
> - entry->flags |= KVM_CPUID_FLAG_SIGNIFCANT_INDEX;
> - entry->eax = 0;
> - entry->ecx = F(RDPID);
> - ++array->nent;
> - default:
> + entry.flags = KVM_CPUID_FLAG_SIGNIFCANT_INDEX;
> + entry.ecx = F(RDPID);
> break;
> + default:
> + goto out;
> }
>
> + /* This check is performed only when func is valid */

Sorry to keep nitpicking and bikeshedding. Funcs aren't really "invalid", KVM
just doesn't have any features it emulates in other leafs. Maybe be more literal
in describing what triggers the check?

/* Check the array capacity iff the entry is being copied over. */

Not a sticking point, so either way:

Reviewed-by: Sean Christopherson <[email protected]>

> + if (array->nent >= array->maxnent)
> + return -E2BIG;
> +
> + entry.function = func;
> + memcpy(&array->entries[array->nent++], &entry, sizeof(entry));
> +
> +out:
> return 0;
> }
>
> --
> 2.30.2
>

Subject: Re: [PATCH v4 1/4] KVM: x86: Fix a spurious -E2BIG in KVM_GET_EMULATED_CPUID



On 08/04/2021 22:29, Sean Christopherson wrote:
> On Thu, Apr 08, 2021, Emanuele Giuseppe Esposito wrote:
>> When retrieving emulated CPUID entries, check for an insufficient array
>> size if and only if KVM is actually inserting an entry.
>> If userspace has a priori knowledge of the exact array size,
>> KVM_GET_EMULATED_CPUID will incorrectly fail due to effectively requiring
>> an extra, unused entry.
>>
>> Fixes: 433f4ba19041 ("KVM: x86: fix out-of-bounds write in KVM_GET_EMULATED_CPUID (CVE-2019-19332)")
>> Signed-off-by: Emanuele Giuseppe Esposito <[email protected]>
>> ---
>> arch/x86/kvm/cpuid.c | 33 ++++++++++++++++-----------------
>> 1 file changed, 16 insertions(+), 17 deletions(-)
>>
>> diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
>> index 6bd2f8b830e4..d30194081892 100644
>> --- a/arch/x86/kvm/cpuid.c
>> +++ b/arch/x86/kvm/cpuid.c
>> @@ -567,34 +567,33 @@ static struct kvm_cpuid_entry2 *do_host_cpuid(struct kvm_cpuid_array *array,
>>
>> static int __do_cpuid_func_emulated(struct kvm_cpuid_array *array, u32 func)
>> {
>> - struct kvm_cpuid_entry2 *entry;
>> -
>> - if (array->nent >= array->maxnent)
>> - return -E2BIG;
>> + struct kvm_cpuid_entry2 entry;
>>
>> - entry = &array->entries[array->nent];
>> - entry->function = func;
>> - entry->index = 0;
>> - entry->flags = 0;
>> + memset(&entry, 0, sizeof(entry));
>>
>> switch (func) {
>> case 0:
>> - entry->eax = 7;
>> - ++array->nent;
>> + entry.eax = 7;
>> break;
>> case 1:
>> - entry->ecx = F(MOVBE);
>> - ++array->nent;
>> + entry.ecx = F(MOVBE);
>> break;
>> case 7:
>> - entry->flags |= KVM_CPUID_FLAG_SIGNIFCANT_INDEX;
>> - entry->eax = 0;
>> - entry->ecx = F(RDPID);
>> - ++array->nent;
>> - default:
>> + entry.flags = KVM_CPUID_FLAG_SIGNIFCANT_INDEX;
>> + entry.ecx = F(RDPID);
>> break;
>> + default:
>> + goto out;
>> }
>>
>> + /* This check is performed only when func is valid */
>
> Sorry to keep nitpicking and bikeshedding.

No problem at all. Any comment is very welcome :)

Funcs aren't really "invalid", KVM
> just doesn't have any features it emulates in other leafs. Maybe be more literal
> in describing what triggers the check?
>
> /* Check the array capacity iff the entry is being copied over. */

What I mean here is that a func is "valid" if it matches one of the
cases of the switch statement. If it is not valid, it ends up in the
default case. But I agree, will change the comment your suggestion and
resend.

Thank you,
Emanuele

>
> Not a sticking point, so either way:
>
> Reviewed-by: Sean Christopherson <[email protected]>
>
>> + if (array->nent >= array->maxnent)
>> + return -E2BIG;
>> +
>> + entry.function = func;
>> + memcpy(&array->entries[array->nent++], &entry, sizeof(entry));
>> +
>> +out:
>> return 0;
>> }
>>
>> --
>> 2.30.2
>>
>