KVM_SEV_SEND_UPDATE_DATA and KVM_SEV_RECEIVE_UPDATE_DATA have an integer overflow issue. Params.guest_len and offset are both 32bite wide, with a large params.guest_len the check to confirm a page boundary is not crossed can falsely pass:
/* Check if we are crossing the page boundary * offset = params.guest_uaddr & (PAGE_SIZE - 1); if ((params.guest_len + offset > PAGE_SIZE))
Add an additional check to this conditional to confirm that params.guest_len itself is not greater than PAGE_SIZE.
The current code is can only overflow with a params.guest_len of greater than 0xfffff000. And the FW spec says these commands fail with lengths greater than 16KB. So this issue should not be a security concern
Fixes: 15fb7de1a7f5 ("KVM: SVM: Add KVM_SEV_RECEIVE_UPDATE_DATA command") Fixes: d3d1af85e2c7 ("KVM: SVM: Add KVM_SEND_UPDATE_DATA command") Reported-by: Andy Nguyen theflow@google.com Suggested-by: Thomas Lendacky thomas.lendacky@amd.com Signed-off-by: Peter Gonda pgonda@google.com Cc: David Rientjes rientjes@google.com Cc: Paolo Bonzini pbonzini@redhat.com Cc: Sean Christopherson seanjc@google.com Cc: kvm@vger.kernel.org Cc: stable@vger.kernel.org Cc: linux-kernel@vger.kernel.org ---
V2 * Updated conditional based on feedback from Tom.
--- arch/x86/kvm/svm/sev.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c index 273cba809328..3d74facaead8 100644 --- a/arch/x86/kvm/svm/sev.c +++ b/arch/x86/kvm/svm/sev.c @@ -1294,7 +1294,7 @@ static int sev_send_update_data(struct kvm *kvm, struct kvm_sev_cmd *argp)
/* Check if we are crossing the page boundary */ offset = params.guest_uaddr & (PAGE_SIZE - 1); - if ((params.guest_len + offset > PAGE_SIZE)) + if (params.guest_len > PAGE_SIZE || (params.guest_len + offset) > PAGE_SIZE) return -EINVAL;
/* Pin guest memory */ @@ -1474,7 +1474,7 @@ static int sev_receive_update_data(struct kvm *kvm, struct kvm_sev_cmd *argp)
/* Check if we are crossing the page boundary */ offset = params.guest_uaddr & (PAGE_SIZE - 1); - if ((params.guest_len + offset > PAGE_SIZE)) + if (params.guest_len > PAGE_SIZE || (params.guest_len + offset) > PAGE_SIZE) return -EINVAL;
hdr = psp_copy_user_blob(params.hdr_uaddr, params.hdr_len);
On 2/7/23 11:13, Peter Gonda wrote:
KVM_SEV_SEND_UPDATE_DATA and KVM_SEV_RECEIVE_UPDATE_DATA have an integer overflow issue. Params.guest_len and offset are both 32bite wide, with a large params.guest_len the check to confirm a page boundary is not crossed can falsely pass:
/* Check if we are crossing the page boundary * offset = params.guest_uaddr & (PAGE_SIZE - 1); if ((params.guest_len + offset > PAGE_SIZE))
Add an additional check to this conditional to confirm that params.guest_len itself is not greater than PAGE_SIZE.
The current code is can only overflow with a params.guest_len of greater than 0xfffff000. And the FW spec says these commands fail with lengths greater than 16KB. So this issue should not be a security concern
Fixes: 15fb7de1a7f5 ("KVM: SVM: Add KVM_SEV_RECEIVE_UPDATE_DATA command") Fixes: d3d1af85e2c7 ("KVM: SVM: Add KVM_SEND_UPDATE_DATA command") Reported-by: Andy Nguyen theflow@google.com Suggested-by: Thomas Lendacky thomas.lendacky@amd.com Signed-off-by: Peter Gonda pgonda@google.com Cc: David Rientjes rientjes@google.com Cc: Paolo Bonzini pbonzini@redhat.com Cc: Sean Christopherson seanjc@google.com Cc: kvm@vger.kernel.org Cc: stable@vger.kernel.org Cc: linux-kernel@vger.kernel.org
Reviewed-by: Tom Lendacky thomas.lendacky@amd.com
V2
- Updated conditional based on feedback from Tom.
arch/x86/kvm/svm/sev.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c index 273cba809328..3d74facaead8 100644 --- a/arch/x86/kvm/svm/sev.c +++ b/arch/x86/kvm/svm/sev.c @@ -1294,7 +1294,7 @@ static int sev_send_update_data(struct kvm *kvm, struct kvm_sev_cmd *argp) /* Check if we are crossing the page boundary */ offset = params.guest_uaddr & (PAGE_SIZE - 1);
- if ((params.guest_len + offset > PAGE_SIZE))
- if (params.guest_len > PAGE_SIZE || (params.guest_len + offset) > PAGE_SIZE) return -EINVAL;
/* Pin guest memory */ @@ -1474,7 +1474,7 @@ static int sev_receive_update_data(struct kvm *kvm, struct kvm_sev_cmd *argp) /* Check if we are crossing the page boundary */ offset = params.guest_uaddr & (PAGE_SIZE - 1);
- if ((params.guest_len + offset > PAGE_SIZE))
- if (params.guest_len > PAGE_SIZE || (params.guest_len + offset) > PAGE_SIZE) return -EINVAL;
hdr = psp_copy_user_blob(params.hdr_uaddr, params.hdr_len);
For now at least, I want to keep with "KVM: SVM:" instead of using "KVM: SEV:". Many commits that touch SEV aren't strictly isolated to SEV, which means the "SEV" tag is unreliable. There's also the question of taggin SEV vs. SEV-ES vs. SEV-SNP. It's usually easy enough to squeeze SEV (or SEV-ES or SNP) into the shortlog, e.g.
KVM: SVM: Fix potential overflow in SEV's send|receive_update_data()
On Tue, Feb 07, 2023, Peter Gonda wrote:
KVM_SEV_SEND_UPDATE_DATA and KVM_SEV_RECEIVE_UPDATE_DATA have an integer overflow issue. Params.guest_len and offset are both 32bite wide, with a
"32 bits"
large params.guest_len the check to confirm a page boundary is not crossed can falsely pass:
/* Check if we are crossing the page boundary * offset = params.guest_uaddr & (PAGE_SIZE - 1); if ((params.guest_len + offset > PAGE_SIZE))
Add an additional check to this conditional to confirm that
Eh, "to this conditional" is unnecessarily precise.
params.guest_len itself is not greater than PAGE_SIZE.
The current code is can only overflow with a params.guest_len of greater
"is can", though I vote to omit the "current code" part entirely, it should be obvious that this is talking about the pre-patched code.
than 0xfffff000. And the FW spec says these commands fail with lengths greater than 16KB. So this issue should not be a security concern
Slightly reworded, how about this for the "not a security concern" disclaimer?
Note, this isn't a security concern as overflow can happen if and only if params.guest_len is greater than 0xfffff000, and the FW spec says these commands fail with lengths greater than 16KB, i.e. the PSP will detect KVM's goof.
No need to send a v3, I'll fix up the changelog when applying. Holler if you disagree with anything though.
Thanks!
On Tue, 07 Feb 2023 09:13:54 -0800, Peter Gonda wrote:
KVM_SEV_SEND_UPDATE_DATA and KVM_SEV_RECEIVE_UPDATE_DATA have an integer overflow issue. Params.guest_len and offset are both 32bite wide, with a large params.guest_len the check to confirm a page boundary is not crossed can falsely pass:
/* Check if we are crossing the page boundary * offset = params.guest_uaddr & (PAGE_SIZE - 1); if ((params.guest_len + offset > PAGE_SIZE))
[...]
Applied to kvm-x86 svm, thanks!
[1/1] KVM: sev: Fix potential overflow send|recieve_update_data https://github.com/kvm-x86/linux/commit/f94f053aa3a5
-- https://github.com/kvm-x86/linux/tree/next https://github.com/kvm-x86/linux/tree/fixes
linux-stable-mirror@lists.linaro.org