From: Nicholas Piggin
stable inclusion
from linux-4.19.199
commit 0493b10c06021796ba80cbe53c961defd5aca6e5
category: bugfix
issue: #I4CBD6
CVE: CVE-2021-37576
--------------------------------
commit f62f3c20647ebd5fb6ecb8f0b477b9281c44c10a upstream.
The kvmppc_rtas_hcall() sets the host rtas_args.rets pointer based on
the rtas_args.nargs that was provided by the guest. That guest nargs
value is not range checked, so the guest can cause the host rets pointer
to be pointed outside the args array. The individual rtas function
handlers check the nargs and nrets values to ensure they are correct,
but if they are not, the handlers store a -3 (0xfffffffd) failure
indication in rets[0] which corrupts host memory.
Fix this by testing up front whether the guest supplied nargs and nret
would exceed the array size, and fail the hcall directly without storing
a failure indication to rets[0].
Also expand on a comment about why we kill the guest and try not to
return errors directly if we have a valid rets[0] pointer.
Fixes: 8e591cb72047 ("KVM: PPC: Book3S: Add infrastructure to implement kernel-side RTAS calls")
Cc: stable@vger.kernel.org # v3.10+
Reported-by: Alexey Kardashevskiy
Signed-off-by: Nicholas Piggin
Signed-off-by: Michael Ellerman
Signed-off-by: Greg Kroah-Hartman
Signed-off-by: Yang Yingliang
Reviewed-by: Xiu Jianfeng
Signed-off-by: Yang Yingliang
Signed-off-by: Yu Changchun
---
arch/powerpc/kvm/book3s_rtas.c | 25 ++++++++++++++++++++++---
1 file changed, 22 insertions(+), 3 deletions(-)
diff --git a/arch/powerpc/kvm/book3s_rtas.c b/arch/powerpc/kvm/book3s_rtas.c
index 8f2355138f80..a56c56aa829c 100644
--- a/arch/powerpc/kvm/book3s_rtas.c
+++ b/arch/powerpc/kvm/book3s_rtas.c
@@ -243,6 +243,17 @@ int kvmppc_rtas_hcall(struct kvm_vcpu *vcpu)
* value so we can restore it on the way out.
*/
orig_rets = args.rets;
+ if (be32_to_cpu(args.nargs) >= ARRAY_SIZE(args.args)) {
+ /*
+ * Don't overflow our args array: ensure there is room for
+ * at least rets[0] (even if the call specifies 0 nret).
+ *
+ * Each handler must then check for the correct nargs and nret
+ * values, but they may always return failure in rets[0].
+ */
+ rc = -EINVAL;
+ goto fail;
+ }
args.rets = &args.args[be32_to_cpu(args.nargs)];
mutex_lock(&vcpu->kvm->arch.rtas_token_lock);
@@ -270,9 +281,17 @@ int kvmppc_rtas_hcall(struct kvm_vcpu *vcpu)
fail:
/*
* We only get here if the guest has called RTAS with a bogus
- * args pointer. That means we can't get to the args, and so we
- * can't fail the RTAS call. So fail right out to userspace,
- * which should kill the guest.
+ * args pointer or nargs/nret values that would overflow the
+ * array. That means we can't get to the args, and so we can't
+ * fail the RTAS call. So fail right out to userspace, which
+ * should kill the guest.
+ *
+ * SLOF should actually pass the hcall return value from the
+ * rtas handler call in r3, so enter_rtas could be modified to
+ * return a failure indication in r3 and we could return such
+ * errors to the guest rather than failing to host userspace.
+ * However old guests that don't test for failure could then
+ * continue silently after errors, so for now we won't do this.
*/
return rc;
}
--
2.22.0