Blob Blame History Raw
From 742fee241938f6089d67c4e779ba0d608a9d88e3 Mon Sep 17 00:00:00 2001
From: Joe Lawrence <joe.lawrence@redhat.com>
Date: Mon, 30 Aug 2021 16:54:36 -0400
Subject: [KPATCH CVE-2021-37576] powerpc: kpatch fixes for CVE-2021-37576

Kernels:
4.18.0-305.el8
4.18.0-305.3.1.el8_4
4.18.0-305.7.1.el8_4
4.18.0-305.10.2.el8_4
4.18.0-305.12.1.el8_4

arches: ppc64le
Changes since last build:
[ppc64le]:
book3s_rtas.o: changed function: kvmppc_rtas_hcall

---------------------------

Kernels:
4.18.0-305.el8
4.18.0-305.3.1.el8_4
4.18.0-305.7.1.el8_4
4.18.0-305.10.2.el8_4
4.18.0-305.12.1.el8_4

Modifications: none
Approved-by: Yannick Cote (@ycote1)
Approved-by: Artem Savkov (@artem.savkov)
KPATCH-MR: https://gitlab.com/kpatch-dev/rhel-8/-/merge_requests/2

KT0 test PASS: https://beaker.engineering.redhat.com/jobs/5756102
for kpatch-patch-4_18_0-305-1-5.el8 scratch build:
https://brewweb.engineering.redhat.com/brew/taskinfo?taskID=39394966

commit 82faab596fc8f92648f20e2fbc4211557b115c13
Author: Jon Maloy <jmaloy@redhat.com>
Date:   Thu Aug 12 19:22:51 2021 -0400

    KVM: PPC: Book3S: Fix H_RTAS rets buffer overflow

    Bugzilla: https://bugzilla.redhat.com/1988225
    Upstream Status: Merged
    Build Info: https://brewweb.engineering.redhat.com/brew/taskinfo?taskID=38936146
    CVE: CVE-2021-37576

    commit f62f3c20647ebd5fb6ecb8f0b477b9281c44c10a
    Author: Nicholas Piggin <npiggin@gmail.com>
    Date:   Tue Jul 20 20:43:09 2021 +1000

        KVM: PPC: Book3S: Fix H_RTAS rets buffer overflow

        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 <aik@ozlabs.ru>
        Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
        Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

    Signed-off-by: Jon Maloy <jmaloy@redhat.com>

Signed-off-by: Joe Lawrence <joe.lawrence@redhat.com>
---
 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 ceccacbf028e..52095f765e32 100644
--- a/arch/powerpc/kvm/book3s_rtas.c
+++ b/arch/powerpc/kvm/book3s_rtas.c
@@ -245,6 +245,17 @@ 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);
@@ -272,9 +283,17 @@ 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;
 }
-- 
2.31.1