linux-kernel - [PATCH 1/2] arm64: ARM_SMCCC_ARCH_WORKAROUND_1 doesn't return SMCCC_RET_NOT

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20201020214544.3206838-2-swboyd@chromium.org>
Date:   Tue, 20 Oct 2020 14:45:43 -0700
From:   Stephen Boyd <swboyd@...omium.org>
To:     Will Deacon <will@...nel.org>,
        Catalin Marinas <catalin.marinas@....com>
Cc:     linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
        Andre Przywara <andre.przywara@....com>,
        Steven Price <steven.price@....com>,
        Marc Zyngier <maz@...nel.org>, stable@...r.kernel.org
Subject: [PATCH 1/2] arm64: ARM_SMCCC_ARCH_WORKAROUND_1 doesn't return SMCCC_RET_NOT_REQUIRED

According to the SMCCC spec (7.5.2 Discovery) the
ARM_SMCCC_ARCH_WORKAROUND_1 function id only returns 0, 1, and
SMCCC_RET_NOT_SUPPORTED corresponding to "workaround required",
"workaround not required but implemented", and "who knows, you're on
your own" respectively. For kvm hypercalls (hvc), we've implemented this
function id to return SMCCC_RET_NOT_SUPPORTED, 1, and
SMCCC_RET_NOT_REQUIRED. The SMCCC_RET_NOT_REQUIRED return value is not a
thing for this function id, and is probably copy/pasted from the
SMCCC_ARCH_WORKAROUND_2 function id that does support it.

Clean this up by returning 0, 1, and SMCCC_RET_NOT_SUPPORTED
appropriately. Changing this exposes the problem that
spectre_v2_get_cpu_fw_mitigation_state() assumes a
SMCCC_RET_NOT_SUPPORTED return value means we are vulnerable, but really
it means we have no idea and should assume we can't do anything about
mitigation. Put another way, it better be unaffected because it can't be
mitigated in the firmware (in this case kvm) as the call isn't
implemented!

Cc: Andre Przywara <andre.przywara@....com>
Cc: Steven Price <steven.price@....com>
Cc: Marc Zyngier <maz@...nel.org>
Cc: stable@...r.kernel.org
Fixes: c118bbb52743 ("arm64: KVM: Propagate full Spectre v2 workaround state to KVM guests")
Fixes: 73f381660959 ("arm64: Advertise mitigation of Spectre-v2, or lack thereof")
Signed-off-by: Stephen Boyd <swboyd@...omium.org>
---

This will require a slightly different backport to stable kernels, but
at least it looks like this is a problem given that this return value
isn't valid per the spec and we've been going around it by returning
something invalid for some time.

 arch/arm64/kernel/proton-pack.c | 3 +--
 arch/arm64/kvm/hypercalls.c     | 2 +-
 2 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/arch/arm64/kernel/proton-pack.c b/arch/arm64/kernel/proton-pack.c
index 68b710f1b43f..00bd54f63f4f 100644
--- a/arch/arm64/kernel/proton-pack.c
+++ b/arch/arm64/kernel/proton-pack.c
@@ -149,10 +149,9 @@ static enum mitigation_state spectre_v2_get_cpu_fw_mitigation_state(void)
 	case SMCCC_RET_SUCCESS:
 		return SPECTRE_MITIGATED;
 	case SMCCC_ARCH_WORKAROUND_RET_UNAFFECTED:
+	case SMCCC_RET_NOT_SUPPORTED: /* Good luck w/ the Gatekeeper of Gozer */
 		return SPECTRE_UNAFFECTED;
 	default:
-		fallthrough;
-	case SMCCC_RET_NOT_SUPPORTED:
 		return SPECTRE_VULNERABLE;
 	}
 }
diff --git a/arch/arm64/kvm/hypercalls.c b/arch/arm64/kvm/hypercalls.c
index 9824025ccc5c..868486957808 100644
--- a/arch/arm64/kvm/hypercalls.c
+++ b/arch/arm64/kvm/hypercalls.c
@@ -31,7 +31,7 @@ int kvm_hvc_call_handler(struct kvm_vcpu *vcpu)
 				val = SMCCC_RET_SUCCESS;
 				break;
 			case SPECTRE_UNAFFECTED:
-				val = SMCCC_RET_NOT_REQUIRED;
+				val = SMCCC_RET_NOT_SUPPORTED;
 				break;
 			}
 			break;
-- 
Sent by a computer, using git, on the internet