Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Akihiko Odaki <akihiko.odaki@daynix.com>
Cc:     linux-kernel@vger.kernel.org, kvmarm@lists.linux.dev,
        kvmarm@lists.cs.columbia.edu, linux-arm-kernel@lists.infradead.org,
        Mathieu Poirier <mathieu.poirier@linaro.org>,
        Oliver Upton <oliver.upton@linux.dev>,
        Suzuki K Poulose <suzuki.poulose@arm.com>,
        Alexandru Elisei <alexandru.elisei@arm.com>,
        James Morse <james.morse@arm.com>,
        Marc Zyngier <maz@kernel.org>, Will Deacon <will@kernel.org>,
        Catalin Marinas <catalin.marinas@arm.com>,
        asahi@lists.linux.dev, Alyssa Rosenzweig <alyssa@rosenzweig.io>,
        Sven Peter <sven@svenpeter.dev>,
        Hector Martin <marcan@marcan.st>,
        Akihiko Odaki <akihiko.odaki@daynix.com>
Subject: [PATCH 1/3] KVM: arm64: Make CCSIDRs consistent
Date:   Thu,  1 Dec 2022 19:49:12 +0900
Message-Id: <20221201104914.28944-2-akihiko.odaki@daynix.com>
In-Reply-To: <20221201104914.28944-1-akihiko.odaki@daynix.com>
References: <20221201104914.28944-1-akihiko.odaki@daynix.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
To:     unlisted-recipients:; (no To-header on input)
Precedence: bulk

A vCPU sees masked CCSIDRs when the physical CPUs has mismatched
cache types or the vCPU has 32-bit EL1. Perform the same masking for
ioctls too so that ioctls shows values consistent with the values the
vCPU actually sees.

Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
---
 arch/arm64/include/asm/kvm_emulate.h |  9 ++++--
 arch/arm64/kvm/sys_regs.c            | 45 ++++++++++++++--------------
 2 files changed, 30 insertions(+), 24 deletions(-)

diff --git a/arch/arm64/include/asm/kvm_emulate.h b/arch/arm64/include/asm/kvm_emulate.h
index 9bdba47f7e14..b45cf8903190 100644
--- a/arch/arm64/include/asm/kvm_emulate.h
+++ b/arch/arm64/include/asm/kvm_emulate.h
@@ -61,6 +61,12 @@ static __always_inline bool vcpu_el1_is_32bit(struct kvm_vcpu *vcpu)
 }
 #endif
 
+static inline bool vcpu_cache_overridden(struct kvm_vcpu *vcpu)
+{
+	return cpus_have_const_cap(ARM64_MISMATCHED_CACHE_TYPE) ||
+	       vcpu_el1_is_32bit(vcpu);
+}
+
 static inline void vcpu_reset_hcr(struct kvm_vcpu *vcpu)
 {
 	vcpu->arch.hcr_el2 = HCR_GUEST_FLAGS;
@@ -88,8 +94,7 @@ static inline void vcpu_reset_hcr(struct kvm_vcpu *vcpu)
 	if (vcpu_el1_is_32bit(vcpu))
 		vcpu->arch.hcr_el2 &= ~HCR_RW;
 
-	if (cpus_have_const_cap(ARM64_MISMATCHED_CACHE_TYPE) ||
-	    vcpu_el1_is_32bit(vcpu))
+	if (vcpu_cache_overridden(vcpu))
 		vcpu->arch.hcr_el2 |= HCR_TID2;
 
 	if (kvm_has_mte(vcpu->kvm))
diff --git a/arch/arm64/kvm/sys_regs.c b/arch/arm64/kvm/sys_regs.c
index f4a7c5abcbca..273ed1aaa6b3 100644
--- a/arch/arm64/kvm/sys_regs.c
+++ b/arch/arm64/kvm/sys_regs.c
@@ -88,7 +88,7 @@ static u32 cache_levels;
 #define CSSELR_MAX 14
 
 /* Which cache CCSIDR represents depends on CSSELR value. */
-static u32 get_ccsidr(u32 csselr)
+static u32 get_ccsidr(struct kvm_vcpu *vcpu, u32 csselr)
 {
 	u32 ccsidr;
 
@@ -99,6 +99,21 @@ static u32 get_ccsidr(u32 csselr)
 	ccsidr = read_sysreg(ccsidr_el1);
 	local_irq_enable();
 
+	/*
+	 * Guests should not be doing cache operations by set/way at all, and
+	 * for this reason, we trap them and attempt to infer the intent, so
+	 * that we can flush the entire guest's address space at the appropriate
+	 * time.
+	 * To prevent this trapping from causing performance problems, let's
+	 * expose the geometry of all data and unified caches (which are
+	 * guaranteed to be PIPT and thus non-aliasing) as 1 set and 1 way.
+	 * [If guests should attempt to infer aliasing properties from the
+	 * geometry (which is not permitted by the architecture), they would
+	 * only do so for virtually indexed caches.]
+	 */
+	if (vcpu_cache_overridden(vcpu) && !(csselr & 1)) // data or unified cache
+		ccsidr &= ~GENMASK(27, 3);
+
 	return ccsidr;
 }
 
@@ -1300,22 +1315,8 @@ static bool access_ccsidr(struct kvm_vcpu *vcpu, struct sys_reg_params *p,
 		return write_to_read_only(vcpu, p, r);
 
 	csselr = vcpu_read_sys_reg(vcpu, CSSELR_EL1);
-	p->regval = get_ccsidr(csselr);
+	p->regval = get_ccsidr(vcpu, csselr);
 
-	/*
-	 * Guests should not be doing cache operations by set/way at all, and
-	 * for this reason, we trap them and attempt to infer the intent, so
-	 * that we can flush the entire guest's address space at the appropriate
-	 * time.
-	 * To prevent this trapping from causing performance problems, let's
-	 * expose the geometry of all data and unified caches (which are
-	 * guaranteed to be PIPT and thus non-aliasing) as 1 set and 1 way.
-	 * [If guests should attempt to infer aliasing properties from the
-	 * geometry (which is not permitted by the architecture), they would
-	 * only do so for virtually indexed caches.]
-	 */
-	if (!(csselr & 1)) // data or unified cache
-		p->regval &= ~GENMASK(27, 3);
 	return true;
 }
 
@@ -2686,7 +2687,7 @@ static bool is_valid_cache(u32 val)
 	}
 }
 
-static int demux_c15_get(u64 id, void __user *uaddr)
+static int demux_c15_get(struct kvm_vcpu *vcpu, u64 id, void __user *uaddr)
 {
 	u32 val;
 	u32 __user *uval = uaddr;
@@ -2705,13 +2706,13 @@ static int demux_c15_get(u64 id, void __user *uaddr)
 		if (!is_valid_cache(val))
 			return -ENOENT;
 
-		return put_user(get_ccsidr(val), uval);
+		return put_user(get_ccsidr(vcpu, val), uval);
 	default:
 		return -ENOENT;
 	}
 }
 
-static int demux_c15_set(u64 id, void __user *uaddr)
+static int demux_c15_set(struct kvm_vcpu *vcpu, u64 id, void __user *uaddr)
 {
 	u32 val, newval;
 	u32 __user *uval = uaddr;
@@ -2734,7 +2735,7 @@ static int demux_c15_set(u64 id, void __user *uaddr)
 			return -EFAULT;
 
 		/* This is also invariant: you can't change it. */
-		if (newval != get_ccsidr(val))
+		if (newval != get_ccsidr(vcpu, val))
 			return -EINVAL;
 		return 0;
 	default:
@@ -2773,7 +2774,7 @@ int kvm_arm_sys_reg_get_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg
 	int err;
 
 	if ((reg->id & KVM_REG_ARM_COPROC_MASK) == KVM_REG_ARM_DEMUX)
-		return demux_c15_get(reg->id, uaddr);
+		return demux_c15_get(vcpu, reg->id, uaddr);
 
 	err = get_invariant_sys_reg(reg->id, uaddr);
 	if (err != -ENOENT)
@@ -2817,7 +2818,7 @@ int kvm_arm_sys_reg_set_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg
 	int err;
 
 	if ((reg->id & KVM_REG_ARM_COPROC_MASK) == KVM_REG_ARM_DEMUX)
-		return demux_c15_set(reg->id, uaddr);
+		return demux_c15_set(vcpu, reg->id, uaddr);
 
 	err = set_invariant_sys_reg(reg->id, uaddr);
 	if (err != -ENOENT)
-- 
2.38.1