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;
Date:   Wed, 20 Jul 2022 22:57:23 -0700
In-Reply-To: <20220721055728.718573-1-kaleshsingh@google.com>
Message-Id: <20220721055728.718573-13-kaleshsingh@google.com>
Mime-Version: 1.0
References: <20220721055728.718573-1-kaleshsingh@google.com>
Subject: [PATCH v5 12/17] KVM: arm64: Save protected-nVHE (pKVM) hyp stacktrace
From:   Kalesh Singh <kaleshsingh@google.com>
To:     maz@kernel.org, mark.rutland@arm.com, broonie@kernel.org,
        madvenka@linux.microsoft.com, tabba@google.com
Cc:     will@kernel.org, qperret@google.com, kaleshsingh@google.com,
        james.morse@arm.com, alexandru.elisei@arm.com,
        suzuki.poulose@arm.com, catalin.marinas@arm.com,
        andreyknvl@gmail.com, vincenzo.frascino@arm.com,
        mhiramat@kernel.org, ast@kernel.org, drjones@redhat.com,
        wangkefeng.wang@huawei.com, elver@google.com, keirf@google.com,
        yuzenghui@huawei.com, ardb@kernel.org, oupton@google.com,
        linux-arm-kernel@lists.infradead.org, kvmarm@lists.cs.columbia.edu,
        linux-kernel@vger.kernel.org, android-mm@google.com,
        kernel-team@android.com
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

In protected nVHE mode, the host cannot access private owned hypervisor
memory. Also the hypervisor aims to remains simple to reduce the attack
surface and does not provide any printk support.

For the above reasons, the approach taken to provide hypervisor stacktraces
in protected mode is:
   1) Unwind and save the hyp stack addresses in EL2 to a shared buffer
      with the host (done in this patch).
   2) Delegate the dumping and symbolization of the addresses to the
      host in EL1 (later patch in the series).

On hyp_panic(), the hypervisor prepares the stacktrace before returning to
the host.

Signed-off-by: Kalesh Singh <kaleshsingh@google.com>
---

Changes in v5:
  - Comment/clarify pkvm_save_backtrace_entry(), per Fuad
  - kvm_nvhe_unwind_init(), doesn't need to be always inline, make it
    inline instead to avoid linking issues, per Marc
  - Use regular comments instead of doc comments, per Fuad

 arch/arm64/include/asm/stacktrace/nvhe.h | 17 ++++++
 arch/arm64/kvm/hyp/nvhe/stacktrace.c     | 78 ++++++++++++++++++++++++
 arch/arm64/kvm/hyp/nvhe/switch.c         |  6 ++
 3 files changed, 101 insertions(+)

diff --git a/arch/arm64/include/asm/stacktrace/nvhe.h b/arch/arm64/include/asm/stacktrace/nvhe.h
index 3078501f8e22..05d7e03e0a8c 100644
--- a/arch/arm64/include/asm/stacktrace/nvhe.h
+++ b/arch/arm64/include/asm/stacktrace/nvhe.h
@@ -21,6 +21,23 @@
 
 #include <asm/stacktrace/common.h>
 
+/*
+ * kvm_nvhe_unwind_init - Start an unwind from the given nVHE HYP fp and pc
+ *
+ * @state : unwind_state to initialize
+ * @fp    : frame pointer at which to start the unwinding.
+ * @pc    : program counter at which to start the unwinding.
+ */
+static inline void kvm_nvhe_unwind_init(struct unwind_state *state,
+					unsigned long fp,
+					unsigned long pc)
+{
+	unwind_init_common(state, NULL);
+
+	state->fp = fp;
+	state->pc = pc;
+}
+
 static inline bool on_accessible_stack(const struct task_struct *tsk,
 				       unsigned long sp, unsigned long size,
 				       struct stack_info *info)
diff --git a/arch/arm64/kvm/hyp/nvhe/stacktrace.c b/arch/arm64/kvm/hyp/nvhe/stacktrace.c
index 96c8b93320eb..60461c033a04 100644
--- a/arch/arm64/kvm/hyp/nvhe/stacktrace.c
+++ b/arch/arm64/kvm/hyp/nvhe/stacktrace.c
@@ -11,4 +11,82 @@ DEFINE_PER_CPU(unsigned long [OVERFLOW_STACK_SIZE/sizeof(long)], overflow_stack)
 
 #ifdef CONFIG_PROTECTED_NVHE_STACKTRACE
 DEFINE_PER_CPU(unsigned long [NVHE_STACKTRACE_SIZE/sizeof(long)], pkvm_stacktrace);
+
+/*
+ * pkvm_save_backtrace_entry - Saves a protected nVHE HYP stacktrace entry
+ *
+ * @arg    : the position of the entry in the stacktrace buffer
+ * @where  : the program counter corresponding to the stack frame
+ *
+ * Save the return address of a stack frame to the shared stacktrace buffer.
+ * The host can access this shared buffer from EL1 to dump the backtrace.
+ */
+static bool pkvm_save_backtrace_entry(void *arg, unsigned long where)
+{
+	unsigned long **stacktrace_entry = (unsigned long **)arg;
+	int nr_entries = NVHE_STACKTRACE_SIZE / sizeof(long);
+	unsigned long *stacktrace_start, *stacktrace_end;
+
+	stacktrace_start = (unsigned long *)this_cpu_ptr(pkvm_stacktrace);
+	stacktrace_end = stacktrace_start + nr_entries;
+
+	/*
+	 * Need 2 free slots: 1 for current entry and 1 for the
+	 * trailing zero entry delimiter.
+	 */
+	if (*stacktrace_entry > stacktrace_end - 2)
+		return false;
+
+	/* Save the current entry */
+	**stacktrace_entry = where;
+
+	/* Add trailing zero entry delimiter */
+	*(*stacktrace_entry + 1) = 0UL;
+
+	/*
+	 * Increment the current entry position. The zero entry
+	 * will be overwritten by the next backtrace entry (if any)
+	 */
+	++*stacktrace_entry;
+
+	return true;
+}
+
+/*
+ * pkvm_save_backtrace - Saves the protected nVHE HYP stacktrace
+ *
+ * @fp : frame pointer at which to start the unwinding.
+ * @pc : program counter at which to start the unwinding.
+ *
+ * Save the unwinded stack addresses to the shared stacktrace buffer.
+ * The host can access this shared buffer from EL1 to dump the backtrace.
+ */
+static void pkvm_save_backtrace(unsigned long fp, unsigned long pc)
+{
+	void *stacktrace_entry = (void *)this_cpu_ptr(pkvm_stacktrace);
+	struct unwind_state state;
+
+	kvm_nvhe_unwind_init(&state, fp, pc);
+
+	unwind(&state, pkvm_save_backtrace_entry, &stacktrace_entry);
+}
+#else /* !CONFIG_PROTECTED_NVHE_STACKTRACE */
+static void pkvm_save_backtrace(unsigned long fp, unsigned long pc)
+{
+}
 #endif /* CONFIG_PROTECTED_NVHE_STACKTRACE */
+
+/*
+ * kvm_nvhe_prepare_backtrace - prepare to dump the nVHE backtrace
+ *
+ * @fp : frame pointer at which to start the unwinding.
+ * @pc : program counter at which to start the unwinding.
+ *
+ * Saves the information needed by the host to dump the nVHE hypervisor
+ * backtrace.
+ */
+void kvm_nvhe_prepare_backtrace(unsigned long fp, unsigned long pc)
+{
+	if (is_protected_kvm_enabled())
+		pkvm_save_backtrace(fp, pc);
+}
diff --git a/arch/arm64/kvm/hyp/nvhe/switch.c b/arch/arm64/kvm/hyp/nvhe/switch.c
index 6db801db8f27..64e13445d0d9 100644
--- a/arch/arm64/kvm/hyp/nvhe/switch.c
+++ b/arch/arm64/kvm/hyp/nvhe/switch.c
@@ -34,6 +34,8 @@ DEFINE_PER_CPU(struct kvm_host_data, kvm_host_data);
 DEFINE_PER_CPU(struct kvm_cpu_context, kvm_hyp_ctxt);
 DEFINE_PER_CPU(unsigned long, kvm_hyp_vector);
 
+extern void kvm_nvhe_prepare_backtrace(unsigned long fp, unsigned long pc);
+
 static void __activate_traps(struct kvm_vcpu *vcpu)
 {
 	u64 val;
@@ -375,6 +377,10 @@ asmlinkage void __noreturn hyp_panic(void)
 		__sysreg_restore_state_nvhe(host_ctxt);
 	}
 
+	/* Prepare to dump kvm nvhe hyp stacktrace */
+	kvm_nvhe_prepare_backtrace((unsigned long)__builtin_frame_address(0),
+				   _THIS_IP_);
+
 	__hyp_do_panic(host_ctxt, spsr, elr, par);
 	unreachable();
 }
-- 
2.37.0.170.g444d1eabd0-goog