Merge branch kvm-arm64/nvhe-stacktrace into kvmarm-master/next

* kvm-arm64/nvhe-stacktrace: (27 commits)
  : .
  : Add an overflow stack to the nVHE EL2 code, allowing
  : the implementation of an unwinder, courtesy of
  : Kalesh Singh. From the cover letter (slightly edited):
  :
  : "nVHE has two modes of operation: protected (pKVM) and unprotected
  : (conventional nVHE). Depending on the mode, a slightly different approach
  : is used to dump the hypervisor stacktrace but the core unwinding logic
  : remains the same.
  :
  : * Protected nVHE (pKVM) stacktraces:
  :
  : In protected nVHE mode, the host cannot directly access hypervisor memory.
  :
  : The hypervisor stack unwinding happens in EL2 and is made accessible to
  : the host via a shared buffer. Symbolizing and printing the stacktrace
  : addresses is delegated to the host and happens in EL1.
  :
  : * Non-protected (Conventional) nVHE stacktraces:
  :
  : In non-protected mode, the host is able to directly access the hypervisor
  : stack pages.
  :
  : The hypervisor stack unwinding and dumping of the stacktrace is performed
  : by the host in EL1, as this avoids the memory overhead of setting up
  : shared buffers between the host and hypervisor."
  :
  : Additional patches from Oliver Upton and Marc Zyngier, tidying up
  : the initial series.
  : .
  arm64: Update 'unwinder howto'
  KVM: arm64: Don't open code ARRAY_SIZE()
  KVM: arm64: Move nVHE-only helpers into kvm/stacktrace.c
  KVM: arm64: Make unwind()/on_accessible_stack() per-unwinder functions
  KVM: arm64: Move nVHE stacktrace unwinding into its own compilation unit
  KVM: arm64: Move PROTECTED_NVHE_STACKTRACE around
  KVM: arm64: Introduce pkvm_dump_backtrace()
  KVM: arm64: Implement protected nVHE hyp stack unwinder
  KVM: arm64: Save protected-nVHE (pKVM) hyp stacktrace
  KVM: arm64: Stub implementation of pKVM HYP stack unwinder
  KVM: arm64: Allocate shared pKVM hyp stacktrace buffers
  KVM: arm64: Add PROTECTED_NVHE_STACKTRACE Kconfig
  KVM: arm64: Introduce hyp_dump_backtrace()
  KVM: arm64: Implement non-protected nVHE hyp stack unwinder
  KVM: arm64: Prepare non-protected nVHE hypervisor stacktrace
  KVM: arm64: Stub implementation of non-protected nVHE HYP stack unwinder
  KVM: arm64: On stack overflow switch to hyp overflow_stack
  arm64: stacktrace: Add description of stacktrace/common.h
  arm64: stacktrace: Factor out common unwind()
  arm64: stacktrace: Handle frame pointer from different address spaces
  ...

Signed-off-by: Marc Zyngier <maz@kernel.org>

1 files changed, 84 insertions, 100 deletions

diff --git a/arch/arm64/kernel/stacktrace.c b/arch/arm64/kernel/stacktrace.c
index 0467cb79f080..ce190ee18a20 100644
--- a/arch/arm64/kernel/stacktrace.c
+++ b/arch/arm64/kernel/stacktrace.c
@@ -7,72 +7,90 @@
 #include <linux/kernel.h>
 #include <linux/export.h>
 #include <linux/ftrace.h>
-#include <linux/kprobes.h>
 #include <linux/sched.h>
 #include <linux/sched/debug.h>
 #include <linux/sched/task_stack.h>
 #include <linux/stacktrace.h>
 
 #include <asm/irq.h>
-#include <asm/pointer_auth.h>
 #include <asm/stack_pointer.h>
 #include <asm/stacktrace.h>
 
 /*
- * A snapshot of a frame record or fp/lr register values, along with some
- * accounting information necessary for robust unwinding.
+ * Start an unwind from a pt_regs.
  *
- * @fp:          The fp value in the frame record (or the real fp)
- * @pc:          The lr value in the frame record (or the real lr)
+ * The unwind will begin at the PC within the regs.
  *
- * @stacks_done: Stacks which have been entirely unwound, for which it is no
- *               longer valid to unwind to.
+ * The regs must be on a stack currently owned by the calling task.
+ */
+static inline void unwind_init_from_regs(struct unwind_state *state,
+					 struct pt_regs *regs)
+{
+	unwind_init_common(state, current);
+
+	state->fp = regs->regs[29];
+	state->pc = regs->pc;
+}
+
+/*
+ * Start an unwind from a caller.
  *
- * @prev_fp:     The fp that pointed to this frame record, or a synthetic value
- *               of 0. This is used to ensure that within a stack, each
- *               subsequent frame record is at an increasing address.
- * @prev_type:   The type of stack this frame record was on, or a synthetic
- *               value of STACK_TYPE_UNKNOWN. This is used to detect a
- *               transition from one stack to another.
+ * The unwind will begin at the caller of whichever function this is inlined
+ * into.
  *
- * @kr_cur:      When KRETPROBES is selected, holds the kretprobe instance
- *               associated with the most recently encountered replacement lr
- *               value.
+ * The function which invokes this must be noinline.
  */
-struct unwind_state {
-	unsigned long fp;
-	unsigned long pc;
-	DECLARE_BITMAP(stacks_done, __NR_STACK_TYPES);
-	unsigned long prev_fp;
-	enum stack_type prev_type;
-#ifdef CONFIG_KRETPROBES
-	struct llist_node *kr_cur;
-#endif
-};
+static __always_inline void unwind_init_from_caller(struct unwind_state *state)
+{
+	unwind_init_common(state, current);
+
+	state->fp = (unsigned long)__builtin_frame_address(1);
+	state->pc = (unsigned long)__builtin_return_address(0);
+}
 
-static notrace void unwind_init(struct unwind_state *state, unsigned long fp,
-				unsigned long pc)
+/*
+ * Start an unwind from a blocked task.
+ *
+ * The unwind will begin at the blocked tasks saved PC (i.e. the caller of
+ * cpu_switch_to()).
+ *
+ * The caller should ensure the task is blocked in cpu_switch_to() for the
+ * duration of the unwind, or the unwind will be bogus. It is never valid to
+ * call this for the current task.
+ */
+static inline void unwind_init_from_task(struct unwind_state *state,
+					 struct task_struct *task)
 {
-	state->fp = fp;
-	state->pc = pc;
-#ifdef CONFIG_KRETPROBES
-	state->kr_cur = NULL;
-#endif
+	unwind_init_common(state, task);
+
+	state->fp = thread_saved_fp(task);
+	state->pc = thread_saved_pc(task);
+}
 
-	/*
-	 * Prime the first unwind.
-	 *
-	 * In unwind_next() we'll check that the FP points to a valid stack,
-	 * which can't be STACK_TYPE_UNKNOWN, and the first unwind will be
-	 * treated as a transition to whichever stack that happens to be. The
-	 * prev_fp value won't be used, but we set it to 0 such that it is
-	 * definitely not an accessible stack address.
-	 */
-	bitmap_zero(state->stacks_done, __NR_STACK_TYPES);
-	state->prev_fp = 0;
-	state->prev_type = STACK_TYPE_UNKNOWN;
+/*
+ * We can only safely access per-cpu stacks from current in a non-preemptible
+ * context.
+ */
+static bool on_accessible_stack(const struct task_struct *tsk,
+				unsigned long sp, unsigned long size,
+				struct stack_info *info)
+{
+	if (info)
+		info->type = STACK_TYPE_UNKNOWN;
+
+	if (on_task_stack(tsk, sp, size, info))
+		return true;
+	if (tsk != current || preemptible())
+		return false;
+	if (on_irq_stack(sp, size, info))
+		return true;
+	if (on_overflow_stack(sp, size, info))
+		return true;
+	if (on_sdei_stack(sp, size, info))
+		return true;
+
+	return false;
 }
-NOKPROBE_SYMBOL(unwind_init);
 
 /*
  * Unwind from one frame record (A) to the next frame record (B).
@@ -81,53 +99,20 @@ NOKPROBE_SYMBOL(unwind_init);
  * records (e.g. a cycle), determined based on the location and fp value of A
  * and the location (but not the fp value) of B.
  */
-static int notrace unwind_next(struct task_struct *tsk,
-			       struct unwind_state *state)
+static int notrace unwind_next(struct unwind_state *state)
 {
+	struct task_struct *tsk = state->task;
 	unsigned long fp = state->fp;
 	struct stack_info info;
+	int err;
 
 	/* Final frame; nothing to unwind */
 	if (fp == (unsigned long)task_pt_regs(tsk)->stackframe)
 		return -ENOENT;
 
-	if (fp & 0x7)
-		return -EINVAL;
-
-	if (!on_accessible_stack(tsk, fp, 16, &info))
-		return -EINVAL;
-
-	if (test_bit(info.type, state->stacks_done))
-		return -EINVAL;
-
-	/*
-	 * As stacks grow downward, any valid record on the same stack must be
-	 * at a strictly higher address than the prior record.
-	 *
-	 * Stacks can nest in several valid orders, e.g.
-	 *
-	 * TASK -> IRQ -> OVERFLOW -> SDEI_NORMAL
-	 * TASK -> SDEI_NORMAL -> SDEI_CRITICAL -> OVERFLOW
-	 *
-	 * ... but the nesting itself is strict. Once we transition from one
-	 * stack to another, it's never valid to unwind back to that first
-	 * stack.
-	 */
-	if (info.type == state->prev_type) {
-		if (fp <= state->prev_fp)
-			return -EINVAL;
-	} else {
-		set_bit(state->prev_type, state->stacks_done);
-	}
-
-	/*
-	 * Record this frame record's values and location. The prev_fp and
-	 * prev_type are only meaningful to the next unwind_next() invocation.
-	 */
-	state->fp = READ_ONCE_NOCHECK(*(unsigned long *)(fp));
-	state->pc = READ_ONCE_NOCHECK(*(unsigned long *)(fp + 8));
-	state->prev_fp = fp;
-	state->prev_type = info.type;
+	err = unwind_next_common(state, &info, on_accessible_stack, NULL);
+	if (err)
+		return err;
 
 	state->pc = ptrauth_strip_insn_pac(state->pc);
 
@@ -157,8 +142,7 @@ static int notrace unwind_next(struct task_struct *tsk,
 }
 NOKPROBE_SYMBOL(unwind_next);
 
-static void notrace unwind(struct task_struct *tsk,
-			   struct unwind_state *state,
+static void notrace unwind(struct unwind_state *state,
 			   stack_trace_consume_fn consume_entry, void *cookie)
 {
 	while (1) {
@@ -166,7 +150,7 @@ static void notrace unwind(struct task_struct *tsk,
 
 		if (!consume_entry(cookie, state->pc))
 			break;
-		ret = unwind_next(tsk, state);
+		ret = unwind_next(state);
 		if (ret < 0)
 			break;
 	}
@@ -212,15 +196,15 @@ noinline notrace void arch_stack_walk(stack_trace_consume_fn consume_entry,
 {
 	struct unwind_state state;
 
-	if (regs)
-		unwind_init(&state, regs->regs[29], regs->pc);
-	else if (task == current)
-		unwind_init(&state,
-				(unsigned long)__builtin_frame_address(1),
-				(unsigned long)__builtin_return_address(0));
-	else
-		unwind_init(&state, thread_saved_fp(task),
-				thread_saved_pc(task));
-
-	unwind(task, &state, consume_entry, cookie);
+	if (regs) {
+		if (task != current)
+			return;
+		unwind_init_from_regs(&state, regs);
+	} else if (task == current) {
+		unwind_init_from_caller(&state);
+	} else {
+		unwind_init_from_task(&state, task);
+	}
+
+	unwind(&state, consume_entry, cookie);
 }