s390/unwind: introduce stack unwind API

Rework the dump_trace() stack unwinder interface to support different
unwinding algorithms. The new interface looks like this:

	struct unwind_state state;
	unwind_for_each_frame(&state, task, regs, start_stack)
		do_something(state.sp, state.ip, state.reliable);

The unwind_bc.c file contains the implementation for the classic
back-chain unwinder.

One positive side effect of the new code is it now handles ftraced
functions gracefully. It prints the real name of the return function
instead of 'return_to_handler'.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

11 files changed, 297 insertions, 119 deletions

diff --git a/arch/s390/kernel/Makefile b/arch/s390/kernel/Makefile
index 19425605a83d..b0478d01a0c5 100644
--- a/arch/s390/kernel/Makefile
+++ b/arch/s390/kernel/Makefile
@@ -39,6 +39,7 @@ CFLAGS_smp.o		:= -Wno-nonnull
 #
 CFLAGS_stacktrace.o	+= -fno-optimize-sibling-calls
 CFLAGS_dumpstack.o	+= -fno-optimize-sibling-calls
+CFLAGS_unwind_bc.o	+= -fno-optimize-sibling-calls
 
 #
 # Pass UTS_MACHINE for user_regset definition
@@ -51,7 +52,7 @@ obj-y	+= debug.o irq.o ipl.o dis.o diag.o vdso.o early_nobss.o
 obj-y	+= sysinfo.o lgr.o os_info.o machine_kexec.o pgm_check.o
 obj-y	+= runtime_instr.o cache.o fpu.o dumpstack.o guarded_storage.o sthyi.o
 obj-y	+= entry.o reipl.o relocate_kernel.o kdebugfs.o alternative.o
-obj-y	+= nospec-branch.o ipl_vmparm.o machine_kexec_reloc.o
+obj-y	+= nospec-branch.o ipl_vmparm.o machine_kexec_reloc.o unwind_bc.o
 
 extra-y				+= head64.o vmlinux.lds
 
diff --git a/arch/s390/kernel/asm-offsets.c b/arch/s390/kernel/asm-offsets.c
index 164bec175628..41ac4ad21311 100644
--- a/arch/s390/kernel/asm-offsets.c
+++ b/arch/s390/kernel/asm-offsets.c
@@ -16,6 +16,7 @@
 #include <asm/pgtable.h>
 #include <asm/gmap.h>
 #include <asm/nmi.h>
+#include <asm/stacktrace.h>
 
 int main(void)
 {
diff --git a/arch/s390/kernel/dumpstack.c b/arch/s390/kernel/dumpstack.c
index cb7f55bbe06e..9e87b68be21c 100644
--- a/arch/s390/kernel/dumpstack.c
+++ b/arch/s390/kernel/dumpstack.c
@@ -21,95 +21,124 @@
 #include <asm/debug.h>
 #include <asm/dis.h>
 #include <asm/ipl.h>
+#include <asm/unwind.h>
 
-/*
- * For dump_trace we have tree different stack to consider:
- *   - the panic stack which is used if the kernel stack has overflown
- *   - the asynchronous interrupt stack (cpu related)
- *   - the synchronous kernel stack (process related)
- * The stack trace can start at any of the three stacks and can potentially
- * touch all of them. The order is: panic stack, async stack, sync stack.
- */
-static unsigned long __no_sanitize_address
-__dump_trace(dump_trace_func_t func, void *data, unsigned long sp,
-	     unsigned long low, unsigned long high)
+const char *stack_type_name(enum stack_type type)
 {
-	struct stack_frame *sf;
-	struct pt_regs *regs;
-
-	while (1) {
-		if (sp < low || sp > high - sizeof(*sf))
-			return sp;
-		sf = (struct stack_frame *) sp;
-		if (func(data, sf->gprs[8], 0))
-			return sp;
-		/* Follow the backchain. */
-		while (1) {
-			low = sp;
-			sp = sf->back_chain;
-			if (!sp)
-				break;
-			if (sp <= low || sp > high - sizeof(*sf))
-				return sp;
-			sf = (struct stack_frame *) sp;
-			if (func(data, sf->gprs[8], 1))
-				return sp;
-		}
-		/* Zero backchain detected, check for interrupt frame. */
-		sp = (unsigned long) (sf + 1);
-		if (sp <= low || sp > high - sizeof(*regs))
-			return sp;
-		regs = (struct pt_regs *) sp;
-		if (!user_mode(regs)) {
-			if (func(data, regs->psw.addr, 1))
-				return sp;
-		}
-		low = sp;
-		sp = regs->gprs[15];
+	switch (type) {
+	case STACK_TYPE_TASK:
+		return "task";
+	case STACK_TYPE_IRQ:
+		return "irq";
+	case STACK_TYPE_NODAT:
+		return "nodat";
+	case STACK_TYPE_RESTART:
+		return "restart";
+	default:
+		return "unknown";
 	}
 }
 
-void dump_trace(dump_trace_func_t func, void *data, struct task_struct *task,
-		unsigned long sp)
+static inline bool in_stack(unsigned long sp, struct stack_info *info,
+			    enum stack_type type, unsigned long low,
+			    unsigned long high)
+{
+	if (sp < low || sp >= high)
+		return false;
+	info->type = type;
+	info->begin = low;
+	info->end = high;
+	return true;
+}
+
+static bool in_task_stack(unsigned long sp, struct task_struct *task,
+			  struct stack_info *info)
+{
+	unsigned long stack;
+
+	stack = (unsigned long) task_stack_page(task);
+	return in_stack(sp, info, STACK_TYPE_TASK, stack, stack + THREAD_SIZE);
+}
+
+static bool in_irq_stack(unsigned long sp, struct stack_info *info)
 {
-	unsigned long frame_size;
+	unsigned long frame_size, top;
 
 	frame_size = STACK_FRAME_OVERHEAD + sizeof(struct pt_regs);
-#ifdef CONFIG_CHECK_STACK
-	sp = __dump_trace(func, data, sp,
-			  S390_lowcore.nodat_stack + frame_size - THREAD_SIZE,
-			  S390_lowcore.nodat_stack + frame_size);
-#endif
-	sp = __dump_trace(func, data, sp,
-			  S390_lowcore.async_stack + frame_size - THREAD_SIZE,
-			  S390_lowcore.async_stack + frame_size);
-	task = task ?: current;
-	__dump_trace(func, data, sp,
-		     (unsigned long)task_stack_page(task),
-		     (unsigned long)task_stack_page(task) + THREAD_SIZE);
+	top = S390_lowcore.async_stack + frame_size;
+	return in_stack(sp, info, STACK_TYPE_IRQ, top - THREAD_SIZE, top);
+}
+
+static bool in_nodat_stack(unsigned long sp, struct stack_info *info)
+{
+	unsigned long frame_size, top;
+
+	frame_size = STACK_FRAME_OVERHEAD + sizeof(struct pt_regs);
+	top = S390_lowcore.nodat_stack + frame_size;
+	return in_stack(sp, info, STACK_TYPE_NODAT, top - THREAD_SIZE, top);
 }
-EXPORT_SYMBOL_GPL(dump_trace);
 
-static int show_address(void *data, unsigned long address, int reliable)
+static bool in_restart_stack(unsigned long sp, struct stack_info *info)
 {
-	if (reliable)
-		printk(" [<%016lx>] %pSR \n", address, (void *)address);
-	else
-		printk("([<%016lx>] %pSR)\n", address, (void *)address);
+	unsigned long frame_size, top;
+
+	frame_size = STACK_FRAME_OVERHEAD + sizeof(struct pt_regs);
+	top = S390_lowcore.restart_stack + frame_size;
+	return in_stack(sp, info, STACK_TYPE_RESTART, top - THREAD_SIZE, top);
+}
+
+int get_stack_info(unsigned long sp, struct task_struct *task,
+		   struct stack_info *info, unsigned long *visit_mask)
+{
+	if (!sp)
+		goto unknown;
+
+	task = task ? : current;
+
+	/* Check per-task stack */
+	if (in_task_stack(sp, task, info))
+		goto recursion_check;
+
+	if (task != current)
+		goto unknown;
+
+	/* Check per-cpu stacks */
+	if (!in_irq_stack(sp, info) &&
+	    !in_nodat_stack(sp, info) &&
+	    !in_restart_stack(sp, info))
+		goto unknown;
+
+recursion_check:
+	/*
+	 * Make sure we don't iterate through any given stack more than once.
+	 * If it comes up a second time then there's something wrong going on:
+	 * just break out and report an unknown stack type.
+	 */
+	if (*visit_mask & (1UL << info->type)) {
+		printk_deferred_once(KERN_WARNING
+			"WARNING: stack recursion on stack type %d\n",
+			info->type);
+		goto unknown;
+	}
+	*visit_mask |= 1UL << info->type;
 	return 0;
+unknown:
+	info->type = STACK_TYPE_UNKNOWN;
+	return -EINVAL;
 }
 
 void show_stack(struct task_struct *task, unsigned long *stack)
 {
-	unsigned long sp = (unsigned long) stack;
+	struct unwind_state state;
 
-	if (!sp)
-		sp = task ? task->thread.ksp : current_stack_pointer();
 	printk("Call Trace:\n");
-	dump_trace(show_address, NULL, task, sp);
 	if (!task)
 		task = current;
-	debug_show_held_locks(task);
+	unwind_for_each_frame(&state, task, NULL, (unsigned long) stack)
+		printk(state.reliable ? " [<%016lx>] %pSR \n" :
+					"([<%016lx>] %pSR)\n",
+		       state.ip, (void *) state.ip);
+	debug_show_held_locks(task ? : current);
 }
 
 static void show_last_breaking_event(struct pt_regs *regs)
diff --git a/arch/s390/kernel/irq.c b/arch/s390/kernel/irq.c
index 150964f91183..8371855042dc 100644
--- a/arch/s390/kernel/irq.c
+++ b/arch/s390/kernel/irq.c
@@ -26,6 +26,7 @@
 #include <asm/lowcore.h>
 #include <asm/irq.h>
 #include <asm/hw_irq.h>
+#include <asm/stacktrace.h>
 #include "entry.h"
 
 DEFINE_PER_CPU_SHARED_ALIGNED(struct irq_stat, irq_stat);
diff --git a/arch/s390/kernel/machine_kexec.c b/arch/s390/kernel/machine_kexec.c
index e2ba7b7f574e..2f3a742a71a5 100644
--- a/arch/s390/kernel/machine_kexec.c
+++ b/arch/s390/kernel/machine_kexec.c
@@ -27,6 +27,7 @@
 #include <asm/cacheflush.h>
 #include <asm/os_info.h>
 #include <asm/set_memory.h>
+#include <asm/stacktrace.h>
 #include <asm/switch_to.h>
 #include <asm/nmi.h>
 
diff --git a/arch/s390/kernel/perf_event.c b/arch/s390/kernel/perf_event.c
index 0d770e513abf..fcb6c2e92b07 100644
--- a/arch/s390/kernel/perf_event.c
+++ b/arch/s390/kernel/perf_event.c
@@ -21,6 +21,7 @@
 #include <asm/lowcore.h>
 #include <asm/processor.h>
 #include <asm/sysinfo.h>
+#include <asm/unwind.h>
 
 const char *perf_pmu_name(void)
 {
@@ -219,20 +220,13 @@ static int __init service_level_perf_register(void)
 }
 arch_initcall(service_level_perf_register);
 
-static int __perf_callchain_kernel(void *data, unsigned long address, int reliable)
-{
-	struct perf_callchain_entry_ctx *entry = data;
-
-	perf_callchain_store(entry, address);
-	return 0;
-}
-
 void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry,
 			   struct pt_regs *regs)
 {
-	if (user_mode(regs))
-		return;
-	dump_trace(__perf_callchain_kernel, entry, NULL, regs->gprs[15]);
+	struct unwind_state state;
+
+	unwind_for_each_frame(&state, current, regs, 0)
+		perf_callchain_store(entry, state.ip);
 }
 
 /* Perf definitions for PMU event attributes in sysfs */
diff --git a/arch/s390/kernel/process.c b/arch/s390/kernel/process.c
index 6e758bb6cd29..63873aa6693f 100644
--- a/arch/s390/kernel/process.c
+++ b/arch/s390/kernel/process.c
@@ -37,6 +37,7 @@
 #include <asm/irq.h>
 #include <asm/nmi.h>
 #include <asm/smp.h>
+#include <asm/stacktrace.h>
 #include <asm/switch_to.h>
 #include <asm/runtime_instr.h>
 #include "entry.h"
diff --git a/arch/s390/kernel/setup.c b/arch/s390/kernel/setup.c
index 64e4bc9dd130..f8544d517430 100644
--- a/arch/s390/kernel/setup.c
+++ b/arch/s390/kernel/setup.c
@@ -66,6 +66,7 @@
 #include <asm/diag.h>
 #include <asm/os_info.h>
 #include <asm/sclp.h>
+#include <asm/stacktrace.h>
 #include <asm/sysinfo.h>
 #include <asm/numa.h>
 #include <asm/alternative.h>
diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c
index 88634fb0cc50..35fafa2b91a8 100644
--- a/arch/s390/kernel/smp.c
+++ b/arch/s390/kernel/smp.c
@@ -53,6 +53,7 @@
 #include <asm/sigp.h>
 #include <asm/idle.h>
 #include <asm/nmi.h>
+#include <asm/stacktrace.h>
 #include <asm/topology.h>
 #include "entry.h"
 
diff --git a/arch/s390/kernel/stacktrace.c b/arch/s390/kernel/stacktrace.c
index 460dcfba7d4e..89f9f63dca18 100644
--- a/arch/s390/kernel/stacktrace.c
+++ b/arch/s390/kernel/stacktrace.c
@@ -11,40 +11,21 @@
 #include <linux/stacktrace.h>
 #include <linux/kallsyms.h>
 #include <linux/export.h>
-
-static int __save_address(void *data, unsigned long address, int nosched)
-{
-	struct stack_trace *trace = data;
-
-	if (nosched && in_sched_functions(address))
-		return 0;
-	if (trace->skip > 0) {
-		trace->skip--;
-		return 0;
-	}
-	if (trace->nr_entries < trace->max_entries) {
-		trace->entries[trace->nr_entries++] = address;
-		return 0;
-	}
-	return 1;
-}
-
-static int save_address(void *data, unsigned long address, int reliable)
-{
-	return __save_address(data, address, 0);
-}
-
-static int save_address_nosched(void *data, unsigned long address, int reliable)
-{
-	return __save_address(data, address, 1);
-}
+#include <asm/stacktrace.h>
+#include <asm/unwind.h>
 
 void save_stack_trace(struct stack_trace *trace)
 {
-	unsigned long sp;
+	struct unwind_state state;
 
-	sp = current_stack_pointer();
-	dump_trace(save_address, trace, NULL, sp);
+	unwind_for_each_frame(&state, current, NULL, 0) {
+		if (trace->nr_entries >= trace->max_entries)
+			break;
+		if (trace->skip > 0)
+			trace->skip--;
+		else
+			trace->entries[trace->nr_entries++] = state.ip;
+	}
 	if (trace->nr_entries < trace->max_entries)
 		trace->entries[trace->nr_entries++] = ULONG_MAX;
 }
@@ -52,12 +33,18 @@ EXPORT_SYMBOL_GPL(save_stack_trace);
 
 void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace)
 {
-	unsigned long sp;
+	struct unwind_state state;
 
-	sp = tsk->thread.ksp;
-	if (tsk == current)
-		sp = current_stack_pointer();
-	dump_trace(save_address_nosched, trace, tsk, sp);
+	unwind_for_each_frame(&state, tsk, NULL, 0) {
+		if (trace->nr_entries >= trace->max_entries)
+			break;
+		if (in_sched_functions(state.ip))
+			continue;
+		if (trace->skip > 0)
+			trace->skip--;
+		else
+			trace->entries[trace->nr_entries++] = state.ip;
+	}
 	if (trace->nr_entries < trace->max_entries)
 		trace->entries[trace->nr_entries++] = ULONG_MAX;
 }
@@ -65,10 +52,16 @@ EXPORT_SYMBOL_GPL(save_stack_trace_tsk);
 
 void save_stack_trace_regs(struct pt_regs *regs, struct stack_trace *trace)
 {
-	unsigned long sp;
+	struct unwind_state state;
 
-	sp = kernel_stack_pointer(regs);
-	dump_trace(save_address, trace, NULL, sp);
+	unwind_for_each_frame(&state, current, regs, 0) {
+		if (trace->nr_entries >= trace->max_entries)
+			break;
+		if (trace->skip > 0)
+			trace->skip--;
+		else
+			trace->entries[trace->nr_entries++] = state.ip;
+	}
 	if (trace->nr_entries < trace->max_entries)
 		trace->entries[trace->nr_entries++] = ULONG_MAX;
 }
diff --git a/arch/s390/kernel/unwind_bc.c b/arch/s390/kernel/unwind_bc.c
new file mode 100644
index 000000000000..cf5a630f3aa9
--- /dev/null
+++ b/arch/s390/kernel/unwind_bc.c
@@ -0,0 +1,155 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#include <linux/sched.h>
+#include <linux/sched/task.h>
+#include <linux/sched/task_stack.h>
+#include <linux/interrupt.h>
+#include <asm/sections.h>
+#include <asm/ptrace.h>
+#include <asm/bitops.h>
+#include <asm/stacktrace.h>
+#include <asm/unwind.h>
+
+unsigned long unwind_get_return_address(struct unwind_state *state)
+{
+	if (unwind_done(state))
+		return 0;
+	return __kernel_text_address(state->ip) ? state->ip : 0;
+}
+EXPORT_SYMBOL_GPL(unwind_get_return_address);
+
+static bool outside_of_stack(struct unwind_state *state, unsigned long sp)
+{
+	return (sp <= state->sp) ||
+		(sp + sizeof(struct stack_frame) > state->stack_info.end);
+}
+
+static bool update_stack_info(struct unwind_state *state, unsigned long sp)
+{
+	struct stack_info *info = &state->stack_info;
+	unsigned long *mask = &state->stack_mask;
+
+	/* New stack pointer leaves the current stack */
+	if (get_stack_info(sp, state->task, info, mask) != 0 ||
+	    !on_stack(info, sp, sizeof(struct stack_frame)))
+		/* 'sp' does not point to a valid stack */
+		return false;
+	return true;
+}
+
+bool unwind_next_frame(struct unwind_state *state)
+{
+	struct stack_info *info = &state->stack_info;
+	struct stack_frame *sf;
+	struct pt_regs *regs;
+	unsigned long sp, ip;
+	bool reliable;
+
+	regs = state->regs;
+	if (unlikely(regs)) {
+		sp = READ_ONCE_TASK_STACK(state->task, regs->gprs[15]);
+		if (unlikely(outside_of_stack(state, sp))) {
+			if (!update_stack_info(state, sp))
+				goto out_err;
+		}
+		sf = (struct stack_frame *) sp;
+		ip = READ_ONCE_TASK_STACK(state->task, sf->gprs[8]);
+		reliable = false;
+		regs = NULL;
+	} else {
+		sf = (struct stack_frame *) state->sp;
+		sp = READ_ONCE_TASK_STACK(state->task, sf->back_chain);
+		if (likely(sp)) {
+			/* Non-zero back-chain points to the previous frame */
+			if (unlikely(outside_of_stack(state, sp))) {
+				if (!update_stack_info(state, sp))
+					goto out_err;
+			}
+			sf = (struct stack_frame *) sp;
+			ip = READ_ONCE_TASK_STACK(state->task, sf->gprs[8]);
+			reliable = true;
+		} else {
+			/* No back-chain, look for a pt_regs structure */
+			sp = state->sp + STACK_FRAME_OVERHEAD;
+			if (!on_stack(info, sp, sizeof(struct pt_regs)))
+				goto out_stop;
+			regs = (struct pt_regs *) sp;
+			if (user_mode(regs))
+				goto out_stop;
+			ip = READ_ONCE_TASK_STACK(state->task, regs->psw.addr);
+			reliable = true;
+		}
+	}
+
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+	/* Decode any ftrace redirection */
+	if (ip == (unsigned long) return_to_handler)
+		ip = ftrace_graph_ret_addr(state->task, &state->graph_idx,
+					   ip, NULL);
+#endif
+
+	/* Update unwind state */
+	state->sp = sp;
+	state->ip = ip;
+	state->regs = regs;
+	state->reliable = reliable;
+	return true;
+
+out_err:
+	state->error = true;
+out_stop:
+	state->stack_info.type = STACK_TYPE_UNKNOWN;
+	return false;
+}
+EXPORT_SYMBOL_GPL(unwind_next_frame);
+
+void __unwind_start(struct unwind_state *state, struct task_struct *task,
+		    struct pt_regs *regs, unsigned long sp)
+{
+	struct stack_info *info = &state->stack_info;
+	unsigned long *mask = &state->stack_mask;
+	struct stack_frame *sf;
+	unsigned long ip;
+	bool reliable;
+
+	memset(state, 0, sizeof(*state));
+	state->task = task;
+	state->regs = regs;
+
+	/* Don't even attempt to start from user mode regs: */
+	if (regs && user_mode(regs)) {
+		info->type = STACK_TYPE_UNKNOWN;
+		return;
+	}
+
+	/* Get current stack pointer and initialize stack info */
+	if (get_stack_info(sp, task, info, mask) != 0 ||
+	    !on_stack(info, sp, sizeof(struct stack_frame))) {
+		/* Something is wrong with the stack pointer */
+		info->type = STACK_TYPE_UNKNOWN;
+		state->error = true;
+		return;
+	}
+
+	/* Get the instruction pointer from pt_regs or the stack frame */
+	if (regs) {
+		ip = READ_ONCE_TASK_STACK(state->task, regs->psw.addr);
+		reliable = true;
+	} else {
+		sf = (struct stack_frame *) sp;
+		ip = READ_ONCE_TASK_STACK(state->task, sf->gprs[8]);
+		reliable = false;
+	}
+
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+	/* Decode any ftrace redirection */
+	if (ip == (unsigned long) return_to_handler)
+		ip = ftrace_graph_ret_addr(state->task, &state->graph_idx,
+					   ip, NULL);
+#endif
+
+	/* Update unwind state */
+	state->sp = sp;
+	state->ip = ip;
+	state->reliable = reliable;
+}
+EXPORT_SYMBOL_GPL(__unwind_start);