Merge tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux

Pull arm64 updates from Will Deacon:
 "Here is the core arm64 queue for 4.5.  As you might expect, the
  Christmas break resulted in a number of patches not making the final
  cut, so 4.6 is likely to be larger than usual.  There's still some
  useful stuff here, however, and it's detailed below.

  The EFI changes have been Reviewed-by Matt and the memblock change got
  an "OK" from akpm.

  Summary:

   - Support for a separate IRQ stack, although we haven't reduced the
     size of our thread stack just yet since we don't have enough data
     to determine a safe value

   - Refactoring of our EFI initialisation and runtime code into
     drivers/firmware/efi/ so that it can be reused by arch/arm/.

   - Ftrace improvements when unwinding in the function graph tracer

   - Document our silicon errata handling process

   - Cache flushing optimisation when mapping executable pages

   - Support for hugetlb mappings using the contiguous hint in the pte"

* tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux: (45 commits)
  arm64: head.S: use memset to clear BSS
  efi: stub: define DISABLE_BRANCH_PROFILING for all architectures
  arm64: entry: remove pointless SPSR mode check
  arm64: mm: move pgd_cache initialisation to pgtable_cache_init
  arm64: module: avoid undefined shift behavior in reloc_data()
  arm64: module: fix relocation of movz instruction with negative immediate
  arm64: traps: address fallout from printk -> pr_* conversion
  arm64: ftrace: fix a stack tracer's output under function graph tracer
  arm64: pass a task parameter to unwind_frame()
  arm64: ftrace: modify a stack frame in a safe way
  arm64: remove irq_count and do_softirq_own_stack()
  arm64: hugetlb: add support for PTE contiguous bit
  arm64: Use PoU cache instr for I/D coherency
  arm64: Defer dcache flush in __cpu_copy_user_page
  arm64: reduce stack use in irq_handler
  arm64: mm: ensure that the zero page is visible to the page table walker
  arm64: Documentation: add list of software workarounds for errata
  arm64: mm: place __cpu_setup in .text
  arm64: cmpxchg: Don't incldue linux/mmdebug.h
  arm64: mm: fold alternatives into .init
  ...

1 files changed, 69 insertions, 6 deletions

diff --git a/arch/arm64/kernel/stacktrace.c b/arch/arm64/kernel/stacktrace.c
index ccb6078ed9f2..4fad9787ab46 100644
--- a/arch/arm64/kernel/stacktrace.c
+++ b/arch/arm64/kernel/stacktrace.c
@@ -17,9 +17,11 @@
  */
 #include <linux/kernel.h>
 #include <linux/export.h>
+#include <linux/ftrace.h>
 #include <linux/sched.h>
 #include <linux/stacktrace.h>
 
+#include <asm/irq.h>
 #include <asm/stacktrace.h>
 
 /*
@@ -35,25 +37,83 @@
  *	ldp	x29, x30, [sp]
  *	add	sp, sp, #0x10
  */
-int notrace unwind_frame(struct stackframe *frame)
+int notrace unwind_frame(struct task_struct *tsk, struct stackframe *frame)
 {
 	unsigned long high, low;
 	unsigned long fp = frame->fp;
+	unsigned long irq_stack_ptr;
+
+	/*
+	 * Use raw_smp_processor_id() to avoid false-positives from
+	 * CONFIG_DEBUG_PREEMPT. get_wchan() calls unwind_frame() on sleeping
+	 * task stacks, we can be pre-empted in this case, so
+	 * {raw_,}smp_processor_id() may give us the wrong value. Sleeping
+	 * tasks can't ever be on an interrupt stack, so regardless of cpu,
+	 * the checks will always fail.
+	 */
+	irq_stack_ptr = IRQ_STACK_PTR(raw_smp_processor_id());
 
 	low  = frame->sp;
-	high = ALIGN(low, THREAD_SIZE);
+	/* irq stacks are not THREAD_SIZE aligned */
+	if (on_irq_stack(frame->sp, raw_smp_processor_id()))
+		high = irq_stack_ptr;
+	else
+		high = ALIGN(low, THREAD_SIZE) - 0x20;
 
-	if (fp < low || fp > high - 0x18 || fp & 0xf)
+	if (fp < low || fp > high || fp & 0xf)
 		return -EINVAL;
 
 	frame->sp = fp + 0x10;
 	frame->fp = *(unsigned long *)(fp);
 	frame->pc = *(unsigned long *)(fp + 8);
 
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+	if (tsk && tsk->ret_stack &&
+			(frame->pc == (unsigned long)return_to_handler)) {
+		/*
+		 * This is a case where function graph tracer has
+		 * modified a return address (LR) in a stack frame
+		 * to hook a function return.
+		 * So replace it to an original value.
+		 */
+		frame->pc = tsk->ret_stack[frame->graph--].ret;
+	}
+#endif /* CONFIG_FUNCTION_GRAPH_TRACER */
+
+	/*
+	 * Check whether we are going to walk through from interrupt stack
+	 * to task stack.
+	 * If we reach the end of the stack - and its an interrupt stack,
+	 * unpack the dummy frame to find the original elr.
+	 *
+	 * Check the frame->fp we read from the bottom of the irq_stack,
+	 * and the original task stack pointer are both in current->stack.
+	 */
+	if (frame->sp == irq_stack_ptr) {
+		struct pt_regs *irq_args;
+		unsigned long orig_sp = IRQ_STACK_TO_TASK_STACK(irq_stack_ptr);
+
+		if (object_is_on_stack((void *)orig_sp) &&
+		   object_is_on_stack((void *)frame->fp)) {
+			frame->sp = orig_sp;
+
+			/* orig_sp is the saved pt_regs, find the elr */
+			irq_args = (struct pt_regs *)orig_sp;
+			frame->pc = irq_args->pc;
+		} else {
+			/*
+			 * This frame has a non-standard format, and we
+			 * didn't fix it, because the data looked wrong.
+			 * Refuse to output this frame.
+			 */
+			return -EINVAL;
+		}
+	}
+
 	return 0;
 }
 
-void notrace walk_stackframe(struct stackframe *frame,
+void notrace walk_stackframe(struct task_struct *tsk, struct stackframe *frame,
 		     int (*fn)(struct stackframe *, void *), void *data)
 {
 	while (1) {
@@ -61,7 +121,7 @@ void notrace walk_stackframe(struct stackframe *frame,
 
 		if (fn(frame, data))
 			break;
-		ret = unwind_frame(frame);
+		ret = unwind_frame(tsk, frame);
 		if (ret < 0)
 			break;
 	}
@@ -112,8 +172,11 @@ void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace)
 		frame.sp = current_stack_pointer;
 		frame.pc = (unsigned long)save_stack_trace_tsk;
 	}
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+	frame.graph = tsk->curr_ret_stack;
+#endif
 
-	walk_stackframe(&frame, save_trace, &data);
+	walk_stackframe(tsk, &frame, save_trace, &data);
 	if (trace->nr_entries < trace->max_entries)
 		trace->entries[trace->nr_entries++] = ULONG_MAX;
 }