Patch "x86/unwind: Fix last frame check for aligned function stacks" has been added to the 4.10-stable tree

This is a note to let you know that I've just added the patch titled

x86/unwind: Fix last frame check for aligned function stacks

to the 4.10-stable tree which can be found at:
http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary

The filename of the patch is:
x86-unwind-fix-last-frame-check-for-aligned-function-stacks.patch
and it can be found in the queue-4.10 subdirectory.

If you, or anyone else, feels it should not be added to the stable tree, please let <[email protected]> know about it.

From: Josh Poimboeuf <[email protected]>
Date: Mon, 13 Mar 2017 23:27:47 -0500
Subject: x86/unwind: Fix last frame check for aligned function stacks

From: Josh Poimboeuf <[email protected]>

commit 87a6b2975f0d340c75b7488d22d61d2f98fb8abf upstream.

Pavel Machek reported the following warning on x86-32:

WARNING: kernel stack frame pointer at f50cdf98 in swapper/2:0 has bad value (null)

The warning is caused by the unwinder not realizing that it reached the end of the stack, due to an unusual prologue which gcc sometimes generates for aligned stacks. The prologue is based on a gcc feature called the Dynamic Realign Argument Pointer (DRAP). It's almost always enabled for aligned stacks when -maccumulate-outgoing-args isn't set.

This issue is similar to the one fixed by the following commit:

8023e0e2a48d ("x86/unwind: Adjust last frame check for aligned function stacks")

... but that fix was specific to x86-64.

Make the fix more generic to cover x86-32 as well, and also ensure that the return address referred to by the frame pointer is a copy of the original return address.

---
 arch/x86/kernel/unwind_frame.c |   36 ++++++++++++++++++++++++++++++------
 1 file changed, 30 insertions(+), 6 deletions(-)

@@ -80,19 +80,43 @@ static size_t regs_size(struct pt_regs *
 	return sizeof(*regs);
 }
 
+#ifdef CONFIG_X86_32
+#define GCC_REALIGN_WORDS 3
+#else
+#define GCC_REALIGN_WORDS 1
+#endif
+
 static bool is_last_task_frame(struct unwind_state *state)
 {
-	unsigned long bp = (unsigned long)state->bp;
-	unsigned long regs = (unsigned long)task_pt_regs(state->task);
+	unsigned long *last_bp = (unsigned long *)task_pt_regs(state->task) - 2;
+	unsigned long *aligned_bp = last_bp - GCC_REALIGN_WORDS;
 
 	/*
 	 * We have to check for the last task frame at two different locations
 	 * because gcc can occasionally decide to realign the stack pointer and
-	 * change the offset of the stack frame by a word in the prologue of a
-	 * function called by head/entry code.
+	 * change the offset of the stack frame in the prologue of a function
+	 * called by head/entry code.  Examples:
+	 *
+	 * <start_secondary>:
+	 *      push   %edi
+	 *      lea    0x8(%esp),%edi
+	 *      and    $0xfffffff8,%esp
+	 *      pushl  -0x4(%edi)
+	 *      push   %ebp
+	 *      mov    %esp,%ebp
+	 *
+	 * <x86_64_start_kernel>:
+	 *      lea    0x8(%rsp),%r10
+	 *      and    $0xfffffffffffffff0,%rsp
+	 *      pushq  -0x8(%r10)
+	 *      push   %rbp
+	 *      mov    %rsp,%rbp
+	 *
+	 * Note that after aligning the stack, it pushes a duplicate copy of
+	 * the return address before pushing the frame pointer.
 	 */
-	return bp == regs - FRAME_HEADER_SIZE ||
-	       bp == regs - FRAME_HEADER_SIZE - sizeof(long);
+	return (state->bp == last_bp ||
+		(state->bp == aligned_bp && *(aligned_bp+1) == *(last_bp+1)));
 }
 
 /*

Patches currently in stable-queue which might be from [email protected] are

queue-4.10/x86-unwind-fix-last-frame-check-for-aligned-function-stacks.patch