selftests/x86: Add a test for vDSO unwinding

While the kernel itself doesn't use DWARF unwinding, user code
expects to be able to unwind the vDSO.  The vsyscall
(AT_SYSINFO) entry is manually CFI-annotated, and this tests
that it unwinds correctly.

I tested the test by incorrectly annotating __kernel_vsyscall,
and the test indeed fails if I do that.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Shuah Khan <shuahkh@osg.samsung.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Link: http://lkml.kernel.org/r/8bf736d1925cdd165c0f980156a4248e55af47a1.1444091584.git.luto@kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>

2 files changed, 210 insertions, 1 deletions

diff --git a/tools/testing/selftests/x86/Makefile b/tools/testing/selftests/x86/Makefile
index fd55bc3..7541352 100644
--- a/tools/testing/selftests/x86/Makefile
+++ b/tools/testing/selftests/x86/Makefile
@@ -5,7 +5,7 @@ include ../lib.mk
 .PHONY: all all_32 all_64 warn_32bit_failure clean
 
 TARGETS_C_BOTHBITS := single_step_syscall sysret_ss_attrs ldt_gdt syscall_nt
-TARGETS_C_32BIT_ONLY := entry_from_vm86 syscall_arg_fault sigreturn test_syscall_vdso
+TARGETS_C_32BIT_ONLY := entry_from_vm86 syscall_arg_fault sigreturn test_syscall_vdso unwind_vdso
 
 TARGETS_C_32BIT_ALL := $(TARGETS_C_BOTHBITS) $(TARGETS_C_32BIT_ONLY)
 BINARIES_32 := $(TARGETS_C_32BIT_ALL:%=%_32)
diff --git a/tools/testing/selftests/x86/unwind_vdso.c b/tools/testing/selftests/x86/unwind_vdso.c
new file mode 100644
index 0000000..5992ff2
--- /dev/null
+++ b/tools/testing/selftests/x86/unwind_vdso.c
@@ -0,0 +1,209 @@
+/*
+ * unwind_vdso.c - tests unwind info for AT_SYSINFO in the vDSO
+ * Copyright (c) 2014-2015 Andrew Lutomirski
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * This tests __kernel_vsyscall's unwind info.
+ */
+
+#define _GNU_SOURCE
+
+#include <features.h>
+#include <stdio.h>
+
+#if defined(__GLIBC__) && __GLIBC__ == 2 && __GLIBC_MINOR__ < 16
+
+int main()
+{
+	/* We need getauxval(). */
+	printf("[SKIP]\tGLIBC before 2.16 cannot compile this test\n");
+	return 0;
+}
+
+#else
+
+#include <sys/time.h>
+#include <stdlib.h>
+#include <syscall.h>
+#include <unistd.h>
+#include <string.h>
+#include <inttypes.h>
+#include <sys/mman.h>
+#include <signal.h>
+#include <sys/ucontext.h>
+#include <err.h>
+#include <stddef.h>
+#include <stdbool.h>
+#include <sys/ptrace.h>
+#include <sys/user.h>
+#include <sys/ucontext.h>
+#include <link.h>
+#include <sys/auxv.h>
+#include <dlfcn.h>
+#include <unwind.h>
+
+static void sethandler(int sig, void (*handler)(int, siginfo_t *, void *),
+		       int flags)
+{
+	struct sigaction sa;
+	memset(&sa, 0, sizeof(sa));
+	sa.sa_sigaction = handler;
+	sa.sa_flags = SA_SIGINFO | flags;
+	sigemptyset(&sa.sa_mask);
+	if (sigaction(sig, &sa, 0))
+		err(1, "sigaction");
+}
+
+#ifdef __x86_64__
+# define WIDTH "q"
+#else
+# define WIDTH "l"
+#endif
+
+static unsigned long get_eflags(void)
+{
+	unsigned long eflags;
+	asm volatile ("pushf" WIDTH "\n\tpop" WIDTH " %0" : "=rm" (eflags));
+	return eflags;
+}
+
+static void set_eflags(unsigned long eflags)
+{
+	asm volatile ("push" WIDTH " %0\n\tpopf" WIDTH
+		      : : "rm" (eflags) : "flags");
+}
+
+#define X86_EFLAGS_TF (1UL << 8)
+
+static volatile sig_atomic_t nerrs;
+static unsigned long sysinfo;
+static bool got_sysinfo = false;
+static unsigned long return_address;
+
+struct unwind_state {
+	unsigned long ip;	/* trap source */
+	int depth;		/* -1 until we hit the trap source */
+};
+
+_Unwind_Reason_Code trace_fn(struct _Unwind_Context * ctx, void *opaque)
+{
+	struct unwind_state *state = opaque;
+	unsigned long ip = _Unwind_GetIP(ctx);
+
+	if (state->depth == -1) {
+		if (ip == state->ip)
+			state->depth = 0;
+		else
+			return _URC_NO_REASON;	/* Not there yet */
+	}
+	printf("\t  0x%lx\n", ip);
+
+	if (ip == return_address) {
+		/* Here we are. */
+		unsigned long eax = _Unwind_GetGR(ctx, 0);
+		unsigned long ecx = _Unwind_GetGR(ctx, 1);
+		unsigned long edx = _Unwind_GetGR(ctx, 2);
+		unsigned long ebx = _Unwind_GetGR(ctx, 3);
+		unsigned long ebp = _Unwind_GetGR(ctx, 5);
+		unsigned long esi = _Unwind_GetGR(ctx, 6);
+		unsigned long edi = _Unwind_GetGR(ctx, 7);
+		bool ok = (eax == SYS_getpid || eax == getpid()) &&
+			ebx == 1 && ecx == 2 && edx == 3 &&
+			esi == 4 && edi == 5 && ebp == 6;
+
+		if (!ok)
+			nerrs++;
+		printf("[%s]\t  NR = %ld, args = %ld, %ld, %ld, %ld, %ld, %ld\n",
+		       (ok ? "OK" : "FAIL"),
+		       eax, ebx, ecx, edx, esi, edi, ebp);
+
+		return _URC_NORMAL_STOP;
+	} else {
+		state->depth++;
+		return _URC_NO_REASON;
+	}
+}
+
+static void sigtrap(int sig, siginfo_t *info, void *ctx_void)
+{
+	ucontext_t *ctx = (ucontext_t*)ctx_void;
+	struct unwind_state state;
+	unsigned long ip = ctx->uc_mcontext.gregs[REG_EIP];
+
+	if (!got_sysinfo && ip == sysinfo) {
+		got_sysinfo = true;
+
+		/* Find the return address. */
+		return_address = *(unsigned long *)(unsigned long)ctx->uc_mcontext.gregs[REG_ESP];
+
+		printf("\tIn vsyscall at 0x%lx, returning to 0x%lx\n",
+		       ip, return_address);
+	}
+
+	if (!got_sysinfo)
+		return;		/* Not there yet */
+
+	if (ip == return_address) {
+		ctx->uc_mcontext.gregs[REG_EFL] &= ~X86_EFLAGS_TF;
+		printf("\tVsyscall is done\n");
+		return;
+	}
+
+	printf("\tSIGTRAP at 0x%lx\n", ip);
+
+	state.ip = ip;
+	state.depth = -1;
+	_Unwind_Backtrace(trace_fn, &state);
+}
+
+int main()
+{
+	sysinfo = getauxval(AT_SYSINFO);
+	printf("\tAT_SYSINFO is 0x%lx\n", sysinfo);
+
+	Dl_info info;
+	if (!dladdr((void *)sysinfo, &info)) {
+		printf("[WARN]\tdladdr failed on AT_SYSINFO\n");
+	} else {
+		printf("[OK]\tAT_SYSINFO maps to %s, loaded at 0x%p\n",
+		       info.dli_fname, info.dli_fbase);
+	}
+
+	sethandler(SIGTRAP, sigtrap, 0);
+
+	syscall(SYS_getpid);  /* Force symbol binding without TF set. */
+	printf("[RUN]\tSet TF and check a fast syscall\n");
+	set_eflags(get_eflags() | X86_EFLAGS_TF);
+	syscall(SYS_getpid, 1, 2, 3, 4, 5, 6);
+	if (!got_sysinfo) {
+		set_eflags(get_eflags() & ~X86_EFLAGS_TF);
+
+		/*
+		 * The most likely cause of this is that you're on Debian or
+		 * a Debian-based distro, you're missing libc6-i686, and you're
+		 * affected by libc/19006 (https://sourceware.org/PR19006).
+		 */
+		printf("[WARN]\tsyscall(2) didn't enter AT_SYSINFO\n");
+	} if (get_eflags() & X86_EFLAGS_TF) {
+		printf("[FAIL]\tTF is still set\n");
+		nerrs++;
+	}
+
+	if (nerrs) {
+		printf("[FAIL]\tThere were errors\n");
+		return 1;
+	} else {
+		printf("[OK]\tAll is well\n");
+		return 0;
+	}
+}
+
+#endif	/* New enough libc */