ARM: smp: add support for per-task stack canaries

On ARM, we currently only change the value of the stack canary when
switching tasks if the kernel was built for UP. On SMP kernels, this
is impossible since the stack canary value is obtained via a global
symbol reference, which means
a) all running tasks on all CPUs must use the same value
b) we can only modify the value when no kernel stack frames are live
   on any CPU, which is effectively never.

So instead, use a GCC plugin to add a RTL pass that replaces each
reference to the address of the __stack_chk_guard symbol with an
expression that produces the address of the 'stack_canary' field
that is added to struct thread_info. This way, each task will use
its own randomized value.

Cc: Russell King <linux@armlinux.org.uk>
Cc: Kees Cook <keescook@chromium.org>
Cc: Emese Revfy <re.emese@gmail.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Laura Abbott <labbott@redhat.com>
Cc: kernel-hardening@lists.openwall.com
Acked-by: Nicolas Pitre <nico@linaro.org>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Kees Cook <keescook@chromium.org>

arch/arm/Kconfig

@@ -1810,6 +1810,21 @@ config XEN
 	help
 	  Say Y if you want to run Linux in a Virtual Machine on Xen on ARM.
 
+config STACKPROTECTOR_PER_TASK
+	bool "Use a unique stack canary value for each task"
+	depends on GCC_PLUGINS && STACKPROTECTOR && SMP && !XIP_DEFLATED_DATA
+	select GCC_PLUGIN_ARM_SSP_PER_TASK
+	default y
+	help
+	  Due to the fact that GCC uses an ordinary symbol reference from
+	  which to load the value of the stack canary, this value can only
+	  change at reboot time on SMP systems, and all tasks running in the
+	  kernel's address space are forced to use the same canary value for
+	  the entire duration that the system is up.
+
+	  Enable this option to switch to a different method that uses a
+	  different canary value for each task.
+
 endmenu
 
 menu "Boot options"

arch/arm/Makefile

@@ -303,6 +303,18 @@ else
 KBUILD_IMAGE := $(boot)/zImage
 endif
 
+ifeq ($(CONFIG_STACKPROTECTOR_PER_TASK),y)
+prepare: stack_protector_prepare
+stack_protector_prepare: prepare0
+	$(eval KBUILD_CFLAGS += \
+		-fplugin-arg-arm_ssp_per_task_plugin-tso=$(shell	\
+			awk '{if ($$2 == "THREAD_SZ_ORDER") print $$3;}'\
+				include/generated/asm-offsets.h)	\
+		-fplugin-arg-arm_ssp_per_task_plugin-offset=$(shell	\
+			awk '{if ($$2 == "TI_STACK_CANARY") print $$3;}'\
+				include/generated/asm-offsets.h))
+endif
+
 all:	$(notdir $(KBUILD_IMAGE))
 
 

arch/arm/boot/compressed/Makefile

@@ -101,6 +101,7 @@ clean-files += piggy_data lib1funcs.S ashldi3.S bswapsdi2.S \
 		$(libfdt) $(libfdt_hdrs) hyp-stub.S
 
 KBUILD_CFLAGS += -DDISABLE_BRANCH_PROFILING
+KBUILD_CFLAGS += $(DISABLE_ARM_SSP_PER_TASK_PLUGIN)
 
 ifeq ($(CONFIG_FUNCTION_TRACER),y)
 ORIG_CFLAGS := $(KBUILD_CFLAGS)

arch/arm/include/asm/stackprotector.h

@@ -6,8 +6,10 @@
  * the stack frame and verifying that it hasn't been overwritten when
  * returning from the function.  The pattern is called stack canary
  * and gcc expects it to be defined by a global variable called
- * "__stack_chk_guard" on ARM.  This unfortunately means that on SMP
- * we cannot have a different canary value per task.
+ * "__stack_chk_guard" on ARM.  This prevents SMP systems from using a
+ * different value for each task unless we enable a GCC plugin that
+ * replaces these symbol references with references to each task's own
+ * value.
  */
 
 #ifndef _ASM_STACKPROTECTOR_H
@@ -16,6 +18,8 @@
 #include <linux/random.h>
 #include <linux/version.h>
 
+#include <asm/thread_info.h>
+
 extern unsigned long __stack_chk_guard;
 
 /*
@@ -33,7 +37,11 @@ static __always_inline void boot_init_stack_canary(void)
 	canary ^= LINUX_VERSION_CODE;
 
 	current->stack_canary = canary;
+#ifndef CONFIG_STACKPROTECTOR_PER_TASK
 	__stack_chk_guard = current->stack_canary;
+#else
+	current_thread_info()->stack_canary = current->stack_canary;
+#endif
 }
 
 #endif	/* _ASM_STACKPROTECTOR_H */

arch/arm/include/asm/thread_info.h

@@ -53,6 +53,9 @@ struct thread_info {
 	struct task_struct	*task;		/* main task structure */
 	__u32			cpu;		/* cpu */
 	__u32			cpu_domain;	/* cpu domain */
+#ifdef CONFIG_STACKPROTECTOR_PER_TASK
+	unsigned long		stack_canary;
+#endif
 	struct cpu_context_save	cpu_context;	/* cpu context */
 	__u32			syscall;	/* syscall number */
 	__u8			used_cp[16];	/* thread used copro */

arch/arm/kernel/asm-offsets.c

@@ -79,6 +79,10 @@ int main(void)
 #ifdef CONFIG_CRUNCH
   DEFINE(TI_CRUNCH_STATE,	offsetof(struct thread_info, crunchstate));
 #endif
+#ifdef CONFIG_STACKPROTECTOR_PER_TASK
+  DEFINE(TI_STACK_CANARY,	offsetof(struct thread_info, stack_canary));
+#endif
+  DEFINE(THREAD_SZ_ORDER,	THREAD_SIZE_ORDER);
   BLANK();
   DEFINE(S_R0,			offsetof(struct pt_regs, ARM_r0));
   DEFINE(S_R1,			offsetof(struct pt_regs, ARM_r1));

arch/arm/kernel/process.c

@@ -39,7 +39,7 @@
 #include <asm/tls.h>
 #include <asm/vdso.h>
 
-#ifdef CONFIG_STACKPROTECTOR
+#if defined(CONFIG_STACKPROTECTOR) && !defined(CONFIG_STACKPROTECTOR_PER_TASK)
 #include <linux/stackprotector.h>
 unsigned long __stack_chk_guard __read_mostly;
 EXPORT_SYMBOL(__stack_chk_guard);
@@ -267,6 +267,10 @@ copy_thread(unsigned long clone_flags, unsigned long stack_start,
 
 	thread_notify(THREAD_NOTIFY_COPY, thread);
 
+#ifdef CONFIG_STACKPROTECTOR_PER_TASK
+	thread->stack_canary = p->stack_canary;
+#endif
+
 	return 0;
 }
 

scripts/Makefile.gcc-plugins

@@ -36,6 +36,12 @@ ifdef CONFIG_GCC_PLUGIN_STACKLEAK
 endif
 export DISABLE_STACKLEAK_PLUGIN
 
+gcc-plugin-$(CONFIG_GCC_PLUGIN_ARM_SSP_PER_TASK) += arm_ssp_per_task_plugin.so
+ifdef CONFIG_GCC_PLUGIN_ARM_SSP_PER_TASK
+    DISABLE_ARM_SSP_PER_TASK_PLUGIN += -fplugin-arg-arm_ssp_per_task_plugin-disable
+endif
+export DISABLE_ARM_SSP_PER_TASK_PLUGIN
+
 # All the plugin CFLAGS are collected here in case a build target needs to
 # filter them out of the KBUILD_CFLAGS.
 GCC_PLUGINS_CFLAGS := $(strip $(addprefix -fplugin=$(objtree)/scripts/gcc-plugins/, $(gcc-plugin-y)) $(gcc-plugin-cflags-y))

scripts/gcc-plugins/Kconfig

@@ -190,4 +190,8 @@ config STACKLEAK_RUNTIME_DISABLE
 	  runtime to control kernel stack erasing for kernels built with
 	  CONFIG_GCC_PLUGIN_STACKLEAK.
 
+config GCC_PLUGIN_ARM_SSP_PER_TASK
+	bool
+	depends on GCC_PLUGINS && ARM
+
 endif

scripts/gcc-plugins/arm_ssp_per_task_plugin.c

+// SPDX-License-Identifier: GPL-2.0
+
+#include "gcc-common.h"
+
+__visible int plugin_is_GPL_compatible;
+
+static unsigned int sp_mask, canary_offset;
+
+static unsigned int arm_pertask_ssp_rtl_execute(void)
+{
+	rtx_insn *insn;
+
+	for (insn = get_insns(); insn; insn = NEXT_INSN(insn)) {
+		const char *sym;
+		rtx body;
+		rtx masked_sp;
+
+		/*
+		 * Find a SET insn involving a SYMBOL_REF to __stack_chk_guard
+		 */
+		if (!INSN_P(insn))
+			continue;
+		body = PATTERN(insn);
+		if (GET_CODE(body) != SET ||
+		    GET_CODE(SET_SRC(body)) != SYMBOL_REF)
+			continue;
+		sym = XSTR(SET_SRC(body), 0);
+		if (strcmp(sym, "__stack_chk_guard"))
+			continue;
+
+		/*
+		 * Replace the source of the SET insn with an expression that
+		 * produces the address of the copy of the stack canary value
+		 * stored in struct thread_info
+		 */
+		masked_sp = gen_reg_rtx(Pmode);
+
+		emit_insn_before(gen_rtx_SET(masked_sp,
+					     gen_rtx_AND(Pmode,
+							 stack_pointer_rtx,
+							 GEN_INT(sp_mask))),
+				 insn);
+
+		SET_SRC(body) = gen_rtx_PLUS(Pmode, masked_sp,
+					     GEN_INT(canary_offset));
+	}
+	return 0;
+}
+
+#define PASS_NAME arm_pertask_ssp_rtl
+
+#define NO_GATE
+#include "gcc-generate-rtl-pass.h"
+
+__visible int plugin_init(struct plugin_name_args *plugin_info,
+			  struct plugin_gcc_version *version)
+{
+	const char * const plugin_name = plugin_info->base_name;
+	const int argc = plugin_info->argc;
+	const struct plugin_argument *argv = plugin_info->argv;
+	int tso = 0;
+	int i;
+
+	if (!plugin_default_version_check(version, &gcc_version)) {
+		error(G_("incompatible gcc/plugin versions"));
+		return 1;
+	}
+
+	for (i = 0; i < argc; ++i) {
+		if (!strcmp(argv[i].key, "disable"))
+			return 0;
+
+		/* all remaining options require a value */
+		if (!argv[i].value) {
+			error(G_("no value supplied for option '-fplugin-arg-%s-%s'"),
+			      plugin_name, argv[i].key);
+			return 1;
+		}
+
+		if (!strcmp(argv[i].key, "tso")) {
+			tso = atoi(argv[i].value);
+			continue;
+		}
+
+		if (!strcmp(argv[i].key, "offset")) {
+			canary_offset = atoi(argv[i].value);
+			continue;
+		}
+		error(G_("unknown option '-fplugin-arg-%s-%s'"),
+		      plugin_name, argv[i].key);
+		return 1;
+	}
+
+	/* create the mask that produces the base of the stack */
+	sp_mask = ~((1U << (12 + tso)) - 1);
+
+	PASS_INFO(arm_pertask_ssp_rtl, "expand", 1, PASS_POS_INSERT_AFTER);
+
+	register_callback(plugin_info->base_name, PLUGIN_PASS_MANAGER_SETUP,
+			  NULL, &arm_pertask_ssp_rtl_pass_info);
+
+	return 0;
+}