riscv: Implement Shadow Call Stack

Implement CONFIG_SHADOW_CALL_STACK for RISC-V. When enabled, the
compiler injects instructions to all non-leaf C functions to
store the return address to the shadow stack and unconditionally
load it again before returning, which makes it harder to corrupt
the return address through a stack overflow, for example.

The active shadow call stack pointer is stored in the gp
register, which makes SCS incompatible with gp relaxation. Use
--no-relax-gp to ensure gp relaxation is disabled and disable
global pointer loading.  Add SCS pointers to struct thread_info,
implement SCS initialization, and task switching

Signed-off-by: Sami Tolvanen <samitolvanen@google.com>
Tested-by: Nathan Chancellor <nathan@kernel.org>
Link: https://lore.kernel.org/r/20230927224757.1154247-12-samitolvanen@google.com
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

arch/riscv/Kconfig

@@ -48,6 +48,7 @@ config RISCV
 	select ARCH_SUPPORTS_HUGETLBFS if MMU
 	select ARCH_SUPPORTS_PAGE_TABLE_CHECK if MMU
 	select ARCH_SUPPORTS_PER_VMA_LOCK if MMU
+	select ARCH_SUPPORTS_SHADOW_CALL_STACK if HAVE_SHADOW_CALL_STACK
 	select ARCH_USE_MEMTEST
 	select ARCH_USE_QUEUED_RWLOCKS
 	select ARCH_USES_CFI_TRAPS if CFI_CLANG
@@ -174,6 +175,11 @@ config GCC_SUPPORTS_DYNAMIC_FTRACE
 	def_bool CC_IS_GCC
 	depends on $(cc-option,-fpatchable-function-entry=8)
 
+config HAVE_SHADOW_CALL_STACK
+	def_bool $(cc-option,-fsanitize=shadow-call-stack)
+	# https://github.com/riscv-non-isa/riscv-elf-psabi-doc/commit/a484e843e6eeb51f0cb7b8819e50da6d2444d769
+	depends on $(ld-option,--no-relax-gp)
+
 config ARCH_MMAP_RND_BITS_MIN
 	default 18 if 64BIT
 	default 8

arch/riscv/Makefile

@@ -55,6 +55,10 @@ endif
 endif
 endif
 
+ifeq ($(CONFIG_SHADOW_CALL_STACK),y)
+	KBUILD_LDFLAGS += --no-relax-gp
+endif
+
 # ISA string setting
 riscv-march-$(CONFIG_ARCH_RV32I)	:= rv32ima
 riscv-march-$(CONFIG_ARCH_RV64I)	:= rv64ima

arch/riscv/include/asm/asm.h

@@ -109,6 +109,11 @@
 	REG_L \dst, 0(\dst)
 .endm
 
+#ifdef CONFIG_SHADOW_CALL_STACK
+/* gp is used as the shadow call stack pointer instead */
+.macro load_global_pointer
+.endm
+#else
 /* load __global_pointer to gp */
 .macro load_global_pointer
 .option push
@@ -116,6 +121,7 @@
 	la gp, __global_pointer$
 .option pop
 .endm
+#endif /* CONFIG_SHADOW_CALL_STACK */
 
 	/* save all GPs except x1 ~ x5 */
 	.macro save_from_x6_to_x31

arch/riscv/include/asm/scs.h

@@ -0,0 +1,47 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _ASM_SCS_H
+#define _ASM_SCS_H
+
+#ifdef __ASSEMBLY__
+#include <asm/asm-offsets.h>
+
+#ifdef CONFIG_SHADOW_CALL_STACK
+
+/* Load init_shadow_call_stack to gp. */
+.macro scs_load_init_stack
+	la	gp, init_shadow_call_stack
+	XIP_FIXUP_OFFSET gp
+.endm
+
+/* Load task_scs_sp(current) to gp. */
+.macro scs_load_current
+	REG_L	gp, TASK_TI_SCS_SP(tp)
+.endm
+
+/* Load task_scs_sp(current) to gp, but only if tp has changed. */
+.macro scs_load_current_if_task_changed prev
+	beq	\prev, tp, _skip_scs
+	scs_load_current
+_skip_scs:
+.endm
+
+/* Save gp to task_scs_sp(current). */
+.macro scs_save_current
+	REG_S	gp, TASK_TI_SCS_SP(tp)
+.endm
+
+#else /* CONFIG_SHADOW_CALL_STACK */
+
+.macro scs_load_init_stack
+.endm
+.macro scs_load_current
+.endm
+.macro scs_load_current_if_task_changed prev
+.endm
+.macro scs_save_current
+.endm
+
+#endif /* CONFIG_SHADOW_CALL_STACK */
+#endif /* __ASSEMBLY__ */
+
+#endif /* _ASM_SCS_H */

arch/riscv/include/asm/thread_info.h

@@ -57,8 +57,20 @@ struct thread_info {
 	long			user_sp;	/* User stack pointer */
 	int			cpu;
 	unsigned long		syscall_work;	/* SYSCALL_WORK_ flags */
+#ifdef CONFIG_SHADOW_CALL_STACK
+	void			*scs_base;
+	void			*scs_sp;
+#endif
 };
 
+#ifdef CONFIG_SHADOW_CALL_STACK
+#define INIT_SCS							\
+	.scs_base	= init_shadow_call_stack,			\
+	.scs_sp		= init_shadow_call_stack,
+#else
+#define INIT_SCS
+#endif
+
 /*
  * macros/functions for gaining access to the thread information structure
  *
@@ -68,6 +80,7 @@ struct thread_info {
 {						\
 	.flags		= 0,			\
 	.preempt_count	= INIT_PREEMPT_COUNT,	\
+	INIT_SCS				\
 }
 
 void arch_release_task_struct(struct task_struct *tsk);

arch/riscv/kernel/asm-offsets.c

@@ -39,6 +39,9 @@ void asm_offsets(void)
 	OFFSET(TASK_TI_PREEMPT_COUNT, task_struct, thread_info.preempt_count);
 	OFFSET(TASK_TI_KERNEL_SP, task_struct, thread_info.kernel_sp);
 	OFFSET(TASK_TI_USER_SP, task_struct, thread_info.user_sp);
+#ifdef CONFIG_SHADOW_CALL_STACK
+	OFFSET(TASK_TI_SCS_SP, task_struct, thread_info.scs_sp);
+#endif
 
 	OFFSET(TASK_TI_CPU_NUM, task_struct, thread_info.cpu);
 	OFFSET(TASK_THREAD_F0,  task_struct, thread.fstate.f[0]);

arch/riscv/kernel/entry.S

@@ -9,6 +9,7 @@
 
 #include <asm/asm.h>
 #include <asm/csr.h>
+#include <asm/scs.h>
 #include <asm/unistd.h>
 #include <asm/page.h>
 #include <asm/thread_info.h>
@@ -77,6 +78,9 @@ _save_context:
 	/* Load the global pointer */
 	load_global_pointer
 
+	/* Load the kernel shadow call stack pointer if coming from userspace */
+	scs_load_current_if_task_changed s5
+
 	move a0, sp /* pt_regs */
 	la ra, ret_from_exception
 
@@ -123,6 +127,9 @@ SYM_CODE_START_NOALIGN(ret_from_exception)
 	addi s0, sp, PT_SIZE_ON_STACK
 	REG_S s0, TASK_TI_KERNEL_SP(tp)
 
+	/* Save the kernel shadow call stack pointer */
+	scs_save_current
+
 	/*
 	 * Save TP into the scratch register , so we can find the kernel data
 	 * structures again.
@@ -275,6 +282,8 @@ SYM_FUNC_START(__switch_to)
 	REG_S s9,  TASK_THREAD_S9_RA(a3)
 	REG_S s10, TASK_THREAD_S10_RA(a3)
 	REG_S s11, TASK_THREAD_S11_RA(a3)
+	/* Save the kernel shadow call stack pointer */
+	scs_save_current
 	/* Restore context from next->thread */
 	REG_L ra,  TASK_THREAD_RA_RA(a4)
 	REG_L sp,  TASK_THREAD_SP_RA(a4)
@@ -292,6 +301,8 @@ SYM_FUNC_START(__switch_to)
 	REG_L s11, TASK_THREAD_S11_RA(a4)
 	/* The offset of thread_info in task_struct is zero. */
 	move tp, a1
+	/* Switch to the next shadow call stack */
+	scs_load_current
 	ret
 SYM_FUNC_END(__switch_to)
 

arch/riscv/kernel/head.S

@@ -14,6 +14,7 @@
 #include <asm/cpu_ops_sbi.h>
 #include <asm/hwcap.h>
 #include <asm/image.h>
+#include <asm/scs.h>
 #include <asm/xip_fixup.h>
 #include "efi-header.S"
 
@@ -153,6 +154,7 @@ secondary_start_sbi:
 	XIP_FIXUP_OFFSET a3
 	add a3, a3, a1
 	REG_L sp, (a3)
+	scs_load_current
 
 .Lsecondary_start_common:
 
@@ -289,6 +291,7 @@ clear_bss_done:
 	la sp, init_thread_union + THREAD_SIZE
 	XIP_FIXUP_OFFSET sp
 	addi sp, sp, -PT_SIZE_ON_STACK
+	scs_load_init_stack
 #ifdef CONFIG_BUILTIN_DTB
 	la a0, __dtb_start
 	XIP_FIXUP_OFFSET a0
@@ -307,6 +310,7 @@ clear_bss_done:
 	la tp, init_task
 	la sp, init_thread_union + THREAD_SIZE
 	addi sp, sp, -PT_SIZE_ON_STACK
+	scs_load_current
 
 #ifdef CONFIG_KASAN
 	call kasan_early_init

arch/riscv/kernel/vdso/Makefile

@@ -36,7 +36,7 @@ CPPFLAGS_vdso.lds += -DHAS_VGETTIMEOFDAY
 endif
 
 # Disable -pg to prevent insert call site
-CFLAGS_REMOVE_vgettimeofday.o = $(CC_FLAGS_FTRACE)
+CFLAGS_REMOVE_vgettimeofday.o = $(CC_FLAGS_FTRACE) $(CC_FLAGS_SCS)
 
 # Disable profiling and instrumentation for VDSO code
 GCOV_PROFILE := n

arch/riscv/purgatory/Makefile

@@ -81,6 +81,10 @@ ifdef CONFIG_CFI_CLANG
 PURGATORY_CFLAGS_REMOVE		+= $(CC_FLAGS_CFI)
 endif
 
+ifdef CONFIG_SHADOW_CALL_STACK
+PURGATORY_CFLAGS_REMOVE		+= $(CC_FLAGS_SCS)
+endif
+
 CFLAGS_REMOVE_purgatory.o	+= $(PURGATORY_CFLAGS_REMOVE)
 CFLAGS_purgatory.o		+= $(PURGATORY_CFLAGS)