riscv: add support for SECCOMP and SECCOMP_FILTER

This patch was extensively tested on Fedora/RISCV (applied by default on
top of 5.2-rc7 kernel for <2 months). The patch was also tested with 5.3-rc
on QEMU and SiFive Unleashed board.

libseccomp (userspace) was rebased:
https://github.com/seccomp/libseccomp/pull/134

Fully passes libseccomp regression testing (simulation and live).

There is one failing kernel selftest: global.user_notification_signal

v1 -> v2:
  - return immediately if secure_computing(NULL) returns -1
  - fixed whitespace issues
  - add missing seccomp.h
  - remove patch #2 (solved now)
  - add riscv to seccomp kernel selftest

Signed-off-by: David Abdurachmanov <david.abdurachmanov@sifive.com>
Cc: keescook@chromium.org
Cc: me@carlosedp.com
Tested-by: Carlos de Paula <me@carlosedp.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/linux-riscv/CAEn-LTp=ss0Dfv6J00=rCAy+N78U2AmhqJNjfqjr2FDpPYjxEQ@mail.gmail.com/
Link: https://lore.kernel.org/linux-riscv/CAJr-aD=UnCN9E_mdVJ2H5nt=6juRSWikZnA5HxDLQxXLbsRz-w@mail.gmail.com/
[paul.walmsley@sifive.com: cleaned up Cc: lines; fixed spelling and
 checkpatch issues; updated to apply]
Signed-off-by: Paul Walmsley <paul.walmsley@sifive.com>

arch/riscv/Kconfig

@@ -31,6 +31,7 @@ config RISCV
 	select GENERIC_SMP_IDLE_THREAD
 	select GENERIC_ATOMIC64 if !64BIT
 	select HAVE_ARCH_AUDITSYSCALL
+	select HAVE_ARCH_SECCOMP_FILTER
 	select HAVE_ASM_MODVERSIONS
 	select HAVE_MEMBLOCK_NODE_MAP
 	select HAVE_DMA_CONTIGUOUS
@@ -272,6 +273,19 @@ menu "Kernel features"
 
 source "kernel/Kconfig.hz"
 
+config SECCOMP
+	bool "Enable seccomp to safely compute untrusted bytecode"
+	help
+	  This kernel feature is useful for number crunching applications
+	  that may need to compute untrusted bytecode during their
+	  execution. By using pipes or other transports made available to
+	  the process as file descriptors supporting the read/write
+	  syscalls, it's possible to isolate those applications in
+	  their own address space using seccomp. Once seccomp is
+	  enabled via prctl(PR_SET_SECCOMP), it cannot be disabled
+	  and the task is only allowed to execute a few safe syscalls
+	  defined by each seccomp mode.
+
 endmenu
 
 menu "Boot options"

arch/riscv/include/asm/seccomp.h

@@ -0,0 +1,10 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef _ASM_SECCOMP_H
+#define _ASM_SECCOMP_H
+
+#include <asm/unistd.h>
+
+#include <asm-generic/seccomp.h>
+
+#endif /* _ASM_SECCOMP_H */

arch/riscv/include/asm/thread_info.h

@@ -75,6 +75,7 @@ struct thread_info {
 #define TIF_MEMDIE		5	/* is terminating due to OOM killer */
 #define TIF_SYSCALL_TRACEPOINT  6       /* syscall tracepoint instrumentation */
 #define TIF_SYSCALL_AUDIT	7	/* syscall auditing */
+#define TIF_SECCOMP		8	/* syscall secure computing */
 
 #define _TIF_SYSCALL_TRACE	(1 << TIF_SYSCALL_TRACE)
 #define _TIF_NOTIFY_RESUME	(1 << TIF_NOTIFY_RESUME)
@@ -82,11 +83,13 @@ struct thread_info {
 #define _TIF_NEED_RESCHED	(1 << TIF_NEED_RESCHED)
 #define _TIF_SYSCALL_TRACEPOINT	(1 << TIF_SYSCALL_TRACEPOINT)
 #define _TIF_SYSCALL_AUDIT	(1 << TIF_SYSCALL_AUDIT)
+#define _TIF_SECCOMP		(1 << TIF_SECCOMP)
 
 #define _TIF_WORK_MASK \
 	(_TIF_NOTIFY_RESUME | _TIF_SIGPENDING | _TIF_NEED_RESCHED)
 
 #define _TIF_SYSCALL_WORK \
-	(_TIF_SYSCALL_TRACE | _TIF_SYSCALL_TRACEPOINT | _TIF_SYSCALL_AUDIT)
+	(_TIF_SYSCALL_TRACE | _TIF_SYSCALL_TRACEPOINT | _TIF_SYSCALL_AUDIT | \
+	 _TIF_SECCOMP)
 
 #endif /* _ASM_RISCV_THREAD_INFO_H */

arch/riscv/kernel/entry.S

@@ -226,8 +226,25 @@ check_syscall_nr:
 	/* Check to make sure we don't jump to a bogus syscall number. */
 	li t0, __NR_syscalls
 	la s0, sys_ni_syscall
-	/* Syscall number held in a7 */
-	bgeu a7, t0, 1f
+	/*
+	 * The tracer can change syscall number to valid/invalid value.
+	 * We use syscall_set_nr helper in syscall_trace_enter thus we
+	 * cannot trust the current value in a7 and have to reload from
+	 * the current task pt_regs.
+	 */
+	REG_L a7, PT_A7(sp)
+	/*
+	 * Syscall number held in a7.
+	 * If syscall number is above allowed value, redirect to ni_syscall.
+	 */
+	bge a7, t0, 1f
+	/*
+	 * Check if syscall is rejected by tracer or seccomp, i.e., a7 == -1.
+	 * If yes, we pretend it was executed.
+	 */
+	li t1, -1
+	beq a7, t1, ret_from_syscall_rejected
+	/* Call syscall */
 	la s0, sys_call_table
 	slli t0, a7, RISCV_LGPTR
 	add s0, s0, t0
@@ -238,6 +255,12 @@ check_syscall_nr:
 ret_from_syscall:
 	/* Set user a0 to kernel a0 */
 	REG_S a0, PT_A0(sp)
+	/*
+	 * We didn't execute the actual syscall.
+	 * Seccomp already set return value for the current task pt_regs.
+	 * (If it was configured with SECCOMP_RET_ERRNO/TRACE)
+	 */
+ret_from_syscall_rejected:
 	/* Trace syscalls, but only if requested by the user. */
 	REG_L t0, TASK_TI_FLAGS(tp)
 	andi t0, t0, _TIF_SYSCALL_WORK

arch/riscv/kernel/ptrace.c

@@ -154,6 +154,16 @@ __visible void do_syscall_trace_enter(struct pt_regs *regs)
 		if (tracehook_report_syscall_entry(regs))
 			syscall_set_nr(current, regs, -1);
 
+	/*
+	 * Do the secure computing after ptrace; failures should be fast.
+	 * If this fails we might have return value in a0 from seccomp
+	 * (via SECCOMP_RET_ERRNO/TRACE).
+	 */
+	if (secure_computing(NULL) == -1) {
+		syscall_set_nr(current, regs, -1);
+		return;
+	}
+
 #ifdef CONFIG_HAVE_SYSCALL_TRACEPOINTS
 	if (test_thread_flag(TIF_SYSCALL_TRACEPOINT))
 		trace_sys_enter(regs, syscall_get_nr(current, regs));

tools/testing/selftests/seccomp/seccomp_bpf.c

@@ -112,6 +112,8 @@ struct seccomp_data {
 #  define __NR_seccomp 383
 # elif defined(__aarch64__)
 #  define __NR_seccomp 277
+# elif defined(__riscv)
+#  define __NR_seccomp 277
 # elif defined(__hppa__)
 #  define __NR_seccomp 338
 # elif defined(__powerpc__)
@@ -1587,6 +1589,10 @@ TEST_F(TRACE_poke, getpid_runs_normally)
 # define ARCH_REGS	struct user_pt_regs
 # define SYSCALL_NUM	regs[8]
 # define SYSCALL_RET	regs[0]
+#elif defined(__riscv) && __riscv_xlen == 64
+# define ARCH_REGS	struct user_regs_struct
+# define SYSCALL_NUM	a7
+# define SYSCALL_RET	a0
 #elif defined(__hppa__)
 # define ARCH_REGS	struct user_regs_struct
 # define SYSCALL_NUM	gr[20]
@@ -1676,7 +1682,7 @@ void change_syscall(struct __test_metadata *_metadata,
 	EXPECT_EQ(0, ret) {}
 
 #if defined(__x86_64__) || defined(__i386__) || defined(__powerpc__) || \
-    defined(__s390__) || defined(__hppa__)
+	defined(__s390__) || defined(__hppa__) || defined(__riscv)
 	{
 		regs.SYSCALL_NUM = syscall;
 	}