[PATCH v2 0/4] um: Add seccomp support

[PATCH v2 0/4] um: Add seccomp support

[Mickaël Salaün]

This series add seccomp support to User-mode Linux (i386 and x86_64
subarchitectures) and fix ptrace issues. This apply on v4.4-rc4 and pass all
the 48 tests from selftest/seccomp plus the UML ptsc test.

Changes since v1; addressed Richard Weinberger's comments:
* fix a new PTRACE_SETREGS bug on x86_64 [1/4]
* fix an old PTRACE_SETREGS bug when updating orig_ax on i386 [1/4]

Regards,
 Mickaël

Mickaël Salaün (4):
  um: Fix ptrace GETREGS/SETREGS bugs
  selftests/seccomp: Remove the need for HAVE_ARCH_TRACEHOOK
  um: Add full asm/syscall.h support
  um: Add seccomp support

 .../seccomp/seccomp-filter/arch-support.txt        |   2 +-
 arch/um/Kconfig.common                             |   1 +
 arch/um/Kconfig.um                                 |  16 +++
 arch/um/include/asm/syscall-generic.h              | 138 +++++++++++++++++++++
 arch/um/include/asm/thread_info.h                  |   2 +
 arch/um/kernel/skas/syscall.c                      |  30 +++--
 arch/um/os-Linux/skas/process.c                    |   7 --
 arch/x86/um/asm/syscall.h                          |   1 +
 arch/x86/um/ptrace_32.c                            |   8 +-
 tools/testing/selftests/seccomp/seccomp_bpf.c      |  27 +++-
 10 files changed, 205 insertions(+), 27 deletions(-)
 create mode 100644 arch/um/include/asm/syscall-generic.h

-- 
2.6.4

[PATCH v2 1/4] um: Fix ptrace GETREGS/SETREGS bugs

[Mickaël Salaün]

This fix two related bugs:
* PTRACE_GETREGS doesn't get the right orig_ax (syscall) value
* PTRACE_SETREGS can't set the orig_ax value (erased by initial value)

Remove the now useless and error-prone get_syscall().

Fix inconsistent behavior in the ptrace implementation for i386 when
updating orig_eax automatically update the syscall number as well. This
is now updated in handle_syscall().

Signed-off-by: Mickaël Salaün <mic@digikod.net>
Cc: Jeff Dike <jdike@addtoit.com>
Cc: Richard Weinberger <richard@nod.at>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Kees Cook <keescook@chromium.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Will Drewry <wad@chromium.org>
Cc: Thomas Meyer <thomas@m3y3r.de>
Cc: Nicolas Iooss <nicolas.iooss_linux@m4x.org>
Cc: Anton Ivanov <aivanov@brocade.com>
Cc: Meredydd Luff <meredydd@senatehouse.org>
Cc: David Drysdale <drysdale@google.com>
---
 arch/um/kernel/skas/syscall.c   | 25 ++++++++++++++-----------
 arch/um/os-Linux/skas/process.c |  7 -------
 arch/x86/um/ptrace_32.c         |  8 +++-----
 3 files changed, 17 insertions(+), 23 deletions(-)

diff --git a/arch/um/kernel/skas/syscall.c b/arch/um/kernel/skas/syscall.c
index 1683b8e..0d624e0 100644
--- a/arch/um/kernel/skas/syscall.c
+++ b/arch/um/kernel/skas/syscall.c
@@ -7,29 +7,32 @@
 #include <linux/ptrace.h>
 #include <kern_util.h>
 #include <sysdep/ptrace.h>
+#include <sysdep/ptrace_user.h>
 #include <sysdep/syscalls.h>
 #include <os.h>
 
 void handle_syscall(struct uml_pt_regs *r)
 {
 	struct pt_regs *regs = container_of(r, struct pt_regs, regs);
-	long result;
 	int syscall;
 
-	if (syscall_trace_enter(regs)) {
-		result = -ENOSYS;
+	/* Initialize the syscall number and default return value. */
+	UPT_SYSCALL_NR(r) = PT_SYSCALL_NR(r->gp);
+	PT_REGS_SET_SYSCALL_RETURN(regs, -ENOSYS);
+
+	if (syscall_trace_enter(regs))
 		goto out;
-	}
 
-	syscall = get_syscall(r);
+	/* Update the syscall number after orig_ax has potentially been updated
+	 * with ptrace.
+	 */
+	UPT_SYSCALL_NR(r) = PT_SYSCALL_NR(r->gp);
+	syscall = UPT_SYSCALL_NR(r);
 
-	if ((syscall > __NR_syscall_max) || syscall < 0)
-		result = -ENOSYS;
-	else
-		result = EXECUTE_SYSCALL(syscall, regs);
+	if (syscall >= 0 && syscall <= __NR_syscall_max)
+		PT_REGS_SET_SYSCALL_RETURN(regs,
+				EXECUTE_SYSCALL(syscall, regs));
 
 out:
-	PT_REGS_SET_SYSCALL_RETURN(regs, result);
-
 	syscall_trace_leave(regs);
 }
diff --git a/arch/um/os-Linux/skas/process.c b/arch/um/os-Linux/skas/process.c
index b856c66..23025d6 100644
--- a/arch/um/os-Linux/skas/process.c
+++ b/arch/um/os-Linux/skas/process.c
@@ -172,13 +172,6 @@ static void handle_trap(int pid, struct uml_pt_regs *regs,
 	handle_syscall(regs);
 }
 
-int get_syscall(struct uml_pt_regs *regs)
-{
-	UPT_SYSCALL_NR(regs) = PT_SYSCALL_NR(regs->gp);
-
-	return UPT_SYSCALL_NR(regs);
-}
-
 extern char __syscall_stub_start[];
 
 static int userspace_tramp(void *stack)
diff --git a/arch/x86/um/ptrace_32.c b/arch/x86/um/ptrace_32.c
index a29756f..47c78d5 100644
--- a/arch/x86/um/ptrace_32.c
+++ b/arch/x86/um/ptrace_32.c
@@ -68,6 +68,7 @@ static const int reg_offsets[] = {
 	[EFL] = HOST_EFLAGS,
 	[UESP] = HOST_SP,
 	[SS] = HOST_SS,
+	[ORIG_EAX] = HOST_ORIG_AX,
 };
 
 int putreg(struct task_struct *child, int regno, unsigned long value)
@@ -83,6 +84,7 @@ int putreg(struct task_struct *child, int regno, unsigned long value)
 	case EAX:
 	case EIP:
 	case UESP:
+	case ORIG_EAX:
 		break;
 	case FS:
 		if (value && (value & 3) != 3)
@@ -108,9 +110,6 @@ int putreg(struct task_struct *child, int regno, unsigned long value)
 		value &= FLAG_MASK;
 		child->thread.regs.regs.gp[HOST_EFLAGS] |= value;
 		return 0;
-	case ORIG_EAX:
-		child->thread.regs.regs.syscall = value;
-		return 0;
 	default :
 		panic("Bad register in putreg() : %d\n", regno);
 	}
@@ -143,8 +142,6 @@ unsigned long getreg(struct task_struct *child, int regno)
 
 	regno >>= 2;
 	switch (regno) {
-	case ORIG_EAX:
-		return child->thread.regs.regs.syscall;
 	case FS:
 	case GS:
 	case DS:
@@ -163,6 +160,7 @@ unsigned long getreg(struct task_struct *child, int regno)
 	case EDI:
 	case EBP:
 	case EFL:
+	case ORIG_EAX:
 		break;
 	default:
 		panic("Bad register in getreg() : %d\n", regno);
-- 
2.6.4

[PATCH v2 2/4] selftests/seccomp: Remove the need for HAVE_ARCH_TRACEHOOK

[Mickaël Salaün]

Some architectures do not implement PTRACE_GETREGSET nor
PTRACE_SETREGSET (required by HAVE_ARCH_TRACEHOOK) but only implement
PTRACE_GETREGS and PTRACE_SETREGS (e.g. User-mode Linux).

This improve seccomp selftest portability for architectures without
HAVE_ARCH_TRACEHOOK support by defining a new trigger HAVE_GETREGS. For
now, this is only enabled for i386 and x86_64 architectures. This is
required to be able to run this tests on User-mode Linux.

Signed-off-by: Mickaël Salaün <mic@digikod.net>
Cc: Jeff Dike <jdike@addtoit.com>
Cc: Richard Weinberger <richard@nod.at>
Cc: Kees Cook <keescook@chromium.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Will Drewry <wad@chromium.org>
Cc: Shuah Khan <shuahkh@osg.samsung.com>
Cc: Meredydd Luff <meredydd@senatehouse.org>
Cc: David Drysdale <drysdale@google.com>
---
 tools/testing/selftests/seccomp/seccomp_bpf.c | 27 ++++++++++++++++++++++++---
 1 file changed, 24 insertions(+), 3 deletions(-)

diff --git a/tools/testing/selftests/seccomp/seccomp_bpf.c b/tools/testing/selftests/seccomp/seccomp_bpf.c
index 882fe83..b9453b8 100644
--- a/tools/testing/selftests/seccomp/seccomp_bpf.c
+++ b/tools/testing/selftests/seccomp/seccomp_bpf.c
@@ -1246,11 +1246,24 @@ TEST_F(TRACE_poke, getpid_runs_normally)
 # error "Do not know how to find your architecture's registers and syscalls"
 #endif
 
+/* Use PTRACE_GETREGS and PTRACE_SETREGS when available. This is useful for
+ * architectures without HAVE_ARCH_TRACEHOOK (e.g. User-mode Linux).
+ */
+#if defined(__x86_64__) || defined(__i386__)
+#define HAVE_GETREGS
+#endif
+
 /* Architecture-specific syscall fetching routine. */
 int get_syscall(struct __test_metadata *_metadata, pid_t tracee)
 {
-	struct iovec iov;
 	ARCH_REGS regs;
+#ifdef HAVE_GETREGS
+	EXPECT_EQ(0, ptrace(PTRACE_GETREGS, tracee, 0, &regs)) {
+		TH_LOG("PTRACE_GETREGS failed");
+		return -1;
+	}
+#else
+	struct iovec iov;
 
 	iov.iov_base = &regs;
 	iov.iov_len = sizeof(regs);
@@ -1258,6 +1271,7 @@ int get_syscall(struct __test_metadata *_metadata, pid_t tracee)
 		TH_LOG("PTRACE_GETREGSET failed");
 		return -1;
 	}
+#endif
 
 	return regs.SYSCALL_NUM;
 }
@@ -1266,13 +1280,16 @@ int get_syscall(struct __test_metadata *_metadata, pid_t tracee)
 void change_syscall(struct __test_metadata *_metadata,
 		    pid_t tracee, int syscall)
 {
-	struct iovec iov;
 	int ret;
 	ARCH_REGS regs;
-
+#ifdef HAVE_GETREGS
+	ret = ptrace(PTRACE_GETREGS, tracee, 0, &regs);
+#else
+	struct iovec iov;
 	iov.iov_base = &regs;
 	iov.iov_len = sizeof(regs);
 	ret = ptrace(PTRACE_GETREGSET, tracee, NT_PRSTATUS, &iov);
+#endif
 	EXPECT_EQ(0, ret);
 
 #if defined(__x86_64__) || defined(__i386__) || defined(__powerpc__) || \
@@ -1312,9 +1329,13 @@ void change_syscall(struct __test_metadata *_metadata,
 	if (syscall == -1)
 		regs.SYSCALL_RET = 1;
 
+#ifdef HAVE_GETREGS
+	ret = ptrace(PTRACE_SETREGS, tracee, 0, &regs);
+#else
 	iov.iov_base = &regs;
 	iov.iov_len = sizeof(regs);
 	ret = ptrace(PTRACE_SETREGSET, tracee, NT_PRSTATUS, &iov);
+#endif
 	EXPECT_EQ(0, ret);
 }
 
-- 
2.6.4

[PATCH v2 3/4] um: Add full asm/syscall.h support

[Mickaël Salaün]

Add subarchitecture-independent implementation of asm-generic/syscall.h
allowing access to user system call parameters and results:
* syscall_get_nr()
* syscall_rollback()
* syscall_get_error()
* syscall_get_return_value()
* syscall_set_return_value()
* syscall_get_arguments()
* syscall_set_arguments()
* syscall_get_arch() provided by arch/x86/um/asm/syscall.h

This provides the necessary syscall helpers needed by
HAVE_ARCH_SECCOMP_FILTER plus syscall_get_error().

This is inspired from Meredydd Luff's patch
(https://gerrit.chromium.org/gerrit/21425).

Signed-off-by: Mickaël Salaün <mic@digikod.net>
Cc: Jeff Dike <jdike@addtoit.com>
Cc: Richard Weinberger <richard@nod.at>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Will Drewry <wad@chromium.org>
Cc: Meredydd Luff <meredydd@senatehouse.org>
Cc: David Drysdale <drysdale@google.com>
---
 arch/um/include/asm/syscall-generic.h | 138 ++++++++++++++++++++++++++++++++++
 arch/x86/um/asm/syscall.h             |   1 +
 2 files changed, 139 insertions(+)
 create mode 100644 arch/um/include/asm/syscall-generic.h

diff --git a/arch/um/include/asm/syscall-generic.h b/arch/um/include/asm/syscall-generic.h
new file mode 100644
index 0000000..9fb9cf8
--- /dev/null
+++ b/arch/um/include/asm/syscall-generic.h
@@ -0,0 +1,138 @@
+/*
+ * Access to user system call parameters and results
+ *
+ * See asm-generic/syscall.h for function descriptions.
+ *
+ * Copyright (C) 2015 Mickaël Salaün <mic@digikod.net>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __UM_SYSCALL_GENERIC_H
+#define __UM_SYSCALL_GENERIC_H
+
+#include <asm/ptrace.h>
+#include <linux/err.h>
+#include <linux/sched.h>
+#include <sysdep/ptrace.h>
+
+static inline int syscall_get_nr(struct task_struct *task, struct pt_regs *regs)
+{
+
+	return PT_REGS_SYSCALL_NR(regs);
+}
+
+static inline void syscall_rollback(struct task_struct *task,
+				    struct pt_regs *regs)
+{
+	/* do nothing */
+}
+
+static inline long syscall_get_error(struct task_struct *task,
+				     struct pt_regs *regs)
+{
+	const long error = regs_return_value(regs);
+
+	return IS_ERR_VALUE(error) ? error : 0;
+}
+
+static inline long syscall_get_return_value(struct task_struct *task,
+					    struct pt_regs *regs)
+{
+	return regs_return_value(regs);
+}
+
+static inline void syscall_set_return_value(struct task_struct *task,
+					    struct pt_regs *regs,
+					    int error, long val)
+{
+	PT_REGS_SET_SYSCALL_RETURN(regs, (long) error ?: val);
+}
+
+static inline void syscall_get_arguments(struct task_struct *task,
+					 struct pt_regs *regs,
+					 unsigned int i, unsigned int n,
+					 unsigned long *args)
+{
+	const struct uml_pt_regs *r = &regs->regs;
+
+	switch (i) {
+	case 0:
+		if (!n--)
+			break;
+		*args++ = UPT_SYSCALL_ARG1(r);
+	case 1:
+		if (!n--)
+			break;
+		*args++ = UPT_SYSCALL_ARG2(r);
+	case 2:
+		if (!n--)
+			break;
+		*args++ = UPT_SYSCALL_ARG3(r);
+	case 3:
+		if (!n--)
+			break;
+		*args++ = UPT_SYSCALL_ARG4(r);
+	case 4:
+		if (!n--)
+			break;
+		*args++ = UPT_SYSCALL_ARG5(r);
+	case 5:
+		if (!n--)
+			break;
+		*args++ = UPT_SYSCALL_ARG6(r);
+	case 6:
+		if (!n--)
+			break;
+	default:
+		BUG();
+		break;
+	}
+}
+
+static inline void syscall_set_arguments(struct task_struct *task,
+					 struct pt_regs *regs,
+					 unsigned int i, unsigned int n,
+					 const unsigned long *args)
+{
+	struct uml_pt_regs *r = &regs->regs;
+
+	switch (i) {
+	case 0:
+		if (!n--)
+			break;
+		UPT_SYSCALL_ARG1(r) = *args++;
+	case 1:
+		if (!n--)
+			break;
+		UPT_SYSCALL_ARG2(r) = *args++;
+	case 2:
+		if (!n--)
+			break;
+		UPT_SYSCALL_ARG3(r) = *args++;
+	case 3:
+		if (!n--)
+			break;
+		UPT_SYSCALL_ARG4(r) = *args++;
+	case 4:
+		if (!n--)
+			break;
+		UPT_SYSCALL_ARG5(r) = *args++;
+	case 5:
+		if (!n--)
+			break;
+		UPT_SYSCALL_ARG6(r) = *args++;
+	case 6:
+		if (!n--)
+			break;
+	default:
+		BUG();
+		break;
+	}
+}
+
+/* See arch/x86/um/asm/syscall.h for syscall_get_arch() definition. */
+
+#endif	/* __UM_SYSCALL_GENERIC_H */
diff --git a/arch/x86/um/asm/syscall.h b/arch/x86/um/asm/syscall.h
index 81d6562..11ab90d 100644
--- a/arch/x86/um/asm/syscall.h
+++ b/arch/x86/um/asm/syscall.h
@@ -1,6 +1,7 @@
 #ifndef __UM_ASM_SYSCALL_H
 #define __UM_ASM_SYSCALL_H
 
+#include <asm/syscall-generic.h>
 #include <uapi/linux/audit.h>
 
 typedef asmlinkage long (*sys_call_ptr_t)(unsigned long, unsigned long,
-- 
2.6.4

[PATCH v2 4/4] um: Add seccomp support

[Mickaël Salaün]

This brings SECCOMP_MODE_STRICT and SECCOMP_MODE_FILTER support through
prctl(2) and seccomp(2) to User-mode Linux for i386 and x86_64
subarchitectures.

secure_computing() is called first in handle_syscall() so that the
syscall emulation will be aborted quickly if matching a seccomp rule.

This is inspired from Meredydd Luff's patch
(https://gerrit.chromium.org/gerrit/21425).

Signed-off-by: Mickaël Salaün <mic@digikod.net>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Jeff Dike <jdike@addtoit.com>
Cc: Richard Weinberger <richard@nod.at>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Will Drewry <wad@chromium.org>
Cc: Chris Metcalf <cmetcalf@ezchip.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: James Hogan <james.hogan@imgtec.com>
Cc: Meredydd Luff <meredydd@senatehouse.org>
Cc: David Drysdale <drysdale@google.com>
---
 .../features/seccomp/seccomp-filter/arch-support.txt     |  2 +-
 arch/um/Kconfig.common                                   |  1 +
 arch/um/Kconfig.um                                       | 16 ++++++++++++++++
 arch/um/include/asm/thread_info.h                        |  2 ++
 arch/um/kernel/skas/syscall.c                            |  5 +++++
 5 files changed, 25 insertions(+), 1 deletion(-)

diff --git a/Documentation/features/seccomp/seccomp-filter/arch-support.txt b/Documentation/features/seccomp/seccomp-filter/arch-support.txt
index 76d39d6..4f66ec1 100644
--- a/Documentation/features/seccomp/seccomp-filter/arch-support.txt
+++ b/Documentation/features/seccomp/seccomp-filter/arch-support.txt
@@ -33,7 +33,7 @@
     |          sh: | TODO |
     |       sparc: | TODO |
     |        tile: |  ok  |
-    |          um: | TODO |
+    |          um: |  ok  |
     |   unicore32: | TODO |
     |         x86: |  ok  |
     |      xtensa: | TODO |
diff --git a/arch/um/Kconfig.common b/arch/um/Kconfig.common
index d195a87..cc00134 100644
--- a/arch/um/Kconfig.common
+++ b/arch/um/Kconfig.common
@@ -2,6 +2,7 @@ config UML
 	bool
 	default y
 	select HAVE_ARCH_AUDITSYSCALL
+	select HAVE_ARCH_SECCOMP_FILTER
 	select HAVE_UID16
 	select HAVE_FUTEX_CMPXCHG if FUTEX
 	select GENERIC_IRQ_SHOW
diff --git a/arch/um/Kconfig.um b/arch/um/Kconfig.um
index 28a9885..4b2ed58 100644
--- a/arch/um/Kconfig.um
+++ b/arch/um/Kconfig.um
@@ -104,3 +104,19 @@ config PGTABLE_LEVELS
 	int
 	default 3 if 3_LEVEL_PGTABLES
 	default 2
+
+config SECCOMP
+	def_bool y
+	prompt "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.
+
+	  If unsure, say Y.
diff --git a/arch/um/include/asm/thread_info.h b/arch/um/include/asm/thread_info.h
index 53968aa..053baff 100644
--- a/arch/um/include/asm/thread_info.h
+++ b/arch/um/include/asm/thread_info.h
@@ -62,11 +62,13 @@ static inline struct thread_info *current_thread_info(void)
 #define TIF_SYSCALL_AUDIT	6
 #define TIF_RESTORE_SIGMASK	7
 #define TIF_NOTIFY_RESUME	8
+#define TIF_SECCOMP		9	/* secure computing */
 
 #define _TIF_SYSCALL_TRACE	(1 << TIF_SYSCALL_TRACE)
 #define _TIF_SIGPENDING		(1 << TIF_SIGPENDING)
 #define _TIF_NEED_RESCHED	(1 << TIF_NEED_RESCHED)
 #define _TIF_MEMDIE		(1 << TIF_MEMDIE)
 #define _TIF_SYSCALL_AUDIT	(1 << TIF_SYSCALL_AUDIT)
+#define _TIF_SECCOMP		(1 << TIF_SECCOMP)
 
 #endif
diff --git a/arch/um/kernel/skas/syscall.c b/arch/um/kernel/skas/syscall.c
index 0d624e0..5abc677 100644
--- a/arch/um/kernel/skas/syscall.c
+++ b/arch/um/kernel/skas/syscall.c
@@ -5,6 +5,7 @@
 
 #include <linux/kernel.h>
 #include <linux/ptrace.h>
+#include <linux/seccomp.h>
 #include <kern_util.h>
 #include <sysdep/ptrace.h>
 #include <sysdep/ptrace_user.h>
@@ -20,6 +21,10 @@ void handle_syscall(struct uml_pt_regs *r)
 	UPT_SYSCALL_NR(r) = PT_SYSCALL_NR(r->gp);
 	PT_REGS_SET_SYSCALL_RETURN(regs, -ENOSYS);
 
+	/* Do the secure computing check first; failures should be fast. */
+	if (secure_computing() == -1)
+		return;
+
 	if (syscall_trace_enter(regs))
 		goto out;
 
-- 
2.6.4