Initial snapshot of the new BPF-enabled seccomp sandbox. This code is

sandbox/linux/seccomp_bpf/sandbox_bpf.h

Index: sandbox/linux/seccomp_bpf/sandbox_bpf.h
===================================================================
--- sandbox/linux/seccomp_bpf/sandbox_bpf.h	(revision 0)
+++ sandbox/linux/seccomp_bpf/sandbox_bpf.h	(revision 0)
@@ -0,0 +1,218 @@
+// Copyright (c) 2012 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef SANDBOX_BPF_H__
+#define SANDBOX_BPF_H__
+
+#include <algorithm>
+#include <endian.h>
+#include <errno.h>
+#include <fcntl.h>
+#include <linux/audit.h>
+#include <linux/filter.h>
+//#include <linux/seccomp.h>
+#include <linux/unistd.h>
+#include <netinet/in.h>
+#include <netinet/tcp.h>
+#include <netinet/udp.h>
+#include <sched.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/ioctl.h>
+#include <sys/ipc.h>
+#include <sys/mman.h>
+#include <sys/prctl.h>
+#include <sys/shm.h>
+#include <sys/stat.h>
+#include <sys/types.h>
+#include <sys/uio.h>
+#include <sys/wait.h>
+#include <unistd.h>
+#include <vector>
+
+#include "util.h"
+
+
+// The Seccomp2 kernel ABI is not part of older versions of glibc.
+// As we can't break compilation with these versions of the library,
+// we explicitly define all missing symbols.
+
+#ifndef PR_SET_NO_NEW_PRIVS
+#define PR_SET_NO_NEW_PRIVS          38
+#define PR_GET_NO_NEW_PRIVS          39
+#endif
+#ifndef IPC_64
+#define IPC_64                   0x0100
+#endif
+#ifndef SECCOMP_MODE_FILTER
+#define SECCOMP_MODE_DISABLED         0
+#define SECCOMP_MODE_STRICT           1
+#define SECCOMP_MODE_FILTER           2  // User user-supplied filter
+#define SECCOMP_RET_KILL    0x00000000U  // Kill the task immediately
+#define SECCOMP_RET_TRAP    0x00030000U  // Disallow and force a SIGSYS
+#define SECCOMP_RET_ERRNO   0x00050000U  // Returns an errno
+#define SECCOMP_RET_TRACE   0x7ff00000U  // Pass to a tracer or disallow
+#define SECCOMP_RET_ALLOW   0x7fff0000U  // Allow
+#define SECCOMP_RET_ACTION  0xffff0000U  // Masks for the return value
+#define SECCOMP_RET_DATA    0x0000ffffU  //   sections
+#endif
+#define SECCOMP_RET_DENY   (SECCOMP_RET_ERRNO|EPERM)
+#ifndef SYS_SECCOMP
+#define SYS_SECCOMP                   1
+#endif
+
+#if defined(__i386__)
+#define MAX_SYSCALL  512
+#define SECCOMP_ARCH AUDIT_ARCH_I386
+#define REG_RESULT   REG_EAX
+#define REG_SYSCALL  REG_EAX
+#define REG_PARM1    REG_EBX
+#define REG_PARM2    REG_ECX
+#define REG_PARM3    REG_EDX
+#define REG_PARM4    REG_ESI
+#define REG_PARM5    REG_EDI
+#define REG_PARM6    REG_EBP
+#elif defined(__x86_64__)
+#define MAX_SYSCALL  512
+#define SECCOMP_ARCH AUDIT_ARCH_X86_64
+#define REG_RESULT   REG_RAX
+#define REG_SYSCALL  REG_RAX
+#define REG_PARM1    REG_RDI
+#define REG_PARM2    REG_RSI
+#define REG_PARM3    REG_RDX
+#define REG_PARM4    REG_R10
+#define REG_PARM5    REG_R8
+#define REG_PARM6    REG_R9
+#else
+#define Unsupported target platform
+#endif
+
+struct arch_seccomp_data {
+  int nr;
+  uint32_t arch;
+  uint64_t instruction_pointer;
+  uint64_t args[6];
+};
+
+
+#define ARRAYSIZE(x) \
+  (int)(sizeof(x)/sizeof(*(x)))
+
+
+namespace playground2 {
+
+class Sandbox {
+ friend class Util;
+
+ public:
+  enum ErrorCode {
+    SB_TRAP          = -1,
+    SB_ALLOWED       = 0x0000,
+    SB_INSPECT_ARG_1 = 0x8001,
+    SB_INSPECT_ARG_2 = 0x8002,
+    SB_INSPECT_ARG_3 = 0x8004,
+    SB_INSPECT_ARG_4 = 0x8008,
+    SB_INSPECT_ARG_5 = 0x8010,
+    SB_INSPECT_ARG_6 = 0x8020
+  };
+
+  enum Operation {
+    OP_NOP, OP_EQUAL, OP_NOTEQUAL, OP_LESS,
+    OP_LESS_EQUAL, OP_GREATER, OP_GREATER_EQUAL,
+    OP_HAS_BITS, OP_DOES_NOT_HAVE_BITS
+  };
+
+  struct Constraint {
+    bool      is32bit;
+    Operation op;
+    uint32_t  value;
+    ErrorCode passed;
+    ErrorCode failed;
+  };
+
+  typedef ErrorCode (*EvaluateSyscall)(int sysno);
+  typedef int       (*EvaluateArguments)(int sysno, int arg,
+                                         Constraint *constraint);
+
+  // There are a lot of reasons why the Seccomp sandbox might not be available.
+  // This could be because the kernel does not support Seccomp mode, or it
+  // could be because another sandbox is already active.
+  // "proc_fd" should be a file descriptor for "/proc", or -1 if not
+  // provided by the caller.
+  static int supportsSeccompSandbox(int proc_fd)
+                                         asm("SupportsSeccomp2Sandbox");
+
+  // The sandbox needs to be able to access files in "/proc/self". If this
+  // directory is not accessible when "startSandbox()" gets called, the caller
+  // can provide an already opened file descriptor by calling "setProcFd()".
+  // The sandbox becomes the newer owner of this file descriptor and will
+  // eventually close it when "startSandbox()" executes.
+  static void setProcFd(int proc_fd)     asm("Seccomp2SandboxSetProcFd");
+
+  // This is the main public entry point. It finds all system calls that
+  // need rewriting, sets up the resources needed by the sandbox, and
+  // enters Seccomp mode.
+  static void startSandbox()             asm("StartSeccomp2Sandbox");
+
+  // This optional call can be used to provide a non-standard sandbox
+  // policy.
+  // The system call evaluator function is called with the system
+  // call number. It can decide to allow the system call unconditionally
+  // by returning "0"; it can deny the system call unconditionally by
+  // returning an appropriate "errno" value; or it can request inspection
+  // of system call argument(s) by returning a suitable combination of
+  // SB_INSPECT_ARG_x bits.
+  // The system argument evaluator is called (if needed) to query additional
+  // constraints for the system call arguments. In the vast majority of
+  // cases, it will set a "Constraint" that forces a new "errno" value.
+  // But for more complex filters, it is possible to return another mask
+  // of SB_INSPECT_ARG_x bits.
+  static void setSandboxPolicy(EvaluateSyscall syscallEvaluator,
+                               EvaluateArguments argumentEvaluator);
+
+ protected:
+  // Print an error message and terminate the program. Used for fatal errors.
+  static void die(const char *msg = 0) __attribute__((noreturn)) {
+    if (msg) {
+      TEMP_FAILURE_RETRY(write(2, msg, strlen(msg)));
+      TEMP_FAILURE_RETRY(write(2, "\n", 1));
+    }
+    for (;;) {
+      syscall(__NR_exit_group, 1);
+      _exit(1);
+    }
+  }
+
+  // Get a file descriptor pointing to "/proc", if currently available.
+  static int getProcFd() { return proc_fd_; }
+
+ private:
+  enum SandboxStatus {
+    STATUS_UNKNOWN, STATUS_UNSUPPORTED, STATUS_AVAILABLE, STATUS_ENABLED
+  };
+
+  static bool    isSingleThreaded(int proc_fd);
+  static bool    disableFilesystem();
+  static int     jumpTableSize(int numSyscalls, bool recursing = false);
+  static void    verifyJumpTable(struct sock_filter *filter, int numInsn,
+                                 const int *syscallList, int numSyscalls);
+  static int     jumpTable(struct sock_filter *filter, int *idx,
+                           const int *syscallList, int numSyscalls,
+                           int ret, bool sorted = false,
+                           bool recursing = false);
+  static void    installFilter();
+  static void    sigSys(int nr, siginfo_t *info, void *void_context);
+
+  static SandboxStatus                  status_;
+  static int                            proc_fd_;
+  static std::vector<std::pair<EvaluateSyscall,
+                               EvaluateArguments> > evaluators_;
+};
+
+} // namespace
+
+#endif // SANDBOX_BPF_H__
Property changes on: sandbox/linux/seccomp_bpf/sandbox_bpf.h
___________________________________________________________________
Added: svn:eol-style
   + LF