1 /*- 2 * SPDX-License-Identifier: BSD-2-Clause 3 * 4 * Copyright (c) 1993, David Greenman 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 19 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 20 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 21 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 22 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 23 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 26 * SUCH DAMAGE. 27 */ 28 29 #include <sys/cdefs.h> 30 #include <sys/param.h> 31 #include <sys/exec.h> 32 #include <sys/imgact.h> 33 #include <sys/imgact_aout.h> 34 #include <sys/kernel.h> 35 #include <sys/limits.h> 36 #include <sys/lock.h> 37 #include <sys/malloc.h> 38 #include <sys/mutex.h> 39 #include <sys/proc.h> 40 #include <sys/racct.h> 41 #include <sys/resourcevar.h> 42 #include <sys/signalvar.h> 43 #include <sys/syscall.h> 44 #include <sys/sysent.h> 45 #include <sys/systm.h> 46 #include <sys/vnode.h> 47 48 #include <machine/frame.h> 49 #include <machine/md_var.h> 50 51 #include <vm/vm.h> 52 #include <vm/pmap.h> 53 #include <vm/vm_map.h> 54 #include <vm/vm_object.h> 55 #include <vm/vm_param.h> 56 57 #ifdef __amd64__ 58 #include <compat/freebsd32/freebsd32_signal.h> 59 #include <compat/freebsd32/freebsd32_util.h> 60 #include <compat/freebsd32/freebsd32_proto.h> 61 #include <compat/freebsd32/freebsd32_syscall.h> 62 #include <compat/ia32/ia32_signal.h> 63 #endif 64 65 static int exec_aout_imgact(struct image_params *imgp); 66 static int aout_fixup(uintptr_t *stack_base, struct image_params *imgp); 67 68 #define AOUT32_USRSTACK 0xbfc00000 69 70 #if defined(__i386__) 71 72 #define AOUT32_PS_STRINGS (AOUT32_USRSTACK - sizeof(struct ps_strings)) 73 74 struct sysentvec aout_sysvec = { 75 .sv_size = SYS_MAXSYSCALL, 76 .sv_table = sysent, 77 .sv_fixup = aout_fixup, 78 .sv_sendsig = sendsig, 79 .sv_sigcode = sigcode, 80 .sv_szsigcode = &szsigcode, 81 .sv_name = "FreeBSD a.out", 82 .sv_coredump = NULL, 83 .sv_minsigstksz = MINSIGSTKSZ, 84 .sv_minuser = VM_MIN_ADDRESS, 85 .sv_maxuser = AOUT32_USRSTACK, 86 .sv_usrstack = AOUT32_USRSTACK, 87 .sv_psstrings = AOUT32_PS_STRINGS, 88 .sv_psstringssz = sizeof(struct ps_strings), 89 .sv_stackprot = VM_PROT_ALL, 90 .sv_copyout_strings = exec_copyout_strings, 91 .sv_setregs = exec_setregs, 92 .sv_fixlimit = NULL, 93 .sv_maxssiz = NULL, 94 .sv_flags = SV_ABI_FREEBSD | SV_AOUT | SV_IA32 | SV_ILP32, 95 .sv_set_syscall_retval = cpu_set_syscall_retval, 96 .sv_fetch_syscall_args = cpu_fetch_syscall_args, 97 .sv_syscallnames = syscallnames, 98 .sv_schedtail = NULL, 99 .sv_thread_detach = NULL, 100 .sv_trap = NULL, 101 .sv_onexec_old = exec_onexec_old, 102 .sv_onexit = exit_onexit, 103 .sv_set_fork_retval = x86_set_fork_retval, 104 }; 105 106 #elif defined(__amd64__) 107 108 #include "vdso_ia32_offsets.h" 109 110 extern const char _binary_elf_vdso32_so_1_start[]; 111 extern const char _binary_elf_vdso32_so_1_end[]; 112 extern char _binary_elf_vdso32_so_1_size; 113 114 #define AOUT32_PS_STRINGS \ 115 (AOUT32_USRSTACK - sizeof(struct freebsd32_ps_strings)) 116 #define AOUT32_MINUSER FREEBSD32_MINUSER 117 118 extern const char *freebsd32_syscallnames[]; 119 extern u_long ia32_maxssiz; 120 121 static int aout_szsigcode; 122 123 struct sysentvec aout_sysvec = { 124 .sv_size = FREEBSD32_SYS_MAXSYSCALL, 125 .sv_table = freebsd32_sysent, 126 .sv_fixup = aout_fixup, 127 .sv_sendsig = ia32_sendsig, 128 .sv_sigcode = _binary_elf_vdso32_so_1_start, 129 .sv_szsigcode = &aout_szsigcode, 130 .sv_name = "FreeBSD a.out", 131 .sv_coredump = NULL, 132 .sv_minsigstksz = MINSIGSTKSZ, 133 .sv_minuser = AOUT32_MINUSER, 134 .sv_maxuser = AOUT32_USRSTACK, 135 .sv_usrstack = AOUT32_USRSTACK, 136 .sv_psstrings = AOUT32_PS_STRINGS, 137 .sv_psstringssz = sizeof(struct freebsd32_ps_strings), 138 .sv_stackprot = VM_PROT_ALL, 139 .sv_copyout_strings = freebsd32_copyout_strings, 140 .sv_setregs = ia32_setregs, 141 .sv_fixlimit = ia32_fixlimit, 142 .sv_maxssiz = &ia32_maxssiz, 143 .sv_flags = SV_ABI_FREEBSD | SV_AOUT | SV_IA32 | SV_ILP32, 144 .sv_set_syscall_retval = ia32_set_syscall_retval, 145 .sv_fetch_syscall_args = ia32_fetch_syscall_args, 146 .sv_syscallnames = freebsd32_syscallnames, 147 .sv_onexec_old = exec_onexec_old, 148 .sv_onexit = exit_onexit, 149 .sv_set_fork_retval = x86_set_fork_retval, 150 }; 151 152 static void 153 aout_sysent(void *arg __unused) 154 { 155 aout_szsigcode = (int)(uintptr_t)&_binary_elf_vdso32_so_1_size; 156 } 157 SYSINIT(aout_sysent, SI_SUB_EXEC, SI_ORDER_ANY, aout_sysent, NULL); 158 #else 159 #error "Only ia32 arch is supported" 160 #endif 161 162 static int 163 aout_fixup(uintptr_t *stack_base, struct image_params *imgp) 164 { 165 166 *stack_base -= sizeof(uint32_t); 167 if (suword32((void *)*stack_base, imgp->args->argc) != 0) 168 return (EFAULT); 169 return (0); 170 } 171 172 static int 173 exec_aout_imgact(struct image_params *imgp) 174 { 175 const struct exec *a_out; 176 struct vmspace *vmspace; 177 vm_map_t map; 178 vm_object_t object; 179 vm_offset_t text_end, data_end; 180 unsigned long virtual_offset; 181 unsigned long file_offset; 182 unsigned long bss_size; 183 int error; 184 185 a_out = (const struct exec *)imgp->image_header; 186 187 /* 188 * Linux and *BSD binaries look very much alike, 189 * only the machine id is different: 190 * 0x64 for Linux, 0x86 for *BSD, 0x00 for BSDI. 191 * NetBSD is in network byte order.. ugh. 192 */ 193 if (((a_out->a_midmag >> 16) & 0xff) != 0x86 && 194 ((a_out->a_midmag >> 16) & 0xff) != 0 && 195 ((((int)ntohl(a_out->a_midmag)) >> 16) & 0xff) != 0x86) 196 return (-1); 197 198 /* 199 * Set file/virtual offset based on a.out variant. 200 * We do two cases: host byte order and network byte order 201 * (for NetBSD compatibility) 202 */ 203 switch ((int)(a_out->a_midmag & 0xffff)) { 204 case ZMAGIC: 205 virtual_offset = 0; 206 if (a_out->a_text) { 207 file_offset = PAGE_SIZE; 208 } else { 209 /* Bill's "screwball mode" */ 210 file_offset = 0; 211 } 212 break; 213 case QMAGIC: 214 virtual_offset = PAGE_SIZE; 215 file_offset = 0; 216 /* Pass PS_STRINGS for BSD/OS binaries only. */ 217 if (N_GETMID(*a_out) == MID_ZERO) 218 imgp->ps_strings = (void *)aout_sysvec.sv_psstrings; 219 break; 220 default: 221 /* NetBSD compatibility */ 222 switch ((int)(ntohl(a_out->a_midmag) & 0xffff)) { 223 case ZMAGIC: 224 case QMAGIC: 225 virtual_offset = PAGE_SIZE; 226 file_offset = 0; 227 break; 228 default: 229 return (-1); 230 } 231 } 232 233 bss_size = roundup(a_out->a_bss, PAGE_SIZE); 234 235 /* 236 * Check various fields in header for validity/bounds. 237 */ 238 if (/* entry point must lay with text region */ 239 a_out->a_entry < virtual_offset || 240 a_out->a_entry >= virtual_offset + a_out->a_text || 241 242 /* text and data size must each be page rounded */ 243 a_out->a_text & PAGE_MASK || a_out->a_data & PAGE_MASK 244 245 #ifdef __amd64__ 246 || 247 /* overflows */ 248 virtual_offset + a_out->a_text + a_out->a_data + bss_size > UINT_MAX 249 #endif 250 ) 251 return (-1); 252 253 /* text + data can't exceed file size */ 254 if (a_out->a_data + a_out->a_text > imgp->attr->va_size) 255 return (EFAULT); 256 257 /* 258 * text/data/bss must not exceed limits 259 */ 260 PROC_LOCK(imgp->proc); 261 if (/* text can't exceed maximum text size */ 262 a_out->a_text > maxtsiz || 263 264 /* data + bss can't exceed rlimit */ 265 a_out->a_data + bss_size > lim_cur_proc(imgp->proc, RLIMIT_DATA) || 266 racct_set(imgp->proc, RACCT_DATA, a_out->a_data + bss_size) != 0) { 267 PROC_UNLOCK(imgp->proc); 268 return (ENOMEM); 269 } 270 PROC_UNLOCK(imgp->proc); 271 272 /* 273 * Avoid a possible deadlock if the current address space is destroyed 274 * and that address space maps the locked vnode. In the common case, 275 * the locked vnode's v_usecount is decremented but remains greater 276 * than zero. Consequently, the vnode lock is not needed by vrele(). 277 * However, in cases where the vnode lock is external, such as nullfs, 278 * v_usecount may become zero. 279 */ 280 VOP_UNLOCK(imgp->vp); 281 282 /* 283 * Destroy old process VM and create a new one (with a new stack) 284 */ 285 error = exec_new_vmspace(imgp, &aout_sysvec); 286 287 vn_lock(imgp->vp, LK_SHARED | LK_RETRY); 288 if (error) 289 return (error); 290 291 /* 292 * The vm space can be changed by exec_new_vmspace 293 */ 294 vmspace = imgp->proc->p_vmspace; 295 296 object = imgp->object; 297 map = &vmspace->vm_map; 298 vm_map_lock(map); 299 vm_object_reference(object); 300 301 text_end = virtual_offset + a_out->a_text; 302 error = vm_map_insert(map, object, 303 file_offset, 304 virtual_offset, text_end, 305 VM_PROT_READ | VM_PROT_EXECUTE, VM_PROT_ALL, 306 MAP_COPY_ON_WRITE | MAP_PREFAULT | MAP_VN_EXEC); 307 if (error) { 308 vm_map_unlock(map); 309 vm_object_deallocate(object); 310 return (error); 311 } 312 VOP_SET_TEXT_CHECKED(imgp->vp); 313 data_end = text_end + a_out->a_data; 314 if (a_out->a_data) { 315 vm_object_reference(object); 316 error = vm_map_insert(map, object, 317 file_offset + a_out->a_text, 318 text_end, data_end, 319 VM_PROT_ALL, VM_PROT_ALL, 320 MAP_COPY_ON_WRITE | MAP_PREFAULT | MAP_VN_EXEC); 321 if (error) { 322 vm_map_unlock(map); 323 vm_object_deallocate(object); 324 return (error); 325 } 326 VOP_SET_TEXT_CHECKED(imgp->vp); 327 } 328 329 if (bss_size) { 330 error = vm_map_insert(map, NULL, 0, 331 data_end, data_end + bss_size, 332 VM_PROT_ALL, VM_PROT_ALL, 0); 333 if (error) { 334 vm_map_unlock(map); 335 return (error); 336 } 337 } 338 vm_map_unlock(map); 339 340 /* Fill in process VM information */ 341 vmspace->vm_tsize = a_out->a_text >> PAGE_SHIFT; 342 vmspace->vm_dsize = (a_out->a_data + bss_size) >> PAGE_SHIFT; 343 vmspace->vm_taddr = (caddr_t) (uintptr_t) virtual_offset; 344 vmspace->vm_daddr = (caddr_t) (uintptr_t) 345 (virtual_offset + a_out->a_text); 346 347 error = exec_map_stack(imgp); 348 if (error != 0) 349 return (error); 350 351 /* Fill in image_params */ 352 imgp->interpreted = 0; 353 imgp->entry_addr = a_out->a_entry; 354 355 imgp->proc->p_sysent = &aout_sysvec; 356 357 return (0); 358 } 359 360 /* 361 * Tell kern_execve.c about it, with a little help from the linker. 362 */ 363 static struct execsw aout_execsw = { 364 .ex_imgact = exec_aout_imgact, 365 .ex_name = "a.out" 366 }; 367 EXEC_SET(aout, aout_execsw); 368