imgact_shell.c revision 147151
1139804Simp/*- 2885Swollman * Copyright (c) 1993, David Greenman 3885Swollman * All rights reserved. 4885Swollman * 5885Swollman * Redistribution and use in source and binary forms, with or without 6885Swollman * modification, are permitted provided that the following conditions 7885Swollman * are met: 8885Swollman * 1. Redistributions of source code must retain the above copyright 9885Swollman * notice, this list of conditions and the following disclaimer. 10885Swollman * 2. Redistributions in binary form must reproduce the above copyright 11885Swollman * notice, this list of conditions and the following disclaimer in the 12885Swollman * documentation and/or other materials provided with the distribution. 13885Swollman * 14885Swollman * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 15885Swollman * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 16885Swollman * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 1710625Sdg * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 18885Swollman * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 19885Swollman * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 20885Swollman * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 21885Swollman * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 22885Swollman * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 23885Swollman * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 24885Swollman * SUCH DAMAGE. 25885Swollman */ 26885Swollman 27116182Sobrien#include <sys/cdefs.h> 28116182Sobrien__FBSDID("$FreeBSD: head/sys/kern/imgact_shell.c 147151 2005-06-09 00:27:02Z gad $"); 29116182Sobrien 302056Swollman#include <sys/param.h> 31142448Ssobomax#include <sys/vnode.h> 32142448Ssobomax#include <sys/proc.h> 3340435Speter#include <sys/systm.h> 3411332Sswallace#include <sys/sysproto.h> 351549Srgrimes#include <sys/exec.h> 362056Swollman#include <sys/imgact.h> 372056Swollman#include <sys/kernel.h> 38885Swollman 39886Swollman#if BYTE_ORDER == LITTLE_ENDIAN 40885Swollman#define SHELLMAGIC 0x2123 /* #! */ 41886Swollman#else 42886Swollman#define SHELLMAGIC 0x2321 43886Swollman#endif 44886Swollman 45885Swollman/* 46146731Sgad * At the time of this writing, MAXSHELLCMDLEN == PAGE_SIZE. This is 47146731Sgad * significant because the caller has only mapped in one page of the 48146731Sgad * file we're reading. This code should be changed to know how to 49146731Sgad * read in the second page, but I'm not doing that just yet... 50885Swollman */ 51146731Sgad#if MAXSHELLCMDLEN > PAGE_SIZE 52146731Sgad#error "MAXSHELLCMDLEN is larger than a single page!" 53146731Sgad#endif 54146731Sgad 55146731Sgad/** 56146731Sgad * Shell interpreter image activator. An interpreter name beginning at 57146731Sgad * imgp->args->begin_argv is the minimal successful exit requirement. 58146731Sgad * 59146731Sgad * If the given file is a shell-script, then the first line will start 60146731Sgad * with the two characters `#!' (aka SHELLMAGIC), followed by the name 61146731Sgad * of the shell-interpreter to run, followed by zero or more tokens. 62146731Sgad * 63146731Sgad * The interpreter is then started up such that it will see: 64146731Sgad * arg[0] -> The name of interpreter as specified after `#!' in the 65146731Sgad * first line of the script. The interpreter name must 66146731Sgad * not be longer than MAXSHELLCMDLEN bytes. 67146731Sgad * arg[1] -> *If* there are any additional tokens on the first line, 68146731Sgad * then we add a new arg[1], which is a copy of the rest of 69146731Sgad * that line. The copy starts at the first token after the 70146731Sgad * interpreter name. We leave it to the interpreter to 71146731Sgad * parse the tokens in that value. 72146731Sgad * arg[x] -> the full pathname of the script. This will either be 73146731Sgad * arg[2] or arg[1], depending on whether or not tokens 74146731Sgad * were found after the interpreter name. 75146731Sgad * arg[x+1] -> all the arguments that were specified on the original 76146731Sgad * command line. 77146731Sgad * 78146731Sgad * This processing is described in the execve(2) man page. 79146731Sgad */ 80146731Sgad 81146731Sgad/* 82146731Sgad * HISTORICAL NOTE: From 1993 to mid-2005, FreeBSD parsed out the tokens as 83146731Sgad * found on the first line of the script, and setup each token as a separate 84146731Sgad * value in arg[]. This extra processing did not match the behavior of other 85146731Sgad * OS's, and caused a few subtle problems. For one, it meant the kernel was 86146731Sgad * deciding how those values should be parsed (wrt characters for quoting or 87146731Sgad * comments, etc), while the interpreter might have other rules for parsing. 88146731Sgad * It also meant the interpreter had no way of knowing which arguments came 89146731Sgad * from the first line of the shell script, and which arguments were specified 90147151Sgad * by the user on the command line. That extra processing was dropped in the 91147151Sgad * 6.x branch on May 28, 2005 (matching __FreeBSD_version 600029). 92146731Sgad */ 9359663Sdillonint 9412130Sdgexec_shell_imgact(imgp) 9512130Sdg struct image_params *imgp; 96885Swollman{ 9712130Sdg const char *image_header = imgp->image_header; 98146731Sgad const char *ihp, *interpb, *interpe, *maxp, *optb, *opte; 99141028Ssobomax int error, offset; 100142448Ssobomax size_t length, clength; 101142448Ssobomax struct vattr vattr; 102885Swollman 103885Swollman /* a shell script? */ 10417974Sbde if (((const short *) image_header)[0] != SHELLMAGIC) 105885Swollman return(-1); 106885Swollman 107885Swollman /* 108885Swollman * Don't allow a shell script to be the shell for a shell 109885Swollman * script. :-) 110885Swollman */ 11112130Sdg if (imgp->interpreted) 112885Swollman return(ENOEXEC); 113885Swollman 11412130Sdg imgp->interpreted = 1; 115885Swollman 116885Swollman /* 117142448Ssobomax * At this point we have the first page of the file mapped. 118142448Ssobomax * However, we don't know how far into the page the contents are 119142448Ssobomax * valid -- the actual file might be much shorter than the page. 120142448Ssobomax * So find out the file size. 121142448Ssobomax */ 122142448Ssobomax error = VOP_GETATTR(imgp->vp, &vattr, imgp->proc->p_ucred, curthread); 123142448Ssobomax if (error) 124142448Ssobomax return (error); 125142448Ssobomax 126142448Ssobomax /* 127146731Sgad * Copy shell name and arguments from image_header into a string 128146731Sgad * buffer. Remember that the caller has mapped only the 129146731Sgad * first page of the file into memory. 130146731Sgad */ 131146731Sgad clength = (vattr.va_size > PAGE_SIZE) ? PAGE_SIZE : vattr.va_size; 132146731Sgad 133146731Sgad maxp = &image_header[clength]; 134146731Sgad ihp = &image_header[2]; 135146731Sgad 136146731Sgad /* 137146731Sgad * Find the beginning and end of the interpreter_name. If the 138146731Sgad * line does not include any interpreter, or if the name which 139146731Sgad * was found is too long, we bail out. 140146731Sgad */ 141146731Sgad while (ihp < maxp && ((*ihp == ' ') || (*ihp == '\t'))) 142146731Sgad ihp++; 143146731Sgad interpb = ihp; 144146731Sgad while (ihp < maxp && ((*ihp != ' ') && (*ihp != '\t') && (*ihp != '\n') 145146731Sgad && (*ihp != '\0'))) 146146731Sgad ihp++; 147146731Sgad interpe = ihp; 148146731Sgad if (interpb == interpe) 149146731Sgad return (ENOEXEC); 150146731Sgad if ((interpe - interpb) >= MAXSHELLCMDLEN) 151146731Sgad return (ENAMETOOLONG); 152885Swollman 153140992Ssobomax /* 154146731Sgad * Find the beginning of the options (if any), and the end-of-line. 155146731Sgad * Then trim the trailing blanks off the value. Note that some 156146731Sgad * other operating systems do *not* trim the trailing whitespace... 157146731Sgad */ 158146731Sgad while (ihp < maxp && ((*ihp == ' ') || (*ihp == '\t'))) 159146731Sgad ihp++; 160146731Sgad optb = ihp; 161146731Sgad while (ihp < maxp && ((*ihp != '\n') && (*ihp != '\0'))) 162146731Sgad ihp++; 163146731Sgad opte = ihp; 164146731Sgad while (--ihp > interpe && ((*ihp == ' ') || (*ihp == '\t'))) 165146731Sgad opte = ihp; 166146731Sgad 167146731Sgad /* 168146731Sgad * We need to "pop" (remove) the present value of arg[0], and "push" 169146731Sgad * either two or three new values in the arg[] list. To do this, 170146731Sgad * we first shift all the other values in the `begin_argv' area to 171146731Sgad * provide the exact amount of room for the values added. Set up 172146731Sgad * `offset' as the number of bytes to be added to the `begin_argv' 173146731Sgad * area, and 'length' as the number of bytes being removed. 174146731Sgad */ 175146731Sgad offset = interpe - interpb + 1; /* interpreter */ 176146731Sgad if (opte != optb) /* options (if any) */ 177146731Sgad offset += opte - optb + 1; 178146731Sgad offset += strlen(imgp->args->fname) + 1; /* fname of script */ 179146731Sgad length = (imgp->args->argc == 0) ? 0 : 180146731Sgad strlen(imgp->args->begin_argv) + 1; /* bytes to delete */ 181146731Sgad 182146731Sgad if (offset - length > imgp->args->stringspace) 183146731Sgad return (E2BIG); 184146731Sgad 185146731Sgad bcopy(imgp->args->begin_argv + length, imgp->args->begin_argv + offset, 186146731Sgad imgp->args->endp - (imgp->args->begin_argv + length)); 187146731Sgad 188146731Sgad offset -= length; /* calculate actual adjustment */ 189146731Sgad imgp->args->begin_envv += offset; 190146731Sgad imgp->args->endp += offset; 191146731Sgad imgp->args->stringspace -= offset; 192146731Sgad 193146731Sgad /* 194146731Sgad * If there was no arg[0] when we started, then the interpreter_name 195146731Sgad * is adding an argument (instead of replacing the arg[0] we started 196146731Sgad * with). And we're always adding an argument when we include the 197146731Sgad * full pathname of the original script. 198146731Sgad */ 199146731Sgad if (imgp->args->argc == 0) 200146731Sgad imgp->args->argc = 1; 201146731Sgad imgp->args->argc++; 202146731Sgad 203146731Sgad /* 204146731Sgad * The original arg[] list has been shifted appropriately. Copy in 205146731Sgad * the interpreter name and options-string. 206146731Sgad */ 207146731Sgad length = interpe - interpb; 208146731Sgad bcopy(interpb, imgp->args->buf, length); 209146731Sgad *(imgp->args->buf + length) = '\0'; 210146731Sgad offset = length + 1; 211146731Sgad if (opte != optb) { 212146731Sgad length = opte - optb; 213146731Sgad bcopy(optb, imgp->args->buf + offset, length); 214146731Sgad *(imgp->args->buf + offset + length) = '\0'; 215146731Sgad offset += length + 1; 216146731Sgad imgp->args->argc++; 217146731Sgad } 218146731Sgad 219146731Sgad /* 220140992Ssobomax * Finally, add the filename onto the end for the interpreter to 221140992Ssobomax * use and copy the interpreter's name to imgp->interpreter_name 222140992Ssobomax * for exec to use. 223140992Ssobomax */ 224140992Ssobomax error = copystr(imgp->args->fname, imgp->args->buf + offset, 225140992Ssobomax imgp->args->stringspace, &length); 226885Swollman 227140992Ssobomax if (error == 0) 228140992Ssobomax error = copystr(imgp->args->begin_argv, imgp->interpreter_name, 229140992Ssobomax MAXSHELLCMDLEN, &length); 230140992Ssobomax 231140992Ssobomax return (error); 232885Swollman} 233886Swollman 234886Swollman/* 235886Swollman * Tell kern_execve.c about it, with a little help from the linker. 236886Swollman */ 23743402Sdillonstatic struct execsw shell_execsw = { exec_shell_imgact, "#!" }; 23840435SpeterEXEC_SET(shell, shell_execsw); 239