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: stable/11/sys/kern/imgact_shell.c 325513 2017-11-07 14:25:09Z pfg $"); 29116182Sobrien 302056Swollman#include <sys/param.h> 31142448Ssobomax#include <sys/vnode.h> 32142448Ssobomax#include <sys/proc.h> 33182191Skib#include <sys/sbuf.h> 3440435Speter#include <sys/systm.h> 3511332Sswallace#include <sys/sysproto.h> 361549Srgrimes#include <sys/exec.h> 372056Swollman#include <sys/imgact.h> 382056Swollman#include <sys/kernel.h> 39885Swollman 40886Swollman#if BYTE_ORDER == LITTLE_ENDIAN 41885Swollman#define SHELLMAGIC 0x2123 /* #! */ 42886Swollman#else 43886Swollman#define SHELLMAGIC 0x2321 44886Swollman#endif 45886Swollman 46885Swollman/* 47146731Sgad * At the time of this writing, MAXSHELLCMDLEN == PAGE_SIZE. This is 48146731Sgad * significant because the caller has only mapped in one page of the 49212965Salc * file we're reading. 50885Swollman */ 51146731Sgad#if MAXSHELLCMDLEN > PAGE_SIZE 52146731Sgad#error "MAXSHELLCMDLEN is larger than a single page!" 53146731Sgad#endif 54146731Sgad 55212965Salc/* 56212965Salc * MAXSHELLCMDLEN must be at least MAXINTERP plus the size of the `#!' 57212965Salc * prefix and terminating newline. 58212965Salc */ 59212965SalcCTASSERT(MAXSHELLCMDLEN >= MAXINTERP + 3); 60212965Salc 61146731Sgad/** 62146731Sgad * Shell interpreter image activator. An interpreter name beginning at 63146731Sgad * imgp->args->begin_argv is the minimal successful exit requirement. 64146731Sgad * 65146731Sgad * If the given file is a shell-script, then the first line will start 66146731Sgad * with the two characters `#!' (aka SHELLMAGIC), followed by the name 67146731Sgad * of the shell-interpreter to run, followed by zero or more tokens. 68146731Sgad * 69146731Sgad * The interpreter is then started up such that it will see: 70146731Sgad * arg[0] -> The name of interpreter as specified after `#!' in the 71146731Sgad * first line of the script. The interpreter name must 72146731Sgad * not be longer than MAXSHELLCMDLEN bytes. 73146731Sgad * arg[1] -> *If* there are any additional tokens on the first line, 74146731Sgad * then we add a new arg[1], which is a copy of the rest of 75146731Sgad * that line. The copy starts at the first token after the 76146731Sgad * interpreter name. We leave it to the interpreter to 77146731Sgad * parse the tokens in that value. 78146731Sgad * arg[x] -> the full pathname of the script. This will either be 79146731Sgad * arg[2] or arg[1], depending on whether or not tokens 80146731Sgad * were found after the interpreter name. 81146731Sgad * arg[x+1] -> all the arguments that were specified on the original 82146731Sgad * command line. 83146731Sgad * 84146731Sgad * This processing is described in the execve(2) man page. 85146731Sgad */ 86146731Sgad 87146731Sgad/* 88146731Sgad * HISTORICAL NOTE: From 1993 to mid-2005, FreeBSD parsed out the tokens as 89146731Sgad * found on the first line of the script, and setup each token as a separate 90146731Sgad * value in arg[]. This extra processing did not match the behavior of other 91146731Sgad * OS's, and caused a few subtle problems. For one, it meant the kernel was 92146731Sgad * deciding how those values should be parsed (wrt characters for quoting or 93146731Sgad * comments, etc), while the interpreter might have other rules for parsing. 94146731Sgad * It also meant the interpreter had no way of knowing which arguments came 95146731Sgad * from the first line of the shell script, and which arguments were specified 96147151Sgad * by the user on the command line. That extra processing was dropped in the 97147151Sgad * 6.x branch on May 28, 2005 (matching __FreeBSD_version 600029). 98146731Sgad */ 9959663Sdillonint 100325513Spfgexec_shell_imgact(struct image_params *imgp) 101885Swollman{ 10212130Sdg const char *image_header = imgp->image_header; 103182191Skib const char *ihp, *interpb, *interpe, *maxp, *optb, *opte, *fname; 104141028Ssobomax int error, offset; 105212965Salc size_t length; 106142448Ssobomax struct vattr vattr; 107182191Skib struct sbuf *sname; 108885Swollman 109885Swollman /* a shell script? */ 110212965Salc if (((const short *)image_header)[0] != SHELLMAGIC) 111212965Salc return (-1); 112885Swollman 113885Swollman /* 114885Swollman * Don't allow a shell script to be the shell for a shell 115885Swollman * script. :-) 116885Swollman */ 117271141Ssbruno if (imgp->interpreted & IMGACT_SHELL) 118212965Salc return (ENOEXEC); 119885Swollman 120271141Ssbruno imgp->interpreted |= IMGACT_SHELL; 121885Swollman 122885Swollman /* 123142448Ssobomax * At this point we have the first page of the file mapped. 124142448Ssobomax * However, we don't know how far into the page the contents are 125142448Ssobomax * valid -- the actual file might be much shorter than the page. 126142448Ssobomax * So find out the file size. 127325513Spfg */ 128182371Sattilio error = VOP_GETATTR(imgp->vp, &vattr, imgp->proc->p_ucred); 129142448Ssobomax if (error) 130142448Ssobomax return (error); 131142448Ssobomax 132142448Ssobomax /* 133146731Sgad * Copy shell name and arguments from image_header into a string 134212965Salc * buffer. 135146731Sgad */ 136212965Salc maxp = &image_header[MIN(vattr.va_size, MAXSHELLCMDLEN)]; 137146731Sgad ihp = &image_header[2]; 138146731Sgad 139146731Sgad /* 140146731Sgad * Find the beginning and end of the interpreter_name. If the 141146731Sgad * line does not include any interpreter, or if the name which 142146731Sgad * was found is too long, we bail out. 143146731Sgad */ 144146731Sgad while (ihp < maxp && ((*ihp == ' ') || (*ihp == '\t'))) 145146731Sgad ihp++; 146146731Sgad interpb = ihp; 147146731Sgad while (ihp < maxp && ((*ihp != ' ') && (*ihp != '\t') && (*ihp != '\n') 148146731Sgad && (*ihp != '\0'))) 149146731Sgad ihp++; 150146731Sgad interpe = ihp; 151146731Sgad if (interpb == interpe) 152146731Sgad return (ENOEXEC); 153212965Salc if (interpe - interpb >= MAXINTERP) 154146731Sgad return (ENAMETOOLONG); 155885Swollman 156140992Ssobomax /* 157146731Sgad * Find the beginning of the options (if any), and the end-of-line. 158146731Sgad * Then trim the trailing blanks off the value. Note that some 159146731Sgad * other operating systems do *not* trim the trailing whitespace... 160146731Sgad */ 161146731Sgad while (ihp < maxp && ((*ihp == ' ') || (*ihp == '\t'))) 162146731Sgad ihp++; 163146731Sgad optb = ihp; 164146731Sgad while (ihp < maxp && ((*ihp != '\n') && (*ihp != '\0'))) 165146731Sgad ihp++; 166146731Sgad opte = ihp; 167212965Salc if (opte == maxp) 168212965Salc return (ENOEXEC); 169147479Sgad while (--ihp > optb && ((*ihp == ' ') || (*ihp == '\t'))) 170146731Sgad opte = ihp; 171146731Sgad 172182191Skib if (imgp->args->fname != NULL) { 173182191Skib fname = imgp->args->fname; 174182191Skib sname = NULL; 175182191Skib } else { 176182191Skib sname = sbuf_new_auto(); 177182191Skib sbuf_printf(sname, "/dev/fd/%d", imgp->args->fd); 178182191Skib sbuf_finish(sname); 179182191Skib fname = sbuf_data(sname); 180182191Skib } 181182191Skib 182146731Sgad /* 183146731Sgad * We need to "pop" (remove) the present value of arg[0], and "push" 184146731Sgad * either two or three new values in the arg[] list. To do this, 185146731Sgad * we first shift all the other values in the `begin_argv' area to 186146731Sgad * provide the exact amount of room for the values added. Set up 187146731Sgad * `offset' as the number of bytes to be added to the `begin_argv' 188146731Sgad * area, and 'length' as the number of bytes being removed. 189146731Sgad */ 190146731Sgad offset = interpe - interpb + 1; /* interpreter */ 191147479Sgad if (opte > optb) /* options (if any) */ 192146731Sgad offset += opte - optb + 1; 193182191Skib offset += strlen(fname) + 1; /* fname of script */ 194146731Sgad length = (imgp->args->argc == 0) ? 0 : 195146731Sgad strlen(imgp->args->begin_argv) + 1; /* bytes to delete */ 196146731Sgad 197219352Skib if (offset > imgp->args->stringspace + length) { 198182191Skib if (sname != NULL) 199182191Skib sbuf_delete(sname); 200146731Sgad return (E2BIG); 201182191Skib } 202146731Sgad 203146731Sgad bcopy(imgp->args->begin_argv + length, imgp->args->begin_argv + offset, 204146731Sgad imgp->args->endp - (imgp->args->begin_argv + length)); 205146731Sgad 206146731Sgad offset -= length; /* calculate actual adjustment */ 207146731Sgad imgp->args->begin_envv += offset; 208146731Sgad imgp->args->endp += offset; 209146731Sgad imgp->args->stringspace -= offset; 210146731Sgad 211146731Sgad /* 212146731Sgad * If there was no arg[0] when we started, then the interpreter_name 213146731Sgad * is adding an argument (instead of replacing the arg[0] we started 214146731Sgad * with). And we're always adding an argument when we include the 215146731Sgad * full pathname of the original script. 216146731Sgad */ 217146731Sgad if (imgp->args->argc == 0) 218146731Sgad imgp->args->argc = 1; 219146731Sgad imgp->args->argc++; 220146731Sgad 221146731Sgad /* 222146731Sgad * The original arg[] list has been shifted appropriately. Copy in 223146731Sgad * the interpreter name and options-string. 224146731Sgad */ 225146731Sgad length = interpe - interpb; 226210475Salc bcopy(interpb, imgp->args->begin_argv, length); 227210475Salc *(imgp->args->begin_argv + length) = '\0'; 228146731Sgad offset = length + 1; 229147479Sgad if (opte > optb) { 230146731Sgad length = opte - optb; 231210475Salc bcopy(optb, imgp->args->begin_argv + offset, length); 232210475Salc *(imgp->args->begin_argv + offset + length) = '\0'; 233146731Sgad offset += length + 1; 234146731Sgad imgp->args->argc++; 235146731Sgad } 236146731Sgad 237146731Sgad /* 238140992Ssobomax * Finally, add the filename onto the end for the interpreter to 239140992Ssobomax * use and copy the interpreter's name to imgp->interpreter_name 240140992Ssobomax * for exec to use. 241140992Ssobomax */ 242210475Salc error = copystr(fname, imgp->args->begin_argv + offset, 243210475Salc imgp->args->stringspace, NULL); 244885Swollman 245140992Ssobomax if (error == 0) 246210545Salc imgp->interpreter_name = imgp->args->begin_argv; 247140992Ssobomax 248182191Skib if (sname != NULL) 249182191Skib sbuf_delete(sname); 250140992Ssobomax return (error); 251885Swollman} 252886Swollman 253886Swollman/* 254886Swollman * Tell kern_execve.c about it, with a little help from the linker. 255886Swollman */ 25643402Sdillonstatic struct execsw shell_execsw = { exec_shell_imgact, "#!" }; 25740435SpeterEXEC_SET(shell, shell_execsw); 258