Mercurial > vim
view src/termlib.c @ 33811:06219b3bdaf3 v9.0.2121
patch 9.0.2121: [security]: use-after-free in ex_substitute
Commit: https://github.com/vim/vim/commit/26c11c56888d01e298cd8044caf860f3c26f57bb
Author: Christian Brabandt <cb@256bit.org>
Date: Wed Nov 22 21:26:41 2023 +0100
patch 9.0.2121: [security]: use-after-free in ex_substitute
Problem: [security]: use-after-free in ex_substitute
Solution: always allocate memory
closes: #13552
A recursive :substitute command could cause a heap-use-after free in Vim
(CVE-2023-48706).
The whole reproducible test is a bit tricky, I can only reproduce this
reliably when no previous substitution command has been used yet
(which is the reason, the test needs to run as first one in the
test_substitute.vim file) and as a combination of the `:~` command
together with a :s command that contains the special substitution atom `~\=`
which will make use of a sub-replace special atom and calls a vim script
function.
There was a comment in the existing :s code, that already makes the
`sub` variable allocate memory so that a recursive :s call won't be able
to cause any issues here, so this was known as a potential problem
already. But for the current test-case that one does not work, because
the substitution does not start with `\=` but with `~\=` (and since
there does not yet exist a previous substitution atom, Vim will simply
increment the `sub` pointer (which then was not allocated dynamically)
and later one happily use a sub-replace special expression (which could
then free the `sub` var).
The following commit fixes this, by making the sub var always using
allocated memory, which also means we need to free the pointer whenever
we leave the function. Since sub is now always an allocated variable,
we also do no longer need the sub_copy variable anymore, since this one
was used to indicated when sub pointed to allocated memory (and had
therefore to be freed on exit) and when not.
Github Security Advisory:
https://github.com/vim/vim/security/advisories/GHSA-c8qm-x72m-q53q
Signed-off-by: Christian Brabandt <cb@256bit.org>
author | Christian Brabandt <cb@256bit.org> |
---|---|
date | Wed, 22 Nov 2023 22:15:05 +0100 |
parents | 50555279168b |
children |
line wrap: on
line source
/* vi:set ts=8 sts=4 sw=4 noet: */ /* * The following software is (C) 1984 Peter da Silva, the Mad Australian, in * the public domain. It may be re-distributed for any purpose with the * inclusion of this notice. */ // Modified by Bram Moolenaar for use with VIM - Vi Improved. // A few bugs removed by Olaf 'Rhialto' Seibert. // TERMLIB: Terminal independent database. #include "vim.h" #include "termlib.pro" #if !defined(AMIGA) && !defined(VMS) # include <sgtty.h> #endif static int getent(char *, char *, FILE *, int); static int nextent(char *, FILE *, int); static int _match(char *, char *); static char *_addfmt(char *, char *, int); static char *_find(char *, char *); /* * Global variables for termlib */ char *tent; // Pointer to terminal entry, set by tgetent char PC = 0; // Pad character, default NULL char *UP = 0, *BC = 0; // Pointers to UP and BC strings from database short ospeed; // Baud rate (1-16, 1=300, 16=19200), as in stty /* * Module: tgetent * * Purpose: Get termcap entry for <term> into buffer at <tbuf>. * * Calling conventions: char tbuf[TBUFSZ+], term=canonical name for terminal. * * Returned values: 1 = success, -1 = can't open file, * 0 = can't find terminal. * * Notes: * - Should probably supply static buffer. * - Uses environment variables "TERM" and "TERMCAP". If TERM = term (that is, * if the argument matches the environment) then it looks at TERMCAP. * - If TERMCAP begins with a slash, then it assumes this is the file to * search rather than /etc/termcap. * - If TERMCAP does not begin with a slash, and it matches TERM, then this is * used as the entry. * - This could be simplified considerably for non-UNIX systems. */ #ifndef TERMCAPFILE # ifdef AMIGA # define TERMCAPFILE "s:termcap" # else # ifdef VMS # define TERMCAPFILE "VIMRUNTIME:termcap" # else # define TERMCAPFILE "/etc/termcap" # endif # endif #endif int tgetent( char *tbuf, // Buffer to hold termcap entry, TBUFSZ bytes max char *term) // Name of terminal { char tcbuf[32]; // Temp buffer to handle char *tcptr = tcbuf; // extended entries char *tcap = TERMCAPFILE; // Default termcap file char *tmp; FILE *termcap; int retval = 0; int len; if ((tmp = (char *)mch_getenv((char_u *)"TERMCAP")) != NULL) { if (*tmp == '/') // TERMCAP = name of termcap file { tcap = tmp ; #if defined(AMIGA) // Convert /usr/share/lib/termcap to usr:share/lib/termcap tcap++; tmp = strchr(tcap, '/'); if (tmp) *tmp = ':'; #endif } else // TERMCAP = termcap entry itself { int tlen = strlen(term); while (*tmp && *tmp != ':') // Check if TERM matches { char *nexttmp; while (*tmp == '|') tmp++; nexttmp = _find(tmp, ":|"); // Rhialto if (tmp+tlen == nexttmp && _match(tmp, term) == tlen) { strcpy(tbuf, tmp); tent = tbuf; return 1; } else tmp = nexttmp; } } } if (!(termcap = mch_fopen(tcap, "r"))) { strcpy(tbuf, tcap); return -1; } len = 0; while (getent(tbuf + len, term, termcap, TBUFSZ - len)) { tcptr = tcbuf; // Rhialto if ((term = tgetstr("tc", &tcptr))) // extended entry { rewind(termcap); len = strlen(tbuf); } else { retval = 1; tent = tbuf; // reset it back to the beginning break; } } fclose(termcap); return retval; } static int getent(char *tbuf, char *term, FILE *termcap, int buflen) { char *tptr; int tlen = strlen(term); while (nextent(tbuf, termcap, buflen)) // For each possible entry { tptr = tbuf; while (*tptr && *tptr != ':') // : terminates name field { char *nexttptr; while (*tptr == '|') // | separates names tptr++; nexttptr = _find(tptr, ":|"); // Rhialto if (tptr + tlen == nexttptr && _match(tptr, term) == tlen) // FOUND! { tent = tbuf; return 1; } else // Look for next name tptr = nexttptr; } } return 0; } /* * Read 1 entry from TERMCAP file. */ static int nextent(char *tbuf, FILE *termcap, int buflen) { char *lbuf = tbuf; // lbuf=line buffer // read lines straight into buffer while (lbuf < tbuf+buflen && // There's room and fgets(lbuf, (int)(tbuf+buflen-lbuf), termcap)) // another line { int llen = strlen(lbuf); if (*lbuf == '#') // eat comments continue; if (lbuf[-1] == ':' && // and whitespace lbuf[0] == '\t' && lbuf[1] == ':') { STRMOVE(lbuf, lbuf + 2); llen -= 2; } if (lbuf[llen-2] == '\\') // and continuations lbuf += llen-2; else { lbuf[llen-1]=0; // no continuation, return return 1; } } return 0; // ran into end of file } /* * Module: tgetflag * * Purpose: returns flag true or false as to the existence of a given entry. * used with 'bs', 'am', etc... * * Calling conventions: id is the 2 character capability id. * * Returned values: 1 for success, 0 for failure. */ int tgetflag(char *id) { char buf[256], *ptr = buf; return tgetstr(id, &ptr) ? 1 : 0; } /* * Module: tgetnum * * Purpose: get numeric value such as 'li' or 'co' from termcap. * * Calling conventions: id = 2 character id. * * Returned values: -1 for failure, else numerical value. */ int tgetnum(char *id) { char *ptr, buf[256]; ptr = buf; if (tgetstr(id, &ptr)) return atoi(buf); else return 0; } /* * Module: tgetstr * * Purpose: get terminal capability string from database. * * Calling conventions: id is the two character capability id. * (*buf) points into a hold buffer for the * id. the capability is copied into the buffer * and (*buf) is advanced to point to the next * free byte in the buffer. * * Returned values: 0 = no such entry, otherwise returns original * (*buf) (now a pointer to the string). * * Notes * It also decodes certain escape sequences in the buffer. * they should be obvious from the code: * \E = escape. * \n, \r, \t, \f, \b match the 'c' escapes. * ^x matches control-x (^@...^_). * \nnn matches nnn octal. * \x, where x is anything else, matches x. I differ * from the standard library here, in that I allow ^: to match * :. * */ char * tgetstr(char *id, char **buf) { int len = strlen(id); char *tmp=tent; char *hold; int i; do { tmp = _find(tmp, ":"); // For each field while (*tmp == ':') // skip empty fields tmp++; if (!*tmp) break; if (_match(id, tmp) == len) { tmp += len; // find '=' '@' or '#' if (*tmp == '@') // :xx@: entry for tc return 0; // deleted entry hold= *buf; while (*++tmp && *tmp != ':') // not at end of field { switch(*tmp) { case '\\': // Expand escapes here switch(*++tmp) { case 0: // ignore backslashes tmp--; // at end of entry break; // shouldn't happen case 'e': case 'E': // ESC *(*buf)++ = ESC; break; case 'n': // \n *(*buf)++ = '\n'; break; case 'r': // \r *(*buf)++ = '\r'; break; case 't': // \t *(*buf)++ = '\t'; break; case 'b': // \b *(*buf)++ = '\b'; break; case 'f': // \f *(*buf)++ = '\f'; break; case '0': // \nnn case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': **buf = 0; // get up to three digits for (i = 0; i < 3 && VIM_ISDIGIT(*tmp); ++i) **buf = **buf * 8 + *tmp++ - '0'; (*buf)++; tmp--; break; default: // \x, for all other x *(*buf)++= *tmp; } break; case '^': // control characters ++tmp; *(*buf)++ = Ctrl_chr(*tmp); break; default: *(*buf)++ = *tmp; } } *(*buf)++ = 0; return hold; } } while (*tmp); return 0; } /* * Module: tgoto * * Purpose: decode cm cursor motion string. * * Calling conventions: cm is cursor motion string. line, col, are the * desired destination. * * Returned values: a string pointing to the decoded string, or "OOPS" if it * cannot be decoded. * * Notes * The accepted escapes are: * %d as in printf, 0 origin. * %2, %3 like %02d, %03d in printf. * %. like %c * %+x adds <x> to value, then %. * %>xy if value>x, adds y. No output. * %i increments line& col, no output. * %r reverses order of line&col. No output. * %% prints as a single %. * %n exclusive or row & col with 0140. * %B BCD, no output. * %D reverse coding (x-2*(x%16)), no output. */ char * tgoto( char *cm, // cm string, from termcap int col, // column, x position int line) // line, y position { char gx, gy, // x, y *ptr, // pointer in 'cm' reverse = 0, // reverse flag *bufp, // pointer in returned string addup = 0, // add upline addbak = 0, // add backup c; static char buffer[32]; if (!cm) return "OOPS"; // Kludge, but standard bufp = buffer; ptr = cm; while (*ptr) { if ((c = *ptr++) != '%') // normal char { *bufp++ = c; } else { // % escape switch(c = *ptr++) { case 'd': // decimal bufp = _addfmt(bufp, "%d", line); line = col; break; case '2': // 2 digit decimal bufp = _addfmt(bufp, "%02d", line); line = col; break; case '3': // 3 digit decimal bufp = _addfmt(bufp, "%03d", line); line = col; break; case '>': // %>xy: if >x, add y gx = *ptr++; gy = *ptr++; if (col>gx) col += gy; if (line>gx) line += gy; break; case '+': // %+c: add c line += *ptr++; case '.': // print x/y if (line == '\t' || // these are line == '\n' || // chars that line == '\004' || // UNIX hates line == '\0') { line++; // so go to next pos if (reverse == (line == col)) addup=1; // and mark UP else addbak=1; // or BC } *bufp++=line; line = col; break; case 'r': // r: reverse gx = line; line = col; col = gx; reverse = 1; break; case 'i': // increment (1-origin screen) col++; line++; break; case '%': // %%=% literally *bufp++='%'; break; case 'n': // magic DM2500 code line ^= 0140; col ^= 0140; break; case 'B': // bcd encoding line = line/10<<4+line%10; col = col/10<<4+col%10; break; case 'D': // magic Delta Data code line = line-2*(line&15); col = col-2*(col&15); break; default: // Unknown escape return "OOPS"; } } } if (addup) // add upline if (UP) { ptr=UP; while (VIM_ISDIGIT(*ptr) || *ptr == '.') ptr++; if (*ptr == '*') ptr++; while (*ptr) *bufp++ = *ptr++; } if (addbak) // add backspace if (BC) { ptr=BC; while (VIM_ISDIGIT(*ptr) || *ptr == '.') ptr++; if (*ptr == '*') ptr++; while (*ptr) *bufp++ = *ptr++; } else *bufp++='\b'; *bufp = 0; return(buffer); } /* * Module: tputs * * Purpose: decode padding information * * Calling conventions: cp = string to be padded, affcnt = # of items affected * (lines, characters, whatever), outc = routine to output 1 character. * * Returned values: none * * Notes * cp has padding information ahead of it, in the form * nnnTEXT or nnn*TEXT. nnn is the number of milliseconds to delay, * and may be a decimal (nnn.mmm). If the asterisk is given, then * the delay is multiplied by afcnt. The delay is produced by outputting * a number of nulls (or other padding char) after printing the * TEXT. * */ long _bauds[16]={ 0, 50, 75, 110, 134, 150, 200, 300, 600, 1200, 1800, 2400, 4800, 9600, 19200, 19200 }; int tputs( char *cp, // string to print int affcnt, // Number of lines affected void (*outc)(unsigned int)) // routine to output 1 character { long frac, // 10^(#digits after decimal point) counter, // digits atol(const char *); if (VIM_ISDIGIT(*cp)) { counter = 0; frac = 1000; while (VIM_ISDIGIT(*cp)) counter = counter * 10L + (long)(*cp++ - '0'); if (*cp == '.') while (VIM_ISDIGIT(*++cp)) { counter = counter * 10L + (long)(*cp++ - '0'); frac = frac * 10; } if (*cp!='*') // multiply by affected lines { if (affcnt>1) affcnt = 1; } else cp++; // Calculate number of characters for padding counter/frac ms delay if (ospeed) counter = (counter * _bauds[ospeed] * (long)affcnt) / frac; while (*cp) // output string (*outc)(*cp++); if (ospeed) while (counter--) // followed by pad characters (*outc)(PC); } else while (*cp) (*outc)(*cp++); return 0; } /* * Module: tutil.c * * Purpose: Utility routines for TERMLIB functions. * Returns length of text common to s1 and s2. */ static int _match(char *s1, char *s2) { int i = 0; while (s1[i] && s1[i] == s2[i]) i++; return i; } /* * finds next c in s that's a member of set, returns pointer */ static char * _find(char *s, char *set) { for (; *s; s++) { char *ptr = set; while (*ptr && *s != *ptr) ptr++; if (*ptr) return s; } return s; } /* * add val to buf according to format fmt */ static char * _addfmt(char *buf, char *fmt, int val) { sprintf(buf, fmt, val); while (*buf) buf++; return buf; }