X-Git-Url: https://git.exim.org/exim.git/blobdiff_plain/5976eb9983e5f88f22d55f26ddac53c23aeb7f3d..1e1ddfac79fbcd052f199500a6493c7f79cb8462:/src/src/parse.c

diff --git a/src/src/parse.c b/src/src/parse.c
index a7231f39b..e3b471f1a 100644
--- a/src/src/parse.c
+++ b/src/src/parse.c
@@ -2,7 +2,8 @@
 *     Exim - an Internet mail transport agent    *
 *************************************************/
 
-/* Copyright (c) University of Cambridge 1995 - 2017 */
+/* Copyright (c) University of Cambridge 1995 - 2018 */
+/* Copyright (c) The Exim Maintainers 2020 */
 /* See the file NOTICE for conditions of use and distribution. */
 
 /* Functions for parsing addresses */
@@ -23,7 +24,7 @@ redundant apparatus. */
 
 address_item *deliver_make_addr(uschar *address, BOOL copy)
 {
-address_item *addr = store_get(sizeof(address_item));
+address_item *addr = store_get(sizeof(address_item), FALSE);
 addr->next = NULL;
 addr->parent = NULL;
 addr->address = address;
@@ -197,10 +198,10 @@ last_comment_position = s;
 while (*s)
   {
   int c, level;
-  while (isspace(*s)) s++;
-  if (*s != '(') break;
+
+  if (Uskip_whitespace(&s) != '(') break;
   level = 1;
-  while((c = *(++s)) != 0)
+  while((c = *(++s)))
     {
     if (c == '(') level++;
     else if (c == ')') { if (--level <= 0) { s++; break; } }
@@ -421,10 +422,10 @@ for (;;)
   if (*s == '\"')
     {
     *t++ = '\"';
-    while ((c = *(++s)) != 0 && c != '\"')
+    while ((c = *++s) && c != '\"')
       {
       *t++ = c;
-      if (c == '\\' && s[1] != 0) *t++ = *(++s);
+      if (c == '\\' && s[1]) *t++ = *++s;
       }
     if (c == '\"')
       {
@@ -443,7 +444,7 @@ for (;;)
   else while (!mac_iscntrl_or_special(*s) || *s == '\\')
     {
     c = *t++ = *s++;
-    if (c == '\\' && *s != 0) *t++ = *s++;
+    if (c == '\\' && *s) *t++ = *s++;
     }
 
   /* Terminate the word and skip subsequent comment */
@@ -618,10 +619,10 @@ uschar *
 parse_extract_address(uschar *mailbox, uschar **errorptr, int *start, int *end,
   int *domain, BOOL allow_null)
 {
-uschar *yield = store_get(Ustrlen(mailbox) + 1);
+uschar *yield = store_get(Ustrlen(mailbox) + 1, is_tainted(mailbox));
 uschar *startptr, *endptr;
-uschar *s = (uschar *)mailbox;
-uschar *t = (uschar *)yield;
+uschar *s = US mailbox;
+uschar *t = US yield;
 
 *domain = 0;
 
@@ -638,7 +639,7 @@ RESTART:   /* Come back here after passing a group name */
 s = skip_comment(s);
 startptr = s;                                 /* In case addr-spec */
 s = read_local_part(s, t, errorptr, TRUE);    /* Dot separated words */
-if (*errorptr != NULL) goto PARSE_FAILED;
+if (*errorptr) goto PARSE_FAILED;
 
 /* If the terminator is neither < nor @ then the format of the address
 must either be a bare local-part (we are now at the end), or a phrase
@@ -658,7 +659,7 @@ if (*s != '@' && *s != '<')
   end of string will produce a null local_part and therefore fail. We don't
   need to keep updating t, as the phrase isn't to be kept. */
 
-  while (*s != '<' && (!parse_allow_group || *s != ':'))
+  while (*s != '<' && (!f.parse_allow_group || *s != ':'))
     {
     s = read_local_part(s, t, errorptr, FALSE);
     if (*errorptr)
@@ -670,8 +671,8 @@ if (*s != '@' && *s != '<')
 
   if (*s == ':')
     {
-    parse_found_group = TRUE;
-    parse_allow_group = FALSE;
+    f.parse_found_group = TRUE;
+    f.parse_allow_group = FALSE;
     s++;
     goto RESTART;
     }
@@ -745,7 +746,7 @@ if (*s == '<')
     *errorptr = s[-1] == 0
       ? US"'>' missing at end of address"
       : string_sprintf("malformed address: %.32s may not follow %.*s",
-	  s-1, s - (uschar *)mailbox - 1, mailbox);
+	  s-1, (int)(s - US mailbox - 1), mailbox);
     goto PARSE_FAILED;
     }
 
@@ -790,21 +791,21 @@ move it back past white space if necessary. */
 PARSE_SUCCEEDED:
 if (*s != 0)
   {
-  if (parse_found_group && *s == ';')
+  if (f.parse_found_group && *s == ';')
     {
-    parse_found_group = FALSE;
-    parse_allow_group = TRUE;
+    f.parse_found_group = FALSE;
+    f.parse_allow_group = TRUE;
     }
   else
     {
     *errorptr = string_sprintf("malformed address: %.32s may not follow %.*s",
-      s, s - (uschar *)mailbox, mailbox);
+      s, (int)(s - US mailbox), mailbox);
     goto PARSE_FAILED;
     }
   }
-*start = startptr - (uschar *)mailbox;      /* Return offsets */
+*start = startptr - US mailbox;      /* Return offsets */
 while (isspace(endptr[-1])) endptr--;
-*end = endptr - (uschar *)mailbox;
+*end = endptr - US mailbox;
 
 /* Although this code has no limitation on the length of address extracted,
 other parts of Exim may have limits, and in any case, RFC 2821 limits local
@@ -824,10 +825,10 @@ We might have an empty address in a group - the caller can choose to ignore
 this. We must, however, keep the flags correct. */
 
 PARSE_FAILED:
-if (parse_found_group && *s == ';')
+if (f.parse_found_group && *s == ';')
   {
-  parse_found_group = FALSE;
-  parse_allow_group = TRUE;
+  f.parse_found_group = FALSE;
+  f.parse_allow_group = TRUE;
   }
 return NULL;
 }
@@ -876,7 +877,7 @@ int hlen;
 BOOL coded = FALSE;
 BOOL first_byte = FALSE;
 
-if (charset == NULL) charset = US"iso-8859-1";
+if (!charset) charset = US"iso-8859-1";
 
 /* We don't expect this to fail! */
 
@@ -925,7 +926,7 @@ for (; len > 0; len--)
 *t++ = '=';
 *t = 0;
 
-return coded? buffer : string;
+return coded ? buffer : string;
 }
 
 
@@ -1277,10 +1278,10 @@ for (;;)
   However, if the list is empty only because syntax errors were skipped, we
   return FF_DELIVERED. */
 
-  if (*s == 0)
+  if (!*s)
     {
-    return (count > 0 || (syntax_errors != NULL && *syntax_errors != NULL))?
-      FF_DELIVERED : FF_NOTDELIVERED;
+    return (count > 0 || (syntax_errors && *syntax_errors))
+      ?  FF_DELIVERED : FF_NOTDELIVERED;
 
     /* This previous code returns FF_ERROR if nothing is generated but a
     syntax error has been skipped. I now think it is the wrong approach, but
@@ -1396,7 +1397,7 @@ for (;;)
 
     if (flen <= 0)
       {
-      *error = string_sprintf("file name missing after :include:");
+      *error = US"file name missing after :include:";
       return FF_ERROR;
       }
 
@@ -1411,7 +1412,7 @@ for (;;)
 
     /* Insist on absolute path */
 
-    if (filename[0]!= '/')
+    if (filename[0] != '/')
       {
       *error = string_sprintf("included file \"%s\" is not an absolute path",
         filename);
@@ -1420,12 +1421,19 @@ for (;;)
 
     /* Check if include is permitted */
 
-    if ((options & RDO_INCLUDE) != 0)
+    if (options & RDO_INCLUDE)
       {
       *error = US"included files not permitted";
       return FF_ERROR;
       }
 
+    if (is_tainted(filename))
+      {
+      *error = string_sprintf("Tainted name '%s' for included file  not permitted\n",
+       filename);
+      return FF_ERROR;
+      }
+
     /* Check file name if required */
 
     if (directory)
@@ -1447,7 +1455,7 @@ for (;;)
       with a flag that fails symlinks. */
 
       {
-      int fd = open(CS directory, O_RDONLY);
+      int fd = exim_open2(CS directory, O_RDONLY);
       if (fd < 0)
 	{
 	*error = string_sprintf("failed to open directory %s", directory);
@@ -1463,7 +1471,7 @@ for (;;)
 	temp = *p;
 	*p = '\0';
 
-	fd2 = openat(fd, CS q, O_RDONLY|O_NOFOLLOW);
+	fd2 = exim_openat(fd, CS q, O_RDONLY|O_NOFOLLOW);
 	close(fd);
 	*p = temp;
 	if (fd2 < 0)
@@ -1547,7 +1555,7 @@ for (;;)
       return FF_ERROR;
       }
 
-    filebuf = store_get(statbuf.st_size + 1);
+    filebuf = store_get(statbuf.st_size + 1, is_tainted(filename));
     if (fread(filebuf, 1, statbuf.st_size, f) != statbuf.st_size)
       {
       *error = string_sprintf("error while reading included file %s: %s",
@@ -1607,14 +1615,14 @@ for (;;)
       {
       recipient =
         parse_extract_address(s+1, error, &start, &end, &domain, FALSE);
-      if (recipient != NULL)
-        recipient = (domain != 0)? NULL :
+      if (recipient)
+        recipient = domain != 0 ? NULL :
           string_sprintf("%s@%s", recipient, incoming_domain);
       }
 
     /* Try parsing the item as an address. */
 
-    if (recipient == NULL) recipient =
+    if (!recipient) recipient =
       parse_extract_address(s, error, &start, &end, &domain, FALSE);
 
     /* If item starts with / or | and is not a valid address, or there
@@ -1623,7 +1631,7 @@ for (;;)
 
     if ((*s == '|' || *s == '/') && (recipient == NULL || domain == 0))
       {
-      uschar *t = store_get(Ustrlen(s) + 1);
+      uschar *t = store_get(Ustrlen(s) + 1, is_tainted(s));
       uschar *p = t;
       uschar *q = s;
       while (*q != 0)
@@ -1662,7 +1670,7 @@ for (;;)
 
         if (syntax_errors != NULL)
           {
-          error_block *e = store_get(sizeof(error_block));
+          error_block *e = store_get(sizeof(error_block), FALSE);
           error_block *last = *syntax_errors;
           if (last == NULL) *syntax_errors = e; else
             {
@@ -1730,6 +1738,7 @@ parse_message_id(uschar *str, uschar **yield, uschar **error)
 {
 uschar *domain = NULL;
 uschar *id;
+rmark reset_point;
 
 str = skip_comment(str);
 if (*str != '<')
@@ -1742,27 +1751,28 @@ if (*str != '<')
 for the answer, but it may also be very long if we are processing a header
 line. Therefore, take care to release unwanted store afterwards. */
 
-id = *yield = store_get(Ustrlen(str) + 1);
+reset_point = store_mark();
+id = *yield = store_get(Ustrlen(str) + 1, is_tainted(str));
 *id++ = *str++;
 
 str = read_addr_spec(str, id, '>', error, &domain);
 
-if (*error == NULL)
+if (!*error)
   {
   if (*str != '>') *error = US"Missing '>' after message-id";
     else if (domain == NULL) *error = US"domain missing in message-id";
   }
 
-if (*error != NULL)
+if (*error)
   {
-  store_reset(*yield);
+  store_reset(reset_point);
   return NULL;
   }
 
-while (*id != 0) id++;
+while (*id) id++;
 *id++ = *str++;
 *id++ = 0;
-store_reset(id);
+store_release_above(id);
 
 str = skip_comment(str);
 return str;
@@ -2118,7 +2128,9 @@ while (Ufgets(buffer, sizeof(buffer), stdin) != NULL)
   buffer[Ustrlen(buffer) - 1] = 0;
   if (buffer[0] == 0) break;
   out = parse_extract_address(buffer, &errmess, &start, &end, &domain, FALSE);
-  if (out == NULL) printf("*** bad address: %s\n", errmess); else
+  if (!out)
+    printf("*** bad address: %s\n", errmess);
+  else
     {
     uschar extract[1024];
     Ustrncpy(extract, buffer+start, end-start);
@@ -2137,7 +2149,9 @@ while (Ufgets(buffer, sizeof(buffer), stdin) != NULL)
   buffer[Ustrlen(buffer) - 1] = 0;
   if (buffer[0] == 0) break;
   out = parse_extract_address(buffer, &errmess, &start, &end, &domain, FALSE);
-  if (out == NULL) printf("*** bad address: %s\n", errmess); else
+  if (!out)
+    printf("*** bad address: %s\n", errmess);
+  else
     {
     uschar extract[1024];
     Ustrncpy(extract, buffer+start, end-start);
@@ -2149,7 +2163,7 @@ allow_utf8_domains = FALSE;
 
 printf("Testing parse_extract_address with group syntax\n");
 
-parse_allow_group = TRUE;
+f.parse_allow_group = TRUE;
 while (Ufgets(buffer, sizeof(buffer), stdin) != NULL)
   {
   uschar *out;
@@ -2158,7 +2172,7 @@ while (Ufgets(buffer, sizeof(buffer), stdin) != NULL)
   buffer[Ustrlen(buffer) - 1] = 0;
   if (buffer[0] == 0) break;
   s = buffer;
-  while (*s != 0)
+  while (*s)
     {
     uschar *ss = parse_find_address_end(s, FALSE);
     int terminator = *ss;
@@ -2166,7 +2180,9 @@ while (Ufgets(buffer, sizeof(buffer), stdin) != NULL)
     out = parse_extract_address(buffer, &errmess, &start, &end, &domain, FALSE);
     *ss = terminator;
 
-    if (out == NULL) printf("*** bad address: %s\n", errmess); else
+    if (!out)
+      printf("*** bad address: %s\n", errmess);
+    else
       {
       uschar extract[1024];
       Ustrncpy(extract, buffer+start, end-start);
@@ -2175,7 +2191,7 @@ while (Ufgets(buffer, sizeof(buffer), stdin) != NULL)
       }
 
     s = ss + (terminator? 1:0);
-    while (isspace(*s)) s++;
+    Uskip_whitespace(&s);
     }
   }