Fix regerror() small buffer issues.

ChangeLog

@@ -226,6 +226,9 @@ overflow.
 make it easier to test long repetitive patterns. The tests for 63 above are 
 converted to use the new feature.
 
+66. In the POSIX wrapper, if regerror() was given too small a buffer, it could 
+misbehave.
+
 
 Version 10.20 30-June-2015
 --------------------------

doc/pcre2posix.3

@@ -1,4 +1,4 @@
-.TH PCRE2POSIX 3 "03 September 2015" "PCRE2 10.21"
+.TH PCRE2POSIX 3 "30 October 2015" "PCRE2 10.21"
 .SH NAME
 PCRE2 - Perl-compatible regular expressions (revised API)
 .SH "SYNOPSIS"
@@ -240,9 +240,11 @@ header file, of which REG_NOMATCH is the "expected" failure code.
 The \fBregerror()\fP function maps a non-zero errorcode from either
 \fBregcomp()\fP or \fBregexec()\fP to a printable message. If \fIpreg\fP is not
 NULL, the error should have arisen from the use of that structure. A message
-terminated by a binary zero is placed in \fIerrbuf\fP. The length of the
+terminated by a binary zero is placed in \fIerrbuf\fP. If the buffer is too
-message, including the zero, is limited to \fIerrbuf_size\fP. The yield of the
+short, only the first \fIerrbuf_size\fP - 1 characters of the error message are
-function is the size of buffer needed to hold the whole message.
+used. The yield of the function is the size of buffer needed to hold the whole
+message, including the terminating zero. This value is greater than
+\fIerrbuf_size\fP if the message was truncated.
 .
 .
 .SH MEMORY USAGE
@@ -267,6 +269,6 @@ Cambridge, England.
 .rs
 .sp
 .nf
-Last updated: 03 September 2015
+Last updated: 30 October 2015
 Copyright (c) 1997-2015 University of Cambridge.
 .fi

src/pcre2posix.c

@@ -144,29 +144,23 @@ static const char *const pstring[] = {
 PCRE2POSIX_EXP_DEFN size_t PCRE2_CALL_CONVENTION
 regerror(int errcode, const regex_t *preg, char *errbuf, size_t errbuf_size)
 {
-const char *message, *addmessage;
+int used;
-size_t length, addlength;
+const char *message;
 
 message = (errcode >= (int)(sizeof(pstring)/sizeof(char *)))?
   "unknown error code" : pstring[errcode];
-length = strlen(message) + 1;
 
-addmessage = " at offset ";
+if (preg != NULL && (int)preg->re_erroffset != -1)
-addlength = (preg != NULL && (int)preg->re_erroffset != -1)?
-  strlen(addmessage) + 6 : 0;
-
-if (errbuf_size > 0)
   {
-  if (addlength > 0 && errbuf_size >= length + addlength)
+  used = snprintf(errbuf, errbuf_size, "%s at offset %-6d", message, 
-    sprintf(errbuf, "%s%s%-6d", message, addmessage, (int)preg->re_erroffset);
+    (int)preg->re_erroffset);
-  else
+  }
-    {
+else
-    strncpy(errbuf, message, errbuf_size - 1);
+  {
-    errbuf[errbuf_size-1] = 0;
+  used = snprintf(errbuf, errbuf_size, "%s", message);
-    }
   }
   
-return length + addlength;
+return used + 1;      
 }
 
 

src/pcre2test.c

@@ -445,6 +445,7 @@ typedef struct patctl {    /* Structure for pattern modifiers. */
   uint32_t  jit;
   uint32_t  stackguard_test;
   uint32_t  tables_id;
+  uint32_t  regerror_buffsize; 
    uint8_t  locale[LOCALESIZE];
 } patctl;
 
@@ -566,6 +567,7 @@ static modstruct modlist[] = {
   { "ps",                  MOD_DAT,  MOD_OPT, PCRE2_PARTIAL_SOFT,        DO(options) },
   { "push",                MOD_PAT,  MOD_CTL, CTL_PUSH,                  PO(control) },
   { "recursion_limit",     MOD_CTM,  MOD_INT, 0,                         MO(recursion_limit) },
+  { "regerror_buffsize",   MOD_PAT,  MOD_INT, 0,                         PO(regerror_buffsize) },
   { "replace",             MOD_PND,  MOD_STR, REPLACE_MODSIZE,           PO(replacement) },
   { "stackguard",          MOD_PAT,  MOD_INT, 0,                         PO(stackguard_test) },
   { "startchar",           MOD_PND,  MOD_CTL, CTL_STARTCHAR,             PO(control) },
@@ -774,7 +776,7 @@ buffer is where all input lines are read. Its size is the same as pbuffer8.
 Pattern lines are always copied to pbuffer8 for use in callouts, even if they
 are actually compiled from pbuffer16 or pbuffer32. */
 
-static int       pbuffer8_size  = 50000;        /* Initial size, bytes */
+static size_t    pbuffer8_size  = 50000;        /* Initial size, bytes */
 static uint8_t  *pbuffer8 = NULL;
 static uint8_t  *buffer = NULL;
 
@@ -4575,8 +4577,21 @@ if ((pat_patctl.control & CTL_POSIX) != 0)
   rc = regcomp(&preg, (char *)pbuffer8, cflags);
   if (rc != 0)   /* Failure */
     {
-    (void)regerror(rc, &preg, (char *)pbuffer8, pbuffer8_size);
+    size_t bsize, usize;
+     
+    bsize = (pat_patctl.regerror_buffsize != 0)?
+      pat_patctl.regerror_buffsize : pbuffer8_size; 
+    if (bsize + 8 < pbuffer8_size)
+      memcpy(pbuffer8 + bsize, "DEADBEEF", 8);
+    usize = regerror(rc, &preg, (char *)pbuffer8, bsize);
+     
     fprintf(outfile, "Failed: POSIX code %d: %s\n", rc, pbuffer8);
+    if (usize > bsize)
+      {
+      fprintf(outfile, "** regerror() message truncated\n");
+      if (memcmp(pbuffer8 + bsize, "DEADBEEF", 8) != 0)
+        fprintf(outfile, "** regerror() buffer overflow\n");  
+      }    
     return PR_SKIP;
     }
   return PR_OK;

testdata/testinput18

@@ -94,4 +94,8 @@
 
 /abcd/substitute_extended
 
+/\[A]{1000000}**/expand,regerror_buffsize=31
+
+/\[A]{1000000}**/expand,regerror_buffsize=32
+
 # End of testdata/testinput18

testdata/testoutput18

@@ -143,4 +143,11 @@ Failed: POSIX code 3: pattern error at offset 2
 /abcd/substitute_extended
 ** Ignored with POSIX interface: substitute_extended
 
+/\[A]{1000000}**/expand,regerror_buffsize=31
+Failed: POSIX code 4: ? * + invalid at offset 100000
+** regerror() message truncated
+
+/\[A]{1000000}**/expand,regerror_buffsize=32
+Failed: POSIX code 4: ? * + invalid at offset 1000001
+
 # End of testdata/testinput18