]> git.sur5r.net Git - openocd/commitdiff
SVF: all content between parentheses is one parameter
authorsimon qian <simonqian.openocd@gmail.com>
Sat, 23 Jan 2010 20:08:47 +0000 (04:08 +0800)
committerDavid Brownell <dbrownell@users.sourceforge.net>
Mon, 25 Jan 2010 21:06:46 +0000 (13:06 -0800)
More SVF fixes:

 * Treat all content between parentheses as part of the same
   parameter; don't (wrongly) treat whitespace as a delimiter.

 * Use isspace() to catch that whitespace; it's not all single
   spaces, newlines etc are also valid.

 * When parsing bitstrings, strip leading whitespace too.

So for example, these are equivalent and should (now) be OK:

  "TDI( 1234 )"
  "TDI( 1 2 3 4 )"
  "TDI(00 12 34 )"
  "TDI(
   00 12
34)"

[dbrownell@users.sourceforge.net: comment updates; trivial cleanup]

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
src/svf/svf.c

index 275bced8eca39355ffd8fea07eee3302fada9894..ea56a88aa4bd019a768bfb5a35217f2866b5cc16 100644 (file)
@@ -500,7 +500,9 @@ static int svf_read_command_from_file(int fd)
                case '\r':
                        slash = 0;
                        comment = 0;
-                       break;
+                       /* Don't save '\r' and '\n' if no data is parsed */
+                       if (!cmd_pos)
+                               break;
                default:
                        if (!comment)
                        {
@@ -565,25 +567,30 @@ static int svf_read_command_from_file(int fd)
 
 static int svf_parse_cmd_string(char *str, int len, char **argus, int *num_of_argu)
 {
-       int pos = 0, num = 0, space_found = 1;
+       int pos = 0, num = 0, space_found = 1, in_bracket = 0;
 
        while (pos < len)
        {
                switch (str[pos])
                {
-               case '\n':
-               case '\r':
                case '!':
                case '/':
                        LOG_ERROR("fail to parse svf command");
                        return ERROR_FAIL;
-                       break;
-               case ' ':
-                       space_found = 1;
-                       str[pos] = '\0';
-                       break;
+               case '(':
+                       in_bracket = 1;
+                       goto parse_char;
+               case ')':
+                       in_bracket = 0;
+                       goto parse_char;
                default:
-                       if (space_found)
+parse_char:
+                       if (!in_bracket && isspace(str[pos]))
+                       {
+                               space_found = 1;
+                               str[pos] = '\0';
+                       }
+                       else if (space_found)
                        {
                                argus[num++] = &str[pos];
                                space_found = 0;
@@ -651,6 +658,7 @@ static int svf_copy_hexstring_to_binary(char *str, uint8_t **bin, int orig_bit_l
                return ERROR_FAIL;
        }
 
+       /* fill from LSB (end of str) to MSB (beginning of str) */
        for (i = 0; i < str_hbyte_len; i++)
        {
                ch = 0;
@@ -658,7 +666,13 @@ static int svf_copy_hexstring_to_binary(char *str, uint8_t **bin, int orig_bit_l
                {
                        ch = str[--str_len];
 
-                       if (!isblank(ch))
+                       /* Skip whitespace.  The SVF specification (rev E) is
+                        * deficient in terms of basic lexical issues like
+                        * where whitespace is allowed.  Long bitstrings may
+                        * require line ends for correctness, since there is
+                        * a hard limit on line length.
+                        */
+                       if (!isspace(ch))
                        {
                                if ((ch >= '0') && (ch <= '9'))
                                {
@@ -694,11 +708,12 @@ static int svf_copy_hexstring_to_binary(char *str, uint8_t **bin, int orig_bit_l
                }
        }
 
-       // consume optional leading '0' characters
-       while (str_len > 0 && str[str_len - 1] == '0')
+       /* consume optional leading '0' MSBs or whitespace */
+       while (str_len > 0 && ((str[str_len - 1] == '0')
+                               || isspace(str[str_len - 1])))
                str_len--;
 
-       // check valid
+       /* check validity: we must have consumed everything */
        if (str_len > 0 || (ch & ~((2 << ((bit_len - 1) % 4)) - 1)) != 0)
        {
                LOG_ERROR("value execeeds length");