- } else {
- // no room for more
+ } else if (3 < (s_end - sentences)) {
+ // no room for more, but room for '...'
*--s_end = '.';
*--s_end = '.';
> I've looked at that code a lot. Seems solid to me. What do you not like?
It's hard to understand and it doesn't fix the warning. I would revert
that change and add a comment saying the gcc 10 gives a bogus warnings.
Try to figure out what that new code is doing without knowing that it is
trying to dodge a warning.
The old code had a catch-all else at the end. The new code doesn't. What
should happen if the new else fails?
--------
I see a way that the warning gets close to real. Suppose sentences is
empty and fields[0] is big so it doesn't fit. s_end will point to the
beginning of sentences so backing up is evil.
But field0_len comes from:
size_t field0_len = strnlen(fields[0], NMEA_MAX_FLD);
include/gpsd.h:#define NMEA_MAX_FLD 100 // max fields in an
NMEA sentence
include/gpsd.h: char *field[NMEA_MAX_FLD];
I can't figure that out. Is that max fields or max field length?
Assming max field length, then it will fit because 100<132
static char sentences[132];
If the problem is that the compiler doesn't figure out that the fields[0]
will always fit, that would explain why your patch didn't fix the warning.
--------
if ((s_len + field0_len + 2) < max) {
// room for more
*s_end++ = ' ';
*s_end = '\0';
(void)strlcpy(s_end, fields[0], field0_len + 2);
Is that +2 on the strlcpy corrent? You just bumped s_end by one to insert
the space.
--
These are my opinions. I hate spam.
