2016-06-07 19:11 GMT+05:00 Gert Doering <g...@greenie.muc.de>: > Hi, > > On Tue, Jun 07, 2016 at 06:04:54PM +0500, ???????? ?????????????? wrote: > > as I see, there's call to format_hex_ex with separator=NULL here: > > Interesting find. > > This code is funny - format_hex_ex() is called from various places with > separator=NULL, and has been that way since at least 2005... it's part of > the oldest commit we have in git. > > Now, that expression > > struct buffer out = alloc_buf_gc (maxoutput ? maxoutput : > ((size * 2) + (size / > (space_break_flags & F > HE_SPACE_BREAK_MASK)) * (int) strlen (separator) + 2), > gc); > > will of course *only* call strlen(separator) if "maxoutput" is 0 > (C short-circuit evaluation of conditionals). So if all callers respect > that and *either* set "maxoutput > 0" *or* pass "separator != NULL", we're > fine. > > The line 152 in ssl_openssl.c is passing in "len", which is not 0... > > const char *key = format_hex_ex (ekm, size, len, 0, NULL, &gc); > > ... so there will not be a strlen() call. > > > I wouldn't mind an ASSERT( maxoutput > 0 || separator != NULL ) call > in format_hex_ex(), though. >
ASSERT is more readable that ternary operator here. ok, I'll propose new patch > > gert > > -- > USENET is *not* the non-clickable part of WWW! > // > www.muc.de/~gert/ > Gert Doering - Munich, Germany > g...@greenie.muc.de > fax: +49-89-35655025 > g...@net.informatik.tu-muenchen.de >
