Datum: Wed, 07 Mar 2012 08:50:20 +0100
Von: Laurent Sallafranque<laurent.sallafranque@xxxxxxx>
I've compiled with the patch you suggest.
Compile runs with 2 warnings:
[ 21%] Building C object src/cpu/CMakeFiles/UaeCpu.dir/fpp.c.o
/home/laurent/Atari/hatari/src/cpu/fpp.c: In function ‘to_pack’:
/home/laurent/Atari/hatari/src/cpu/fpp.c:339:2: attention : format
‘%Le’
expects argument of type ‘long double *’, but argument 3 has type
‘fptype *’ [-Wformat]
/home/laurent/Atari/hatari/src/cpu/fpp.c: In function ‘from_pack’:
/home/laurent/Atari/hatari/src/cpu/fpp.c:350:2: attention : format
‘%Le’
expects argument of type ‘long double’, but argument 3 has type
‘fptype’
[-Wformat]
Ah, sorry, my bad, I was working with the "previous" sources when I wrote my mail. Actually that problem only happens when using USE_LONG_DOUBLE=1. This is the case for previous. Since we're using USE_LONG_DOUBLE=0 in Hatari, we are not affected of this format string problem. But to be sure, we should maybe add a patch like this:
diff -r 22d3c9f30097 src/cpu/fpp.c
--- a/src/cpu/fpp.c Wed Mar 07 10:08:50 2012 +0100
+++ b/src/cpu/fpp.c Wed Mar 07 18:50:43 2012 +0100
@@ -336,7 +336,11 @@
*cp++ = ((wrd1>> 20)& 0xf) + '0';
*cp++ = ((wrd1>> 16)& 0xf) + '0';
*cp = 0;
+#if USE_LONG_DOUBLE
+ sscanf (str, "%Le",&d);
+#else
sscanf (str, "%le",&d);
+#endif
return d;
}
@@ -347,7 +351,11 @@
char *cp;
char str[100];
+#if USE_LONG_DOUBLE
+ sprintf (str, "%.16Le", src);
+#else
sprintf (str, "%.16e", src);
+#endif
cp = str;
*wrd1 = *wrd2 = *wrd3 = 0;
if (*cp == '-') {
What do you think?
Thomas