I'm not sure if this is a bug or just questionable coding, so I'll
ask here before reporting it as a bug. I'm trying to figure out why someone
elses code tha works on snaps less than 5559 but not after 5560 is broken (so
far without success :-)). While poking at that I came across this change where
the same code sometimes gets a warning and sometimes doesn't:

$ cat test4.c
#include <w7100.h>

typedef unsigned char uint8;
typedef volatile unsigned char vuint8;

#define COMMON_BASE 0x0000
#define UIPR0               ((volatile __xdata uint8 *)(COMMON_BASE + 0x002A))


// required for IINCHIP_WRITE and IINCHIP_READ port for SDCC to
// avoid compiler use of non mapped memory access while banking

static __data uint8 w_d;

uint8 IINCHIP_READ(__xdata vuint8 *addr) __reentrant __critical
{
        w_d = *((__xdata vuint8 *)(addr));
        return w_d;
}

main()
{
        static __xdata uint8 invalid_ip[4];

        invalid_ip[0] = IINCHIP_READ(UIPR0+1);
        return (0);
}

        This compiler is snap 5559 compiled from source

pe...@pc700 ~/wiznet/nos_test
$ sdcc -v
SDCC : mcs51/gbz80/z80/avr/ds390/pic16/pic14/TININative/xa51/ds400/hc08 2.9.4 #
(Dec 31 2010) (CYGWIN)

pe...@pc700 ~/wiznet/nos_test
$ /usr/local//bin/sdcc -c test4.c

pe...@pc700 ~/wiznet/nos_test
$  

Note it doesn't complain about the conversion without a cast in the line

invalid_ip[0] = IINCHIP_READ(UIPR0+1);

where UIPR0 is in fact a pointer to an unsigned char and thus is doing pointer
arithmatic (nor am I a good enough C coder to know if it should be complaining
or not :-)). 

nor does a recent 3_0_1 
snapshot 

pe...@pc700 ~/wiznet/nos_test
$ "/cygdrive/c/program files/SDCC3_0_1"/bin/sdcc -v
SDCC : mcs51/gbz80/z80/ds390/pic16/pic14/TININative/ds400/hc08 3.0.1 #6066 (Nov
26 2010) (MINGW32)

pe...@pc700 ~/wiznet/nos_test
$ "/cygdrive/c/program files/SDCC3_0_1"/bin/sdcc test4.c

pe...@pc700 ~/wiznet/nos_test

but snap 5560 (and I think some later ones, I've been binary searching for the
failure point through previous snaps :-)) issues a warning about the lack of 
a cast. 

This compiler is snap 5560 compiled from source (which also breaks the full code
for reasons as yet unknown) which does complain about the lack of a cast.

pe...@pc700 ~/wiznet/nos_test
$ /usr/local//bin/sdcc -c test4.c
test4.c:25: warning 154: converting integral to pointer without a cast
from type 'const-unsigned-char literal'
to type 'volatile-unsigned-char xdata* auto'

pe...@pc700 ~/wiznet/nos_test
$ sdcc -v
SDCC : mcs51/gbz80/z80/avr/ds390/pic16/pic14/TININative/xa51/ds400/hc08 2.9.4 #
(Dec 30 2010) (CYGWIN)

        Comments on a reportable bug or operator error in the form of poor 
coding practice (the original code from Wiznet that this is based on is full 
of poor coding practices :-))?

Peter Van Epp

------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl
_______________________________________________
Sdcc-user mailing list
Sdcc-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sdcc-user

Reply via email to