Build break in xdm.

[mcnichol at austin.ibm.com]

> From: Alan Coopersmith <Alan.Coopersmith at Sun.COM>
> 
> mcnichol at austin.ibm.com wrote:
> > I noticed that also.  This is looking pretty strange right now.
> > 
> > While I'm poking around my compiler flagged a couple of other problems:
> > 
> > 
> >         xlc_r -O -D__STR31__ -DNDEBUG      -I../.. -I../../exports/include   -DSYSV -DAIXV3 -DAIXV4 -DAIXV5 -D_ALL_SOURCE  -DFUNCPROTO=15   -DBINDIR=\"/xorg/X1132/bin\" -DXDMDIR=\"/xorg/X1132/lib/X11/xdm\"      -DHASXDMAUTH     -DUSE_PAM -DTCPCONN -DUNIXCONN   -DHAS_STICKY_DIR_BIT -DHAS_FCHOWN -DIPv6  -DGREET_USER_STATIC      -DOSMAJORVERSION=5  -DOSMINORVERSION=3     -DXPM        -c access.c
> > "access.c", line 260.38: 1506-1298 (W) The subscript 256 is out of range. The valid range is 0 to 255.
> 
> That code should be safe on most machines, since it's checking that the 
> subscript is in range:
> 
>      204 #define WORD_LEN    256
>      205 static char     wordBuffer[WORD_LEN];
> [...]
>      260             if (wordp < &(wordBuffer[WORD_LEN]))
>      261               *wordp++ = c;
> 
> I suppose 'if (wordp <= &(wordBuffer[WORD_LEN - 1]))' would be just as effective
> without flagging compiler warnings.
> 

Hmmm...  That does look like it should be valid.

I'll have to send a bug report into our compiler guys.

Dan