Posted on July 29, 2014 at 16:27Hi,
I would like to share some knowledge about a point that puzzled me for 3 days.
In Keil µVision (C51), if you write the following:
char *ptr = (char *)0x8000;
for accessing data located in some Flash memory for example... You won't get the expected results: the pointer will read elsewhere.
Actually the pointer does not hold on 2 bytes as expected, but on 3 (generic pointer), however the explicit cast does not provide you the 3rd byte necessary to access XDATA (for example).
Some pointers hold on 1 byte: for accessing IDATA.
In such cases, advice is to _always_ specify which address space it is pointing to:
char xdata *ptr = (char xdata *)0x8000;
so that the generated assembly code makes use of the appropriate opcodes (DPTR in the case of XDATA). This pointer is really 2-byte long and will allow you to read/write at the right location.
Note that any function prototype taking a pointer as argument is concerned with this issue.
If the destination is not known at compile-time (because of a generic function), use the keyword ''generic'' (but don't try to cast an integer to a generic pointer unless you really know how it will be decoded).
_____
References:
http://www.keil.com/support/man/docs/c51/c51_le_ptrs.htmhttp://www.keil.com/support/man/docs/c51/c51_le_memspecificptrs.htmhttp://www.keil.com/support/docs/1012.htm
