[Mspgcc-users] [msp430] Re: 1611 RAM overlapping?

[Matthias W. <ms...@pe...>]

Some news from real samples. The RAM in the big RAM chips is overlapping.
Read below...

M.

----- Forwarded message from michelqv <michel@q.........> -----

Date: Sat, 28 Feb 2004 17:15:59 -0000
From: "michelqv" <michel@q...........>
Reply-To: ms...@ya...
To: ms...@ya...
Subject: [msp430] Re: 1611 RAM overlapping?

Indeed. Now that I finally got my hands on a 1611 sample, I 
understand the TI "documentation" relating to the RAM map for F161x.
The lower RAM is indeed mapped into the upper RAM area, so that the 
actual RAM usable by the C compiler really starts at 0x1100, and is 
contiguous.
For the 1610, this means:
RAM: 0x1100 to 0x24FF, with the first 2K bytes mirrored at 0x200 
through 0x9FF
For the 1611:
RAM 0x1100 to 0x38FF, with the first 2K bytes mirrored at 0x200 
through 0x9FF

Michel
--- In ms...@ya..., onestone <onestone@b...> wrote:
> Paul Curtis wrote:
> > The first 2K of the big RAM bank is mirrored into the 2K of standard RAM
> > at 0200 on each device.  This is the most sensible choice that could
> > have been made.
> 
> Why is this sensible? since the memory addressing limit of 64k has not 
> been breached why wouldn't Ti use both RAM areas to give 12K. It would 
> make more sense to have the 2K module and an 8k module, since they 
> appear to be stuck with data flash at 1100H.
> 
> Al
> 
> > 
> > I've no idea what the TI DLL tells you, it seems to be wrong and have
> > gone through various incarnations.  We don't need to use it, luckily.
> > 
> > Regards,
> > 
> > -- Paul.
> > 
> > 
> >>-----Original Message-----
> >>From: michelqv [mailto:michel@q...]
> >>Sent: 27 February 2004 18:44
> >>To: ms...@ya...
> >>Subject: [msp430] Re: 1611 RAM overlapping?
> >>
> >>
> >>I asked TI Germany the same question this morning , because their 
> >>newly released DLL gives this 0x38FF number also, which seems 
wrong. 
> >>(Should be 30FF?)
> >>All 161x chips have a "first RAM area" between 0x200 and 0x9FF, 
and a 
> >>second memory area starting at 0x1100 to complete the RAM.
> >>For the 1612, the dll says that the second RAM area is from 
0x1100 to 
> >>0x24FF,
> >> and the main memory starts at 0x2500.
> >>
> >>We will release a new version of AQ430 early next week to support 
the 
> >>new chips, but we need some clarification on this memory size 
first.
> >>
> >>Michel
> >>
> >>
> >>--- In ms...@ya..., Matthias Weingart <msp430@p...> 
wrote:
> >>
> >>>Due to the datasheet, there are actually more than 10k RAM in 
the 
> >>
> >>1611
> >>
> >>>address space:
> >>>
> >>>0000h-000fH SFR
> >>>0010h-00ffH 8bit per.
> >>>0100h-01ffh 16bit per.
> >>>0200h-09ffh RAM 2k  <-############
> >>>0A00h-0bffh gap  512 bytes
> >>>0C00h-0fffh ROM 1k
> >>>1000h-10ffh flash 256bytes
> >>>1100h-38ffh RAM 10k <-############
> >>>3900h-3fffh gap  1792 bytes
> >>>4000h-ffffh flash
> >>>
> >>>Does the range at 200h map into the RAM at 1100h? How? The 
> >>
> >>datasheet is not
> >>
> >>>clear (it states that 1100h-38ffh is 8k, what is obviously 
wrong).
> >>>
> >>>How is the address space in the 1612 assigned?
> >>>
> >>>        Matthias
> >>