Re: [avr-gcc-list] The "Temporary Buffer" (back to the key qeustion)

[Christoph Plattner]

Back to the key question !

Ken wanted to know, where this buffer space is, and if he must be
careful
not to overwrite is. My idea was (I don't know it), that this memory is
an
additional isolated one, not beeing used for other stuff.

So try to find out that fact for Ken.

With friendly regards
Christoph P.


"Theodore A. Roth" wrote:
> 
> This is mostly true. It depends on whether Atmel based the flash core on
> the 29xxx or the 49xxx flash devices.
> 
> For the 29xxx, to program a page, you write a special data/address
> sequence and then clock in a pages worth data (and you must give all data
> for the page). The data is stored in temporary data buffer the size of the
> page and the entire page is written to the flash at once. If I remember
> rightly, the buffer is implemented with sram. Erasing of the page in flash
> is handled by internal logic when you enter programming mode via the
> special sequence.
> 
> For the 49xxx, you must first erase the page, then you can write a byte at
> a time. You still need to give a special sequence for each byte and the
> erase.
> 
> If you need to know more, check the data sheets for the 29lv512 and the
> 49bv512 and compare.
> 
> As for eeprom, you write a byte, wait, write a byte, wait, etc.
> 
> The megas hide the special sequence stuff from you via the spm
> instruction.
> 
> Hope this helps.
> 
> Ted Roth
> 
> On Tue, 4 Dec 2001, Christoph Plattner wrote:
> 
> :)Hello !
> :)
> :)I don't know anything about the architecture of the write
> :)algorism and the SPM structure, but I really think, that
> :)such a buffer is implemented as a separate area in the
> :)chip, not colliding to any part of RAM or ROM. But this
> :)depends on the size of it.
> :)Dependent on the page-size (sector size), there a flash
> :)(eeprom) chips, where the user has to write without
> :)interruption bytes in the programming algorithm for one
> :)sector. Perhaps the buffer has the size of such one sector ...
> :)
> :)Sorry, that the answer is so weak here.
> :)
> :)Bye
> :)Christoph P.
> :)
> :)ken restivo wrote:
> :)>
> :)> -----BEGIN PGP SIGNED MESSAGE-----
> :)> Hash: SHA1
> :)>
> :)> I am writing a little boot loader for an ATmega163. I have
> :)> implemented a modified version of the page write routine from the data
> :)> book, in assembler. So far so good.
> :)>
> :)> I am concerned about the mention in the databook of a "temporary
> :)> buffer". The docs claim that the SPM instruction reads and writes a
> :)> temporary buffer.  However, a search through the docs for the text
> :)> "temporary buffer" didn't turn up any description of *what* exactly
> :)> this "temporary buffer" is, or where it is located.
> :)>
> :)> I have also read the AVR109 app note and looked through Atmel's
> :)> sample bootloader source code as well, with no answers found there.
> :)>
> :)> This worries me. For example, I may want to load more than one page
> :)> at a time of app code into RAM from my serial port, do a crc check,
> :)> etc, and only then start iteratively calling the page write routine,
> :)> but I don't want to stomp on anything inadvertently.
> :)>
> :)> Any ideas what/where this "temporary buffer" actually is so I can
> :)> stay out of its way?
> :)>
> 
> _______________________________________________
> avr-gcc-list mailing list
> address@hidden
> http://avr.jpk.co.nz/mailman/listinfo/avr-gcc-list

-- 
-------------------------------------------------------
private:        address@hidden
company:        address@hidden