RE: [avr-gcc-list] Re: "progmem_far" attribute

[Weddington, Eric]

Hi Anatoly,

Patches look good so far.

Is <stdbool.h> included in avr.c? Should it be for the bool type, and
'true' and 'false'?

FYI, it would be nice if you could use the -p switch in producing
patches. This prints out the C function that the hunk belongs to. It
helps in reviewing patches.

Thanks,
Eric Weddington

 

> -----Original Message-----
> From: 
> address@hidden 
> [mailto:address@hidden
> org] On Behalf Of Anatoly Sokolov
> Sent: Sunday, July 20, 2008 9:41 AM
> To: Anatoly Sokolov; avr-GCC
> Subject: [avr-gcc-list] Re: "progmem_far" attribute
> 
> Hi.
> 
> The patch add "progmem_far" attribute in GCC.
> 
> Anatoly.
> 
> ----- Original Message ----- 
> From: "Anatoly Sokolov" <address@hidden>
> To: "avr-GCC" <address@hidden>
> Cc: "Anatoliy Sokolov" <address@hidden>
> Sent: Friday, May 02, 2008 8:36 PM
> Subject: "progmem_far" attribute
> 
> 
> > Hello.
> > 
> >  Now GCC have "progmem" attribute to place data in the 
> '.progmem' section.
> > The 'avr-libc' have set of '*_P' string functions for 
> access to "progmem"
> > data. But pointer operand in  '*_P' functions is 16-bit 
> size and hence size
> > of data in  '.progmem' section it is limited by about 64KB.
> > 
> >  This restriction was rare a source of problem for 128K 
> devices but is very
> > hard for 256K and up devices.
> > 
> >  I suggest to add "progmem_far" attribute, and place data 
> defined with it
> > attribute in the new '.progmem_far' section place it in end 
> of '.text'
> > section after ".fini0" (It is some section which should be 
> located in low
> > 128K ".trampolines*" and ".lowtext*"). For access to data 
> in this section
> > to use set of  '*_PF' string functions (patch #6352: Far 
> pointer library).
> > 
> >  Now the programmer can define often used data with 
> "progmem" attribute and
> > define seldom used data with  "progmem_far" attribute.
> > 
> >  Yes I understand what to have two similar sets of 
> attributes and functions
> > is error prone, but it is necessary to allow to the 
> programmer to store more
> > them 64KB data in program memory, not forcing it to write 
> the linker script.
> > 
> > For 'xmega' devices it is necessary to add  '.apptable' 
> section which is
> > located in the top 8(4)KB of FLASH.
> > 
> > What your opinion on this offer? 
> > 
> > Anatoly.
> >
>