Enable MMCFS by default on Mizar32

[Martin Guy]

This patch enables MMCFS on Mizar32 platform, since it (just!) fits
into the 120K bare-bones system.

   M

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[BogdanM]

It doesn't work for me (SVN version 881):

src/platform/avr32/crt0.o: warning: input is not relaxable
src/platform/avr32/trampoline.o: warning: input is not relaxable
src/platform/avr32/exception.o: warning: input is not relaxable
/usr/lib/gcc/avr32/4.3.2/../../../../avr32/bin/ld:
elua_lualong_at32uc3a0128.elf section .text will not fit in region
flash
/usr/lib/gcc/avr32/4.3.2/../../../../avr32/bin/ld: region flash
overflowed by 512 bytes

Recently there was a large merge of a development branch
(remotefs_int) to trunk, it's possible that the merge added some code
to the port (although this shouldn't have happened in theory).

Best,
Bogdan

On Sat, Jan 15, 2011 at 11:58 PM, Martin Guy <[hidden email]> wrote:

> This patch enables MMCFS on Mizar32 platform, since it (just!) fits
> into the 120K bare-bones system.
>
>   M
>
> _______________________________________________
> eLua-dev mailing list
> [hidden email]
> https://lists.berlios.de/mailman/listinfo/elua-dev
>
>

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[BogdanM]

I can confirm that I get the exact same message with SVN revision 880
(before the branch merge) so the problem is somewhere else.

Best,
Bogdn

On Sun, Jan 16, 2011 at 3:06 PM, Bogdan Marinescu
<[hidden email]> wrote:

> It doesn't work for me (SVN version 881):
>
> src/platform/avr32/crt0.o: warning: input is not relaxable
> src/platform/avr32/trampoline.o: warning: input is not relaxable
> src/platform/avr32/exception.o: warning: input is not relaxable
> /usr/lib/gcc/avr32/4.3.2/../../../../avr32/bin/ld:
> elua_lualong_at32uc3a0128.elf section .text will not fit in region
> flash
> /usr/lib/gcc/avr32/4.3.2/../../../../avr32/bin/ld: region flash
> overflowed by 512 bytes
>
> Recently there was a large merge of a development branch
> (remotefs_int) to trunk, it's possible that the merge added some code
> to the port (although this shouldn't have happened in theory).
>
> Best,
> Bogdan
>
> On Sat, Jan 15, 2011 at 11:58 PM, Martin Guy <[hidden email]> wrote:
>> This patch enables MMCFS on Mizar32 platform, since it (just!) fits
>> into the 120K bare-bones system.
>>
>>   M
>>
>> _______________________________________________
>> eLua-dev mailing list
>> [hidden email]
>> https://lists.berlios.de/mailman/listinfo/elua-dev
>>
>>
>

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[Martin Guy]

On 16 January 2011 14:06, Bogdan Marinescu <[hidden email]> wrote:
> It doesn't work for me (SVN version 881):
> elua_lualong_at32uc3a0128.elf section .text will not fit in region
> flash
> /usr/lib/gcc/avr32/4.3.2/../../../../avr32/bin/ld: region flash
> overflowed by 512 bytes
>
> Recently there was a large merge of a development branch
> (remotefs_int) to trunk, it's possible that the merge added some code
> to the port (although this shouldn't have happened in theory).

Strange, I updated it all to r880,and it (just) fit: 122132 bytes out
of a maximum of 122880.
Maybe we have different toolchains? I remember you said that adding a
compilation flag -fno-common gave you no code size reduction, while it
saves 1KB for me.  I'm building on Ubuntu with
scons board=mizar32 target=lualong allocator=newlib optram=0
and
martin@fon:~/mizar32/elua-svn$ avr32-gcc --version
avr32-gcc (atmel.1.2.0-(linux-i686)) 4.3.2
_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[Martin Guy]

On 16 January 2011 16:03, Martin Guy <[hidden email]> wrote:
> Maybe we have different toolchains? I remember you said that adding a
> compilation flag -fno-common gave you no code size reduction, while it
> saves 1KB for me.
(but no, I'm wasn't using it here, since I'm staying as close to svn as I can)

    M
_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[BogdanM]

> scons board=mizar32 target=lualong allocator=newlib optram=0

Ah OK, I completely forgout about "allocator=newlib" and "optram=0".
It works now. Interesting though:

> martin@fon:~/mizar32/elua-svn$ avr32-gcc --version
> avr32-gcc (atmel.1.2.0-(linux-i686)) 4.3.2

bogdanm@bogdanm-desktop:~/work/elua/r880$ avr32-gcc --version
avr32-gcc (-(Linux-amd64)) 4.3.2

Also, when using the build command above I get this:

  text   data    bss    dec    hex filename
 119296   1364   1440 122100  1dcf4 elua_lualong_at32uc3a0128.elf

Without "optram=0" I get:

   text   data    bss    dec    hex filename
 120320   1364   1440 123124  1e0f4 elua_lualong_at32uc3a0128.elf

(and obviously it still fits).

Best,
Bogdan
_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[BogdanM]

Patch commited to trunk.

On Sun, Jan 16, 2011 at 5:09 PM, Bogdan Marinescu
<[hidden email]> wrote:

>> scons board=mizar32 target=lualong allocator=newlib optram=0
>
> Ah OK, I completely forgout about "allocator=newlib" and "optram=0".
> It works now. Interesting though:
>
>> martin@fon:~/mizar32/elua-svn$ avr32-gcc --version
>> avr32-gcc (atmel.1.2.0-(linux-i686)) 4.3.2
>
> bogdanm@bogdanm-desktop:~/work/elua/r880$ avr32-gcc --version
> avr32-gcc (-(Linux-amd64)) 4.3.2
>
> Also, when using the build command above I get this:
>
>  text     data     bss     dec     hex filename
>  119296    1364    1440  122100   1dcf4 elua_lualong_at32uc3a0128.elf
>
> Without "optram=0" I get:
>
>   text    data     bss     dec     hex filename
>  120320    1364    1440  123124   1e0f4 elua_lualong_at32uc3a0128.elf
>
> (and obviously it still fits).
>
> Best,
> Bogdan
>

_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev

[Martin Guy]

On 16 January 2011 16:09, Bogdan Marinescu <[hidden email]> wrote:
>> scons board=mizar32 target=lualong allocator=newlib optram=0
>
> Also, when using the build command above I get this:
>
>  text     data     bss     dec     hex filename
>  119296    1364    1440  122100   1dcf4 elua_lualong_at32uc3a0128.elf

Idem:
   text   data    bss    dec    hex filename
 119296   1364   1440 122100  1dcf4 elua_lualong_at32uc3a0128.elf

with -fno-common in src/platform/avr32/conf.py -> CFLAGS I now get
exactly the same size as without, so i don't know what I did before.

    M
_______________________________________________
eLua-dev mailing list
[hidden email]
https://lists.berlios.de/mailman/listinfo/elua-dev