New Atmel toolchain 3.5.0 available, featuring gcc 4.9.2 and avrdude 6.1

[Federico Fissore]

Hi all

I've just published a test AVR core 9.6.8 with gcc 4.9.2 and avrdude 6.1

You can test it by adding this url to Boards Manager pref
http://downloads.arduino.cc/packages/package_3.5.0_index.json

I've compiled three random sketches: Blink, GSM->ReceiveVoiceCall and
Bridge->Bridge

Results are not encouraging. New toolchain makes sketches slightly
bigger: from ~2% bigger for Bridge and GSM to ~11% bigger for Blink. See
attachment

Andrew, you opened this issue [1] an year ago: may you give this
toolchain a spin and see if it, at least, fixes it?

Regards

Federico

[1] https://github.com/arduino/Arduino/issues/2413

[Andrew Kroll]

I'll be able to test sometime today, and I'll let you know.

--
You received this message because you are subscribed to the Google Groups "Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to developers+...@arduino.cc.

[Federico Fissore]

Andrew Kroll ha scritto il 07/10/2015 alle 09:42:
> I'll be able to test sometime today, and I'll let you know.
>

Thank you in advance :)

Federico

[Mikael Patel]

Nice work!

Does this include the LTO plugin? I would like to test it with the Cosa build script (which uses LTO). There seems to be changes to the ar(chive) handling of LTO compiled files. I guess that a new setup is needed.

Cheers! Mikael

[Federico Fissore]

I'm not expert with LTO, but LTO.so file is included with GCC
Btw, this is true for current toolchain as well

Federico

[Mikael Patel]

Thanks! I found the issue in my build script; 1) option -fuse-linker-plugin needed for the linker, 2) avr-ar should be avr-gcc-ar (to pass necessary LTO plugin info). The results for Cosa looks very promising. Compared to AVR-GCC-4.8.1 with LTO enabled it uses less SRAM and PROGMEM. I will get back with more details. Cheers!

[Mikael Patel]

Found one issue when building for ATtiny85. The power reduction management register functions, e.g. power_timer0_enable(), generate a linker error (LTO). Below is the very cryptic error message.

/tmp/ccRaqJQG.s: Assembler messages:
/tmp/ccRaqJQG.s:829: Error: number must be positive and less than 32

Cheers!

Den onsdag 7 oktober 2015 kl. 09:33:25 UTC+2 skrev f.fissore:

[Andrew Kroll]

Still fails. :-(

In file included from /root/Arduino/libraries/UHS_FS/UHS_FS.h:146:0,
                 from UHS_FS_NEW_DEMO.ino:165:
/root/Arduino/libraries/UHS_FS/FAT/FatFS/src/ff.c: In function 'FRESULT f_write(FIL*, const void*, UINT, UINT*)':
/root/Arduino/libraries/UHS_FS/FAT/FatFS/src/ff.c:2883:1: error: unable to find a register to spill in class 'POINTER_REGS'
 }
 ^
/root/Arduino/libraries/UHS_FS/FAT/FatFS/src/ff.c:2883:1: error: this is the insn:
(insn 202 201 203 32 (set (reg:HI 26 r26)
        (reg/f:HI 194)) /root/Arduino/libraries/UHS_FS/FAT/FatFS/src/ff.c:2840 83 {*movhi}
     (nil))
/root/Arduino/libraries/UHS_FS/FAT/FatFS/src/ff.c:2883: confused by earlier errors, bailing out

--
You received this message because you are subscribed to the Google Groups "Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to developers+...@arduino.cc.

--

Visit my github for awesome Arduino code @ https://github.com/xxxajk

[Andrew Kroll]

Oh and yeah, here's your example code to make it fail:
https://github.com/felis/UHS30

examples -> UHS_FS -> USB_HOST_SHIELD -> UHS_FS_NEW_DEMO

I can fix this by tagging the compiler version as bad (see report and sources for my fix) but it would be best to simply nail this bug.

[Federico Fissore]

I've received an email from Atmel, notifying me that the toolchain was
updated

"One of the fixes is to address 'out of range' error reported in
following forums.
https://groups.google.com/a/arduino.cc/forum/m/#!topic/developers/ATDTccou4cM
http://www.avrfreaks.net/forum/operand-out-range-error-when-using-prr"

I've updated the json file for boards manager, which will now download
gcc 4.9.2-arduino2 and avrdude 6-1-arduino2

Regards

Federico