ARM assembly cannot use immediate values and ADDS/ADCS together

Question

I am currently trying to speed up some of my C functions on a Cortex-M0 (Freescale KL25Z) using assembly. I get a problem with this minimal test program:

@.syntax unified
.cpu cortex-m0
.text
  .global test
  .code 16
test:
  mov r0, #0
  adds r0, r0, #1
  bx lr

When I try to assemble my .s file to a .o file, I get this error

$ arm-none-eabi-as test.s -o test.o
test.s: Assembler messages:
test.s:8: Error: instruction not supported in Thumb16 mode -- `adds r0,r0,#1'

The error message doesn't make sense to me, ADDS is a valid instruction acording to this document. I found a possible answer on stackoverflow and added the line at the beginning of the program (".syntax unified" worked, like the second answer suggested). This resulted in fixing this problem, I can now use instructions like ADDS and ADCS, but I did get a new error:

$ arm-none-eabi-as test.s -o test.o
test.s: Assembler messages:
test.s:7: Error: cannot honor width suffix -- `mov r0,#0'

Some instructions using immediate values get this error. I am compiling on Mac OS 10.9.5. I can't find a solution to this via Google or Stackoverflow and have no idea how to get around these errors.

$ arm-none-eabi-gcc --version
arm-none-eabi-gcc (GNU Tools for ARM Embedded Processors) 4.9.3 20150303 (release) [ARM/embedded-4_9-branch revision 221220]
Copyright (C) 2014 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

$ arm-none-eabi-as --version
GNU assembler (GNU Tools for ARM Embedded Processors) 2.24.0.20150304
Copyright 2013 Free Software Foundation, Inc.
This program is free software; you may redistribute it under the terms of
the GNU General Public License version 3 or later.
This program has absolutely no warranty.
This assembler was configured for a target of `arm-none-eabi'.

Answer 1

what Jester said:

.cpu cortex-m0
.text
  .thumb

  .thumb_func
  .global test
test:
  mov r0, #0
  add r0, r0, #1
  bx lr

which gives

   0:   2000        movs    r0, #0
   2:   3001        adds    r0, #1
   4:   4770        bx  lr

or if you go with syntax unified then you have to put the s on there

.syntax unified
.cpu cortex-m0
.text
  .thumb

  .thumb_func
  .global test
test:
  movs r0, #0
  adds r0, r0, #1
  bx lr

which also gives

00000000 <test>:
   0:   2000        movs    r0, #0
   2:   3001        adds    r0, #1
   4:   4770        bx  lr

Answer 2

Somewhat ironically, by using UAL syntax to solve the first problem you've now hit pretty much the same thing, but the other way round and with a rather more cryptic symptom.

The only Thumb encodings for (non-flag-setting) mov with an immediate operand are 32-bit ones, however Cortex-M0 doesn't support those, so the assembler ends up choking on its own constraints. In UAL you have to explicitly use movs to get the only "move immediate" instruction Cortex-M0 actually has.

Answer 3

In thumb mode, you can not use adds, you only have add. As such, the correct code is:

.cpu cortex-m0
.text
  .global test
  .code 16
test:
  mov r0, #0
  add r0, r0, #1
  bx lr