There are two new functions: get_valid_toolchain_names and
find_valid_toolchain. These functions are used to figure out if a
fallback is possible and necessary. find_valid_toolchain is expected to
be used by the front-end scripts.
get_toolchain_name was updated with some different logic and comments.
On Windows 7 using --preproc option in ARMC6 assembler doesn't work
when -MD option is also specified. Compiler creates incorrect filename for
dependency file and compilation files.
To workaround this issue, this change returns to using a temporary file and
separately calling preprocessor and assembler (as in a case of ARMC5).
The "hook tools" were capable of hooking into many commands run by the build system.
To my knowlage, the only hook is the "post-build-hook". The post build hook could be
easier to reason about if the implementation is specialized for just post-build
hooking.
This commit make it much easier to point out where post build hooks are called by
making the call explicit.
Below are the options read from the toolchains/arm
armclang --target=arm-arm-none-eabi -mcpu=list
The following arguments to option 'mcpu' can be selected:
-mcpu=cortex-m0
-mcpu=cortex-m0plus
-mcpu=cortex-m1
-mcpu=cortex-m3
-mcpu=cortex-m4
-mcpu=cortex-m7
-mcpu=cortex-m23
-mcpu=cortex-m33
...
armlink --cpu=list
The following arguments to option 'cpu' can be selected:
--cpu=Cortex-M0
--cpu=Cortex-M0plus
--cpu=Cortex-M1
--cpu=Cortex-M1.os_extension
--cpu=Cortex-M1.no_os_extension
--cpu=Cortex-M4
--cpu=Cortex-M4.no_fp
--cpu=Cortex-M7
--cpu=Cortex-M7.fp.sp
--cpu=Cortex-M7.no_fp
--cpu=Cortex-M23
--cpu=Cortex-M33
--cpu=Cortex-M33.no_fp
--cpu=Cortex-M33.no_dsp
--cpu=Cortex-M33.no_dsp.no_fp
...
armclang --target=arm-arm-none-eabi -mfpu=list
The following arguments to option 'mfpu' can be selected:
-mfpu=fpv4-sp-d16
-mfpu=fpv5-sp-d16
-mfpu=fpv5-d16
...
For binary compatibility with ARMC5, use the hard ABI variant whenever
we have FP hardware - this is ARMC5's default behaviour, which we do not
override.
Cortex-M4F was already using hard; this brings M7F and M33F into line.
The 'archive' method of the toolchain class 'ARM' expects that self.ar
is a string, but the constructor of the class 'ARMC6' initializes it
with a list. This patch fixes the issue.
### Description
The prior fix made the assumption that you wanted to compute all of the
parents for a give header file going all the way up the path. This is
not true: you probably want to stop when the project stops. We already
keep track of a virtual name within the project, so instead, we compute
parents of the name, and generate the actual location of these files in
your FS as the path. This makes the solution robust offline and online
(I tested it with my local copy of os.mbed.com)
### Pull request type
[x] Fix
[ ] Refactor
[ ] Target update
[ ] Functionality change
[ ] Docs update
[ ] Test update
[ ] Breaking change
### Description
I noticed that there was quite a bit missing from the mbed2 release of
the microbit today. That's because the microbit uses ARM by default and
the GCC_ARM small library. The release script was trying to release for
ARM using scan rules for uARM. Turns out we're stuck with 2 configuration
parameters for the same thing: `default_lib` for GCC and `default_toolchain`
for ARM. Dang
### Pull request type
[x] Fix
[ ] Refactor
[ ] Target update
[ ] Functionality change
[ ] Breaking change
-mcpu option if set for v8M CPU;s it will add DSP feature as default
which is optional. Hence setting just the architecture for Cortex-M23
and Cortex-M33
Cruz Monrreal
Brian Daniels
Leszek Rusinowicz
Jimmy Brisson
Senthil Ramakrishnan
Oren Cohen
Michael Schwarcz
Martin Kojtal
Deepika
Kevin Bracey
Mikhail Maltsev
jeromecoutant
Jammu Kekkonen