[Mspgcc-users] New MSP430 GCC version release available!

[Ghannouchi, Youssef]

Hello GCC-enthusiasts,
We are pleased to announce a new stand-alone & open source release (v3.5.0.0) of MSP430 GCC available for download here<http://software-dl.ti.com/msp430/msp430_public_sw/mcu/msp430/MSPGCC/3_05_00_00/index_FDS.html>!
The stand-alone package includes all binaries, the source code and MSP430 device specific support files.
The MSP430 GCC build is also available in CCSv6 App Center on all platforms (Windows, Linux and Mac OSX).


· Highlights

o MSP430 GCC compiler latest build (14r1-364)

o MSP430 GCC for Mac OS X (beta release)

o A new MSP430 GCC User Guide



· Package Content

o Full Packages

§ msp430-gcc-full-linux-installer-3.5.0.0.run: MSP430 GCC Linux installer incl. support files and debug stack and USB drivers - apply chmod x before executing the package

§ msp430-gcc-full-osx-installer-3.5.0.0-beta.zip: MSP430 GCC Mac OS X installer incl. support files and debug stack and USB drivers - beta release

§ msp430-gcc-full-windows-installer-3.5.0.0.exe: MSP430 GCC Windows installer incl. support files and debug stack and USB drivers

o Compiler Only Packages

§ msp430-gcc-linux-installer-3.5.0.0.run: MSP430 GCC Linux installer - compiler only - apply chmod x before executing the package

§ msp430-gcc-osx-installer-3.5.0.0-beta.zip: MSP430 GCC Mac OS X installer - compiler only - beta release

§ msp430-gcc-windows-installer-3.5.0.0.exe: MSP430 GCC Windows installer - compiler only

o MSP430 Support Files

§ msp430-gcc-support-files.zip: MSP430 Header and linker Files

o MSP430 GCC Sources files

§ msp430-gcc-source.tar.bz2: MSP430 GCC source files



· What's New

o Key Changes to MSP430 GCC

§ Updated GCC compiler (14r1-364)

· Code to disable watchdog timer has been removed from start-up sequence (See examples).

· Fixes for Hardware Multiplier issues for F5 series (Details<https://e2e.ti.com/support/development_tools/compiler/f/343/t/417242#pi317851=2>).

· Please refer to the changelogs.txt in the install directory for more details about all changes vs. the previous build 14r1-167.

§ MSP430 GCC for Mac OS X (beta release)

§ Updated GDB agent (6.0.14.5)

· Added support for following GDB monitor commands:

o 'monitor reset' - reset the CPU

o 'monitor halt' - halt the CPU

o MSP Debug Stack v3.5.1.1 support

§ Please refer to the release notes<http://software-dl.ti.com/msp430/msp430_public_sw/mcu/msp430/MSPDS/3_05_001_001/index_FDS.html> for more details.

o Key Changes to Header & Support Files

§ F673xA: added missing 3rd SD24 Channel definition

§ FR59xx and FR69xx: removed unavailable defines for SYSSNIV_SVS and SYSSNIV_ACCTEIFG

§ F677x: fixed wrong define DMAxTSEL__USCIB3TX to DMAxTSEL__USCIA3TX

§ added: MSP430FR2433

o A new MSP430 GCC User Guide

§ This manual describes only the setup and basic operation of the MSP430™ GCC compiler and the software development environment. This manual applies to the use of MSP430 GCC as stand-alone package or within the Code Composer Studio™ (CCS) IDE v6.1

§ Building MSP430 GCC from sources instructions are also described in this manual.

§ http://www.ti.com/lit/pdf/slau646

§ Please submit your feedbacks: http://www.go-dsp.com/forms/techdoc/doc_feedback.htm?litnum=SLAU646

o The MSP430 GCC standalone package version scheme is updated starting from v3.5.0.0

§ Version is handled by the following scheme (4-digits): major.minor.bug.build

· major: Indicates a major release. A change in this number would indicate either a complete re-write or a major functional change which breaks compatibility with a previous version.

· minor: Indicates a new feature integration.

· bug: Indicates a pure bug fix.

· build: As such the build number has no significance, but might be use to convey component specific information.



· New Devices Support

o MSP430FR2433


· Known Issues

o The MSP430FR4133 (Gradon) is currently not fully supported by GDB + GDB Agent.

o When running the GNU debugger on Microsoft Windows, pressing Ctrl-C may terminate the debugger.

o The GDB Agent is not able to connect to MSP430L092 target devices.

o The GDB Agent is not able to connect to RF430FRL15xH target devices.

o MSP430FR2433 header file shows some bit defines which are not supported by the device and are not described in the user's guide.

o #define SYNCSEL0 (0x0040u)

o #define SYNCSEL1 (0x0080u)


o #define SYNCSEL0_L (0x0040u)

o #define SYNCSEL1_L (0x0080u)


o #define SYNCSEL_0 (0x0000u)

o #define SYNCSEL_1 (0x0040u)

o #define SYNCSEL_2 (0x0080u)

o #define SYNCSEL_3 (0x00C0u)


o #define SYNCSEL__EXT (0x0000u)

o #define SYNCSEL__ADC (0x0040u)

o #define SYNCSEL__COMP (0x0080u)

o #define SYNCSEL_3 (0x00C0u)


o #define CAPTIVATE_VECTOR (40 * 2u)


o It is recommended not to use them as they will be removed in the next release.


· How to get it?

o Use the Stand-alone & Open Source downloads from the GCC tools folder<http://www.ti.com/tool/msp430-gcc-opensource>

o Update through CCS v6.x installation.

§ Update your CCS v6.x installation (Help -> Search for updates) to the latest version and visit the CCS App Center under (Help -> Getting Started)
Basic migration advice and links to GCC documentation is provided in the CCSv6 for MSP430 user’s guide<http://www.ti.com/lit/pdf/slau157>

o Update through CCS v6.x installation for Mac OS X

§ Download the Code Composer Studio Mac OS X beta 2 for MSP430: Link<https://www.ti.com/licreg/docs/swlicexportcontrol.tsp?form_type=2&prod_no=CCS6.1.1.00026_osx.zip&ref_url=http://software-dl.ti.com/ccs/esd/CCSv6/CCS_6_1_1/>

· More details can be found on the e2e community<https://e2e.ti.com/blogs_/b/toolsinsider/archive/2015/03/20/code-composer-studio-beta-for-msp430>.

§ Update your CCS v6.x installation (Help -> Search for updates) to the latest version and visit the CCS App Center under (Help -> Getting Started)


Best regards,
________________________________
Youssef Ghannouchi
Product Manager - Texas Instruments Inc.


Texas Instruments Deutschland GmbH, Haggertystr. 1, D-85356 Freising. Amtsgericht MÌnchen HRB 40960. GeschÀftsfÌhrer: Dr. Wolfram Tietscher. Vorsitzender des Aufsichtsrates: Andreas Schwaiger

[Mark Rages]

On Sun, Sep 20, 2015 at 10:17 AM, Ghannouchi, Youssef

Post by Ghannouchi, Youssef
Hello GCC-enthusiasts,
We are pleased to announce a new stand-alone & open source release (v3.5.0.0) of MSP430 GCC available for download here<http://software-dl.ti.com/msp430/msp430_public_sw/mcu/msp430/MSPGCC/3_05_00_00/index_FDS.html>!
The stand-alone package includes all binaries, the source code and MSP430 device specific support files.
The MSP430 GCC build is also available in CCSv6 App Center on all platforms (Windows, Linux and Mac OSX).

What version of GCC is it? Neither 3.5.0 nor 14r1-364 are GCC version
numbers. Development of GCC 3.x ended a decade ago.

Regards,
Mark

------------------------------------------------------------------------------

[David Brown]

Post by Mark Rages
On Sun, Sep 20, 2015 at 10:17 AM, Ghannouchi, Youssef

Hello GCC-enthusiasts, We are pleased to announce a new stand-alone
& open source release (v3.5.0.0) of MSP430 GCC available for
download
here<http://software-dl.ti.com/msp430/msp430_public_sw/mcu/msp430/MSPGCC/3_05_00_00/index_FDS.html>!
The stand-alone package includes all binaries, the source code and >>

MSP430 device specific support files.

Post by Mark Rages

The MSP430 GCC build is also available in CCSv6 App Center on all
platforms (Windows, Linux and Mac OSX).

What version of GCC is it? Neither 3.5.0 nor 14r1-364 are GCC
version numbers. Development of GCC 3.x ended a decade ago.
Regards, Mark

It looks like it is gcc 4.9 (based on
<http://www.ti.com/tool/msp430-gcc-opensource> - I haven't downloaded
the new release).

It seems strange that they have stuck to 4.9, since the 5.x series has
been out for a while, but there is not a huge difference (5.x has some
nice new warnings, and more complete C++14, as well as a few more
optimisations).


------------------------------------------------------------------------------

[Peter Bigot]

My archive of msp430-elf releases at https://github.com/pabigot/msp430-elf
has been updated to include 3.05.00.00. The bottom of the README includes
a description of the version numbering used by TI. This new release does
include a few fixes/enhancements in the compiler, the first since
3.02.03.00, but is fundamentally the same GCC version as the original TI
release one year ago.

I no longer use MSP430 and at some point will stop updating my archive.
Earlier this summer I was using the upstream gcc+binutils (5.0.0-based at
the time, selected specifically for C++14 support), built as documented at
http://pabigot.github.io/bsp430/msp430elf.html. Stock GNU tools work fine
with MSP430 now, though they sometimes do lag in applying patches that
appear in the TI fork. Only the device-specific headers and linker scripts
from TI (also in my archive) are necessary when using upstream versions.

Peter

Post by Mark Rages
On Sun, Sep 20, 2015 at 10:17 AM, Ghannouchi, Youssef

Post by Ghannouchi, Youssef
Hello GCC-enthusiasts,
We are pleased to announce a new stand-alone & open source release

(v3.5.0.0) of MSP430 GCC available for download here<
http://software-dl.ti.com/msp430/msp430_public_sw/mcu/msp430/MSPGCC/3_05_00_00/index_FDS.html

Post by Ghannouchi, Youssef
!
The stand-alone package includes all binaries, the source code and

MSP430 device specific support files.

Post by Ghannouchi, Youssef
The MSP430 GCC build is also available in CCSv6 App Center on all

platforms (Windows, Linux and Mac OSX).
What version of GCC is it? Neither 3.5.0 nor 14r1-364 are GCC version
numbers. Development of GCC 3.x ended a decade ago.
Regards,
Mark
------------------------------------------------------------------------------
_______________________________________________
Mspgcc-users mailing list
https://lists.sourceforge.net/lists/listinfo/mspgcc-users

[Orlando Arias]

Greetings,

First and foremost I would like to personally thank Peter Bigot for his
efforts on maintaining the mspgcc toolchain before TI/RedHat took over
and developed their own and then for still caring enough to aid with
community builds for the msp430-elf toolchain. I feel this must be
addressed, given that this effort has been made even when, I believe,
there is some friction between TI/RedHat and Mr. Bigot himself.

That being said, I would like to mention that I have released a few
PKGBUILDs for Arch Linux which are able to build packages that can be
installed in that system. These may be found in the AUR [1]. I am using
binutils 2.25, gcc 5.2.0, newlib 2.2.0.20150824 and the latest snapshot
of TI headers and linker scripts.

If you decide to use these in your Arch Linux system, please report any
bugs you find with these PKGBUILDs and packages over in the AUR and not
here. Also, I am sorry, but I am unable to provide packages for other
distributions as I only utilize Arch Linux.

With that being said, thank you again for all your efforts, good sir.

Cheers,
Orlando.


[1] https://aur.archlinux.org/packages/?K=msp430-elf

Post by Peter Bigot
My archive of msp430-elf releases at https://github.com/pabigot/msp430-elf
has been updated to include 3.05.00.00. The bottom of the README includes
a description of the version numbering used by TI. This new release does
include a few fixes/enhancements in the compiler, the first since
3.02.03.00, but is fundamentally the same GCC version as the original TI
release one year ago.
I no longer use MSP430 and at some point will stop updating my archive.
Earlier this summer I was using the upstream gcc+binutils (5.0.0-based at
the time, selected specifically for C++14 support), built as documented at
http://pabigot.github.io/bsp430/msp430elf.html. Stock GNU tools work fine
with MSP430 now, though they sometimes do lag in applying patches that
appear in the TI fork. Only the device-specific headers and linker scripts
from TI (also in my archive) are necessary when using upstream versions.
Peter

Post by Mark Rages
On Sun, Sep 20, 2015 at 10:17 AM, Ghannouchi, Youssef

Post by Ghannouchi, Youssef
Hello GCC-enthusiasts,
We are pleased to announce a new stand-alone & open source release

(v3.5.0.0) of MSP430 GCC available for download here<
http://software-dl.ti.com/msp430/msp430_public_sw/mcu/msp430/MSPGCC/3_05_00_00/index_FDS.html

Post by Ghannouchi, Youssef
!
The stand-alone package includes all binaries, the source code and

MSP430 device specific support files.

Post by Ghannouchi, Youssef
The MSP430 GCC build is also available in CCSv6 App Center on all

platforms (Windows, Linux and Mac OSX).
What version of GCC is it? Neither 3.5.0 nor 14r1-364 are GCC version
numbers. Development of GCC 3.x ended a decade ago.
Regards,
Mark
------------------------------------------------------------------------------
_______________________________________________
Mspgcc-users mailing list
https://lists.sourceforge.net/lists/listinfo/mspgcc-users

------------------------------------------------------------------------------
_______________________________________________
Mspgcc-users mailing list
https://lists.sourceforge.net/lists/listinfo/mspgcc-users

[Grant Edwards]

Post by Peter Bigot
I no longer use MSP430 and at some point will stop updating my
archive. Earlier this summer I was using the upstream gcc+binutils
(5.0.0-based at the time, selected specifically for C++14 support),
built as documented at
http://pabigot.github.io/bsp430/msp430elf.html. Stock GNU tools work
fine with MSP430 now, though they sometimes do lag in applying
patches that appear in the TI fork. Only the device-specific headers
and linker scripts from TI (also in my archive) are necessary when
using upstream versions.

Things have come a long way since I first started using mspgcc about
11 years ago. I'm glad TI stopped dragging their feet and acting
generally obstructive and got behind gcc. I'm no longer an active
user of the '430, but I'd like to extend my personal thanks to
everybody who worked on mspgcc over the years (especially in the old
days) -- and thanks/congratulations to whoever at TI finally slapped
some sense into management at TI and got them to support gcc. Back in
the day it must have been embarassing to many at TI to see how highly
everbody regarded Atmel due to thier commitment to gcc and
open-source. [Expecially considering how annoying and hard to support
the AVR architecture was compared to the '430.]

[Ghannouchi, Youssef]

Hello GCC-enthusiasts,

I have seen some questions and feedbacks related to the email below and I would like to clarify some points.

1. Versioning

a. The MSP430 GCC standalone package is container which has its own version. The latest version is v3.5.0.0. Please refer to versioning scheme for the package described below.

b. The MSP430 GCC standalone package has a software manifest included in the package (attached also GCC_14r1_manifest.pdf) which describe in details all sub components and their version.

c. The package contains:

i. GGC 4.9.1

ii. GDB 7.5.0

iii. binutils 2.23

iv. Tcl 8.4.1

v. MSP430 Debug Stack 3.5.1.1

vi. MSP430 Support Files 1.173

vii. MSP430 GDB Agent 6.0.14.5



2. MSP430 GCC 5.1

a. The MSP430 GCC standalone package v3.5.0.0 uses the GCC 4.9.1. The purpose of this package was to give a quick update on GCC compiler including the latest bug fixes to allow our customers continuing their development easily.

b. We are working on preparing an MSP430 compiler based on GCC 5.1. This package is expected to be released by end of this year. I will keep informed.


Thank you for the valuable feedbacks, we really appreciate it.
We would like to continue improving MSP430 GCC offer, so please feel free to send any suggestions or recommendations.
For better tracking and handling please use TI e2e community<https://e2e.ti.com/> to submit your requests.

Best regards,
________________________________
Youssef Ghannouchi
Product Manager - Texas Instruments Inc.



Texas Instruments Deutschland GmbH, Haggertystr. 1, D-85356 Freising. Amtsgericht MÌnchen HRB 40960. GeschÀftsfÌhrer: Dr. Wolfram Tietscher. Vorsitzender des Aufsichtsrates: Andreas Schwaiger


From: Ghannouchi, Youssef
Sent: Sunday, September 20, 2015 6:18 PM
To: 'GCC for MSP430 - http://mspgcc.sf.net'
Cc: 'Brendan Conoboy (***@redhat.com)'; Miller, Greg; Mitnacht, Thomas
Subject: New MSP430 GCC version release available!

Hello GCC-enthusiasts,
We are pleased to announce a new stand-alone & open source release (v3.5.0.0) of MSP430 GCC available for download here<http://software-dl.ti.com/msp430/msp430_public_sw/mcu/msp430/MSPGCC/3_05_00_00/index_FDS.html>!
The stand-alone package includes all binaries, the source code and MSP430 device specific support files.
The MSP430 GCC build is also available in CCSv6 App Center on all platforms (Windows, Linux and Mac OSX).


· Highlights

o MSP430 GCC compiler latest build (14r1-364)

o MSP430 GCC for Mac OS X (beta release)

o A new MSP430 GCC User Guide



· Package Content

o Full Packages

§ msp430-gcc-full-linux-installer-3.5.0.0.run: MSP430 GCC Linux installer incl. support files and debug stack and USB drivers - apply chmod x before executing the package

§ msp430-gcc-full-osx-installer-3.5.0.0-beta.zip: MSP430 GCC Mac OS X installer incl. support files and debug stack and USB drivers - beta release

§ msp430-gcc-full-windows-installer-3.5.0.0.exe: MSP430 GCC Windows installer incl. support files and debug stack and USB drivers

o Compiler Only Packages

§ msp430-gcc-linux-installer-3.5.0.0.run: MSP430 GCC Linux installer - compiler only - apply chmod x before executing the package

§ msp430-gcc-osx-installer-3.5.0.0-beta.zip: MSP430 GCC Mac OS X installer - compiler only - beta release

§ msp430-gcc-windows-installer-3.5.0.0.exe: MSP430 GCC Windows installer - compiler only

o MSP430 Support Files

§ msp430-gcc-support-files.zip: MSP430 Header and linker Files

o MSP430 GCC Sources files

§ msp430-gcc-source.tar.bz2: MSP430 GCC source files



· What's New

o Key Changes to MSP430 GCC

§ Updated GCC compiler (14r1-364)

· Code to disable watchdog timer has been removed from start-up sequence (See examples).

· Fixes for Hardware Multiplier issues for F5 series (Details<https://e2e.ti.com/support/development_tools/compiler/f/343/t/417242#pi317851=2>).

· Please refer to the changelogs.txt in the install directory for more details about all changes vs. the previous build 14r1-167.

§ MSP430 GCC for Mac OS X (beta release)

§ Updated GDB agent (6.0.14.5)

· Added support for following GDB monitor commands:

o 'monitor reset' - reset the CPU

o 'monitor halt' - halt the CPU

o MSP Debug Stack v3.5.1.1 support

§ Please refer to the release notes<http://software-dl.ti.com/msp430/msp430_public_sw/mcu/msp430/MSPDS/3_05_001_001/index_FDS.html> for more details.

o Key Changes to Header & Support Files

§ F673xA: added missing 3rd SD24 Channel definition

§ FR59xx and FR69xx: removed unavailable defines for SYSSNIV_SVS and SYSSNIV_ACCTEIFG

§ F677x: fixed wrong define DMAxTSEL__USCIB3TX to DMAxTSEL__USCIA3TX

§ added: MSP430FR2433

o A new MSP430 GCC User Guide

§ This manual describes only the setup and basic operation of the MSP430™ GCC compiler and the software development environment. This manual applies to the use of MSP430 GCC as stand-alone package or within the Code Composer Studio™ (CCS) IDE v6.1

§ Building MSP430 GCC from sources instructions are also described in this manual.

§ http://www.ti.com/lit/pdf/slau646

§ Please submit your feedbacks: http://www.go-dsp.com/forms/techdoc/doc_feedback.htm?litnum=SLAU646

o The MSP430 GCC standalone package version scheme is updated starting from v3.5.0.0

§ Version is handled by the following scheme (4-digits): major.minor.bug.build

· major: Indicates a major release. A change in this number would indicate either a complete re-write or a major functional change which breaks compatibility with a previous version.

· minor: Indicates a new feature integration.

· bug: Indicates a pure bug fix.

· build: As such the build number has no significance, but might be use to convey component specific information.



· New Devices Support

o MSP430FR2433


· Known Issues

o The MSP430FR4133 (Gradon) is currently not fully supported by GDB + GDB Agent.

o When running the GNU debugger on Microsoft Windows, pressing Ctrl-C may terminate the debugger.

o The GDB Agent is not able to connect to MSP430L092 target devices.

o The GDB Agent is not able to connect to RF430FRL15xH target devices.

o MSP430FR2433 header file shows some bit defines which are not supported by the device and are not described in the user's guide.

o #define SYNCSEL0 (0x0040u)

o #define SYNCSEL1 (0x0080u)


o #define SYNCSEL0_L (0x0040u)

o #define SYNCSEL1_L (0x0080u)


o #define SYNCSEL_0 (0x0000u)

o #define SYNCSEL_1 (0x0040u)

o #define SYNCSEL_2 (0x0080u)

o #define SYNCSEL_3 (0x00C0u)


o #define SYNCSEL__EXT (0x0000u)

o #define SYNCSEL__ADC (0x0040u)

o #define SYNCSEL__COMP (0x0080u)

o #define SYNCSEL_3 (0x00C0u)


o #define CAPTIVATE_VECTOR (40 * 2u)


o It is recommended not to use them as they will be removed in the next release.


· How to get it?

o Use the Stand-alone & Open Source downloads from the GCC tools folder<http://www.ti.com/tool/msp430-gcc-opensource>

o Update through CCS v6.x installation.

§ Update your CCS v6.x installation (Help -> Search for updates) to the latest version and visit the CCS App Center under (Help -> Getting Started)
Basic migration advice and links to GCC documentation is provided in the CCSv6 for MSP430 user’s guide<http://www.ti.com/lit/pdf/slau157>

o Update through CCS v6.x installation for Mac OS X

§ Download the Code Composer Studio Mac OS X beta 2 for MSP430: Link<https://www.ti.com/licreg/docs/swlicexportcontrol.tsp?form_type=2&prod_no=CCS6.1.1.00026_osx.zip&ref_url=http://software-dl.ti.com/ccs/esd/CCSv6/CCS_6_1_1/>

· More details can be found on the e2e community<https://e2e.ti.com/blogs_/b/toolsinsider/archive/2015/03/20/code-composer-studio-beta-for-msp430>.

§ Update your CCS v6.x installation (Help -> Search for updates) to the latest version and visit the CCS App Center under (Help -> Getting Started)


Best regards,
________________________________
Youssef Ghannouchi
Product Manager - Texas Instruments Inc.

[Peter Bigot]

Thanks for the update.

Please note the mspgcc-users mailing list has historically had a maximum
message length of 40 kB, and does not allow attachments. A couple of your
messages have exceeded this limit and I've passed them through, but that's
why you may see delays on your postings.

Also may I suggest TI make the effort to rebase the next compiler release
on gcc 5.2 since that was released last July (three months after 5.1). I
don't know if they have a 5.3 coming up before you expect your version to
be out, but you might want to coordinate with the GNU folks on that.

Post by Ghannouchi, Youssef
Hello GCC-enthusiasts,
I have seen some questions and feedbacks related to the email below and I
would like to clarify some points.
1. Versioning
a. The MSP430 GCC standalone package is container which has its own
version. The latest version is v3.5.0.0. Please refer to versioning scheme
for the package described below.
b. The MSP430 GCC standalone package has a software manifest included
in the package (attached also GCC_14r1_manifest.pdf) which describe in
details all sub components and their version.
i. GGC 4.9.1
ii. GDB 7.5.0
iii.
binutils 2.23
iv. Tcl 8.4.1
v.
MSP430 Debug Stack 3.5.1.1
vi. MSP430
Support Files 1.173
vii. MSP430
GDB Agent 6.0.14.5
2. MSP430 GCC 5.1
a. The MSP430 GCC standalone package v3.5.0.0 uses the GCC 4.9.1.
The purpose of this package was to give a quick update on GCC compiler
including the latest bug fixes to allow our customers continuing their
development easily.
b. We are working on preparing an MSP430 compiler based on GCC 5.1.
This package is expected to be released by end of this year. I will keep
informed.
Thank you for the valuable feedbacks, we really appreciate it.
We would like to continue improving MSP430 GCC offer, so please feel free
to send any suggestions or recommendations.
For better tracking and handling please use TI e2e community<
https://e2e.ti.com/> to submit your requests.
Best regards,
________________________________
Youssef Ghannouchi
Product Manager - Texas Instruments Inc.
Texas Instruments Deutschland GmbH, Haggertystr. 1, D-85356 Freising.
Amtsgericht MÌnchen HRB 40960. GeschÀftsfÌhrer: Dr. Wolfram Tietscher.
Vorsitzender des Aufsichtsrates: Andreas Schwaiger
From: Ghannouchi, Youssef
Sent: Sunday, September 20, 2015 6:18 PM
To: 'GCC for MSP430 - http://mspgcc.sf.net'
Subject: New MSP430 GCC version release available!
Hello GCC-enthusiasts,
We are pleased to announce a new stand-alone & open source release
(v3.5.0.0) of MSP430 GCC available for download here<
http://software-dl.ti.com/msp430/msp430_public_sw/mcu/msp430/MSPGCC/3_05_00_00/index_FDS.html

!

The stand-alone package includes all binaries, the source code and MSP430
device specific support files.
The MSP430 GCC build is also available in CCSv6 App Center on all
platforms (Windows, Linux and Mac OSX).
· Highlights
o MSP430 GCC compiler latest build (14r1-364)
o MSP430 GCC for Mac OS X (beta release)
o A new MSP430 GCC User Guide
· Package Content
o Full Packages
§ msp430-gcc-full-linux-installer-3.5.0.0.run: MSP430 GCC Linux installer
incl. support files and debug stack and USB drivers - apply chmod x before
executing the package
§ msp430-gcc-full-osx-installer-3.5.0.0-beta.zip: MSP430 GCC Mac OS X
installer incl. support files and debug stack and USB drivers - beta release
§ msp430-gcc-full-windows-installer-3.5.0.0.exe: MSP430 GCC Windows
installer incl. support files and debug stack and USB drivers
o Compiler Only Packages
§ msp430-gcc-linux-installer-3.5.0.0.run: MSP430 GCC Linux installer -
compiler only - apply chmod x before executing the package
§ msp430-gcc-osx-installer-3.5.0.0-beta.zip: MSP430 GCC Mac OS X
installer - compiler only - beta release
§ msp430-gcc-windows-installer-3.5.0.0.exe: MSP430 GCC Windows installer
- compiler only
o MSP430 Support Files
§ msp430-gcc-support-files.zip: MSP430 Header and linker Files
o MSP430 GCC Sources files
§ msp430-gcc-source.tar.bz2: MSP430 GCC source files
· What's New
o Key Changes to MSP430 GCC
§ Updated GCC compiler (14r1-364)
· Code to disable watchdog timer has been removed from start-up
sequence (See examples).
· Fixes for Hardware Multiplier issues for F5 series (Details<
https://e2e.ti.com/support/development_tools/compiler/f/343/t/417242#pi317851=2

).

· Please refer to the changelogs.txt in the install directory for
more details about all changes vs. the previous build 14r1-167.
§ MSP430 GCC for Mac OS X (beta release)
§ Updated GDB agent (6.0.14.5)
o 'monitor reset' - reset the CPU
o 'monitor halt' - halt the CPU
o MSP Debug Stack v3.5.1.1 support
§ Please refer to the release notes<
http://software-dl.ti.com/msp430/msp430_public_sw/mcu/msp430/MSPDS/3_05_001_001/index_FDS.html>
for more details.
o Key Changes to Header & Support Files
§ F673xA: added missing 3rd SD24 Channel definition
§ FR59xx and FR69xx: removed unavailable defines for SYSSNIV_SVS and SYSSNIV_ACCTEIFG
§ F677x: fixed wrong define DMAxTSEL__USCIB3TX to DMAxTSEL__USCIA3TX
§ added: MSP430FR2433
o A new MSP430 GCC User Guide
§ This manual describes only the setup and basic operation of the MSP430™
GCC compiler and the software development environment. This manual applies
to the use of MSP430 GCC as stand-alone package or within the Code Composer
Studio™ (CCS) IDE v6.1
§ Building MSP430 GCC from sources instructions are also described in this manual.
§ http://www.ti.com/lit/pdf/slau646
http://www.go-dsp.com/forms/techdoc/doc_feedback.htm?litnum=SLAU646
o The MSP430 GCC standalone package version scheme is updated starting from v3.5.0.0
major.minor.bug.build
· major: Indicates a major release. A change in this number would
indicate either a complete re-write or a major functional change which
breaks compatibility with a previous version.
· minor: Indicates a new feature integration.
· bug: Indicates a pure bug fix.
· build: As such the build number has no significance, but might
be use to convey component specific information.
· New Devices Support
o MSP430FR2433
· Known Issues
o The MSP430FR4133 (Gradon) is currently not fully supported by GDB + GDB Agent.
o When running the GNU debugger on Microsoft Windows, pressing Ctrl-C
may terminate the debugger.
o The GDB Agent is not able to connect to MSP430L092 target devices.
o The GDB Agent is not able to connect to RF430FRL15xH target devices.
o MSP430FR2433 header file shows some bit defines which are not
supported by the device and are not described in the user's guide.
o #define SYNCSEL0 (0x0040u)
o #define SYNCSEL1 (0x0080u)
o #define SYNCSEL0_L (0x0040u)
o #define SYNCSEL1_L (0x0080u)
o #define SYNCSEL_0 (0x0000u)
o #define SYNCSEL_1 (0x0040u)
o #define SYNCSEL_2 (0x0080u)
o #define SYNCSEL_3 (0x00C0u)
o #define SYNCSEL__EXT (0x0000u)
o #define SYNCSEL__ADC (0x0040u)
o #define SYNCSEL__COMP (0x0080u)
o #define SYNCSEL_3 (0x00C0u)
o #define CAPTIVATE_VECTOR (40 * 2u)
o It is recommended not to use them as they will be removed in the next release.
· How to get it?
o Use the Stand-alone & Open Source downloads from the GCC tools folder<
http://www.ti.com/tool/msp430-gcc-opensource>
o Update through CCS v6.x installation.
§ Update your CCS v6.x installation (Help -> Search for updates) to the
latest version and visit the CCS App Center under (Help -> Getting Started)
Basic migration advice and links to GCC documentation is provided in the
CCSv6 for MSP430 user’s guide<http://www.ti.com/lit/pdf/slau157>
o Update through CCS v6.x installation for Mac OS X
§ Download the Code Composer Studio Mac OS X beta 2 for MSP430: Link<
https://www.ti.com/licreg/docs/swlicexportcontrol.tsp?form_type=2&prod_no=CCS6.1.1.00026_osx.zip&ref_url=http://software-dl.ti.com/ccs/esd/CCSv6/CCS_6_1_1/
· More details can be found on the e2e community<
https://e2e.ti.com/blogs_/b/toolsinsider/archive/2015/03/20/code-composer-studio-beta-for-msp430

.

§ Update your CCS v6.x installation (Help -> Search for updates) to the
latest version and visit the CCS App Center under (Help -> Getting Started)
Best regards,
________________________________
Youssef Ghannouchi
Product Manager - Texas Instruments Inc.
------------------------------------------------------------------------------
_______________________________________________
Mspgcc-users mailing list
https://lists.sourceforge.net/lists/listinfo/mspgcc-users

[David W. Schultz]

Hello GCC-enthusiasts, We are pleased to announce a new stand-alone &
open source release (v3.5.0.0) of MSP430 GCC

I don't see any mention of a fix to the compiler saving more registers
than required when entering an interrupt routine. Its like it completely
forgot about the EABI.

[Lev Serebryakov]

On 20.09.2015 19:17, Ghannouchi, Youssef wrote:

I could not build this version on FreeBSD. It could not link "sim"
(multiple symbol definitions):

cc -DHAVE_CONFIG_H -DPROFILE=1 -DWITH_PROFILE=-1
-DWITH_DEFAULT_ALIGNMENT=STRICT_ALIGNMENT
-DWITH_TARGET_BYTE_ORDER=LITTLE_ENDIAN -DDEFAULT_INLINE=0
-I. -I../../.././sim/msp430 -I../common
-I../../.././sim/msp430/../common -I../../include
-I../../.././sim/msp430/../../include -I../../bfd
-I../../.././sim/msp430/../../bfd -I../../opcodes
-I../../.././sim/msp430/../../opcodes -O2 -pipe -DLIBICONV_PLUG
-fstack-protector -fno-strict-aliasing -fstack-protector -o run \
nrun.o libsim.a ../../bfd/libbfd.a ../../opcodes/libopcodes.a
../../libiberty/libiberty.a -lz
libsim.a(sim-info.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-info.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-module.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-module.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-options.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-options.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-profile.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-profile.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-signal.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-signal.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-trace.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-trace.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-utils.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-utils.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(msp430-sim.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/msp430-sim.c:(.text+0x0): multiple definition of
`sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-cpu.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-cpu.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-engine.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-engine.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-hload.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-hload.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-hrw.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-hrw.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-reason.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-reason.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-resume.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-resume.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-stop.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-stop.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-command.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-command.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-config.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-config.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-core.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-core.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-endian.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-endian.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-events.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-events.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-fpu.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-fpu.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-io.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-io.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-load.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-load.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-memopt.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-memopt.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
libsim.a(sim-bits.o): In function `sim_addr_range_hit_p':
../../.././sim/msp430/../common/sim-bits.c:(.text+0x0): multiple
definition of `sim_addr_range_hit_p'
nrun.o:../../.././sim/msp430/../common/nrun.c:(.text+0x0): first defined
here
cc: error: linker command failed with exit code 1 (use -v to see invocation)
Makefile:311: recipe for target 'run' failed
gmake[4]: *** [run] Error 1
gmake[4]: Leaving directory
'/usr/home/lev/FreeBSD/ports/devel/gcc-msp430-ti-toolchain/work/sources/tools/host-amd64-portbld-freebsd10.2/sim/msp430'
Makefile:129: recipe for target 'all' failed
gmake[3]: *** [all] Error 1
gmake[3]: Leaving directory
'/usr/home/lev/FreeBSD/ports/devel/gcc-msp430-ti-toolchain/work/sources/tools/host-amd64-portbld-freebsd10.2/sim'

Post by Ghannouchi, Youssef
Hello GCC-enthusiasts,
We are pleased to announce a new stand-alone & open source release (v3.5.0.0) of MSP430 GCC available for download here<http://software-dl.ti.com/msp430/msp430_public_sw/mcu/msp430/MSPGCC/3_05_00_00/index_FDS.html>!
The stand-alone package includes all binaries, the source code and MSP430 device specific support files.
The MSP430 GCC build is also available in CCSv6 App Center on all platforms (Windows, Linux and Mac OSX).

[DJ Delorie]

You're probably running into a common problem with "static inline" vs
"extern inline" - the semantics of those changed with the recent C
standards, so not all compilers do the same thing. I assume you're
not using GCC?

------------------------------------------------------------------------------

[Lev Serebryakov]

Hello DJ,

Post by DJ Delorie
You're probably running into a common problem with "static inline" vs
"extern inline" - the semantics of those changed with the recent C
standards, so not all compilers do the same thing. I assume you're
not using GCC?

It is clang 3.4.1, and I don't want to install gcc, if it is possible to
avoid with some "simple" patch :)

[DJ Delorie]

Post by Lev Serebryakov
It is clang 3.4.1, and I don't want to install gcc, if it is possible to
avoid with some "simple" patch :)

IIRC there's an option in clang to use the old semantics (-std=gnu89).

------------------------------------------------------------------------------

[Lev Serebryakov]

Hello DJ,

Post by DJ Delorie
You're probably running into a common problem with "static inline" vs
"extern inline" - the semantics of those changed with the recent C
standards, so not all compilers do the same thing. I assume you're
not using GCC?

Ok, tinkering with "extern" and "inline" helps, but now I have more serious
problem:

gmake[3]: Leaving directory '/usr/home/lev/FreeBSD/ports/devel/gcc-msp430-ti-toolchain/work/sources/tools/host-amd64-portbld-freebsd10.2/gdb'
Checking multilib configuration for libgcc...
mkdir msp430-elf
mkdir msp430-elf/libgcc
Configuring in msp430-elf/libgcc
[SKIPPED]
configure: updating cache ./config.cache
configure: creating ./config.status
config.status: creating Makefile
config.status: creating auto-target.h
config.status: linking ../../.././libgcc/enable-execute-stack-empty.c to enable-execute-stack.c
config.status: linking ../../.././libgcc/unwind-generic.h to unwind.h
config.status: linking ../../.././libgcc/config/no-unwind.h to md-unwind-support.h
config.status: linking ../../.././libgcc/config/no-sfp-machine.h to sfp-machine.h
config.status: linking ../../.././libgcc/gthr-single.h to gthr-default.h
config.status: executing default commands
Adding multilib support to Makefile in ../../.././libgcc
with_multisubdir=large
gmake[3]: Entering directory '/usr/home/lev/FreeBSD/ports/devel/gcc-msp430-ti-toolchain/work/sources/tools/msp430-elf/libgcc'
Makefile:165: ../.././gcc/libgcc.mvars: No such file or directory
gmake[3]: *** No rule to make target '../.././gcc/libgcc.mvars'. Stop.
gmake[3]: Leaving directory '/usr/home/lev/FreeBSD/ports/devel/gcc-msp430-ti-toolchain/work/sources/tools/msp430-elf/libgcc'
Makefile:17333: recipe for target 'all-target-libgcc' failed
gmake[2]: *** [all-target-libgcc] Error 2
gmake[2]: Leaving directory '/usr/home/lev/FreeBSD/ports/devel/gcc-msp430-ti-toolchain/work/sources/tools'
Makefile:860: recipe for target 'all' failed

[DJ Delorie]

That should be built as part of building gcc/

so you probably have a problem earlier on that you didn't see in your logs...

------------------------------------------------------------------------------

[Lev Serebryakov]

Hello DJ,

Friday, October 16, 2015, 9:12:20 PM, you wrote:


BTW, "gnu89" helps, thank you.

Post by DJ Delorie
That should be built as part of building gcc/
so you probably have a problem earlier on that you didn't see in your logs...

Nope. It was build here ("root" is sources/tools from tarball):

"host-amd64-portbld-freebsd10.2/gcc/libgcc.mvars"

But later build try to find it here:

gmake[3]: Entering directory '/usr/home/lev/FreeBSD/ports/devel/gcc-msp430-ti-toolchain/work/sources/tools/msp430-elf/libgcc'
Makefile:165: ../.././gcc/libgcc.mvars: No such file or directory

Which is (again, path is trimmed):

"msp430-elf/libgcc/../.././gcc" -> "gcc", which is not
"host-amd64-portbld-freebsd10.2/gcc/libgcc.mvars" :)

So, it looks like everyhting but msp430-elf/libgcc is built out-of-tree and
this library is build in-tree and here are some problems with that.

Makefile in "msp430-elf/libgcc" contains:

====
libgcc_topdir = ../.././libgcc/..
host_subdir = .

gcc_objdir = $(MULTIBUILDTOP)../../$(host_subdir)/gcc
====

"host_subdir" is clearly invalid :)

[Lev Serebryakov]

Hello Lev,

Post by Lev Serebryakov
====
libgcc_topdir = ../.././libgcc/..
host_subdir = .
gcc_objdir = $(MULTIBUILDTOP)../../$(host_subdir)/gcc
====
"host_subdir" is clearly invalid :)

Really, This problem exists in each "msp430/**/libgcc/Makefile".

[DJ Delorie]

Two questions:

1. Are you cross-building a compiler? I.e. is the host you're
building *on* different than the host (not target, which is msp430)
you're building *for*? (This is called a "canadian cross" and is
much more complicated than a regular cross compiler).

(one typically cross-builds by accident by specifying --build or
--host but not both, or by specifying both but not having them the
same.)

2. Are you building in the source tree? If so, that is not supported,
you must always build in a separate (empty) directory.

------------------------------------------------------------------------------

[Lev Serebryakov]

Hello DJ,

Post by DJ Delorie
1. Are you cross-building a compiler? I.e. is the host you're
building *on* different than the host (not target, which is msp430)
you're building *for*? (This is called a "canadian cross" and is
much more complicated than a regular cross compiler).

Nope. *host* and *build* are both the same.

Post by DJ Delorie
(one typically cross-builds by accident by specifying --build or
--host but not both, or by specifying both but not having them the
same.)
2. Are you building in the source tree? If so, that is not supported,
you must always build in a separate (empty) directory.

Oh, it looks like my error. I'm trying to build out-of-tree now.

[David W. Schultz]

Post by Lev Serebryakov
Hello DJ,

Post by DJ Delorie
1. Are you cross-building a compiler? I.e. is the host you're
building *on* different than the host (not target, which is msp430)
you're building *for*? (This is called a "canadian cross" and is
much more complicated than a regular cross compiler).

Nope. *host* and *build* are both the same.

Post by DJ Delorie
(one typically cross-builds by accident by specifying --build or
--host but not both, or by specifying both but not having them the
same.)
2. Are you building in the source tree? If so, that is not supported,
you must always build in a separate (empty) directory.

Oh, it looks like my error. I'm trying to build out-of-tree now.

I built the TI version from source a while back. I don't recall exactly
what I did but fortunately, it remembers.

I created the directory ti/gcc/sources/tools/msp430-build

The config.log reveals that I ran from that directory:

../configure --target=msp430-elf --program-prefix=msp430-elf --with-newlib

gcc has a lot of build options and by default it is going to build a lot
of stuff you have no interest in. Taking a long time to do it. I also
played with building a version for the 68000 with:

../configure --target=m68k-elf --with-arch=m68k --program-prefix=m68k-
--enable-languages=c --disable-libmudflap --disable-libssp
--disable-libgomp --disable-libstdcxx-pch --disable-threads

And since that was in its own build directory, it didn't conflict at all
with the msp430 build. It is almost certain that there is an optimal set
of options for the task but I don't know what they are.

[Lev Serebryakov]

Hello Lev,

Post by Lev Serebryakov

Post by DJ Delorie
2. Are you building in the source tree? If so, that is not supported,
you must always build in a separate (empty) directory.

Oh, it looks like my error. I'm trying to build out-of-tree now.

Yep, it works, modulo DESTDIR problems, which could be easily patched.
I'm not sure, that toolchain need all these separate tcl and tk stuff (system
already has them!), but it is better than nothing (though, very
Windows-like!).

[DJ Delorie]

Post by Lev Serebryakov
Yep, it works, modulo DESTDIR problems, which could be easily patched.

We've always used a separate --prefix for each release (typically
/opt/redhat/msp430-YYMMDD/) so we wouldn't notice.

Post by Lev Serebryakov
I'm not sure, that toolchain need all these separate tcl and tk
stuff (system already has them!), but it is better than nothing
(though, very Windows-like!).

No surprise, we support windows :-)

And not all Unix-y systems have those, either. Even Linux is the user
hasn't chosen to install them yet (which is commonly the case).

------------------------------------------------------------------------------

[Lev Serebryakov]

Hello DJ,

Post by DJ Delorie

Post by Lev Serebryakov
Yep, it works, modulo DESTDIR problems, which could be easily patched.

We've always used a separate --prefix for each release (typically
/opt/redhat/msp430-YYMMDD/) so we wouldn't notice.

When it is build for system package (like RPM, DEB or Gentoo Portage) it
should has proper (final) prefix (like /usr/local/ti-gcc-${version}) but
"make install" must place it in "stage" directory for packaging. It is
typically done with DESTDIR (so, destination becomes $(DESTDIR)$(PREFIX),
not just $(PREFIX), and it is supported by all autotools/libtools
projects.

But, as I say, it is very minor problem, as it could be easily patched by
me (package maintainer), it is only 3KiB of patches, which is negligible :)

Post by DJ Delorie

Post by Lev Serebryakov
I'm not sure, that toolchain need all these separate tcl and tk
stuff (system already has them!), but it is better than nothing
(though, very Windows-like!).

No surprise, we support windows :-)

Yep, I know :)

Post by DJ Delorie
And not all Unix-y systems have those, either. Even Linux is the user
hasn't chosen to install them yet (which is commonly the case).

Yes. But typical *IX way is to use package managers for each separate
component. Ideally, all this stuff, which is not MSP430-dependand (build for
host and could be used by many oither packages and programs in same system):

libgmp
libmpfr
zlib
tcl
tk
itlc

should be mentioned in requirements and installed by means of system
package manager (rmp, apt-get, Gentoo Portages, NetBSD pkg-src, FreeBSD
ports, etc). For example, msp430 package contains old versions of gmp and mpfr.
Are you sure, that packaged versions doesnt have bugs and, may be, even
security ones?

IMHO, ideal source-based releasing of toolcahin looks like this:

(1) List of requirements, like "libgmp > 5.1.0, mprf > 3.1.0" and so on.
NO SOURCES of all these libraries should be provided, as they are
SEPARATE projects.

(2) Links to some official binutils / gcc / gdb RELEASE tarballs (gcc 4.9.1,
for example).

(3) Minimal patches for (2), which is msp430-dependant.

(4) System headers, linker scripts, stuff like this.

(5) Some script to build toolchain from all parts above
(unpack-patch-configure-build in proper order).

It is ideal situation, which allows to package your work for differnet
distributives in most natural way.

Second-to-best way is like ARM does: script, which, really, allows to
build toolchain with system libraries (but binutls/gcc/gdb are packaged as
whole).

To be honest, your (RedHat + TI) way is worst possible one :-(

[DJ Delorie]

Post by Lev Serebryakov
To be honest, your (RedHat + TI) way is worst possible one :-(

Worst for your purposes, perhaps. We just have a different goal - a
turnkey custom collection that "just works" for our customers. That
means we normally include things that wouldn't be included in a system
package.

What you want is an "upstream" collection, broken down by packages,
with the latest greatest of each package. Over time, the upstream
releases with msp430 support will happen and allow that. Often we do
releases for customers that can't go upstream right away (usually
because the chips are still NDA, which is obviously not the case for
MSP430 ;) so our workflow is geared towards self-sufficiency.

And BTW I know all about packaging rules, I do after all work for Red
Hat, we have a distro or two ourselves :-)

------------------------------------------------------------------------------

[Lev Serebryakov]

Hello DJ,

Post by DJ Delorie

Post by Lev Serebryakov
To be honest, your (RedHat + TI) way is worst possible one :-(

Worst for your purposes, perhaps. We just have a different goal - a
turnkey custom collection that "just works" for our customers. That
means we normally include things that wouldn't be included in a system
package.

Yep, because my goal is to do system package for yet another system :)

Post by DJ Delorie
And BTW I know all about packaging rules, I do after all work for Red
Hat, we have a distro or two ourselves :-)

:-)

Anyway, I'm pleased, that it could be built on BSD system with clang
compiler almost flawlessly. I've remember days, when such package for Agenda
VR3 (did you remember this very first Linuix-based PDA with MIPS-based CPU?)
was hopeless to built anywhere but one distributive of Linux!

Could you please clear situation with debugging (in separate thread in
mailing list)?

[DJ Delorie]

Agenda VR3 (did you remember this very first Linuix-based PDA with
MIPS-based CPU?)

Yup, I was involved with the project way back then.

Could you please clear situation with debugging (in separate thread in
mailing list)?

Sorry, that's a TI question, they provide the DLLs and SOs for that.

------------------------------------------------------------------------------

[Ghannouchi, Youssef]

Hello Lev,

The MSP Debug Stack is an open source. Please refer to this link: http://www.ti.com/tool/mspds Please see README-BUILD.txt for build instructions.

I will double check the gdb_agent_console license and let you know asap.

Best regards,
Youssef Ghannouchi


Texas Instruments Deutschland GmbH, Haggertystr. 1, D-85356 Freising. Amtsgericht M?nchen HRB 40960. Gesch?ftsf?hrer: Dr. Wolfram Tietscher. Vorsitzender des Aufsichtsrates: Andreas Schwaiger

-----Original Message-----
From: DJ Delorie [mailto:***@redhat.com]
Sent: Monday, October 19, 2015 12:21 AM
To: Lev Serebryakov
Cc: Ghannouchi, Youssef; mspgcc-***@lists.sourceforge.net; Miller, Greg
Subject: Re: [Mspgcc-users] New MSP430 GCC version release available!

Agenda VR3 (did you remember this very first Linuix-based PDA with
MIPS-based CPU?)

Yup, I was involved with the project way back then.

Could you please clear situation with debugging (in separate thread
in mailing list)?

Sorry, that's a TI question, they provide the DLLs and SOs for that.


------------------------------------------------------------------------------

[Lev Serebryakov]

Hello Youssef,

Post by Ghannouchi, Youssef
http://www.ti.com/tool/mspds Please see README-BUILD.txt for build instructions.

Yep, I was able to build it on FreeBSD with some small additions to
Makefiles (adding new $(PLATFORM) branch). Sources were compiled without any
problems!

Post by Ghannouchi, Youssef
I will double check the gdb_agent_console license and let you know asap.

Thank you! Looks like it is last component to build full-featured MSP430
developemnt stack for yet another host platform :)

[Lev Serebryakov]

Hello DJ,

Post by DJ Delorie
That should be built as part of building gcc/
so you probably have a problem earlier on that you didn't see in your logs...

Other problem: tcl doesn't respect DESTDIR setting, so whole
package could not be properly "staged" for package building, "tcl" try to
install itself directly to $(PREFIX).

Same problem exists in ictl/* and tk.

Also, "libgloss" doesn't respect DESTDIR.

It could be patched, though.