I'm taking this back to tinyos-devel: I don't subscribe to so can't post to tinyos-help (my reply to your original email is sitting somewhere), and this should be worked out in a smaller circle before inviting the whole world to try it.
I maintain mspgcc, but don't build distributions of it for any platform, let alone TinyOS. I don't use Ubuntu, so can't even test the TinyOS-built packages without a lot of extra work. I and others can confirm that mspgcc 20110716 does work with TinyOS, though at the moment I don't use TinyOS for anything so my testing of it is somewhat limited. With that in mind, I can only say that you need the msp430 variants of binutils, gcc, and libc, as well as the msp430mcu package, in order to build. msp430.h comes from msp430mcu. The error you get suggests you don't have msp430mcu installed. There should be no need for special environment variables other than to put the installation bin directory into your PATH. Peter On Fri, Aug 19, 2011 at 3:35 AM, Kevin Daly <[email protected]> wrote: > Hi, > > Thanks for the info. > > A quick test this morning shows something is very broken for me. > > Code that was compiling and working no problem for me yesterday is not > compiling for me now. I'm 100% certain that the only difference is the > updates to msp430-gcc, msp430-libc-tinyos and msp430-binutils-tinyos from > the stanford repository (ttp://tinyos.stanford.edu/tinyos/dists/ubuntulucid > main). I don't have a record of the versions I was running yesterday, > but it was the version that was in the repository before these. > > I updated tinyos-main from trunk this morning (after the first failed > compiling attempt) so I'm using revision 5695. > > See the attached text file for the errors I get when compiling Null for > telosb. > The first error that jumps out at me is msp430.h is not being found, but I > do have it at > /usr/lib/gcc/msp430/4.5.3/plugin/include/config/msp430/msp430.h. Any idea as > to why the build system is not finding this. Am I missing an environment > variable or what? Or any other ideas as to what is going wrong? > > Versions I have installed now: > msp430-gcc: 4.5.3-20110811 > msp430-libc-tinyos: 20110612-20110811 > msp430-binutils-tinyos: 2.21.1-20110811 > > I'll be very surprised if this problem is specific to my system, as I would > have thought it was fairly standard (on Ubuntu 10.04 64-bit). > > Thanks, > Kev > > > > > On Fri, Aug 19, 2011 at 1:23 AM, Eric Decker <[email protected]> wrote: > >> >> >> On Thu, Aug 18, 2011 at 12:54 PM, Kevin Daly <[email protected]>wrote: >> >>> Hi, >>> >>> Moving this to tinyos-help, as I'm sure it is is a more appropriate >>> location for this question. >>> >>> In practical terms, for those of us who use msp430-gcc, but are not >>> platform maintainers, what does this change to watchdog timer behaviour >>> mean? Is there some change needed to the core platform files before this >>> version of msp430-gcc can be used? >>> >> >> It depends on your platform code. Peter put in modifications for >> existing platforms that preserves the existing old toolchain behaviour. >> Namely it disables the watchdog. >> >> In practice for production systems, the watchdog should get used because >> it protects against the code going south in a major way. But for the >> current trunk it is disabled. This was inherited from the way the old >> toolchain worked. >> >> In the SVN trunk, the code is there for the shimmer, span, and telosb >> platforms. Those are the ones I checked. I'm pretty sure that Peter got >> all the major msp430 platforms supported in the trunk. >> >> >> >>> The package manager on my linux box just updated the msp430-gcc version >>> to 4.x when I wasn't paying attention, so, what do I need to do to stop the >>> watchdog timer from breaking all my apps? Is there a flag I can add to the >>> makefile re-disable the watchdog timer if it proves problematic, or do I >>> need to disable it in Boot.booted() (and if so exaclty how), or if I did >>> want to tickle it, what is the easiest way to do this? I'll go read through >>> the msp430f1611 user guide in the morning, but i'm on a fairly tight >>> deadline at the moment so any advice is welcome. >>> >> >> If you are using the main trunk you should be okay. Or take a look at >> the main trunk and search for WDTCTL. You'll see where to put it if you >> aren't on the main trunk. >> >> Take a look at tos/platforms/shimmer2/PlatformP.nc, look for WDTCTL. >> >> >>> Incidentally, if it matters, I'm primarily using telosbs and tmotes, with >>> a few shimmers scattered around. >>> >>> I could revert back to the version 3.x msp430-gcc, but the hardware >>> multiplier fix could be fairly compelling for me. >>> >> >> It would be better to use the new toolchain. The more time we get on it >> the better. It also has significantly better optimization. >> >> >>> Thanks in advance, >>> Kevin >>> >>> On Thu, Aug 18, 2011 at 5:51 PM, Peter Bigot <[email protected]> wrote: >>> >>>> To save time and questions, I'd like to remind those responsible for >>>> MSP430-based platforms that watchdog is NOT disabled on startup >>>> anymore<https://www.millennium.berkeley.edu/pipermail/tinyos-devel/2011-June/004952.html>, >>>> and that hardware multiply support should be fully >>>> stable<http://www.mail-archive.com/[email protected]/msg09978.html>and >>>> -mdisable-hwmul should be removed from any Makefiles or other places >>>> they might be hiding. >>>> >>>> There is a bug in the original LTS-20110716 mspgcc release that affects >>>> multiplication when using emulation instead of the hardware multiplier (in >>>> other words, -mdisable-hwmul will generate bad code while removing it will >>>> not). This was discovered in the last few days and is probably not in the >>>> uploaded toolchain packages. >>>> >>>> The fix for this and all other known LTS-20110716 releases are available >>>> as >>>> patches<https://sourceforge.net/projects/mspgcc/files/Patches/LTS/20110716/>. >>>> If you browse up a directory from that link you'll see instructions for >>>> using git to track the current set of LTS patches. >>>> >>>> If you determine that the new mspgcc generates bad code, please report >>>> the problem here or on the mspgcc bug >>>> tracker<https://sourceforge.net/tracker/?group_id=42303&atid=432701> >>>> . >>>> >>>> Peter >>>> >>>> On Thu, Aug 18, 2011 at 9:40 AM, Philip Levis <[email protected]>wrote: >>>> >>>>> The first step for the 2.1.2 release is updating the compiler >>>>> toolchains to support newer microcontrollers (and fix many bugs!). >>>>> Specifically, the TinyOS Ubuntu repository on tinyos.stanford.edu will >>>>> include new compiler toolchains for the AVR, MSP430, and SAM3 Cortex M3. >>>>> >>>>> The new MSP430 and AVR toolchains are up in the repository now. I'd >>>>> suggest that everyone switch over: 2.1.2 may not work with the old tools >>>>> (e.g., the old MSP430 compiler requires hardware multiply to be disabled >>>>> or >>>>> it generates bad assembly sometimes, 2.1.2 will not disable it). As a >>>>> short-term benefit, you should generally see smaller code when using the >>>>> new >>>>> compilers. >>>>> >>>>> Once Thomas and Razvan have the SAM3 packages ready, I'll tag the >>>>> TinyOS tree for 2.1.2 and we'll start testing. >>>>> >>>>> Phil >>>>> _______________________________________________ >>>>> Tinyos-2.0wg mailing list >>>>> [email protected] >>>>> >>>>> https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-2.0wg >>>>> >>>> >>>> >>>> _______________________________________________ >>>> Tinyos-devel mailing list >>>> [email protected] >>>> >>>> https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-devel >>>> >>>> >>> >>> _______________________________________________ >>> Tinyos-help mailing list >>> [email protected] >>> https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help >>> >> >> >> >> -- >> Eric B. Decker >> Senior (over 50 :-) Researcher >> >> >> >
_______________________________________________ Tinyos-devel mailing list [email protected] https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-devel
