This is the mail archive of the ecos-discuss@sources.redhat.com mailing list for the eCos project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Synth environment malloc seg faults under Fedora Linux


Bart,

Thanks for the help.  One last piece of information, and then I'll lay
this one to rest on my side: I've got the synth working now on my
Fedora box with gcc-3.2.1 (to match my arm-elf cross-compiler).  Here's
what I did:

1) Extracted the gcc-core and gcc-g++ 3.2.1 source.

2) Applied relevant patches from the eCos toolchain build instructions.

3) Did a "configure --prefix <...>".

4) Did a "make -w all install"

5) Put the new gcc bin directory at the head of my path.


Note that I did not replace my binutils or use newlib or any of that. 
All I did was replace my gcc 3.3.2 with a "generic" native 3.2.1 build.
 So it looks to be strictly a problem with gcc.

Thanks again for the help,

--
Dan Jakubiec
Systech Corp


--- Bart Veer <bartv@ecoscentric.com> wrote:
> >>>>> "Dan" == Dan Jakubiec <djakubiec@yahoo.com> writes:
> 
>     Dan> Hi Bart,
>     Dan> Thanks for the suggestions. I went ahead and built a synth
>     Dan> executable under the older RH9 system using the older
>     Dan> toolchain. I used the same eCos code base I've been working
>     Dan> with under Fedora. I then copied the resulting executable to
>     Dan> my Fedora box and sure enough: it worked! Good call!
> 
>     Dan> So it appears that the new GNU toolchain is causing me some
>     Dan> strange problems with the synth environment build. Just FYI,
>     Dan> my gcc versions are:
> 
>     Dan> RH9, works:
>     Dan>  gcc (GCC) 3.2.2 20030222 (Red Hat Linux 3.2.2-5)
> 
>     Dan> Fedora, doesn't: 
>     Dan>  gcc (GCC) 3.3.2 20031022 (Red Hat Linux 3.3.2-1)
> 
>     Dan> I suppose the "quick" solution for me is to install the gcc
>     Dan> 3.2.2 compiler on my Fedora box and use that for synth
>     Dan> builds. This should be fine for my purposes.
> 
> That is what I would recommend for now. Note that the compiler is the
> most likely culprit, but it could also be the linker or some other
> binutils utility.
> 
>     Dan> Is there someone in particular who deals with the compiler
>     Dan> version migration issues? Can I provide any more information
>     Dan> about this or help out in some way?
> 
> This is not the first time there have been problems. The issue is
> that
> the native gcc is intended to produce ordinary Linux applications
> linked with glibc. Hence the toolchain makes certain assumptions
> about
> the run-time environment which do not hold for the synthetic target.
> The Linux kernel tends to suffer similar problems.
> 
> Of course it could also be a more generic problem with eCos vs. gcc
> 3.3.2, but I have not seen any other problem reports.
> 
> There is no easy way to track down and fix these problems. I do have
> plans to upgrade one of my machines to Fedora, but don't know exactly
> when yet. If nobody else has fixed things by then I'll investigate
> further.
> 
> Bart
> 
> -- 
> Bart Veer                       eCos Configuration Architect
> http://www.ecoscentric.com/     The eCos and RedBoot experts


__________________________________
Do you Yahoo!?
Yahoo! SiteBuilder - Free web site building tool. Try it!
http://webhosting.yahoo.com/ps/sb/

-- 
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]