[ECOS] "new" heap allocation for synthetic target doesn't work for me

[ECOS] "new" heap allocation for synthetic target doesn't work for me

[Alexander Neundorf]

Hi,

the "new" (ok, already 1 1/2 years old) way to allocate the memory for the 
heap doesn't work for me (kubuntu 6.10, gcc 4.1.2), i.e. malloc() fails with 
a SIGSEGV in dlmalloc.cxx, line 1282. Reverting this change:
http://ecos.sourceware.org/cgi-bin/cvsweb.cgi/ecos/packages/hal/synth/arch/current/src/synth_entry.c.diff?r1=1.6&r2=1.7&cvsroot=ecos&f=h

makes it work again. 

Alex

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

Re: [ECOS] "new" heap allocation for synthetic target doesn't work for me

[Andrew Lunn]

On Wed, Jul 11, 2007 at 08:55:00AM -0400, Alexander Neundorf wrote:
> Hi,
> 
> the "new" (ok, already 1 1/2 years old) way to allocate the memory for the 
> heap doesn't work for me (kubuntu 6.10, gcc 4.1.2), i.e. malloc() fails with 
> a SIGSEGV in dlmalloc.cxx, line 1282. Reverting this change:
> http://ecos.sourceware.org/cgi-bin/cvsweb.cgi/ecos/packages/hal/synth/arch/current/src/synth_entry.c.diff?r1=1.6&r2=1.7&cvsroot=ecos&f=h
> 
> makes it work again. 

Do we know what makes the difference. Is it gcc, binutils, kernel?

We need some way to decide at runtime if we need to brk() memory or
not.

        Andrew

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

Re: [ECOS] "new" heap allocation for synthetic target doesn't work for me

[Alexander Neundorf]

On Wednesday 11 July 2007 09:02, Andrew Lunn wrote:
> On Wed, Jul 11, 2007 at 08:55:00AM -0400, Alexander Neundorf wrote:
> > Hi,
> >
> > the "new" (ok, already 1 1/2 years old) way to allocate the memory for
> > the heap doesn't work for me (kubuntu 6.10, gcc 4.1.2), i.e. malloc()
> > fails with a SIGSEGV in dlmalloc.cxx, line 1282. Reverting this change:
> > http://ecos.sourceware.org/cgi-bin/cvsweb.cgi/ecos/packages/hal/synth/arc
> >h/current/src/synth_entry.c.diff?r1=1.6&r2=1.7&cvsroot=ecos&f=h
> >
> > makes it work again.
>
> Do we know what makes the difference. Is it gcc, binutils, kernel?

I don't know.

> We need some way to decide at runtime if we need to brk() memory or
> not.

At least the values reported by mallinfo() were ok in both cases.

Alex

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