Hi All,

In brief : POV-ray compiles fine, segmentation faults at runtime.

Setup :

PPC 7400 (G4) with
    YellowDog Linux ,
    2.4.20 kernel
    gcc 3.2,
    jpeg-6b,
    libpng-1.2.5,
    tiff-v3.5.7,
    zlib-1.1.4,
    povray-3.50c

Makefile mods :
src/Makefile :
DEFS
= -DPREFIX=\"$(prefix)\" -DPOV_LIB_DIR=\"${prefix}/share/povray-3.5\" -DCOMP
ILER_VER=\".Linux.$(CC)\" -DSYSCONFDIR=\"$(sysconfdir)\" -DUSE_IO_RESTRICTIO
NS=\"$(ac_use_io_restrictions)\" -DX_DISPLAY_MISSING=1
CFLAGS
= -O3 -mcpu=7400 -finline-functions -ffast-math -fomit-frame-pointer -funrol
l-loops -fexpensive-optimizations -foptimize-sibling-calls -Wno-multichar
CXXFLAGS =
$(NOMULTICHAR) -O3 -mcpu=7400  -finline-functions -ffast-math -fomit-frame-p
ointer -funroll-loops -fexpensive-optimizations -foptimize-sibling-calls -Wn
o-multichar

When rendering the abyss demo with :

povray +F -D abyss.pov (slightly adapted, i admit)

i get :

Creating bounding slabs.
Scene contains 44551 frame level objects; 1 infinite.

Creating vista buffer.
Creating light buffers
  -:--:-- Rendering line    1 of  240. Segmentation fault

Post a reply to this message

Sorry, hit the send button before completing the post.

It should read :

Any help MOSTLY appreciated, thanks !



bye !

Jan Dewerchin



"Jan Dewerchin" <jdw### [at] fabricom-idsbe> wrote in message
news:3e1b082d$1@news.povray.org...
> Hi All,
>
> In brief : POV-ray compiles fine, segmentation faults at runtime.
>
> Setup :
>
> PPC 7400 (G4) with
>     YellowDog Linux ,
>     2.4.20 kernel
>     gcc 3.2,
>     jpeg-6b,
>     libpng-1.2.5,
>     tiff-v3.5.7,
>     zlib-1.1.4,
>     povray-3.50c
>
> Makefile mods :
> src/Makefile :
> DEFS
>
= -DPREFIX=\"$(prefix)\" -DPOV_LIB_DIR=\"${prefix}/share/povray-3.5\" -DCOMP
>
ILER_VER=\".Linux.$(CC)\" -DSYSCONFDIR=\"$(sysconfdir)\" -DUSE_IO_RESTRICTIO
> NS=\"$(ac_use_io_restrictions)\" -DX_DISPLAY_MISSING=1
> CFLAGS
>
= -O3 -mcpu=7400 -finline-functions -ffast-math -fomit-frame-pointer -funrol
> l-loops -fexpensive-optimizations -foptimize-sibling-calls -Wno-multichar
> CXXFLAGS =
>
$(NOMULTICHAR) -O3 -mcpu=7400  -finline-functions -ffast-math -fomit-frame-p
>
ointer -funroll-loops -fexpensive-optimizations -foptimize-sibling-calls -Wn
> o-multichar
>
> When rendering the abyss demo with :
>
> povray +F -D abyss.pov (slightly adapted, i admit)
>
> i get :
>
> Creating bounding slabs.
> Scene contains 44551 frame level objects; 1 infinite.
>
> Creating vista buffer.
> Creating light buffers
>   -:--:-- Rendering line    1 of  240. Segmentation fault
>
>
>
>
>

Post a reply to this message

One more comment :

I only get this with abyss.pov until now....

strange........

Jan Dewerchin

"Jan Dewerchin" <jdw### [at] fabricom-idsbe> wrote in message
news:3e1b082d$1@news.povray.org...
> Hi All,
>
> In brief : POV-ray compiles fine, segmentation faults at runtime.
>
> Setup :
>
> PPC 7400 (G4) with
>     YellowDog Linux ,
>     2.4.20 kernel
>     gcc 3.2,
>     jpeg-6b,
>     libpng-1.2.5,
>     tiff-v3.5.7,
>     zlib-1.1.4,
>     povray-3.50c
>
> Makefile mods :
> src/Makefile :
> DEFS
>
= -DPREFIX=\"$(prefix)\" -DPOV_LIB_DIR=\"${prefix}/share/povray-3.5\" -DCOMP
>
ILER_VER=\".Linux.$(CC)\" -DSYSCONFDIR=\"$(sysconfdir)\" -DUSE_IO_RESTRICTIO
> NS=\"$(ac_use_io_restrictions)\" -DX_DISPLAY_MISSING=1
> CFLAGS
>
= -O3 -mcpu=7400 -finline-functions -ffast-math -fomit-frame-pointer -funrol
> l-loops -fexpensive-optimizations -foptimize-sibling-calls -Wno-multichar
> CXXFLAGS =
>
$(NOMULTICHAR) -O3 -mcpu=7400  -finline-functions -ffast-math -fomit-frame-p
>
ointer -funroll-loops -fexpensive-optimizations -foptimize-sibling-calls -Wn
> o-multichar
>
> When rendering the abyss demo with :
>
> povray +F -D abyss.pov (slightly adapted, i admit)
>
> i get :
>
> Creating bounding slabs.
> Scene contains 44551 frame level objects; 1 infinite.
>
> Creating vista buffer.
> Creating light buffers
>   -:--:-- Rendering line    1 of  240. Segmentation fault
>
>
>
>
>

Post a reply to this message

In article <3e1b082d$1@news.povray.org> , "Jan Dewerchin" 
<jdw### [at] fabricom-idsbe> wrote:

> In brief : POV-ray compiles fine, segmentation faults at runtime.
>
> Setup :
>
> PPC 7400 (G4) with

Do you really have a G4 or did you only compile for it?

    Thorsten

Post a reply to this message

Hi Thorsten,

It's a Totalimpact Briq G4 with altivec.

(check www.totalimpact.com to see it, if you like)

It's as G4 as a G4 gets, i guess.

I would like to stress that I get this problem ONLY with abyss.pov .
As I am a complete newbie on POV-ray, I was rendering some intresting
scenes,
and abyss.pov was the only one wich gave me trouble until now.

At first I thought it was because I added a clock related camera movement
(360 spin around the sub),
but POV-ray halts whith every imaginable option.

Mind you, the same scene renders just fine with the windows version.

Jan Dewerchin

"Thorsten Froehlich" <tho### [at] trfde> wrote in message
news:3e1b27cb@news.povray.org...
> In article <3e1b082d$1@news.povray.org> , "Jan Dewerchin"
> <jdw### [at] fabricom-idsbe> wrote:
>
> > In brief : POV-ray compiles fine, segmentation faults at runtime.
> >
> > Setup :
> >
> > PPC 7400 (G4) with
>
> Do you really have a G4 or did you only compile for it?
>
>     Thorsten

Post a reply to this message

In article <3e1be87e$1@news.povray.org> , "Jan Dewerchin" 
<jdw### [at] fabricom-idsbe> wrote:

> It's as G4 as a G4 gets, i guess.

Did you compile all libraries for the G4 as well?

> I would like to stress that I get this problem ONLY with abyss.pov .

How about using the debugger?  That will reveal where your compile crashes.

> Mind you, the same scene renders just fine with the windows version.

Well, that is an official compile and obviously it has been tested with all
sample scenes...

    Thorsten

____________________________________________________
Thorsten Froehlich, Duisburg, Germany
e-mail: tho### [at] trfde

Visit POV-Ray on the web: http://mac.povray.org

Post a reply to this message

Thorsten,

Most likely, the problem lies in one of the external libraries as you
pointed out.

What puzzles me is the fact that apparently, there is a special feature used
in abyss.pov
that isn't used in other scenes. Or perhaps it's the number of objects in
the scene.

The thing I will look into is the fatc that I get loads of warnings like :
bsphere.cpp:516: warning: assignment to `int' from `double'

I'll try to fix this after I recompiled the libs for G4.

Thanks for your comments !

Jan Dewerchin.


"Thorsten Froehlich" <tho### [at] trfde> wrote in message
news:3e1bf1c2@news.povray.org...
> In article <3e1be87e$1@news.povray.org> , "Jan Dewerchin"
> <jdw### [at] fabricom-idsbe> wrote:
>
> > It's as G4 as a G4 gets, i guess.
>
> Did you compile all libraries for the G4 as well?
>
> > I would like to stress that I get this problem ONLY with abyss.pov .
>
> How about using the debugger?  That will reveal where your compile
crashes.
>
> > Mind you, the same scene renders just fine with the windows version.
>
> Well, that is an official compile and obviously it has been tested with
all
> sample scenes...
>
>     Thorsten
>
> ____________________________________________________
> Thorsten Froehlich, Duisburg, Germany
> e-mail: tho### [at] trfde
>
> Visit POV-Ray on the web: http://mac.povray.org

Post a reply to this message

Thorsten,

The problem is located here, according to gdb :

#0  0x1009c9a8 in buildsturm (ord=-25, sseq=0x7fffd7b8) at polysolv.cpp:273
273           *fp /= f;

Tracing backwards I found a very strange situation :

#0  0x1009c9a8 in buildsturm (ord=-25, sseq=0x7fffd7b8) at polysolv.cpp:273
#1  0x1009c91c in buildsturm (ord=-25, sseq=0x7fffd7b8) at polysolv.cpp:265
#2  0x1009bb68 in Solve_Polynomial (n=2147473744, c0=0x10100000,
r=0x7fffe0d8,
    sturm=269481264, epsilon=-0) at polysolv.cpp:1674
#3  0x10046b18 in intersect_lathe (Ray=0x7fffe898, Lathe=0x108c6fb0,
    Depth_Stack=0x10302f10) at lathe.cpp:440

look at #2, the value of n (2147473744), where all the calls to
Solve_Polynomial are done with values
2,4 or 6 in the function intersect_lathe. Notice the value of sturm
(269481264). Looks like a pretty serious corruption to me.

It's obvious that the corruption is not caused by a coding error , but by
some compiler setting. I'll try to build without the -mcpu setting.

Any further ideas ?

Jan Dewerchin





"Thorsten Froehlich" <tho### [at] trfde> wrote in message
news:3e1bf1c2@news.povray.org...
> In article <3e1be87e$1@news.povray.org> , "Jan Dewerchin"
> <jdw### [at] fabricom-idsbe> wrote:
>
> > It's as G4 as a G4 gets, i guess.
>
> Did you compile all libraries for the G4 as well?
>
> > I would like to stress that I get this problem ONLY with abyss.pov .
>
> How about using the debugger?  That will reveal where your compile
crashes.
>
> > Mind you, the same scene renders just fine with the windows version.
>
> Well, that is an official compile and obviously it has been tested with
all
> sample scenes...
>
>     Thorsten
>
> ____________________________________________________
> Thorsten Froehlich, Duisburg, Germany
> e-mail: tho### [at] trfde
>
> Visit POV-Ray on the web: http://mac.povray.org

Post a reply to this message

In article <3e1c2b0d@news.povray.org> , "Jan Dewerchin" 
<jdw### [at] fabricom-idsbe> wrote:

> The thing I will look into is the fatc that I get loads of warnings like :
> bsphere.cpp:516: warning: assignment to `int' from `double'

Don't worry about those warnings, just turn them off.

    Thorsten

____________________________________________________
Thorsten Froehlich, Duisburg, Germany
e-mail: tho### [at] trfde

Visit POV-Ray on the web: http://mac.povray.org

Post a reply to this message