NVIZ Segmentation Fault

Philip_Warner · February 21, 2000, 11:51pm

I may be being optimistic, but I am running Grass on the beta release of
SuSE Linux for the PowerPC. So far, it all seems to build and run fine (not
that I'd know if there was a problem, but it seems to do what I ask).

However, NVIZ segfaults whenever I try to run it. Nothing is displayed, and
no error messages (other than seg fault). If I try to run NVIZ without X or
Grass, it complains, so I assume that it is at least getting as far as
checking it's environment etc.

I vaguely recall reading somewhere that NVIS had certain video hardware
requirements, but can no longer find anything to that effect. If this is
the case, then the PPC probably does not meet those requirements, and the
problem is solved.

Any helpful suggestions would be greatly appreciated - even if they are
only suggestions as to how I can find out more...

Thanks,

Philip Warner.

----------------------------------------------------------------
Philip Warner | __---_____
Albatross Consulting Pty. Ltd. |----/ - \
(A.C.N. 008 659 498) | /(@) ______---_
Tel: +61-03-5367 7422 | _________ \
Fax: +61-03-5367 7430 | ___________ |
Http://www.rhyme.com.au | / \|
| --________--
PGP key available upon request, | /
and from pgp5.ai.mit.edu:11371 |/

Eric_G_Miller · February 22, 2000, 1:06am

On Tue, Feb 22, 2000 at 10:51:11AM +1100, Philip Warner wrote:

I may be being optimistic, but I am running Grass on the beta release of
SuSE Linux for the PowerPC. So far, it all seems to build and run fine (not
that I'd know if there was a problem, but it seems to do what I ask).

However, NVIZ segfaults whenever I try to run it. Nothing is displayed, and
no error messages (other than seg fault). If I try to run NVIZ without X or
Grass, it complains, so I assume that it is at least getting as far as
checking it's environment etc.

Been having the same problem here for months!

I vaguely recall reading somewhere that NVIS had certain video hardware
requirements, but can no longer find anything to that effect. If this is
the case, then the PPC probably does not meet those requirements, and the
problem is solved.

Don't think it's PPC specific since my machine is i686/Linux.

Any helpful suggestions would be greatly appreciated - even if they are
only suggestions as to how I can find out more...

Basically, I think the NVIZ code has some problems. Sure it compiles,
but I get many warnings about incompatible pointer types (clue 1). My
last strace on NVIZ seemed to indicate it fails shortly after loading
the tk.tcl interface. It appears to send an inappropriate IO request:

access("/usr/lib/tk8.0/tk.tcl", F_OK) = 0
stat("/usr/lib/tk8.0/tk.tcl", {st_mode=S_IFREG|0644, st_size=5697, ...})
= 0
open("/usr/lib/tk8.0/tk.tcl", O_RDONLY) = 7
fcntl(7, F_SETFD, FD_CLOEXEC) = 0
ioctl(7, TCGETS, 0xbfffd8c8) = -1 ENOTTY (Inappropriate ioctl
for device)
read(7, "# tk.tcl --\n#\n# Initialization s"..., 4096) = 4096
read(7, "vent add <<Clear>> <Clear>\n }"..., 4096) = 1601
close(7) = 0
--- SIGSEGV (Segmentation fault) ---
+++ killed by SIGSEGV +++

I wonder if ENOTTY has any relation to UNIX 98 PTYs? Just speculatin'...

--
+----------------------------------------------------+
| Eric G. Miller egm2@jps.net |
| GnuPG public key: http://www.jps.net/egm2/gpg.asc |
+----------------------------------------------------+

Philip_Warner · February 22, 2000, 1:25am

At 17:06 21/02/00 -0800, Eric G . Miller wrote:

On Tue, Feb 22, 2000 at 10:51:11AM +1100, Philip Warner wrote:

I vaguely recall reading somewhere that NVIS had certain video hardware
requirements, but can no longer find anything to that effect. If this is
the case, then the PPC probably does not meet those requirements, and the
problem is solved.

Don't think it's PPC specific since my machine is i686/Linux.

This is good to know!

Any helpful suggestions would be greatly appreciated - even if they are
only suggestions as to how I can find out more...

Basically, I think the NVIZ code has some problems. Sure it compiles,
but I get many warnings about incompatible pointer types (clue 1).

Ditto. Lots of software produces these errors, and I assumed that they
would be OK. I also vaguely recall an 'unwritable parameter' - I probably
should go back and find it...

My
last strace on NVIZ seemed to indicate it fails shortly after loading
the tk.tcl interface. It appears to send an inappropriate IO request:

access("/usr/lib/tk8.0/tk.tcl", F_OK) = 0
stat("/usr/lib/tk8.0/tk.tcl", {st_mode=S_IFREG|0644, st_size=5697, ...})
= 0
open("/usr/lib/tk8.0/tk.tcl", O_RDONLY) = 7
fcntl(7, F_SETFD, FD_CLOEXEC) = 0
ioctl(7, TCGETS, 0xbfffd8c8) = -1 ENOTTY (Inappropriate ioctl
for device)
read(7, "# tk.tcl --\n#\n# Initialization s"..., 4096) = 4096
read(7, "vent add <<Clear>> <Clear>\n }"..., 4096) = 1601
close(7) = 0
--- SIGSEGV (Segmentation fault) ---
+++ killed by SIGSEGV +++

This is also good to see - an strace on my machine looks pretty close to
identical. I guess I need to recompile and look for any suspect messages.

I wonder if ENOTTY has any relation to UNIX 98 PTYs? Just speculatin'...

Hopefully there is an NVIZ person out there who may recognize where this is
happening.

Thanks for your help,

Philip Warner.

----------------------------------------------------------------
Philip Warner | __---_____
Albatross Consulting Pty. Ltd. |----/ - \
(A.C.N. 008 659 498) | /(@) ______---_
Tel: +61-03-5367 7422 | _________ \
Fax: +61-03-5367 7430 | ___________ |
Http://www.rhyme.com.au | / \|
| --________--
PGP key available upon request, | /
and from pgp5.ai.mit.edu:11371 |/

Philip_Warner · February 25, 2000, 9:48am

At 10:17 25/02/00 +0000, Markus Neteler wrote:

Hi there,

I am still searching for someone with programming knowledge to
fix this

What I found here:

The GNU C Library
http://gnu.senet.com.au/glibc/libc_2.html

Macro: int ENOTTY
Inappropriate I/O control operation, such as trying to set terminal
modes on an ordinary file.

Maybe this is useful to somebody. Could it be a wrong
sprintf, fprintf call somewhere in
src.contrib/GMSL/NVIZ2.2/src/
?

Note: On some Linux boxes it works (luckily on my system).

Thanks for the reply.

I am happy to try and track it down if there is no 'NVIZ-person' available.
But I have a problem: under Linux 2.2.7/i686 box/SuSE 6.1 it works fine,
and on my PPC/SuSE 6.3beta box it dies. Unfortunately, the PPC debugger
seems not to work.

If the people who who get the error on an Intel machine can tell me the
Linux version etc that they are running, then I will try to reproduce it.

Looking forward to hearing from people...

Philip Warner.

> My
>last strace on NVIZ seemed to indicate it fails shortly after loading
>the tk.tcl interface. It appears to send an inappropriate IO request:
>
>access("/usr/lib/tk8.0/tk.tcl", F_OK) = 0
>stat("/usr/lib/tk8.0/tk.tcl", {st_mode=S_IFREG|0644, st_size=5697, ...})
>= 0
>open("/usr/lib/tk8.0/tk.tcl", O_RDONLY) = 7
>fcntl(7, F_SETFD, FD_CLOEXEC) = 0
>ioctl(7, TCGETS, 0xbfffd8c8) = -1 ENOTTY (Inappropriate ioctl
>for device)
>read(7, "# tk.tcl --\n#\n# Initialization s"..., 4096) = 4096
>read(7, "vent add <<Clear>> <Clear>\n }"..., 4096) = 1601
>close(7) = 0
>--- SIGSEGV (Segmentation fault) ---
>+++ killed by SIGSEGV +++

This is also good to see - an strace on my machine looks pretty close to
identical. I guess I need to recompile and look for any suspect messages.

>I wonder if ENOTTY has any relation to UNIX 98 PTYs? Just speculatin'...

Hopefully there is an NVIZ person out there who may recognize where this is
happening.

----------------------------------------------------------------
Philip Warner | __---_____
Albatross Consulting Pty. Ltd. |----/ - \
(A.C.N. 008 659 498) | /(@) ______---_
Tel: +61-03-5367 7422 | _________ \
Fax: +61-03-5367 7430 | ___________ |
Http://www.rhyme.com.au | / \|
| --________--
PGP key available upon request, | /
and from pgp5.ai.mit.edu:11371 |/

Markus_Neteler3 · February 25, 2000, 10:17am

Hi there,

> My
>last strace on NVIZ seemed to indicate it fails shortly after loading
>the tk.tcl interface. It appears to send an inappropriate IO request:
>
>access("/usr/lib/tk8.0/tk.tcl", F_OK) = 0
>stat("/usr/lib/tk8.0/tk.tcl", {st_mode=S_IFREG|0644, st_size=5697, ...})
>= 0
>open("/usr/lib/tk8.0/tk.tcl", O_RDONLY) = 7
>fcntl(7, F_SETFD, FD_CLOEXEC) = 0
>ioctl(7, TCGETS, 0xbfffd8c8) = -1 ENOTTY (Inappropriate ioctl
>for device)
>read(7, "# tk.tcl --\n#\n# Initialization s"..., 4096) = 4096
>read(7, "vent add <<Clear>> <Clear>\n }"..., 4096) = 1601
>close(7) = 0
>--- SIGSEGV (Segmentation fault) ---
>+++ killed by SIGSEGV +++

This is also good to see - an strace on my machine looks pretty close to
identical. I guess I need to recompile and look for any suspect messages.

>I wonder if ENOTTY has any relation to UNIX 98 PTYs? Just speculatin'...

Hopefully there is an NVIZ person out there who may recognize where this is
happening.

I am still searching for someone with programming knowledge to
fix this

What I found here:

The GNU C Library
http://gnu.senet.com.au/glibc/libc_2.html

Macro: int ENOTTY
Inappropriate I/O control operation, such as trying to set terminal
modes on an ordinary file.

Maybe this is useful to somebody. Could it be a wrong
sprintf, fprintf call somewhere in
src.contrib/GMSL/NVIZ2.2/src/
?

Note: On some Linux boxes it works (luckily on my system).

Regards

Markus Neteler

Kurt_Gerber2 · February 25, 2000, 9:15pm

Hi Philipp
I have the same problem. I was running Suse 6.0 on Linux 2.2.9 on an Intel
Pentium II and Nviz2.2 worked fine. Then I upgraded to Suse 6.3 and Linux
2.2.13 and since the Nviz2.2 doesn't work anymore.
Since some weeks I'm mailing with Markus Neteler to try to track down the
problem....

Kurt Gerber

On Fre, 25 Feb 2000, you wrote:

At 10:17 25/02/00 +0000, Markus Neteler wrote:

Hi there,

I am still searching for someone with programming knowledge to
fix this

What I found here:

The GNU C Library
http://gnu.senet.com.au/glibc/libc_2.html

Macro: int ENOTTY
   Inappropriate I/O control operation, such as trying to set terminal
   modes on an ordinary file.

Maybe this is useful to somebody. Could it be a wrong
sprintf, fprintf call somewhere in
src.contrib/GMSL/NVIZ2.2/src/
?

Note: On some Linux boxes it works (luckily on my system).

Thanks for the reply.

I am happy to try and track it down if there is no 'NVIZ-person' available.
But I have a problem: under Linux 2.2.7/i686 box/SuSE 6.1 it works fine,
and on my PPC/SuSE 6.3beta box it dies. Unfortunately, the PPC debugger
seems not to work.

If the people who who get the error on an Intel machine can tell me the
Linux version etc that they are running, then I will try to reproduce it.

Looking forward to hearing from people...

Philip Warner.

> My
>last strace on NVIZ seemed to indicate it fails shortly after loading
>the tk.tcl interface. It appears to send an inappropriate IO request:
>
>access("/usr/lib/tk8.0/tk.tcl", F_OK) = 0
>stat("/usr/lib/tk8.0/tk.tcl", {st_mode=S_IFREG|0644, st_size=5697, ...})
>= 0
>open("/usr/lib/tk8.0/tk.tcl", O_RDONLY) = 7
>fcntl(7, F_SETFD, FD_CLOEXEC) = 0
>ioctl(7, TCGETS, 0xbfffd8c8) = -1 ENOTTY (Inappropriate ioctl
>for device)
>read(7, "# tk.tcl --\n#\n# Initialization s"..., 4096) = 4096
>read(7, "vent add <<Clear>> <Clear>\n }"..., 4096) = 1601
>close(7) = 0
>--- SIGSEGV (Segmentation fault) ---
>+++ killed by SIGSEGV +++

This is also good to see - an strace on my machine looks pretty close to
identical. I guess I need to recompile and look for any suspect messages.

>I wonder if ENOTTY has any relation to UNIX 98 PTYs? Just speculatin'...

Hopefully there is an NVIZ person out there who may recognize where this is
happening.

----------------------------------------------------------------
Philip Warner | __---_____
Albatross Consulting Pty. Ltd. |----/ - \
(A.C.N. 008 659 498) | /(@) ______---_
Tel: +61-03-5367 7422 | _________ \
Fax: +61-03-5367 7430 | ___________ |
Http://www.rhyme.com.au | / \|
                                | --________--
PGP key available upon request, | /
and from pgp5.ai.mit.edu:11371 |/