rtrace seg fault

Hello all,

rtrace has been giving me some trouble lately, but it doesn't seem to be consistent. I get the following error from time to time:

rtrace: signal - Broken pipe
Segmentation Fault

It usually happens while running rtcontrib, but not always. Sometimes it will run fine with no problems, but then the same geometry with the same command produces and error and it is not always at the same point. Sometimes it is right away, sometimes a few minutes into it, and sometimes a few hours into it. Is this more a memory/hardware issue, or could it be a problem with the way I arrange my files? Any help you could offer would be greatly appreciated.

Thanks,

Dave

Hi Dave,

It sounds like rtcontrib is bombing, not rtrace. Are you on the latest HEAD release? We had some problems with the official release of rtcontrib under Linux, I know. I don't remember any seg faults, but the Cygwin binaries seem to be prone to those if you're under Cygwin.

-Greg

···

From: "J. David Maino" <[email protected]>
Date: March 22, 2006 10:09:13 AM PST

Hello all,

rtrace has been giving me some trouble lately, but it doesn't seem to be consistent. I get the following error from time to time:

rtrace: signal - Broken pipe
Segmentation Fault

It usually happens while running rtcontrib, but not always. Sometimes it will run fine with no problems, but then the same geometry with the same command produces and error and it is not always at the same point. Sometimes it is right away, sometimes a few minutes into it, and sometimes a few hours into it. Is this more a memory/hardware issue, or could it be a problem with the way I arrange my files? Any help you could offer would be greatly appreciated.

Thanks,

Dave

Greg,

After upgrading to the latest HEAD release I've been running several different runs over the past day and it seems that the seg fault is gone, but now there is another message that I am getting:

*** glibc detected *** realloc(): invalid next size: 0x08060248
Aborted
rtrace: signal - Broken pipe

It seems to work OK when I split my sensor input file up into smaller segments (of 100-200 calculation points) but when I try the whole file (around 2000 points) is when it has problems now. I'm running the latest HEAD release on Ubuntu Linux (Breezy Badger) just for reference. For now, I'm fine with using the "split" command as a workaround and calculating the contributions in segments, but I was wondering if there is again perhaps something that I'm doing wrong?

Thanks,

Dave

Gregory J. Ward wrote:

···

Hi Dave,

It sounds like rtcontrib is bombing, not rtrace. Are you on the latest HEAD release? We had some problems with the official release of rtcontrib under Linux, I know. I don't remember any seg faults, but the Cygwin binaries seem to be prone to those if you're under Cygwin.

-Greg

From: "J. David Maino" <[email protected]>
Date: March 22, 2006 10:09:13 AM PST

Hello all,

rtrace has been giving me some trouble lately, but it doesn't seem to be consistent. I get the following error from time to time:

rtrace: signal - Broken pipe
Segmentation Fault

It usually happens while running rtcontrib, but not always. Sometimes it will run fine with no problems, but then the same geometry with the same command produces and error and it is not always at the same point. Sometimes it is right away, sometimes a few minutes into it, and sometimes a few hours into it. Is this more a memory/hardware issue, or could it be a problem with the way I arrange my files? Any help you could offer would be greatly appreciated.

Thanks,

Dave

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

Actually, I had the same problem with rtcontrib. It can easily be invoked by setting -ad higher than 26000.

Martin Moeck, Penn state

···

________________________________

From: [email protected] on behalf of J. David Maino
Sent: Fri 3/24/2006 9:31 AM
To: Radiance general discussion
Subject: Re: [Radiance-general] rtrace seg fault

Greg,

After upgrading to the latest HEAD release I've been running several
different runs over the past day and it seems that the seg fault is
gone, but now there is another message that I am getting:

*** glibc detected *** realloc(): invalid next size: 0x08060248
Aborted
rtrace: signal - Broken pipe

It seems to work OK when I split my sensor input file up into smaller
segments (of 100-200 calculation points) but when I try the whole file
(around 2000 points) is when it has problems now. I'm running the latest
HEAD release on Ubuntu Linux (Breezy Badger) just for reference. For
now, I'm fine with using the "split" command as a workaround and
calculating the contributions in segments, but I was wondering if there
is again perhaps something that I'm doing wrong?

Thanks,

Dave

Gregory J. Ward wrote:

Hi Dave,

It sounds like rtcontrib is bombing, not rtrace. Are you on the
latest HEAD release? We had some problems with the official release
of rtcontrib under Linux, I know. I don't remember any seg faults,
but the Cygwin binaries seem to be prone to those if you're under Cygwin.

-Greg

From: "J. David Maino" <[email protected]>
Date: March 22, 2006 10:09:13 AM PST

Hello all,

rtrace has been giving me some trouble lately, but it doesn't seem
to be consistent. I get the following error from time to time:

rtrace: signal - Broken pipe
Segmentation Fault

It usually happens while running rtcontrib, but not always. Sometimes
it will run fine with no problems, but then the same geometry with
the same command produces and error and it is not always at the same
point. Sometimes it is right away, sometimes a few minutes into it,
and sometimes a few hours into it. Is this more a memory/hardware
issue, or could it be a problem with the way I arrange my files? Any
help you could offer would be greatly appreciated.

Thanks,

Dave

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

Hi Dave,

This appears to be a glibc internal error (consistency check), which indicates a bug in the malloc library on your system. I don't think there's much I can do about it if that's the case. If someone has a copy of the malloc.c module from glibc, I can take a look to see exactly where this check is happening, but that still doesn't give me much chance of fixing the problem if I can't reproduce it on my system. There's nothing particularly funny or strange about my use of realloc() in rtcontrib.

The only possibility I can think of is that I'm passing an invalid pointer to realloc(), which is somehow generating this error rather than the expected one. I checked my code, though, and I can't see how this could happen.

-Greg

···

From: "J. David Maino" <[email protected]>
Date: March 24, 2006 6:31:06 AM PST

Greg,

After upgrading to the latest HEAD release I've been running several different runs over the past day and it seems that the seg fault is gone, but now there is another message that I am getting:

*** glibc detected *** realloc(): invalid next size: 0x08060248
Aborted
rtrace: signal - Broken pipe

It seems to work OK when I split my sensor input file up into smaller segments (of 100-200 calculation points) but when I try the whole file (around 2000 points) is when it has problems now. I'm running the latest HEAD release on Ubuntu Linux (Breezy Badger) just for reference. For now, I'm fine with using the "split" command as a workaround and calculating the contributions in segments, but I was wondering if there is again perhaps something that I'm doing wrong?

Thanks,

Dave