ccdproc PANIC/segmentation violation

vmagalhaes · Joined: 06 Jan 2011 Posts: 7

I have a instalation of 2.16 with both 32 and 64 bits mac os x binaries.
i have been trying to correct some flat images with an averaged bias image.
When i try i get the following error messages:
64 bits:
ecl> ccdproc
List of CCD images to correct (flat*):
ERROR: segmentation violation
32 bits:
ecl> ccdproc
List of CCD images to correct (flat*):

PANIC in `/iraf/noao/bin.macosx/x_ccdred.e': Memory fwa has been corrupted

In neither cases does iraf crash or close, its working fine.
My belief is that this is a problem with x_ccdred.e
is there a patch for this executable?
or a recipe for me to recompile it locally?

Thank you for any advice.

fitz · Posted: Tue May 01, 2012 10:10 pm Post subject:

There's been a similar report recently (http://iraf.net/phpBB2/viewtopic.php?t=90895) but no followup yet that identified the problem or the fix. Would it be possible for you to upload some sample images and parameters to the anonftp at ftp://iraf.noao.edu/pub so we can try to reproduce the problem?

You can be sure you're running the latest version by doing:

vmagalhaes · Joined: 06 Jan 2011 Posts: 7

Thanks for the rapid response.

My script for starting iraf deliberately started ecl instead of cl
I changed it and now it starts in vocl.
But this change didn't make any difference...

i tried updating with the make latest command.
it downloaded the new versions of some parts of IRAF but still no difference...

I tried redownloading all the distribution and still got the same error...
I also tried using other data, but still from the same observational run, and got the same error...

i'm trying to upload a file with the images but the pub ftp tells me i don't have permission...
do you have another alternative?

fitz · Posted: Wed May 02, 2012 5:42 pm Post subject:

I asked about the ecl vs. vocl thing only to be sure we're talking about v2.16, the error is in the CCDPROC binary itself and so this shouldn't matter.

The anonftp doesn't allow you to create subdirs you can write to, just upload everything to the toplevel ftp://iraf.noao.edu/pub directory or else send me a URL where I can nab it (or email fitz@iraf.net).

vmagalhaes · Joined: 06 Jan 2011 Posts: 7

The problem with uploading to the ftp was fixed when i updated filezilla...
the file is there with the name ccdproc_error.tar.gz

it contains an averaged bias image created with zerocombine.
and one of the flat images i was trying to use.

fitz · Posted: Tue May 08, 2012 10:02 pm Post subject:

I can't seem to reproduce the problem. Would you mind sending the parameters you used (e.g. the output of "dpar ccdproc") as well as the command

valdes · IRAF Guru Joined: 11 Nov 2005 Posts: 677

I made an attempt to reproduce the problem but no luck. This was on an iMac.

marco.03 · Joined: 19 Dec 2010 Posts: 5

hi
I am having a similar problem with ubuntu 64bit 12.04 (kernel 3.2.0-38 ).
I have installed the linux64 binaries.
Running ccdproc just to trim any image gives me the segmentation violation error.
I have tried running d_trace before ccdproc. This is what I get

fitz · Posted: Fri Feb 22, 2013 5:46 pm Post subject:

As you see from the previous posts, we weren't able to reproduce the problem locally on mac systems, perhaps linux64 will be different. Could you upload a sample image to the anonftp at ftp://iraf.noao.edu/pub in case the problem is data related and we'll try again.

marco.03 · Joined: 19 Dec 2010 Posts: 5

I have uploaded the ccdproc_sv.tar.gz. Howevere I think the problem is not data related, because I get the same error every time I try to crop any image.

vall · Posted: Tue Mar 05, 2013 5:28 pm Post subject:

a couple of week ago I installed IRAF2.16 on Fedora 18 64-bit. For fat I was doing photometry on already pre-redduced images and everything was working fine. Today I tried oevrscan and trim corrections with ccdproc and I get exactly the same problem.

vall · Posted: Thu Mar 07, 2013 4:36 pm Post subject:

Some more issues popped up the more I use IRAF 2.16 on Fedora 18. Aperture photometry seems to work fine, but when I tried PSF, it fails to write the computed PSF. Then I used 'wcsctran' but it claims that the input coordinate files do not exist, but they surely exist. Then I moved all the jobs to another laptop with Fedora 17 instead of Fedora 18, and it all worked fine.

I do not understand why is this happening. Sure, the new laptop has SSD, ext4, while the old normal HDD, ext3, but this should not matter.

fitz · Posted: Thu Mar 07, 2013 10:31 pm Post subject:

In reply to marco.03:

Thanks for the data, however I am still not able to reproduce the problem using either the v2.15, v2.16 or current development versions. I do notice that you have the 'ccdtype' parameter set to a space and not a null string (i.e. " " and not ""), but this doesn't seem to make a difference. Is this an IRAF system you got from scisoft or some other source? Was it recompiled locally in any way? Are you likewise on a Fedora 18 or similar system as vall (the kernel version is very recent)?? Could you both please post the output of the command

vall · Posted: Thu Mar 07, 2013 11:26 pm Post subject:

For me both the 64bit and the 32-bit binaries did not work. My problems still persist. My original binary was:

fitz · Posted: Fri Mar 08, 2013 12:21 am Post subject:

That appears to be the last released binary, and the same one that worked for me. If you're up to trying one more binary, see

ftp://iraf.noao.edu/pub/x_ccdred.e.LINUX64-STATIC

This is statically linked and if it works may point to something in a newer gcc/glibc (could you post those versions as well) release.