PANIC in … /x_ccdred.e: Out of memory

Sarah Sweet wrote on Jan 25, 2010

We are using mscred to reduce data recently taken at CTIO and are encountering some issues as per below. If you are able to assist that would be much appreciated!

ms> show imcombine_maxmemory
100000000
ms> flatcombine
List of flat field images to combine (@dflat_glist): dflat1010,dflat1011

PANIC in `/scisoft/share/iraf/extern/mscred/bin.redhat/x_ccdred.e': Out of memory
ERROR: Abnormal termination of child process 'mscbin$x_ccdred.e'
    flprcache ()
    ccdproc (images=dflat1010,dflat1011, output=, bpmasks=, ccdtype=dome flat, noproc=no)
    flatcombine ()
ms> lpar mscred
   (pixeltype = "real real")    Output and calculation pixel datatypes
     (verbose = no)             Print log information to the standard output?
     (logfile = "logfile")      Text log file
    (plotfile = "")             Log metacode plot file
      (backup = "all")          Backup data (none|once|all)?
    (bkuproot = "Raw/")         Backup root (directory or prefix)
  (instrument = "mscred$lib/mosaic.dat") CCD instrument file
     (ampfile = "amps")         Amplifier translation file
      (ssfile = "subsets")      Subset translation file
  (im_bufsize = 0.065536)       Image I/O buffer size (in Mbytes)
    (graphics = "stdgraph")     Interactive graphics output device
      (cursor = "")             Graphics cursor input
     (version = "V4.9: August 12, 2008") 
        (mode = "ql")           
      ($nargs = 0)   

I have 9 flats to combine but here have attempted the task with two, with the same error message resulting. I successfully ran zerocombine and darkcombine previously to this, following along as per http://www.noao.edu/noao/noaodeep/ReductionOpt/frames.html .
I have also tried setting imcombine_maxmemory to smaller and smaller values as http://iraf.net/ftp/iraf/extern/mscred.readme suggests, down as far as 10000 but to no avail.
The IRAF version is 2.14.1.

If you need further clarification please let me know.

Sarah

---

Mike Fitzpatrick wrote on Jan 25, 2010

If this is a 64-bit OS then I suspect the problem is a pointer overflow, i.e. the malloc() call returns a 64-bit address which overflows the 32-bit pointer value and IRAF sees the negative value as a failed malloc (which is normally an 'out of memory' condition). If so I could supply a 64-bit ccdred binary for you to try, but to be sure it doesn't happen in another task you would need a full set of binaries. I'm working on the distribution now but it'll be a week or two before something is ready.

OTOH, if this is a 32-bit OS then Frank may have some ideas about other causes.

-Mike

---

Sarah Sweet wrote on Jan 25, 2010

thanks Mike,
it is 64bit. Although since then I've tried setting the image i/o buffer in mscred to 4Mb (was 0.065536), and now it is working so far. But I would be interested in having the 64bit binaries for future use when ready!

Thanks,
Sarah

---

Last post on Jan 25, 2010