frequent “memory has been corrupted” errors
Guest wrote on Oct 07, 2005
Greetings,
I seem to be getting this error quite often now with iraf 2.12:
grosa@xi:/sandbox/grosa$ /iraf/irafbin/noao.bin.linux/x_artdata.e 'mkobjects input=tmpl3.fits output=t1.fits objects=out1 xoffset=0 yoffset=0 star=moffat distance=1 background=0.0 gain=1.0 ranbuf=0 rdnoise=0.0 poisson=no exptime=1.0 magzero=33.75 seed=1 comments=yes nxc=5 nyc=5 nxsub=10 nysub=10 nxgsub=5 nygsub=5 dynrange=100000 psfrange=10 beta=2.5 radius=1 ar=1 pa=0'
PANIC in `/iraf/irafbin/noao.bin.linux/x_artdata.e': Memory has been corrupted
I've found mention of this error in at least 2 places:
http://iraf.noao.edu/iraf/web/new_stuff/v2122-beta2.html
http://iraf.anu.edu.au/iraf/ftp/iraf/v212/PCIX/?C=S;O=D
However, my kernel does not have exec-shield support, and my stack limit
seems to be unlimited:
grosa@xi:/sandbox/grosa$ ulimit -s
unlimited
strace'ing the process indicates it's not dying anywhere [reasonably]
important. Running it in gdb shows that it exits with error code 072,
but I'm unable to provide a backtrace for it. I could, conceivably,
compile my own version with more debugging enabled, but I'm not thrilled
about doing that.
I'm at a bit of a loss. Can you shed some light into this? Any help
would be appreciated.
thanks,
-Gabe
--
Gabriel G. Rosa | For some reason, normal sane people seem to lose all of
Programmer, Physics | their common sense when sitting in front of a computer.
512 Physics/Geology | To the point where plain messages e.g. "Please Insert a
(530) 754-8235 | Floppy", evoke a response of "What does that mean?" -AM
Post generated using Mail2Forum (http://www.mail2forum.com)
I seem to be getting this error quite often now with iraf 2.12:
grosa@xi:/sandbox/grosa$ /iraf/irafbin/noao.bin.linux/x_artdata.e 'mkobjects input=tmpl3.fits output=t1.fits objects=out1 xoffset=0 yoffset=0 star=moffat distance=1 background=0.0 gain=1.0 ranbuf=0 rdnoise=0.0 poisson=no exptime=1.0 magzero=33.75 seed=1 comments=yes nxc=5 nyc=5 nxsub=10 nysub=10 nxgsub=5 nygsub=5 dynrange=100000 psfrange=10 beta=2.5 radius=1 ar=1 pa=0'
PANIC in `/iraf/irafbin/noao.bin.linux/x_artdata.e': Memory has been corrupted
I've found mention of this error in at least 2 places:
http://iraf.noao.edu/iraf/web/new_stuff/v2122-beta2.html
http://iraf.anu.edu.au/iraf/ftp/iraf/v212/PCIX/?C=S;O=D
However, my kernel does not have exec-shield support, and my stack limit
seems to be unlimited:
grosa@xi:/sandbox/grosa$ ulimit -s
unlimited
strace'ing the process indicates it's not dying anywhere [reasonably]
important. Running it in gdb shows that it exits with error code 072,
but I'm unable to provide a backtrace for it. I could, conceivably,
compile my own version with more debugging enabled, but I'm not thrilled
about doing that.
I'm at a bit of a loss. Can you shed some light into this? Any help
would be appreciated.
thanks,
-Gabe
--
Gabriel G. Rosa | For some reason, normal sane people seem to lose all of
Programmer, Physics | their common sense when sitting in front of a computer.
512 Physics/Geology | To the point where plain messages e.g. "Please Insert a
(530) 754-8235 | Floppy", evoke a response of "What does that mean?" -AM
Post generated using Mail2Forum (http://www.mail2forum.com)
Guest wrote on Oct 07, 2005
Hi Gabriel,
The ulimit trick usually fixes this but only when the memory
corrupted comes from all tasks. If this is limited to MKOBJECTS then
it may be related to the data itself and be no more than a bug in the
task. Does it happen reliably with your e.g. object list but not when
trying one of the help page examples? Does it happen for any other task?
If it is limited to the one task and your data files, could you
please put your sample input image and object list in the /pub of our
ftp://iraf.noao.edu archive so we can have a look? Also, it would help
to know what distribution of Linux you're using and the IRAF version
(i.e. V2.12.2 or the V2.12.2a patch from Jul04)? Note that there are
some known issues with the Debian package created by one of the users,
if you're using this try installing the binaries from our archive instead.
Hope this helps.
Cheers,
Mike Fitzpatrick
Post generated using Mail2Forum (http://www.mail2forum.com)
The ulimit trick usually fixes this but only when the memory
corrupted comes from all tasks. If this is limited to MKOBJECTS then
it may be related to the data itself and be no more than a bug in the
task. Does it happen reliably with your e.g. object list but not when
trying one of the help page examples? Does it happen for any other task?
If it is limited to the one task and your data files, could you
please put your sample input image and object list in the /pub of our
ftp://iraf.noao.edu archive so we can have a look? Also, it would help
to know what distribution of Linux you're using and the IRAF version
(i.e. V2.12.2 or the V2.12.2a patch from Jul04)? Note that there are
some known issues with the Debian package created by one of the users,
if you're using this try installing the binaries from our archive instead.
Hope this helps.
Cheers,
Mike Fitzpatrick
Post generated using Mail2Forum (http://www.mail2forum.com)
Last post on Oct 07, 2005