Dear Bas,
I can't test that here, but you are probably correct that it might be an
permission problem. Most Linuxes will mount NTFS/fat32 filesystems with
the "noexec" flag - which will prevent MATLAB and other Linux binaries
from loading and executing binary code from such a filesystem. You can
change this mount behaviour by either omitting the "noexec" flag or even
forcing "exec" permissions.
Volkmar
On Thu, 7 Sep 2006, Neggers, S.F.W. (Bas) wrote:
> Dear List, Volkmar, Doerte,
>
> just for the record and the archives I react to this somewhat older thread below.
>
> Just installed spm5 on a Linux & Windows dual boot laptop (Fedora core 5, 32 bit centrino Acer), and got exactly the same problem & error message as reported below (error message, blank screen, no image displays), a mex-file error message (probably) when starting spm5 from linux. Stumbled on the email below when searching for a solution, but found no answer, and did some testing.
> I also have spm5 installed on a fat32 partition as you had (eg the old windows file system that linux can read & write, still often used for C:/ and D:/ drives), to be able to start the same version from windows (don't know why I should want that, probably will never use, but hey, cool when you can). That got me thinking.
>
> I then moved the spm5 folder over to an ext3 linux partition, started spm5 again, and then I could load & display the images perfectly (that I moved to an ext3 partition too). So it does have something to do with the file system, it seems. To test it further, I now started spm5 from the folder on the ext3 partition, and loaded an image from the fat32 partition, no error, perfect display. So somehow it does have to do something with the filesystem the spm5 mex and/or m-files are installed to. Don't know why, probably some rights management (for executing files?) is garbled on fat32 that doesn't support rights management. Never had that with spm2.
>
> Take home message: when using linux, run spm5 from your linux (ext3/ext2 etc) partition, NOT from a fat32 partition, and all is groovy. Data itself can be anywhere.
> I am not sure whether my conclusions generalize to all linux/windows flavours, but I thought I should share the experience.
>
> Good night,
>
> Bas Neggers
>
> Utrecht University Medical Center
>
>
--
Volkmar Glauche
-
Department of Neurology [log in to unmask]
Universitaetsklinikum Freiburg Phone 49(0)761-270-5331
Breisacher Str. 64 Fax 49(0)761-270-5416
79106 Freiburg
|