Page 1 of 1

EEGLab Import segfault

Posted: 02 Dec 2009, 08:37
by alenhardt
Hello everyone!

I am having trouble with the EEGLab import of BCI2k data. The import works with all applications i have written so far except my most recent one. I dont think i did something different in this application regarding setting state variables.
Still, when i try to import them in matlab i get a segfault with this stack trace:

Code: Select all

  [0] kernel32.dll:0x7c812afb(0x0eefface, 1, 3, 0x00d08344)
  [1] load_bcidat.dll:0x11038b8d(0x1101f2e4, 0x00d08410, 0, 0)
  [2] load_bcidat.dll:0x11038bcd(0x110ad60c, 50, 32, 0x11080000)
  [3] load_bcidat.dll:0x1102cd16(0x110ad60c, 0x00d08478, 0x11080000, 7)
  [4] load_bcidat.dll:0x1102ca59(0x110ad60c, 1, 0x00d084f0, 0x110315ff)
  [5] load_bcidat.dll:0x1102caf2(0x110ad60c, 0x110845b0, 1, 0)
  [6] load_bcidat.dll:0x1101e0df(0x11084578, 0x00d086c0, 0x11014581, 0)
  [7] load_bcidat.dll:0x1101d25d(0x11084578, 0x1644c6b0 "C:\bci2k\data\Name001\NameS001R3..", 50000, 0x0fccd720)
  [8] load_bcidat.dll:0x1101170c(3, 0x00d093d0, 1, 0x00d09460)
  [9] libmex.dll:_mexRunMexFile(3, 0x00d093d0, 1, 0x00d09460) + 119 bytes
...
The .dat file can be viewed with the BCI2000Viewer program though. Converting it to .ascii is also possible. Only the matlab import fails.
I guess i did something in my application module that leads to inconsistent data. Do you have any hints what type of coding-error could lead to this behaviour? I assume it must have something to do with setting certain state variables.

The recording was done in this way:
(tested with Signalgenerator and gUSBAmp as signal source)
pressed start
let it run a few seconds
pressed stop
quit


The small data file can be downloaded at
https://ni.www.techfak.uni-bielefeld.de ... 01R357.zip

Thanks in advance
Alex[/url]

Posted: 03 Dec 2009, 09:57
by mellinger
Alex,

thanks for the comprehensive bug report. Most likely, you are using a rather old version of BCI2000 -- I could not reproduce the crash with the current version of the load_bcidat tool.

I suggest you update your local version of the EEGLabImport contribution to the most recent one from SVN.

Best regards,
Juergen