[Playerstage-developers] Segmentation fault in sicklms200.cc - CVS HEAD

[Peter N. <jpn...@ho...>]

Hello.

I think i have found a bug in the sicklms200.cc file from the CVS HEAD ve=
rsion.
I am using Debian Stable 4.0 with kernel 2.6.18-5-686.
The CVS id tag is CVS: $Id: sicklms200.cc,v 1.65 2007/11/01 22:16:19 gerk=
ey Exp $
When i start the player server with a laser-only.cfg file this is what ha=
ppens. =20

"player -d 9 laser.cfg"
-------------------------------------------------------------------------=
-----
Laser initialising (/dev/ttyUSB0)
connecting at 9600
sending configuration mode request to laser
waiting for acknowledge
laser operating at 9600; changing to 38400
SICK laser type [LMS200;301063;V02.32 ] at [/dev/ttyUSB0:38400]
laser ready
listening on 6665
Listening on ports: 6665
receiving data
Segmentation fault
-------------------------------------------------------------------------=
-----

I have been able to track down the line were the error occurs by using  s=
ome quick "printf debugging".
The line is 706 in sicklms200.cc were it says: =20
       =20
 703    for (int i =3D 0; i < this->scan_max_segment - this->scan_min_seg=
ment + 1; i++)
 704     {
 705       data.intensity[i] =3D ((mm_ranges[i] >> 13) & 0x0007);
 706       data.ranges[i] =3D  (mm_ranges[i] & 0x1FFF)  * this->range_res=
 / 1e3;
 707     }

This is were the laserdata is stored in "data.ranges[i]". I think that da=
ta.ranges never gets any memory allocated the first time the program runs=
.
(data.ranges seem to be a pointer to float according to libplayercore/int=
erfaces/006_laser.def).

A quick fix was to add "data.ranges_count =3D 0 " after line 670. By doin=
g this I can force memory to be allocated by the IF on lines 695-702
This would "solve" that problem but after the loop on line 703 is complet=
e a new segmentation fault error occurs on line 731-733 in th Publish cal=
l.
I dont know what is causing this. Could it perhaps be the use of sizeof(d=
ata). Does that work when data has members with dynamic memory allocation=
 ?
(Well I don't know, its just a thought) =20
=20
 668   if (first)
 669     {
 670       PLAYER_MSG0(2, "receiving data");
             data.ranges_count =3D 0;
 671       first =3D false;
 672     }

 692    double old_count =3D data.ranges_count;
 693     data.ranges_count =3D data.intensity_count =3D
 694             this->scan_max_segment - this->scan_min_segment + 1;
 695     if (old_count < data.ranges_count)
 696     {
 697       delete [] data.ranges;
 698       delete [] data.intensity;
 699       data.ranges =3D new float[data.ranges_count];
 700       data.intensity =3D new uint8_t[data.intensity_count];
 701
 702     }


 730    // Make data available
 731     this->Publish(this->device_addr, =20
 732                   PLAYER_MSGTYPE_DATA, PLAYER_LASER_DATA_SCAN,
 733                   (void*)&data, sizeof(data), &time);
 734     delete [] data.ranges;
 735     delete [] data.intensity;


I hope this will be helpful in the hunt for bugs.
/Peter

Nu =E4r det K=E4rleksvecka p=E5 Spray Date. Vinn en dr=F6mdejt fr=E5n K=E4=
rleksfonden. http://www.spraydate.se