I'm trying to migrate over to 8.2 and have noticed that things seem to be broken under 8.2. Particularly I take the 3DES example and pull it down under both 8.1 and 8.2 (different machines) along with identical EDK directories (thank you Subversion) and begin my build process. The 8.1 generated platform behaves correctly but the 8.2 generated platform burps out "malloc failed" several times and no further output from the FPGA 3DES routine is seen. I've seen similar malloc errors on other projects but haven't gotten to put them side-by-side just yet. Any ideas?
1
behavior change between EDK/ISE 8.1 and 8.2
Started by Brett, Apr 11 2007 09:25 AM
4 replies to this topic
#1
-
- Members
-
- 36 posts
Advanced Member
Posted 11 April 2007 - 09:25 AM
Brett Boren
Computer Engineering Graduate Student
University of Alabama, Huntsville
Computer Engineering Graduate Student
University of Alabama, Huntsville
#2
-
- Members
-
- 35 posts
Advanced Member
Posted 11 April 2007 - 06:18 PM
i think if you modify the software *.ld file can solve the problem.
#3
-
- Members
-
- 36 posts
Advanced Member
Posted 12 April 2007 - 05:07 AM
QUOTE (mypowlar @ Apr 11 2007, 09:18 PM) <{POST_SNAPBACK}>
i think if you modify the software *.ld file can solve the problem.
Ok....? modify how?
I did find the following Xilinx Answer Record (#23345)
http://www.xilinx.com/xlnx/xil_ans_display...tPagePath=23345
Basically admitting that the malloc implementation prior to 8.2 was broken. This we suspected from experiences both Adderly and I had seen regarding hardware resets. After so many resets (sometimes just 1) malloc errors started flying. I don't know conclusively that this is related to what I'm currently seeing but it seems likely.
Brett
Brett Boren
Computer Engineering Graduate Student
University of Alabama, Huntsville
Computer Engineering Graduate Student
University of Alabama, Huntsville
#4
-
- Members
-
- 36 posts
Advanced Member
Posted 12 April 2007 - 02:04 PM
QUOTE (Brett @ Apr 12 2007, 08:07 AM) <{POST_SNAPBACK}>
Ok....? modify how?
I did find the following Xilinx Answer Record (#23345)
http://www.xilinx.com/xlnx/xil_ans_display...tPagePath=23345
Basically admitting that the malloc implementation prior to 8.2 was broken. This we suspected from experiences both Adderly and I had seen regarding hardware resets. After so many resets (sometimes just 1) malloc errors started flying. I don't know conclusively that this is related to what I'm currently seeing but it seems likely.
Brett
I did find the following Xilinx Answer Record (#23345)
http://www.xilinx.com/xlnx/xil_ans_display...tPagePath=23345
Basically admitting that the malloc implementation prior to 8.2 was broken. This we suspected from experiences both Adderly and I had seen regarding hardware resets. After so many resets (sometimes just 1) malloc errors started flying. I don't know conclusively that this is related to what I'm currently seeing but it seems likely.
Brett
It's a good thing I'm around to make all the stupid mistakes for everyone else. Turns out you do need to edit the ld script and put some sufficiently large value in for the heap size. With this I'm getting some things working, but not yet all.
Brett
Brett Boren
Computer Engineering Graduate Student
University of Alabama, Huntsville
Computer Engineering Graduate Student
University of Alabama, Huntsville
#5
-
- Members
-
- 35 posts
Advanced Member
Posted 16 April 2007 - 06:08 PM
another problem may cause this may be the cache. You can take a care to use cache.
1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users
