5.1 Lab 4: lab  (Page 3/4)

This process has shown you how to set breakpoints and single-step through Java code.

Debugging in c

Debugging in C is done using ndk-gdb , which is supported in Eclipse. To do this, first terminate any existing debug session and then on the Android device, exit the current application by hitting the home button.

To configure GDB, switch back to the C/C++ Perspective and do the following:

1. Click Run>Debug Configurations...
2. Select Android NDK Application and press the New icon, and rename the configuration to Lab4 (1) .
3. On the Android tab:
   1. Under Project , hit Browse... and select the current project.
   2. Under Misc. , select Attach to the running application .
4. On the Debugger tab:
   1. Select the GDBserver Settings tab.
   2. Use the APK bundled GDBserver.
5. Apply the settings and Close .

In order to launch GDB, you must first run the application on the Android device, and then attach GDB to the running application. To do this:

1. Set a breakpoint inside the for loop in process.c .
2. Click Run>Run . You should see the application launch on the Android device, ignoring any Java breakpoints that you have set.
3. After the application starts, click Run>Debug As...>Android NDK Application . The Debug Perspective should launch again, but this time with gdbserver debugger . The processor will halt at the set C breakpoint.

In the Expressions window, the sb.get(0) (which is a Java method) will have generated an error as we are now debugging in C. Verify that inBuf has the same values as the test vector look-up table by adding inBuf[0] to the list of expressions; check several different array indices.

Exporting variables to a file

A useful feature that is fully supported in CCS is the ability to export processor memory to a file, which can then be imported into MATLAB for further analysis. To enable this feature in Eclipse, the lab machines have the eVars plugin installed. As an example, to export the inBuf array to file:

1. In the Debug Perspective , go to Window>Show View>Variables .
2. Right-click on the inBuf pointer and select Display as Array .
3. Click the Expand Variables icon multiple times until the entire array has been expanded.
4. Once the array is fully expanded, click on the Export Variables icon, and save to a txt file.
5. Use evars2array.m to read in the text file into a MATLAB vector.

In the next lab, we will see how to create and write to a file on the Android device, and use adb to download this file to the host machine, directly from MATLAB.

Part 3: a spectrogram algorithm in matlab

As an initial step towards implementing the spectrogram in Android, you will first implement it in MATLAB. [link] shows the components that your spectrogram algorithm should have; the ability to overlap, while important, will be left for extra credit, and is therefore optional.

Spectrogram components