I am working on software to control MMAPD devices on embedded ARM systems, but in some situations where debugging And development tools available are not enough. Equipment tools such as valgrind and high end thread profiler are unavailable.
What I want to do is compile my code on an x86 machine, using the same size, a 'dummy' section of memory mmap and then the proxy that writes / writes in the embedded machine in the network Which can then respond accordingly.
I know that this will require the client / server system and it will be very slow but the advantage of this option is the use of mechanics (ASCN event handling, thread management) XMK development tools outside of the MAMP interface The tools available to do will be very useful.
I have heard about this technique for simulation in some ASIC development but never used whatever this functionality is used here the main thing is that I can write a group of stuff again I want to use the same code on both platforms, or write a kernel module in which it has a hardware handling argument. I want to keep all device control logic in USA via MMAP
If you think through this , Basically boils down to implementing the remote file system.
- If your embedded device is a simple version of the NFS client side, and
mmap
a file on your desktop machine - inversion It should also be possible, that is, it is possible to export your nfs server module running on the embedded system and
/ dev / shm
as an NFS mount point to your desktop.
For both methods (using any other mmap
), you should really take precautions on data synchronization. But as long as you have a specific debug setting originally with a writer and a reader, this may be possible, if it becomes more complicated, then remote file locking is available as a NSF extension.
Comments
Post a Comment