This is the mail archive of the gdb@sourceware.cygnus.com mailing list for the GDB project. See the GDB home page for more information.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Date: Fri, 05 Mar 1999 08:30:06 -0500 From: Mohan Pakkurti <pmohan@cadence.com> I am wondering if there is currently any work being done in the area of supporting gdb for machines with multiple processors? What is involved? can someone comment on this? thanks. There are a couple ways to accomplish this. One is to use threads, have different threads run on different processors. GDB's thread support will generally work unchanged on these kinds of machines, since the OS handles the thread/processor mapping invisibly. Another way is to have multiple processes. GDB has nothing for this, although the idea has been talked about. GDB would need some hefty modifications, since much of its internal data about the inferior process is tied to global variables, and all this would have to change. Finally, there is the coprocessor approach, more often seen in embedded environments, in which multiple CPUs are executing instructions from different programs in the same physical address space. They may or may not have the same architecture; for instance there are cell phone designs featuring an ARM and a DSP on the same chip, and using the same RAM. Supporting debugging in this kind of environment is a big challenge for GDB, and will require a lot more work on the underpinnings. I'm encouraging people to change GDB in ways that will facilitate this, but I don't know of anybody actually tackling this multiple-processor problem. Stan