Hello, > So what I need is a group of people who can help by > giving me > feedback or by writing some parts of the > documentation for or with > me. You don't need to be a kernel expert (i'm not) > to help, it's > just common sense. A bit of feedback. I've looked over your interface so far and it looks good. I would suggest however that you reconsider the thread part of the interface. Many kernels don't support threads natively and I wonder if it might be better to try and implement threading in the jvm rather than the kernel. If I recall, one of the goals is to try and keep the kernel as small as possible and move any other functionality up into the JVM or higher (allowing them to be written in java optimally). I also see the same issue for the virtual memory interface. Much of this is coming from two OS's I've been looking at lately, RTEMS and QNX. RTEMS because its a possible kernel target for the interface, and QNX because I really like their OS design. Neither supports threads or virtual memory. I'd think that the best option is to assume a message passing micro kernel that only does task switching (for multi-processing), resource access protection, and message passing from process to process. All other tasks will be handled by the JVM or higher. This would include threading, virtual memory, and device drivers (device drivers are simply separate programs). This is following from the QNX design. I'm also assuming that we even want/need multi-processing. In this design, I'd imagine many aspects of a full JOS system to be running as multiple processes with full system integration being done using RMI or jini (using RMI). In fact, perhaps even resource access protection should be moved up into the JVM. Just some thoughts. -iain _________________________________________________________ Do You Yahoo!? Get your free @yahoo.com address at http://mail.yahoo.com _______________________________________________ Kernel maillist - [EMAIL PROTECTED] http://jos.org/mailman/listinfo/kernel
