creating processes - process adress space - and console hand
Posted: Sat Mar 29, 2003 1:07 pm
Well, there is good news and there is bad news?
Why should I start the bad news first? cuz then you have something to be happy after having heared the good news *ssssffg*
here come the bad news: I 've been too silly till now to understand where to put console & cursor information. To bad I didn't realize that when I say I have 8 consoles, only 8 processes can get a console to display things->I would need 8 structures where local console information as well as a pointer to memory location for video memory buffer copy upon console switch (which is a task switch). The processes which need a video console, get such a structure, but only 8 processes can have one. How to handle TTY consoles I ll have to figure out. Is this what Pype wanted to tell me?
aaand here comes the good news:
multitasking runs. I got it doing what i expect it to do: switching tasks in a happy round robin manner. I also have created a set of functions to create tasks. At the moment it's easy: the tasks are compiled into the kernel file. And yes, I use tss's and they are a real convenient way to implement multitasking. since I don't intend to have my kernel thing beeing portable, I'll stay with this concept. Although I am now about to implement software task switching too. It is important for me to learn how to handle Both methods. Because what happens If I have to build an os for a processor which is not capable o hardware task switching?
Now: do i have to think about some kind of process memory management? Sort of book keeping about the memory allocated FOR a process - creating page directory and page tables for a certain process and marking the memory space as used in the global physical memory management bitmap? - entering the page directory for a process in his tss? Since I think process memory space manegement is a task of the kernel too - ithas to allocate memory for a process, its heap and its stack. Do I understand this right?
stay safe
Why should I start the bad news first? cuz then you have something to be happy after having heared the good news *ssssffg*
here come the bad news: I 've been too silly till now to understand where to put console & cursor information. To bad I didn't realize that when I say I have 8 consoles, only 8 processes can get a console to display things->I would need 8 structures where local console information as well as a pointer to memory location for video memory buffer copy upon console switch (which is a task switch). The processes which need a video console, get such a structure, but only 8 processes can have one. How to handle TTY consoles I ll have to figure out. Is this what Pype wanted to tell me?
aaand here comes the good news:
multitasking runs. I got it doing what i expect it to do: switching tasks in a happy round robin manner. I also have created a set of functions to create tasks. At the moment it's easy: the tasks are compiled into the kernel file. And yes, I use tss's and they are a real convenient way to implement multitasking. since I don't intend to have my kernel thing beeing portable, I'll stay with this concept. Although I am now about to implement software task switching too. It is important for me to learn how to handle Both methods. Because what happens If I have to build an os for a processor which is not capable o hardware task switching?
Now: do i have to think about some kind of process memory management? Sort of book keeping about the memory allocated FOR a process - creating page directory and page tables for a certain process and marking the memory space as used in the global physical memory management bitmap? - entering the page directory for a process in his tss? Since I think process memory space manegement is a task of the kernel too - ithas to allocate memory for a process, its heap and its stack. Do I understand this right?
stay safe