Tuesday, April 19, 2011

An Approach to core dump in minix3

Hey people,
I have been trying to work around implementing core dump in minix3.A clear understanding of what core dump is necessary for its design. I will start with what a core file is.

The core file contains the memory image of the process, which can be used for debugging purposes.A core file is created when various errors(depends on how dumping is implemented) occur. Errors such as memory-address violations, illegal instructions, bus errors, and user-generated quit signals, commonly cause this core dump. The core file that is created contains a memory image of the terminated process.

The contents of a core dump are organized sequentially in the core file.It may follow any data structure to produce dump.Below is a rough idea of what things can be included.

Core header -> Defines basic information about the core dump, and contains offsets that locate the remainder of the core dump information.
User stack -> Contains a copy of the user stack at the time of the core dump.
Data area -> Contains the user data section.
Memory mapped regions -> Contains the anonymously mapped regions.

Process Control Blocks :

A process control block (PCB) exists for each process, describing the state of that process by maintaining the LDT, as well as saved register values, used to save the process state while a process is waiting to be scheduled. The PCB is a struct proc, containing important information about one process, including the process number and status. Also accessible through the PCB, via the p_map memory map, are the base addresses and physical and virtual sizes of the three segments of each process: (1)the text, (2)data, and (3)stack segments.

In minix architecture the microkernel, file system, and memory manager all maintain their own process tables but keep strict correspondence between these entries.The complete state of a process is defined by the process' data in memory, plus the information in its process table slot.

No comments:

Post a Comment