CS695 Topics in Virtualization and Cloud Computing
Spring 2019

Lecture 6  23.1.2019
--------------------

0. Exercise #2 due 30th January 2019
https://www.cse.iitb.ac.in/~puru/
https://www.cse.iitb.ac.in/~puru/courses/spring19/cs695/exercises/ex2.html

references for this lecture:
[lkd] ... Linux Kernel Development, Robert Lowe (book)
[ulk] ... Understanding the Linux Kernel, Daniel P. Bovet and Marco Cesati (book)

1. Recap

- techniques for CPU virtualization,
  0. trap and emulate
  1. scan and patch
  2. para-virtualization
  3. hardware-assisted CPU virtualziation

2. Linux digression

(a) a basic unit for control and management of execution and resources
is a process.

- what is a process for an OS?
   - state (metadata): required to identify entities, to tag resoures,
                       for controlling runtime execution
   - actions: process specific actions, actions for processes etc.
     scheduling, context switching, allocating memory, file operations,
     delivering signals etc.

   - all (process) state and actions managed by the OS as part of 
   providing the process abstraction

- OS representation (state) of a process exists in a task_struct structure

   struct task_struct
      pid, ppid, tgid
      status, priority
      mm_struct
      fs_struct
      signal_struct
      thread_info
      pointer to tasklist
      ...

* Homework: look inside task_struct and mm_struct and understant important fields.

- kernel threads (ps -ef)
  - special processes
  - have a task_struct and hence schedulable
  - mm_struct = NULL (no address space of its own)

  - do non-user intiated work for the kernel
    page cache flush to disk, CPU runqueue migration, interrupt processing, 
    watchdog timers ...

* Homework: How do kernel threads work with no address space? (mm_struct == NULL)

- process creation
  
  fork->clone->do_fork->copy_process->dup_task_struct
  fork + exec  ... create a process, replace process with new program

  copy-on-write memory optimization used by fork

* What does the call 'current' do? What is its return value? How is it implemented?

(b) memory mamanagement

- protected mode of operation of the CPU (as opposed to real mode)
  enables support for virtual address space via paging (and MMU based hardware translation)

  - per process virtual address space
  - isolation, sharing, swapping, physical memory mutliplexing, memory protection, 
    file mapping ...

  - terminology: mapped address, pte, pmd, pud, ...

- Linux specifics: mm_struct in task_struct holds state/metadata of memory allocation
and manangement of a process
  
  task mm_struct
     pgd_t  pgd;                    // page directory pointer
     struct vm_area_struct* mmap;   // list of VMAs

- memory layout (example for 32-bit arch.)

  for each process, the following virtual address space ranges hold
  - 0x0000 0000 to 0xbfff ffff: user addresses (user space)
  - 0xc000 0000 to 0xffff ffff: kernel addresses (kernel space)

  - the kernel space mappings are same across all processes,
    hence all processes share/have the same kernel.

  - kernel space holds the kernel state and code for actions
  - all, kernel addressing is via the virtual address space

  - linux keywords: __pa, kmalloc, vmalloc, virt_to_page, pgd_offset,
                    pte_present, page_address ...

- paging and PTE flags

  - 32-bit virtual to physical address mapping
    v2p mapping and translation

  - PTE_FLAGs: LSB 12-bits of PTE
    - PAGE_PRESENT, PAGE_PROTNONE, PAGE_RW, PAGER_USER, PAGE_DIRTY,
      PAGE_ACCESSED ...

  - Note: On a virtual address issue, the MMU in hardware attempts a
    translation and access. Interventions are possible only on exceptions!