Expos Review

[Akhil S]

Hi,

Some review points till Stage 8 are given below.  

I have tried to stress on the points of confusion a novice reader might get.

 

Landing Page ( http://exposnitc.github.io/index.html )

--------------

1. Better to divide into sections.

2. short introduction, ( better not to use jargons like XSM etc in the first para as it may overwhelm the reader).

3. A section of what exactly you gain by doing this exercise.

4. This can be followed by the technical overview of the final product.

  

Stage 2

-----------

What is the relevance of XFS-interface

Why have we offloaded some major features about the File System from the kernel, (ideally the xfs-interface should only deal with writing  disk blocks, not files)

Stage 3

----------

Mention why an assembler is not required.

XSM privileged mode execution tutorial

A diagram showing the interrupt vector table would be good.

Stage 4

----------------

Some general statement like, SPL is the language you will be writing the kernel in

Ask the reader to go through the generated machine code

Stage 6

----------------

A para about what exactly an OS Startup code is and what it is supposed to do, before introducing the jargon for the first time.

Can we remove the INT10 and exhandler from this stage, (eg the user pgm can go to an infinite loop and we can do Ctrl+c) for now

I personally feel that this stage is a bit too large.

Also, how about using the memory area as well, for eg the code can be like:

0   MOV R0, 1
2   MOV R2, [19]
4   GE R2, R0
6   JZ R2, 18
8   MOV R1, R0
10  MUL R1, R0
12  BRKP
14  ADD R0, 1
16  JMP 2
18  JMP 18     // Infinite loop
19  5 0        // Dont know if xfs-handler supports this

Stage 7

-------------------

2048  0
2049  2056
2050  0
2051  0

This seems to be confusing. Is it possible to change into

2048  0 2056
2050  0 0
2052  0 0

Who has written the Library ? Will the reader be asked to write it in future ( Since we are building an OS from scratch )

 

--

Regards,

Akhil S

[Murali Krishnan]

Work done so far on these points: 

On Sat, Mar 31, 2018 at 11:34 AM, Akhil S <akhil...@gmail.com> wrote:

Hi,

Some review points till Stage 8 are given below.  

I have tried to stress on the points of confusion a novice reader might get.

 

Landing Page ( http://exposnitc.github.io/index.html )

--------------

1. Better to divide into sections.

2. short introduction, ( better not to use jargons like XSM etc in the first para as it may overwhelm the reader).

3. A section of what exactly you gain by doing this exercise.

4. This can be followed by the technical overview of the final product.

 

  Introduction is being modified, in which one sentence on what is done is added, but not much.  I believe giving a "technical" overview

  of the final object is difficult to digest for an S5 student. 

 

  

Stage 2

-----------

What is the relevance of XFS-interface

    explanation added. 

 

Why have we offloaded some major features about the File System from the kernel, (ideally the xfs-interface should only deal with writing  disk blocks, not files)

   This is somewhat difficult to explain in a simple way at that stage.  So, not done. 

 
 

Stage 3

----------

Mention why an assembler is not required.

   explanation added.  

XSM privileged mode execution tutorial

A diagram showing the interrupt vector table would be good.

  diagram will be added. 

 

Stage 4

----------------

Some general statement like, SPL is the language you will be writing the kernel in

Ask the reader to go through the generated machine code

   added. 

 

Stage 6

----------------

A para about what exactly an OS Startup code is and what it is supposed to do, before introducing the jargon for the first time.

   the jargon OS start up code has been introduced in stage 3 and stage 4 already.  An explanation is added as a general introduction to the

   work done by OS startup code. 

 

Can we remove the INT10 and exhandler from this stage, (eg the user pgm can go to an infinite loop and we can do Ctrl+c) for now

   Decided not to do that. 

 

I personally feel that this stage is a bit too large.

  Yes, but don't see a very simple way to improve things (unless lot of "repair" is done, and we fear side effects). 

 

Also, how about using the memory area as well, for eg the code can be like:

0   MOV R0, 1
2   MOV R2, [19]
4   GE R2, R0
6   JZ R2, 18
8   MOV R1, R0
10  MUL R1, R0
12  BRKP
14  ADD R0, 1
16  JMP 2
18  JMP 18     // Infinite loop
19  5 0        // Dont know if xfs-handler supports this

   Not yet done, but planning to incorporate memory access. 

 

Stage 7

-------------------

2048  0
2049  2056
2050  0
2051  0

This seems to be confusing. Is it possible to change into

2048  0 2056
2050  0 0
2052  0 0

  Well, this may be confusing as well.  So, decided not to change this. 

 

Who has written the Library ? Will the reader be asked to write it in future ( Since we are building an OS from scratch )

 

   There is an exercise in stage 11 to write a part of the library.  (Moreover, in S6 compiler lab, they have to write the library interface for Alloc, Free etc. and Read, Write system calls).  

--

Regards,

Akhil S