GATE CSE 1999 | Question: 2.11

option B says: Processes "can be" given protected address spaces.

without  concept of virtual memory then  CPU can also generates virtual address (y/n) ?? - yes we can give virtual address( i am not sure i am thinking virtual adress as a logical address). 

@ mcjoshi  even "can " present in c stmt also so c can also be true if we consider b as true because of "can be". 

This qstn is asking advantage over other memory management techniues, i think only d will be correct

@arjun Sir I think C should also be correct as question is asking only for advantages and not comparitive advantage over traditional methods. Also , by using VM

Linker "can" assign addresses independent of where the program will be loaded in physical memory.

@Arjun sir if procees size is 100 MB,what or when does that process need more than 100 MB of memory in future?

But @Arjun sir protection in not the feature that is limited to VM, even in contiguous allocation we can have protection by that limit and base register.. Isn't it so?

@ Arjun sir, Why cannot we use variable partition with Base and Relocation registers. That will also provide protected address space to the processes

Yes, it will work for uniprocessor system. @pradeep_reddy But not for multiprocessor system.

I think only D is the correct answer. Protection to the memory can also be given using segmentation and using descriptor tables. No need of virtual memory for it.

reference: https://pdos.csail.mit.edu/6.828/2005/readings/i386/s05_01.htm

@khushitshah I believe that segmentation also comes under virtual memory. So, without virtual memory being applied how can you use segmentation?

Virtual Memory implements 4 systems : 

1] Paging , 2] Segmentation , 3] Paged Segmentation and 4] Segmented Paging.

@Arjun Sir, segments are placed in a common virtual address space defined by their names and lengths. They can reside in the main memory or in the auxiliary store. So, segmentation is performed in the context of virtual memory while implementing physical memory. Hence, virtual memory is needed for implementing segmentation. Is my reasoning correct Sir?

@ankit3009 
Segmentation means we are dividing memory in segments. 

Virtual Memory is using part of secondary memory as an extension to main memory. 

Virtual memory is not needed to implement segmentation.
 

@ankit3009 Yes, virtual memory is a concept and segmentation is a way to implement it. This should be given in any standard OS book. 

@khushitshah

Thanks, I read Galvin and you were right.

Segmentation works on logical addressing with the combination of different hardware. Sorry for the wrong comments. 

@Arjun Sir, what's the reason behind this statement: "Segmentation can be used as protection for uniprocessor system but can't be used for multiprocessor system"?

What I think is segmentation is capable of using the shared memory concept, that's the reason segmentation can't be used for multiprocessors. Correct me if I am wrong.

Please help.

@ankit3009 (I didn't notice your doubt was solved, glad to help).
@gatecse
Here is the explanation, if it helps.

Yes, segmentation does not need virtual memory to implement it.

In segmentation, for every segment we define a (base, limit) pair. Now, how is this useful?

1. We can provide access protection by enforcing: $ip < limit$.
2. We can provide protected address space using local descriptor tables.
3.  It allows code relocatability.

Segmentation and cache are totally different concepts. Cache is used to store frequently used data in faster memory

segmentation may use virtual memory. But, it doesn’t need it. OS may swap segments in and out of memory. But virtual memory is not what provides protected address space. Segmentation provides protected address space. 

@Arjun @gatecse sir I have thought in this way:

→ Through Memory Management Unit(MMU) we can check that if CPU generates some address that is outside the range allowed for P. Then such access will be blocked.

→ Suppose the CPU generates an address within the range of P and some write operation needs to be done on that address. But suppose for that address write operation is not permitted. How to restrict such activity? For that, we need a page table. The page table has an entry with permission bits and also through paging we implement the virtual memory.

https://www.slideserve.com/abbott/virtual-memory-powerpoint-ppt-presentation

https://computationstructures.org/lectures/vm/vm.html#7

So option B is also true.

@samarpita If we are allocating separate address space for a process $P$ why can’t we also allocate separate “read only address space” and “read/write address space”?

@gatecse sir we know that the code segment of a process is given read permission. Now if the CPU generates an address of this code segment of the process P and wants to write on this code segment ….so such action needs to be checked.

@gatecse Sir, I am just telling the purpose of “permission bits of PT” and MMU and just trying to tell that though we have MMU why do we need “permission bits of PT.”