mmap.c and region.c in minix 3.1.4

[snell]

hi all,
Can anyone tell me the work done in mmap.c and region.c??

and from where is this do_map() of mmap.c called ??... CALLMAP
(VM_MMAP, do_mmap, ANYEPM);

[Guanqun Lu]

On Sat, May 16, 2009 at 2:30 PM, snell <snel...@gmail.com> wrote:
>
> hi all,
> Can anyone tell me the work done in mmap.c and region.c??

mmap.c, i'm afraid is easy to understand, as long as you have a try
with 'mmap()' and munmap() function in unix-like operating system.
region.c contains the operation of managing virtual addresses in process.

>
> and from where is this do_map() of mmap.c called ??... CALLMAP
> (VM_MMAP, do_mmap, ANYEPM);

ANYEPM means every process can call 'do_mmap'.
do_mmap() is called by c library implementation.

> >
>



--
Guanqun

[snell]

what does map_page_region() do ??

[Guanqun Lu]

On Sat, May 16, 2009 at 3:14 PM, snell <snel...@gmail.com> wrote:
>
> what does map_page_region() do ??

It checks whether a region is available to use, technically whether a
hole between two consecutive regions is availabe to fit the lengh
required. And map the physical address into the page table. The
physical address can be slab-allocated or user specified.

HTH.

But I'm afraid you should get a overview of what's going on in memory
management first. Otherwise, you will be lost in the mass of details.

>
> >
>



--
Guanqun

[snell]

When does page fault occurs and from where is the pagefault called??

[Guanqun Lu]

On Sat, May 16, 2009 at 6:20 PM, snell <snel...@gmail.com> wrote:
>
> When does page fault occurs and from where is the pagefault called??

Look at servers/vm/pagefaults.c, you'll get some clues.

> >
>



--
Guanqun

[snell]

Do we have a fixed virtual memory in minix 3.1.4 .... if so what it
is ...??
in windows we can change the virtual memory .. can we do that in minix
also ??????

[Guanqun Lu]

Sorry, I don't understand your problem.
What do you mean by "fixed" virtual memory?

PS: DO NOT use so many question marks, it looks annoying, and it won't
help you get more answers.

--
Guanqun

[snell]

sorry , if that irritates you ..

am trying to run this simple program

#include<stdio.h>
int main()
{
int a[1000][100];
int b[1000][100];
return 0;
}

it gives the error ::
VM:pagefault : SIGSEGV 71175 bad addr 0x7fedaf2c err 0x6 nopage write
a.out 71175 0x59 0x49
PM:signal 11 for 173 / a.out
dumpcore for 173/ a.out
memory fault (core dumped)

[Alexander Ray]

I just compiled and ran your test program and it worked fine (compiled with cc).

What version are you using?

[Rafael Vargas]

On Sun, May 17, 2009 at 1:04 PM, snell <snel...@gmail.com> wrote:
> it gives the error ::
> VM:pagefault : SIGSEGV 71175 bad addr 0x7fedaf2c err 0x6 nopage write
> a.out 71175 0x59 0x49
> PM:signal 11 for 173 / a.out
> dumpcore for 173/ a.out
> memory fault (core dumped)

Where are you running this program at (Virtual Machine, Bare metal, etc)?

--
Rafael Vargas
http://rafavargas.com

[Rahmat M. Samik-Ibrahim]

Rafael Vargas wrote:

> snell wrote:
>> it gives the error ::
>> VM:pagefault : SIGSEGV 71175 bad addr 0x7fedaf2c err 0x6 nopage write
>> a.out 71175 0x59 0x49
>> PM:signal 11 for 173 / a.out
>> dumpcore for 173/ a.out
>> memory fault (core dumped)
> Where are you running this program at (Virtual Machine, Bare metal, etc)?

I have just tried to replicate this.
I am running it on a Pentium4/HT 3GHz, Ubuntu 9.04,
VMware-server 2.0.1 with Minix 3.1.4

The /var/log/messages output is:
May 23 10:13:46 172 kernel: VM: pagefault: SIGSEGV 71109 bad addr
0x7ff3c908 err 0x6 nopage write
May 23 10:13:46 172 kernel: coba 71109 0x59 0x49
May 23 10:13:46 172 kernel: PM: signal 11 for 119 / coba
May 23 10:13:46 172 kernel: dumpcore for 119 / coba

when executing the program "coba" above
(+ "Memory fault (core dumped)").

--
Rahmat M. Samik-Ibrahim --- vLSM.org

[ravi prakash]

as far as in know i dont thin minix has virtual memory thats the reason when v have many processes the page tables become full and v can run any more processes v have to suspend the virtaul machine completely
pls correct me if its wrong

--
RAVIPRAKASH D
NITK Surathkal

[Ben Gras]

This is because the given stack size (in virtual address space) isn't
big enough to contain the

int a[1000][100];
int b[1000][100];

so increase the virtual stack region (currently the chmem value is
used, default 128k) or declare those variables globally.