Memory problems in schedule()

Memory problems in schedule()

[Vishal Patil]

Hi

I am trying to implement a different process scheduling algorithm in linux
kernel 2.4.30. I am doing this as a hobby and am using User Mode Linux
to test my changes. I have the following problems with the implement

1) Whenever I select a process to run using my algorithm the kernel
panics with "Segfault with no mm" basically the "mm" field of the
task_struct that I selected is empty. I don't understand why this
should happen, since I have just added code to select a process and
haven't modified any memory related code in the schedule() function.

2) I am able to run the UMLfied kernel under gdb, however the execution
never halts even though I set several breakpoints. Also these
breakpoints have not been set in interrupt handling code and I have
compiled the code with -g option.

I have been stuck with these problems since a long time and would
appreciate if someone could help me with it.

Thank you all.

- Vishal 

-- 
A dream is just a dream. A goal is a dream with a plan and a deadline.

Re: Memory problems in schedule()

[Jeff Dike]

On Sun, Jun 05, 2005 at 09:01:19PM -0400, Vishal Patil wrote:
> 1) Whenever I select a process to run using my algorithm the kernel
> panics with "Segfault with no mm" basically the "mm" field of the
> task_struct that I selected is empty. I don't understand why this
> should happen, since I have just added code to select a process and
> haven't modified any memory related code in the schedule() function.

Start with a backtrace from the panic and debug it from there.

> 2) I am able to run the UMLfied kernel under gdb, however the execution
> never halts even though I set several breakpoints. Also these
> breakpoints have not been set in interrupt handling code and I have
> compiled the code with -g option.

UML and gdb versions?  tt or skas mode?

				Jeff

Re: Memory problems in schedule()

[Vishal Patil]

Jeff
Please find the answer to questions below,

On 6/5/05, Jeff Dike <jdike@addtoit.com> wrote:
> On Sun, Jun 05, 2005 at 09:01:19PM -0400, Vishal Patil wrote:
> > 1) Whenever I select a process to run using my algorithm the kernel
> > panics with "Segfault with no mm" basically the "mm" field of the
> > task_struct that I selected is empty. I don't understand why this
> > should happen, since I have just added code to select a process and
> > haven't modified any memory related code in the schedule() function.
> 
> Start with a backtrace from the panic and debug it from there.
> 
> > 2) I am able to run the UMLfied kernel under gdb, however the execution
> > never halts even though I set several breakpoints. Also these
> > breakpoints have not been set in interrupt handling code and I have
> > compiled the code with -g option.
> 
> UML and gdb versions?  tt or skas mode?


GNU gdb Red Hat Linux (6.1post-1.20040607.41rh)
uml-patch-2.4.27-1
skas mode.

Thank you.
>                                 Jeff
> 


-- 
A dream is just a dream. A goal is a dream with a plan and a deadline.

Re: Memory problems in schedule()

[Jeff Dike]

On Mon, Jun 06, 2005 at 08:24:39AM -0400, Vishal Patil wrote:
> GNU gdb Red Hat Linux (6.1post-1.20040607.41rh)
> uml-patch-2.4.27-1
> skas mode.

Disable CONFIG_MODE_TT.  gdb seems confused by the exec that tt mode does
early on.

				Jeff