--- On Fri, 6/25/10, Mulyadi Santosa <mulyadi.santosa@xxxxxxxxx> wrote:
> From: Mulyadi Santosa <mulyadi.santosa@xxxxxxxxx>
> Subject: Re: Accessing kernel memory from multiple modules
> To: "Usman S. Ansari" <uansari@xxxxxxxxx>
> Cc: "Linux Kernel List" <kernelnewbies@xxxxxxxxxxxx>
> Date: Friday, June 25, 2010, 9:01 PM
> Dear Usman...
>
> Sorry if I can't give complete explanation, but hopefully
> it's enough...
>
> On Sat, Jun 26, 2010 at 02:31, Usman S. Ansari <uansari@xxxxxxxxx>
> wrote:
> > I have two kernel modules running.
> >
> > Module 1:
> > - struct a
> > {
> > int val1;
> > char *buffer
> > ...
> > };
> >
> > - Then in module 1, I have following global
> declaration
> > struct a buf[20];
> >
> > - Module 1, calls a exported function
> "module2_function(int *)" from modules 2:
> > it passes &buff[0].val1 as a parameter to the
> module2_functon
> >
> > Module 2:
> > - I assign buff[0].val1 to variable defined as *i
> > - I print *i, and see same data which was assigned by
> module 1, i.e. 0x9595
> > - Have following instruction
> > - *i = 1;
> > - print *i again and new values i.e. 1 is printed
> > - after this in less than a second whole freezes
> >
> > After debugging for whole day, I tried following,
> which seems to fix this issue
> > - I changed struct a buf[20]; to kmalloc
> > - now system does not freez
>
> If my eyes doesn't trick me, initially it's both uninitialized and
> unallocated. Are we agree here?
No. This array to struct is defined as a global in module # 1, where is it accessed (read and written). Issue comes in when I pass pointer to module # 2 and it is written to (read is OK).
> So then, when you gave it a value, kernel tries to store
> it, but realized the "container" isn't prepared. Unfortunately,
> kernels doesn't do page fault the same way we see in user space.
Don't understand what you are saying.
--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ
