i386: Why putting __USER_DS in kernel threads stack initialization?

i386: Why putting __USER_DS in kernel threads stack initialization?

[Ahmed S. Darwish]

Hi list,

Reading the kernel threads initialization code I see:

int kernel_thread(...) {

	struct pt_regs regs;
	memset(&regs, 0, sizeof(regs));
	[...]
**	regs.xds = __USER_DS;
**	regs.xes = __USER_DS;
	[...]
	/* Ok, create the new process.. */
	return do_fork(flags | CLONE_VM | CLONE_UNTRACED, 0, &regs, \
	       	       0, NULL, NULL);

Continuing with the code, the threads stack (beginning from %esp) is
initialized with the passed *regs from do_fork:

int copy_thread(..., struct task_struct *p, struct pt_regs *regs) {

	struct pt_regs * childregs;
	struct task_struct *tsk;
 	childregs = task_pt_regs(p);
**	*childregs = *regs;
	[...]
** 	p->thread.esp = (unsigned long) childregs;


So the question is what will a _kernel_ thread do with the Usermode Segment
address ?

Thanks,

P.S. I've tried commenting out both lines which led to a non functional init,
Also setting them to __USER_DS made init start but stopped issuing the error:
`Panic: Segment violation at 0x8049798 - Sleeping for 30 seconds'

-- 
Ahmed S. Darwish
http://darwish.07.googlepages.com

Re: i386: Why putting __USER_DS in kernel threads stack initialization?

[Ahmed S. Darwish]

On Mon, Mar 19, 2007 at 12:58:31AM +0200, ahmed wrote:
> 
> P.S. I've tried commenting out both lines which led to a non functional init,
> Also setting them to __USER_DS made init start but stopped issuing the error:
> `Panic: Segment violation at 0x8049798 - Sleeping for 30 seconds'
> 

Sorry, I meant setting them to __KERNEL_DS.

Thanks,

-- 
Ahmed S. Darwish
http://darwish.07.googlepages.com

Re: i386: Why putting __USER_DS in kernel threads stack initialization?

[linux-os (Dick Johnson)]

On Sun, 18 Mar 2007, Ahmed S. Darwish wrote:

> Hi list,
>
> Reading the kernel threads initialization code I see:
>
> int kernel_thread(...) {
>
> 	struct pt_regs regs;
> 	memset(&regs, 0, sizeof(regs));
> 	[...]
> **	regs.xds = __USER_DS;
> **	regs.xes = __USER_DS;
> 	[...]
> 	/* Ok, create the new process.. */
> 	return do_fork(flags | CLONE_VM | CLONE_UNTRACED, 0, &regs, \
> 	       	       0, NULL, NULL);
>
> Continuing with the code, the threads stack (beginning from %esp) is
> initialized with the passed *regs from do_fork:
>
> int copy_thread(..., struct task_struct *p, struct pt_regs *regs) {
>
> 	struct pt_regs * childregs;
> 	struct task_struct *tsk;
> 	childregs = task_pt_regs(p);
> **	*childregs = *regs;
> 	[...]
> ** 	p->thread.esp = (unsigned long) childregs;
>
>
> So the question is what will a _kernel_ thread do with the Usermode Segment
> address ?
>
> Thanks,
>
> P.S. I've tried commenting out both lines which led to a non functional init,
> Also setting them to __USER_DS made init start but stopped issuing the error:
> `Panic: Segment violation at 0x8049798 - Sleeping for 30 seconds'
>
> -- 
> Ahmed S. Darwish
> http://darwish.07.googlepages.com

You might be confusing two routines. The kernel thread routine sets
DS and ES to the kernel data segment, __KERNEL_DS, not the user data
segment. This is so the kernel thread can access the kernel data. Note
that this is done by putting the values in the pt_regs structure so
it doesn't happen 'now', but after the fork.

Cheers,
Dick Johnson
Penguin : Linux version 2.6.16.24 on an i686 machine (5592.68 BogoMips).
New book: http://www.AbominableFirebug.com/
_
\x1a\x04

****************************************************************
The information transmitted in this message is confidential and may be privileged.  Any review, retransmission, dissemination, or other use of this information by persons or entities other than the intended recipient is prohibited.  If you are not the intended recipient, please notify Analogic Corporation immediately - by replying to this message or by sending an email to DeliveryErrors@analogic.com - and destroy all copies of this information, including any attachments, without reading or disclosing them.

Thank you.

Re: i386: Why putting __USER_DS in kernel threads stack initialization?

[Ahmed S. Darwish]

On Mon, Mar 19, 2007 at 07:23:25AM -0400, linux-os (Dick Johnson) wrote:
> 
> On Sun, 18 Mar 2007, Ahmed S. Darwish wrote:
> 
> > Hi list,
> >
> > Reading the kernel threads initialization code I see:
> >
> > int kernel_thread(...) {
> >
> > 	struct pt_regs regs;
> > 	memset(&regs, 0, sizeof(regs));
> > 	[...]
> > **	regs.xds = __USER_DS;
> > **	regs.xes = __USER_DS;
> > 	[...]
> > 	/* Ok, create the new process.. */
> > 	return do_fork(flags | CLONE_VM | CLONE_UNTRACED, 0, &regs, \
> > 	       	       0, NULL, NULL);
> >
> > Continuing with the code, the threads stack (beginning from %esp) is
> > initialized with the passed *regs from do_fork:
> >
> > int copy_thread(..., struct task_struct *p, struct pt_regs *regs) {
> >
> > 	struct pt_regs * childregs;
> > 	struct task_struct *tsk;
> > 	childregs = task_pt_regs(p);
> > **	*childregs = *regs;
> > 	[...]
> > ** 	p->thread.esp = (unsigned long) childregs;
> >
> >
> > So the question is what will a _kernel_ thread do with the Usermode Segment
> > address ?
> >
> > Thanks,
> >
> > P.S. I've tried commenting out both lines which led to a non functional init,
> > Also setting them to __USER_DS made init start but stopped issuing the error:
> > `Panic: Segment violation at 0x8049798 - Sleeping for 30 seconds'
> >

Sorry, I meant "setting them to __KERNEL_DS" here.

> 
> You might be confusing two routines. The kernel thread routine sets
> DS and ES to the kernel data segment, __KERNEL_DS, not the user data
> segment. 

And that's what's _not_ happening in the code as I mentioned in original post.

> This is so the kernel thread can access the kernel data. Note
> that this is done by putting the values in the pt_regs structure so
> it doesn't happen 'now', but after the fork.

I've searched the code for such case (setting xds to __KERNEL_DS _After_ 
copy_thread()) with no success. As I understand, the kernel thread 
executes the passed function immediately (when given control by scheduler):

i386/kernel/process::kernel_thread():
**	regs.ebx = (unsigned long) fn;
	regs.edx = (unsigned long) arg;
	regs.xds = __USER_DS;
	regs.xes = __USER_DS;
	regs.xfs = __KERNEL_PDA;
	regs.orig_eax = -1;
**	regs.eip = (unsigned long) kernel_thread_helper;
	do_fork(...)

entry.S::kernel_thread_helper (removing CFI_* pseudo ops):

   ENTRY(kernel_thread_helper)
	pushl $0		
	movl %edx,%eax
	push %edx
**	call *%ebx
	push %eax
	call do_exit
	
Am I interpreting the forking process completely wrong?. I'm just curious why 
the __USER_DS is playing a vital rule in kernel threads regs/stack ?

Thanks alot,

-- 
Ahmed S. Darwish
http://darwish.07.googlepages.com