use_mm/unuse_mm correctness

use_mm/unuse_mm correctness

[Suparna Bhattacharya]

Can anyone spot a problem in the following routines ?

These are used by AIO workqueue routines to take on the 
caller's address space when executing certain operations,
and then to switch back to the workqueue thread's original 
mm context. 

We are seeing some strange bugs in -mm lately, and this
code is one of the suspects. Can't yet see what could be
wrong ...

/*
 * use_mm
 * 	Makes the calling kernel thread take on the specified 
 * 	mm context. 
 * 	Called by the retry thread execute retries within the 
 * 	iocb issuer's mm context, so that copy_from/to_user
 * 	operations work seamlessly for aio.
 * 	(Note: this routine is intended to be called only 
 * 	from a kernel thread context)
 */
static void use_mm(struct mm_struct *mm)
{
	struct mm_struct *active_mm = current->active_mm;
	atomic_inc(&mm->mm_count);
	current->mm = mm;

	current->active_mm = mm;
	activate_mm(active_mm, mm);

	mmdrop(active_mm);
}

/*
 * unuse_mm
 * 	Reverses the effect of use_mm, i.e. releases the
 * 	specified mm context which was earlier taken on
 * 	by the calling kernel thread 
 * 	(Note: this routine is intended to be called only 
 * 	from a kernel thread context)
 */
void unuse_mm(struct mm_struct *mm)
{
	current->mm = NULL;
	/* active_mm is still 'mm' */
	enter_lazy_tlb(mm, current, smp_processor_id());
}

Regards
Suparna

-- 
Suparna Bhattacharya (suparna@in.ibm.com)
Linux Technology Center
IBM Software Labs, India

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: use_mm/unuse_mm correctness

[Hugh Dickins]

On Mon, 16 Jun 2003, Suparna Bhattacharya wrote:
> Can anyone spot a problem in the following routines ?

If CONFIG_PREEMPT=y, then this might help:

--- 2.5.71-mm1/fs/aio.c	Sun Jun 15 12:36:09 2003
+++ linux/fs/aio.c	Mon Jun 16 07:05:53 2003
@@ -582,7 +582,8 @@ void unuse_mm(struct mm_struct *mm)
 {
 	current->mm = NULL;
 	/* active_mm is still 'mm' */
-	enter_lazy_tlb(mm, current, smp_processor_id());
+	enter_lazy_tlb(mm, current, get_cpu());
+	put_cpu();
 }
 
 /*

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: use_mm/unuse_mm correctness

[Suparna Bhattacharya]

get_cpu() would be the right thing to do - thanks for 
pointing that out.

However, in the aio case, use_mm and unuse_mm are called 
only by workqueue threads, so there shouldn't be any 
migration even if a pre-empt occurs (cpus_allowed is fixed 
to a particular cpu), should it ?

Regards
Suparna 

On Mon, Jun 16, 2003 at 07:16:12AM +0100, Hugh Dickins wrote:
> On Mon, 16 Jun 2003, Suparna Bhattacharya wrote:
> > Can anyone spot a problem in the following routines ?
> 
> If CONFIG_PREEMPT=y, then this might help:
> 
> --- 2.5.71-mm1/fs/aio.c	Sun Jun 15 12:36:09 2003
> +++ linux/fs/aio.c	Mon Jun 16 07:05:53 2003
> @@ -582,7 +582,8 @@ void unuse_mm(struct mm_struct *mm)
>  {
>  	current->mm = NULL;
>  	/* active_mm is still 'mm' */
> -	enter_lazy_tlb(mm, current, smp_processor_id());
> +	enter_lazy_tlb(mm, current, get_cpu());
> +	put_cpu();
>  }
>  
>  /*
> 

-- 
Suparna Bhattacharya (suparna@in.ibm.com)
Linux Technology Center
IBM Software Labs, India

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: use_mm/unuse_mm correctness

[Hugh Dickins]

On Mon, 16 Jun 2003, Suparna Bhattacharya wrote:
> 
> However, in the aio case, use_mm and unuse_mm are called 
> only by workqueue threads, so there shouldn't be any 
> migration even if a pre-empt occurs (cpus_allowed is fixed 
> to a particular cpu), should it ?

Ah, yes, I certainly hope the cpu can't change in such a case!
Sorry for the noise, I hope someone else can help,
Hugh

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: use_mm/unuse_mm correctness

[Venu Vadapalli]

Do kernel threads have 'mm'? I thought they inherit the active_mm from the
previously running task or in this case just use the desired mm...


This email message (including attachments) contains Company Proprietary
and/or Confidential Information and is intended only for the addressee.  Any
unauthorized review, use, disclosure or distribution is prohibited.  If you
are not the intended recipient, please contact the sender by reply email and
destroy all copies of the original message.  Thank you. 

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>