Should page->count ever be -1?

Should page->count ever be -1?

[Timur Tabi]

I'm experiencing some kind of memory leaks playing with ioremap and iounmap,
and I've narrowed down the problem to iounmap refusing to unmap the memory that
I just mapped.  The line of code in question is

	if (!PageReserved(page) && atomic_dec_and_test(&page->count)) {

from page_alloc.c (this is 2.2.18pre15).  It appears that page->count is
already zero when this code is executed, and after it's executed, page->count
becomes -1 (or more accurately, 0xFFFFFFFF).  Is this acceptable, or is it an
error condition?


-- 
Timur Tabi - ttabi@interactivesi.com
Interactive Silicon - http://www.interactivesi.com

When replying to a mailing-list message, please direct the reply to the mailing list only.  Don't send another copy to me.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Should page->count ever be -1?

[Andrea Arcangeli]

On Wed, Jan 03, 2001 at 03:07:03PM -0600, Timur Tabi wrote:
> I'm experiencing some kind of memory leaks playing with ioremap and iounmap,
> and I've narrowed down the problem to iounmap refusing to unmap the memory that
> I just mapped.  The line of code in question is
> 
> 	if (!PageReserved(page) && atomic_dec_and_test(&page->count)) {
> 
> from page_alloc.c (this is 2.2.18pre15).  It appears that page->count is
> already zero when this code is executed, and after it's executed, page->count
> becomes -1 (or more accurately, 0xFFFFFFFF).  Is this acceptable, or is it an
> error condition?

It's an error condition. Make sure you marked the page as reserved in the mmap
callback if it's not an mmio region outside RAM.

Andrea
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Should page->count ever be -1?

[Timur Tabi]

** Reply to message from Andrea Arcangeli <andrea@suse.de> on Wed, 3 Jan 2001
22:55:05 +0100


> > from page_alloc.c (this is 2.2.18pre15).  It appears that page->count is
> > already zero when this code is executed, and after it's executed, page->count
> > becomes -1 (or more accurately, 0xFFFFFFFF).  Is this acceptable, or is it an
> > error condition?
> 
> It's an error condition. Make sure you marked the page as reserved in the mmap
> callback if it's not an mmio region outside RAM.

I mark the page as reserved when I call ioremap and then unmark it after
ioremap returns, but iounmap will fail if it's still reserved (the same line of
code also checks the reserved bit):

 	if (!PageReserved(page) && atomic_dec_and_test(&page->count)) {
            ^^^^^^^^^^^^^^^^^^^

Is this what you're saying?

iounmap just calls vfree, which does all the work.  However, I'm confused at
this code in vfree:

	vmfree_area_pages(VMALLOC_VMADDR(tmp->addr), tmp->size);
	kfree(tmp);

What's the difference between vmfree_area_pages and kfree?


-- 
Timur Tabi - ttabi@interactivesi.com
Interactive Silicon - http://www.interactivesi.com

When replying to a mailing-list message, please direct the reply to the mailing list only.  Don't send another copy to me.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/