question about iotable_init( ) function

[Wang, Yalin]

Hi  Will, 

Oh,  I  see,

Thanks for your clarification very much !

-----Original Message-----
From: Will Deacon [mailto:will.deacon at arm.com] 
Sent: Monday, July 01, 2013 5:27 PM
To: Wang, Yalin
Cc: 'Russell King - ARM Linux'; 'linux-arm-kernel at lists.infradead.org'; 'linux-arch at vger.kernel.org'
Subject: Re: question about iotable_init( ) function

On Mon, Jul 01, 2013 at 08:31:38AM +0100, Wang, Yalin wrote:
> Hi  Will,
> 
> I have a question about iotable_init()  function (arch/arm/mm/mmu.c) I 
> found  in this function it only allocate a vm_struct and insert it 
> into vmlist , But it didn’t allocate a vmap_area  to insert it  into vmap_area_root tree (mm/vmalloc.c).
> 
> So this means that the virtual address reserverd by iotable_init is 
> not declared In vmap_area_root  ,  this is not safe ,  maybe some driver will use this virtual address as vmalloc again,
> I think it ‘s not safe in this condition ,  should also add a vmap_area   node into vmap_area_root to
> Hold this virtual address not used by others .

The vmap_area structures are allocated and inserted later on by vmalloc_init (iotable init is for static mappings early during boot).

Will