Re: [Lhms-devel] [PATCH] memory hotadd fixes [4/5] avoid check in acpi

[keith mannthey <kmannth@us.ibm.com>]

On Thu, 2006-08-03 at 11:28 -0700, keith mannthey wrote:
> On Thu, 2006-08-03 at 12:36 +0900, KAMEZAWA Hiroyuki wrote:
> > add_memory() does all necessary check to avoid collision.
> > then, acpi layer doesn't have to check region by itself.
> > 
> > (*) pfn_valid() just returns page struct is valid or not. It returns 0
> >     if a section has been already added even is ioresource is not added.
> >     ioresource collision check in mm/memory_hotplug.c can do more precise
> >     collistion check.
> >     added enabled bit check just for sanity check..
> > 
> > Signed-Off-By: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> > 
> > 
> >  drivers/acpi/acpi_memhotplug.c |    9 +--------
> >  1 files changed, 1 insertion(+), 8 deletions(-)
> > 
> > Index: linux-2.6.18-rc3/drivers/acpi/acpi_memhotplug.c
> > ===================================================================
> > --- linux-2.6.18-rc3.orig/drivers/acpi/acpi_memhotplug.c	2006-08-01 16:11:47.000000000 +0900
> > +++ linux-2.6.18-rc3/drivers/acpi/acpi_memhotplug.c	2006-08-02 14:12:45.000000000 +0900
> > @@ -230,17 +230,10 @@
> >  	 * (i.e. memory-hot-remove function)
> >  	 */
> >  	list_for_each_entry(info, &mem_device->res_list, list) {
> > -		u64 start_pfn, end_pfn;
> > -
> > -		start_pfn = info->start_addr >> PAGE_SHIFT;
> > -		end_pfn = (info->start_addr + info->length - 1) >> PAGE_SHIFT;
> > -
> > -		if (pfn_valid(start_pfn) || pfn_valid(end_pfn)) {
> > -			/* already enabled. try next area */
> > +		if (info->enabled) { /* just sanity check...*/
> >  			num_enabled++;
> >  			continue;
> >  		}
> 
> This check needs to go.  pfn_valid is a sparsemem specific check. Sanity
> checking should be done it the the add_memory code. 
> 
> I will test and let you know. This is going to expose some baddness I
> see already with my RESERVE path work. (Extra add_memory calls from this
> driver during boot....)

Ok.  This pfn_valid check needs to be inserted somewhere in the code
path for sparsemem hotadd.

with a debug statement in add_memory

Hotplug Mem Device
add_memory 0 400000000 70000000
System RAM resource 400000000 - 46fffffff cannot be added
add_memory 0 380000000 80000000
System RAM resource 380000000 - 3ffffffff cannot be added
add_memory 0 300000000 80000000
System RAM resource 300000000 - 37fffffff cannot be added
add_memory 0 280000000 80000000
System RAM resource 280000000 - 2ffffffff cannot be added
add_memory 0 200000000 80000000
System RAM resource 200000000 - 27fffffff cannot be added
add_memory 0 180000000 80000000
System RAM resource 180000000 - 1ffffffff cannot be added
add_memory 0 100000000 80000000
System RAM resource 100000000 - 17fffffff cannot be added
add_memory 0 100000 7ff00000

The box doesn't boot.   I am going to drop this patch and see about the
rest of the set.  (They seem sane and look ok but I want to test)

 The kernel needs to protect against bad calls to add_memory (and/or the
acpi driver needs to correctly id devices?)


keith mannthey <kmannth@us.ibm.com>
Linux Technology Center IBM