Re: [PATCH] mm: make reserve_bootmem can crossed the nodes

["Yinghai Lu" <yhlu.kernel@gmail.com>]

On Thu, Mar 13, 2008 at 6:55 PM, KAMEZAWA Hiroyuki
<kamezawa.hiroyu@jp.fujitsu.com> wrote:
> On Thu, 13 Mar 2008 18:35:51 -0700
>
> "Yinghai Lu" <yhlu.kernel@gmail.com> wrote:
>
>  > On Thu, Mar 13, 2008 at 6:34 PM, KAMEZAWA Hiroyuki
>  > <kamezawa.hiroyu@jp.fujitsu.com> wrote:
>  > > On Thu, 13 Mar 2008 16:45:42 -0700
>  > >  "Yinghai Lu" <yhlu.kernel@gmail.com> wrote:
>  > >  > int __init reserve_bootmem(unsigned long addr, unsigned long size,
>  > >  >                           int flags)
>  > >  >  {
>  > >  > -     return reserve_bootmem_core(NODE_DATA(0)->bdata, addr, size, flags);
>  > >  > +     int ret;
>  > >  > +     bootmem_data_t *bdata;
>  > >  > +     list_for_each_entry(bdata, &bdata_list, list) {
>  > >  > +             ret = can_reserve_bootmem_core(bdata, addr, size, flags);
>  > >  > +             if (ret < 0)
>  > >  > +                     return ret;
>  > >  > +     }
>  > >  > +     list_for_each_entry(bdata, &bdata_list, list)
>  > >  > +             reserve_bootmem_core(bdata, addr, size, flags);
>  > >  > +     return 0;
>  > >  >  }
>  > >
>  > >  why list_for_each twice ?
>  >
>  > first_for_each only check if we can reserve that.
>  > second will do the reserve job.
>  >
>  Hmm, can a call to reserve_bootmem() return a memory region which spread across
>  prural nodes ?

Yes. there is one reserve_bootmem in reserve_crashkernel. and position
is passed via commandline...
there is some chance for ramdisk too. it is called via reserve_bootmem_generic.

YH