Re: Weirdness in __alloc_bootmem_node_high

[Sam Ravnborg <sam@ravnborg.org>]

On Fri, Apr 20, 2012 at 12:30:54PM -0700, Yinghai Lu wrote:
> On Fri, Apr 20, 2012 at 12:14 PM, Sam Ravnborg <sam@ravnborg.org> wrote:
> >
> > I took a quick look at this.
> > __alloc_bootmem_node_high() is used in mm/sparse.c - but only
> > if SPARSEMEM_VMEMMAP is enabled.
> >
> > mips has this:
> >
> > config ARCH_SPARSEMEM_ENABLE
> >        bool
> >        select SPARSEMEM_STATIC
> >
> > So SPARSEMEM_VMEMMAP is not enabled.
> >
> > __alloc_bootmem_node_high() is used in mm/sparse-vmemmap.c which
> > also depends on CONFIG_SPARSEMEM_VMEMMAP.
> >
> >
> > So I really do not see the logic in __alloc_bootmem_node_high()
> > being used anymore and it can be replaced by __alloc_bootmem_node()
> 
> Yes, you are right. __alloc_bootmem_node_high could be removed.
> 
> BTW, x86 is still the only one that use NO_BOOTMEM.
> 
> Are you working on making sparc to use NO_BOOTMEM?

For now I am trying to convert sparc32 to
use memblock and NO_BOOTMEM in one step.

I have it almost finished - except that it does not work :-(
We have limitations in what area we can allocate very early,
and here I had to use the alloc_bootmem_low() variant.
I had preferred a variant that allowed me to allocate
bottom-up in this case.

For now I assume something is fishy in my code where I
hand over memory to the buddyallocator.
But before posting anything I need time to go through
my code and divide it up in smaller patches.

There is so far no changes to nobootmem / memblock code.

I will most likely convert sparc64 to NO_BOOTMEM next,
if it looks reasonable simple that is.
But first step is to get sparc32 working.

	Sam

--
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/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>