Re: [PATCH v2 3/3] m68k: switch to MEMBLOCK + NO_BOOTMEM
From: Michal Hocko
Date: Wed Jul 04 2018 - 09:14:12 EST
On Wed 04-07-18 15:05:08, Geert Uytterhoeven wrote:
> Hi Michal,
> On Wed, Jul 4, 2018 at 2:36 PM Michal Hocko <mhocko@xxxxxxxxxx> wrote:
> > [CC Andrew - email thread starts
> > http://lkml.kernel.org/r/1530685696-14672-1-git-send-email-rppt@xxxxxxxxxxxxxxxxxx]
> > OK, so here we go with the full patch.
> > From 0e8432b875d98a7a0d3f757fce2caa8d16a8de15 Mon Sep 17 00:00:00 2001
> > From: Michal Hocko <mhocko@xxxxxxxx>
> > Date: Wed, 4 Jul 2018 14:31:46 +0200
> > Subject: [PATCH] memblock: do not complain about top-down allocations for
> > !MEMORY_HOTREMOVE
> > Mike Rapoport is converting architectures from bootmem to noboodmem
> > allocator. While doing so for m68k Geert has noticed that he gets
> > a scary looking warning
> > WARNING: CPU: 0 PID: 0 at mm/memblock.c:230
> > memblock_find_in_range_node+0x11c/0x1be
> > memblock: bottom-up allocation failed, memory hotunplug may be affected
> > The warning is basically saying that a top-down allocation can break
> > memory hotremove because memblock allocation is not movable. But m68k
> > doesn't even support MEMORY_HOTREMOVE is there is no point to warn
> so there is
> > about it.
> > Make the warning conditional only to configurations that care.
> Still, I'm wondering if the warning is really that unlikely on systems
> that support
> hotremove. Or is it due to the low amount of RAM on m68k boxes?
Most likely yes. If you want to have full NUMA nodes hot-removable then
the BIOS/FW is supposed to mark them hotplug and then we rely on the
available memory on the low physical memory ranges (usually on not 0)
to cover all early boot allocations. Hack? Sure thing like the whole
memory hotremove, if you ask me.