On 03/02/2016 01:32 AM, Dan Williams wrote:
> ZONE_DEVICE (merged in 4.3) and ZONE_CMA (proposed) are examples of new
> mm zones that are bumping up against the current maximum limit of 4
> zones, i.e. 2 bits in page->flags for the GFP_ZONE_TABLE.
>
> The GFP_ZONE_TABLE poses an interesting constraint since
> include/linux/gfp.h gets included by the 32-bit portion of a 64-bit
> build. We need to be careful to only build the table for zones that
> have a corresponding gfp_t flag. GFP_ZONES_SHIFT is introduced for this
> purpose. This patch does not attempt to solve the problem of adding a
> new zone that also has a corresponding GFP_ flag.
>
> Vlastimil points out that ZONE_DEVICE, by depending on x86_64 and
> SPARSEMEM_VMEMMAP implies that SECTIONS_WIDTH is zero. In other words
^ by default
Because CONFIG_SPARSEMEM_VMEMMAP can still be disabled by the user.
> even though ZONE_DEVICE does not fit in GFP_ZONE_TABLE it is free to
> consume another bit in page->flags (expand ZONES_WIDTH) with room to
> spare.
So it's still possible to configure the x86_64 kernel such that you get
"#warning Unfortunate NUMA and NUMA Balancing config". But it requires
some effort to override the defaults, and it's not breaking build or
runtime. BTW I was able to get that warning even with your previous
patch that limited NODES_WIDTH, so that wasn't a solution for this
anyway. This patch is simpler and better.
> Link: https://bugzilla.kernel.org/show_bug.cgi?id=110931
> Fixes: 033fbae988fc ("mm: ZONE_DEVICE for "device memory"")
> Cc: Mel Gorman <[email protected]>
> Cc: Rik van Riel <[email protected]>
> Cc: Joonsoo Kim <[email protected]>
> Cc: Dave Hansen <[email protected]>
> Cc: Sudip Mukherjee <[email protected]>
> Reported-by: Mark <[email protected]>
> Reported-by: Vlastimil Babka <[email protected]>
> Signed-off-by: Dan Williams <[email protected]>
Acked-by: Vlastimil Babka <[email protected]>
