On Wed 17-08-22 08:00:45, Christoph Hellwig wrote:
On Mon, Aug 15, 2022 at 10:42:01AM +0200, Michal Hocko wrote:
Anyway, you seem to be not thrilled about the __GFP_NOWARN approach and I won't push it. But is the existing inconsistency really desirable? I mean we can get pretty vocal warning if the allocation fails but no information when the zone doesn't have any managed memory. Why should we treat them differently?
How could we end up having ZONE_DMA without any managed memory to start with except for the case where the total memory is smaller than what fits into ZONE_DMA? If we have such a case we really should warn about it as well.
This can be an early memory reservation from this physical address range. My original report http://lkml.kernel.org/r/Yj28gjonUa9+0yae@dhcp22.suse.cz was referring to such a system (a different one than what I am dealing with now): present:636kB managed:0kB
There is only 636kB present in that ZONE_DMA physical range but nothing has made it to the page allocator in the end.