Subject: Re: drm/radeon spamming alloc_contig_range: [xxx, yyy) PFNs busy busy
To: Michal Nazarewicz <mina86@mina86.com>,
        Michal Hocko <mhocko@kernel.org>
References: <20161130092239.GD18437@dhcp22.suse.cz>
 <xa1ty4012k0f.fsf@mina86.com> <20161130132848.GG18432@dhcp22.suse.cz>
 <robbat2-20161130T195244-998539995Z@orbis-terrarum.net>
 <robbat2-20161130T195846-190979177Z@orbis-terrarum.net>
 <20161201071507.GC18272@dhcp22.suse.cz>
 <20161201072119.GD18272@dhcp22.suse.cz>
 <9f2aa4e4-d7d5-e24f-112e-a4b43f0a0ccc@suse.cz>
 <20161201141125.GB20966@dhcp22.suse.cz> <xa1t37i7ocuv.fsf@mina86.com>
 <20161201161117.GD20966@dhcp22.suse.cz> <xa1twpfjmkhc.fsf@mina86.com>
Cc: "Robin H. Johnson" <robbat2@gentoo.org>, linux-kernel@vger.kernel.org,
        linux-mm@kvack.org, dri-devel@lists.freedesktop.org,
        Joonsoo Kim <js1304@gmail.com>,
        Marek Szyprowski <m.szyprowski@samsung.com>
From: Vlastimil Babka <vbabka@suse.cz>
Message-ID: <445bd49a-e9ff-2db4-b5ab-700f6c72bcdc@suse.cz>
Date: Fri, 2 Dec 2016 07:36:50 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101
 Thunderbird/45.5.0
MIME-Version: 1.0
In-Reply-To: <xa1twpfjmkhc.fsf@mina86.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 804
Lines: 16

On 12/01/2016 10:02 PM, Michal Nazarewicz wrote:
> On Thu, Dec 01 2016, Michal Hocko wrote:
>> I am not familiar with this code so I cannot really argue but a quick
>> look at rmem_cma_setup doesn't suggest any speicific placing or
>> anything...
>
> early_cma parses ‘cma’ command line argument which can specify where
> exactly the default CMA area is to be located.  Furthermore, CMA areas
> can be assigned per-device (via the Device Tree IIRC).

OK, but the context of this bug report is a generic cma pool and generic 
dma alloc, which tries cma first and then fallback to 
alloc_pages_node(). If a device really requires specific placing as you 
suggest, then it probably uses a different allocation interface, 
otherwise there would be some flag to disallow the alloc_pages_node() 
fallback?