From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-wm0-f72.google.com (mail-wm0-f72.google.com [74.125.82.72]) by kanga.kvack.org (Postfix) with ESMTP id 33B416B025E for ; Fri, 2 Dec 2016 07:32:18 -0500 (EST) Received: by mail-wm0-f72.google.com with SMTP id m203so2769611wma.2 for ; Fri, 02 Dec 2016 04:32:18 -0800 (PST) Received: from mail-wj0-x22c.google.com (mail-wj0-x22c.google.com. [2a00:1450:400c:c01::22c]) by mx.google.com with ESMTPS id o9si2778135wmo.54.2016.12.02.04.32.16 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Fri, 02 Dec 2016 04:32:17 -0800 (PST) Received: by mail-wj0-x22c.google.com with SMTP id xy5so230749567wjc.0 for ; Fri, 02 Dec 2016 04:32:16 -0800 (PST) From: Michal Nazarewicz Subject: Re: [PATCH] mm: alloc_contig: demote PFN busy message to debug level In-Reply-To: <1480676263.17003.55.camel@pengutronix.de> References: <20161202095742.32449-1-l.stach@pengutronix.de> <74234427-005f-609e-3f33-cdf9a739c1d2@suse.cz> <1480675271.17003.50.camel@pengutronix.de> <20161202104851.GH6830@dhcp22.suse.cz> <1480676263.17003.55.camel@pengutronix.de> Date: Fri, 02 Dec 2016 13:32:13 +0100 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Sender: owner-linux-mm@kvack.org List-ID: To: Lucas Stach , Michal Hocko Cc: Vlastimil Babka , linux-mm@kvack.org, "Robin H. Johnson" , kernel@pengutronix.de, Andrew Morton , patchwork-lst@pengutronix.de, Joonsoo Kim , Marek Szyprowski >>> Am Freitag, den 02.12.2016, 11:18 +0100 schrieb Vlastimil Babka: >>>> I don't think we should just hide the issue like this, as getting high= =20 >>>> volume reports from this is also very likely associated with high=20 >>>> overhead for the allocations. If it's the generic dma-cma context, lik= e=20 >>>> in [1] where it attempts CMA for order-0 allocations, we should first = do=20 >>>> something about that, before tweaking the logging. That was also my concern. Ideally we would have a counter which increments whenever isolation failure happens and some monitoring of that counter but this is kernel so that=E2=80=99s just a pipe dream. >> On Fri 02-12-16 11:41:11, Lucas Stach wrote: >>> Still this message is really disturbing as page isolation failures can >>> be caused by lots of other reasons like temporarily pinned pages. Just so we=E2=80=99re on the same page, lots of allocations is not a *reaso= n* of isolation failures. It only surfaces it. This is not to disagree about better having code that is smart about allocating DMA buffers. This is true regardless. > Am Freitag, den 02.12.2016, 11:48 +0100 schrieb Michal Hocko: >> Hmm, then I think that what Robin has proposed [1] should be a generally >> better solution because it both ratelimits and points to the user who is >> triggering this path.=20 On Fri, Dec 02 2016, Lucas Stach wrote: > Dumping a stacktrace at this point is only going to increase the noise > from this message, as it can be trigger under normal operating > conditions of CMA. If someone temporarily locked a previously movable > page with GUP or something alike, the stacktrace will point to the > victim rather than the offender, so I think the value of the stackstrace > is rather limited. I agree, which is why I suggested printing the stack only if CONFIG_CMA_DEBUG is enabled. --=20 Best regards =E3=83=9F=E3=83=8F=E3=82=A6 =E2=80=9C=F0=9D=93=B6=F0=9D=93=B2=F0=9D=93=B7= =F0=9D=93=AA86=E2=80=9D =E3=83=8A=E3=82=B6=E3=83=AC=E3=83=B4=E3=82=A4=E3=83= =84 =C2=ABIf at first you don=E2=80=99t succeed, give up skydiving=C2=BB -- 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/ . Don't email: email@kvack.org