From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jan Kara Subject: Re: [PATCH] Describe race of direct read and fork for unaligned buffers Date: Wed, 2 May 2012 11:18:37 +0200 Message-ID: <20120502091837.GC16976@quack.suse.cz> References: <1335778207-6511-1-git-send-email-jack@suse.cz> <20120502081705.GB16976@quack.suse.cz> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable Return-path: Content-Disposition: inline In-Reply-To: Sender: owner-linux-mm@kvack.org To: Nick Piggin Cc: Jan Kara , Jeff Moyer , KOSAKI Motohiro , Michael Kerrisk , LKML , linux-man@vger.kernel.org, linux-mm@kvack.org, mgorman@suse.de, Andrea Arcangeli , Woodman List-Id: linux-man@vger.kernel.org On Wed 02-05-12 19:09:54, Nick Piggin wrote: > On 2 May 2012 18:17, Jan Kara wrote: > > On Wed 02-05-12 01:50:46, Nick Piggin wrote: >=20 > >> KOSAKI-san is correct, I think. > >> > >> The race is something like this: > >> > >> DIO-read > >> =A0 =A0 page =3D get_user_pages() > >> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 = =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 fork() > >> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 = =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 COW(page) > >> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 = =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0touch(page) > >> =A0 =A0 DMA(page) > >> =A0 =A0 page_cache_release(page); > >> > >> So whether parent or child touches the page, determines who gets the > >> actual DMA target, and who gets the copy. > > =A0OK, this is roughly what I understood from original threads as wel= l. So > > if our buffer is page aligned and its size is page aligned, you would= hit > > the corruption only if you do modify the buffer while IO to / from th= at buffer > > is in progress. And that would seem like a really bad programming pra= ctice > > anyway. So I still believe that having everything page size aligned w= ill > > effectively remove the problem although I agree it does not aim at th= e core > > of it. >=20 > I see what you mean. >=20 > I'm not sure, though. For most apps it's bad practice I think. If you g= et into > realm of sophisticated, performance critical IO/storage managers, it wo= uld > not surprise me if such concurrent buffer modifications could be allowe= d. > We allow exactly such a thing in our pagecache layer. Although probably > those would be using shared mmaps for their buffer cache. >=20 > I think it is safest to make a default policy of asking for IOs against= private > cow-able mappings to be quiesced before fork, so there are no surprises > or reliance on COW details in the mm. Do you think? Yes, I agree that (and MADV_DONTFORK) is probably the best thing to hav= e in documentation. Otherwise it's a bit too hairy... Honza --=20 Jan Kara SUSE Labs, CR -- 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/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter= .ca/ Don't email: email@kvack.org