From mboxrd@z Thu Jan 1 00:00:00 1970 From: Joerg Roedel Subject: Re: [PATCH 1/3] mmu_notifier: Add mmu_notifier_invalidate_range() Date: Fri, 25 Jul 2014 23:57:18 +0200 Message-ID: <20140725215718.GO14017@8bytes.org> References: <1406212541-25975-1-git-send-email-joro@8bytes.org> <1406212541-25975-2-git-send-email-joro@8bytes.org> <20140725131639.698f18ff@jbarnes-desktop> <20140725213806.GN14017@8bytes.org> <20140725144213.773474e4@jbarnes-desktop> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <20140725144213.773474e4@jbarnes-desktop> Sender: owner-linux-mm@kvack.org To: Jesse Barnes Cc: Andrew Morton , Andrea Arcangeli , Peter Zijlstra , Rik van Riel , Hugh Dickins , Mel Gorman , Johannes Weiner , Jerome Glisse , jroedel@suse.de, Jay.Cornwall@amd.com, Oded.Gabbay@amd.com, John.Bridgman@amd.com, Suravee.Suthikulpanit@amd.com, ben.sander@amd.com, David Woodhouse , linux-kernel@vger.kernel.org, linux-mm@kvack.org, iommu@lists.linux-foundation.org List-Id: iommu@lists.linux-foundation.org On Fri, Jul 25, 2014 at 02:42:13PM -0700, Jesse Barnes wrote: > On Fri, 25 Jul 2014 23:38:06 +0200 > Joerg Roedel wrote: > > I though about removing the need for invalidate_range_end too when > > writing the patches, and possible solutions are > > > > 1) Add mmu_notifier_invalidate_range() to all places where > > start/end is called too. This might add some unnecessary > > overhead. > > > > 2) Call the invalidate_range() call-back from the > > mmu_notifier_invalidate_range_end too. > > > > 3) Just let the user register the same function for > > invalidate_range and invalidate_range_end > > > > I though that option 1) adds overhead that is not needed (but it might > > not be too bad, the overhead is an additional iteration over the > > mmu_notifer list when there are no call-backs registered). > > > > Option 2) might also be overhead if a user registers different functions > > for invalidate_range() and invalidate_range_end(). In the end I came to > > the conclusion that option 3) is the best one from an overhead POV. > > > > But probably targeting better usability with one of the other options is > > a better choice? I am open for thoughts and suggestions on that. > > Making the _end callback just do another TLB flush is fine too, but it > would be nice to have the consistency of (1). I can live with either > though, as long as the callbacks are well documented. You are right, having this consistency would be good. The more I think about it, the more it makes sense to go with option 2). Option 1) would mean that invalidate_range is explicitly called right before invalidate_range_end at some places. Doing this implicitly like in option 2) is cleaner and less error-prone. And the list of mmu_notifiers needs only be traversed once in invalidate_range_end(), so additional overhead is minimal. I'll update patch 3 for this, unless there are other opinions. Joerg -- 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