From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ben Widawsky Subject: Re: [PATCH 4/9] drm/i915: Make clear/insert vfuncs args absolute Date: Wed, 12 Feb 2014 16:34:25 -0800 Message-ID: <20140213003425.GB7401@bwidawsk.net> References: <1392244132-6806-1-git-send-email-benjamin.widawsky@intel.com> <1392244132-6806-5-git-send-email-benjamin.widawsky@intel.com> <20140213001415.GJ14909@nuc-i3427.alporthouse.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Received: from mail.bwidawsk.net (bwidawsk.net [166.78.191.112]) by gabe.freedesktop.org (Postfix) with ESMTP id 172BDFAC73 for ; Wed, 12 Feb 2014 16:34:32 -0800 (PST) Content-Disposition: inline In-Reply-To: <20140213001415.GJ14909@nuc-i3427.alporthouse.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: intel-gfx-bounces@lists.freedesktop.org Errors-To: intel-gfx-bounces@lists.freedesktop.org To: Chris Wilson , Ben Widawsky , Intel GFX List-Id: intel-gfx@lists.freedesktop.org On Thu, Feb 13, 2014 at 12:14:15AM +0000, Chris Wilson wrote: > On Wed, Feb 12, 2014 at 02:28:47PM -0800, Ben Widawsky wrote: > > This patch converts insert_entries and clear_range, both functions which > > are specific to the VM. These functions tend to encapsulate the gen > > specific PTE writes. Passing absolute addresses to the insert_entries, > > and clear_range will help make the logic clearer within the functions as > > to what's going on. Currently, all callers simply do the appropriate > > page shift, which IMO, ends up looking weird with an upcoming change for > > the gen8 page table allocations. > > > > Up until now, the PPGTT was a funky 2 level page table. GEN8 changes > > this to look more like a 3 level page table, and to that extent we need > > a significant amount more memory simply for the page tables. To address > > this, the allocations will be split up in finer amounts. > > Why size_t? Having a type that changes size irrespective of the GPU > seems like asking for trouble. > -Chris > For GGTT, I agree with you. The only trouble I could think of is if you used 48b PPGTT addressing on i386. I suspect there is some reason this won't work, but I've not read the spec. In the case of a field named, "length" I don't really think we need the size_t type to help self-document. So that's not useful... I can live with it if you want the change, I do think size_t still makes sense. > -- > Chris Wilson, Intel Open Source Technology Centre -- Ben Widawsky, Intel Open Source Technology Center