From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Hellstrom Subject: Re: [PATCH RFC] drm: Make the drm_vma_manager authentication mechanism a bit more versatile Date: Wed, 19 Mar 2014 13:39:57 +0100 Message-ID: <5329901D.6000508@shipmail.org> References: <1395228825-28908-1-git-send-email-thellstrom@vmware.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Received: from mx1.bahnhof.se (mx1.bahnhof.se [213.80.101.11]) by gabe.freedesktop.org (Postfix) with ESMTP id EBF458F0BE for ; Wed, 19 Mar 2014 05:51:21 -0700 (PDT) In-Reply-To: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dri-devel-bounces@lists.freedesktop.org Sender: "dri-devel" To: David Herrmann Cc: Thomas Hellstrom , "dri-devel@lists.freedesktop.org" List-Id: dri-devel@lists.freedesktop.org On 03/19/2014 12:45 PM, David Herrmann wrote: > Hi > > On Wed, Mar 19, 2014 at 12:33 PM, Thomas Hellstrom > wrote: >> By handing the authentication mechanism tokens in the form of const void >> pointers instead of struct file pointers, drivers are free to provide whatever >> file pointers they find appropriate. The pointers are never dereferenced in >> the drm_vma_manager so that shouldn't be a problem. >> >> As an example, the vmwgfx driver would like to provide struct ttm_object_file >> pointers. > I guess a "struct file*" pointer in tmm_object_file would be easier, > but if that reverse dependency is not wanted, I'm fine with that. Just > make sure you don't use drm_vma_node_verify_access for TTM directly as > TTM passes in a file pointer. > > Reviewed-by: David Herrmann > > Thanks > David Thanks for reviewing, David. Although it turned out that I won't be needing this, ATM since the TTM object mechanism already has a per file object-to-handle mapping that I could reuse for this. Thanks, Thomas