From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jason Gunthorpe Date: Wed, 24 Nov 2010 22:18:45 +0000 Subject: Re: [patch v4] infiniband: uverbs: handle large number of entries Message-Id: <20101124221845.GH2369@obsidianresearch.com> List-Id: References: <20101007071610.GC11681@bicker> <20101007161649.GD21206@obsidianresearch.com> <20101007165947.GD11681@bicker> <20101009231607.GA24649@obsidianresearch.com> <20101012113117.GB6742@bicker> <20101012210118.GR24268@obsidianresearch.com> <20101013091312.GB6060@bicker> <20101123071025.GI1522@bicker> In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Roland Dreier Cc: Dan Carpenter , Roland Dreier , Sean Hefty , Hal Rosenstock , linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, kernel-janitors-u79uwXL29TY76Z2rM5mHXA@public.gmane.org On Wed, Nov 24, 2010 at 02:07:40PM -0800, Roland Dreier wrote: > - Fixing the potential integer overflow by capping cmd.ne at some high > but safe value (1000 say) This breaks how userspace wants to use poll, ie if you return anything less than what I asked for then that means there is no more work to do. Apps make this assumption, and is a reasonable thing to do. Otherwise apps need a dummy call to ibv_poll to avoid races. So if you are worried about how many times ib_poll_cq is called then bound the kzalloc size and wrap the whole thing in a loop, but realistically I have to think the performance trade off of kzalloc/free vs calling ib_poll more often is not entirely obvious. Who uses this path anyhow? Jason