On Sat, Apr 27, 2019 at 06:23:39AM +0800, Ming Lei wrote: > On Fri, Apr 26, 2019 at 04:49:28PM +0200, Christoph Hellwig wrote: > > On Fri, Apr 26, 2019 at 08:53:44AM +0800, Ming Lei wrote: > > > Introduces __sg_alloc_table_chained() and __sg_free_table_chained() with > > > one extra parameter to specify size of the pre-allocated SGL, then the > > > 'first_chunk' SGL can include any number of entries. > > > > Can you just update the existing sg_alloc_table_chained and > > sg_free_table_chained chained instead? We only have a handful users > > of them in total, and most of them should eventually use your new > > semantics anyway. > > Yeah, I'd like to do that too. > > Not do that in V3 just because both nvme and network change are > involved. > > Will do this way if our maintainers don't object. sunrpc is maintained by the network maintainers, and the rdma code that is calling this in particular has been herded by Chuck Lever. Add them to Cc, I bet they won't object to the trivial paramter change. In fact I wouldn't be surprised if they'd be interested in the smaller inline SGLs as well for next merge window.
