On 07/07/2015 02:25 AM, Nicholas A. Bellinger wrote: > On Tue, 2015-06-23 at 11:05 +0200, Hannes Reinecke wrote: >> On 06/23/2015 10:29 AM, Nicholas A. Bellinger wrote: >>> On Fri, 2015-06-19 at 09:13 +0200, Hannes Reinecke wrote: >>>> On 06/19/2015 08:48 AM, Christoph Hellwig wrote: >>>>> What's the benefit of the SAS transport class writeout? I honestly >>>>> always saw tcm_loop as a simple loopback driver, with the different >>>>> transport IDs in the PR code as a gimmick. Note that vhost and >>>>> xen-blkback copies that style and I did plan to consolidate it >>>>> in common code. >>>>> >>>> The benefit is that tcm_loop will show up in the system as a 'real' >>>> SAS hba; long-term goal is to simulate SAS multipathing here. >>>> I was even planning on adding simlated FC infrastructure, too; >>>> with that we could simulate FC multipathing, too, and our QA would >>>> be _so_ happy... >>>> >>> >>> Sounds like a reasonable use-case to support for loopback testing. >>> >>>> Again, these patches are mainly a collection of patches I've done to >>>> test various scenarios, in the hope others might find them useful, >>>> too. So I can easily hold off these patches until you've posted your >>>> rework. >>>> >>> >>> How different do you expect sas, fc, and iscsi transports to be..? >>> >>> Do you think this would this be better served by a simple tcm_loop LLD >>> specific API for different multipath transports..? >>> >> Actually, I would split off the various transport functions into >> separate files (tcm_loop_sas, tcm_loop_fc, etc), but keep a common >> tcm_loop module. >> We can even make transport classes optional by adding an explicit >> 'sas.XXX' prefix scanning when creating the device similar to what >> we do with the 'fc.XXX' prefix already. >> With that we would have a 'sas.XXX', 'fc.XXX', and 'iqn.XXX' WWN >> which would attach to the respective transport class, and any other >> WWN (which would be the default) would be getting the standard >> emulation without any transport class attached. > > I'm open to merging the tcm_loop patches #1-#6 as-is for the sas > transport pieces, or wait until you've done a large split based on > transport class types. > > It's really your call how the initial merge should look. > Probably leave out the transport class stuff for now; I kinda like the idea of having all types of transport classes available for tcm_loop. But this is actually not related to the rest of the patchset, so you can skip those for the time being. Cheers, Hannes -- Dr. Hannes Reinecke zSeries & Storage hare@xxxxxxx +49 911 74053 688 SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg GF: F. Imendörffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton HRB 21284 (AG Nürnberg) -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
