On 10/26/2011 10:37 AM, Alexander Nezhinsky wrote: >>>> With iser being built in, the tgtd package pulls in two more >>>> libraries, libibverbs and librdmacm, is the motivation here to >>>> eliminate that as of the overhead? >>> Yes. > I think that in general the idea of modularizaton is a good and a > very appropriate one. > > Still, i have a few reservations about the suggested implementation. > It singles out iser as the only module, which is "modularized-out". > This is ok for now, and this is surely what a multitude of users > would like to do, but i would opt for a change which is > framework-wide. Hi Alexander, The only objective I wanted to accomplish with this patch was fewer runtime library dependencies, which was best done by making iser a shared object. Beyond that, it's not clear to me that the added complexity of implementation and additional cmdline parameters justify the benefits. [driver and bs modularization proposal snipped] > * I am willing to co-operate actively in implementation of these > features, when all agree on their design. My only comment would be that since tgt is open source, the need for loadable modules is lessened. A new bs or driver can be added and tgt recompiled easily for a user's unique needs, and more widely useful new drivers should be included upstream. In fact, in looking at my original patch, I'm wondering now if *it* meets the above criteria. Should I have just made this change in my distro's packaging, instead of submitting it upstream? I'm not sure anymore :) BTW the tgt_driver state check patches you just posted look good to me. Regards -- Andy -- To unsubscribe from this list: send the line "unsubscribe stgt" in the body of a message to majordomo at vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html |