On 16/06/2016 16:10, Colin Lord wrote: > On 06/16/2016 10:05 AM, Paolo Bonzini wrote: >> >> >> On 16/06/2016 16:00, Colin Lord wrote: >>>>>> The only block drivers that can be converted into modules are the drivers >>>>>> that don't perform any init operation except for registering themselves. >>>>>> This >>>>>> is why libiscsi has been disabled as a module. >>>> >>>> I don't think it has in this patch :) but you can also move the >>>> iscsi_opts registration to vl.c. >>> >>> Yeah I think Stefan mentioned this point in one of the earlier threads >>> but he said to do it in a separate commit, which I took to mean I >>> shouldn't include it here. Should I add it as a third part to this patch >>> series or leave it for a completely separate commit? >> >> The patches in the series are left separate when including them in QEMU. >> Therefore, a separate patch *is* (or will become :)) a separate commit. > > Yep, mostly just wasn't sure whether it was considered to be related > enough to include with the other two.
I think it is, since you had to refer to (the lack of) it in a commit message. In general, the beauty of patch series is that it's relatively cheap to add a patch. One should not overdo it, but a long series of dependencies benefits neither the author nor the reviewer. Paolo >> Therefore, putting the change before this patch, or alternatively as the >> first patch in the series, will be fine. >> >> Thanks, >> >> Paolo >> > > Sounds good, I'll work on putting that in the next version then. > > Colin >
