On Thu, 2011-01-20 at 12:18 +0200, Or Gerlitz wrote: > David Dillow wrote: > > The size could be 1, 2, 4, 8 MB or whatever. But if there isn't a HW > > limit, then why limit yourself at all? > > There are others limits in the echo system this patch live with, e.g some > buffer allocation and usage scheme with IB RC done by the target...
Keep in mind this is not directly seen by the target -- this is purely an optimization for the block stack. The target has to deal with > 1 MB requests from an initiator anyway, or it is buggy -- we can generate those today, just not reliably. And we have the proper tools to deal with that -- it's called max_hw_sectors_kb, as set by the max_sect option to SRP. > IB RC message is limited by spec to 2G and with mlx4 to 1GB, what not > remaining > in the (say) 8MB area with this patch as well? I would call that a HW/spec limit, and would be fine if mlx4 set it to 1 GB. Or even 512 MB or 256 MB. 8 MB would work, but to give room to grow I'd rather see it higher. -- Dave Dillow National Center for Computational Science Oak Ridge National Laboratory (865) 241-6602 office -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
