Re: [PATCH 22/37] IB/rdmavt: Add queue pair data structure to rdmavt

[Dennis Dalessandro <dennis.dalessandro-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>]

On Thu, Dec 10, 2015 at 01:12:51PM -0600, Hefty, Sean wrote:
>> >> +struct rvt_rwqe {
>> >> +	u64 wr_id;
>> >> +	u8 num_sge;
>> >> +	struct ib_sge sg_list[0];
>> >> +};
>> >> +
>> >> +/*
>> >> + * This structure is used to contain the head pointer, tail pointer,
>> >> + * and receive work queue entries as a single memory allocation so
>> >> + * it can be mmap'ed into user space.
>> >> + * Note that the wq array elements are variable size so you can't
>> >> + * just index into the array to get the N'th element;
>> >> + * use get_rwqe_ptr() instead.
>> >
>> >Can you add/use an entry_size field?
>> 
>> I think we could work something like that, however what we have in
>> qib/hfi1
>> also works.  Any reason we really should change this?
>
>I did not check to see what the drivers do.  Using entry_size is 
>straightforward, may provide the best performance, and can be done in 
>common code, versus adding callbacks to all users. 

Are you concerned that we have to do something different in each driver? If 
so, I do plan to move get_rwqe_ptr to rdmavt in a later patch.  It won't be 
part of the drivers any more.

I kind of like what we have for get_rwqe_ptr, it illustrates how we are 
fishing out the pointer. I'll take another look at it though for the coming 
patch, it could save us from doing a bit of math on the fly.

-Denny
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html