[openib-general] Re: Re: [PATCH (updated)] ipoib: dont lock tx on completion

[Michael S. Tsirkin]

Quoting r. Sean Hefty <mshefty at ichips.intel.com>:
> Subject: Re: Re: [PATCH (updated)] ipoib: dont lock tx on completion
> 
> Michael S. Tsirkin wrote:
> 
> >>I don't see a specific race -- I'm just not totally comfortable yet,
> >>even with that fixed.  The issue is not necessarily reads being
> >>delayed, since both an out-of-order CPU and a compiler may move reads
> >>much _earlier_ than they appear in the code (for example a read may be
> >>speculatively executed because of branch prediction for a test in an
> >>if statement).
> >>
> >>- R.
> >>
> >Can they be moved to before the wmb? I think not.
> 
> Given that there's some concern here regarding races, what sort of 
> performance improvement do we see with this change?
> 
> - Sean
> 

Nothing major, I expect, especially given that only the send flow is
affected, and that is already apparently using less CPU than receive.

But I am just thinking, with tx_head/tx_taiil we have a circular
buffer here, how come we can not use it without races and locks?
There must be a way ...

-- 
MST - Michael S. Tsirkin