[ofa-general] missed cq event

[Or Gerlitz]

Pete Wyckoff wrote:
> You're certainly right that the need to reap even unsignaled sends
> is needed for some apps.  I'm just looking for a shortcut if my app
> does not ever need to find out about send completions.  Just to
> simplify the code, not expecting anything will go faster or require
> less memory.
One way we could go is enhance the verbs API to let an app tell the 
driver to post a NOP work request into the send queue once every n 
posts,  it has some devil in the details but I don't think it impossible 
or too ugly...

Anyway, looking in the IB spec (eg section 11.4.1.1 POST SEND REQUEST 
table Table 93 Operation Type Matrix) I don't see an evidence that such 
opcode exists... the mlx4 driver defines MLX4_NOP_OPCODE as a possible 
value to a send WR, so I assume this specific HW supports it. Does 
anyone know if NOP is supported by IB and/or iWARP?

Or.