[ofa-general] ibv_post_send fails when using malloc in a special way

[Asmund Ostvold]

Roland Dreier wrote:
> The most obvious explanation is that the physical pages underlying your
> allocation are different after the free/re-valloc.  This could happen
> without a system call I guess if a page is faulted in.

This is in the general sense true; that is way we pin down the memory by 
calling ibv_reg_mr().  Having done this we expect the virtual to 
physical relationship to remain constant.  However, we are concerned 
that ibv_reg_mr() does not call madvise() as appropriate, as mentioned 
in our posting.

Thanks, Åsmund