H5 Sasha,
Sasha Khapyorsky wrote:
On 18:17 Tue 01 Dec , Sasha Khapyorsky wrote:
On 16:45 Mon 30 Nov , Yevgeny Kliteynik wrote:
Fix compilation error of that was introduced with patch
that added configurable transaction retries - osm_vendor_mlx
didn't have a definition of OSM_DEFAULT_RETRY_COUNT.
BTW as far as I understand osm_vendor_mlx is used only with ibmgtsim.
Right? If so could this be switched to osm_vendor_ibumad?
There is already libumad2sim.so (in ibsim) which catches all user_mad
calls and redirects this to where you want. Adoption of such model to
ibmgtsim would have a many advantages: it doesn't require OpenSM
rebuild, it supports not only OpenSM and osmtest, but virtually any
application which uses user_mad, it is really simple.
Finally this will leave us with only one vendor layer -
osm_vendor_ibumad. (AFAIK windows OpenSM is moving this direction too).
That is the plan. It's not as trivial as it sounds.
Some things that exist in vendor_mlx do not exist
in vendor_ibumad (such as RMPP implementation),
but this is certainly the plan.
AFAIK, WinSM and ibmgtsim are the only two things
that are using non-umad vendor. Once they both will
switch to umad, the whole vendor layers as it exists
today in OpenSM won't be necessary.
--Yevgeny
Sasha
--
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
