Re: [PATCH for-next 5/9] IB/core, rdmavt, hfi1: Send OPA cap_mask3 in trap

On Wed, May 31, 2017 at 10:58:21AM -0700, Vishwanathapura, Niranjana wrote:

On Wed, May 31, 2017 at 07:32:35AM +0300, Leon Romanovsky wrote:

quoted

On Tue, May 30, 2017 at 11:32:22AM -0700, Henry, Michael N wrote:

quoted

On 5/29/2017 9:14 AM, Leon Romanovsky wrote:

quoted

On Fri, May 26, 2017 at 05:35:37AM -0700, Dennis Dalessandro wrote:

quoted

From: Michael N. Henry <redacted>

Provide the ability for IB clients to modify the OPA specific
capability mask and include this mask in the subsequent trap data.

diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 785e927..6c4f6da 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h

@@ -576,7 +576,8 @@ struct ib_device_modify {
  enum ib_port_modify_flags {
  	IB_PORT_SHUTDOWN		= 1,
  	IB_PORT_INIT_TYPE		= (1<<2),
-	IB_PORT_RESET_QKEY_CNTR		= (1<<3)
+	IB_PORT_RESET_QKEY_CNTR		= (1<<3),
+	IB_PORT_OPA_MASK_CHG		= (1<<4)

How will this flag be set?

This flag can be set as the port_modify_mask when calling ib_modify_port(..)

My question is more general. I'm as a representative of these "IB
clients" would like to know when and why should I use this flag.

Hi,
We have an OPA VNIC use case (patch coming up) to set/clear OPA capmask as
below.

pm.set/clear_port_cap_mask = OPA_CAP_MASK3_IsEthOnFabricSupported;
ib_modify_port(hca, port, IB_PORT_OPA_MASK_CHG, &pm);

This patch support such usecase and we want to get comments early (if any)
for that, hence this patch.

The comment is pretty simple, please don't post patches which are not
backed up by actual usage.

Right now, we have IB_PORT_SHUTDOWN, IB_PORT_INIT_TYPE and
IB_PORT_RESET_QKEY_CNTR are not written in any place. We don't want to
see another one.

And semantic of usage are more than welcomed.

Thanks

Niranjana