[ofa-general] Re: [PATCH 2/2] osm: QoS - support for MPR in qos policy

[Sasha Khapyorsky]

On 12:14 Thu 06 Sep     , Yevgeny Kliteynik wrote:
> Hi Sasha,
> 
> This patch adds osm_qos_policy_get_qos_level_by_mpr() wrapper function that
> basically does the same thing as the osm_qos_policy_get_qos_level_by_pr(),
> only for MultiPathRecord instead of PathRecord..
> 
> -- Yevgeny
> 
> Signed-off-by: Yevgeny Kliteynik <kliteyn at dev.mellanox.co.il>
> ---
>  opensm/include/opensm/osm_qos_policy.h |    8 ++++++++
>  opensm/opensm/osm_qos_policy.c         |   32 ++++++++++++++++++++++++++++++++
>  2 files changed, 40 insertions(+), 0 deletions(-)
> 
> diff --git a/opensm/include/opensm/osm_qos_policy.h b/opensm/include/opensm/osm_qos_policy.h
> index 11598be..0c220ee 100644
> --- a/opensm/include/opensm/osm_qos_policy.h
> +++ b/opensm/include/opensm/osm_qos_policy.h
> @@ -51,6 +51,7 @@
>  #include <complib/cl_list.h>
>  #include <opensm/osm_port.h>
>  #include <opensm/osm_sa_path_record.h>
> +#include <opensm/osm_sa_multipath_record.h>
> 
>  #define YYSTYPE char *
>  #define OSM_QOS_POLICY_MAX_PORTS_ON_SWITCH  128
> @@ -179,6 +180,13 @@ osm_qos_level_t * osm_qos_policy_get_qos_level_by_pr(
>  	IN const osm_physp_t * p_dest_physp,
>  	IN ib_net64_t comp_mask);
> 
> +osm_qos_level_t * osm_qos_policy_get_qos_level_by_mpr(
> +	IN const osm_qos_policy_t * p_qos_policy,
> +	IN const ib_multipath_rec_t * p_mpr,
> +	IN const osm_physp_t * p_src_physp,
> +	IN const osm_physp_t * p_dest_physp,
> +	IN ib_net64_t comp_mask);
> +
>  /***************************************************/
> 
>  int osm_qos_parse_policy_file(IN osm_subn_t * const p_subn);
> diff --git a/opensm/opensm/osm_qos_policy.c b/opensm/opensm/osm_qos_policy.c
> index 74628a5..a778bcb 100644
> --- a/opensm/opensm/osm_qos_policy.c
> +++ b/opensm/opensm/osm_qos_policy.c
> @@ -957,3 +957,35 @@ osm_qos_level_t * osm_qos_policy_get_qos_level_by_pr(
>  /***************************************************
>   ***************************************************/
> 
> +osm_qos_level_t * osm_qos_policy_get_qos_level_by_mpr(
> +	IN const osm_qos_policy_t * p_qos_policy,
> +	IN const ib_multipath_rec_t * p_mpr,
> +	IN const osm_physp_t * p_src_physp,
> +	IN const osm_physp_t * p_dest_physp,
> +	IN ib_net64_t comp_mask)
> +{
> +	uint8_t params_comp_mask = 0;
> +
> +	if (!p_qos_policy)
> +		return NULL;
> +
> +	if (comp_mask & IB_MPR_COMPMASK_QOS_CLASS)
> +		params_comp_mask |= QOS_PARAMS_COMPMASK_QOS_CLASS;
> +		
> +	if (comp_mask & IB_MPR_COMPMASK_SERVICEID_MSB &&
> +	    comp_mask & IB_MPR_COMPMASK_SERVICEID_LSB)
> +		params_comp_mask |= QOS_PARAMS_COMPMASK_SERVICEID;
> +
> +	if (comp_mask & IB_MPR_COMPMASK_PKEY)
> +		params_comp_mask |= QOS_PARAMS_COMPMASK_PKEY;
> +
> +	return __qos_policy_get_qos_level_by_params(
> +		p_qos_policy, p_src_physp, p_dest_physp,
> +		cl_ntoh64(ib_multipath_rec_service_id(p_mpr)),
> +		ib_multipath_rec_qos_class(p_mpr),
> +		cl_ntoh16(p_mpr->pkey), params_comp_mask);
> +}
> +
> +/***************************************************
> + ***************************************************/
> +
> -- 
> 1.5.1.4

This patch does not apply. The reason is trailing newline in
osm_qos_policy.c file (introduced in the previous patch). I'm using
'git-am --whitespace=strip', so this new line was stripped and the next
patch (this one) does not apply. It is better to not put empty new lines
at the end.

Sasha