Hi Hal,
On 16:51 Tue 21 Jul , Hal Rosenstock wrote:
> diff --git a/opensm/opensm/osm_ucast_lash.c b/opensm/opensm/osm_ucast_lash.c
> index 12b5e34..b785551 100644
> --- a/opensm/opensm/osm_ucast_lash.c
> +++ b/opensm/opensm/osm_ucast_lash.c
> @@ -486,6 +486,7 @@ static void balance_virtual_lanes(lash_t * p_lash,
> unsigned lanes_needed)
> int next_switch2, output_link2;
> int stop = 0, cycle_found;
> int cycle_found2;
> + unsigned start_vl = p_lash->p_osm->subn.opt.lash_start_vl;
>
> max_filled_lane = 0;
> min_filled_lane = lanes_needed - 1;
> @@ -572,8 +573,8 @@ static void balance_virtual_lanes(lash_t * p_lash,
> unsigned lanes_needed)
> virtual_location[dest][src][max_filled_lane] = 0;
> virtual_location[src][dest][min_filled_lane] = 1;
> virtual_location[dest][src][min_filled_lane] = 1;
> - p_lash->switches[src]->routing_table[dest].lane =
> min_filled_lane;
> - p_lash->switches[dest]->routing_table[src].lane =
> min_filled_lane;
> + p_lash->switches[src]->routing_table[dest].lane =
> min_filled_lane + start_vl;
> + p_lash->switches[dest]->routing_table[src].lane =
> min_filled_lane + start_vl;
> }
>
> if (trials == 0)
> @@ -804,6 +805,7 @@ static int lash_core(lash_t * p_lash)
> int cycle_found2 = 0;
> int status = 0;
> int *switch_bitmap = NULL; /* Bitmap to check if we have processed
> this pair */
> + unsigned start_vl = p_lash->p_osm->subn.opt.lash_start_vl;
>
> OSM_LOG_ENTER(p_log);
>
> @@ -838,7 +840,9 @@ static int lash_core(lash_t * p_lash)
> }
>
> for (i = 0; i < num_switches; i++) {
> - for (dest_switch = 0; dest_switch < num_switches; dest_switch++)
> + for (dest_switch = 0; dest_switch < num_switches;
> dest_switch++) {
> + if (lanes_needed > p_lash->vl_min)
> + goto Error_Not_Enough_Lanes;
Hmm, such check is already performed in place where lanes_needed is
increased (below). Why do we need another one?
> if (dest_switch != i && switch_bitmap[i * num_switches
> + dest_switch] == 0) {
> v_lane = 0;
> stop = 0;
> @@ -902,8 +906,8 @@ static int lash_core(lash_t * p_lash)
> }
> }
>
> - switches[i]->routing_table[dest_switch].lane =
> v_lane;
> - switches[dest_switch]->routing_table[i].lane =
> v_lane;
> + switches[i]->routing_table[dest_switch].lane =
> v_lane + start_vl;
> + switches[dest_switch]->routing_table[i].lane =
> v_lane + start_vl;
>
> if (cycle_found == 1 || cycle_found2 == 1) {
> if (++lanes_needed > p_lash->vl_min)
This one.
> @@ -926,16 +930,20 @@ static int lash_core(lash_t * p_lash)
> switch_bitmap[i * num_switches + dest_switch] =
> 1;
> switch_bitmap[dest_switch * num_switches + i] =
> 1;
> }
> + }
> }
>
> - OSM_LOG(p_log, OSM_LOG_INFO,
> - "Lanes needed: %d, Balancing\n", lanes_needed);
> + if (lanes_needed > p_lash->vl_min)
> + goto Error_Not_Enough_Lanes;
The same question, why do we need yet another lanes_needed > vl_min
check?
>
> for (i = 0; i < lanes_needed; i++) {
> OSM_LOG(p_log, OSM_LOG_INFO, "Lanes in layer %d: %d\n",
> i, p_lash->num_mst_in_lane[i]);
> }
>
> + OSM_LOG(p_log, OSM_LOG_INFO,
> + "Lanes needed: %d, Balancing\n", lanes_needed);
> +
> balance_virtual_lanes(p_lash, lanes_needed);
>
> for (i = 0; i < lanes_needed; i++) {
> @@ -948,8 +956,9 @@ static int lash_core(lash_t * p_lash)
> Error_Not_Enough_Lanes:
> status = -1;
> OSM_LOG(p_log, OSM_LOG_ERROR, "ERR 4D02: "
> - "Lane requirements (%d) exceed available lanes (%d)\n",
> - lanes_needed, p_lash->vl_min);
> + "Lane requirements (%d) exceed available lanes (%d)"
> + " with starting lane (%d)\n",
> + lanes_needed, p_lash->vl_min, start_vl);
> Exit:
> if (switch_bitmap)
> free(switch_bitmap);
> @@ -1177,10 +1186,11 @@ static int discover_network_properties(lash_t *
> p_lash)
> if (vl_min > 15)
> vl_min = 15;
>
> - p_lash->vl_min = vl_min;
> + p_lash->vl_min = vl_min - p_lash->p_osm->subn.opt.lash_start_vl;
vl_min =< lash_start_vl could be a legal case and lash->vl_min has
unsigned type. You will get an invalid value in this case. I suppose
that check should be here with LASH termination if no VLs are available.
Sasha
_______________________________________________
general mailing list
[email protected]
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
