[ofa-general] RE: [PATCH] opensm: Bug in coding of VL Arbitration tables

[Eitan Zahavi]

Thanks Hal. Good catch. Should have seen this.
Sorry
 
Eitan


________________________________

	From: Hal Rosenstock [mailto:hal.rosenstock at gmail.com] 
	Sent: Wednesday, July 18, 2007 7:36 PM
	To: Eitan Zahavi
	Cc: OPENIB; sashak at voltaire.com; Yevgeny Kliteynik
	Subject: Re: [PATCH] opensm: Bug in coding of VL Arbitration
tables
	
	
	Hi Eitan,
	
	
	On 7/18/07, Eitan Zahavi <eitan at mellanox.co.il> wrote: 

		Hi Sasha
		
		Discovered a bug in coding of the VL Arbitration table
"index".
		According to spec should be:
		1 for low part of low table
		2 for high part of low table
		3 for low part of high table
		4 for high part of high table
		
		the patch below fixes it:
		
		Eitan
		
		Signed-off-by:  Eitan Zahavi <eitan at mellanox.co.il>
		
		diff --git a/opensm/opensm/osm_qos.c
b/opensm/opensm/osm_qos.c 
		index bbb1608..413e200 100644
		--- a/opensm/opensm/osm_qos.c
		+++ b/opensm/opensm/osm_qos.c
		@@ -116,14 +116,14 @@ static ib_api_status_t
vlarb_update(osm_req_t * p_req,
		                    p_pi->vl_arb_low_cap :
IB_NUM_VL_ARB_ELEMENTS_IN_BLOCK;
		                if ((status =
vlarb_update_table_block(p_req, p, port_num,
	
&qcfg->vlarb_low[0],
		-
len, 0)) != IB_SUCCESS)
		+
len, 1)) != IB_SUCCESS)
		                        return status;
		        }
		        if (p_pi->vl_arb_low_cap >
IB_NUM_VL_ARB_ELEMENTS_IN_BLOCK) {
		                len = p_pi->vl_arb_low_cap %
IB_NUM_VL_ARB_ELEMENTS_IN_BLOCK;
		                if ((status =
vlarb_update_table_block(p_req, p, port_num,
	
&qcfg->vlarb_low[1],
		-
len, 1)) != IB_SUCCESS)
		+
len, 2)) != IB_SUCCESS)
		                        return status;
		        }
		        if (p_pi->vl_arb_high_cap > 0) {
		@@ -131,14 +131,14 @@ static ib_api_status_t
vlarb_update(osm_req_t * p_req,
		                    p_pi->vl_arb_high_cap :
IB_NUM_VL_ARB_ELEMENTS_IN_BLOCK;
		                if ((status =
vlarb_update_table_block(p_req, p, port_num,
	
&qcfg->vlarb_high[0],
		-
len, 2)) != IB_SUCCESS)
		+
len, 3)) != IB_SUCCESS)
		                        return status;
		        }
		        if (p_pi->vl_arb_high_cap >
IB_NUM_VL_ARB_ELEMENTS_IN_BLOCK) {
		                len = p_pi->vl_arb_high_cap %
IB_NUM_VL_ARB_ELEMENTS_IN_BLOCK;
		                if ((status =
vlarb_update_table_block(p_req, p, port_num,
	
&qcfg->vlarb_high[1],
		-
len, 3)) != IB_SUCCESS)
		+
len, 4)) != IB_SUCCESS)
		                        return status;
		        }



		Are you sure ? It looks to me like this is already
handled in vlarb_update_table_block as follows: 
		


	        if (!memcmp(&p->vl_arb[block_num], &block,
	                     block_length * sizeof(block.vl_entry[0])))
	                return IB_SUCCESS;
	
	but
	
	        attr_mod = ((block_num + 1) << 16) | port_num;
	
	        return osm_req_set(p_req, osm_physp_get_dr_path_ptr(p),
	                           (uint8_t *) & block, sizeof(block),
	                           IB_MAD_ATTR_VL_ARBITRATION,
	                           cl_hton32(attr_mod),
CL_DISP_MSGID_NONE, &context);
	
	-- Hal 



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openfabrics.org/pipermail/general/attachments/20070718/fa82c1f8/attachment.html>