ibnetdiscover: add '-f' flag to show full information (ports' speed and width).

add '-f' flag to show full information (ports' speed and witdh). mainly to work with ibsim (using links real speed and width).

Signed-off-by Doron Shoham <dorons-smomgflXvOZWk0Htik3J/[email protected]>

---
 infiniband-diags/src/ibnetdiscover.c |   10 +++++++++-
 1 files changed, 9 insertions(+), 1 deletions(-)

infiniband-diags/src/ibnetdiscover.c +1 -0

index f20058c..0a020a2 100644
@@ -77,6 +77,7 @@ static char *diff_cache_file = NULL;
 static unsigned diffcheck_flags = DIFF_FLAG_DEFAULT;
 
 static int report_max_hops = 0;
+static int full_info = 0;
 
 /**
  * Define our own conversion functions to maintain compatibility with the old

infiniband-diags/src/ibnetdiscover.c +4 -0

 	fprintf(f, "\t\t# lid %d lmc %d \"%s\" lid %d %s%s\n",
 		port->base_lid, port->lmc, rem_nodename,
 		port->remoteport->node->type == IB_NODE_SWITCH ?

@@ -926,6 +930,9 @@ static int process_opt(void *context, int ch, char *optarg)
 	case 's':
 		cfg->show_progress = 1;
 		break;
+	case 'f':
+		full_info = 1;
+		break;
 	case 'l':
 		list = LIST_CA_NODE | LIST_SWITCH_NODE | LIST_ROUTER_NODE;
 		break;

@@ -964,6 +971,7 @@ int main(int argc, char **argv)
 	ibnd_fabric_t *diff_fabric = NULL;
 
 	const struct ibdiag_opt opts[] = {
+		{"full", 'f', 0, NULL, "show full information (ports' speed and witdh)"},
 		{"show", 's', 0, NULL, "show more information"},
 		{"list", 'l', 0, NULL, "list of connected nodes"},
 		{"grouping", 'g', 0, NULL, "show grouping"},

Signature · 1 line

1.5.4

Quoted 2 lines · Doron Shoham

add '-f' flag to show full information (ports' speed and witdh). mainly to work with ibsim (using links real speed and width).

Seems almost fine. However see the comments below.

Quoted 16 lines

Signed-off-by Doron Shoham <dorons-smomgflXvOZWk0Htik3J/[email protected]>

---
 infiniband-diags/src/ibnetdiscover.c |   10 +++++++++-
 1 files changed, 9 insertions(+), 1 deletions(-)

infiniband-diags/src/ibnetdiscover.c +1 -0

index f20058c..0a020a2 100644
@@ -77,6 +77,7 @@ static char *diff_cache_file = NULL;
 static unsigned diffcheck_flags = DIFF_FLAG_DEFAULT;
 
 static int report_max_hops = 0;
+static int full_info = 0;
 

infiniband-diags/src/ibnetdiscover.c void out_switch_port(ibnd_port_t * port, int group, char *out_prefix) · line 362

Quoted 8 lines

infiniband-diags/src/ibnetdiscover.c +2 -0

 /**
  * Define our own conversion functions to maintain compatibility with the old

@@ -357,6 +358,8 @@ void out_switch_port(ibnd_port_t * port, int group, char *out_prefix)
 		ext_port_str ? ext_port_str : "");
 	if (port->remoteport->node->type != IB_NODE_SWITCH)
 		fprintf(f, "(%" PRIx64 ") ", port->remoteport->guid);
+	if (full_info)
+			fprintf(f, " s=%d w=%d", ispeed, iwidth);

I think that in order to not potentially break any ibnetdiscover output parsers it would be better to put such "f" output after a comment line. Would it work with ibsim in a same way?

infiniband-diags/src/ibnetdiscover.c void out_ca_port(ibnd_port_t * port, int group, char *out_prefix) · line 403

Quoted 11 lines

infiniband-diags/src/ibnetdiscover.c +2 -1

 	fprintf(f, "\t\t# \"%s\" lid %d %s%s",
 		rem_nodename,
 		port->remoteport->node->type == IB_NODE_SWITCH ?

@@ -396,7 +399,8 @@ void out_ca_port(ibnd_port_t * port, int group, char *out_prefix)
 	rem_nodename = remap_node_name(node_name_map,
 				       port->remoteport->node->guid,
 				       port->remoteport->node->nodedesc);
-
+	if (full_info)
+			fprintf(f, " s=%d w=%d", ispeed, iwidth);

Ditto.

Sasha

Quoted 16 lines

infiniband-diags/src/ibnetdiscover.c +3 -0

 	fprintf(f, "\t\t# lid %d lmc %d \"%s\" lid %d %s%s\n",
 		port->base_lid, port->lmc, rem_nodename,
 		port->remoteport->node->type == IB_NODE_SWITCH ?

@@ -926,6 +930,9 @@ static int process_opt(void *context, int ch, char *optarg)
 	case 's':
 		cfg->show_progress = 1;
 		break;
+	case 'f':
+		full_info = 1;
+		break;
 	case 'l':
 		list = LIST_CA_NODE | LIST_SWITCH_NODE | LIST_ROUTER_NODE;
 		break;

@@ -964,6 +971,7 @@ int main(int argc, char **argv)
 	ibnd_fabric_t *diff_fabric = NULL;
 

Quoted 8 lines

infiniband-diags/src/ibnetdiscover.c +1 -0

 	const struct ibdiag_opt opts[] = {
+		{"full", 'f', 0, NULL, "show full information (ports' speed and witdh)"},
 		{"show", 's', 0, NULL, "show more information"},
 		{"list", 'l', 0, NULL, "list of connected nodes"},
 		{"grouping", 'g', 0, NULL, "show grouping"},

Signature · 1 line

1.5.4

Quoted 22 lines · Sasha Khapyorsky

Quoted 2 lines · Doron Shoham

add '-f' flag to show full information (ports' speed and witdh). mainly to work with ibsim (using links real speed and width).

Seems almost fine. However see the comments below.

Quoted 16 lines

Signed-off-by Doron Shoham <dorons-smomgflXvOZWk0Htik3J/[email protected]>

---
 infiniband-diags/src/ibnetdiscover.c |   10 +++++++++-

 1 files changed, 9 insertions(+), 1 deletions(-)

infiniband-diags/src/ibnetdiscover.c +1 -0

index f20058c..0a020a2 100644
@@ -77,6 +77,7 @@ static char *diff_cache_file = NULL;
  static unsigned diffcheck_flags = DIFF_FLAG_DEFAULT;
 
  static int report_max_hops = 0;
+static int full_info = 0;
 

Quoted 12 lines

infiniband-diags/src/ibnetdiscover.c void out_switch_port(ibnd_port_t * port, int group, char *out_prefix) · line 362

Quoted 8 lines

infiniband-diags/src/ibnetdiscover.c +2 -0

  /**
   * Define our own conversion functions to maintain compatibility with the old

@@ -357,6 +358,8 @@ void out_switch_port(ibnd_port_t * port, int group, char *out_prefix)
              ext_port_str ? ext_port_str : "");
      if (port->remoteport->node->type != IB_NODE_SWITCH)
              fprintf(f, "(%" PRIx64 ") ", port->remoteport->guid);
+    if (full_info)
+                    fprintf(f, " s=%d w=%d", ispeed, iwidth);

I think that in order to not potentially break any ibnetdiscover output parsers it would be better to put such "f" output after a comment line. Would it work with ibsim in a same way?

ibsim should still work if the output is after the comment.
I still think that it can potentially break ibnetdiscover output parsers (that why I used '-f' flag).
Do you want it with '-f' flag and after the comment?
for example:
[1](2c903000020a9)      "S-0008f10500650272"[19]                #  s=2 w=2 lid 6 lmc 0 "Voltaire sLB-4018    Line 10  Chip 1 4700 #4700-B778" lid 3 4xDDR

Quoted 32 lines

infiniband-diags/src/ibnetdiscover.c void out_ca_port(ibnd_port_t * port, int group, char *out_prefix) · line 403

Quoted 11 lines

infiniband-diags/src/ibnetdiscover.c +0 -0

      fprintf(f, "\t\t# \"%s\" lid %d %s%s",
              rem_nodename,

             port->remoteport->node->type == IB_NODE_SWITCH ?

infiniband-diags/src/ibnetdiscover.c +2 -1

@@ -396,7 +399,8 @@ void out_ca_port(ibnd_port_t * port, int group, char *out_prefix)
      rem_nodename = remap_node_name(node_name_map,
                                     port->remoteport->node->guid,
                                     port->remoteport->node->nodedesc);
-
+    if (full_info)
+                    fprintf(f, " s=%d w=%d", ispeed, iwidth);

Ditto.

Sasha

Quoted 16 lines

infiniband-diags/src/ibnetdiscover.c +0 -0

      fprintf(f, "\t\t# lid %d lmc %d \"%s\" lid %d %s%s\n",
              port->base_lid, port->lmc, rem_nodename,

             port->remoteport->node->type == IB_NODE_SWITCH ?

infiniband-diags/src/ibnetdiscover.c +0 -0

@@ -926,6 +930,9 @@ static int process_opt(void *context, int ch, char *optarg)

     case 's':
             cfg->show_progress = 1;
             break;
+    case 'f':
+            full_info = 1;
+            break;
     case 'l':
             list = LIST_CA_NODE | LIST_SWITCH_NODE | LIST_ROUTER_NODE; break;

infiniband-diags/src/ibnetdiscover.c +0 -0

@@ -964,6 +971,7 @@ int main(int argc, char **argv)
      ibnd_fabric_t *diff_fabric = NULL;
 

Quoted 8 lines

Quoted 8 lines

infiniband-diags/src/ibnetdiscover.c +1 -1

      const struct ibdiag_opt opts[] = {
+            {"full", 'f', 0, NULL, "show full information (ports' speed and witdh)"},
              {"show", 's', 0, NULL, "show more information"},
              {"list", 'l', 0, NULL, "list of connected nodes"},
              {"grouping", 'g', 0, NULL, "show grouping"},
--

1.5.4

--
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

Quoted 36 lines · Doron Shoham

Quoted 22 lines · Sasha Khapyorsky

Quoted 2 lines · Doron Shoham

add '-f' flag to show full information (ports' speed and witdh). mainly to work with ibsim (using links real speed and width).

Seems almost fine. However see the comments below.

Quoted 16 lines

Signed-off-by Doron Shoham <dorons-smomgflXvOZWk0Htik3J/[email protected]>

---
 infiniband-diags/src/ibnetdiscover.c |   10 +++++++++-
 1 files changed, 9 insertions(+), 1 deletions(-)

infiniband-diags/src/ibnetdiscover.c +1 -0

index f20058c..0a020a2 100644
@@ -77,6 +77,7 @@ static char *diff_cache_file = NULL;
 static unsigned diffcheck_flags = DIFF_FLAG_DEFAULT;
 
 static int report_max_hops = 0;
+static int full_info = 0;
 

Quoted 12 lines

infiniband-diags/src/ibnetdiscover.c void out_switch_port(ibnd_port_t * port, int group, char *out_prefix) · line 362

Quoted 8 lines

infiniband-diags/src/ibnetdiscover.c +2 -0

 /**
  * Define our own conversion functions to maintain compatibility with the old

@@ -357,6 +358,8 @@ void out_switch_port(ibnd_port_t * port, int group, char *out_prefix)
             ext_port_str ? ext_port_str : "");
     if (port->remoteport->node->type != IB_NODE_SWITCH)
             fprintf(f, "(%" PRIx64 ") ", port->remoteport->guid);
+    if (full_info)
+                    fprintf(f, " s=%d w=%d", ispeed, iwidth);

I think that in order to not potentially break any ibnetdiscover output parsers it would be better to put such "f" output after a comment line. Would it work with ibsim in a same way?

Quoted 6 lines

ibsim should still work if the output is after the comment.
I still think that it can potentially break ibnetdiscover output parsers (that why I used '-f' flag). Do you want it with '-f' flag and after the comment?
for example:
[1](2c903000020a9) "S-0008f10500650272"[19] # s=2 w=2 lid 6 lmc 0 "Voltaire sLB-4018 Line 10 Chip 1 4700 #4700-B778" lid 3 4xDDR

Personally, I'd prefer it at the end of the line after 4xDDR and use speed/width rather than s=/w=:

[1](2c903000020a9) "S-0008f10500650272"[19] # lid
6 lmc 0 "Voltaire sLB-4018 Line 10 Chip 1 4700 #4700-B778" lid 3
4xDDR speed 2 width 2

I think that looks better and is more readable. Just my $0.02...

-- Hal

Quoted 33 lines

Quoted 33 lines

infiniband-diags/src/ibnetdiscover.c void out_ca_port(ibnd_port_t * port, int group, char *out_prefix) · line 403

Quoted 12 lines

infiniband-diags/src/ibnetdiscover.c +2 -1

     fprintf(f, "\t\t# \"%s\" lid %d %s%s",
             rem_nodename,
             port->remoteport->node->type == IB_NODE_SWITCH ?

@@ -396,7 +399,8 @@ void out_ca_port(ibnd_port_t * port, int group, char *out_prefix)
     rem_nodename = remap_node_name(node_name_map,
                                    port->remoteport->node->guid,
                                    port->remoteport->node->nodedesc);
-
+    if (full_info)
+                    fprintf(f, " s=%d w=%d", ispeed, iwidth);

Ditto.

Sasha

Quoted 16 lines

infiniband-diags/src/ibnetdiscover.c +3 -0

     fprintf(f, "\t\t# lid %d lmc %d \"%s\" lid %d %s%s\n",
             port->base_lid, port->lmc, rem_nodename,
             port->remoteport->node->type == IB_NODE_SWITCH ?

@@ -926,6 +930,9 @@ static int process_opt(void *context, int ch, char *optarg)
     case 's':
             cfg->show_progress = 1;
             break;
+    case 'f':
+            full_info = 1;
+            break;
     case 'l':
             list = LIST_CA_NODE | LIST_SWITCH_NODE | LIST_ROUTER_NODE;
             break;

@@ -964,6 +971,7 @@ int main(int argc, char **argv)
     ibnd_fabric_t *diff_fabric = NULL;
 

Quoted 9 lines

Quoted 8 lines

Quoted 8 lines

infiniband-diags/src/ibnetdiscover.c +1 -1

     const struct ibdiag_opt opts[] = {
+            {"full", 'f', 0, NULL, "show full information (ports' speed and witdh)"},
             {"show", 's', 0, NULL, "show more information"},
             {"list", 'l', 0, NULL, "list of connected nodes"},
             {"grouping", 'g', 0, NULL, "show grouping"},
--

1.5.4

Quoted 5 lines

--
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

Quoted 51 lines · Hal Rosenstock

Quoted 36 lines · Doron Shoham

Quoted 22 lines · Sasha Khapyorsky

Quoted 2 lines · Doron Shoham

add '-f' flag to show full information (ports' speed and witdh). mainly to work with ibsim (using links real speed and width).

Seems almost fine. However see the comments below.

Quoted 16 lines

Signed-off-by Doron Shoham <dorons-smomgflXvOZWk0Htik3J/[email protected]>

---
 infiniband-diags/src/ibnetdiscover.c |   10 +++++++++-

 1 files changed, 9 insertions(+), 1 deletions(-)

infiniband-diags/src/ibnetdiscover.c +1 -0

index f20058c..0a020a2 100644
@@ -77,6 +77,7 @@ static char *diff_cache_file = NULL;
  static unsigned diffcheck_flags = DIFF_FLAG_DEFAULT;
 
  static int report_max_hops = 0;
+static int full_info = 0;
 

Quoted 12 lines

infiniband-diags/src/ibnetdiscover.c void out_switch_port(ibnd_port_t * port, int group, char *out_prefix) · line 362

Quoted 8 lines

infiniband-diags/src/ibnetdiscover.c +2 -0

  /**
   * Define our own conversion functions to maintain compatibility with the old

@@ -357,6 +358,8 @@ void out_switch_port(ibnd_port_t * port, int group, char *out_prefix)
              ext_port_str ? ext_port_str : "");
      if (port->remoteport->node->type != IB_NODE_SWITCH)
              fprintf(f, "(%" PRIx64 ") ", port->remoteport->guid);
+    if (full_info)
+                    fprintf(f, " s=%d w=%d", ispeed, iwidth);

I think that in order to not potentially break any ibnetdiscover output parsers it would be better to put such "f" output after a comment line. Would it work with ibsim in a same way?

Quoted 6 lines

ibsim should still work if the output is after the comment.
I still think that it can potentially break ibnetdiscover output parsers (that why I used '-f' flag).
Do you want it with '-f' flag and after the comment?
for example:
[1](2c903000020a9)      "S-0008f10500650272"[19]                #  s=2 w=2 lid 6 lmc 0 "Voltaire sLB-4018    Line 10  Chip 1 4700 #4700-B778" lid 3 4xDDR

Personally, I'd prefer it at the end of the line after 4xDDR and use speed/width rather than s=/w=:

[1](2c903000020a9)      "S-0008f10500650272"[19]                #  lid 6 lmc 0 "Voltaire sLB-4018    Line 10  Chip 1 4700 #4700-B778" lid 3 4xDDR speed 2 width 2

Quoted 3 lines

I think that looks better and is more readable. Just my $0.02...

-- Hal

The problem is the ibsim is expecting s=/w= and not speed <number> or width <number>.

What about the flag? do we still need it if we pass the output after the comment?

Doron

Quoted 47 lines

Quoted 47 lines

Quoted 33 lines

infiniband-diags/src/ibnetdiscover.c void out_ca_port(ibnd_port_t * port, int group, char *out_prefix) · line 403

Quoted 12 lines

infiniband-diags/src/ibnetdiscover.c +0 -0

      fprintf(f, "\t\t# \"%s\" lid %d %s%s",
              rem_nodename,

             port->remoteport->node->type == IB_NODE_SWITCH ?

infiniband-diags/src/ibnetdiscover.c +2 -1

@@ -396,7 +399,8 @@ void out_ca_port(ibnd_port_t * port, int group, char *out_prefix)
      rem_nodename = remap_node_name(node_name_map,
                                     port->remoteport->node->guid,
                                     port->remoteport->node->nodedesc);
-
+    if (full_info)
+                    fprintf(f, " s=%d w=%d", ispeed, iwidth);

Ditto.

Sasha

Quoted 16 lines

infiniband-diags/src/ibnetdiscover.c +0 -0

      fprintf(f, "\t\t# lid %d lmc %d \"%s\" lid %d %s%s\n",
              port->base_lid, port->lmc, rem_nodename,

             port->remoteport->node->type == IB_NODE_SWITCH ?

infiniband-diags/src/ibnetdiscover.c +0 -0

@@ -926,6 +930,9 @@ static int process_opt(void *context, int ch, char *optarg)

     case 's':
             cfg->show_progress = 1;
             break;
+    case 'f':
+            full_info = 1;
+            break;
     case 'l':
             list = LIST_CA_NODE | LIST_SWITCH_NODE | LIST_ROUTER_NODE; break;

infiniband-diags/src/ibnetdiscover.c +0 -0

@@ -964,6 +971,7 @@ int main(int argc, char **argv)
      ibnd_fabric_t *diff_fabric = NULL;
 

Quoted 8 lines

Quoted 8 lines

infiniband-diags/src/ibnetdiscover.c +1 -1

      const struct ibdiag_opt opts[] = {
+            {"full", 'f', 0, NULL, "show full information (ports' speed and witdh)"},
              {"show", 's', 0, NULL, "show more information"},
              {"list", 'l', 0, NULL, "list of connected nodes"},
              {"grouping", 'g', 0, NULL, "show grouping"},
--

1.5.4

--
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

Quoted 2 lines

What about the flag? do we still need it if we pass the output after the comment?

I wouldn't think so. I also think we've made commentary changes to the ibnetdiscover output format like this before. If we wanted to be absolutely sure it wouldn't break anything, we'd keep the flag though. It's up to Sasha.

add '-f' flag to show full information (ports' speed and witdh). mainly to work with ibsim (using links real speed and width).

Signed-off-by Doron Shoham <dorons-smomgflXvOZWk0Htik3J/[email protected]>

---
 infiniband-diags/src/ibnetdiscover.c |   16 ++++++++++++++--

 1 files changed, 14 insertions(+), 2 deletions(-)

infiniband-diags/src/ibnetdiscover.c +1 -0

index f20058c..df9e5a6 100644
@@ -77,6 +77,7 @@ static char *diff_cache_file = NULL;
  static unsigned diffcheck_flags = DIFF_FLAG_DEFAULT;
 
  static int report_max_hops = 0;
+static int full_info;
 
  /**
  * Define our own conversion functions to maintain compatibility with the old

infiniband-diags/src/ibnetdiscover.c +0 -0

                rem_nodename,

               port->remoteport->node->type == IB_NODE_SWITCH ? port->remoteport->node->smalid :
               port->remoteport->base_lid, dump_linkwidth_compat(iwidth), dump_linkspeed_compat(ispeed));

+       if (full_info)
+               fprintf(f, "s=%d w=%d", ispeed, iwidth); +

infiniband-diags/src/ibnetdiscover.c +0 -0

                port->base_lid, port->lmc, rem_nodename,

               port->remoteport->node->type == IB_NODE_SWITCH ? port->remoteport->node->smalid :
               port->remoteport->base_lid, dump_linkwidth_compat(iwidth), dump_linkspeed_compat(ispeed));

+       if (full_info)
+               fprintf(f, "s=%d w=%d", ispeed, iwidth);

+       fprintf(f, "\n");
+
       free(rem_nodename);
 }

infiniband-diags/src/ibnetdiscover.c +0 -0

@@ -926,6 +934,9 @@ static int process_opt(void *context, int ch, char *optarg)

       case 's':
               cfg->show_progress = 1;
               break;
+       case 'f':
+               full_info = 1;
+               break;
       case 'l':
               list = LIST_CA_NODE | LIST_SWITCH_NODE | LIST_ROUTER_NODE; break;

infiniband-diags/src/ibnetdiscover.c +1 -1

@@ -964,6 +975,7 @@ int main(int argc, char **argv)
        ibnd_fabric_t *diff_fabric = NULL;
 
        const struct ibdiag_opt opts[] = {
+               {"full", 'f', 0, NULL, "show full information (ports' speed and witdh)"},
                {"show", 's', 0, NULL, "show more information"},
                {"list", 'l', 0, NULL, "list of connected nodes"},
                {"grouping", 'g', 0, NULL, "show grouping"},
--

1.5.4

add '-f' flag to show full information (ports' speed and witdh). mainly to work with ibsim (using links real speed and width).

Signed-off-by Doron Shoham <dorons-smomgflXvOZWk0Htik3J/[email protected]>

Quoted 5 lines · Hal Rosenstock

Quoted 1 line

What about the flag? do we still need it if we pass the output after the comment?

I wouldn't think so. I also think we've made commentary changes to the ibnetdiscover output format like this before. If we wanted to be absolutely sure it wouldn't break anything, we'd keep the flag though. It's up to Sasha.

The '-f' flag could make sense since those information actually duplicates previous 4xDDR, etc. and used by ibsim's parser only.

Also it would be nice to make ibsim's parser to understand both
possibilities: s=2 and speed=2 (Or even to parse "*x*DR" strings :))

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

Quoted 7 lines · Sasha Khapyorsky

Quoted 5 lines · Hal Rosenstock

Quoted 1 line

What about the flag? do we still need it if we pass the output after the comment?

I wouldn't think so. I also think we've made commentary changes to the ibnetdiscover output format like this before. If we wanted to be absolutely sure it wouldn't break anything, we'd keep the flag though. It's up to Sasha.

Quoted 5 lines

The '-f' flag could make sense since those information actually duplicates previous 4xDDR, etc. and used by ibsim's parser only.

Also it would be nice to make ibsim's parser to understand both
possibilities: s=2 and speed=2 (Or even to parse "*x*DR" strings :))

Is it valid to assume that "*x*DR" string is always the last token?

If so, the patch should be changed

Eli

Quoted 6 lines

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

Quoted 7 lines · Sasha Khapyorsky

Quoted 5 lines · Hal Rosenstock

Quoted 1 line

What about the flag? do we still need it if we pass the output after the comment?

I wouldn't think so. I also think we've made commentary changes to the ibnetdiscover output format like this before. If we wanted to be absolutely sure it wouldn't break anything, we'd keep the flag though. It's up to Sasha.

Quoted 5 lines

The '-f' flag could make sense since those information actually duplicates previous 4xDDR, etc. and used by ibsim's parser only.

Also it would be nice to make ibsim's parser to understand both
possibilities: s=2 and speed=2 (Or even to parse "*x*DR" strings :))

Is it valid to assume that "*x*DR" string is always the last token?

If so, the patch should be changed

Eli

Quoted 6 lines

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