opensm: fixed port order configuration in torus routing engine

opensm: fixed port order configuration in torus routing engine

[Alex Netes]

Commit 1c2a298b295eba7e24205519abc24e47106d15df broke port order
configuration for torus routing engine. order was incorrectly initiated,
causing setting LFTs to fail.

Signed-off-by: Alex Netes <alexne-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
 opensm/osm_torus.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/opensm/osm_torus.c b/opensm/osm_torus.c
index cd3d490..75724d2 100644
--- a/opensm/osm_torus.c
+++ b/opensm/osm_torus.c
@@ -8484,7 +8484,7 @@ bool torus_lft(struct torus *t, struct t_switch *sw)
 	struct port_grp *pgrp;
 	struct t_switch *dsw;
 	osm_switch_t *osm_sw;
-	unsigned order[IB_NODE_NUM_PORTS_MAX+1];
+	unsigned char order[IB_NODE_NUM_PORTS_MAX+1];
 
 	if (!(sw->osm_switch && sw->osm_switch->priv == sw)) {
 		OSM_LOG(&t->osm->log, OSM_LOG_ERROR,
@@ -8506,7 +8506,7 @@ bool torus_lft(struct torus *t, struct t_switch *sw)
 
 		for (p = 0; p < ARRAY_SIZE(order); p++) {
 
-			unsigned px = order[t->port_order[p]];
+			unsigned char px = order[t->port_order[p]];
 
 			if (px == IB_INVALID_PORT_NUM)
 				continue;
-- 
1.7.5.2

-- 

-- Alex
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: opensm: fixed port order configuration in torus routing engine

[Jim Schutt]

Hi Alex,

Alex Netes wrote:
> Commit 1c2a298b295eba7e24205519abc24e47106d15df broke port order
> configuration for torus routing engine. order was incorrectly initiated,
> causing setting LFTs to fail.
> 
> Signed-off-by: Alex Netes <alexne-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
> ---
>  opensm/osm_torus.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/opensm/osm_torus.c b/opensm/osm_torus.c
> index cd3d490..75724d2 100644
> --- a/opensm/osm_torus.c
> +++ b/opensm/osm_torus.c
> @@ -8484,7 +8484,7 @@ bool torus_lft(struct torus *t, struct t_switch *sw)
>  	struct port_grp *pgrp;
>  	struct t_switch *dsw;
>  	osm_switch_t *osm_sw;
> -	unsigned order[IB_NODE_NUM_PORTS_MAX+1];
> +	unsigned char order[IB_NODE_NUM_PORTS_MAX+1];
>  
>  	if (!(sw->osm_switch && sw->osm_switch->priv == sw)) {
>  		OSM_LOG(&t->osm->log, OSM_LOG_ERROR,
> @@ -8506,7 +8506,7 @@ bool torus_lft(struct torus *t, struct t_switch *sw)
>  
>  		for (p = 0; p < ARRAY_SIZE(order); p++) {
>  
> -			unsigned px = order[t->port_order[p]];
> +			unsigned char px = order[t->port_order[p]];
>  
>  			if (px == IB_INVALID_PORT_NUM)
>  				continue;

I guess the memset(order, IB_INVALID_PORT_NUM, sizeof(order))
a few lines up does the wrong thing without your fix,
since we compare here with IB_INVALID_PORT_NUM?

Nice catch - I'm sorry I missed it.

FWIW, ib_types.h uses uint8_t for ports - maybe should do
that here as well rather than unsigned char?

Acked-by: Jim Schutt <jaschut-4OHPYypu0djtX7QSmKvirg@public.gmane.org>

-- Jim

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 1v2] opensm: fixed port order configuration in torus routing engine

[Alex Netes]

Commit 1c2a298b295eba7e24205519abc24e47106d15df broke port order
configuration for torus routing engine. order was incorrectly initiated,
causing setting LFTs to fail.

Signed-off-by: Alex Netes <alexne-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
---
Changes since v1:
1. Defined order[] as uint8_t
2. Removed error message while parsing port_order. It gives false alarms, 
   because strtok() returns NULL when no tokens left.

--
 opensm/osm_torus.c |   10 ++++------
 1 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/opensm/osm_torus.c b/opensm/osm_torus.c
index cd3d490..29c1bb4 100644
--- a/opensm/osm_torus.c
+++ b/opensm/osm_torus.c
@@ -868,11 +868,9 @@ bool parse_port_order(struct torus *t, const char *parse_sep)
 	unsigned i, j, k, n;
 
 	for (i = 0; i < ARRAY_SIZE(t->port_order); i++) {
-		if (!parse_port(&(t->port_order[i]), parse_sep)) {
-			OSM_LOG(&t->osm->log, OSM_LOG_ERROR,
-				"Error: cannot parse port_order");
+		if (!parse_port(&(t->port_order[i]), parse_sep))
 			break;
-		}
+
 		for (j = 0; j < i; j++) {
 			if (t->port_order[j] == t->port_order[i]) {
 				OSM_LOG(&t->osm->log, OSM_LOG_ERROR,
@@ -8484,7 +8482,7 @@ bool torus_lft(struct torus *t, struct t_switch *sw)
 	struct port_grp *pgrp;
 	struct t_switch *dsw;
 	osm_switch_t *osm_sw;
-	unsigned order[IB_NODE_NUM_PORTS_MAX+1];
+	uint8_t order[IB_NODE_NUM_PORTS_MAX+1];
 
 	if (!(sw->osm_switch && sw->osm_switch->priv == sw)) {
 		OSM_LOG(&t->osm->log, OSM_LOG_ERROR,
@@ -8506,7 +8504,7 @@ bool torus_lft(struct torus *t, struct t_switch *sw)
 
 		for (p = 0; p < ARRAY_SIZE(order); p++) {
 
-			unsigned px = order[t->port_order[p]];
+			uint8_t px = order[t->port_order[p]];
 
 			if (px == IB_INVALID_PORT_NUM)
 				continue;
-- 
1.7.5.2

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1v2] opensm: fixed port order configuration in torus routing engine

[Jim Schutt]

Alex Netes wrote:
> Commit 1c2a298b295eba7e24205519abc24e47106d15df broke port order
> configuration for torus routing engine. order was incorrectly initiated,
> causing setting LFTs to fail.
> 
> Signed-off-by: Alex Netes <alexne-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>

Acked-by: Jim Schutt <jaschut-4OHPYypu0djtX7QSmKvirg@public.gmane.org>

Also, while reviewing this I noticed a couple things in the
port-order patch that I should have noticed earlier, but
didn't.  I have a couple of minor fixup patches that go on
top of this one, to send along in a minute.

-- Jim

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 1/2] opensm: fail if configured torus port order references a port not available in all switches

[Jim Schutt]

Signed-off-by: Jim Schutt <jaschut-4OHPYypu0djtX7QSmKvirg@public.gmane.org>
---
 opensm/osm_torus.c |   24 ++++++++++++++++++++----
 1 files changed, 20 insertions(+), 4 deletions(-)

diff --git a/opensm/osm_torus.c b/opensm/osm_torus.c
index 29c1bb4..47654ce 100644
--- a/opensm/osm_torus.c
+++ b/opensm/osm_torus.c
@@ -856,9 +856,6 @@ bool parse_port(unsigned *pnum, const char *parse_sep)
 	if (!val)
 		return false;
 	*pnum = strtoul(val, &nextchar, 0);
-	if (*pnum > IB_NODE_NUM_PORTS_MAX) {
-		*pnum = 0;
-	}
 	return true;
 }
 
@@ -7018,7 +7015,8 @@ static
 bool verify_setup(struct torus *t, struct fabric *f)
 {
 	struct coord_dirs *o;
-	unsigned n = 0;
+	struct f_switch *sw;
+	unsigned p, s, n = 0;
 	bool success = false;
 	bool all_sw_present, need_seed = true;
 
@@ -7044,6 +7042,24 @@ bool verify_setup(struct torus *t, struct fabric *f)
 			"with two QoS levels (have %d need 8)\n",
 			(int)t->osm->subn.min_data_vls);
 	/*
+	 * Be sure all the switches in the torus support the port
+	 * ordering that might have been configured.
+	 */
+	for (s = 0; s < f->switch_cnt; s++) {
+		sw = f->sw[s];
+		for (p = 0; p < sw->port_cnt; p++) {
+			if (t->port_order[p] >= sw->port_cnt) {
+				OSM_LOG(&t->osm->log, OSM_LOG_ERROR,
+					"Error: port_order configured using "
+					"port %u, but only %u ports in "
+					"switch w/ GUID 0x%04"PRIx64"\n",
+					t->port_order[p],  sw->port_cnt - 1,
+					cl_ntoh64(sw->n_id));
+				goto out;
+			}
+		}
+	}
+	/*
 	 * Unfortunately, there is a problem with non-unique topology for any
 	 * torus dimension which has radix four.  This problem requires extra
 	 * input, in the form of specifying both the positive and negative
-- 
1.6.2.2


--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 1/2] opensm: fail if configured torus port order references a port not available in all switches

[Alex Netes]

On 16:33 Wed 01 Jun     , Jim Schutt wrote:
> 
> Signed-off-by: Jim Schutt <jaschut-4OHPYypu0djtX7QSmKvirg@public.gmane.org>
> ---

Applied, thanks.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 2/2] opensm: use OSM_LOG_INFO on duplicate torus port order parsing

[Jim Schutt]

Since it doesn't cause OpenSM to exit, it isn't an error.
We just want to give the admin information that will allow them
to clean up their configuration.

Signed-off-by: Jim Schutt <jaschut-4OHPYypu0djtX7QSmKvirg@public.gmane.org>
---
 opensm/osm_torus.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/opensm/osm_torus.c b/opensm/osm_torus.c
index 47654ce..d06cea9 100644
--- a/opensm/osm_torus.c
+++ b/opensm/osm_torus.c
@@ -870,8 +870,8 @@ bool parse_port_order(struct torus *t, const char *parse_sep)
 
 		for (j = 0; j < i; j++) {
 			if (t->port_order[j] == t->port_order[i]) {
-				OSM_LOG(&t->osm->log, OSM_LOG_ERROR,
-					"Error: ignoring duplicate port %u in "
+				OSM_LOG(&t->osm->log, OSM_LOG_INFO,
+					"Ignored duplicate port %u in"
 					" port_order parsing\n",
 					t->port_order[j]);
 				i--;	/* Ignore duplicate port number */
-- 
1.6.2.2


--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 2/2] opensm: use OSM_LOG_INFO on duplicate torus port order parsing

[Alex Netes]

On 16:33 Wed 01 Jun     , Jim Schutt wrote:
> Since it doesn't cause OpenSM to exit, it isn't an error.
> We just want to give the admin information that will allow them
> to clean up their configuration.
> 
> Signed-off-by: Jim Schutt <jaschut-4OHPYypu0djtX7QSmKvirg@public.gmane.org>
> ---

Applied, thanks.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html