| Message ID | 90ba6b41-2a1a-4419-bb49-4ed8c216e2a7@CMEXHTCAS2.ad.emulex.com (mailing list archive) |
|---|---|
| State | Rejected |
| Headers | show |
Hi Roland, Could you please ack/respond? Thanks Som > -----Original Message----- > From: linux-rdma-owner@vger.kernel.org [mailto:linux-rdma- > owner@vger.kernel.org] On Behalf Of Somnath Kotur > Sent: Tuesday, January 28, 2014 12:34 PM > To: roland@kernel.org > Cc: linux-rdma@vger.kernel.org; Somnath Kotur > Subject: [PATCH v2] IB/CMA: Fix initialization of hop_limit instead of > hardcode to 1. > > As suggested by Sean Hefty, take into account src/dst addresses on different > subnets and set hop_limit in cma_resolve_iboe_route accordingly instead of > blind hardcode to 1 or any other value. > > Signed-off-by: Somnath Kotur <somnath.kotur@emulex.com> > --- > Incorporated comments by Sean Hefty > > drivers/infiniband/core/addr.c | 5 +++++ > drivers/infiniband/core/cma.c | 2 +- > include/rdma/ib_addr.h | 1 + > 3 files changed, 7 insertions(+), 1 deletions(-) > > diff --git a/drivers/infiniband/core/addr.c b/drivers/infiniband/core/addr.c > index e90f2b2..6fa4721 100644 > --- a/drivers/infiniband/core/addr.c > +++ b/drivers/infiniband/core/addr.c > @@ -250,6 +250,8 @@ static int addr4_resolve(struct sockaddr_in *src_in, > goto put; > } > > + addr->uses_gateway = rt->rt_uses_gateway; > + > ret = dst_fetch_ha(&rt->dst, addr, &fl4.daddr); > put: > ip_rt_put(rt); > @@ -263,6 +265,7 @@ static int addr6_resolve(struct sockaddr_in6 *src_in, > struct rdma_dev_addr *addr) > { > struct flowi6 fl6; > + struct rt6_info *rt; > struct dst_entry *dst; > int ret; > > @@ -275,6 +278,7 @@ static int addr6_resolve(struct sockaddr_in6 *src_in, > if ((ret = dst->error)) > goto put; > > + rt = (struct rt6_info *)dst; > if (ipv6_addr_any(&fl6.saddr)) { > ret = ipv6_dev_get_saddr(&init_net, ip6_dst_idev(dst)- > >dev, > &fl6.daddr, 0, &fl6.saddr); > @@ -298,6 +302,7 @@ static int addr6_resolve(struct sockaddr_in6 *src_in, > goto put; > } > > + addr->uses_gateway = rt->rt6i_flags & RTF_GATEWAY; > ret = dst_fetch_ha(dst, addr, &fl6.daddr); > put: > dst_release(dst); > diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c > index 8e49db6..ae60104 100644 > --- a/drivers/infiniband/core/cma.c > +++ b/drivers/infiniband/core/cma.c > @@ -1902,7 +1902,7 @@ static int cma_resolve_iboe_route(struct > rdma_id_private *id_priv) > iboe_mac_vlan_to_ll(&route->path_rec->sgid, addr- > >dev_addr.src_dev_addr, vid); > iboe_mac_vlan_to_ll(&route->path_rec->dgid, addr- > >dev_addr.dst_dev_addr, vid); > > - route->path_rec->hop_limit = 1; > + route->path_rec->hop_limit = addr->uses_gateway ? 0xFF : 1; > route->path_rec->reversible = 1; > route->path_rec->pkey = cpu_to_be16(0xffff); > route->path_rec->mtu_selector = IB_SA_EQ; diff --git > a/include/rdma/ib_addr.h b/include/rdma/ib_addr.h index f3ac0f2..24bc985 > 100644 > --- a/include/rdma/ib_addr.h > +++ b/include/rdma/ib_addr.h > @@ -64,6 +64,7 @@ struct rdma_dev_addr { > unsigned char dst_dev_addr[MAX_ADDR_LEN]; > unsigned char broadcast[MAX_ADDR_LEN]; > unsigned short dev_type; > + bool uses_gateway; > int bound_dev_if; > enum rdma_transport_type transport; > }; > -- > 1.7.1 > > -- > To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the > body of a message to majordomo@vger.kernel.org More majordomo info at > http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Feb 12, 2014 at 9:44 PM, Somnath Kotur <Somnath.Kotur@emulex.com> wrote: > Hi Roland, > > Yes you are correct, currently it is not routable , and this > would probably be the first gentle step in enabling it to be so. > > Also, I did see that in ib_init_ah_from_wc() currently , it is already being > set to 0xff , so this would add symmetry and enable it both ways Iff needed > of course. > > Do you think this should go into for-next instead ? I'm still trying to understand why this patch should go anywhere. Repeating my previous question: is there any practical situation where this makes a difference? If not then wouldn't it make sense to design how routable IBoE will work overall before tinkering with minor stuff like this? - R. -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/infiniband/core/addr.c b/drivers/infiniband/core/addr.c index e90f2b2..6fa4721 100644 --- a/drivers/infiniband/core/addr.c +++ b/drivers/infiniband/core/addr.c @@ -250,6 +250,8 @@ static int addr4_resolve(struct sockaddr_in *src_in, goto put; } + addr->uses_gateway = rt->rt_uses_gateway; + ret = dst_fetch_ha(&rt->dst, addr, &fl4.daddr); put: ip_rt_put(rt); @@ -263,6 +265,7 @@ static int addr6_resolve(struct sockaddr_in6 *src_in, struct rdma_dev_addr *addr) { struct flowi6 fl6; + struct rt6_info *rt; struct dst_entry *dst; int ret; @@ -275,6 +278,7 @@ static int addr6_resolve(struct sockaddr_in6 *src_in, if ((ret = dst->error)) goto put; + rt = (struct rt6_info *)dst; if (ipv6_addr_any(&fl6.saddr)) { ret = ipv6_dev_get_saddr(&init_net, ip6_dst_idev(dst)->dev, &fl6.daddr, 0, &fl6.saddr); @@ -298,6 +302,7 @@ static int addr6_resolve(struct sockaddr_in6 *src_in, goto put; } + addr->uses_gateway = rt->rt6i_flags & RTF_GATEWAY; ret = dst_fetch_ha(dst, addr, &fl6.daddr); put: dst_release(dst); diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index 8e49db6..ae60104 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -1902,7 +1902,7 @@ static int cma_resolve_iboe_route(struct rdma_id_private *id_priv) iboe_mac_vlan_to_ll(&route->path_rec->sgid, addr->dev_addr.src_dev_addr, vid); iboe_mac_vlan_to_ll(&route->path_rec->dgid, addr->dev_addr.dst_dev_addr, vid); - route->path_rec->hop_limit = 1; + route->path_rec->hop_limit = addr->uses_gateway ? 0xFF : 1; route->path_rec->reversible = 1; route->path_rec->pkey = cpu_to_be16(0xffff); route->path_rec->mtu_selector = IB_SA_EQ; diff --git a/include/rdma/ib_addr.h b/include/rdma/ib_addr.h index f3ac0f2..24bc985 100644 --- a/include/rdma/ib_addr.h +++ b/include/rdma/ib_addr.h @@ -64,6 +64,7 @@ struct rdma_dev_addr { unsigned char dst_dev_addr[MAX_ADDR_LEN]; unsigned char broadcast[MAX_ADDR_LEN]; unsigned short dev_type; + bool uses_gateway; int bound_dev_if; enum rdma_transport_type transport; };
As suggested by Sean Hefty, take into account src/dst addresses on different subnets and set hop_limit in cma_resolve_iboe_route accordingly instead of blind hardcode to 1 or any other value. Signed-off-by: Somnath Kotur <somnath.kotur@emulex.com> --- Incorporated comments by Sean Hefty drivers/infiniband/core/addr.c | 5 +++++ drivers/infiniband/core/cma.c | 2 +- include/rdma/ib_addr.h | 1 + 3 files changed, 7 insertions(+), 1 deletions(-)