| Message ID | 168330138101.5953.12575990094340826016.stgit@oracle-102.nfsv4bat.org (mailing list archive) |
|---|---|
| State | RFC |
| Headers | show |
| Series | siw on tunnel devices | expand |
On Fri, May 05, 2023 at 11:43:11AM -0400, Chuck Lever wrote: > From: Chuck Lever <chuck.lever@oracle.com> > > In the past, LOOPBACK and NONE (tunnel) devices had all-zero MAC > addresses. siw_device_create() would fall back to copying the > device's name in those cases, because an all-zero MAC address breaks > the RDMA core IP-to-device lookup mechanism. Why not just make up a dummy address in SIW? It shouldn't need to leak out of it.. It is just some artifact of how the iWarp stuff has been designed Jason
> On May 5, 2023, at 3:58 PM, Jason Gunthorpe <jgg@ziepe.ca> wrote: > > On Fri, May 05, 2023 at 11:43:11AM -0400, Chuck Lever wrote: >> From: Chuck Lever <chuck.lever@oracle.com> >> >> In the past, LOOPBACK and NONE (tunnel) devices had all-zero MAC >> addresses. siw_device_create() would fall back to copying the >> device's name in those cases, because an all-zero MAC address breaks >> the RDMA core IP-to-device lookup mechanism. > > Why not just make up a dummy address in SIW? It shouldn't need to leak > out of it.. It is just some artifact of how the iWarp stuff has been > designed I've been trying that. Even though the siw0 device is now registered with a non-zero GID, cma_acquire_dev_by_src_ip() still comes up with a zero GID which matches no device. Address resolution then fails. I'm still looking into why. -- Chuck Lever
> On May 5, 2023, at 4:03 PM, Chuck Lever III <chuck.lever@oracle.com> wrote: > >> On May 5, 2023, at 3:58 PM, Jason Gunthorpe <jgg@ziepe.ca> wrote: >> >> On Fri, May 05, 2023 at 11:43:11AM -0400, Chuck Lever wrote: >>> From: Chuck Lever <chuck.lever@oracle.com> >>> >>> In the past, LOOPBACK and NONE (tunnel) devices had all-zero MAC >>> addresses. siw_device_create() would fall back to copying the >>> device's name in those cases, because an all-zero MAC address breaks >>> the RDMA core IP-to-device lookup mechanism. >> >> Why not just make up a dummy address in SIW? It shouldn't need to leak >> out of it.. It is just some artifact of how the iWarp stuff has been >> designed > > I've been trying that. > > Even though the siw0 device is now registered with a non-zero GID, > cma_acquire_dev_by_src_ip() still comes up with a zero GID which > matches no device. Address resolution then fails. > > I'm still looking into why. The tun0 device's flags are: UP|POINTOPOINT|NOARP|MULTICAST That flag combination turns addr_resolve_neigh() into a no-op, so that the returned GIDs and addresses are uninitialized. Cc'ing Parav because he's the last person who did significant work on this code path. I can hack this to make it work, but I have no idea what the proper solution would be. -- Chuck Lever
> On May 5, 2023, at 3:58 PM, Jason Gunthorpe <jgg@ziepe.ca> wrote: > > On Fri, May 05, 2023 at 11:43:11AM -0400, Chuck Lever wrote: >> From: Chuck Lever <chuck.lever@oracle.com> >> >> In the past, LOOPBACK and NONE (tunnel) devices had all-zero MAC >> addresses. siw_device_create() would fall back to copying the >> device's name in those cases, because an all-zero MAC address breaks >> the RDMA core IP-to-device lookup mechanism. > > Why not just make up a dummy address in SIW? It shouldn't need to leak > out of it.. It is just some artifact of how the iWarp stuff has been > designed So that approach is already being done in siw_device_create(), even though it is broken (the device name hasn't been initialized when the phony MAC is created, so it is all zeroes). I've fixed that and it still doesn't help. siw cannot modify the underlying net_device to add a made-up MAC address. The core address resolution code wants to find an L2 address for the egress device. The underlying ib_device, where a made-up GID might be stored, is not involved with address resolution AFAICT. tun devices have no L2 address. Neither do loopback devices, but address resolution makes an exception for LOOPBACK devices by redirecting to a local physical Ethernet device. Redirecting tun traffic to the local Ethernet device seems dodgy at best. I wasn't sure that an L2 address was required for siw before, but now I'm pretty confident that it is required by our implementation. -- Chuck Lever
On 5/23/2023 3:18 PM, Chuck Lever III wrote: > >> On May 5, 2023, at 3:58 PM, Jason Gunthorpe <jgg@ziepe.ca> wrote: >> >> On Fri, May 05, 2023 at 11:43:11AM -0400, Chuck Lever wrote: >>> From: Chuck Lever <chuck.lever@oracle.com> >>> >>> In the past, LOOPBACK and NONE (tunnel) devices had all-zero MAC >>> addresses. siw_device_create() would fall back to copying the >>> device's name in those cases, because an all-zero MAC address breaks >>> the RDMA core IP-to-device lookup mechanism. >> >> Why not just make up a dummy address in SIW? It shouldn't need to leak >> out of it.. It is just some artifact of how the iWarp stuff has been >> designed > > So that approach is already being done in siw_device_create(), > even though it is broken (the device name hasn't been initialized > when the phony MAC is created, so it is all zeroes). I've fixed > that and it still doesn't help. > > siw cannot modify the underlying net_device to add a made-up > MAC address. > > The core address resolution code wants to find an L2 address > for the egress device. The underlying ib_device, where a made-up > GID might be stored, is not involved with address resolution > AFAICT. > > tun devices have no L2 address. Neither do loopback devices, > but address resolution makes an exception for LOOPBACK devices > by redirecting to a local physical Ethernet device. > > Redirecting tun traffic to the local Ethernet device seems > dodgy at best. > > I wasn't sure that an L2 address was required for siw before, > but now I'm pretty confident that it is required by our > implementation. Does rxe work over tunnels? Seems like it would have the same issue. int rxe_register_device(struct rxe_dev *rxe, const char *ibdev_name) { ... addrconf_addr_eui48((unsigned char *)&dev->node_guid, rxe->ndev->dev_addr); static struct siw_device *siw_device_create(struct net_device *netdev) { ... addrconf_addr_eui48((unsigned char *)&base_dev->node_guid, netdev->dev_addr); Tom.
> On May 23, 2023, at 3:44 PM, Tom Talpey <tom@talpey.com> wrote: > > On 5/23/2023 3:18 PM, Chuck Lever III wrote: >>> On May 5, 2023, at 3:58 PM, Jason Gunthorpe <jgg@ziepe.ca> wrote: >>> >>> On Fri, May 05, 2023 at 11:43:11AM -0400, Chuck Lever wrote: >>>> From: Chuck Lever <chuck.lever@oracle.com> >>>> >>>> In the past, LOOPBACK and NONE (tunnel) devices had all-zero MAC >>>> addresses. siw_device_create() would fall back to copying the >>>> device's name in those cases, because an all-zero MAC address breaks >>>> the RDMA core IP-to-device lookup mechanism. >>> >>> Why not just make up a dummy address in SIW? It shouldn't need to leak >>> out of it.. It is just some artifact of how the iWarp stuff has been >>> designed >> So that approach is already being done in siw_device_create(), >> even though it is broken (the device name hasn't been initialized >> when the phony MAC is created, so it is all zeroes). I've fixed >> that and it still doesn't help. >> siw cannot modify the underlying net_device to add a made-up >> MAC address. >> The core address resolution code wants to find an L2 address >> for the egress device. The underlying ib_device, where a made-up >> GID might be stored, is not involved with address resolution >> AFAICT. >> tun devices have no L2 address. Neither do loopback devices, >> but address resolution makes an exception for LOOPBACK devices >> by redirecting to a local physical Ethernet device. >> Redirecting tun traffic to the local Ethernet device seems >> dodgy at best. >> I wasn't sure that an L2 address was required for siw before, >> but now I'm pretty confident that it is required by our >> implementation. > > Does rxe work over tunnels? (It's not tunnels per se, it's devices that don't have L2 addresses... and tun happens to be one instance of that class). My (brief) reading of the source code is that the use of devices that do not have L2 addresses is prohibited for rxe. > Seems like it would have the same issue. Agreed, if rxe did not prohibit them, it would have the same issue. To be clear: siw itself and the family of iWARP protocols shouldn't have any problem at all with such devices. The issue seems to be with the Linux implementation of address resolution. > int rxe_register_device(struct rxe_dev *rxe, const char *ibdev_name) > { > ... > addrconf_addr_eui48((unsigned char *)&dev->node_guid, > rxe->ndev->dev_addr); > > static struct siw_device *siw_device_create(struct net_device *netdev) > { > ... > addrconf_addr_eui48((unsigned char *)&base_dev->node_guid, > netdev->dev_addr); > > Tom. -- Chuck Lever
On Tue, May 23, 2023 at 07:18:18PM +0000, Chuck Lever III wrote: > The core address resolution code wants to find an L2 address > for the egress device. The underlying ib_device, where a made-up > GID might be stored, is not involved with address resolution > AFAICT. Where are you hitting this? Jason
> On May 31, 2023, at 3:04 PM, Jason Gunthorpe <jgg@ziepe.ca> wrote: > > On Tue, May 23, 2023 at 07:18:18PM +0000, Chuck Lever III wrote: > >> The core address resolution code wants to find an L2 address >> for the egress device. The underlying ib_device, where a made-up >> GID might be stored, is not involved with address resolution >> AFAICT. > > Where are you hitting this? kworker/2:0-26 [002] 551.962874: funcgraph_entry: | addr_resolve() { kworker/2:0-26 [002] 551.962874: bprint: addr_resolve: resolve_neigh=true resolve_by_gid_attr=false kworker/2:0-26 [002] 551.962874: funcgraph_entry: | addr4_resolve.constprop.0() { kworker/2:0-26 [002] 551.962875: bprint: addr4_resolve.constprop.0: src_in=0.0.0.0:35173 dst_in=100.72.1.2:20049 kworker/2:0-26 [002] 551.962875: funcgraph_entry: | ip_route_output_flow() { kworker/2:0-26 [002] 551.962875: funcgraph_entry: | ip_route_output_key_hash() { kworker/2:0-26 [002] 551.962876: funcgraph_entry: | ip_route_output_key_hash_rcu() { kworker/2:0-26 [002] 551.962876: funcgraph_entry: 4.526 us | __fib_lookup(); kworker/2:0-26 [002] 551.962881: funcgraph_entry: 0.264 us | fib_select_path(); kworker/2:0-26 [002] 551.962881: funcgraph_entry: 1.022 us | __mkroute_output(); kworker/2:0-26 [002] 551.962882: funcgraph_exit: 6.705 us | } kworker/2:0-26 [002] 551.962882: funcgraph_exit: 7.283 us | } kworker/2:0-26 [002] 551.962883: funcgraph_exit: 7.624 us | } kworker/2:0-26 [002] 551.962883: funcgraph_exit: 8.395 us | } kworker/2:0-26 [002] 551.962883: funcgraph_entry: | rdma_set_src_addr_rcu.constprop.0() { kworker/2:0-26 [002] 551.962883: bprint: rdma_set_src_addr_rcu.constprop.0: ndev=0xffff91f5135a4000 name=tailscale0 kworker/2:0-26 [002] 551.962884: funcgraph_entry: | copy_src_l2_addr() { kworker/2:0-26 [002] 551.962884: funcgraph_entry: 0.984 us | iff_flags2string(); kworker/2:0-26 [002] 551.962885: bprint: copy_src_l2_addr: ndev=0xffff91f5135a4000 dst_in=100.72.1.2:20049 flags=UP|POINTOPOINT|NOARP|MULTICAST kworker/2:0-26 [002] 551.962885: funcgraph_entry: | rdma_copy_src_l2_addr() { kworker/2:0-26 [002] 551.962886: funcgraph_entry: 0.148 us | devtype2string(); kworker/2:0-26 [002] 551.962887: bprint: rdma_copy_src_l2_addr: name=tailscale0 type=NONE src_dev_addr=00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 broadcast=00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ifindex=3 kworker/2:0-26 [002] 551.962887: funcgraph_exit: 1.488 us | } kworker/2:0-26 [002] 551.962887: bprint: copy_src_l2_addr: network type=IB kworker/2:0-26 [002] 551.962887: funcgraph_exit: 3.636 us | } kworker/2:0-26 [002] 551.962887: funcgraph_exit: 4.275 us | } Address resolution finds the right device, but there's a zero-value L2 address. Thus it cannot form a unique GID from that. Perhaps there needs to be a call to query_gid in here? -- Chuck Lever
On Wed, May 31, 2023 at 07:11:52PM +0000, Chuck Lever III wrote: > > > > On May 31, 2023, at 3:04 PM, Jason Gunthorpe <jgg@ziepe.ca> wrote: > > > > On Tue, May 23, 2023 at 07:18:18PM +0000, Chuck Lever III wrote: > > > >> The core address resolution code wants to find an L2 address > >> for the egress device. The underlying ib_device, where a made-up > >> GID might be stored, is not involved with address resolution > >> AFAICT. > > > > Where are you hitting this? > > kworker/2:0-26 [002] 551.962874: funcgraph_entry: | addr_resolve() { > kworker/2:0-26 [002] 551.962874: bprint: addr_resolve: resolve_neigh=true resolve_by_gid_attr=false > kworker/2:0-26 [002] 551.962874: funcgraph_entry: | addr4_resolve.constprop.0() { > kworker/2:0-26 [002] 551.962875: bprint: addr4_resolve.constprop.0: src_in=0.0.0.0:35173 dst_in=100.72.1.2:20049 > kworker/2:0-26 [002] 551.962875: funcgraph_entry: | ip_route_output_flow() { > kworker/2:0-26 [002] 551.962875: funcgraph_entry: | ip_route_output_key_hash() { > kworker/2:0-26 [002] 551.962876: funcgraph_entry: | ip_route_output_key_hash_rcu() { > kworker/2:0-26 [002] 551.962876: funcgraph_entry: 4.526 us | __fib_lookup(); > kworker/2:0-26 [002] 551.962881: funcgraph_entry: 0.264 us | fib_select_path(); > kworker/2:0-26 [002] 551.962881: funcgraph_entry: 1.022 us | __mkroute_output(); > kworker/2:0-26 [002] 551.962882: funcgraph_exit: 6.705 us | } > kworker/2:0-26 [002] 551.962882: funcgraph_exit: 7.283 us | } > kworker/2:0-26 [002] 551.962883: funcgraph_exit: 7.624 us | } > kworker/2:0-26 [002] 551.962883: funcgraph_exit: 8.395 us | } > kworker/2:0-26 [002] 551.962883: funcgraph_entry: | rdma_set_src_addr_rcu.constprop.0() { > kworker/2:0-26 [002] 551.962883: bprint: rdma_set_src_addr_rcu.constprop.0: ndev=0xffff91f5135a4000 name=tailscale0 > kworker/2:0-26 [002] 551.962884: funcgraph_entry: | copy_src_l2_addr() { > kworker/2:0-26 [002] 551.962884: funcgraph_entry: 0.984 us | iff_flags2string(); > kworker/2:0-26 [002] 551.962885: bprint: copy_src_l2_addr: ndev=0xffff91f5135a4000 dst_in=100.72.1.2:20049 flags=UP|POINTOPOINT|NOARP|MULTICAST > kworker/2:0-26 [002] 551.962885: funcgraph_entry: | rdma_copy_src_l2_addr() { > kworker/2:0-26 [002] 551.962886: funcgraph_entry: 0.148 us | devtype2string(); > kworker/2:0-26 [002] 551.962887: bprint: rdma_copy_src_l2_addr: name=tailscale0 type=NONE src_dev_addr=00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 broadcast=00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ifindex=3 > kworker/2:0-26 [002] 551.962887: funcgraph_exit: 1.488 us | } > kworker/2:0-26 [002] 551.962887: bprint: copy_src_l2_addr: network type=IB > kworker/2:0-26 [002] 551.962887: funcgraph_exit: 3.636 us | } > kworker/2:0-26 [002] 551.962887: funcgraph_exit: 4.275 us | } > > > Address resolution finds the right device, but there's > a zero-value L2 address. Sure, but why is that a problem? This got to rdma_set_src_addr_rcu, so the resolution suceeded, where is the failure? From the above trace I think addr_resolve() succeeded? > Thus it cannot form a unique GID from that. Perhaps there needs to > be a call to query_gid in here? So your issue is cma_iw_acquire_dev() which looks like it is encoding the MAC into the GID for some reason? We don't do that on rocee, the GID encodes the IP address I have no idea how iWarp works, but this is surprising that it puts a MAC in the GID.. If the iwarp device has only one GID ever and it is always the "MAC" the cma_iw_acquire_dev()'s logic is simply wrong, it should check that the dev_addr's netdev matches the one and only GID and just use the GID. No reason to search for GIDs. A small edit to cma_validate_port() might make sense, it is kind of wrong to force the gid_type to IB_GID_TYPE_IB for whatever ARPHRD type the tunnel is using. Jason
> On May 31, 2023, at 4:09 PM, Jason Gunthorpe <jgg@ziepe.ca> wrote: > > On Wed, May 31, 2023 at 07:11:52PM +0000, Chuck Lever III wrote: >> >> >>> On May 31, 2023, at 3:04 PM, Jason Gunthorpe <jgg@ziepe.ca> wrote: >>> >>> On Tue, May 23, 2023 at 07:18:18PM +0000, Chuck Lever III wrote: >>> >>>> The core address resolution code wants to find an L2 address >>>> for the egress device. The underlying ib_device, where a made-up >>>> GID might be stored, is not involved with address resolution >>>> AFAICT. >>> >>> Where are you hitting this? >> >> kworker/2:0-26 [002] 551.962874: funcgraph_entry: | addr_resolve() { >> kworker/2:0-26 [002] 551.962874: bprint: addr_resolve: resolve_neigh=true resolve_by_gid_attr=false >> kworker/2:0-26 [002] 551.962874: funcgraph_entry: | addr4_resolve.constprop.0() { >> kworker/2:0-26 [002] 551.962875: bprint: addr4_resolve.constprop.0: src_in=0.0.0.0:35173 dst_in=100.72.1.2:20049 >> kworker/2:0-26 [002] 551.962875: funcgraph_entry: | ip_route_output_flow() { >> kworker/2:0-26 [002] 551.962875: funcgraph_entry: | ip_route_output_key_hash() { >> kworker/2:0-26 [002] 551.962876: funcgraph_entry: | ip_route_output_key_hash_rcu() { >> kworker/2:0-26 [002] 551.962876: funcgraph_entry: 4.526 us | __fib_lookup(); >> kworker/2:0-26 [002] 551.962881: funcgraph_entry: 0.264 us | fib_select_path(); >> kworker/2:0-26 [002] 551.962881: funcgraph_entry: 1.022 us | __mkroute_output(); >> kworker/2:0-26 [002] 551.962882: funcgraph_exit: 6.705 us | } >> kworker/2:0-26 [002] 551.962882: funcgraph_exit: 7.283 us | } >> kworker/2:0-26 [002] 551.962883: funcgraph_exit: 7.624 us | } >> kworker/2:0-26 [002] 551.962883: funcgraph_exit: 8.395 us | } >> kworker/2:0-26 [002] 551.962883: funcgraph_entry: | rdma_set_src_addr_rcu.constprop.0() { >> kworker/2:0-26 [002] 551.962883: bprint: rdma_set_src_addr_rcu.constprop.0: ndev=0xffff91f5135a4000 name=tailscale0 >> kworker/2:0-26 [002] 551.962884: funcgraph_entry: | copy_src_l2_addr() { >> kworker/2:0-26 [002] 551.962884: funcgraph_entry: 0.984 us | iff_flags2string(); >> kworker/2:0-26 [002] 551.962885: bprint: copy_src_l2_addr: ndev=0xffff91f5135a4000 dst_in=100.72.1.2:20049 flags=UP|POINTOPOINT|NOARP|MULTICAST >> kworker/2:0-26 [002] 551.962885: funcgraph_entry: | rdma_copy_src_l2_addr() { >> kworker/2:0-26 [002] 551.962886: funcgraph_entry: 0.148 us | devtype2string(); >> kworker/2:0-26 [002] 551.962887: bprint: rdma_copy_src_l2_addr: name=tailscale0 type=NONE src_dev_addr=00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 broadcast=00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ifindex=3 >> kworker/2:0-26 [002] 551.962887: funcgraph_exit: 1.488 us | } >> kworker/2:0-26 [002] 551.962887: bprint: copy_src_l2_addr: network type=IB >> kworker/2:0-26 [002] 551.962887: funcgraph_exit: 3.636 us | } >> kworker/2:0-26 [002] 551.962887: funcgraph_exit: 4.275 us | } >> >> >> Address resolution finds the right device, but there's >> a zero-value L2 address. > > Sure, but why is that a problem? > > This got to rdma_set_src_addr_rcu, so the resolution suceeded, where > is the failure? From the above trace I think addr_resolve() succeeded? Possibly it did succeed. But the ULP consumer sees CM_ADDR_ERROR_EVENT, and does not proceed to route resolution. >> Thus it cannot form a unique GID from that. Perhaps there needs to >> be a call to query_gid in here? > > So your issue is cma_iw_acquire_dev() which looks like it is encoding > the MAC into the GID for some reason? We don't do that on rocee, the > GID encodes the IP address Well, I'm not getting there at all on the initiator side. cma_iw_acquire_dev() is called only for listeners, I thought. > > I have no idea how iWarp works, but this is surprising that it puts a > MAC in the GID.. > > If the iwarp device has only one GID ever and it is always the "MAC" > the cma_iw_acquire_dev()'s logic is simply wrong, it should check that > the dev_addr's netdev matches the one and only GID and just use the > GID. No reason to search for GIDs. > > A small edit to cma_validate_port() might make sense, it is kind of > wrong to force the gid_type to IB_GID_TYPE_IB for whatever ARPHRD type > the tunnel is using. I will have a look. -- Chuck Lever
diff --git a/drivers/infiniband/sw/siw/siw_main.c b/drivers/infiniband/sw/siw/siw_main.c index 65b5cda5457b..2c31bf397993 100644 --- a/drivers/infiniband/sw/siw/siw_main.c +++ b/drivers/infiniband/sw/siw/siw_main.c @@ -304,10 +304,15 @@ static const struct ib_device_ops siw_device_ops = { static struct siw_device *siw_device_create(struct net_device *netdev) { + static const u8 zeromac[ETH_ALEN] = { 0 }; struct siw_device *sdev = NULL; struct ib_device *base_dev; int rv; + if ((netdev->addr_len != ETH_ALEN) || + (memcmp(netdev->dev_addr, zeromac, ETH_ALEN) == 0)) + return NULL; + sdev = ib_alloc_device(siw_device, base_dev); if (!sdev) return NULL; @@ -316,21 +321,8 @@ static struct siw_device *siw_device_create(struct net_device *netdev) sdev->netdev = netdev; - if (netdev->type != ARPHRD_LOOPBACK && netdev->type != ARPHRD_NONE) { - addrconf_addr_eui48((unsigned char *)&base_dev->node_guid, - netdev->dev_addr); - } else { - /* - * This device does not have a HW address, - * but connection mangagement lib expects gid != 0 - */ - size_t len = min_t(size_t, strlen(base_dev->name), 6); - char addr[6] = { }; - - memcpy(addr, base_dev->name, len); - addrconf_addr_eui48((unsigned char *)&base_dev->node_guid, - addr); - } + addrconf_addr_eui48((unsigned char *)&base_dev->node_guid, + netdev->dev_addr); base_dev->uverbs_cmd_mask |= BIT_ULL(IB_USER_VERBS_CMD_POST_SEND);