* [PATCH net-next 01/10] ipv4: Replace route in list before notifying
2020-01-07 15:45 [PATCH net-next 00/10] net: Add route offload indication Ido Schimmel
@ 2020-01-07 15:45 ` Ido Schimmel
2020-01-08 18:33 ` David Ahern
2020-01-07 15:45 ` [PATCH net-next 02/10] ipv4: Encapsulate function arguments in a struct Ido Schimmel
` (8 subsequent siblings)
9 siblings, 1 reply; 17+ messages in thread
From: Ido Schimmel @ 2020-01-07 15:45 UTC (permalink / raw)
To: netdev; +Cc: davem, jiri, jakub.kicinski, dsahern, roopa, mlxsw, Ido Schimmel
From: Ido Schimmel <idosch@mellanox.com>
Subsequent patches will add an offload / trap indication to routes which
will signal if the route is present in hardware or not.
After programming the route to the hardware, drivers will have to ask
the IPv4 code to set the flags by passing the route's key.
In the case of route replace, the new route is notified before it is
actually inserted into the FIB alias list. This can prevent simple
drivers (e.g., netdevsim) that program the route to the hardware in the
same context it is notified in from being able to set the flag.
Solve this by first inserting the new route to the list and rollback the
operation in case the route was vetoed.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
---
net/ipv4/fib_trie.c | 11 +++++++----
1 file changed, 7 insertions(+), 4 deletions(-)
diff --git a/net/ipv4/fib_trie.c b/net/ipv4/fib_trie.c
index b92a42433a7d..39f56d68ec19 100644
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -1221,23 +1221,26 @@ int fib_table_insert(struct net *net, struct fib_table *tb,
new_fa->tb_id = tb->tb_id;
new_fa->fa_default = -1;
+ hlist_replace_rcu(&fa->fa_list, &new_fa->fa_list);
+
if (fib_find_alias(&l->leaf, fa->fa_slen, 0, 0,
- tb->tb_id, true) == fa) {
+ tb->tb_id, true) == new_fa) {
enum fib_event_type fib_event;
fib_event = FIB_EVENT_ENTRY_REPLACE;
err = call_fib_entry_notifiers(net, fib_event,
key, plen,
new_fa, extack);
- if (err)
+ if (err) {
+ hlist_replace_rcu(&new_fa->fa_list,
+ &fa->fa_list);
goto out_free_new_fa;
+ }
}
rtmsg_fib(RTM_NEWROUTE, htonl(key), new_fa, plen,
tb->tb_id, &cfg->fc_nlinfo, nlflags);
- hlist_replace_rcu(&fa->fa_list, &new_fa->fa_list);
-
alias_free_mem_rcu(fa);
fib_release_info(fi_drop);
--
2.24.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* Re: [PATCH net-next 01/10] ipv4: Replace route in list before notifying
2020-01-07 15:45 ` [PATCH net-next 01/10] ipv4: Replace route in list before notifying Ido Schimmel
@ 2020-01-08 18:33 ` David Ahern
0 siblings, 0 replies; 17+ messages in thread
From: David Ahern @ 2020-01-08 18:33 UTC (permalink / raw)
To: Ido Schimmel, netdev
Cc: davem, jiri, jakub.kicinski, roopa, mlxsw, Ido Schimmel
On 1/7/20 8:45 AM, Ido Schimmel wrote:
> From: Ido Schimmel <idosch@mellanox.com>
>
> Subsequent patches will add an offload / trap indication to routes which
> will signal if the route is present in hardware or not.
>
> After programming the route to the hardware, drivers will have to ask
> the IPv4 code to set the flags by passing the route's key.
>
> In the case of route replace, the new route is notified before it is
> actually inserted into the FIB alias list. This can prevent simple
> drivers (e.g., netdevsim) that program the route to the hardware in the
> same context it is notified in from being able to set the flag.
>
> Solve this by first inserting the new route to the list and rollback the
> operation in case the route was vetoed.
>
> Signed-off-by: Ido Schimmel <idosch@mellanox.com>
> Reviewed-by: Jiri Pirko <jiri@mellanox.com>
> ---
> net/ipv4/fib_trie.c | 11 +++++++----
> 1 file changed, 7 insertions(+), 4 deletions(-)
>
Reviewed-by: David Ahern <dsahern@gmail.com>
^ permalink raw reply [flat|nested] 17+ messages in thread
* [PATCH net-next 02/10] ipv4: Encapsulate function arguments in a struct
2020-01-07 15:45 [PATCH net-next 00/10] net: Add route offload indication Ido Schimmel
2020-01-07 15:45 ` [PATCH net-next 01/10] ipv4: Replace route in list before notifying Ido Schimmel
@ 2020-01-07 15:45 ` Ido Schimmel
2020-01-08 18:37 ` David Ahern
2020-01-07 15:45 ` [PATCH net-next 03/10] ipv4: Add "offload" and "trap" indications to routes Ido Schimmel
` (7 subsequent siblings)
9 siblings, 1 reply; 17+ messages in thread
From: Ido Schimmel @ 2020-01-07 15:45 UTC (permalink / raw)
To: netdev; +Cc: davem, jiri, jakub.kicinski, dsahern, roopa, mlxsw, Ido Schimmel
From: Ido Schimmel <idosch@mellanox.com>
fib_dump_info() is used to prepare RTM_{NEW,DEL}ROUTE netlink messages
using the passed arguments. Currently, the function takes 11 arguments,
6 of which are attributes of the route being dumped (e.g., prefix, TOS).
The next patch will need the function to also dump to user space an
indication if the route is present in hardware or not. Instead of
passing yet another argument, change the function to take a struct
containing the different route attributes.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
---
net/ipv4/fib_lookup.h | 14 +++++++++++---
net/ipv4/fib_semantics.c | 26 ++++++++++++++++----------
net/ipv4/fib_trie.c | 14 +++++++++-----
net/ipv4/route.c | 12 +++++++++---
4 files changed, 45 insertions(+), 21 deletions(-)
diff --git a/net/ipv4/fib_lookup.h b/net/ipv4/fib_lookup.h
index a68b5e21ec51..b34594a9965f 100644
--- a/net/ipv4/fib_lookup.h
+++ b/net/ipv4/fib_lookup.h
@@ -21,6 +21,15 @@ struct fib_alias {
#define FA_S_ACCESSED 0x01
+struct fib_rt_info {
+ struct fib_info *fi;
+ u32 tb_id;
+ __be32 dst;
+ int dst_len;
+ u8 tos;
+ u8 type;
+};
+
/* Dont write on fa_state unless needed, to keep it shared on all cpus */
static inline void fib_alias_accessed(struct fib_alias *fa)
{
@@ -35,9 +44,8 @@ struct fib_info *fib_create_info(struct fib_config *cfg,
int fib_nh_match(struct fib_config *cfg, struct fib_info *fi,
struct netlink_ext_ack *extack);
bool fib_metrics_match(struct fib_config *cfg, struct fib_info *fi);
-int fib_dump_info(struct sk_buff *skb, u32 pid, u32 seq, int event, u32 tb_id,
- u8 type, __be32 dst, int dst_len, u8 tos, struct fib_info *fi,
- unsigned int);
+int fib_dump_info(struct sk_buff *skb, u32 pid, u32 seq, int event,
+ struct fib_rt_info *fri, unsigned int);
void rtmsg_fib(int event, __be32 key, struct fib_alias *fa, int dst_len,
u32 tb_id, const struct nl_info *info, unsigned int nlm_flags);
diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c
index f1888c683426..3ed1349be428 100644
--- a/net/ipv4/fib_semantics.c
+++ b/net/ipv4/fib_semantics.c
@@ -504,6 +504,7 @@ void rtmsg_fib(int event, __be32 key, struct fib_alias *fa,
int dst_len, u32 tb_id, const struct nl_info *info,
unsigned int nlm_flags)
{
+ struct fib_rt_info fri;
struct sk_buff *skb;
u32 seq = info->nlh ? info->nlh->nlmsg_seq : 0;
int err = -ENOBUFS;
@@ -512,9 +513,13 @@ void rtmsg_fib(int event, __be32 key, struct fib_alias *fa,
if (!skb)
goto errout;
- err = fib_dump_info(skb, info->portid, seq, event, tb_id,
- fa->fa_type, key, dst_len,
- fa->fa_tos, fa->fa_info, nlm_flags);
+ fri.fi = fa->fa_info;
+ fri.tb_id = tb_id;
+ fri.dst = key;
+ fri.dst_len = dst_len;
+ fri.tos = fa->fa_tos;
+ fri.type = fa->fa_type;
+ err = fib_dump_info(skb, info->portid, seq, event, &fri, nlm_flags);
if (err < 0) {
/* -EMSGSIZE implies BUG in fib_nlmsg_size() */
WARN_ON(err == -EMSGSIZE);
@@ -1725,10 +1730,11 @@ static int fib_add_multipath(struct sk_buff *skb, struct fib_info *fi)
#endif
int fib_dump_info(struct sk_buff *skb, u32 portid, u32 seq, int event,
- u32 tb_id, u8 type, __be32 dst, int dst_len, u8 tos,
- struct fib_info *fi, unsigned int flags)
+ struct fib_rt_info *fri, unsigned int flags)
{
- unsigned int nhs = fib_info_num_path(fi);
+ unsigned int nhs = fib_info_num_path(fri->fi);
+ struct fib_info *fi = fri->fi;
+ u32 tb_id = fri->tb_id;
struct nlmsghdr *nlh;
struct rtmsg *rtm;
@@ -1738,22 +1744,22 @@ int fib_dump_info(struct sk_buff *skb, u32 portid, u32 seq, int event,
rtm = nlmsg_data(nlh);
rtm->rtm_family = AF_INET;
- rtm->rtm_dst_len = dst_len;
+ rtm->rtm_dst_len = fri->dst_len;
rtm->rtm_src_len = 0;
- rtm->rtm_tos = tos;
+ rtm->rtm_tos = fri->tos;
if (tb_id < 256)
rtm->rtm_table = tb_id;
else
rtm->rtm_table = RT_TABLE_COMPAT;
if (nla_put_u32(skb, RTA_TABLE, tb_id))
goto nla_put_failure;
- rtm->rtm_type = type;
+ rtm->rtm_type = fri->type;
rtm->rtm_flags = fi->fib_flags;
rtm->rtm_scope = fi->fib_scope;
rtm->rtm_protocol = fi->fib_protocol;
if (rtm->rtm_dst_len &&
- nla_put_in_addr(skb, RTA_DST, dst))
+ nla_put_in_addr(skb, RTA_DST, fri->dst))
goto nla_put_failure;
if (fi->fib_priority &&
nla_put_u32(skb, RTA_PRIORITY, fi->fib_priority))
diff --git a/net/ipv4/fib_trie.c b/net/ipv4/fib_trie.c
index 39f56d68ec19..75af3f8ae50e 100644
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -2194,14 +2194,18 @@ static int fn_trie_dump_leaf(struct key_vector *l, struct fib_table *tb,
if (filter->dump_routes) {
if (!s_fa) {
+ struct fib_rt_info fri;
+
+ fri.fi = fi;
+ fri.tb_id = tb->tb_id;
+ fri.dst = xkey;
+ fri.dst_len = KEYLENGTH - fa->fa_slen;
+ fri.tos = fa->fa_tos;
+ fri.type = fa->fa_type;
err = fib_dump_info(skb,
NETLINK_CB(cb->skb).portid,
cb->nlh->nlmsg_seq,
- RTM_NEWROUTE,
- tb->tb_id, fa->fa_type,
- xkey,
- KEYLENGTH - fa->fa_slen,
- fa->fa_tos, fi, flags);
+ RTM_NEWROUTE, &fri, flags);
if (err < 0)
goto stop;
}
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 87e979f2b74a..167a7357d12a 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -3223,16 +3223,22 @@ static int inet_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh,
skb_reset_mac_header(skb);
if (rtm->rtm_flags & RTM_F_FIB_MATCH) {
+ struct fib_rt_info fri;
+
if (!res.fi) {
err = fib_props[res.type].error;
if (!err)
err = -EHOSTUNREACH;
goto errout_rcu;
}
+ fri.fi = res.fi;
+ fri.tb_id = table_id;
+ fri.dst = res.prefix;
+ fri.dst_len = res.prefixlen;
+ fri.tos = fl4.flowi4_tos;
+ fri.type = rt->rt_type;
err = fib_dump_info(skb, NETLINK_CB(in_skb).portid,
- nlh->nlmsg_seq, RTM_NEWROUTE, table_id,
- rt->rt_type, res.prefix, res.prefixlen,
- fl4.flowi4_tos, res.fi, 0);
+ nlh->nlmsg_seq, RTM_NEWROUTE, &fri, 0);
} else {
err = rt_fill_info(net, dst, src, rt, table_id, &fl4, skb,
NETLINK_CB(in_skb).portid,
--
2.24.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* Re: [PATCH net-next 02/10] ipv4: Encapsulate function arguments in a struct
2020-01-07 15:45 ` [PATCH net-next 02/10] ipv4: Encapsulate function arguments in a struct Ido Schimmel
@ 2020-01-08 18:37 ` David Ahern
2020-01-09 8:26 ` Ido Schimmel
0 siblings, 1 reply; 17+ messages in thread
From: David Ahern @ 2020-01-08 18:37 UTC (permalink / raw)
To: Ido Schimmel, netdev
Cc: davem, jiri, jakub.kicinski, roopa, mlxsw, Ido Schimmel
On 1/7/20 8:45 AM, Ido Schimmel wrote:
> diff --git a/net/ipv4/fib_lookup.h b/net/ipv4/fib_lookup.h
> index a68b5e21ec51..b34594a9965f 100644
> --- a/net/ipv4/fib_lookup.h
> +++ b/net/ipv4/fib_lookup.h
> @@ -21,6 +21,15 @@ struct fib_alias {
>
> #define FA_S_ACCESSED 0x01
>
> +struct fib_rt_info {
> + struct fib_info *fi;
> + u32 tb_id;
> + __be32 dst;
> + int dst_len;
> + u8 tos;
> + u8 type;
> +};
> +
> /* Dont write on fa_state unless needed, to keep it shared on all cpus */
> static inline void fib_alias_accessed(struct fib_alias *fa)
> {
> @@ -35,9 +44,8 @@ struct fib_info *fib_create_info(struct fib_config *cfg,
> int fib_nh_match(struct fib_config *cfg, struct fib_info *fi,
> struct netlink_ext_ack *extack);
> bool fib_metrics_match(struct fib_config *cfg, struct fib_info *fi);
> -int fib_dump_info(struct sk_buff *skb, u32 pid, u32 seq, int event, u32 tb_id,
> - u8 type, __be32 dst, int dst_len, u8 tos, struct fib_info *fi,
> - unsigned int);
> +int fib_dump_info(struct sk_buff *skb, u32 pid, u32 seq, int event,
> + struct fib_rt_info *fri, unsigned int);
since you are modifying this, can you add a name for that last argument?
Otherwise, nice cleanup.
Reviewed-by: David Ahern <dsahern@gmail.com>
^ permalink raw reply [flat|nested] 17+ messages in thread* Re: [PATCH net-next 02/10] ipv4: Encapsulate function arguments in a struct
2020-01-08 18:37 ` David Ahern
@ 2020-01-09 8:26 ` Ido Schimmel
0 siblings, 0 replies; 17+ messages in thread
From: Ido Schimmel @ 2020-01-09 8:26 UTC (permalink / raw)
To: David Ahern
Cc: netdev, davem, jiri, jakub.kicinski, roopa, mlxsw, Ido Schimmel
On Wed, Jan 08, 2020 at 11:37:28AM -0700, David Ahern wrote:
> On 1/7/20 8:45 AM, Ido Schimmel wrote:
> > diff --git a/net/ipv4/fib_lookup.h b/net/ipv4/fib_lookup.h
> > index a68b5e21ec51..b34594a9965f 100644
> > --- a/net/ipv4/fib_lookup.h
> > +++ b/net/ipv4/fib_lookup.h
> > @@ -21,6 +21,15 @@ struct fib_alias {
> >
> > #define FA_S_ACCESSED 0x01
> >
> > +struct fib_rt_info {
> > + struct fib_info *fi;
> > + u32 tb_id;
> > + __be32 dst;
> > + int dst_len;
> > + u8 tos;
> > + u8 type;
> > +};
> > +
> > /* Dont write on fa_state unless needed, to keep it shared on all cpus */
> > static inline void fib_alias_accessed(struct fib_alias *fa)
> > {
> > @@ -35,9 +44,8 @@ struct fib_info *fib_create_info(struct fib_config *cfg,
> > int fib_nh_match(struct fib_config *cfg, struct fib_info *fi,
> > struct netlink_ext_ack *extack);
> > bool fib_metrics_match(struct fib_config *cfg, struct fib_info *fi);
> > -int fib_dump_info(struct sk_buff *skb, u32 pid, u32 seq, int event, u32 tb_id,
> > - u8 type, __be32 dst, int dst_len, u8 tos, struct fib_info *fi,
> > - unsigned int);
> > +int fib_dump_info(struct sk_buff *skb, u32 pid, u32 seq, int event,
> > + struct fib_rt_info *fri, unsigned int);
>
> since you are modifying this, can you add a name for that last argument?
Yes, will add in v2
>
>
> Otherwise, nice cleanup.
>
> Reviewed-by: David Ahern <dsahern@gmail.com>
Thanks!
^ permalink raw reply [flat|nested] 17+ messages in thread
* [PATCH net-next 03/10] ipv4: Add "offload" and "trap" indications to routes
2020-01-07 15:45 [PATCH net-next 00/10] net: Add route offload indication Ido Schimmel
2020-01-07 15:45 ` [PATCH net-next 01/10] ipv4: Replace route in list before notifying Ido Schimmel
2020-01-07 15:45 ` [PATCH net-next 02/10] ipv4: Encapsulate function arguments in a struct Ido Schimmel
@ 2020-01-07 15:45 ` Ido Schimmel
2020-01-09 15:47 ` David Ahern
2020-01-07 15:45 ` [PATCH net-next 04/10] ipv6: " Ido Schimmel
` (6 subsequent siblings)
9 siblings, 1 reply; 17+ messages in thread
From: Ido Schimmel @ 2020-01-07 15:45 UTC (permalink / raw)
To: netdev; +Cc: davem, jiri, jakub.kicinski, dsahern, roopa, mlxsw, Ido Schimmel
From: Ido Schimmel <idosch@mellanox.com>
When performing L3 offload, routes and nexthops are usually programmed
into two different tables in the underlying device. Therefore, the fact
that a nexthop resides in hardware does not necessarily mean that all
the associated routes also reside in hardware and vice-versa.
While the kernel can signal to user space the presence of a nexthop in
hardware (via 'RTNH_F_OFFLOAD'), it does not have a corresponding flag
for routes. In addition, the fact that a route resides in hardware does
not necessarily mean that the traffic is offloaded. For example,
unreachable routes (i.e., 'RTN_UNREACHABLE') are programmed to trap
packets to the CPU so that the kernel will be able to generate the
appropriate ICMP error packet.
This patch adds an "offload" and "trap" indications to IPv4 routes, so
that users will have better visibility into the offload process.
'struct fib_alias' is extended with two new fields that indicate if the
route resides in hardware or not and if it is offloading traffic from
the kernel or trapping packets to it. Note that the new fields are added
in the 6 bytes hole and therefore the struct still fits in a single
cache line [1].
Capable drivers are expected to invoke fib_alias_hw_flags_set() with the
route's key in order to set the flags.
The indications are dumped to user space via a new flags (i.e.,
'RTM_F_OFFLOAD' and 'RTM_F_TRAP') in the 'rtm_flags' field in the
ancillary header.
[1]
struct fib_alias {
struct hlist_node fa_list; /* 0 16 */
struct fib_info * fa_info; /* 16 8 */
u8 fa_tos; /* 24 1 */
u8 fa_type; /* 25 1 */
u8 fa_state; /* 26 1 */
u8 fa_slen; /* 27 1 */
u32 tb_id; /* 28 4 */
s16 fa_default; /* 32 2 */
u8 offload:1; /* 34: 0 1 */
u8 trap:1; /* 34: 1 1 */
u8 unused:6; /* 34: 2 1 */
/* XXX 5 bytes hole, try to pack */
struct callback_head rcu __attribute__((__aligned__(8))); /* 40 16 */
/* size: 56, cachelines: 1, members: 12 */
/* sum members: 50, holes: 1, sum holes: 5 */
/* sum bitfield members: 8 bits (1 bytes) */
/* forced alignments: 1, forced holes: 1, sum forced holes: 5 */
/* last cacheline: 56 bytes */
} __attribute__((__aligned__(8)));
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
---
include/net/ip_fib.h | 4 +++
include/uapi/linux/rtnetlink.h | 2 ++
net/ipv4/fib_lookup.h | 6 ++++
net/ipv4/fib_semantics.c | 7 +++++
net/ipv4/fib_trie.c | 56 ++++++++++++++++++++++++++++++++++
net/ipv4/route.c | 19 ++++++++++++
6 files changed, 94 insertions(+)
diff --git a/include/net/ip_fib.h b/include/net/ip_fib.h
index b9cba41c6d4f..5fc57c015e28 100644
--- a/include/net/ip_fib.h
+++ b/include/net/ip_fib.h
@@ -21,6 +21,7 @@
#include <linux/percpu.h>
#include <linux/notifier.h>
#include <linux/refcount.h>
+#include <linux/bits.h>
struct fib_config {
u8 fc_dst_len;
@@ -464,6 +465,9 @@ int fib_nh_common_init(struct fib_nh_common *nhc, struct nlattr *fc_encap,
void fib_nh_common_release(struct fib_nh_common *nhc);
/* Exported by fib_trie.c */
+void fib_alias_hw_flags_set(struct net *net, u32 dst, int dst_len,
+ const struct fib_info *fi, u8 tos, u8 type,
+ u32 tb_id, bool offload, bool trap);
void fib_trie_init(void);
struct fib_table *fib_trie_table(u32 id, struct fib_table *alias);
diff --git a/include/uapi/linux/rtnetlink.h b/include/uapi/linux/rtnetlink.h
index 1418a8362bb7..cd43321d20dd 100644
--- a/include/uapi/linux/rtnetlink.h
+++ b/include/uapi/linux/rtnetlink.h
@@ -309,6 +309,8 @@ enum rt_scope_t {
#define RTM_F_PREFIX 0x800 /* Prefix addresses */
#define RTM_F_LOOKUP_TABLE 0x1000 /* set rtm_table to FIB lookup result */
#define RTM_F_FIB_MATCH 0x2000 /* return full fib lookup match */
+#define RTM_F_OFFLOAD 0x4000 /* route is offloaded */
+#define RTM_F_TRAP 0x8000 /* route is trapping packets */
/* Reserved table identifiers */
diff --git a/net/ipv4/fib_lookup.h b/net/ipv4/fib_lookup.h
index b34594a9965f..b7839127a43a 100644
--- a/net/ipv4/fib_lookup.h
+++ b/net/ipv4/fib_lookup.h
@@ -16,6 +16,9 @@ struct fib_alias {
u8 fa_slen;
u32 tb_id;
s16 fa_default;
+ u8 offload:1,
+ trap:1,
+ unused:6;
struct rcu_head rcu;
};
@@ -28,6 +31,9 @@ struct fib_rt_info {
int dst_len;
u8 tos;
u8 type;
+ u8 offload:1,
+ trap:1,
+ unused:6;
};
/* Dont write on fa_state unless needed, to keep it shared on all cpus */
diff --git a/net/ipv4/fib_semantics.c b/net/ipv4/fib_semantics.c
index 3ed1349be428..a803cdd9400a 100644
--- a/net/ipv4/fib_semantics.c
+++ b/net/ipv4/fib_semantics.c
@@ -519,6 +519,8 @@ void rtmsg_fib(int event, __be32 key, struct fib_alias *fa,
fri.dst_len = dst_len;
fri.tos = fa->fa_tos;
fri.type = fa->fa_type;
+ fri.offload = fa->offload;
+ fri.trap = fa->trap;
err = fib_dump_info(skb, info->portid, seq, event, &fri, nlm_flags);
if (err < 0) {
/* -EMSGSIZE implies BUG in fib_nlmsg_size() */
@@ -1801,6 +1803,11 @@ int fib_dump_info(struct sk_buff *skb, u32 portid, u32 seq, int event,
goto nla_put_failure;
}
+ if (fri->offload)
+ rtm->rtm_flags |= RTM_F_OFFLOAD;
+ if (fri->trap)
+ rtm->rtm_flags |= RTM_F_TRAP;
+
nlmsg_end(skb, nlh);
return 0;
diff --git a/net/ipv4/fib_trie.c b/net/ipv4/fib_trie.c
index 75af3f8ae50e..272c9f73e4b3 100644
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -1012,6 +1012,56 @@ static struct fib_alias *fib_find_alias(struct hlist_head *fah, u8 slen,
return NULL;
}
+static struct fib_alias *
+fib_find_matching_alias(struct net *net, u32 dst, int dst_len,
+ const struct fib_info *fi, u8 tos, u8 type, u32 tb_id)
+{
+ u8 slen = KEYLENGTH - dst_len;
+ struct key_vector *l, *tp;
+ struct fib_table *tb;
+ struct fib_alias *fa;
+ struct trie *t;
+
+ tb = fib_get_table(net, tb_id);
+ if (!tb)
+ return NULL;
+
+ t = (struct trie *)tb->tb_data;
+ l = fib_find_node(t, &tp, dst);
+ if (!l)
+ return NULL;
+
+ hlist_for_each_entry_rcu(fa, &l->leaf, fa_list) {
+ if (fa->fa_slen == slen && fa->tb_id == tb_id &&
+ fa->fa_tos == tos && fa->fa_info == fi &&
+ fa->fa_type == type)
+ return fa;
+ }
+
+ return NULL;
+}
+
+void fib_alias_hw_flags_set(struct net *net, u32 dst, int dst_len,
+ const struct fib_info *fi, u8 tos, u8 type,
+ u32 tb_id, bool offload, bool trap)
+{
+ struct fib_alias *fa_match;
+
+ rcu_read_lock();
+
+ fa_match = fib_find_matching_alias(net, dst, dst_len, fi, tos, type,
+ tb_id);
+ if (!fa_match)
+ goto out;
+
+ fa_match->offload = offload;
+ fa_match->trap = trap;
+
+out:
+ rcu_read_unlock();
+}
+EXPORT_SYMBOL_GPL(fib_alias_hw_flags_set);
+
static void trie_rebalance(struct trie *t, struct key_vector *tn)
{
while (!IS_TRIE(tn))
@@ -1220,6 +1270,8 @@ int fib_table_insert(struct net *net, struct fib_table *tb,
new_fa->fa_slen = fa->fa_slen;
new_fa->tb_id = tb->tb_id;
new_fa->fa_default = -1;
+ new_fa->offload = 0;
+ new_fa->trap = 0;
hlist_replace_rcu(&fa->fa_list, &new_fa->fa_list);
@@ -1278,6 +1330,8 @@ int fib_table_insert(struct net *net, struct fib_table *tb,
new_fa->fa_slen = slen;
new_fa->tb_id = tb->tb_id;
new_fa->fa_default = -1;
+ new_fa->offload = 0;
+ new_fa->trap = 0;
/* Insert new entry to the list. */
err = fib_insert_alias(t, tp, l, new_fa, fa, key);
@@ -2202,6 +2256,8 @@ static int fn_trie_dump_leaf(struct key_vector *l, struct fib_table *tb,
fri.dst_len = KEYLENGTH - fa->fa_slen;
fri.tos = fa->fa_tos;
fri.type = fa->fa_type;
+ fri.offload = fa->offload;
+ fri.trap = fa->trap;
err = fib_dump_info(skb,
NETLINK_CB(cb->skb).portid,
cb->nlh->nlmsg_seq,
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 167a7357d12a..2010888e68ca 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -3237,6 +3237,25 @@ static int inet_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh,
fri.dst_len = res.prefixlen;
fri.tos = fl4.flowi4_tos;
fri.type = rt->rt_type;
+ fri.offload = 0;
+ fri.trap = 0;
+ if (res.fa_head) {
+ struct fib_alias *fa;
+
+ hlist_for_each_entry_rcu(fa, res.fa_head, fa_list) {
+ u8 slen = 32 - fri.dst_len;
+
+ if (fa->fa_slen == slen &&
+ fa->tb_id == fri.tb_id &&
+ fa->fa_tos == fri.tos &&
+ fa->fa_info == res.fi &&
+ fa->fa_type == fri.type) {
+ fri.offload = fa->offload;
+ fri.trap = fa->trap;
+ break;
+ }
+ }
+ }
err = fib_dump_info(skb, NETLINK_CB(in_skb).portid,
nlh->nlmsg_seq, RTM_NEWROUTE, &fri, 0);
} else {
--
2.24.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* Re: [PATCH net-next 03/10] ipv4: Add "offload" and "trap" indications to routes
2020-01-07 15:45 ` [PATCH net-next 03/10] ipv4: Add "offload" and "trap" indications to routes Ido Schimmel
@ 2020-01-09 15:47 ` David Ahern
0 siblings, 0 replies; 17+ messages in thread
From: David Ahern @ 2020-01-09 15:47 UTC (permalink / raw)
To: Ido Schimmel, netdev
Cc: davem, jiri, jakub.kicinski, roopa, mlxsw, Ido Schimmel
On 1/7/20 8:45 AM, Ido Schimmel wrote:
> diff --git a/net/ipv4/fib_trie.c b/net/ipv4/fib_trie.c
> index 75af3f8ae50e..272c9f73e4b3 100644
> --- a/net/ipv4/fib_trie.c
> +++ b/net/ipv4/fib_trie.c
> @@ -1012,6 +1012,56 @@ static struct fib_alias *fib_find_alias(struct hlist_head *fah, u8 slen,
> return NULL;
> }
>
> +static struct fib_alias *
> +fib_find_matching_alias(struct net *net, u32 dst, int dst_len,
> + const struct fib_info *fi, u8 tos, u8 type, u32 tb_id)
> +{
> + u8 slen = KEYLENGTH - dst_len;
> + struct key_vector *l, *tp;
> + struct fib_table *tb;
> + struct fib_alias *fa;
> + struct trie *t;
> +
> + tb = fib_get_table(net, tb_id);
> + if (!tb)
> + return NULL;
> +
> + t = (struct trie *)tb->tb_data;
> + l = fib_find_node(t, &tp, dst);
> + if (!l)
> + return NULL;
> +
> + hlist_for_each_entry_rcu(fa, &l->leaf, fa_list) {
> + if (fa->fa_slen == slen && fa->tb_id == tb_id &&
> + fa->fa_tos == tos && fa->fa_info == fi &&
> + fa->fa_type == type)
> + return fa;
> + }
> +
> + return NULL;
> +}
> +
> +void fib_alias_hw_flags_set(struct net *net, u32 dst, int dst_len,
> + const struct fib_info *fi, u8 tos, u8 type,
> + u32 tb_id, bool offload, bool trap)
Seem like struct fib_rt_info from the previous patch be used here and
then passed to fib_find_matching_alias.
> +{
> + struct fib_alias *fa_match;
> +
> + rcu_read_lock();
> +
> + fa_match = fib_find_matching_alias(net, dst, dst_len, fi, tos, type,
> + tb_id);
> + if (!fa_match)
> + goto out;
> +
> + fa_match->offload = offload;
> + fa_match->trap = trap;
> +
> +out:
> + rcu_read_unlock();
> +}
> +EXPORT_SYMBOL_GPL(fib_alias_hw_flags_set);
> +
> static void trie_rebalance(struct trie *t, struct key_vector *tn)
> {
> while (!IS_TRIE(tn))
^ permalink raw reply [flat|nested] 17+ messages in thread
* [PATCH net-next 04/10] ipv6: Add "offload" and "trap" indications to routes
2020-01-07 15:45 [PATCH net-next 00/10] net: Add route offload indication Ido Schimmel
` (2 preceding siblings ...)
2020-01-07 15:45 ` [PATCH net-next 03/10] ipv4: Add "offload" and "trap" indications to routes Ido Schimmel
@ 2020-01-07 15:45 ` Ido Schimmel
2020-01-09 15:48 ` David Ahern
2020-01-09 16:24 ` Roopa Prabhu
2020-01-07 15:45 ` [PATCH net-next 05/10] mlxsw: spectrum_router: Separate nexthop offload indication from route Ido Schimmel
` (5 subsequent siblings)
9 siblings, 2 replies; 17+ messages in thread
From: Ido Schimmel @ 2020-01-07 15:45 UTC (permalink / raw)
To: netdev; +Cc: davem, jiri, jakub.kicinski, dsahern, roopa, mlxsw, Ido Schimmel
From: Ido Schimmel <idosch@mellanox.com>
In a similar fashion to previous patch, add "offload" and "trap"
indication to IPv6 routes.
This is done by using two unused bits in 'struct fib6_info' to hold
these indications. Capable drivers are expected to set these when
processing the various in-kernel route notifications.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
---
include/net/ip6_fib.h | 11 ++++++++++-
net/ipv6/route.c | 7 +++++++
2 files changed, 17 insertions(+), 1 deletion(-)
diff --git a/include/net/ip6_fib.h b/include/net/ip6_fib.h
index b579faea41e9..fd60a8ac02ee 100644
--- a/include/net/ip6_fib.h
+++ b/include/net/ip6_fib.h
@@ -192,7 +192,9 @@ struct fib6_info {
dst_nopolicy:1,
dst_host:1,
fib6_destroying:1,
- unused:3;
+ offload:1,
+ trap:1,
+ unused:1;
struct rcu_head rcu;
struct nexthop *nh;
@@ -329,6 +331,13 @@ static inline void fib6_info_release(struct fib6_info *f6i)
call_rcu(&f6i->rcu, fib6_info_destroy_rcu);
}
+static inline void fib6_info_hw_flags_set(struct fib6_info *f6i, bool offload,
+ bool trap)
+{
+ f6i->offload = offload;
+ f6i->trap = trap;
+}
+
enum fib6_walk_state {
#ifdef CONFIG_IPV6_SUBTREES
FWS_S,
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index 0253b702afb7..4fbdc60b4e07 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -5576,6 +5576,13 @@ static int rt6_fill_node(struct net *net, struct sk_buff *skb,
expires -= jiffies;
}
+ if (!dst) {
+ if (rt->offload)
+ rtm->rtm_flags |= RTM_F_OFFLOAD;
+ if (rt->trap)
+ rtm->rtm_flags |= RTM_F_TRAP;
+ }
+
if (rtnl_put_cacheinfo(skb, dst, 0, expires, dst ? dst->error : 0) < 0)
goto nla_put_failure;
--
2.24.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* Re: [PATCH net-next 04/10] ipv6: Add "offload" and "trap" indications to routes
2020-01-07 15:45 ` [PATCH net-next 04/10] ipv6: " Ido Schimmel
@ 2020-01-09 15:48 ` David Ahern
2020-01-09 16:24 ` Roopa Prabhu
1 sibling, 0 replies; 17+ messages in thread
From: David Ahern @ 2020-01-09 15:48 UTC (permalink / raw)
To: Ido Schimmel, netdev
Cc: davem, jiri, jakub.kicinski, roopa, mlxsw, Ido Schimmel
On 1/7/20 8:45 AM, Ido Schimmel wrote:
> From: Ido Schimmel <idosch@mellanox.com>
>
> In a similar fashion to previous patch, add "offload" and "trap"
> indication to IPv6 routes.
>
> This is done by using two unused bits in 'struct fib6_info' to hold
> these indications. Capable drivers are expected to set these when
> processing the various in-kernel route notifications.
>
> Signed-off-by: Ido Schimmel <idosch@mellanox.com>
> Reviewed-by: Jiri Pirko <jiri@mellanox.com>
> ---
> include/net/ip6_fib.h | 11 ++++++++++-
> net/ipv6/route.c | 7 +++++++
> 2 files changed, 17 insertions(+), 1 deletion(-)
>
Reviewed-by: David Ahern <dsahern@gmail.com>
^ permalink raw reply [flat|nested] 17+ messages in thread
* Re: [PATCH net-next 04/10] ipv6: Add "offload" and "trap" indications to routes
2020-01-07 15:45 ` [PATCH net-next 04/10] ipv6: " Ido Schimmel
2020-01-09 15:48 ` David Ahern
@ 2020-01-09 16:24 ` Roopa Prabhu
1 sibling, 0 replies; 17+ messages in thread
From: Roopa Prabhu @ 2020-01-09 16:24 UTC (permalink / raw)
To: Ido Schimmel
Cc: netdev, David Miller, Jiri Pirko, Jakub Kicinski, David Ahern,
mlxsw, Ido Schimmel
On Tue, Jan 7, 2020 at 7:45 AM Ido Schimmel <idosch@idosch.org> wrote:
>
> From: Ido Schimmel <idosch@mellanox.com>
>
> In a similar fashion to previous patch, add "offload" and "trap"
> indication to IPv6 routes.
>
> This is done by using two unused bits in 'struct fib6_info' to hold
> these indications. Capable drivers are expected to set these when
> processing the various in-kernel route notifications.
>
> Signed-off-by: Ido Schimmel <idosch@mellanox.com>
> Reviewed-by: Jiri Pirko <jiri@mellanox.com>
> ---
Acked-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Nice, Thanks Ido!
^ permalink raw reply [flat|nested] 17+ messages in thread
* [PATCH net-next 05/10] mlxsw: spectrum_router: Separate nexthop offload indication from route
2020-01-07 15:45 [PATCH net-next 00/10] net: Add route offload indication Ido Schimmel
` (3 preceding siblings ...)
2020-01-07 15:45 ` [PATCH net-next 04/10] ipv6: " Ido Schimmel
@ 2020-01-07 15:45 ` Ido Schimmel
2020-01-07 15:45 ` [PATCH net-next 06/10] mlxsw: spectrum_router: Set hardware flags for routes Ido Schimmel
` (4 subsequent siblings)
9 siblings, 0 replies; 17+ messages in thread
From: Ido Schimmel @ 2020-01-07 15:45 UTC (permalink / raw)
To: netdev; +Cc: davem, jiri, jakub.kicinski, dsahern, roopa, mlxsw, Ido Schimmel
From: Ido Schimmel <idosch@mellanox.com>
The driver currently uses the 'RTNH_F_OFFLOAD' flag for both routes and
nexthops, which is cumbersome and unnecessary now that we have separate
flag for the route itself.
Separate the offload indication for nexthops from routes and call it
whenever the offload state within the nexthop group changes.
Note that IPv6 (unlike IPv4) does not share the same nexthop group
between different routes, whereas mlxsw does. Therefore, whenever the
offload indication within an IPv6 nexthop group changes, all the linked
routes need to be updated.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
---
.../ethernet/mellanox/mlxsw/spectrum_router.c | 92 +++++++++++++++----
1 file changed, 75 insertions(+), 17 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index 7f70aa799064..cbc23485f487 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
@@ -3235,20 +3235,6 @@ mlxsw_sp_nexthop_fib_entries_update(struct mlxsw_sp *mlxsw_sp,
return 0;
}
-static void
-mlxsw_sp_fib_entry_offload_refresh(struct mlxsw_sp_fib_entry *fib_entry,
- enum mlxsw_reg_ralue_op op, int err);
-
-static void
-mlxsw_sp_nexthop_fib_entries_refresh(struct mlxsw_sp_nexthop_group *nh_grp)
-{
- enum mlxsw_reg_ralue_op op = MLXSW_REG_RALUE_OP_WRITE_WRITE;
- struct mlxsw_sp_fib_entry *fib_entry;
-
- list_for_each_entry(fib_entry, &nh_grp->fib_list, nexthop_group_node)
- mlxsw_sp_fib_entry_offload_refresh(fib_entry, op, 0);
-}
-
static void mlxsw_sp_adj_grp_size_round_up(u16 *p_adj_grp_size)
{
/* Valid sizes for an adjacency group are:
@@ -3352,6 +3338,73 @@ mlxsw_sp_nexthop_group_rebalance(struct mlxsw_sp_nexthop_group *nh_grp)
}
}
+static struct mlxsw_sp_nexthop *
+mlxsw_sp_rt6_nexthop(struct mlxsw_sp_nexthop_group *nh_grp,
+ const struct mlxsw_sp_rt6 *mlxsw_sp_rt6);
+
+static void
+mlxsw_sp_nexthop4_group_offload_refresh(struct mlxsw_sp *mlxsw_sp,
+ struct mlxsw_sp_nexthop_group *nh_grp)
+{
+ int i;
+
+ for (i = 0; i < nh_grp->count; i++) {
+ struct mlxsw_sp_nexthop *nh = &nh_grp->nexthops[i];
+
+ if (nh->offloaded)
+ nh->key.fib_nh->fib_nh_flags |= RTNH_F_OFFLOAD;
+ else
+ nh->key.fib_nh->fib_nh_flags &= ~RTNH_F_OFFLOAD;
+ }
+}
+
+static void
+__mlxsw_sp_nexthop6_group_offload_refresh(struct mlxsw_sp_nexthop_group *nh_grp,
+ struct mlxsw_sp_fib6_entry *fib6_entry)
+{
+ struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
+
+ list_for_each_entry(mlxsw_sp_rt6, &fib6_entry->rt6_list, list) {
+ struct fib6_nh *fib6_nh = mlxsw_sp_rt6->rt->fib6_nh;
+ struct mlxsw_sp_nexthop *nh;
+
+ nh = mlxsw_sp_rt6_nexthop(nh_grp, mlxsw_sp_rt6);
+ if (nh && nh->offloaded)
+ fib6_nh->fib_nh_flags |= RTNH_F_OFFLOAD;
+ else
+ fib6_nh->fib_nh_flags &= ~RTNH_F_OFFLOAD;
+ }
+}
+
+static void
+mlxsw_sp_nexthop6_group_offload_refresh(struct mlxsw_sp *mlxsw_sp,
+ struct mlxsw_sp_nexthop_group *nh_grp)
+{
+ struct mlxsw_sp_fib6_entry *fib6_entry;
+
+ /* Unfortunately, in IPv6 the route and the nexthop are described by
+ * the same struct, so we need to iterate over all the routes using the
+ * nexthop group and set / clear the offload indication for them.
+ */
+ list_for_each_entry(fib6_entry, &nh_grp->fib_list,
+ common.nexthop_group_node)
+ __mlxsw_sp_nexthop6_group_offload_refresh(nh_grp, fib6_entry);
+}
+
+static void
+mlxsw_sp_nexthop_group_offload_refresh(struct mlxsw_sp *mlxsw_sp,
+ struct mlxsw_sp_nexthop_group *nh_grp)
+{
+ switch (mlxsw_sp_nexthop_group_type(nh_grp)) {
+ case AF_INET:
+ mlxsw_sp_nexthop4_group_offload_refresh(mlxsw_sp, nh_grp);
+ break;
+ case AF_INET6:
+ mlxsw_sp_nexthop6_group_offload_refresh(mlxsw_sp, nh_grp);
+ break;
+ }
+}
+
static void
mlxsw_sp_nexthop_group_refresh(struct mlxsw_sp *mlxsw_sp,
struct mlxsw_sp_nexthop_group *nh_grp)
@@ -3425,6 +3478,8 @@ mlxsw_sp_nexthop_group_refresh(struct mlxsw_sp *mlxsw_sp,
goto set_trap;
}
+ mlxsw_sp_nexthop_group_offload_refresh(mlxsw_sp, nh_grp);
+
if (!old_adj_index_valid) {
/* The trap was set for fib entries, so we have to call
* fib entry update to unset it and use adjacency index.
@@ -3446,9 +3501,6 @@ mlxsw_sp_nexthop_group_refresh(struct mlxsw_sp *mlxsw_sp,
goto set_trap;
}
- /* Offload state within the group changed, so update the flags. */
- mlxsw_sp_nexthop_fib_entries_refresh(nh_grp);
-
return;
set_trap:
@@ -3461,6 +3513,7 @@ mlxsw_sp_nexthop_group_refresh(struct mlxsw_sp *mlxsw_sp,
err = mlxsw_sp_nexthop_fib_entries_update(mlxsw_sp, nh_grp);
if (err)
dev_warn(mlxsw_sp->bus_info->dev, "Failed to set traps for fib entries.\n");
+ mlxsw_sp_nexthop_group_offload_refresh(mlxsw_sp, nh_grp);
if (old_adj_index_valid)
mlxsw_sp_kvdl_free(mlxsw_sp, MLXSW_SP_KVDL_ENTRY_TYPE_ADJ,
nh_grp->ecmp_size, nh_grp->adj_index);
@@ -5113,6 +5166,11 @@ static int mlxsw_sp_nexthop6_group_get(struct mlxsw_sp *mlxsw_sp,
&nh_grp->fib_list);
fib6_entry->common.nh_group = nh_grp;
+ /* The route and the nexthop are described by the same struct, so we
+ * need to the update the nexthop offload indication for the new route.
+ */
+ __mlxsw_sp_nexthop6_group_offload_refresh(nh_grp, fib6_entry);
+
return 0;
}
--
2.24.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH net-next 06/10] mlxsw: spectrum_router: Set hardware flags for routes
2020-01-07 15:45 [PATCH net-next 00/10] net: Add route offload indication Ido Schimmel
` (4 preceding siblings ...)
2020-01-07 15:45 ` [PATCH net-next 05/10] mlxsw: spectrum_router: Separate nexthop offload indication from route Ido Schimmel
@ 2020-01-07 15:45 ` Ido Schimmel
2020-01-07 15:45 ` [PATCH net-next 07/10] netdevsim: fib: Add dummy implementation for FIB offload Ido Schimmel
` (3 subsequent siblings)
9 siblings, 0 replies; 17+ messages in thread
From: Ido Schimmel @ 2020-01-07 15:45 UTC (permalink / raw)
To: netdev; +Cc: davem, jiri, jakub.kicinski, dsahern, roopa, mlxsw, Ido Schimmel
From: Ido Schimmel <idosch@mellanox.com>
Previous patches added support for two hardware flags for IPv4 and IPv6
routes: 'RTM_F_OFFLOAD' and 'RTM_F_TRAP'. Both indicate the presence of
the route in hardware. The first indicates that traffic is actually
offloaded from the kernel, whereas the second indicates that packets
hitting such routes are trapped to the kernel for processing (e.g., host
routes).
Use these two flags in mlxsw. The flags are modified in two places.
Firstly, whenever a route is updated in the device's table. This
includes the addition, deletion or update of a route. For example, when
a host route is promoted to perform NVE decapsulation, its action in the
device is updated, the 'RTM_F_OFFLOAD' flag set and the 'RTM_F_TRAP'
flag cleared.
Secondly, when a route is replaced and overwritten by another route, its
flags are cleared.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
---
.../ethernet/mellanox/mlxsw/spectrum_router.c | 141 ++++++++----------
1 file changed, 64 insertions(+), 77 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
index cbc23485f487..7c0e247ab7b6 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c
@@ -4096,131 +4096,115 @@ mlxsw_sp_rt6_nexthop(struct mlxsw_sp_nexthop_group *nh_grp,
}
static void
-mlxsw_sp_fib4_entry_offload_set(struct mlxsw_sp_fib_entry *fib_entry)
+mlxsw_sp_fib4_entry_hw_flags_set(struct mlxsw_sp *mlxsw_sp,
+ struct mlxsw_sp_fib_entry *fib_entry)
{
- struct mlxsw_sp_nexthop_group *nh_grp = fib_entry->nh_group;
- int i;
-
- if (fib_entry->type == MLXSW_SP_FIB_ENTRY_TYPE_LOCAL ||
- fib_entry->type == MLXSW_SP_FIB_ENTRY_TYPE_BLACKHOLE ||
- fib_entry->type == MLXSW_SP_FIB_ENTRY_TYPE_IPIP_DECAP ||
- fib_entry->type == MLXSW_SP_FIB_ENTRY_TYPE_NVE_DECAP) {
- nh_grp->nexthops->key.fib_nh->fib_nh_flags |= RTNH_F_OFFLOAD;
- return;
- }
-
- for (i = 0; i < nh_grp->count; i++) {
- struct mlxsw_sp_nexthop *nh = &nh_grp->nexthops[i];
+ struct fib_info *fi = mlxsw_sp_nexthop4_group_fi(fib_entry->nh_group);
+ u32 *p_dst = (u32 *) fib_entry->fib_node->key.addr;
+ int dst_len = fib_entry->fib_node->key.prefix_len;
+ struct mlxsw_sp_fib4_entry *fib4_entry;
+ bool should_offload;
- if (nh->offloaded)
- nh->key.fib_nh->fib_nh_flags |= RTNH_F_OFFLOAD;
- else
- nh->key.fib_nh->fib_nh_flags &= ~RTNH_F_OFFLOAD;
- }
+ should_offload = mlxsw_sp_fib_entry_should_offload(fib_entry);
+ fib4_entry = container_of(fib_entry, struct mlxsw_sp_fib4_entry,
+ common);
+ fib_alias_hw_flags_set(mlxsw_sp_net(mlxsw_sp), *p_dst, dst_len,
+ fi, fib4_entry->tos, fib4_entry->type,
+ fib4_entry->tb_id, should_offload,
+ !should_offload);
}
static void
-mlxsw_sp_fib4_entry_offload_unset(struct mlxsw_sp_fib_entry *fib_entry)
+mlxsw_sp_fib4_entry_hw_flags_clear(struct mlxsw_sp *mlxsw_sp,
+ struct mlxsw_sp_fib_entry *fib_entry)
{
- struct mlxsw_sp_nexthop_group *nh_grp = fib_entry->nh_group;
- int i;
-
- if (!list_is_singular(&nh_grp->fib_list))
- return;
-
- for (i = 0; i < nh_grp->count; i++) {
- struct mlxsw_sp_nexthop *nh = &nh_grp->nexthops[i];
+ struct fib_info *fi = mlxsw_sp_nexthop4_group_fi(fib_entry->nh_group);
+ u32 *p_dst = (u32 *) fib_entry->fib_node->key.addr;
+ int dst_len = fib_entry->fib_node->key.prefix_len;
+ struct mlxsw_sp_fib4_entry *fib4_entry;
- nh->key.fib_nh->fib_nh_flags &= ~RTNH_F_OFFLOAD;
- }
+ fib4_entry = container_of(fib_entry, struct mlxsw_sp_fib4_entry,
+ common);
+ fib_alias_hw_flags_set(mlxsw_sp_net(mlxsw_sp), *p_dst, dst_len,
+ fi, fib4_entry->tos, fib4_entry->type,
+ fib4_entry->tb_id, false, false);
}
static void
-mlxsw_sp_fib6_entry_offload_set(struct mlxsw_sp_fib_entry *fib_entry)
+mlxsw_sp_fib6_entry_hw_flags_set(struct mlxsw_sp *mlxsw_sp,
+ struct mlxsw_sp_fib_entry *fib_entry)
{
struct mlxsw_sp_fib6_entry *fib6_entry;
struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
+ bool should_offload;
+
+ should_offload = mlxsw_sp_fib_entry_should_offload(fib_entry);
+ /* In IPv6 a multipath route is represented using multiple routes, so
+ * we need to set the flags on all of them.
+ */
fib6_entry = container_of(fib_entry, struct mlxsw_sp_fib6_entry,
common);
-
- if (fib_entry->type == MLXSW_SP_FIB_ENTRY_TYPE_LOCAL ||
- fib_entry->type == MLXSW_SP_FIB_ENTRY_TYPE_BLACKHOLE) {
- list_first_entry(&fib6_entry->rt6_list, struct mlxsw_sp_rt6,
- list)->rt->fib6_nh->fib_nh_flags |= RTNH_F_OFFLOAD;
- return;
- }
-
- list_for_each_entry(mlxsw_sp_rt6, &fib6_entry->rt6_list, list) {
- struct mlxsw_sp_nexthop_group *nh_grp = fib_entry->nh_group;
- struct fib6_nh *fib6_nh = mlxsw_sp_rt6->rt->fib6_nh;
- struct mlxsw_sp_nexthop *nh;
-
- nh = mlxsw_sp_rt6_nexthop(nh_grp, mlxsw_sp_rt6);
- if (nh && nh->offloaded)
- fib6_nh->fib_nh_flags |= RTNH_F_OFFLOAD;
- else
- fib6_nh->fib_nh_flags &= ~RTNH_F_OFFLOAD;
- }
+ list_for_each_entry(mlxsw_sp_rt6, &fib6_entry->rt6_list, list)
+ fib6_info_hw_flags_set(mlxsw_sp_rt6->rt, should_offload,
+ !should_offload);
}
static void
-mlxsw_sp_fib6_entry_offload_unset(struct mlxsw_sp_fib_entry *fib_entry)
+mlxsw_sp_fib6_entry_hw_flags_clear(struct mlxsw_sp *mlxsw_sp,
+ struct mlxsw_sp_fib_entry *fib_entry)
{
struct mlxsw_sp_fib6_entry *fib6_entry;
struct mlxsw_sp_rt6 *mlxsw_sp_rt6;
fib6_entry = container_of(fib_entry, struct mlxsw_sp_fib6_entry,
common);
- list_for_each_entry(mlxsw_sp_rt6, &fib6_entry->rt6_list, list) {
- struct fib6_info *rt = mlxsw_sp_rt6->rt;
-
- rt->fib6_nh->fib_nh_flags &= ~RTNH_F_OFFLOAD;
- }
+ list_for_each_entry(mlxsw_sp_rt6, &fib6_entry->rt6_list, list)
+ fib6_info_hw_flags_set(mlxsw_sp_rt6->rt, false, false);
}
-static void mlxsw_sp_fib_entry_offload_set(struct mlxsw_sp_fib_entry *fib_entry)
+static void
+mlxsw_sp_fib_entry_hw_flags_set(struct mlxsw_sp *mlxsw_sp,
+ struct mlxsw_sp_fib_entry *fib_entry)
{
switch (fib_entry->fib_node->fib->proto) {
case MLXSW_SP_L3_PROTO_IPV4:
- mlxsw_sp_fib4_entry_offload_set(fib_entry);
+ mlxsw_sp_fib4_entry_hw_flags_set(mlxsw_sp, fib_entry);
break;
case MLXSW_SP_L3_PROTO_IPV6:
- mlxsw_sp_fib6_entry_offload_set(fib_entry);
+ mlxsw_sp_fib6_entry_hw_flags_set(mlxsw_sp, fib_entry);
break;
}
}
static void
-mlxsw_sp_fib_entry_offload_unset(struct mlxsw_sp_fib_entry *fib_entry)
+mlxsw_sp_fib_entry_hw_flags_clear(struct mlxsw_sp *mlxsw_sp,
+ struct mlxsw_sp_fib_entry *fib_entry)
{
switch (fib_entry->fib_node->fib->proto) {
case MLXSW_SP_L3_PROTO_IPV4:
- mlxsw_sp_fib4_entry_offload_unset(fib_entry);
+ mlxsw_sp_fib4_entry_hw_flags_clear(mlxsw_sp, fib_entry);
break;
case MLXSW_SP_L3_PROTO_IPV6:
- mlxsw_sp_fib6_entry_offload_unset(fib_entry);
+ mlxsw_sp_fib6_entry_hw_flags_clear(mlxsw_sp, fib_entry);
break;
}
}
static void
-mlxsw_sp_fib_entry_offload_refresh(struct mlxsw_sp_fib_entry *fib_entry,
- enum mlxsw_reg_ralue_op op, int err)
+mlxsw_sp_fib_entry_hw_flags_refresh(struct mlxsw_sp *mlxsw_sp,
+ struct mlxsw_sp_fib_entry *fib_entry,
+ enum mlxsw_reg_ralue_op op)
{
switch (op) {
- case MLXSW_REG_RALUE_OP_WRITE_DELETE:
- return mlxsw_sp_fib_entry_offload_unset(fib_entry);
case MLXSW_REG_RALUE_OP_WRITE_WRITE:
- if (err)
- return;
- if (mlxsw_sp_fib_entry_should_offload(fib_entry))
- mlxsw_sp_fib_entry_offload_set(fib_entry);
- else
- mlxsw_sp_fib_entry_offload_unset(fib_entry);
- return;
+ mlxsw_sp_fib_entry_hw_flags_set(mlxsw_sp, fib_entry);
+ break;
+ case MLXSW_REG_RALUE_OP_WRITE_DELETE:
+ mlxsw_sp_fib_entry_hw_flags_clear(mlxsw_sp, fib_entry);
+ break;
default:
- return;
+ break;
}
}
@@ -4447,7 +4431,10 @@ static int mlxsw_sp_fib_entry_op(struct mlxsw_sp *mlxsw_sp,
{
int err = __mlxsw_sp_fib_entry_op(mlxsw_sp, fib_entry, op);
- mlxsw_sp_fib_entry_offload_refresh(fib_entry, op, err);
+ if (err)
+ return err;
+
+ mlxsw_sp_fib_entry_hw_flags_refresh(mlxsw_sp, fib_entry, op);
return err;
}
@@ -4883,7 +4870,7 @@ mlxsw_sp_router_fib4_replace(struct mlxsw_sp *mlxsw_sp,
if (!replaced)
return 0;
- mlxsw_sp_fib_entry_offload_unset(replaced);
+ mlxsw_sp_fib_entry_hw_flags_clear(mlxsw_sp, replaced);
fib4_replaced = container_of(replaced, struct mlxsw_sp_fib4_entry,
common);
mlxsw_sp_fib4_entry_destroy(mlxsw_sp, fib4_replaced);
@@ -5451,7 +5438,7 @@ static int mlxsw_sp_router_fib6_replace(struct mlxsw_sp *mlxsw_sp,
if (!replaced)
return 0;
- mlxsw_sp_fib_entry_offload_unset(replaced);
+ mlxsw_sp_fib_entry_hw_flags_clear(mlxsw_sp, replaced);
fib6_replaced = container_of(replaced, struct mlxsw_sp_fib6_entry,
common);
mlxsw_sp_fib6_entry_destroy(mlxsw_sp, fib6_replaced);
--
2.24.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH net-next 07/10] netdevsim: fib: Add dummy implementation for FIB offload
2020-01-07 15:45 [PATCH net-next 00/10] net: Add route offload indication Ido Schimmel
` (5 preceding siblings ...)
2020-01-07 15:45 ` [PATCH net-next 06/10] mlxsw: spectrum_router: Set hardware flags for routes Ido Schimmel
@ 2020-01-07 15:45 ` Ido Schimmel
2020-01-07 15:45 ` [PATCH net-next 08/10] selftests: forwarding: Add helpers and tests " Ido Schimmel
` (2 subsequent siblings)
9 siblings, 0 replies; 17+ messages in thread
From: Ido Schimmel @ 2020-01-07 15:45 UTC (permalink / raw)
To: netdev; +Cc: davem, jiri, jakub.kicinski, dsahern, roopa, mlxsw, Ido Schimmel
From: Ido Schimmel <idosch@mellanox.com>
Implement dummy IPv4 and IPv6 FIB "offload" in the driver by storing
currently "programmed" routes in a hash table. Each route in the hash
table is marked with "trap" indication. The indication is cleared when
the route is replaced or when the netdevsim instance is deleted.
This will later allow us to test the route offload API on top of
netdevsim.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
---
drivers/net/Kconfig | 1 +
drivers/net/netdevsim/fib.c | 664 +++++++++++++++++++++++++++++++++++-
2 files changed, 656 insertions(+), 9 deletions(-)
diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 01e2657e4c26..e65ed4d0a7ad 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -549,6 +549,7 @@ source "drivers/net/hyperv/Kconfig"
config NETDEVSIM
tristate "Simulated networking device"
depends on DEBUG_FS
+ depends on IPV6 || IPV6=n
select NET_DEVLINK
help
This driver is a developer testing tool and software model that can
diff --git a/drivers/net/netdevsim/fib.c b/drivers/net/netdevsim/fib.c
index b5df308b4e33..82ee44569d3b 100644
--- a/drivers/net/netdevsim/fib.c
+++ b/drivers/net/netdevsim/fib.c
@@ -14,6 +14,12 @@
* THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
*/
+#include <linux/in6.h>
+#include <linux/kernel.h>
+#include <linux/list.h>
+#include <linux/rhashtable.h>
+#include <linux/spinlock_types.h>
+#include <linux/types.h>
#include <net/fib_notifier.h>
#include <net/ip_fib.h>
#include <net/ip6_fib.h>
@@ -36,6 +42,48 @@ struct nsim_fib_data {
struct notifier_block fib_nb;
struct nsim_per_fib_data ipv4;
struct nsim_per_fib_data ipv6;
+ struct rhashtable fib_rt_ht;
+ struct list_head fib_rt_list;
+ spinlock_t fib_lock; /* Protects hashtable, list and accounting */
+ struct devlink *devlink;
+};
+
+struct nsim_fib_rt_key {
+ unsigned char addr[sizeof(struct in6_addr)];
+ unsigned char prefix_len;
+ int family;
+ u32 tb_id;
+};
+
+struct nsim_fib_rt {
+ struct nsim_fib_rt_key key;
+ struct rhash_head ht_node;
+ struct list_head list; /* Member of fib_rt_list */
+};
+
+struct nsim_fib4_rt {
+ struct nsim_fib_rt common;
+ struct fib_info *fi;
+ u8 tos;
+ u8 type;
+};
+
+struct nsim_fib6_rt {
+ struct nsim_fib_rt common;
+ struct list_head nh_list;
+ unsigned int nhs;
+};
+
+struct nsim_fib6_rt_nh {
+ struct list_head list; /* Member of nh_list */
+ struct fib6_info *rt;
+};
+
+static const struct rhashtable_params nsim_fib_rt_ht_params = {
+ .key_offset = offsetof(struct nsim_fib_rt, key),
+ .head_offset = offsetof(struct nsim_fib_rt, ht_node),
+ .key_len = sizeof(struct nsim_fib_rt_key),
+ .automatic_shrinking = true,
};
u64 nsim_fib_get_val(struct nsim_fib_data *fib_data,
@@ -144,18 +192,549 @@ static int nsim_fib_account(struct nsim_fib_entry *entry, bool add,
return err;
}
+static void nsim_fib_rt_init(struct nsim_fib_data *data,
+ struct nsim_fib_rt *fib_rt, const void *addr,
+ size_t addr_len, unsigned int prefix_len,
+ int family, u32 tb_id)
+{
+ memcpy(fib_rt->key.addr, addr, addr_len);
+ fib_rt->key.prefix_len = prefix_len;
+ fib_rt->key.family = family;
+ fib_rt->key.tb_id = tb_id;
+ list_add(&fib_rt->list, &data->fib_rt_list);
+}
+
+static void nsim_fib_rt_fini(struct nsim_fib_rt *fib_rt)
+{
+ list_del(&fib_rt->list);
+}
+
+static struct nsim_fib_rt *nsim_fib_rt_lookup(struct rhashtable *fib_rt_ht,
+ const void *addr, size_t addr_len,
+ unsigned int prefix_len,
+ int family, u32 tb_id)
+{
+ struct nsim_fib_rt_key key;
+
+ memset(&key, 0, sizeof(key));
+ memcpy(key.addr, addr, addr_len);
+ key.prefix_len = prefix_len;
+ key.family = family;
+ key.tb_id = tb_id;
+
+ return rhashtable_lookup_fast(fib_rt_ht, &key, nsim_fib_rt_ht_params);
+}
+
+static struct nsim_fib4_rt *
+nsim_fib4_rt_create(struct nsim_fib_data *data,
+ struct fib_entry_notifier_info *fen_info)
+{
+ struct nsim_fib4_rt *fib4_rt;
+
+ fib4_rt = kzalloc(sizeof(*fib4_rt), GFP_ATOMIC);
+ if (!fib4_rt)
+ return NULL;
+
+ nsim_fib_rt_init(data, &fib4_rt->common, &fen_info->dst, sizeof(u32),
+ fen_info->dst_len, AF_INET, fen_info->tb_id);
+
+ fib4_rt->fi = fen_info->fi;
+ fib_info_hold(fib4_rt->fi);
+ fib4_rt->tos = fen_info->tos;
+ fib4_rt->type = fen_info->type;
+
+ return fib4_rt;
+}
+
+static void nsim_fib4_rt_destroy(struct nsim_fib4_rt *fib4_rt)
+{
+ fib_info_put(fib4_rt->fi);
+ nsim_fib_rt_fini(&fib4_rt->common);
+ kfree(fib4_rt);
+}
+
+static struct nsim_fib4_rt *
+nsim_fib4_rt_lookup(struct rhashtable *fib_rt_ht,
+ const struct fib_entry_notifier_info *fen_info)
+{
+ struct nsim_fib_rt *fib_rt;
+
+ fib_rt = nsim_fib_rt_lookup(fib_rt_ht, &fen_info->dst, sizeof(u32),
+ fen_info->dst_len, AF_INET,
+ fen_info->tb_id);
+ if (!fib_rt)
+ return NULL;
+
+ return container_of(fib_rt, struct nsim_fib4_rt, common);
+}
+
+static void nsim_fib4_rt_hw_flags_set(struct net *net,
+ const struct nsim_fib4_rt *fib4_rt,
+ bool trap)
+{
+ u32 *p_dst = (u32 *) fib4_rt->common.key.addr;
+ int dst_len = fib4_rt->common.key.prefix_len;
+
+ fib_alias_hw_flags_set(net, *p_dst, dst_len, fib4_rt->fi, fib4_rt->tos,
+ fib4_rt->type, fib4_rt->common.key.tb_id,
+ false, trap);
+}
+
+static int nsim_fib4_rt_add(struct nsim_fib_data *data,
+ struct nsim_fib4_rt *fib4_rt,
+ struct netlink_ext_ack *extack)
+{
+ struct net *net = devlink_net(data->devlink);
+ int err;
+
+ err = nsim_fib_account(&data->ipv4.fib, true, extack);
+ if (err)
+ return err;
+
+ err = rhashtable_insert_fast(&data->fib_rt_ht,
+ &fib4_rt->common.ht_node,
+ nsim_fib_rt_ht_params);
+ if (err) {
+ NL_SET_ERR_MSG_MOD(extack, "Failed to insert IPv4 route");
+ goto err_fib_dismiss;
+ }
+
+ nsim_fib4_rt_hw_flags_set(net, fib4_rt, true);
+
+ return 0;
+
+err_fib_dismiss:
+ nsim_fib_account(&data->ipv4.fib, false, extack);
+ return err;
+}
+
+static int nsim_fib4_rt_replace(struct nsim_fib_data *data,
+ struct nsim_fib4_rt *fib4_rt,
+ struct nsim_fib4_rt *fib4_rt_old,
+ struct netlink_ext_ack *extack)
+{
+ struct net *net = devlink_net(data->devlink);
+ int err;
+
+ /* We are replacing a route, so no need to change the accounting. */
+ err = rhashtable_replace_fast(&data->fib_rt_ht,
+ &fib4_rt_old->common.ht_node,
+ &fib4_rt->common.ht_node,
+ nsim_fib_rt_ht_params);
+ if (err) {
+ NL_SET_ERR_MSG_MOD(extack, "Failed to replace IPv4 route");
+ return err;
+ }
+
+ nsim_fib4_rt_hw_flags_set(net, fib4_rt, true);
+
+ nsim_fib4_rt_hw_flags_set(net, fib4_rt_old, false);
+ nsim_fib4_rt_destroy(fib4_rt_old);
+
+ return 0;
+}
+
+static int nsim_fib4_rt_insert(struct nsim_fib_data *data,
+ struct fib_entry_notifier_info *fen_info)
+{
+ struct netlink_ext_ack *extack = fen_info->info.extack;
+ struct nsim_fib4_rt *fib4_rt, *fib4_rt_old;
+ int err;
+
+ fib4_rt = nsim_fib4_rt_create(data, fen_info);
+ if (!fib4_rt)
+ return -ENOMEM;
+
+ fib4_rt_old = nsim_fib4_rt_lookup(&data->fib_rt_ht, fen_info);
+ if (!fib4_rt_old)
+ err = nsim_fib4_rt_add(data, fib4_rt, extack);
+ else
+ err = nsim_fib4_rt_replace(data, fib4_rt, fib4_rt_old, extack);
+
+ if (err)
+ nsim_fib4_rt_destroy(fib4_rt);
+
+ return err;
+}
+
+static void nsim_fib4_rt_remove(struct nsim_fib_data *data,
+ const struct fib_entry_notifier_info *fen_info)
+{
+ struct netlink_ext_ack *extack = fen_info->info.extack;
+ struct nsim_fib4_rt *fib4_rt;
+
+ fib4_rt = nsim_fib4_rt_lookup(&data->fib_rt_ht, fen_info);
+ if (WARN_ON_ONCE(!fib4_rt))
+ return;
+
+ rhashtable_remove_fast(&data->fib_rt_ht, &fib4_rt->common.ht_node,
+ nsim_fib_rt_ht_params);
+ nsim_fib_account(&data->ipv4.fib, false, extack);
+ nsim_fib4_rt_destroy(fib4_rt);
+}
+
+static int nsim_fib4_event(struct nsim_fib_data *data,
+ struct fib_notifier_info *info,
+ unsigned long event)
+{
+ struct fib_entry_notifier_info *fen_info;
+ int err = 0;
+
+ fen_info = container_of(info, struct fib_entry_notifier_info, info);
+
+ if (fen_info->fi->nh) {
+ NL_SET_ERR_MSG_MOD(info->extack, "IPv4 route with nexthop objects is not supported");
+ return 0;
+ }
+
+ switch (event) {
+ case FIB_EVENT_ENTRY_REPLACE:
+ err = nsim_fib4_rt_insert(data, fen_info);
+ break;
+ case FIB_EVENT_ENTRY_DEL:
+ nsim_fib4_rt_remove(data, fen_info);
+ break;
+ default:
+ break;
+ }
+
+ return err;
+}
+
+static struct nsim_fib6_rt_nh *
+nsim_fib6_rt_nh_find(const struct nsim_fib6_rt *fib6_rt,
+ const struct fib6_info *rt)
+{
+ struct nsim_fib6_rt_nh *fib6_rt_nh;
+
+ list_for_each_entry(fib6_rt_nh, &fib6_rt->nh_list, list) {
+ if (fib6_rt_nh->rt == rt)
+ return fib6_rt_nh;
+ }
+
+ return NULL;
+}
+
+static int nsim_fib6_rt_nh_add(struct nsim_fib6_rt *fib6_rt,
+ struct fib6_info *rt)
+{
+ struct nsim_fib6_rt_nh *fib6_rt_nh;
+
+ fib6_rt_nh = kzalloc(sizeof(*fib6_rt_nh), GFP_ATOMIC);
+ if (!fib6_rt_nh)
+ return -ENOMEM;
+
+ fib6_info_hold(rt);
+ fib6_rt_nh->rt = rt;
+ list_add_tail(&fib6_rt_nh->list, &fib6_rt->nh_list);
+ fib6_rt->nhs++;
+
+ return 0;
+}
+
+static void nsim_fib6_rt_nh_del(struct nsim_fib6_rt *fib6_rt,
+ const struct fib6_info *rt)
+{
+ struct nsim_fib6_rt_nh *fib6_rt_nh;
+
+ fib6_rt_nh = nsim_fib6_rt_nh_find(fib6_rt, rt);
+ if (WARN_ON_ONCE(!fib6_rt_nh))
+ return;
+
+ fib6_rt->nhs--;
+ list_del(&fib6_rt_nh->list);
+#if IS_ENABLED(CONFIG_IPV6)
+ fib6_info_release(fib6_rt_nh->rt);
+#endif
+ kfree(fib6_rt_nh);
+}
+
+static struct nsim_fib6_rt *
+nsim_fib6_rt_create(struct nsim_fib_data *data,
+ struct fib6_entry_notifier_info *fen6_info)
+{
+ struct fib6_info *iter, *rt = fen6_info->rt;
+ struct nsim_fib6_rt *fib6_rt;
+ int i = 0;
+ int err;
+
+ fib6_rt = kzalloc(sizeof(*fib6_rt), GFP_ATOMIC);
+ if (!fib6_rt)
+ return NULL;
+
+ nsim_fib_rt_init(data, &fib6_rt->common, &rt->fib6_dst.addr,
+ sizeof(rt->fib6_dst.addr), rt->fib6_dst.plen, AF_INET6,
+ rt->fib6_table->tb6_id);
+
+ /* We consider a multipath IPv6 route as one entry, but it can be made
+ * up from several fib6_info structs (one for each nexthop), so we
+ * add them all to the same list under the entry.
+ */
+ INIT_LIST_HEAD(&fib6_rt->nh_list);
+
+ err = nsim_fib6_rt_nh_add(fib6_rt, rt);
+ if (err)
+ goto err_fib_rt_fini;
+
+ if (!fen6_info->nsiblings)
+ return fib6_rt;
+
+ list_for_each_entry(iter, &rt->fib6_siblings, fib6_siblings) {
+ if (i == fen6_info->nsiblings)
+ break;
+
+ err = nsim_fib6_rt_nh_add(fib6_rt, iter);
+ if (err)
+ goto err_fib6_rt_nh_del;
+ i++;
+ }
+ WARN_ON_ONCE(i != fen6_info->nsiblings);
+
+ return fib6_rt;
+
+err_fib6_rt_nh_del:
+ list_for_each_entry_continue_reverse(iter, &rt->fib6_siblings,
+ fib6_siblings)
+ nsim_fib6_rt_nh_del(fib6_rt, iter);
+ nsim_fib6_rt_nh_del(fib6_rt, rt);
+err_fib_rt_fini:
+ nsim_fib_rt_fini(&fib6_rt->common);
+ kfree(fib6_rt);
+ return ERR_PTR(err);
+}
+
+static void nsim_fib6_rt_destroy(struct nsim_fib6_rt *fib6_rt)
+{
+ struct nsim_fib6_rt_nh *iter, *tmp;
+
+ list_for_each_entry_safe(iter, tmp, &fib6_rt->nh_list, list)
+ nsim_fib6_rt_nh_del(fib6_rt, iter->rt);
+ WARN_ON_ONCE(!list_empty(&fib6_rt->nh_list));
+ nsim_fib_rt_fini(&fib6_rt->common);
+ kfree(fib6_rt);
+}
+
+static struct nsim_fib6_rt *
+nsim_fib6_rt_lookup(struct rhashtable *fib_rt_ht, const struct fib6_info *rt)
+{
+ struct nsim_fib_rt *fib_rt;
+
+ fib_rt = nsim_fib_rt_lookup(fib_rt_ht, &rt->fib6_dst.addr,
+ sizeof(rt->fib6_dst.addr),
+ rt->fib6_dst.plen, AF_INET6,
+ rt->fib6_table->tb6_id);
+ if (!fib_rt)
+ return NULL;
+
+ return container_of(fib_rt, struct nsim_fib6_rt, common);
+}
+
+static int nsim_fib6_rt_append(struct nsim_fib_data *data,
+ struct fib6_entry_notifier_info *fen6_info)
+{
+ struct fib6_info *iter, *rt = fen6_info->rt;
+ struct nsim_fib6_rt *fib6_rt;
+ int i = 0;
+ int err;
+
+ fib6_rt = nsim_fib6_rt_lookup(&data->fib_rt_ht, rt);
+ if (WARN_ON_ONCE(!fib6_rt))
+ return -EINVAL;
+
+ err = nsim_fib6_rt_nh_add(fib6_rt, rt);
+ if (err)
+ return err;
+ rt->trap = true;
+
+ if (!fen6_info->nsiblings)
+ return 0;
+
+ list_for_each_entry(iter, &rt->fib6_siblings, fib6_siblings) {
+ if (i == fen6_info->nsiblings)
+ break;
+
+ err = nsim_fib6_rt_nh_add(fib6_rt, iter);
+ if (err)
+ goto err_fib6_rt_nh_del;
+ iter->trap = true;
+ i++;
+ }
+ WARN_ON_ONCE(i != fen6_info->nsiblings);
+
+ return 0;
+
+err_fib6_rt_nh_del:
+ list_for_each_entry_continue_reverse(iter, &rt->fib6_siblings,
+ fib6_siblings) {
+ iter->trap = false;
+ nsim_fib6_rt_nh_del(fib6_rt, iter);
+ }
+ rt->trap = false;
+ nsim_fib6_rt_nh_del(fib6_rt, rt);
+ return err;
+}
+
+static void nsim_fib6_rt_hw_flags_set(const struct nsim_fib6_rt *fib6_rt,
+ bool trap)
+{
+ struct nsim_fib6_rt_nh *fib6_rt_nh;
+
+ list_for_each_entry(fib6_rt_nh, &fib6_rt->nh_list, list)
+ fib6_info_hw_flags_set(fib6_rt_nh->rt, false, trap);
+}
+
+static int nsim_fib6_rt_add(struct nsim_fib_data *data,
+ struct nsim_fib6_rt *fib6_rt,
+ struct netlink_ext_ack *extack)
+{
+ int err;
+
+ err = nsim_fib_account(&data->ipv6.fib, true, extack);
+ if (err)
+ return err;
+
+ err = rhashtable_insert_fast(&data->fib_rt_ht,
+ &fib6_rt->common.ht_node,
+ nsim_fib_rt_ht_params);
+ if (err) {
+ NL_SET_ERR_MSG_MOD(extack, "Failed to insert IPv6 route");
+ goto err_fib_dismiss;
+ }
+
+ nsim_fib6_rt_hw_flags_set(fib6_rt, true);
+
+ return 0;
+
+err_fib_dismiss:
+ nsim_fib_account(&data->ipv6.fib, false, extack);
+ return err;
+}
+
+static int nsim_fib6_rt_replace(struct nsim_fib_data *data,
+ struct nsim_fib6_rt *fib6_rt,
+ struct nsim_fib6_rt *fib6_rt_old,
+ struct netlink_ext_ack *extack)
+{
+ int err;
+
+ /* We are replacing a route, so no need to change the accounting. */
+ err = rhashtable_replace_fast(&data->fib_rt_ht,
+ &fib6_rt_old->common.ht_node,
+ &fib6_rt->common.ht_node,
+ nsim_fib_rt_ht_params);
+ if (err) {
+ NL_SET_ERR_MSG_MOD(extack, "Failed to replace IPv6 route");
+ return err;
+ }
+
+ nsim_fib6_rt_hw_flags_set(fib6_rt, true);
+
+ nsim_fib6_rt_hw_flags_set(fib6_rt_old, false);
+ nsim_fib6_rt_destroy(fib6_rt_old);
+
+ return 0;
+}
+
+static int nsim_fib6_rt_insert(struct nsim_fib_data *data,
+ struct fib6_entry_notifier_info *fen6_info)
+{
+ struct netlink_ext_ack *extack = fen6_info->info.extack;
+ struct nsim_fib6_rt *fib6_rt, *fib6_rt_old;
+ int err;
+
+ fib6_rt = nsim_fib6_rt_create(data, fen6_info);
+ if (!fib6_rt)
+ return -ENOMEM;
+
+ fib6_rt_old = nsim_fib6_rt_lookup(&data->fib_rt_ht, fen6_info->rt);
+ if (!fib6_rt_old)
+ err = nsim_fib6_rt_add(data, fib6_rt, extack);
+ else
+ err = nsim_fib6_rt_replace(data, fib6_rt, fib6_rt_old, extack);
+
+ if (err)
+ nsim_fib6_rt_destroy(fib6_rt);
+
+ return err;
+}
+
+static void
+nsim_fib6_rt_remove(struct nsim_fib_data *data,
+ const struct fib6_entry_notifier_info *fen6_info)
+{
+ struct netlink_ext_ack *extack = fen6_info->info.extack;
+ struct nsim_fib6_rt *fib6_rt;
+
+ /* Multipath routes are first added to the FIB trie and only then
+ * notified. If we vetoed the addition, we will get a delete
+ * notification for a route we do not have. Therefore, do not warn if
+ * route was not found.
+ */
+ fib6_rt = nsim_fib6_rt_lookup(&data->fib_rt_ht, fen6_info->rt);
+ if (!fib6_rt)
+ return;
+
+ /* If not all the nexthops are deleted, then only reduce the nexthop
+ * group.
+ */
+ if (fen6_info->nsiblings + 1 != fib6_rt->nhs) {
+ nsim_fib6_rt_nh_del(fib6_rt, fen6_info->rt);
+ return;
+ }
+
+ rhashtable_remove_fast(&data->fib_rt_ht, &fib6_rt->common.ht_node,
+ nsim_fib_rt_ht_params);
+ nsim_fib_account(&data->ipv6.fib, false, extack);
+ nsim_fib6_rt_destroy(fib6_rt);
+}
+
+static int nsim_fib6_event(struct nsim_fib_data *data,
+ struct fib_notifier_info *info,
+ unsigned long event)
+{
+ struct fib6_entry_notifier_info *fen6_info;
+ int err = 0;
+
+ fen6_info = container_of(info, struct fib6_entry_notifier_info, info);
+
+ if (fen6_info->rt->nh) {
+ NL_SET_ERR_MSG_MOD(info->extack, "IPv6 route with nexthop objects is not supported");
+ return 0;
+ }
+
+ if (fen6_info->rt->fib6_src.plen) {
+ NL_SET_ERR_MSG_MOD(info->extack, "IPv6 source-specific route is not supported");
+ return 0;
+ }
+
+ switch (event) {
+ case FIB_EVENT_ENTRY_REPLACE:
+ err = nsim_fib6_rt_insert(data, fen6_info);
+ break;
+ case FIB_EVENT_ENTRY_APPEND:
+ err = nsim_fib6_rt_append(data, fen6_info);
+ break;
+ case FIB_EVENT_ENTRY_DEL:
+ nsim_fib6_rt_remove(data, fen6_info);
+ break;
+ default:
+ break;
+ }
+
+ return err;
+}
+
static int nsim_fib_event(struct nsim_fib_data *data,
- struct fib_notifier_info *info, bool add)
+ struct fib_notifier_info *info, unsigned long event)
{
- struct netlink_ext_ack *extack = info->extack;
int err = 0;
switch (info->family) {
case AF_INET:
- err = nsim_fib_account(&data->ipv4.fib, add, extack);
+ err = nsim_fib4_event(data, info, event);
break;
case AF_INET6:
- err = nsim_fib_account(&data->ipv6.fib, add, extack);
+ err = nsim_fib6_event(data, info, event);
break;
}
@@ -170,6 +749,9 @@ static int nsim_fib_event_nb(struct notifier_block *nb, unsigned long event,
struct fib_notifier_info *info = ptr;
int err = 0;
+ /* IPv6 routes can be added via RAs from softIRQ. */
+ spin_lock_bh(&data->fib_lock);
+
switch (event) {
case FIB_EVENT_RULE_ADD: /* fall through */
case FIB_EVENT_RULE_DEL:
@@ -178,23 +760,74 @@ static int nsim_fib_event_nb(struct notifier_block *nb, unsigned long event,
break;
case FIB_EVENT_ENTRY_REPLACE: /* fall through */
+ case FIB_EVENT_ENTRY_APPEND: /* fall through */
case FIB_EVENT_ENTRY_DEL:
- err = nsim_fib_event(data, info, event != FIB_EVENT_ENTRY_DEL);
+ err = nsim_fib_event(data, info, event);
break;
}
+ spin_unlock_bh(&data->fib_lock);
+
return notifier_from_errno(err);
}
+static void nsim_fib4_rt_free(struct nsim_fib_rt *fib_rt,
+ struct nsim_fib_data *data)
+{
+ struct devlink *devlink = data->devlink;
+ struct nsim_fib4_rt *fib4_rt;
+
+ fib4_rt = container_of(fib_rt, struct nsim_fib4_rt, common);
+ nsim_fib4_rt_hw_flags_set(devlink_net(devlink), fib4_rt, false);
+ nsim_fib_account(&data->ipv4.fib, false, NULL);
+ nsim_fib4_rt_destroy(fib4_rt);
+}
+
+static void nsim_fib6_rt_free(struct nsim_fib_rt *fib_rt,
+ struct nsim_fib_data *data)
+{
+ struct nsim_fib6_rt *fib6_rt;
+
+ fib6_rt = container_of(fib_rt, struct nsim_fib6_rt, common);
+ nsim_fib6_rt_hw_flags_set(fib6_rt, false);
+ nsim_fib_account(&data->ipv6.fib, false, NULL);
+ nsim_fib6_rt_destroy(fib6_rt);
+}
+
+static void nsim_fib_rt_free(void *ptr, void *arg)
+{
+ struct nsim_fib_rt *fib_rt = ptr;
+ struct nsim_fib_data *data = arg;
+
+ switch (fib_rt->key.family) {
+ case AF_INET:
+ nsim_fib4_rt_free(fib_rt, data);
+ break;
+ case AF_INET6:
+ nsim_fib6_rt_free(fib_rt, data);
+ break;
+ default:
+ WARN_ON_ONCE(1);
+ }
+}
+
/* inconsistent dump, trying again */
static void nsim_fib_dump_inconsistent(struct notifier_block *nb)
{
struct nsim_fib_data *data = container_of(nb, struct nsim_fib_data,
fib_nb);
+ struct nsim_fib_rt *fib_rt, *fib_rt_tmp;
+
+ /* The notifier block is still not registered, so we do not need to
+ * take any locks here.
+ */
+ list_for_each_entry_safe(fib_rt, fib_rt_tmp, &data->fib_rt_list, list) {
+ rhashtable_remove_fast(&data->fib_rt_ht, &fib_rt->ht_node,
+ nsim_fib_rt_ht_params);
+ nsim_fib_rt_free(fib_rt, data);
+ }
- data->ipv4.fib.num = 0ULL;
data->ipv4.rules.num = 0ULL;
- data->ipv6.fib.num = 0ULL;
data->ipv6.rules.num = 0ULL;
}
@@ -255,6 +888,13 @@ struct nsim_fib_data *nsim_fib_create(struct devlink *devlink,
data = kzalloc(sizeof(*data), GFP_KERNEL);
if (!data)
return ERR_PTR(-ENOMEM);
+ data->devlink = devlink;
+
+ spin_lock_init(&data->fib_lock);
+ INIT_LIST_HEAD(&data->fib_rt_list);
+ err = rhashtable_init(&data->fib_rt_ht, &nsim_fib_rt_ht_params);
+ if (err)
+ goto err_data_free;
nsim_fib_set_max_all(data, devlink);
@@ -263,7 +903,7 @@ struct nsim_fib_data *nsim_fib_create(struct devlink *devlink,
nsim_fib_dump_inconsistent, extack);
if (err) {
pr_err("Failed to register fib notifier\n");
- goto err_out;
+ goto err_rhashtable_destroy;
}
devlink_resource_occ_get_register(devlink,
@@ -284,7 +924,10 @@ struct nsim_fib_data *nsim_fib_create(struct devlink *devlink,
data);
return data;
-err_out:
+err_rhashtable_destroy:
+ rhashtable_free_and_destroy(&data->fib_rt_ht, nsim_fib_rt_free,
+ data);
+err_data_free:
kfree(data);
return ERR_PTR(err);
}
@@ -300,5 +943,8 @@ void nsim_fib_destroy(struct devlink *devlink, struct nsim_fib_data *data)
devlink_resource_occ_get_unregister(devlink,
NSIM_RESOURCE_IPV4_FIB);
unregister_fib_notifier(devlink_net(devlink), &data->fib_nb);
+ rhashtable_free_and_destroy(&data->fib_rt_ht, nsim_fib_rt_free,
+ data);
+ WARN_ON_ONCE(!list_empty(&data->fib_rt_list));
kfree(data);
}
--
2.24.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH net-next 08/10] selftests: forwarding: Add helpers and tests for FIB offload
2020-01-07 15:45 [PATCH net-next 00/10] net: Add route offload indication Ido Schimmel
` (6 preceding siblings ...)
2020-01-07 15:45 ` [PATCH net-next 07/10] netdevsim: fib: Add dummy implementation for FIB offload Ido Schimmel
@ 2020-01-07 15:45 ` Ido Schimmel
2020-01-07 15:45 ` [PATCH net-next 09/10] selftests: netdevsim: Add test for FIB offload API Ido Schimmel
2020-01-07 15:45 ` [PATCH net-next 10/10] selftests: mlxsw: " Ido Schimmel
9 siblings, 0 replies; 17+ messages in thread
From: Ido Schimmel @ 2020-01-07 15:45 UTC (permalink / raw)
To: netdev; +Cc: davem, jiri, jakub.kicinski, dsahern, roopa, mlxsw, Ido Schimmel
From: Ido Schimmel <idosch@mellanox.com>
Implement a set of common helpers and tests for FIB offload that can be
used by multiple drivers to check their FIB offload implementations.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
---
.../net/forwarding/fib_offload_lib.sh | 873 ++++++++++++++++++
1 file changed, 873 insertions(+)
create mode 100644 tools/testing/selftests/net/forwarding/fib_offload_lib.sh
diff --git a/tools/testing/selftests/net/forwarding/fib_offload_lib.sh b/tools/testing/selftests/net/forwarding/fib_offload_lib.sh
new file mode 100644
index 000000000000..66496659bea7
--- /dev/null
+++ b/tools/testing/selftests/net/forwarding/fib_offload_lib.sh
@@ -0,0 +1,873 @@
+# SPDX-License-Identifier: GPL-2.0
+#
+# Various helpers and tests to verify FIB offload.
+
+__fib_trap_check()
+{
+ local ns=$1; shift
+ local family=$1; shift
+ local route=$1; shift
+ local should_fail=$1; shift
+ local ret
+
+ ip -n $ns -j -p -$family route show $route \
+ | jq -e '.[]["flags"] | contains(["trap"])' &> /dev/null
+ ret=$?
+ if [[ $should_fail == "true" ]]; then
+ if [[ $ret -ne 0 ]]; then
+ return 0
+ else
+ return 1
+ fi
+ fi
+
+ return $ret
+}
+
+fib_trap_check()
+{
+ local ns=$1; shift
+ local family=$1; shift
+ local route=$1; shift
+ local should_fail=$1; shift
+
+ busywait 5000 __fib_trap_check $ns $family "$route" $should_fail
+}
+
+fib4_trap_check()
+{
+ local ns=$1; shift
+ local route=$1; shift
+ local should_fail=$1; shift
+
+ fib_trap_check $ns 4 "$route" $should_fail
+}
+
+fib6_trap_check()
+{
+ local ns=$1; shift
+ local route=$1; shift
+ local should_fail=$1; shift
+
+ fib_trap_check $ns 6 "$route" $should_fail
+}
+
+fib_ipv4_identical_routes_test()
+{
+ local ns=$1; shift
+ local i
+
+ RET=0
+
+ for i in $(seq 1 3); do
+ ip -n $ns link add name dummy$i type dummy
+ ip -n $ns link set dev dummy$i up
+ done
+
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 tos 0 metric 1024
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 tos 0 metric 1024" false
+ check_err $? "Route not in hardware when should"
+
+ ip -n $ns route append 192.0.2.0/24 dev dummy2 tos 0 metric 1024
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy2 tos 0 metric 1024" true
+ check_err $? "Appended route in hardware when should not"
+
+ ip -n $ns route prepend 192.0.2.0/24 dev dummy3 tos 0 metric 1024
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy3 tos 0 metric 1024" false
+ check_err $? "Prepended route not in hardware when should"
+
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 tos 0 metric 1024" true
+ check_err $? "Route was not replaced in hardware by prepended one"
+
+ log_test "IPv4 identical routes"
+
+ for i in $(seq 1 3); do
+ ip -n $ns link del dev dummy$i
+ done
+}
+
+fib_ipv4_tos_test()
+{
+ local ns=$1; shift
+
+ RET=0
+
+ ip -n $ns link add name dummy1 type dummy
+ ip -n $ns link set dev dummy1 up
+
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 tos 0 metric 1024
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 tos 0 metric 1024" false
+ check_err $? "Route not in hardware when should"
+
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 tos 2 metric 1024
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 tos 2 metric 1024" false
+ check_err $? "Highest TOS route not in hardware when should"
+
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 tos 0 metric 1024" true
+ check_err $? "Lowest TOS route still in hardware when should not"
+
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 tos 1 metric 1024
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 tos 1 metric 1024" true
+ check_err $? "Middle TOS route in hardware when should not"
+
+ log_test "IPv4 routes with TOS"
+
+ ip -n $ns link del dev dummy1
+}
+
+fib_ipv4_metric_test()
+{
+ local ns=$1; shift
+
+ RET=0
+
+ ip -n $ns link add name dummy1 type dummy
+ ip -n $ns link set dev dummy1 up
+
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 metric 1024
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 metric 1024" false
+ check_err $? "Route not in hardware when should"
+
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 metric 1022
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 metric 1022" false
+ check_err $? "Lowest metric route not in hardware when should"
+
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 metric 1024" true
+ check_err $? "Highest metric route still in hardware when should not"
+
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 metric 1023
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 metric 1023" true
+ check_err $? "Middle metric route in hardware when should not"
+
+ log_test "IPv4 routes with metric"
+
+ ip -n $ns link del dev dummy1
+}
+
+fib_ipv4_replace_test()
+{
+ local ns=$1; shift
+ local i
+
+ RET=0
+
+ for i in $(seq 1 2); do
+ ip -n $ns link add name dummy$i type dummy
+ ip -n $ns link set dev dummy$i up
+ done
+
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 metric 1024
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 metric 1024" false
+ check_err $? "Route not in hardware when should"
+
+ ip -n $ns route replace 192.0.2.0/24 dev dummy2 metric 1024
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy2 metric 1024" false
+ check_err $? "Replacement route not in hardware when should"
+
+ # Add a route with an higher metric and make sure that replacing it
+ # does not affect the lower metric one.
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 metric 1025
+ ip -n $ns route replace 192.0.2.0/24 dev dummy2 metric 1025
+
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy2 metric 1024" false
+ check_err $? "Lowest metric route not in hardware when should"
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy2 metric 1025" true
+ check_err $? "Highest metric route in hardware when should not"
+
+ log_test "IPv4 route replace"
+
+ for i in $(seq 1 2); do
+ ip -n $ns link del dev dummy$i
+ done
+}
+
+fib_ipv4_delete_test()
+{
+ local ns=$1; shift
+ local metric
+
+ RET=0
+
+ ip -n $ns link add name dummy1 type dummy
+ ip -n $ns link set dev dummy1 up
+
+ # Insert multiple routes with the same prefix and length and varying
+ # metrics. Make sure that throughout delete operations the lowest
+ # metric route is the one in hardware.
+ for metric in $(seq 1024 1026); do
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 metric $metric
+ done
+
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 metric 1024" false
+ check_err $? "Route not in hardware when should"
+
+ ip -n $ns route del 192.0.2.0/24 dev dummy1 metric 1024
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 metric 1025" false
+ check_err $? "Lowest metric route not in hardware when should"
+
+ ip -n $ns route del 192.0.2.0/24 dev dummy1 metric 1026
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 metric 1025" false
+ check_err $? "Sole route not in hardware when should"
+
+ log_test "IPv4 route delete"
+
+ ip -n $ns link del dev dummy1
+}
+
+fib_ipv4_plen_test()
+{
+ local ns=$1; shift
+
+ RET=0
+
+ ip -n $ns link add name dummy1 type dummy
+ ip -n $ns link set dev dummy1 up
+
+ # Add two routes with the same key and different prefix length and
+ # make sure both are in hardware. It can be verfied that both are
+ # sharing the same leaf by checking the /proc/net/fib_trie
+ ip -n $ns route add 192.0.2.0/24 dev dummy1
+ ip -n $ns route add 192.0.2.0/25 dev dummy1
+
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1" false
+ check_err $? "/24 not in hardware when should"
+
+ fib4_trap_check $ns "192.0.2.0/25 dev dummy1" false
+ check_err $? "/25 not in hardware when should"
+
+ log_test "IPv4 routes with different prefix length"
+
+ ip -n $ns link del dev dummy1
+}
+
+fib_ipv4_replay_metric_test()
+{
+ local ns=$1; shift
+ local devlink_dev=$1; shift
+
+ RET=0
+
+ ip -n $ns link add name dummy1 type dummy
+ ip -n $ns link set dev dummy1 up
+
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 metric 1024
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 metric 1025
+
+ devlink -N $ns dev reload $devlink_dev
+
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 metric 1024" false
+ check_err $? "Lowest metric route not in hardware when should"
+
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 metric 1025" true
+ check_err $? "Highest metric route in hardware when should not"
+
+ log_test "IPv4 routes replay - metric"
+
+ ip -n $ns link del dev dummy1
+}
+
+fib_ipv4_replay_tos_test()
+{
+ local ns=$1; shift
+ local devlink_dev=$1; shift
+
+ RET=0
+
+ ip -n $ns link add name dummy1 type dummy
+ ip -n $ns link set dev dummy1 up
+
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 tos 0
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 tos 1
+
+ devlink -N $ns dev reload $devlink_dev
+
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 tos 1" false
+ check_err $? "Highest TOS route not in hardware when should"
+
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1 tos 0" true
+ check_err $? "Lowest TOS route in hardware when should not"
+
+ log_test "IPv4 routes replay - TOS"
+
+ ip -n $ns link del dev dummy1
+}
+
+fib_ipv4_replay_plen_test()
+{
+ local ns=$1; shift
+ local devlink_dev=$1; shift
+
+ RET=0
+
+ ip -n $ns link add name dummy1 type dummy
+ ip -n $ns link set dev dummy1 up
+
+ ip -n $ns route add 192.0.2.0/24 dev dummy1
+ ip -n $ns route add 192.0.2.0/25 dev dummy1
+
+ devlink -N $ns dev reload $devlink_dev
+
+ fib4_trap_check $ns "192.0.2.0/24 dev dummy1" false
+ check_err $? "/24 not in hardware when should"
+
+ fib4_trap_check $ns "192.0.2.0/25 dev dummy1" false
+ check_err $? "/25 not in hardware when should"
+
+ log_test "IPv4 routes replay - prefix length"
+
+ ip -n $ns link del dev dummy1
+}
+
+fib_ipv4_flush_test()
+{
+ local ns=$1; shift
+ local metric
+
+ RET=0
+
+ ip -n $ns link add name dummy1 type dummy
+ ip -n $ns link set dev dummy1 up
+
+ # Exercise the routes flushing code paths by inserting various
+ # prefix routes on a netdev and then deleting it.
+ for metric in $(seq 1 20); do
+ ip -n $ns route add 192.0.2.0/24 dev dummy1 metric $metric
+ done
+
+ ip -n $ns link del dev dummy1
+
+ log_test "IPv4 routes flushing"
+}
+
+fib_ipv6_add_test()
+{
+ local ns=$1; shift
+
+ RET=0
+
+ for i in $(seq 1 2); do
+ ip -n $ns link add name dummy$i type dummy
+ ip -n $ns link set dev dummy$i up
+ done
+
+ ip -n $ns route add 2001:db8:1::/64 dev dummy1 metric 1024
+ fib6_trap_check $ns "2001:db8:1::/64 dev dummy1 metric 1024" false
+ check_err $? "Route not in hardware when should"
+
+ ip -n $ns route append 2001:db8:1::/64 dev dummy2 metric 1024
+ fib6_trap_check $ns "2001:db8:1::/64 dev dummy2 metric 1024" true
+ check_err $? "Route in hardware when should not"
+
+ fib6_trap_check $ns "2001:db8:1::/64 dev dummy1 metric 1024" false
+ check_err $? "Route not in hardware after appending route"
+
+ log_test "IPv6 single route add"
+
+ for i in $(seq 1 2); do
+ ip -n $ns link del dev dummy$i
+ done
+}
+
+fib_ipv6_metric_test()
+{
+ local ns=$1; shift
+
+ RET=0
+
+ ip -n $ns link add name dummy1 type dummy
+ ip -n $ns link set dev dummy1 up
+
+ ip -n $ns route add 2001:db8:1::/64 dev dummy1 metric 1024
+ fib6_trap_check $ns "2001:db8:1::/64 dev dummy1 metric 1024" false
+ check_err $? "Route not in hardware when should"
+
+ ip -n $ns route add 2001:db8:1::/64 dev dummy1 metric 1022
+ fib6_trap_check $ns "2001:db8:1::/64 dev dummy1 metric 1022" false
+ check_err $? "Lowest metric route not in hardware when should"
+
+ fib6_trap_check $ns "2001:db8:1::/64 dev dummy1 metric 1024" true
+ check_err $? "Highest metric route still in hardware when should not"
+
+ ip -n $ns route add 2001:db8:1::/64 dev dummy1 metric 1023
+ fib6_trap_check $ns "2001:db8:1::/64 dev dummy1 metric 1023" true
+ check_err $? "Middle metric route in hardware when should not"
+
+ log_test "IPv6 routes with metric"
+
+ ip -n $ns link del dev dummy1
+}
+
+fib_ipv6_append_single_test()
+{
+ local ns=$1; shift
+
+ # When an IPv6 multipath route is added without the 'nexthop' keyword,
+ # different code paths are taken compared to when the keyword is used.
+ # This test tries to verify the former.
+ RET=0
+
+ for i in $(seq 1 2); do
+ ip -n $ns link add name dummy$i type dummy
+ ip -n $ns link set dev dummy$i up
+ ip -n $ns address add 2001:db8:$i::1/64 dev dummy$i
+ done
+
+ ip -n $ns route add 2001:db8:10::/64 via 2001:db8:1::2 metric 1024
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" false
+ check_err $? "Route not in hardware when should"
+
+ ip -n $ns route append 2001:db8:10::/64 via 2001:db8:2::2 metric 1024
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" false
+ check_err $? "Route not in hardware after appending"
+
+ ip -n $ns route add 2001:db8:10::/64 via 2001:db8:1::2 metric 1025
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1025" true
+ check_err $? "Route in hardware when should not"
+
+ ip -n $ns route append 2001:db8:10::/64 via 2001:db8:2::2 metric 1025
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1025" true
+ check_err $? "Route in hardware when should not after appending"
+
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" false
+ check_err $? "Lowest metric route not in hardware when should"
+
+ log_test "IPv6 append single route without 'nexthop' keyword"
+
+ for i in $(seq 1 2); do
+ ip -n $ns link del dev dummy$i
+ done
+}
+
+fib_ipv6_replace_single_test()
+{
+ local ns=$1; shift
+ local i
+
+ RET=0
+
+ for i in $(seq 1 2); do
+ ip -n $ns link add name dummy$i type dummy
+ ip -n $ns link set dev dummy$i up
+ done
+
+ ip -n $ns route add 2001:db8:1::/64 dev dummy1 metric 1024
+ fib6_trap_check $ns "2001:db8:1::/64 dev dummy1 metric 1024" false
+ check_err $? "Route not in hardware when should"
+
+ ip -n $ns route replace 2001:db8:1::/64 dev dummy2 metric 1024
+ fib6_trap_check $ns "2001:db8:1::/64 dev dummy2 metric 1024" false
+ check_err $? "Replacement route not in hardware when should"
+
+ # Add a route with an higher metric and make sure that replacing it
+ # does not affect the lower metric one.
+ ip -n $ns route add 2001:db8:1::/64 dev dummy1 metric 1025
+ ip -n $ns route replace 2001:db8:1::/64 dev dummy2 metric 1025
+
+ fib6_trap_check $ns "2001:db8:1::/64 dev dummy2 metric 1024" false
+ check_err $? "Lowest metric route not in hardware when should"
+ fib6_trap_check $ns "2001:db8:1::/64 dev dummy2 metric 1025" true
+ check_err $? "Highest metric route in hardware when should not"
+
+ log_test "IPv6 single route replace"
+
+ for i in $(seq 1 2); do
+ ip -n $ns link del dev dummy$i
+ done
+}
+
+fib_ipv6_metric_multipath_test()
+{
+ local ns=$1; shift
+
+ RET=0
+
+ for i in $(seq 1 2); do
+ ip -n $ns link add name dummy$i type dummy
+ ip -n $ns link set dev dummy$i up
+ ip -n $ns address add 2001:db8:$i::1/64 dev dummy$i
+ done
+
+ ip -n $ns route add 2001:db8:10::/64 metric 1024 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" false
+ check_err $? "Route not in hardware when should"
+
+ ip -n $ns route add 2001:db8:10::/64 metric 1022 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1022" false
+ check_err $? "Lowest metric route not in hardware when should"
+
+ ip -n $ns route add 2001:db8:10::/64 metric 1023 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" true
+ check_err $? "Highest metric route still in hardware when should not"
+
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1023" true
+ check_err $? "Middle metric route in hardware when should not"
+
+ log_test "IPv6 multipath routes with metric"
+
+ for i in $(seq 1 2); do
+ ip -n $ns link del dev dummy$i
+ done
+}
+
+fib_ipv6_append_multipath_test()
+{
+ local ns=$1; shift
+
+ RET=0
+
+ for i in $(seq 1 3); do
+ ip -n $ns link add name dummy$i type dummy
+ ip -n $ns link set dev dummy$i up
+ ip -n $ns address add 2001:db8:$i::1/64 dev dummy$i
+ done
+
+ ip -n $ns route add 2001:db8:10::/64 metric 1024 \
+ nexthop via 2001:db8:1::2 dev dummy1
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" false
+ check_err $? "Route not in hardware when should"
+
+ ip -n $ns route append 2001:db8:10::/64 metric 1024 \
+ nexthop via 2001:db8:2::2 dev dummy2 \
+ nexthop via 2001:db8:3::2 dev dummy3
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" false
+ check_err $? "Route not in hardware after appending"
+
+ ip -n $ns route add 2001:db8:10::/64 metric 1025 \
+ nexthop via 2001:db8:1::2 dev dummy1
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1025" true
+ check_err $? "Route in hardware when should not"
+
+ ip -n $ns route append 2001:db8:10::/64 metric 1025 \
+ nexthop via 2001:db8:2::2 dev dummy2 \
+ nexthop via 2001:db8:3::2 dev dummy3
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1025" true
+ check_err $? "Route in hardware when should not after appending"
+
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" false
+ check_err $? "Lowest metric route not in hardware when should"
+
+ log_test "IPv6 append multipath route with 'nexthop' keyword"
+
+ for i in $(seq 1 3); do
+ ip -n $ns link del dev dummy$i
+ done
+}
+
+fib_ipv6_replace_multipath_test()
+{
+ local ns=$1; shift
+ local i
+
+ RET=0
+
+ for i in $(seq 1 3); do
+ ip -n $ns link add name dummy$i type dummy
+ ip -n $ns link set dev dummy$i up
+ ip -n $ns address add 2001:db8:$i::1/64 dev dummy$i
+ done
+
+ ip -n $ns route add 2001:db8:10::/64 metric 1024 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" false
+ check_err $? "Route not in hardware when should"
+
+ ip -n $ns route replace 2001:db8:10::/64 metric 1024 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:3::2 dev dummy3
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" false
+ check_err $? "Replacement route not in hardware when should"
+
+ # Add a route with an higher metric and make sure that replacing it
+ # does not affect the lower metric one.
+ ip -n $ns route add 2001:db8:10::/64 metric 1025 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ ip -n $ns route replace 2001:db8:10::/64 metric 1025 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:3::2 dev dummy3
+
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" false
+ check_err $? "Lowest metric route not in hardware when should"
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1025" true
+ check_err $? "Highest metric route in hardware when should not"
+
+ log_test "IPv6 multipath route replace"
+
+ for i in $(seq 1 3); do
+ ip -n $ns link del dev dummy$i
+ done
+}
+
+fib_ipv6_append_multipath_to_single_test()
+{
+ local ns=$1; shift
+
+ # Test that when the first route in the leaf is not a multipath route
+ # and we try to append a multipath route with the same metric to it, it
+ # is not notified.
+ RET=0
+
+ for i in $(seq 1 2); do
+ ip -n $ns link add name dummy$i type dummy
+ ip -n $ns link set dev dummy$i up
+ ip -n $ns address add 2001:db8:$i::1/64 dev dummy$i
+ done
+
+ ip -n $ns route add 2001:db8:10::/64 dev dummy1 metric 1024
+ fib6_trap_check $ns "2001:db8:10::/64 dev dummy1 metric 1024" false
+ check_err $? "Route not in hardware when should"
+
+ ip -n $ns route append 2001:db8:10::/64 metric 1024 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ fib6_trap_check $ns "2001:db8:10::/64 dev dummy2 metric 1024" true
+ check_err $? "Route in hardware when should not"
+
+ fib6_trap_check $ns "2001:db8:10::/64 dev dummy1 metric 1024" false
+ check_err $? "Route not in hardware after append"
+
+ log_test "IPv6 append multipath route to non-multipath route"
+
+ for i in $(seq 1 2); do
+ ip -n $ns link del dev dummy$i
+ done
+}
+
+fib_ipv6_delete_single_test()
+{
+ local ns=$1; shift
+
+ # Test various deletion scenarios, where only a single route is
+ # deleted from the FIB node.
+ for i in $(seq 1 2); do
+ ip -n $ns link add name dummy$i type dummy
+ ip -n $ns link set dev dummy$i up
+ ip -n $ns address add 2001:db8:$i::1/64 dev dummy$i
+ done
+
+ # Test deletion of a single route when it is the only route in the FIB
+ # node.
+ RET=0
+
+ ip -n $ns route add 2001:db8:10::/64 dev dummy1 metric 1024
+ ip -n $ns route del 2001:db8:10::/64 dev dummy1 metric 1024
+
+ log_test "IPv6 delete sole single route"
+
+ # Test that deletion of last route does not affect the first one.
+ RET=0
+
+ ip -n $ns route add 2001:db8:10::/64 dev dummy1 metric 1024
+ ip -n $ns route add 2001:db8:10::/64 dev dummy1 metric 1025
+ ip -n $ns route del 2001:db8:10::/64 dev dummy1 metric 1025
+
+ fib6_trap_check $ns "2001:db8:10::/64 dev dummy1 metric 1024" false
+ check_err $? "Route not in hardware after deleting higher metric route"
+
+ log_test "IPv6 delete single route not in hardware"
+
+ ip -n $ns route del 2001:db8:10::/64 dev dummy1 metric 1024
+
+ # Test that first route is replaced by next single route in the FIB
+ # node.
+ RET=0
+
+ ip -n $ns route add 2001:db8:10::/64 dev dummy1 metric 1024
+ ip -n $ns route add 2001:db8:10::/64 dev dummy1 metric 1025
+ ip -n $ns route del 2001:db8:10::/64 dev dummy1 metric 1024
+
+ fib6_trap_check $ns "2001:db8:10::/64 dev dummy1 metric 1025" false
+ check_err $? "Route not in hardware after deleting lowest metric route"
+
+ log_test "IPv6 delete single route - replaced by single"
+
+ ip -n $ns route del 2001:db8:10::/64 dev dummy1 metric 1025
+
+ # Test that first route is replaced by next multipath route in the FIB
+ # node.
+ RET=0
+
+ ip -n $ns route add 2001:db8:10::/64 dev dummy1 metric 1024
+ ip -n $ns route add 2001:db8:10::/64 metric 1025 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ ip -n $ns route del 2001:db8:10::/64 dev dummy1 metric 1024
+
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1025" false
+ check_err $? "Route not in hardware after deleting lowest metric route"
+
+ log_test "IPv6 delete single route - replaced by multipath"
+
+ ip -n $ns route del 2001:db8:10::/64 metric 1025
+
+ # Test deletion of a single nexthop from a multipath route.
+ ip -n $ns route add 2001:db8:10::/64 metric 1024 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ ip -n $ns route del 2001:db8:10::/64 metric 1024 \
+ nexthop via 2001:db8:1::2 dev dummy1
+
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" false
+ check_err $? "Route not in hardware after deleting a single nexthop"
+
+ log_test "IPv6 delete single nexthop"
+
+ ip -n $ns route del 2001:db8:10::/64 metric 1024
+
+ for i in $(seq 1 2); do
+ ip -n $ns link del dev dummy$i
+ done
+}
+
+fib_ipv6_delete_multipath_test()
+{
+ local ns=$1; shift
+
+ # Test various deletion scenarios, where an entire multipath route is
+ # deleted from the FIB node.
+ for i in $(seq 1 2); do
+ ip -n $ns link add name dummy$i type dummy
+ ip -n $ns link set dev dummy$i up
+ ip -n $ns address add 2001:db8:$i::1/64 dev dummy$i
+ done
+
+ # Test deletion of a multipath route when it is the only route in the
+ # FIB node.
+ RET=0
+
+ ip -n $ns route add 2001:db8:10::/64 metric 1024 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ ip -n $ns route del 2001:db8:10::/64 metric 1024
+
+ log_test "IPv6 delete sole multipath route"
+
+ # Test that deletion of last route does not affect the first one.
+ RET=0
+
+ ip -n $ns route add 2001:db8:10::/64 metric 1024 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ ip -n $ns route add 2001:db8:10::/64 metric 1025 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ ip -n $ns route del 2001:db8:10::/64 metric 1025
+
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" false
+ check_err $? "Route not in hardware after deleting higher metric route"
+
+ log_test "IPv6 delete multipath route not in hardware"
+
+ ip -n $ns route del 2001:db8:10::/64 metric 1024
+
+ # Test that first route is replaced by next single route in the FIB
+ # node.
+ RET=0
+
+ ip -n $ns route add 2001:db8:10::/64 metric 1024 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ ip -n $ns route add 2001:db8:10::/64 dev dummy1 metric 1025
+ ip -n $ns route del 2001:db8:10::/64 metric 1024
+
+ fib6_trap_check $ns "2001:db8:10::/64 dev dummy1 metric 1025" false
+ check_err $? "Route not in hardware after deleting lowest metric route"
+
+ log_test "IPv6 delete multipath route - replaced by single"
+
+ ip -n $ns route del 2001:db8:10::/64 dev dummy1 metric 1025
+
+ # Test that first route is replaced by next multipath route in the FIB
+ # node.
+ RET=0
+
+ ip -n $ns route add 2001:db8:10::/64 metric 1024 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ ip -n $ns route add 2001:db8:10::/64 metric 1025 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ ip -n $ns route del 2001:db8:10::/64 metric 1024
+
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1025" false
+ check_err $? "Route not in hardware after deleting lowest metric route"
+
+ log_test "IPv6 delete multipath route - replaced by multipath"
+
+ ip -n $ns route del 2001:db8:10::/64 metric 1025
+
+ for i in $(seq 1 2); do
+ ip -n $ns link del dev dummy$i
+ done
+}
+
+fib_ipv6_replay_single_test()
+{
+ local ns=$1; shift
+ local devlink_dev=$1; shift
+
+ RET=0
+
+ for i in $(seq 1 2); do
+ ip -n $ns link add name dummy$i type dummy
+ ip -n $ns link set dev dummy$i up
+ done
+
+ ip -n $ns route add 2001:db8:1::/64 dev dummy1
+ ip -n $ns route append 2001:db8:1::/64 dev dummy2
+
+ devlink -N $ns dev reload $devlink_dev
+
+ fib6_trap_check $ns "2001:db8:1::/64 dev dummy1" false
+ check_err $? "First route not in hardware when should"
+
+ fib6_trap_check $ns "2001:db8:1::/64 dev dummy2" true
+ check_err $? "Second route in hardware when should not"
+
+ log_test "IPv6 routes replay - single route"
+
+ for i in $(seq 1 2); do
+ ip -n $ns link del dev dummy$i
+ done
+}
+
+fib_ipv6_replay_multipath_test()
+{
+ local ns=$1; shift
+ local devlink_dev=$1; shift
+
+ RET=0
+
+ for i in $(seq 1 2); do
+ ip -n $ns link add name dummy$i type dummy
+ ip -n $ns link set dev dummy$i up
+ ip -n $ns address add 2001:db8:$i::1/64 dev dummy$i
+ done
+
+ ip -n $ns route add 2001:db8:10::/64 metric 1024 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+ ip -n $ns route add 2001:db8:10::/64 metric 1025 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2
+
+ devlink -N $ns dev reload $devlink_dev
+
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1024" false
+ check_err $? "First route not in hardware when should"
+
+ fib6_trap_check $ns "2001:db8:10::/64 metric 1025" true
+ check_err $? "Second route in hardware when should not"
+
+ log_test "IPv6 routes replay - multipath route"
+
+ for i in $(seq 1 2); do
+ ip -n $ns link del dev dummy$i
+ done
+}
--
2.24.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH net-next 09/10] selftests: netdevsim: Add test for FIB offload API
2020-01-07 15:45 [PATCH net-next 00/10] net: Add route offload indication Ido Schimmel
` (7 preceding siblings ...)
2020-01-07 15:45 ` [PATCH net-next 08/10] selftests: forwarding: Add helpers and tests " Ido Schimmel
@ 2020-01-07 15:45 ` Ido Schimmel
2020-01-07 15:45 ` [PATCH net-next 10/10] selftests: mlxsw: " Ido Schimmel
9 siblings, 0 replies; 17+ messages in thread
From: Ido Schimmel @ 2020-01-07 15:45 UTC (permalink / raw)
To: netdev; +Cc: davem, jiri, jakub.kicinski, dsahern, roopa, mlxsw, Ido Schimmel
From: Ido Schimmel <idosch@mellanox.com>
Test various aspects of the FIB offload API on top of the netdevsim
implementation. Both good and bad flows are tested.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
---
.../selftests/drivers/net/netdevsim/fib.sh | 341 ++++++++++++++++++
1 file changed, 341 insertions(+)
create mode 100755 tools/testing/selftests/drivers/net/netdevsim/fib.sh
diff --git a/tools/testing/selftests/drivers/net/netdevsim/fib.sh b/tools/testing/selftests/drivers/net/netdevsim/fib.sh
new file mode 100755
index 000000000000..2f87c3be76a9
--- /dev/null
+++ b/tools/testing/selftests/drivers/net/netdevsim/fib.sh
@@ -0,0 +1,341 @@
+#!/bin/bash
+# SPDX-License-Identifier: GPL-2.0
+#
+# This test is for checking the FIB offload API. It makes use of netdevsim
+# which registers a listener to the FIB notification chain.
+
+lib_dir=$(dirname $0)/../../../net/forwarding
+
+ALL_TESTS="
+ ipv4_identical_routes
+ ipv4_tos
+ ipv4_metric
+ ipv4_replace
+ ipv4_delete
+ ipv4_plen
+ ipv4_replay
+ ipv4_flush
+ ipv4_error_path
+ ipv6_add
+ ipv6_metric
+ ipv6_append_single
+ ipv6_replace_single
+ ipv6_metric_multipath
+ ipv6_append_multipath
+ ipv6_replace_multipath
+ ipv6_append_multipath_to_single
+ ipv6_delete_single
+ ipv6_delete_multipath
+ ipv6_replay_single
+ ipv6_replay_multipath
+ ipv6_error_path
+"
+NETDEVSIM_PATH=/sys/bus/netdevsim/
+DEV_ADDR=1337
+DEV=netdevsim${DEV_ADDR}
+DEVLINK_DEV=netdevsim/${DEV}
+SYSFS_NET_DIR=/sys/bus/netdevsim/devices/$DEV/net/
+NUM_NETIFS=0
+source $lib_dir/lib.sh
+source $lib_dir/devlink_lib.sh
+source $lib_dir/fib_offload_lib.sh
+
+ipv4_identical_routes()
+{
+ fib_ipv4_identical_routes_test "testns1"
+}
+
+ipv4_tos()
+{
+ fib_ipv4_tos_test "testns1"
+}
+
+ipv4_metric()
+{
+ fib_ipv4_metric_test "testns1"
+}
+
+ipv4_replace()
+{
+ fib_ipv4_replace_test "testns1"
+}
+
+ipv4_delete()
+{
+ fib_ipv4_delete_test "testns1"
+}
+
+ipv4_plen()
+{
+ fib_ipv4_plen_test "testns1"
+}
+
+ipv4_replay_metric()
+{
+ fib_ipv4_replay_metric_test "testns1" "$DEVLINK_DEV"
+}
+
+ipv4_replay_tos()
+{
+ fib_ipv4_replay_tos_test "testns1" "$DEVLINK_DEV"
+}
+
+ipv4_replay_plen()
+{
+ fib_ipv4_replay_plen_test "testns1" "$DEVLINK_DEV"
+}
+
+ipv4_replay()
+{
+ ipv4_replay_metric
+ ipv4_replay_tos
+ ipv4_replay_plen
+}
+
+ipv4_flush()
+{
+ fib_ipv4_flush_test "testns1"
+}
+
+ipv4_error_path_add()
+{
+ local lsb
+
+ RET=0
+
+ ip -n testns1 link add name dummy1 type dummy
+ ip -n testns1 link set dev dummy1 up
+
+ devlink -N testns1 resource set $DEVLINK_DEV path IPv4/fib size 10
+ devlink -N testns1 dev reload $DEVLINK_DEV
+
+ for lsb in $(seq 1 20); do
+ ip -n testns1 route add 192.0.2.${lsb}/32 dev dummy1 \
+ &> /dev/null
+ done
+
+ log_test "IPv4 error path - add"
+
+ ip -n testns1 link del dev dummy1
+}
+
+ipv4_error_path_replay()
+{
+ local lsb
+
+ RET=0
+
+ ip -n testns1 link add name dummy1 type dummy
+ ip -n testns1 link set dev dummy1 up
+
+ devlink -N testns1 resource set $DEVLINK_DEV path IPv4/fib size 100
+ devlink -N testns1 dev reload $DEVLINK_DEV
+
+ for lsb in $(seq 1 20); do
+ ip -n testns1 route add 192.0.2.${lsb}/32 dev dummy1
+ done
+
+ devlink -N testns1 resource set $DEVLINK_DEV path IPv4/fib size 10
+ devlink -N testns1 dev reload $DEVLINK_DEV &> /dev/null
+
+ log_test "IPv4 error path - replay"
+
+ ip -n testns1 link del dev dummy1
+
+ # Successfully reload after deleting all the routes.
+ devlink -N testns1 resource set $DEVLINK_DEV path IPv4/fib size 100
+ devlink -N testns1 dev reload $DEVLINK_DEV
+}
+
+ipv4_error_path()
+{
+ # Test the different error paths of the notifiers by limiting the size
+ # of the "IPv4/fib" resource.
+ ipv4_error_path_add
+ ipv4_error_path_replay
+}
+
+ipv6_add()
+{
+ fib_ipv6_add_test "testns1"
+}
+
+ipv6_metric()
+{
+ fib_ipv6_metric_test "testns1"
+}
+
+ipv6_append_single()
+{
+ fib_ipv6_append_single_test "testns1"
+}
+
+ipv6_replace_single()
+{
+ fib_ipv6_replace_single_test "testns1"
+}
+
+ipv6_metric_multipath()
+{
+ fib_ipv6_metric_multipath_test "testns1"
+}
+
+ipv6_append_multipath()
+{
+ fib_ipv6_append_multipath_test "testns1"
+}
+
+ipv6_replace_multipath()
+{
+ fib_ipv6_replace_multipath_test "testns1"
+}
+
+ipv6_append_multipath_to_single()
+{
+ fib_ipv6_append_multipath_to_single_test "testns1"
+}
+
+ipv6_delete_single()
+{
+ fib_ipv6_delete_single_test "testns1"
+}
+
+ipv6_delete_multipath()
+{
+ fib_ipv6_delete_multipath_test "testns1"
+}
+
+ipv6_replay_single()
+{
+ fib_ipv6_replay_single_test "testns1" "$DEVLINK_DEV"
+}
+
+ipv6_replay_multipath()
+{
+ fib_ipv6_replay_multipath_test "testns1" "$DEVLINK_DEV"
+}
+
+ipv6_error_path_add_single()
+{
+ local lsb
+
+ RET=0
+
+ ip -n testns1 link add name dummy1 type dummy
+ ip -n testns1 link set dev dummy1 up
+
+ devlink -N testns1 resource set $DEVLINK_DEV path IPv6/fib size 10
+ devlink -N testns1 dev reload $DEVLINK_DEV
+
+ for lsb in $(seq 1 20); do
+ ip -n testns1 route add 2001:db8:1::${lsb}/128 dev dummy1 \
+ &> /dev/null
+ done
+
+ log_test "IPv6 error path - add single"
+
+ ip -n testns1 link del dev dummy1
+}
+
+ipv6_error_path_add_multipath()
+{
+ local lsb
+
+ RET=0
+
+ for i in $(seq 1 2); do
+ ip -n testns1 link add name dummy$i type dummy
+ ip -n testns1 link set dev dummy$i up
+ ip -n testns1 address add 2001:db8:$i::1/64 dev dummy$i
+ done
+
+ devlink -N testns1 resource set $DEVLINK_DEV path IPv6/fib size 10
+ devlink -N testns1 dev reload $DEVLINK_DEV
+
+ for lsb in $(seq 1 20); do
+ ip -n testns1 route add 2001:db8:10::${lsb}/128 \
+ nexthop via 2001:db8:1::2 dev dummy1 \
+ nexthop via 2001:db8:2::2 dev dummy2 &> /dev/null
+ done
+
+ log_test "IPv6 error path - add multipath"
+
+ for i in $(seq 1 2); do
+ ip -n testns1 link del dev dummy$i
+ done
+}
+
+ipv6_error_path_replay()
+{
+ local lsb
+
+ RET=0
+
+ ip -n testns1 link add name dummy1 type dummy
+ ip -n testns1 link set dev dummy1 up
+
+ devlink -N testns1 resource set $DEVLINK_DEV path IPv6/fib size 100
+ devlink -N testns1 dev reload $DEVLINK_DEV
+
+ for lsb in $(seq 1 20); do
+ ip -n testns1 route add 2001:db8:1::${lsb}/128 dev dummy1
+ done
+
+ devlink -N testns1 resource set $DEVLINK_DEV path IPv6/fib size 10
+ devlink -N testns1 dev reload $DEVLINK_DEV &> /dev/null
+
+ log_test "IPv6 error path - replay"
+
+ ip -n testns1 link del dev dummy1
+
+ # Successfully reload after deleting all the routes.
+ devlink -N testns1 resource set $DEVLINK_DEV path IPv6/fib size 100
+ devlink -N testns1 dev reload $DEVLINK_DEV
+}
+
+ipv6_error_path()
+{
+ # Test the different error paths of the notifiers by limiting the size
+ # of the "IPv6/fib" resource.
+ ipv6_error_path_add_single
+ ipv6_error_path_add_multipath
+ ipv6_error_path_replay
+}
+
+setup_prepare()
+{
+ local netdev
+
+ modprobe netdevsim &> /dev/null
+
+ echo "$DEV_ADDR 1" > ${NETDEVSIM_PATH}/new_device
+ while [ ! -d $SYSFS_NET_DIR ] ; do :; done
+
+ ip netns add testns1
+ if [ $? -ne 0 ]; then
+ echo "Failed to add netns \"testns1\""
+ exit 1
+ fi
+
+ devlink dev reload $DEVLINK_DEV netns testns1
+ if [ $? -ne 0 ]; then
+ echo "Failed to reload into netns \"testns1\""
+ exit 1
+ fi
+}
+
+cleanup()
+{
+ pre_cleanup
+ ip netns del testns1
+ echo "$DEV_ADDR" > ${NETDEVSIM_PATH}/del_device
+ modprobe -r netdevsim &> /dev/null
+}
+
+trap cleanup EXIT
+
+setup_prepare
+
+tests_run
+
+exit $EXIT_STATUS
--
2.24.1
^ permalink raw reply related [flat|nested] 17+ messages in thread* [PATCH net-next 10/10] selftests: mlxsw: Add test for FIB offload API
2020-01-07 15:45 [PATCH net-next 00/10] net: Add route offload indication Ido Schimmel
` (8 preceding siblings ...)
2020-01-07 15:45 ` [PATCH net-next 09/10] selftests: netdevsim: Add test for FIB offload API Ido Schimmel
@ 2020-01-07 15:45 ` Ido Schimmel
9 siblings, 0 replies; 17+ messages in thread
From: Ido Schimmel @ 2020-01-07 15:45 UTC (permalink / raw)
To: netdev; +Cc: davem, jiri, jakub.kicinski, dsahern, roopa, mlxsw, Ido Schimmel
From: Ido Schimmel <idosch@mellanox.com>
The test reuses the common FIB offload tests in order to make sure that
mlxsw correctly implements FIB offload.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
---
.../selftests/drivers/net/mlxsw/fib.sh | 180 ++++++++++++++++++
1 file changed, 180 insertions(+)
create mode 100755 tools/testing/selftests/drivers/net/mlxsw/fib.sh
diff --git a/tools/testing/selftests/drivers/net/mlxsw/fib.sh b/tools/testing/selftests/drivers/net/mlxsw/fib.sh
new file mode 100755
index 000000000000..45115f81c2b1
--- /dev/null
+++ b/tools/testing/selftests/drivers/net/mlxsw/fib.sh
@@ -0,0 +1,180 @@
+#!/bin/bash
+# SPDX-License-Identifier: GPL-2.0
+#
+# This test is for checking the FIB offload API on top of mlxsw.
+
+lib_dir=$(dirname $0)/../../../net/forwarding
+
+ALL_TESTS="
+ ipv4_identical_routes
+ ipv4_tos
+ ipv4_metric
+ ipv4_replace
+ ipv4_delete
+ ipv4_plen
+ ipv4_replay
+ ipv4_flush
+ ipv6_add
+ ipv6_metric
+ ipv6_append_single
+ ipv6_replace_single
+ ipv6_metric_multipath
+ ipv6_append_multipath
+ ipv6_replace_multipath
+ ipv6_append_multipath_to_single
+ ipv6_delete_single
+ ipv6_delete_multipath
+ ipv6_replay_single
+ ipv6_replay_multipath
+"
+NUM_NETIFS=0
+source $lib_dir/lib.sh
+source $lib_dir/devlink_lib.sh
+source $lib_dir/fib_offload_lib.sh
+
+ipv4_identical_routes()
+{
+ fib_ipv4_identical_routes_test "testns1"
+}
+
+ipv4_tos()
+{
+ fib_ipv4_tos_test "testns1"
+}
+
+ipv4_metric()
+{
+ fib_ipv4_metric_test "testns1"
+}
+
+ipv4_replace()
+{
+ fib_ipv4_replace_test "testns1"
+}
+
+ipv4_delete()
+{
+ fib_ipv4_delete_test "testns1"
+}
+
+ipv4_plen()
+{
+ fib_ipv4_plen_test "testns1"
+}
+
+ipv4_replay_metric()
+{
+ fib_ipv4_replay_metric_test "testns1" "$DEVLINK_DEV"
+}
+
+ipv4_replay_tos()
+{
+ fib_ipv4_replay_tos_test "testns1" "$DEVLINK_DEV"
+}
+
+ipv4_replay_plen()
+{
+ fib_ipv4_replay_plen_test "testns1" "$DEVLINK_DEV"
+}
+
+ipv4_replay()
+{
+ ipv4_replay_metric
+ ipv4_replay_tos
+ ipv4_replay_plen
+}
+
+ipv4_flush()
+{
+ fib_ipv4_flush_test "testns1"
+}
+
+ipv6_add()
+{
+ fib_ipv6_add_test "testns1"
+}
+
+ipv6_metric()
+{
+ fib_ipv6_metric_test "testns1"
+}
+
+ipv6_append_single()
+{
+ fib_ipv6_append_single_test "testns1"
+}
+
+ipv6_replace_single()
+{
+ fib_ipv6_replace_single_test "testns1"
+}
+
+ipv6_metric_multipath()
+{
+ fib_ipv6_metric_multipath_test "testns1"
+}
+
+ipv6_append_multipath()
+{
+ fib_ipv6_append_multipath_test "testns1"
+}
+
+ipv6_replace_multipath()
+{
+ fib_ipv6_replace_multipath_test "testns1"
+}
+
+ipv6_append_multipath_to_single()
+{
+ fib_ipv6_append_multipath_to_single_test "testns1"
+}
+
+ipv6_delete_single()
+{
+ fib_ipv6_delete_single_test "testns1"
+}
+
+ipv6_delete_multipath()
+{
+ fib_ipv6_delete_multipath_test "testns1"
+}
+
+ipv6_replay_single()
+{
+ fib_ipv6_replay_single_test "testns1" "$DEVLINK_DEV"
+}
+
+ipv6_replay_multipath()
+{
+ fib_ipv6_replay_multipath_test "testns1" "$DEVLINK_DEV"
+}
+
+setup_prepare()
+{
+ ip netns add testns1
+ if [ $? -ne 0 ]; then
+ echo "Failed to add netns \"testns1\""
+ exit 1
+ fi
+
+ devlink dev reload $DEVLINK_DEV netns testns1
+ if [ $? -ne 0 ]; then
+ echo "Failed to reload into netns \"testns1\""
+ exit 1
+ fi
+}
+
+cleanup()
+{
+ pre_cleanup
+ devlink -N testns1 dev reload $DEVLINK_DEV netns $$
+ ip netns del testns1
+}
+
+trap cleanup EXIT
+
+setup_prepare
+
+tests_run
+
+exit $EXIT_STATUS
--
2.24.1
^ permalink raw reply related [flat|nested] 17+ messages in thread