[dpdk-dev] [PATCH v1 0/3] lpm: increase number of next hops for lpm (ipv4)
Vladimir Medvedkin
medvedkinv at gmail.com
Mon Oct 26 15:03:31 CET 2015
Hi Michal,
Forwarding class can help us to classify traffic based on dst prefix, it's
something like Juniper DCU. For example on Juniper MX I can make policy
that install prefix into the FIB with some class and use it on dataplane,
for example with ACL.
On Juniper MX I can make something like that:
#show policy-options
policy-statement community-to-class {
term customer {
from community originate-customer;
then destination-class customer;
}
}
community originate-customer members 12345:11111;
# show routing-options
forwarding-table {
export community-to-class;
}
# show forwarding-options
forwarding-options {
family inet {
filter {
output test-filter;
}
}
}
# show firewall family inet filter test-filter
term 1 {
from {
protocol icmp;
destination-class customer;
}
then {
discard;
}
}
announce route 10.10.10.10/32 next-hop 10.10.10.2 community 12345:11111
After than on dataplane we have
NPC1( vty)# show route ip lookup 10.10.10.10
Route Information (10.10.10.10):
interface : xe-1/0/0.0 (328)
Nexthop prefix : -
Nexthop ID : 1048574
MTU : 0
Class ID : 129 <- That is "forwarding class" in my implementation
This construction discards all ICMP traffic that goes to dst prefixes which
was originated with community 12345:11111. With this mechanism we can make
on control plane different sophisticated policy to control traffic on
dataplane.
The same with as_num, we can have on dataplane AS number that has
originated that prefix, or another 4-byte number e.g. geo-id.
What issue do you mean? I think it is because of table/pipeline/test
frameworks that doesen't want to compile due to changing API/ABI. You can
turn it off for LPM testing, if my patch will be applied I will make
changes in above-mentioned frameworks.
Regards,
Vladimir
2015-10-26 14:57 GMT+03:00 Jastrzebski, MichalX K <
michalx.k.jastrzebski at intel.com>:
> > -----Original Message-----
> > From: Michal Jastrzebski [mailto:michalx.k.jastrzebski at intel.com]
> > Sent: Monday, October 26, 2015 12:55 PM
> > To: Vladimir Medvedkin
> > Subject: Re: [dpdk-dev] [PATCH v1 0/3] lpm: increase number of next hops
> > for lpm (ipv4)
> >
> > On Sun, Oct 25, 2015 at 08:52:04PM +0300, Vladimir Medvedkin wrote:
> > > Hi all,
> > >
> > > Here my implementation
> > >
> > > Signed-off-by: Vladimir Medvedkin <medvedkinv at gmail.com>
> > > ---
> > > config/common_bsdapp | 1 +
> > > config/common_linuxapp | 1 +
> > > lib/librte_lpm/rte_lpm.c | 194
> > > +++++++++++++++++++++++++++++------------------
> > > lib/librte_lpm/rte_lpm.h | 163 +++++++++++++++++++++++----------------
> > > 4 files changed, 219 insertions(+), 140 deletions(-)
> > >
> > > diff --git a/config/common_bsdapp b/config/common_bsdapp
> > > index b37dcf4..408cc2c 100644
> > > --- a/config/common_bsdapp
> > > +++ b/config/common_bsdapp
> > > @@ -344,6 +344,7 @@ CONFIG_RTE_LIBRTE_JOBSTATS=y
> > > #
> > > CONFIG_RTE_LIBRTE_LPM=y
> > > CONFIG_RTE_LIBRTE_LPM_DEBUG=n
> > > +CONFIG_RTE_LIBRTE_LPM_ASNUM=n
> > >
> > > #
> > > # Compile librte_acl
> > > diff --git a/config/common_linuxapp b/config/common_linuxapp
> > > index 0de43d5..1c60e63 100644
> > > --- a/config/common_linuxapp
> > > +++ b/config/common_linuxapp
> > > @@ -352,6 +352,7 @@ CONFIG_RTE_LIBRTE_JOBSTATS=y
> > > #
> > > CONFIG_RTE_LIBRTE_LPM=y
> > > CONFIG_RTE_LIBRTE_LPM_DEBUG=n
> > > +CONFIG_RTE_LIBRTE_LPM_ASNUM=n
> > >
> > > #
> > > # Compile librte_acl
> > > diff --git a/lib/librte_lpm/rte_lpm.c b/lib/librte_lpm/rte_lpm.c
> > > index 163ba3c..363b400 100644
> > > --- a/lib/librte_lpm/rte_lpm.c
> > > +++ b/lib/librte_lpm/rte_lpm.c
> > > @@ -159,9 +159,11 @@ rte_lpm_create(const char *name, int socket_id,
> > int
> > > max_rules,
> > >
> > > lpm_list = RTE_TAILQ_CAST(rte_lpm_tailq.head, rte_lpm_list);
> > >
> > > - RTE_BUILD_BUG_ON(sizeof(struct rte_lpm_tbl24_entry) != 2);
> > > - RTE_BUILD_BUG_ON(sizeof(struct rte_lpm_tbl8_entry) != 2);
> > > -
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + RTE_BUILD_BUG_ON(sizeof(struct rte_lpm_tbl_entry) != 8);
> > > +#else
> > > + RTE_BUILD_BUG_ON(sizeof(struct rte_lpm_tbl_entry) != 4);
> > > +#endif
> > > /* Check user arguments. */
> > > if ((name == NULL) || (socket_id < -1) || (max_rules == 0)){
> > > rte_errno = EINVAL;
> > > @@ -261,7 +263,7 @@ rte_lpm_free(struct rte_lpm *lpm)
> > > */
> > > static inline int32_t
> > > rule_add(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth,
> > > - uint8_t next_hop)
> > > + struct rte_lpm_res *res)
> > > {
> > > uint32_t rule_gindex, rule_index, last_rule;
> > > int i;
> > > @@ -282,8 +284,11 @@ rule_add(struct rte_lpm *lpm, uint32_t
> > ip_masked,
> > > uint8_t depth,
> > >
> > > /* If rule already exists update its next_hop
> and
> > > return. */
> > > if (lpm->rules_tbl[rule_index].ip ==
> ip_masked) {
> > > - lpm->rules_tbl[rule_index].next_hop =
> > > next_hop;
> > > -
> > > + lpm->rules_tbl[rule_index].next_hop =
> > > res->next_hop;
> > > + lpm->rules_tbl[rule_index].fwd_class =
> > > res->fwd_class;
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + lpm->rules_tbl[rule_index].as_num =
> > > res->as_num;
> > > +#endif
> > > return rule_index;
> > > }
> > > }
> > > @@ -320,7 +325,11 @@ rule_add(struct rte_lpm *lpm, uint32_t
> > ip_masked,
> > > uint8_t depth,
> > >
> > > /* Add the new rule. */
> > > lpm->rules_tbl[rule_index].ip = ip_masked;
> > > - lpm->rules_tbl[rule_index].next_hop = next_hop;
> > > + lpm->rules_tbl[rule_index].next_hop = res->next_hop;
> > > + lpm->rules_tbl[rule_index].fwd_class = res->fwd_class;
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + lpm->rules_tbl[rule_index].as_num = res->as_num;
> > > +#endif
> > >
> > > /* Increment the used rules counter for this rule group. */
> > > lpm->rule_info[depth - 1].used_rules++;
> > > @@ -382,10 +391,10 @@ rule_find(struct rte_lpm *lpm, uint32_t
> > ip_masked,
> > > uint8_t depth)
> > > * Find, clean and allocate a tbl8.
> > > */
> > > static inline int32_t
> > > -tbl8_alloc(struct rte_lpm_tbl8_entry *tbl8)
> > > +tbl8_alloc(struct rte_lpm_tbl_entry *tbl8)
> > > {
> > > uint32_t tbl8_gindex; /* tbl8 group index. */
> > > - struct rte_lpm_tbl8_entry *tbl8_entry;
> > > + struct rte_lpm_tbl_entry *tbl8_entry;
> > >
> > > /* Scan through tbl8 to find a free (i.e. INVALID) tbl8 group.
> */
> > > for (tbl8_gindex = 0; tbl8_gindex < RTE_LPM_TBL8_NUM_GROUPS;
> > > @@ -393,12 +402,12 @@ tbl8_alloc(struct rte_lpm_tbl8_entry *tbl8)
> > > tbl8_entry = &tbl8[tbl8_gindex *
> > > RTE_LPM_TBL8_GROUP_NUM_ENTRIES];
> > > /* If a free tbl8 group is found clean it and set as
> VALID.
> > > */
> > > - if (!tbl8_entry->valid_group) {
> > > + if (!tbl8_entry->ext_valid) {
> > > memset(&tbl8_entry[0], 0,
> > > RTE_LPM_TBL8_GROUP_NUM_ENTRIES
> *
> > > sizeof(tbl8_entry[0]));
> > >
> > > - tbl8_entry->valid_group = VALID;
> > > + tbl8_entry->ext_valid = VALID;
> > >
> > > /* Return group index for allocated tbl8
> group. */
> > > return tbl8_gindex;
> > > @@ -410,46 +419,50 @@ tbl8_alloc(struct rte_lpm_tbl8_entry *tbl8)
> > > }
> > >
> > > static inline void
> > > -tbl8_free(struct rte_lpm_tbl8_entry *tbl8, uint32_t tbl8_group_start)
> > > +tbl8_free(struct rte_lpm_tbl_entry *tbl8, uint32_t tbl8_group_start)
> > > {
> > > /* Set tbl8 group invalid*/
> > > - tbl8[tbl8_group_start].valid_group = INVALID;
> > > + tbl8[tbl8_group_start].ext_valid = INVALID;
> > > }
> > >
> > > static inline int32_t
> > > add_depth_small(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
> > > - uint8_t next_hop)
> > > + struct rte_lpm_res *res)
> > > {
> > > uint32_t tbl24_index, tbl24_range, tbl8_index, tbl8_group_end,
> i, j;
> > >
> > > /* Calculate the index into Table24. */
> > > tbl24_index = ip >> 8;
> > > tbl24_range = depth_to_range(depth);
> > > + struct rte_lpm_tbl_entry new_tbl_entry = {
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + .as_num = res->as_num,
> > > +#endif
> > > + .next_hop = res->next_hop,
> > > + .fwd_class = res->fwd_class,
> > > + .ext_valid = 0,
> > > + .depth = depth,
> > > + .valid = VALID,
> > > + };
> > > +
> > >
> > > for (i = tbl24_index; i < (tbl24_index + tbl24_range); i++) {
> > > /*
> > > * For invalid OR valid and non-extended tbl 24
> entries set
> > > * entry.
> > > */
> > > - if (!lpm->tbl24[i].valid || (lpm->tbl24[i].ext_entry
> == 0 &&
> > > + if (!lpm->tbl24[i].valid || (lpm->tbl24[i].ext_valid
> == 0 &&
> > > lpm->tbl24[i].depth <= depth)) {
> > >
> > > - struct rte_lpm_tbl24_entry new_tbl24_entry = {
> > > - { .next_hop = next_hop, },
> > > - .valid = VALID,
> > > - .ext_entry = 0,
> > > - .depth = depth,
> > > - };
> > > -
> > > /* Setting tbl24 entry in one go to avoid race
> > > * conditions
> > > */
> > > - lpm->tbl24[i] = new_tbl24_entry;
> > > + lpm->tbl24[i] = new_tbl_entry;
> > >
> > > continue;
> > > }
> > >
> > > - if (lpm->tbl24[i].ext_entry == 1) {
> > > + if (lpm->tbl24[i].ext_valid == 1) {
> > > /* If tbl24 entry is valid and extended
> calculate
> > > the
> > > * index into tbl8.
> > > */
> > > @@ -461,19 +474,14 @@ add_depth_small(struct rte_lpm *lpm, uint32_t
> > ip,
> > > uint8_t depth,
> > > for (j = tbl8_index; j < tbl8_group_end; j++) {
> > > if (!lpm->tbl8[j].valid ||
> > > lpm->tbl8[j].depth <=
> > > depth) {
> > > - struct rte_lpm_tbl8_entry
> > > - new_tbl8_entry = {
> > > - .valid = VALID,
> > > - .valid_group = VALID,
> > > - .depth = depth,
> > > - .next_hop = next_hop,
> > > - };
> > > +
> > > + new_tbl_entry.ext_valid =
> VALID;
> > >
> > > /*
> > > * Setting tbl8 entry in one
> go to
> > > avoid
> > > * race conditions
> > > */
> > > - lpm->tbl8[j] = new_tbl8_entry;
> > > + lpm->tbl8[j] = new_tbl_entry;
> > >
> > > continue;
> > > }
> > > @@ -486,7 +494,7 @@ add_depth_small(struct rte_lpm *lpm, uint32_t ip,
> > > uint8_t depth,
> > >
> > > static inline int32_t
> > > add_depth_big(struct rte_lpm *lpm, uint32_t ip_masked, uint8_t depth,
> > > - uint8_t next_hop)
> > > + struct rte_lpm_res *res)
> > > {
> > > uint32_t tbl24_index;
> > > int32_t tbl8_group_index, tbl8_group_start, tbl8_group_end,
> > > tbl8_index,
> > > @@ -512,7 +520,11 @@ add_depth_big(struct rte_lpm *lpm, uint32_t
> > ip_masked,
> > > uint8_t depth,
> > > /* Set tbl8 entry. */
> > > for (i = tbl8_index; i < (tbl8_index + tbl8_range);
> i++) {
> > > lpm->tbl8[i].depth = depth;
> > > - lpm->tbl8[i].next_hop = next_hop;
> > > + lpm->tbl8[i].next_hop = res->next_hop;
> > > + lpm->tbl8[i].fwd_class = res->fwd_class;
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + lpm->tbl8[i].as_num = res->as_num;
> > > +#endif
> > > lpm->tbl8[i].valid = VALID;
> > > }
> > >
> > > @@ -522,17 +534,17 @@ add_depth_big(struct rte_lpm *lpm, uint32_t
> > > ip_masked, uint8_t depth,
> > > * so assign whole structure in one go
> > > */
> > >
> > > - struct rte_lpm_tbl24_entry new_tbl24_entry = {
> > > - { .tbl8_gindex = (uint8_t)tbl8_group_index, },
> > > - .valid = VALID,
> > > - .ext_entry = 1,
> > > + struct rte_lpm_tbl_entry new_tbl24_entry = {
> > > + .tbl8_gindex = (uint16_t)tbl8_group_index,
> > > .depth = 0,
> > > + .ext_valid = 1,
> > > + .valid = VALID,
> > > };
> > >
> > > lpm->tbl24[tbl24_index] = new_tbl24_entry;
> > >
> > > }/* If valid entry but not extended calculate the index into
> > > Table8. */
> > > - else if (lpm->tbl24[tbl24_index].ext_entry == 0) {
> > > + else if (lpm->tbl24[tbl24_index].ext_valid == 0) {
> > > /* Search for free tbl8 group. */
> > > tbl8_group_index = tbl8_alloc(lpm->tbl8);
> > >
> > > @@ -551,6 +563,11 @@ add_depth_big(struct rte_lpm *lpm, uint32_t
> > ip_masked,
> > > uint8_t depth,
> > > lpm->tbl8[i].depth =
> lpm->tbl24[tbl24_index].depth;
> > > lpm->tbl8[i].next_hop =
> > >
> lpm->tbl24[tbl24_index].next_hop;
> > > + lpm->tbl8[i].fwd_class =
> > > +
> lpm->tbl24[tbl24_index].fwd_class;
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + lpm->tbl8[i].as_num =
> > > lpm->tbl24[tbl24_index].as_num;
> > > +#endif
> > > }
> > >
> > > tbl8_index = tbl8_group_start + (ip_masked & 0xFF);
> > > @@ -561,7 +578,11 @@ add_depth_big(struct rte_lpm *lpm, uint32_t
> > ip_masked,
> > > uint8_t depth,
> > > lpm->tbl8[i].depth <= depth) {
> > > lpm->tbl8[i].valid = VALID;
> > > lpm->tbl8[i].depth = depth;
> > > - lpm->tbl8[i].next_hop = next_hop;
> > > + lpm->tbl8[i].next_hop = res->next_hop;
> > > + lpm->tbl8[i].fwd_class =
> res->fwd_class;
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + lpm->tbl8[i].as_num = res->as_num;
> > > +#endif
> > >
> > > continue;
> > > }
> > > @@ -573,11 +594,11 @@ add_depth_big(struct rte_lpm *lpm, uint32_t
> > > ip_masked, uint8_t depth,
> > > * so assign whole structure in one go.
> > > */
> > >
> > > - struct rte_lpm_tbl24_entry new_tbl24_entry = {
> > > - { .tbl8_gindex =
> (uint8_t)tbl8_group_index,
> > > },
> > > - .valid = VALID,
> > > - .ext_entry = 1,
> > > + struct rte_lpm_tbl_entry new_tbl24_entry = {
> > > + .tbl8_gindex =
> (uint16_t)tbl8_group_index,
> > > .depth = 0,
> > > + .ext_valid = 1,
> > > + .valid = VALID,
> > > };
> > >
> > > lpm->tbl24[tbl24_index] = new_tbl24_entry;
> > > @@ -595,11 +616,15 @@ add_depth_big(struct rte_lpm *lpm, uint32_t
> > > ip_masked, uint8_t depth,
> > >
> > > if (!lpm->tbl8[i].valid ||
> > > lpm->tbl8[i].depth <= depth) {
> > > - struct rte_lpm_tbl8_entry
> new_tbl8_entry = {
> > > - .valid = VALID,
> > > + struct rte_lpm_tbl_entry
> new_tbl8_entry = {
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + .as_num = res->as_num,
> > > +#endif
> > > + .next_hop = res->next_hop,
> > > + .fwd_class = res->fwd_class,
> > > .depth = depth,
> > > - .next_hop = next_hop,
> > > - .valid_group =
> > > lpm->tbl8[i].valid_group,
> > > + .ext_valid =
> lpm->tbl8[i].ext_valid,
> > > + .valid = VALID,
> > > };
> > >
> > > /*
> > > @@ -621,19 +646,19 @@ add_depth_big(struct rte_lpm *lpm, uint32_t
> > > ip_masked, uint8_t depth,
> > > */
> > > int
> > > rte_lpm_add(struct rte_lpm *lpm, uint32_t ip, uint8_t depth,
> > > - uint8_t next_hop)
> > > + struct rte_lpm_res *res)
> > > {
> > > int32_t rule_index, status = 0;
> > > uint32_t ip_masked;
> > >
> > > /* Check user arguments. */
> > > - if ((lpm == NULL) || (depth < 1) || (depth >
> RTE_LPM_MAX_DEPTH))
> > > + if ((lpm == NULL) || (res == NULL) || (depth < 1) || (depth >
> > > RTE_LPM_MAX_DEPTH))
> > > return -EINVAL;
> > >
> > > ip_masked = ip & depth_to_mask(depth);
> > >
> > > /* Add the rule to the rule table. */
> > > - rule_index = rule_add(lpm, ip_masked, depth, next_hop);
> > > + rule_index = rule_add(lpm, ip_masked, depth, res);
> > >
> > > /* If the is no space available for new rule return error. */
> > > if (rule_index < 0) {
> > > @@ -641,10 +666,10 @@ rte_lpm_add(struct rte_lpm *lpm, uint32_t ip,
> > uint8_t
> > > depth,
> > > }
> > >
> > > if (depth <= MAX_DEPTH_TBL24) {
> > > - status = add_depth_small(lpm, ip_masked, depth,
> next_hop);
> > > + status = add_depth_small(lpm, ip_masked, depth, res);
> > > }
> > > else { /* If depth > RTE_LPM_MAX_DEPTH_TBL24 */
> > > - status = add_depth_big(lpm, ip_masked, depth,
> next_hop);
> > > + status = add_depth_big(lpm, ip_masked, depth, res);
> > >
> > > /*
> > > * If add fails due to exhaustion of tbl8 extensions
> delete
> > > @@ -665,14 +690,14 @@ rte_lpm_add(struct rte_lpm *lpm, uint32_t ip,
> > uint8_t
> > > depth,
> > > */
> > > int
> > > rte_lpm_is_rule_present(struct rte_lpm *lpm, uint32_t ip, uint8_t
> depth,
> > > -uint8_t *next_hop)
> > > + struct rte_lpm_res *res)
> > > {
> > > uint32_t ip_masked;
> > > int32_t rule_index;
> > >
> > > /* Check user arguments. */
> > > if ((lpm == NULL) ||
> > > - (next_hop == NULL) ||
> > > + (res == NULL) ||
> > > (depth < 1) || (depth > RTE_LPM_MAX_DEPTH))
> > > return -EINVAL;
> > >
> > > @@ -681,7 +706,11 @@ uint8_t *next_hop)
> > > rule_index = rule_find(lpm, ip_masked, depth);
> > >
> > > if (rule_index >= 0) {
> > > - *next_hop = lpm->rules_tbl[rule_index].next_hop;
> > > + res->next_hop = lpm->rules_tbl[rule_index].next_hop;
> > > + res->fwd_class = lpm->rules_tbl[rule_index].fwd_class;
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + res->as_num = lpm->rules_tbl[rule_index].as_num;
> > > +#endif
> > > return 1;
> > > }
> > >
> > > @@ -731,7 +760,7 @@ delete_depth_small(struct rte_lpm *lpm, uint32_t
> > > ip_masked,
> > > */
> > > for (i = tbl24_index; i < (tbl24_index + tbl24_range);
> i++)
> > > {
> > >
> > > - if (lpm->tbl24[i].ext_entry == 0 &&
> > > + if (lpm->tbl24[i].ext_valid == 0 &&
> > > lpm->tbl24[i].depth <= depth )
> {
> > > lpm->tbl24[i].valid = INVALID;
> > > }
> > > @@ -761,23 +790,30 @@ delete_depth_small(struct rte_lpm *lpm,
> > uint32_t
> > > ip_masked,
> > > * associated with this rule.
> > > */
> > >
> > > - struct rte_lpm_tbl24_entry new_tbl24_entry = {
> > > - {.next_hop =
> > > lpm->rules_tbl[sub_rule_index].next_hop,},
> > > - .valid = VALID,
> > > - .ext_entry = 0,
> > > + struct rte_lpm_tbl_entry new_tbl24_entry = {
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + .as_num =
> lpm->rules_tbl[sub_rule_index].as_num,
> > > +#endif
> > > + .next_hop =
> lpm->rules_tbl[sub_rule_index].next_hop,
> > > + .fwd_class =
> > > lpm->rules_tbl[sub_rule_index].fwd_class,
> > > .depth = sub_rule_depth,
> > > + .ext_valid = 0,
> > > + .valid = VALID,
> > > };
> > >
> > > - struct rte_lpm_tbl8_entry new_tbl8_entry = {
> > > - .valid = VALID,
> > > + struct rte_lpm_tbl_entry new_tbl8_entry = {
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + .as_num =
> lpm->rules_tbl[sub_rule_index].as_num,
> > > +#endif
> > > + .next_hop =
> lpm->rules_tbl[sub_rule_index].next_hop,
> > > + .fwd_class =
> > > lpm->rules_tbl[sub_rule_index].fwd_class,
> > > .depth = sub_rule_depth,
> > > - .next_hop = lpm->rules_tbl
> > > - [sub_rule_index].next_hop,
> > > + .valid = VALID,
> > > };
> > >
> > > for (i = tbl24_index; i < (tbl24_index + tbl24_range);
> i++)
> > > {
> > >
> > > - if (lpm->tbl24[i].ext_entry == 0 &&
> > > + if (lpm->tbl24[i].ext_valid == 0 &&
> > > lpm->tbl24[i].depth <= depth )
> {
> > > lpm->tbl24[i] = new_tbl24_entry;
> > > }
> > > @@ -814,7 +850,7 @@ delete_depth_small(struct rte_lpm *lpm, uint32_t
> > > ip_masked,
> > > * thus can be recycled
> > > */
> > > static inline int32_t
> > > -tbl8_recycle_check(struct rte_lpm_tbl8_entry *tbl8, uint32_t
> > > tbl8_group_start)
> > > +tbl8_recycle_check(struct rte_lpm_tbl_entry *tbl8, uint32_t
> > > tbl8_group_start)
> > > {
> > > uint32_t tbl8_group_end, i;
> > > tbl8_group_end = tbl8_group_start +
> > RTE_LPM_TBL8_GROUP_NUM_ENTRIES;
> > > @@ -891,11 +927,15 @@ delete_depth_big(struct rte_lpm *lpm, uint32_t
> > > ip_masked,
> > > }
> > > else {
> > > /* Set new tbl8 entry. */
> > > - struct rte_lpm_tbl8_entry new_tbl8_entry = {
> > > - .valid = VALID,
> > > - .depth = sub_rule_depth,
> > > - .valid_group =
> > > lpm->tbl8[tbl8_group_start].valid_group,
> > > + struct rte_lpm_tbl_entry new_tbl8_entry = {
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + .as_num =
> lpm->rules_tbl[sub_rule_index].as_num,
> > > +#endif
> > > + .fwd_class =
> > > lpm->rules_tbl[sub_rule_index].fwd_class,
> > > .next_hop =
> lpm->rules_tbl[sub_rule_index].next_hop,
> > > + .depth = sub_rule_depth,
> > > + .ext_valid =
> lpm->tbl8[tbl8_group_start].ext_valid,
> > > + .valid = VALID,
> > > };
> > >
> > > /*
> > > @@ -923,11 +963,15 @@ delete_depth_big(struct rte_lpm *lpm, uint32_t
> > > ip_masked,
> > > }
> > > else if (tbl8_recycle_index > -1) {
> > > /* Update tbl24 entry. */
> > > - struct rte_lpm_tbl24_entry new_tbl24_entry = {
> > > - { .next_hop =
> > > lpm->tbl8[tbl8_recycle_index].next_hop, },
> > > - .valid = VALID,
> > > - .ext_entry = 0,
> > > + struct rte_lpm_tbl_entry new_tbl24_entry = {
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + .as_num = lpm->tbl8[tbl8_recycle_index].as_num,
> > > +#endif
> > > + .next_hop =
> lpm->tbl8[tbl8_recycle_index].next_hop,
> > > + .fwd_class =
> > > lpm->tbl8[tbl8_recycle_index].fwd_class,
> > > .depth = lpm->tbl8[tbl8_recycle_index].depth,
> > > + .ext_valid = 0,
> > > + .valid = VALID,
> > > };
> > >
> > > /* Set tbl24 before freeing tbl8 to avoid race
> condition. */
> > > diff --git a/lib/librte_lpm/rte_lpm.h b/lib/librte_lpm/rte_lpm.h
> > > index c299ce2..7c615bc 100644
> > > --- a/lib/librte_lpm/rte_lpm.h
> > > +++ b/lib/librte_lpm/rte_lpm.h
> > > @@ -31,8 +31,8 @@
> > > * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
> > DAMAGE.
> > > */
> > >
> > > -#ifndef _RTE_LPM_H_
> > > -#define _RTE_LPM_H_
> > > +#ifndef _RTE_LPM_EXT_H_
> > > +#define _RTE_LPM_EXT_H_
> > >
> > > /**
> > > * @file
> > > @@ -81,57 +81,58 @@ extern "C" {
> > > #define RTE_LPM_RETURN_IF_TRUE(cond, retval)
> > > #endif
> > >
> > > -/** @internal bitmask with valid and ext_entry/valid_group fields set
> */
> > > -#define RTE_LPM_VALID_EXT_ENTRY_BITMASK 0x0300
> > > +/** @internal bitmask with valid and ext_valid/ext_valid fields set */
> > > +#define RTE_LPM_VALID_EXT_ENTRY_BITMASK 0x03
> > >
> > > /** Bitmask used to indicate successful lookup */
> > > -#define RTE_LPM_LOOKUP_SUCCESS 0x0100
> > > +#define RTE_LPM_LOOKUP_SUCCESS 0x01
> > > +
> > > +struct rte_lpm_res {
> > > + uint16_t next_hop;
> > > + uint8_t fwd_class;
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + uint32_t as_num;
> > > +#endif
> > > +};
> > >
> > > #if RTE_BYTE_ORDER == RTE_LITTLE_ENDIAN
> > > -/** @internal Tbl24 entry structure. */
> > > -struct rte_lpm_tbl24_entry {
> > > - /* Stores Next hop or group index (i.e. gindex)into tbl8. */
> > > +struct rte_lpm_tbl_entry {
> > > + uint8_t valid :1;
> > > + uint8_t ext_valid :1;
> > > + uint8_t depth :6;
> > > + uint8_t fwd_class;
> > > union {
> > > - uint8_t next_hop;
> > > - uint8_t tbl8_gindex;
> > > + uint16_t next_hop;
> > > + uint16_t tbl8_gindex;
> > > };
> > > - /* Using single uint8_t to store 3 values. */
> > > - uint8_t valid :1; /**< Validation flag. */
> > > - uint8_t ext_entry :1; /**< External entry. */
> > > - uint8_t depth :6; /**< Rule depth. */
> > > -};
> > > -
> > > -/** @internal Tbl8 entry structure. */
> > > -struct rte_lpm_tbl8_entry {
> > > - uint8_t next_hop; /**< next hop. */
> > > - /* Using single uint8_t to store 3 values. */
> > > - uint8_t valid :1; /**< Validation flag. */
> > > - uint8_t valid_group :1; /**< Group validation flag. */
> > > - uint8_t depth :6; /**< Rule depth. */
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + uint32_t as_num;
> > > +#endif
> > > };
> > > #else
> > > -struct rte_lpm_tbl24_entry {
> > > - uint8_t depth :6;
> > > - uint8_t ext_entry :1;
> > > - uint8_t valid :1;
> > > +struct rte_lpm_tbl_entry {
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + uint32_t as_num;
> > > +#endif
> > > union {
> > > - uint8_t tbl8_gindex;
> > > - uint8_t next_hop;
> > > + uint16_t tbl8_gindex;
> > > + uint16_t next_hop;
> > > };
> > > -};
> > > -
> > > -struct rte_lpm_tbl8_entry {
> > > - uint8_t depth :6;
> > > - uint8_t valid_group :1;
> > > - uint8_t valid :1;
> > > - uint8_t next_hop;
> > > + uint8_t fwd_class;
> > > + uint8_t depth :6;
> > > + uint8_t ext_valid :1;
> > > + uint8_t valid :1;
> > > };
> > > #endif
> > >
> > > /** @internal Rule structure. */
> > > struct rte_lpm_rule {
> > > uint32_t ip; /**< Rule IP address. */
> > > - uint8_t next_hop; /**< Rule next hop. */
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + uint32_t as_num;
> > > +#endif
> > > + uint16_t next_hop; /**< Rule next hop. */
> > > + uint8_t fwd_class;
> > > };
> > >
> > > /** @internal Contains metadata about the rules table. */
> > > @@ -148,9 +149,9 @@ struct rte_lpm {
> > > struct rte_lpm_rule_info rule_info[RTE_LPM_MAX_DEPTH]; /**<
> Rule
> > > info table. */
> > >
> > > /* LPM Tables. */
> > > - struct rte_lpm_tbl24_entry tbl24[RTE_LPM_TBL24_NUM_ENTRIES] \
> > > + struct rte_lpm_tbl_entry tbl24[RTE_LPM_TBL24_NUM_ENTRIES] \
> > > __rte_cache_aligned; /**< LPM tbl24 table. */
> > > - struct rte_lpm_tbl8_entry tbl8[RTE_LPM_TBL8_NUM_ENTRIES] \
> > > + struct rte_lpm_tbl_entry tbl8[RTE_LPM_TBL8_NUM_ENTRIES] \
> > > __rte_cache_aligned; /**< LPM tbl8 table. */
> > > struct rte_lpm_rule rules_tbl[0] \
> > > __rte_cache_aligned; /**< LPM rules. */
> > > @@ -219,7 +220,7 @@ rte_lpm_free(struct rte_lpm *lpm);
> > > * 0 on success, negative value otherwise
> > > */
> > > int
> > > -rte_lpm_add(struct rte_lpm *lpm, uint32_t ip, uint8_t depth, uint8_t
> > > next_hop);
> > > +rte_lpm_add(struct rte_lpm *lpm, uint32_t ip, uint8_t depth, struct
> > > rte_lpm_res *res);
> > >
> > > /**
> > > * Check if a rule is present in the LPM table,
> > > @@ -238,7 +239,7 @@ rte_lpm_add(struct rte_lpm *lpm, uint32_t ip,
> > uint8_t
> > > depth, uint8_t next_hop);
> > > */
> > > int
> > > rte_lpm_is_rule_present(struct rte_lpm *lpm, uint32_t ip, uint8_t
> depth,
> > > -uint8_t *next_hop);
> > > + struct rte_lpm_res *res);
> > >
> > > /**
> > > * Delete a rule from the LPM table.
> > > @@ -277,29 +278,43 @@ rte_lpm_delete_all(struct rte_lpm *lpm);
> > > * -EINVAL for incorrect arguments, -ENOENT on lookup miss, 0 on
> lookup
> > > hit
> > > */
> > > static inline int
> > > -rte_lpm_lookup(struct rte_lpm *lpm, uint32_t ip, uint8_t *next_hop)
> > > +rte_lpm_lookup(struct rte_lpm *lpm, uint32_t ip, struct rte_lpm_res
> *res)
> > > {
> > > unsigned tbl24_index = (ip >> 8);
> > > - uint16_t tbl_entry;
> > > -
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + uint64_t tbl_entry;
> > > +#else
> > > + uint32_t tbl_entry;
> > > +#endif
> > > /* DEBUG: Check user input arguments. */
> > > - RTE_LPM_RETURN_IF_TRUE(((lpm == NULL) || (next_hop == NULL)),
> > > -EINVAL);
> > > + RTE_LPM_RETURN_IF_TRUE(((lpm == NULL) || (res == NULL)), -
> > EINVAL);
> > >
> > > /* Copy tbl24 entry */
> > > - tbl_entry = *(const uint16_t *)&lpm->tbl24[tbl24_index];
> > > -
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + tbl_entry = *(const uint64_t *)&lpm->tbl24[tbl24_index];
> > > +#else
> > > + tbl_entry = *(const uint32_t *)&lpm->tbl24[tbl24_index];
> > > +#endif
> > > /* Copy tbl8 entry (only if needed) */
> > > if (unlikely((tbl_entry & RTE_LPM_VALID_EXT_ENTRY_BITMASK) ==
> > > RTE_LPM_VALID_EXT_ENTRY_BITMASK)) {
> > >
> > > unsigned tbl8_index = (uint8_t)ip +
> > > - ((uint8_t)tbl_entry *
> > > RTE_LPM_TBL8_GROUP_NUM_ENTRIES);
> > > + ((*(struct rte_lpm_tbl_entry
> > > *)&tbl_entry).tbl8_gindex * RTE_LPM_TBL8_GROUP_NUM_ENTRIES);
> > >
> > > - tbl_entry = *(const uint16_t *)&lpm->tbl8[tbl8_index];
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + tbl_entry = *(const uint64_t *)&lpm->tbl8[tbl8_index];
> > > +#else
> > > + tbl_entry = *(const uint32_t *)&lpm->tbl8[tbl8_index];
> > > +#endif
> > > }
> > > -
> > > - *next_hop = (uint8_t)tbl_entry;
> > > + res->next_hop = ((struct rte_lpm_tbl_entry
> *)&tbl_entry)->next_hop;
> > > + res->fwd_class = ((struct rte_lpm_tbl_entry
> > > *)&tbl_entry)->fwd_class;
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + res->as_num = ((struct rte_lpm_tbl_entry
> > > *)&tbl_entry)->as_num;
> > > +#endif
> > > return (tbl_entry & RTE_LPM_LOOKUP_SUCCESS) ? 0 : -ENOENT;
> > > +
> > > }
> > >
> > > /**
> > > @@ -322,19 +337,25 @@ rte_lpm_lookup(struct rte_lpm *lpm, uint32_t ip,
> > > uint8_t *next_hop)
> > > * @return
> > > * -EINVAL for incorrect arguments, otherwise 0
> > > */
> > > -#define rte_lpm_lookup_bulk(lpm, ips, next_hops, n) \
> > > - rte_lpm_lookup_bulk_func(lpm, ips, next_hops, n)
> > > +#define rte_lpm_lookup_bulk(lpm, ips, res_tbl, n) \
> > > + rte_lpm_lookup_bulk_func(lpm, ips, res_tbl, n)
> > >
> > > static inline int
> > > -rte_lpm_lookup_bulk_func(const struct rte_lpm *lpm, const uint32_t *
> ips,
> > > - uint16_t * next_hops, const unsigned n)
> > > +rte_lpm_lookup_bulk_func(const struct rte_lpm *lpm, const uint32_t
> *ips,
> > > + struct rte_lpm_res *res_tbl, const unsigned n)
> > > {
> > > unsigned i;
> > > + int ret = 0;
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + uint64_t tbl_entry;
> > > +#else
> > > + uint32_t tbl_entry;
> > > +#endif
> > > unsigned tbl24_indexes[n];
> > >
> > > /* DEBUG: Check user input arguments. */
> > > RTE_LPM_RETURN_IF_TRUE(((lpm == NULL) || (ips == NULL) ||
> > > - (next_hops == NULL)), -EINVAL);
> > > + (res_tbl == NULL)), -EINVAL);
> > >
> > > for (i = 0; i < n; i++) {
> > > tbl24_indexes[i] = ips[i] >> 8;
> > > @@ -342,20 +363,32 @@ rte_lpm_lookup_bulk_func(const struct rte_lpm
> > *lpm,
> > > const uint32_t * ips,
> > >
> > > for (i = 0; i < n; i++) {
> > > /* Simply copy tbl24 entry to output */
> > > - next_hops[i] = *(const uint16_t
> > > *)&lpm->tbl24[tbl24_indexes[i]];
> > > -
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + tbl_entry = *(const uint64_t
> > > *)&lpm->tbl24[tbl24_indexes[i]];
> > > +#else
> > > + tbl_entry = *(const uint32_t
> > > *)&lpm->tbl24[tbl24_indexes[i]];
> > > +#endif
> > > /* Overwrite output with tbl8 entry if needed */
> > > - if (unlikely((next_hops[i] &
> > > RTE_LPM_VALID_EXT_ENTRY_BITMASK) ==
> > > - RTE_LPM_VALID_EXT_ENTRY_BITMASK)) {
> > > + if (unlikely((tbl_entry &
> RTE_LPM_VALID_EXT_ENTRY_BITMASK)
> > > ==
> > > + RTE_LPM_VALID_EXT_ENTRY_BITMASK)) {
> > >
> > > unsigned tbl8_index = (uint8_t)ips[i] +
> > > - ((uint8_t)next_hops[i] *
> > > -
> RTE_LPM_TBL8_GROUP_NUM_ENTRIES);
> > > + ((*(struct rte_lpm_tbl_entry
> > > *)&tbl_entry).tbl8_gindex * RTE_LPM_TBL8_GROUP_NUM_ENTRIES);
> > >
> > > - next_hops[i] = *(const uint16_t
> > > *)&lpm->tbl8[tbl8_index];
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + tbl_entry = *(const uint64_t
> > > *)&lpm->tbl8[tbl8_index];
> > > +#else
> > > + tbl_entry = *(const uint32_t
> > > *)&lpm->tbl8[tbl8_index];
> > > +#endif
> > > }
> > > + res_tbl[i].next_hop = ((struct rte_lpm_tbl_entry
> > > *)&tbl_entry)->next_hop;
> > > + res_tbl[i].fwd_class = ((struct rte_lpm_tbl_entry
> > > *)&tbl_entry)->next_hop;
> > > +#ifdef RTE_LIBRTE_LPM_ASNUM
> > > + res_tbl[i].as_num = ((struct rte_lpm_tbl_entry
> > > *)&tbl_entry)->as_num;
> > > +#endif
> > > + ret |= 1 << i;
> > > }
> > > - return 0;
> > > + return ret;
> > > }
> > >
> > > /* Mask four results. */
> > > @@ -477,4 +510,4 @@ rte_lpm_lookupx4(const struct rte_lpm *lpm,
> > __m128i ip,
> > > uint16_t hop[4],
> > > }
> > > #endif
> > >
> > > -#endif /* _RTE_LPM_H_ */
> > > +#endif /* _RTE_LPM_EXT_H_ */
> > >
> > > 2015-10-24 9:09 GMT+03:00 Matthew Hall <mhall at mhcomputing.net>:
> > >
> > > > On 10/23/15 9:20 AM, Matthew Hall wrote:
> > > >
> > > >> On Fri, Oct 23, 2015 at 03:51:48PM +0200, Michal Jastrzebski wrote:
> > > >>
> > > >>> From: Michal Kobylinski <michalx.kobylinski at intel.com>
> > > >>>
> > > >>> The current DPDK implementation for LPM for IPv4 and IPv6 limits
> the
> > > >>> number of next hops to 256, as the next hop ID is an 8-bit long
> field.
> > > >>> Proposed extension increase number of next hops for IPv4 to 2^24
> and
> > > >>> also allows 32-bits read/write operations.
> > > >>>
> > > >>> This patchset requires additional change to rte_table library to
> meet
> > > >>> ABI compatibility requirements. A v2 will be sent next week.
> > > >>>
> > > >>
> > > >> I also have a patchset for this.
> > > >>
> > > >> I will send it out as well so we could compare.
> > > >>
> > > >> Matthew.
> > > >>
> > > >
> > > > Sorry about the delay; I only work on DPDK in personal time and not
> as
> > > > part of a job. My patchset is attached to this email.
> > > >
> > > > One possible advantage with my patchset, compared to others, is that
> the
> > > > space problem is fixed in both IPV4 and in IPV6, to prevent asymmetry
> > > > between these two standards, which is something I try to avoid as
> much
> > as
> > > > humanly possible.
> > > >
> > > > This is because my application code is green-field, so I absolutely
> don't
> > > > want to put any ugly hacks or incompatibilities in this code if I can
> > > > possibly avoid it.
> > > >
> > > > Otherwise, I am not necessarily as expert about rte_lpm as some of
> the
> > > > full-time guys, but I think with four or five of us in the thread
> hammering
> > > > out patches we will be able to create something amazing together and
> I
> > am
> > > > very very very very very happy about this.
> > > >
> > > > Matthew.
> > > >
> >
>
> Hi Vladimir,
> Thanks for sharing Your implementation.
> Could You please clarify what as_num and fwd_class fields represent?
> The second issue I have is that Your patch doesn’t want to apply on top of
> current head. Could You check this please?
>
> Best regards
> Michal
>
More information about the dev
mailing list