| Message ID | 1340022131-7827-1-git-send-email-lisovy@gmail.com |
|---|---|
| State | Awaiting Upstream, archived |
| Delegated to: | David Miller |
| Headers | show |
Hello Rostislav, thanks for the new patch. It got indeed pretty short now :-) I found some time for a review. See details inline ... On 18.06.2012 14:22, Rostislav Lisovy wrote: > > With no extra filter/qdisc configured, median of the time spent in can_send() > was about 27 us -- with prio qdisc with 5 bands and 5 appropriate cls_can > filters (previous patch), it was about 30 us -- with prio qdisc with 5 bands > and 5 appropriate em_can filters (this patch), it was about 34 us. Hm that's more than twice the time consumed for classification ... cls_can: 3 us more em_can: 7 us more @Eric: Is this still the better approach then? > diff --git a/net/sched/Kconfig b/net/sched/Kconfig > index 75b58f8..bc0ceab 100644 > --- a/net/sched/Kconfig > +++ b/net/sched/Kconfig > @@ -485,6 +485,16 @@ config NET_EMATCH_TEXT > To compile this code as a module, choose M here: the > module will be called em_text. > > +config NET_EMATCH_CANID > + tristate "CAN ID" i would prefer tristate "Controller Area Network Identifier" or at least tristate "CAN Identifier" > + depends on NET_EMATCH && CAN > + ---help--- > + Say Y here if you want to be able to classify CAN frames based > + on CAN ID. "on the CAN Identifier." (..) > +#include <net/pkt_cls.h> > +#include <linux/can.h> > + > +#define EM_CAN_RULES_SIZE 32 Reduce the gap before '32' to one single space. (..) > +static int em_canid_match(struct sk_buff *skb, struct tcf_ematch *m, > + struct tcf_pkt_info *info) > +{ > + struct canid_match *cm = em_canid_priv(m); > + canid_t can_id; > + unsigned int match = false; You use test_bit() below which returns an int. Better use int instead of unsigned int ... int match = 0; > + int i; > + > + can_id = em_canid_get_id(skb); > + > + if (can_id & CAN_EFF_FLAG) { > + can_id &= CAN_EFF_MASK; > + > + for (i = 0; i < cm->eff_rules_count; i++) { > + if (!(((cm->rules_raw[i].can_id ^ can_id) & > + cm->rules_raw[i].can_mask) & CAN_EFF_MASK)) { Looks really tricky. I'm still pondering if it does what it should do ... > + match = true; match = 1; > + break; > + } > + } > + } else { /* SFF */ > + can_id &= CAN_SFF_MASK; > + match = test_bit(can_id, cm->match_sff); > + } > + return match; > + if (match) > + return 1; > + > + return 0; > +} > + > +static int em_canid_change(struct tcf_proto *tp, void *data, int len, > + struct tcf_ematch *m) > +{ > + struct can_filter *conf = data; /* Array with rules, > + * fixed size EM_CAN_RULES_SIZE > + */ > + struct canid_match *cm; > + int err; > + int i; > + > + if (len < sizeof(struct can_filter)) > + return -EINVAL; if (len % sizeof(struct can_filter)) return -EINVAL; if (len > sizeof(struct can_filter) * EM_CAN_RULES_SIZE) return -EINVAL; All checks done before kzalloc() => no need for error cleanups at the end of the function. > + > + err = -ENOBUFS; remove > + cm = kzalloc(sizeof(*cm), GFP_KERNEL); > + if (cm == NULL) if (!cm) return -ENOMEM; > + goto errout; The only user of errout - to be removed. > + > + cm->sff_rules_count = 0; > + cm->eff_rules_count = 0; > + cm->rules_count = len/sizeof(struct can_filter); > + err = -EINVAL; remove > + > + /* Be sure to fit into the array */ > + if (cm->rules_count > EM_CAN_RULES_SIZE) > + goto errout_free; already checked before => remove > + > + /* > + * We need two for() loops for copying rules into > + * two contiguous areas in rules_raw > + */ > + > + /* Process EFF frame rules*/ > + for (i = 0; i < cm->rules_count; i++) { > + if ((conf[i].can_id & CAN_EFF_FLAG) && > + (conf[i].can_mask & CAN_EFF_FLAG)) { > + memcpy(cm->rules_raw + cm->eff_rules_count, Oops. Shouldn't this be cm->rules_raw + cm->eff_rules_count * sizeof(struct can_filter), ??? > + &conf[i], > + sizeof(struct can_filter)); > + > + cm->eff_rules_count++; > + } else { > + continue; > + } omit { } around continue > + } > + > + /* Process SFF frame rules */ > + for (i = 0; i < cm->rules_count; i++) { > + if ((conf[i].can_id & CAN_EFF_FLAG) && > + (conf[i].can_mask & CAN_EFF_FLAG)) { What if CAN_EFF_FLAG is set in can_id but not in can_mask ? > + continue; > + } else { > + memcpy(cm->rules_raw > + + cm->eff_rules_count > + + cm->sff_rules_count, dito cm->rules_raw + (cm->eff_rules_count + cm->sff_rules_count) * sizeof(struct can_filter), ??? > + &conf[i], sizeof(struct can_filter)); > + > + cm->sff_rules_count++; > + > + em_canid_sff_match_add(cm, > + conf[i].can_id, conf[i].can_mask); > + } > + } > + > + m->datalen = sizeof(*cm); > + m->data = (unsigned long) cm; > + > + return 0; > + > +errout_free: > + kfree(cm); > +errout: > + return err; error handling can be removed with the above changes. > +} > + > +static void em_canid_destroy(struct tcf_proto *tp, struct tcf_ematch *m) > +{ > + struct canid_match *cm = em_canid_priv(m); > + Check for cm == NULL not needed ? > + kfree(cm); > +} > + > +static int em_canid_dump(struct sk_buff *skb, struct tcf_ematch *m) > +{ > + struct canid_match *cm = em_canid_priv(m); > + Check for cm == NULL not needed ? Can a dump happen before the matches are added?? > + /* > + * When configuring this ematch 'rules_count' is set not to exceed > + * 'rules_raw' array size > + */ > + if (nla_put_nohdr(skb, sizeof(cm->rules_raw[0]) * cm->rules_count, better sizeof(struct can_filter) instead of sizeof(cm->rules_raw[0]) ?? > + &cm->rules_raw) < 0) > + goto nla_put_failure; > + > + return 0; > + > +nla_put_failure: > + return -1; > +} > + > +static struct tcf_ematch_ops em_canid_ops = { > + .kind = TCF_EM_CANID, > + .change = em_canid_change, > + .match = em_canid_match, > + .destroy = em_canid_destroy, > + .dump = em_canid_dump, > + .owner = THIS_MODULE, > + .link = LIST_HEAD_INIT(em_canid_ops.link) > +}; > + > +static int __init init_em_canid(void) > +{ > + return tcf_em_register(&em_canid_ops); > +} > + > +static void __exit exit_em_canid(void) > +{ > + tcf_em_unregister(&em_canid_ops); > +} > + > +MODULE_LICENSE("GPL"); > + > +module_init(init_em_canid); > +module_exit(exit_em_canid); > + > +MODULE_ALIAS_TCF_EMATCH(TCF_EM_CANID); Regards, Oliver -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, Jun 26, 2012 at 10:07:56PM +0200, Oliver Hartkopp wrote: > > With no extra filter/qdisc configured, median of the time spent in can_send() > > was about 27 us -- with prio qdisc with 5 bands and 5 appropriate cls_can > > filters (previous patch), it was about 30 us -- with prio qdisc with 5 bands > > and 5 appropriate em_can filters (this patch), it was about 34 us. > > > Hm that's more than twice the time consumed for classification ... > > cls_can: 3 us more > em_can: 7 us more > > @Eric: Is this still the better approach then? If there is overhead, we should get rid of that overhead and not abandon an established subsystem. Rostislav: Can you provide some details on where the time is spent? > > + /* Process EFF frame rules*/ > > + for (i = 0; i < cm->rules_count; i++) { > > + if ((conf[i].can_id & CAN_EFF_FLAG) && > > + (conf[i].can_mask & CAN_EFF_FLAG)) { > > + memcpy(cm->rules_raw + cm->eff_rules_count, > > > Oops. Shouldn't this be > > cm->rules_raw + cm->eff_rules_count * sizeof(struct can_filter), > > ??? Looks like correct pointer arithmetic to me. Your suggestion would only be valid if rules_raw was a void pointer. > > +static void em_canid_destroy(struct tcf_proto *tp, struct tcf_ematch *m) > > +{ > > + struct canid_match *cm = em_canid_priv(m); > > + > > > Check for cm == NULL not needed ? kfree() has that check embeddded. Also, for destroy() can only be called if the match was added to the tree and that requires a successful call to ->change(). Therefore it will never be NULL. > > +static int em_canid_dump(struct sk_buff *skb, struct tcf_ematch *m) > > +{ > > + struct canid_match *cm = em_canid_priv(m); > > + > > > Check for cm == NULL not needed ? > > Can a dump happen before the matches are added?? Nope, ->dump() is only ever called if the match has been added to the tree. > > + /* > > + * When configuring this ematch 'rules_count' is set not to exceed > > + * 'rules_raw' array size > > + */ > > + if (nla_put_nohdr(skb, sizeof(cm->rules_raw[0]) * cm->rules_count, > > > better sizeof(struct can_filter) instead of sizeof(cm->rules_raw[0]) ?? > > > + &cm->rules_raw) < 0) > > + goto nla_put_failure; No need for a goto here, just return -EMSGSIZE. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Oliver, Rostislav, I was just looking into this. I think the matching itself may be simplified by eliminating checks 'that have already been made'. On Tue, Jun 26, 2012 at 10:07:56PM +0200, Oliver Hartkopp wrote: > > > +static int em_canid_match(struct sk_buff *skb, struct tcf_ematch *m, > > + struct tcf_pkt_info *info) > > +{ > > + struct canid_match *cm = em_canid_priv(m); > > + canid_t can_id; > > + unsigned int match = false; > > + int i; > > + > > + can_id = em_canid_get_id(skb); > > + > > + if (can_id & CAN_EFF_FLAG) { > > + can_id &= CAN_EFF_MASK; Why clear the CAN_EFF_FLAG? It needs an extra read-modify-write, and the CAN_EFF_FLAG is set in the match rule too. IMHO, you could leave can_id as it is. > > + > > + for (i = 0; i < cm->eff_rules_count; i++) { to speed things up manually, I tend to use a 'const struct can_filter *lp' and do: for (i = 0, lp = cm->rules_raw; i < cm->eff_rules_count; ++i, ++lp) { The advantage depends on the compiler's optimizations, which I'm not really aware of. The next statement would then be: if (!((lp->can_id ^ can_id) & lp->can_mask)) { for stripping & CAN_EFF_MASK, see below :-) > > + if (!(((cm->rules_raw[i].can_id ^ can_id) & > > + cm->rules_raw[i].can_mask) & CAN_EFF_MASK)) { > > > Looks really tricky. I'm still pondering if it does what it should do ... I think it does, since: cm->rules_raw[i].can_id ^ can_id gives an value where the bits that differ are set. then: & cm->rules_raw[i].can_mask will strip bits that you don't care. But '& CAN_EFF_MASK' is not needed, since: * can_id will have CAN_EFF_FLAG (see comment above) * cm->rules_raw[i].can_id has CAN_EFF_FLAG, otherwise it would not be in the list * can_id will not have CAN_ERR_FLAG * cm->rules_raw should not have CAN_ERR_FLAG you can always clear CAN_ERR_FLAG from the rules during em_canid_change below * maybe distinguishing on CAN_RTR_FLAG makes sense during classification. > > > + match = true; > > + break; > > + } > > + } > > + } else { /* SFF */ > > + can_id &= CAN_SFF_MASK; > > + match = test_bit(can_id, cm->match_sff); > > + } > > + > > > return match; > > > > +} > > + -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Hello Oliver; On Tue, 2012-06-26 at 22:07 +0200, Oliver Hartkopp wrote: > I found some time for a review. See details inline ... > I agree with quite everything except for following... > > + match = true; > > > match = 1; > egrep -r "= true;" ./linux-source | wc -l returns 6770 -- why don't you like "= true"? > > + break; > > + } > > + } > > + } else { /* SFF */ > > + can_id &= CAN_SFF_MASK; > > + match = test_bit(can_id, cm->match_sff); > > + } > > + > > > return match; > match() function must return 1 or 0, however (from my experience) test_bit() returns 0 and non-0 (strictly speaking, in my case, 0 and -1). > > + &conf[i], > > + sizeof(struct can_filter)); > > + > > + cm->eff_rules_count++; > > + } else { > > + continue; > > + } > > > omit { } around continue > http://lxr.linux.no/#linux+v3.4.4/Documentation/CodingStyle#L169 > > + } > > + > > + /* Process SFF frame rules */ > > + for (i = 0; i < cm->rules_count; i++) { > > + if ((conf[i].can_id & CAN_EFF_FLAG) && > > + (conf[i].can_mask & CAN_EFF_FLAG)) { > > > What if CAN_EFF_FLAG is set in can_id but not in can_mask ? > There were small misunderstanding from my side -- this will be rewritten in the way that EFF_FLAG in mask will determine if we care about the value of EFF_FLAG in an identifier -- i.e. when EFF_FLAG is set in the mask, the rule will be added as SFF or EFF only depending on EFF_FLAG value in the identifier. If EFF_FLAG is 0 in the mask, the rule will be added as both SFF and EFF. Regards; Rostislav -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, 2012-06-27 at 11:33 +0200, Kurt Van Dijck wrote: > Oliver, Rostislav, > > I was just looking into this. I think the matching itself > may be simplified by eliminating checks 'that have already been made'. Thank you for your remarks, I have removed all of the unnecessary CAN_EFF_MASK masking. Regards; Rostislav -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/include/linux/can.h b/include/linux/can.h index 9a19bcb..08d1610 100644 --- a/include/linux/can.h +++ b/include/linux/can.h @@ -38,6 +38,9 @@ */ typedef __u32 canid_t; +#define CAN_SFF_ID_BITS 11 +#define CAN_EFF_ID_BITS 29 + /* * Controller Area Network Error Frame Mask structure * diff --git a/include/linux/pkt_cls.h b/include/linux/pkt_cls.h index defbde2..7fbe6c2 100644 --- a/include/linux/pkt_cls.h +++ b/include/linux/pkt_cls.h @@ -451,8 +451,9 @@ enum { #define TCF_EM_U32 3 #define TCF_EM_META 4 #define TCF_EM_TEXT 5 -#define TCF_EM_VLAN 6 -#define TCF_EM_MAX 6 +#define TCF_EM_VLAN 6 +#define TCF_EM_CANID 7 +#define TCF_EM_MAX 7 enum { TCF_EM_PROG_TC diff --git a/net/sched/Kconfig b/net/sched/Kconfig index 75b58f8..bc0ceab 100644 --- a/net/sched/Kconfig +++ b/net/sched/Kconfig @@ -485,6 +485,16 @@ config NET_EMATCH_TEXT To compile this code as a module, choose M here: the module will be called em_text. +config NET_EMATCH_CANID + tristate "CAN ID" + depends on NET_EMATCH && CAN + ---help--- + Say Y here if you want to be able to classify CAN frames based + on CAN ID. + + To compile this code as a module, choose M here: the + module will be called em_canid. + config NET_CLS_ACT bool "Actions" ---help--- diff --git a/net/sched/Makefile b/net/sched/Makefile index 8cdf4e2..47f9262 100644 --- a/net/sched/Makefile +++ b/net/sched/Makefile @@ -53,3 +53,4 @@ obj-$(CONFIG_NET_EMATCH_NBYTE) += em_nbyte.o obj-$(CONFIG_NET_EMATCH_U32) += em_u32.o obj-$(CONFIG_NET_EMATCH_META) += em_meta.o obj-$(CONFIG_NET_EMATCH_TEXT) += em_text.o +obj-$(CONFIG_NET_EMATCH_CANID) += em_canid.o diff --git a/net/sched/em_canid.c b/net/sched/em_canid.c new file mode 100644 index 0000000..5cc6e5e --- /dev/null +++ b/net/sched/em_canid.c @@ -0,0 +1,252 @@ +/* + * em_canid.c Ematch rule to match CAN frames according to their CAN IDs + * + * This program is free software; you can distribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version + * 2 of the License, or (at your option) any later version. + * + * Idea: Oliver Hartkopp <oliver.hartkopp@volkswagen.de> + * Copyright: (c) 2011 Czech Technical University in Prague + * (c) 2011 Volkswagen Group Research + * Authors: Michal Sojka <sojkam1@fel.cvut.cz> + * Pavel Pisa <pisa@cmp.felk.cvut.cz> + * Rostislav Lisovy <lisovy@gmail.cz> + * Funded by: Volkswagen Group Research + */ + +#include <linux/slab.h> +#include <linux/module.h> +#include <linux/types.h> +#include <linux/kernel.h> +#include <linux/string.h> +#include <linux/skbuff.h> +#include <net/pkt_cls.h> +#include <linux/can.h> + +#define EM_CAN_RULES_SIZE 32 + +struct canid_match { + /* + * Raw rules copied from netlink message; Used for sending + * information to userspace (when 'tc filter show' is invoked) + * AND when matching EFF frames + */ + struct can_filter rules_raw[EM_CAN_RULES_SIZE]; + + /* For each SFF CAN ID (11 bit) there is one record in this bitfield */ + DECLARE_BITMAP(match_sff, (1 << CAN_SFF_ID_BITS)); + + int rules_count; + int eff_rules_count; + int sff_rules_count; +}; + +/** + * em_canid_get_id() - Extracts Can ID out of the sk_buff structure. + */ +static canid_t em_canid_get_id(struct sk_buff *skb) +{ + /* CAN ID is stored within the data field */ + struct can_frame *cf = (struct can_frame *)skb->data; + + return cf->can_id; +} + +static void em_canid_sff_match_add(struct canid_match *cm, u32 can_id, + u32 can_mask) +{ + int i; + + /* + * Limit can_mask and can_id to SFF range to + * protect against write after end of array + */ + can_mask &= CAN_SFF_MASK; + can_id &= can_mask; + + /* Single frame */ + if (can_mask == CAN_SFF_MASK) { + set_bit(can_id, cm->match_sff); + return; + } + + /* All frames */ + if (can_mask == 0) { + bitmap_fill(cm->match_sff, (1 << CAN_SFF_ID_BITS)); + return; + } + + /* + * Individual frame filter. + * Add record (set bit to 1) for each ID that + * conforms particular rule + */ + for (i = 0; i < (1 << CAN_SFF_ID_BITS); i++) { + if ((i & can_mask) == can_id) + set_bit(i, cm->match_sff); + } +} + +static inline struct canid_match *em_canid_priv(struct tcf_ematch *m) +{ + return (struct canid_match *) (m)->data; +} + +static int em_canid_match(struct sk_buff *skb, struct tcf_ematch *m, + struct tcf_pkt_info *info) +{ + struct canid_match *cm = em_canid_priv(m); + canid_t can_id; + unsigned int match = false; + int i; + + can_id = em_canid_get_id(skb); + + if (can_id & CAN_EFF_FLAG) { + can_id &= CAN_EFF_MASK; + + for (i = 0; i < cm->eff_rules_count; i++) { + if (!(((cm->rules_raw[i].can_id ^ can_id) & + cm->rules_raw[i].can_mask) & CAN_EFF_MASK)) { + match = true; + break; + } + } + } else { /* SFF */ + can_id &= CAN_SFF_MASK; + match = test_bit(can_id, cm->match_sff); + } + + if (match) + return 1; + + return 0; +} + +static int em_canid_change(struct tcf_proto *tp, void *data, int len, + struct tcf_ematch *m) +{ + struct can_filter *conf = data; /* Array with rules, + * fixed size EM_CAN_RULES_SIZE + */ + struct canid_match *cm; + int err; + int i; + + if (len < sizeof(struct can_filter)) + return -EINVAL; + + err = -ENOBUFS; + cm = kzalloc(sizeof(*cm), GFP_KERNEL); + if (cm == NULL) + goto errout; + + cm->sff_rules_count = 0; + cm->eff_rules_count = 0; + cm->rules_count = len/sizeof(struct can_filter); + err = -EINVAL; + + /* Be sure to fit into the array */ + if (cm->rules_count > EM_CAN_RULES_SIZE) + goto errout_free; + + /* + * We need two for() loops for copying rules into + * two contiguous areas in rules_raw + */ + + /* Process EFF frame rules*/ + for (i = 0; i < cm->rules_count; i++) { + if ((conf[i].can_id & CAN_EFF_FLAG) && + (conf[i].can_mask & CAN_EFF_FLAG)) { + memcpy(cm->rules_raw + cm->eff_rules_count, + &conf[i], + sizeof(struct can_filter)); + + cm->eff_rules_count++; + } else { + continue; + } + } + + /* Process SFF frame rules */ + for (i = 0; i < cm->rules_count; i++) { + if ((conf[i].can_id & CAN_EFF_FLAG) && + (conf[i].can_mask & CAN_EFF_FLAG)) { + continue; + } else { + memcpy(cm->rules_raw + + cm->eff_rules_count + + cm->sff_rules_count, + &conf[i], sizeof(struct can_filter)); + + cm->sff_rules_count++; + + em_canid_sff_match_add(cm, + conf[i].can_id, conf[i].can_mask); + } + } + + m->datalen = sizeof(*cm); + m->data = (unsigned long) cm; + + return 0; + +errout_free: + kfree(cm); +errout: + return err; +} + +static void em_canid_destroy(struct tcf_proto *tp, struct tcf_ematch *m) +{ + struct canid_match *cm = em_canid_priv(m); + + kfree(cm); +} + +static int em_canid_dump(struct sk_buff *skb, struct tcf_ematch *m) +{ + struct canid_match *cm = em_canid_priv(m); + + /* + * When configuring this ematch 'rules_count' is set not to exceed + * 'rules_raw' array size + */ + if (nla_put_nohdr(skb, sizeof(cm->rules_raw[0]) * cm->rules_count, + &cm->rules_raw) < 0) + goto nla_put_failure; + + return 0; + +nla_put_failure: + return -1; +} + +static struct tcf_ematch_ops em_canid_ops = { + .kind = TCF_EM_CANID, + .change = em_canid_change, + .match = em_canid_match, + .destroy = em_canid_destroy, + .dump = em_canid_dump, + .owner = THIS_MODULE, + .link = LIST_HEAD_INIT(em_canid_ops.link) +}; + +static int __init init_em_canid(void) +{ + return tcf_em_register(&em_canid_ops); +} + +static void __exit exit_em_canid(void) +{ + tcf_em_unregister(&em_canid_ops); +} + +MODULE_LICENSE("GPL"); + +module_init(init_em_canid); +module_exit(exit_em_canid); + +MODULE_ALIAS_TCF_EMATCH(TCF_EM_CANID);
This ematch makes it possible to classify CAN frames (AF_CAN) according to their identifiers. This functionality can not be easily achieved with existing classifiers, such as u32, because CAN ID is always stored in native endianness, whereas u32 expects Network byte order. The filtering rules for EFF frames are stored in an array, which is traversed during classification. A bitmap is used to store SFF rules -- one bit for each ID. It is possible to to pass up to 32 'rules' to this ematch during configuration. Signed-off-by: Rostislav Lisovy <lisovy@gmail.com> --- This Patch contains a reworked classifier initially posted in http://www.spinics.net/lists/netdev/msg200114.html The functionality is the same however there is almost 50% reduction in the source code length. There were simple benchmark performed on MPC5200 -- an embedded PowerPC CPU (e300 core, G2 LE), 396 MHz, with 128 MiB of RAM running 3.4.2 Linux kernel The benchmark simply generated CAN frames with different identifiers and the time spent in can_send() function was measured. CAN device was configured as follows: ip link set can0 type can bitrate 1000000 ip link set can0 txqueuelen 1000 CAN frames were generated with command: cangen can0 -I ${ID} -L 8 -D i -g 1 -n 100 With no extra filter/qdisc configured, median of the time spent in can_send() was about 27 us -- with prio qdisc with 5 bands and 5 appropriate cls_can filters (previous patch), it was about 30 us -- with prio qdisc with 5 bands and 5 appropriate em_can filters (this patch), it was about 34 us. --- include/linux/can.h | 3 + include/linux/pkt_cls.h | 5 +- net/sched/Kconfig | 10 ++ net/sched/Makefile | 1 + net/sched/em_canid.c | 252 +++++++++++++++++++++++++++++++++++++++++++++++ 5 files changed, 269 insertions(+), 2 deletions(-) create mode 100644 net/sched/em_canid.c