This little number makes all of the flow dissection functions take
raw input data pointer as const (1-5) and shuffles the branches in
__skb_header_pointer() according to their hit probability.
The result is +20 Mbps per flow/core with one Flow Dissector pass
per packet. This affects RPS (with software hashing), drivers that
use eth_get_headlen() on their Rx path and so on.
Since v1 [0]:
- rebase on top of the latest net-next. This was super-weird, but
I double-checked that the series applies with no conflicts, and
then on Patchwork it didn't;
- no other changes.
[0] https://lore.kernel.org/netdev/[email protected]
Alexander Lobakin (6):
flow_dissector: constify bpf_flow_dissector's data pointers
skbuff: make __skb_header_pointer()'s data argument const
flow_dissector: constify raw input @data argument
linux/etherdevice.h: misc trailing whitespace cleanup
ethernet: constify eth_get_headlen()'s @data argument
skbuff: micro-optimize {,__}skb_header_pointer()
include/linux/etherdevice.h | 4 ++--
include/linux/skbuff.h | 26 +++++++++++------------
include/net/flow_dissector.h | 6 +++---
net/core/flow_dissector.c | 41 +++++++++++++++++++-----------------
net/ethernet/eth.c | 2 +-
5 files changed, 40 insertions(+), 39 deletions(-)
--
2.30.2
BPF Flow dissection programs are read-only and don't touch input
buffers.
Mark @data and @data_end in struct bpf_flow_dissector as const in
preparation for global input constifying.
Signed-off-by: Alexander Lobakin <[email protected]>
---
include/net/flow_dissector.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/include/net/flow_dissector.h b/include/net/flow_dissector.h
index cc10b10dc3a1..bf00e71816ed 100644
--- a/include/net/flow_dissector.h
+++ b/include/net/flow_dissector.h
@@ -368,8 +368,8 @@ static inline void *skb_flow_dissector_target(struct flow_dissector *flow_dissec
struct bpf_flow_dissector {
struct bpf_flow_keys *flow_keys;
const struct sk_buff *skb;
- void *data;
- void *data_end;
+ const void *data;
+ const void *data_end;
};
static inline void
--
2.30.2
The function never modifies the input buffer, so @data argument
can be marked as const.
This implies one harmless cast-away.
Signed-off-by: Alexander Lobakin <[email protected]>
---
include/linux/skbuff.h | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 483e89348f78..d6ea3dc3eddb 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -3678,11 +3678,11 @@ __wsum skb_checksum(const struct sk_buff *skb, int offset, int len,
__wsum csum);
static inline void * __must_check
-__skb_header_pointer(const struct sk_buff *skb, int offset,
- int len, void *data, int hlen, void *buffer)
+__skb_header_pointer(const struct sk_buff *skb, int offset, int len,
+ const void *data, int hlen, void *buffer)
{
if (hlen - offset >= len)
- return data + offset;
+ return (void *)data + offset;
if (!skb ||
skb_copy_bits(skb, offset, buffer, len) < 0)
--
2.30.2
Caught by the text editor. Fix it separately from the actual changes.
Signed-off-by: Alexander Lobakin <[email protected]>
---
include/linux/etherdevice.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/include/linux/etherdevice.h b/include/linux/etherdevice.h
index 2e5debc0373c..bcb2f81baafb 100644
--- a/include/linux/etherdevice.h
+++ b/include/linux/etherdevice.h
@@ -11,7 +11,7 @@
* Authors: Ross Biro
* Fred N. van Kempen, <[email protected]>
*
- * Relocated to include/linux where it belongs by Alan Cox
+ * Relocated to include/linux where it belongs by Alan Cox
* <[email protected]>
*/
#ifndef _LINUX_ETHERDEVICE_H
--
2.30.2
Flow Dissector code never modifies the input buffer, neither skb nor
raw data.
Make @data argument const for all of the Flow dissector's functions.
Signed-off-by: Alexander Lobakin <[email protected]>
---
include/linux/skbuff.h | 15 ++++++-------
include/net/flow_dissector.h | 2 +-
net/core/flow_dissector.c | 41 +++++++++++++++++++-----------------
3 files changed, 30 insertions(+), 28 deletions(-)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index d6ea3dc3eddb..46c61e127e9f 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -1292,10 +1292,10 @@ __skb_set_sw_hash(struct sk_buff *skb, __u32 hash, bool is_l4)
void __skb_get_hash(struct sk_buff *skb);
u32 __skb_get_hash_symmetric(const struct sk_buff *skb);
u32 skb_get_poff(const struct sk_buff *skb);
-u32 __skb_get_poff(const struct sk_buff *skb, void *data,
+u32 __skb_get_poff(const struct sk_buff *skb, const void *data,
const struct flow_keys_basic *keys, int hlen);
__be32 __skb_flow_get_ports(const struct sk_buff *skb, int thoff, u8 ip_proto,
- void *data, int hlen_proto);
+ const void *data, int hlen_proto);
static inline __be32 skb_flow_get_ports(const struct sk_buff *skb,
int thoff, u8 ip_proto)
@@ -1314,9 +1314,8 @@ bool bpf_flow_dissect(struct bpf_prog *prog, struct bpf_flow_dissector *ctx,
bool __skb_flow_dissect(const struct net *net,
const struct sk_buff *skb,
struct flow_dissector *flow_dissector,
- void *target_container,
- void *data, __be16 proto, int nhoff, int hlen,
- unsigned int flags);
+ void *target_container, const void *data,
+ __be16 proto, int nhoff, int hlen, unsigned int flags);
static inline bool skb_flow_dissect(const struct sk_buff *skb,
struct flow_dissector *flow_dissector,
@@ -1338,9 +1337,9 @@ static inline bool skb_flow_dissect_flow_keys(const struct sk_buff *skb,
static inline bool
skb_flow_dissect_flow_keys_basic(const struct net *net,
const struct sk_buff *skb,
- struct flow_keys_basic *flow, void *data,
- __be16 proto, int nhoff, int hlen,
- unsigned int flags)
+ struct flow_keys_basic *flow,
+ const void *data, __be16 proto,
+ int nhoff, int hlen, unsigned int flags)
{
memset(flow, 0, sizeof(*flow));
return __skb_flow_dissect(net, skb, &flow_keys_basic_dissector, flow,
diff --git a/include/net/flow_dissector.h b/include/net/flow_dissector.h
index bf00e71816ed..ffd386ea0dbb 100644
--- a/include/net/flow_dissector.h
+++ b/include/net/flow_dissector.h
@@ -350,7 +350,7 @@ static inline bool flow_keys_have_l4(const struct flow_keys *keys)
u32 flow_hash_from_keys(struct flow_keys *keys);
void skb_flow_get_icmp_tci(const struct sk_buff *skb,
struct flow_dissector_key_icmp *key_icmp,
- void *data, int thoff, int hlen);
+ const void *data, int thoff, int hlen);
static inline bool dissector_uses_key(const struct flow_dissector *flow_dissector,
enum flow_dissector_key_id key_id)
diff --git a/net/core/flow_dissector.c b/net/core/flow_dissector.c
index 2ef2224b3bff..2ed380d096ce 100644
--- a/net/core/flow_dissector.c
+++ b/net/core/flow_dissector.c
@@ -114,7 +114,7 @@ int flow_dissector_bpf_prog_attach_check(struct net *net,
* is the protocol port offset returned from proto_ports_offset
*/
__be32 __skb_flow_get_ports(const struct sk_buff *skb, int thoff, u8 ip_proto,
- void *data, int hlen)
+ const void *data, int hlen)
{
int poff = proto_ports_offset(ip_proto);
@@ -161,7 +161,7 @@ static bool icmp_has_id(u8 type)
*/
void skb_flow_get_icmp_tci(const struct sk_buff *skb,
struct flow_dissector_key_icmp *key_icmp,
- void *data, int thoff, int hlen)
+ const void *data, int thoff, int hlen)
{
struct icmphdr *ih, _ih;
@@ -187,8 +187,8 @@ EXPORT_SYMBOL(skb_flow_get_icmp_tci);
*/
static void __skb_flow_dissect_icmp(const struct sk_buff *skb,
struct flow_dissector *flow_dissector,
- void *target_container,
- void *data, int thoff, int hlen)
+ void *target_container, const void *data,
+ int thoff, int hlen)
{
struct flow_dissector_key_icmp *key_icmp;
@@ -409,8 +409,8 @@ EXPORT_SYMBOL(skb_flow_dissect_hash);
static enum flow_dissect_ret
__skb_flow_dissect_mpls(const struct sk_buff *skb,
struct flow_dissector *flow_dissector,
- void *target_container, void *data, int nhoff, int hlen,
- int lse_index, bool *entropy_label)
+ void *target_container, const void *data, int nhoff,
+ int hlen, int lse_index, bool *entropy_label)
{
struct mpls_label *hdr, _hdr;
u32 entry, label, bos;
@@ -467,7 +467,8 @@ __skb_flow_dissect_mpls(const struct sk_buff *skb,
static enum flow_dissect_ret
__skb_flow_dissect_arp(const struct sk_buff *skb,
struct flow_dissector *flow_dissector,
- void *target_container, void *data, int nhoff, int hlen)
+ void *target_container, const void *data,
+ int nhoff, int hlen)
{
struct flow_dissector_key_arp *key_arp;
struct {
@@ -523,7 +524,7 @@ static enum flow_dissect_ret
__skb_flow_dissect_gre(const struct sk_buff *skb,
struct flow_dissector_key_control *key_control,
struct flow_dissector *flow_dissector,
- void *target_container, void *data,
+ void *target_container, const void *data,
__be16 *p_proto, int *p_nhoff, int *p_hlen,
unsigned int flags)
{
@@ -663,8 +664,8 @@ __skb_flow_dissect_gre(const struct sk_buff *skb,
static enum flow_dissect_ret
__skb_flow_dissect_batadv(const struct sk_buff *skb,
struct flow_dissector_key_control *key_control,
- void *data, __be16 *p_proto, int *p_nhoff, int hlen,
- unsigned int flags)
+ const void *data, __be16 *p_proto, int *p_nhoff,
+ int hlen, unsigned int flags)
{
struct {
struct batadv_unicast_packet batadv_unicast;
@@ -695,7 +696,8 @@ __skb_flow_dissect_batadv(const struct sk_buff *skb,
static void
__skb_flow_dissect_tcp(const struct sk_buff *skb,
struct flow_dissector *flow_dissector,
- void *target_container, void *data, int thoff, int hlen)
+ void *target_container, const void *data,
+ int thoff, int hlen)
{
struct flow_dissector_key_tcp *key_tcp;
struct tcphdr *th, _th;
@@ -719,8 +721,8 @@ __skb_flow_dissect_tcp(const struct sk_buff *skb,
static void
__skb_flow_dissect_ports(const struct sk_buff *skb,
struct flow_dissector *flow_dissector,
- void *target_container, void *data, int nhoff,
- u8 ip_proto, int hlen)
+ void *target_container, const void *data,
+ int nhoff, u8 ip_proto, int hlen)
{
enum flow_dissector_key_id dissector_ports = FLOW_DISSECTOR_KEY_MAX;
struct flow_dissector_key_ports *key_ports;
@@ -744,7 +746,8 @@ __skb_flow_dissect_ports(const struct sk_buff *skb,
static void
__skb_flow_dissect_ipv4(const struct sk_buff *skb,
struct flow_dissector *flow_dissector,
- void *target_container, void *data, const struct iphdr *iph)
+ void *target_container, const void *data,
+ const struct iphdr *iph)
{
struct flow_dissector_key_ip *key_ip;
@@ -761,7 +764,8 @@ __skb_flow_dissect_ipv4(const struct sk_buff *skb,
static void
__skb_flow_dissect_ipv6(const struct sk_buff *skb,
struct flow_dissector *flow_dissector,
- void *target_container, void *data, const struct ipv6hdr *iph)
+ void *target_container, const void *data,
+ const struct ipv6hdr *iph)
{
struct flow_dissector_key_ip *key_ip;
@@ -908,9 +912,8 @@ bool bpf_flow_dissect(struct bpf_prog *prog, struct bpf_flow_dissector *ctx,
bool __skb_flow_dissect(const struct net *net,
const struct sk_buff *skb,
struct flow_dissector *flow_dissector,
- void *target_container,
- void *data, __be16 proto, int nhoff, int hlen,
- unsigned int flags)
+ void *target_container, const void *data,
+ __be16 proto, int nhoff, int hlen, unsigned int flags)
{
struct flow_dissector_key_control *key_control;
struct flow_dissector_key_basic *key_basic;
@@ -1642,7 +1645,7 @@ __u32 skb_get_hash_perturb(const struct sk_buff *skb,
}
EXPORT_SYMBOL(skb_get_hash_perturb);
-u32 __skb_get_poff(const struct sk_buff *skb, void *data,
+u32 __skb_get_poff(const struct sk_buff *skb, const void *data,
const struct flow_keys_basic *keys, int hlen)
{
u32 poff = keys->control.thoff;
--
2.30.2
It's used only for flow dissection, which now takes constant data
pointers.
Signed-off-by: Alexander Lobakin <[email protected]>
---
include/linux/etherdevice.h | 2 +-
net/ethernet/eth.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/include/linux/etherdevice.h b/include/linux/etherdevice.h
index bcb2f81baafb..330345b1be54 100644
--- a/include/linux/etherdevice.h
+++ b/include/linux/etherdevice.h
@@ -29,7 +29,7 @@ struct device;
int eth_platform_get_mac_address(struct device *dev, u8 *mac_addr);
unsigned char *arch_get_platform_mac_address(void);
int nvmem_get_mac_address(struct device *dev, void *addrbuf);
-u32 eth_get_headlen(const struct net_device *dev, void *data, unsigned int len);
+u32 eth_get_headlen(const struct net_device *dev, const void *data, u32 len);
__be16 eth_type_trans(struct sk_buff *skb, struct net_device *dev);
extern const struct header_ops eth_header_ops;
diff --git a/net/ethernet/eth.c b/net/ethernet/eth.c
index 4106373180c6..e01cf766d2c5 100644
--- a/net/ethernet/eth.c
+++ b/net/ethernet/eth.c
@@ -122,7 +122,7 @@ EXPORT_SYMBOL(eth_header);
* Make a best effort attempt to pull the length for all of the headers for
* a given frame in a linear buffer.
*/
-u32 eth_get_headlen(const struct net_device *dev, void *data, unsigned int len)
+u32 eth_get_headlen(const struct net_device *dev, const void *data, u32 len)
{
const unsigned int flags = FLOW_DISSECTOR_F_PARSE_1ST_FRAG;
const struct ethhdr *eth = (const struct ethhdr *)data;
--
2.30.2
{,__}skb_header_pointer() helpers exist mainly for preventing
accesses-beyond-end of the linear data.
In the vast majorify of cases, they bail out on the first condition.
All code going after is mostly a fallback.
Mark the most common branch as 'likely' one to move it in-line.
Also, skb_copy_bits() can return negative values only when the input
arguments are invalid, e.g. offset is greater than skb->len. It can
be safely marked as 'unlikely' branch, assuming that hotpath code
provides sane input to not fail here.
These two bump the throughput with a single Flow Dissector pass on
every packet (e.g. with RPS or driver that uses eth_get_headlen())
on 20 Mbps per flow/core.
Signed-off-by: Alexander Lobakin <[email protected]>
---
include/linux/skbuff.h | 5 ++---
1 file changed, 2 insertions(+), 3 deletions(-)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 46c61e127e9f..ecc029674ae4 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -3680,11 +3680,10 @@ static inline void * __must_check
__skb_header_pointer(const struct sk_buff *skb, int offset, int len,
const void *data, int hlen, void *buffer)
{
- if (hlen - offset >= len)
+ if (likely(hlen - offset >= len))
return (void *)data + offset;
- if (!skb ||
- skb_copy_bits(skb, offset, buffer, len) < 0)
+ if (!skb || unlikely(skb_copy_bits(skb, offset, buffer, len) < 0))
return NULL;
return buffer;
--
2.30.2
From: Alexander Lobakin <[email protected]>
Date: Sat, 13 Mar 2021 11:37:03 +0000
> This little number makes all of the flow dissection functions take
> raw input data pointer as const (1-5) and shuffles the branches in
> __skb_header_pointer() according to their hit probability.
>
> The result is +20 Mbps per flow/core with one Flow Dissector pass
> per packet. This affects RPS (with software hashing), drivers that
> use eth_get_headlen() on their Rx path and so on.
>
> Since v1 [0]:
> - rebase on top of the latest net-next. This was super-weird, but
> I double-checked that the series applies with no conflicts, and
> then on Patchwork it didn't;
Still failing on Patchwork. I rebased it ten thousand times, rebuilt
the patches manually, tried previous stable Git version and the
latest CVS snapshot, and always got the same series that successfully
applies to next-next.
Can you please take a look?
> - no other changes.
>
> [0] https://lore.kernel.org/netdev/[email protected]
>
> Alexander Lobakin (6):
> flow_dissector: constify bpf_flow_dissector's data pointers
> skbuff: make __skb_header_pointer()'s data argument const
> flow_dissector: constify raw input @data argument
> linux/etherdevice.h: misc trailing whitespace cleanup
> ethernet: constify eth_get_headlen()'s @data argument
> skbuff: micro-optimize {,__}skb_header_pointer()
>
> include/linux/etherdevice.h | 4 ++--
> include/linux/skbuff.h | 26 +++++++++++------------
> include/net/flow_dissector.h | 6 +++---
> net/core/flow_dissector.c | 41 +++++++++++++++++++-----------------
> net/ethernet/eth.c | 2 +-
> 5 files changed, 40 insertions(+), 39 deletions(-)
>
> --
> 2.30.2
Thanks,
Al
None of these apply to net-next as per the patchwork automated checks. Any idea why?
Thanks.
From: David Miller <[email protected]>
Date: Sat, 13 Mar 2021 18:10:00 -0800 (PST)
> None of these apply to net-next as per the patchwork automated checks. Any idea why?
No unfortunately. That'why I sent a follow-up mail. All of them
successfully apply to pure net-next on my machine.
Can it be a Git version conflict? I use 2.30.2, but also tried 2.30.1
and the latest development snapshot, and in either case they got
applied with no problems.
I could have more clue if NIPA provided more detailed log, but didn't
find any. And apply_patches stage doesn't seem to be present on NIPA
GitHub repo, so I couldn't reproduce it 1:1.
I can try again on Monday, not sure if it will help.
I also sent another series yesterday, and it has 15 green lights on
Patchwork, so this problem seems to take place only with this
particular series.
> Thanks.
Thanks,
Al