* Re: [PATCH] net/ipv4/netfilter/ip_tables.c: remove some inlines
[not found] <200712161247.53986.vda.linux@googlemail.com>
@ 2007-12-17 14:47 ` Patrick McHardy
2007-12-27 4:01 ` Denys Vlasenko
0 siblings, 1 reply; 7+ messages in thread
From: Patrick McHardy @ 2007-12-17 14:47 UTC (permalink / raw)
To: Denys Vlasenko
Cc: Harald Welte, netdev, linux-kernel,
Netfilter Development Mailinglist
Please CC netfilter-devel on netfilter patches.
Denys Vlasenko wrote:
> Hi Patrick, Harald,
>
> I was working on unrelated problem and noticed that ip_tables.c
> seem to abuse inline. I prepared a patch which removes inlines
> except those which are used by packet matching code
> (and thus are really performance-critical).
> I added comments explaining that remaining inlines are
> performance critical.
>
> Result as reported by size:
>
> text data bss dec hex filename
> - 6451 380 88 6919 1b07 ip_tables.o
> + 6339 348 72 6759 1a67 ip_tables.o
>
> Please take this patch into netfilter queue.
This clashes with my pending patches, which I'll push upstream
today. I also spent some time resyncing ip_tables and ip6_tables
so a diff of both (with some sed'ing) shows only the actual
differences, so please update ip6_tables as well.
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] net/ipv4/netfilter/ip_tables.c: remove some inlines
2007-12-17 14:47 ` [PATCH] net/ipv4/netfilter/ip_tables.c: remove some inlines Patrick McHardy
@ 2007-12-27 4:01 ` Denys Vlasenko
2007-12-28 15:30 ` Patrick McHardy
0 siblings, 1 reply; 7+ messages in thread
From: Denys Vlasenko @ 2007-12-27 4:01 UTC (permalink / raw)
To: Patrick McHardy
Cc: Harald Welte, netdev, linux-kernel,
Netfilter Development Mailinglist
On Monday 17 December 2007 14:47, Patrick McHardy wrote:
> Please CC netfilter-devel on netfilter patches.
>
> Denys Vlasenko wrote:
> > Hi Patrick, Harald,
> >
> > I was working on unrelated problem and noticed that ip_tables.c
> > seem to abuse inline. I prepared a patch which removes inlines
> > except those which are used by packet matching code
> > (and thus are really performance-critical).
> > I added comments explaining that remaining inlines are
> > performance critical.
> >
> > Result as reported by size:
> >
> > text data bss dec hex filename
> > - 6451 380 88 6919 1b07 ip_tables.o
> > + 6339 348 72 6759 1a67 ip_tables.o
> >
> > Please take this patch into netfilter queue.
>
> This clashes with my pending patches, which I'll push upstream
> today. I also spent some time resyncing ip_tables and ip6_tables
> so a diff of both (with some sed'ing) shows only the actual
> differences, so please update ip6_tables as well.
I would be happy to rediff the patch.
Which tree should I sync against in order to not collide
with your changes?
--
vda
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] net/ipv4/netfilter/ip_tables.c: remove some inlines
2007-12-27 4:01 ` Denys Vlasenko
@ 2007-12-28 15:30 ` Patrick McHardy
2007-12-31 16:43 ` [PATCH] ip[6]_tables.c: " Denys Vlasenko
0 siblings, 1 reply; 7+ messages in thread
From: Patrick McHardy @ 2007-12-28 15:30 UTC (permalink / raw)
To: Denys Vlasenko
Cc: Harald Welte, netdev, linux-kernel,
Netfilter Development Mailinglist
Denys Vlasenko wrote:
> On Monday 17 December 2007 14:47, Patrick McHardy wrote:
>
>> This clashes with my pending patches, which I'll push upstream
>> today. I also spent some time resyncing ip_tables and ip6_tables
>> so a diff of both (with some sed'ing) shows only the actual
>> differences, so please update ip6_tables as well.
>>
>
> I would be happy to rediff the patch.
>
> Which tree should I sync against in order to not collide
> with your changes?
>
The net-2.6.25.git tree contains all my changes.
^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH] ip[6]_tables.c: remove some inlines
2007-12-28 15:30 ` Patrick McHardy
@ 2007-12-31 16:43 ` Denys Vlasenko
2007-12-31 17:00 ` Patrick McHardy
0 siblings, 1 reply; 7+ messages in thread
From: Denys Vlasenko @ 2007-12-31 16:43 UTC (permalink / raw)
To: Patrick McHardy
Cc: Harald Welte, netdev, linux-kernel,
Netfilter Development Mailinglist
[-- Attachment #1: Type: text/plain, Size: 767 bytes --]
On Friday 28 December 2007 15:30, Patrick McHardy wrote:
> >> This clashes with my pending patches, which I'll push upstream
> >> today. I also spent some time resyncing ip_tables and ip6_tables
> >> so a diff of both (with some sed'ing) shows only the actual
> >> differences, so please update ip6_tables as well.
> >
> > I would be happy to rediff the patch.
> >
> > Which tree should I sync against in order to not collide
> > with your changes?
>
> The net-2.6.25.git tree contains all my changes.
ip[6]_tables.c seem to abuse inline.
This patch removes inlines except those which are used
by packet matching code and thus are performance-critical.
Please take this patch into netfilter queue.
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
--
vda
[-- Attachment #2: net-2.6.25_iptable_deinlined.patch --]
[-- Type: text/x-diff, Size: 11649 bytes --]
diff --git a/net/ipv4/netfilter/ip_tables.c b/net/ipv4/netfilter/ip_tables.c
index f5b66ec..982b7f9 100644
--- a/net/ipv4/netfilter/ip_tables.c
+++ b/net/ipv4/netfilter/ip_tables.c
@@ -75,6 +75,7 @@ do { \
Hence the start of any table is given by get_table() below. */
/* Returns whether matches rule or not. */
+/* Performance critical - called for every packet */
static inline bool
ip_packet_match(const struct iphdr *ip,
const char *indev,
@@ -153,7 +154,7 @@ ip_packet_match(const struct iphdr *ip,
return true;
}
-static inline bool
+static bool
ip_checkentry(const struct ipt_ip *ip)
{
if (ip->flags & ~IPT_F_MASK) {
@@ -183,8 +184,9 @@ ipt_error(struct sk_buff *skb,
return NF_DROP;
}
-static inline
-bool do_match(struct ipt_entry_match *m,
+/* Performance critical - called for every packet */
+static inline bool
+do_match(struct ipt_entry_match *m,
const struct sk_buff *skb,
const struct net_device *in,
const struct net_device *out,
@@ -199,6 +201,7 @@ bool do_match(struct ipt_entry_match *m,
return false;
}
+/* Performance critical */
static inline struct ipt_entry *
get_entry(void *base, unsigned int offset)
{
@@ -206,6 +209,7 @@ get_entry(void *base, unsigned int offset)
}
/* All zeroes == unconditional rule. */
+/* Mildly perf critical (only if packet tracing is on) */
static inline int
unconditional(const struct ipt_ip *ip)
{
@@ -221,7 +225,7 @@ unconditional(const struct ipt_ip *ip)
#if defined(CONFIG_NETFILTER_XT_TARGET_TRACE) || \
defined(CONFIG_NETFILTER_XT_TARGET_TRACE_MODULE)
-static const char *hooknames[] = {
+static const char *const hooknames[] = {
[NF_INET_PRE_ROUTING] = "PREROUTING",
[NF_INET_LOCAL_IN] = "INPUT",
[NF_INET_FORWARD] = "FORWARD",
@@ -235,7 +239,7 @@ enum nf_ip_trace_comments {
NF_IP_TRACE_COMMENT_POLICY,
};
-static const char *comments[] = {
+static const char *const comments[] = {
[NF_IP_TRACE_COMMENT_RULE] = "rule",
[NF_IP_TRACE_COMMENT_RETURN] = "return",
[NF_IP_TRACE_COMMENT_POLICY] = "policy",
@@ -251,6 +255,7 @@ static struct nf_loginfo trace_loginfo = {
},
};
+/* Mildly perf critical (only if packet tracing is on) */
static inline int
get_chainname_rulenum(struct ipt_entry *s, struct ipt_entry *e,
char *hookname, char **chainname,
@@ -567,7 +572,7 @@ mark_source_chains(struct xt_table_info *newinfo,
return 1;
}
-static inline int
+static int
cleanup_match(struct ipt_entry_match *m, unsigned int *i)
{
if (i && (*i)-- == 0)
@@ -579,7 +584,7 @@ cleanup_match(struct ipt_entry_match *m, unsigned int *i)
return 0;
}
-static inline int
+static int
check_entry(struct ipt_entry *e, const char *name)
{
struct ipt_entry_target *t;
@@ -600,7 +605,8 @@ check_entry(struct ipt_entry *e, const char *name)
return 0;
}
-static inline int check_match(struct ipt_entry_match *m, const char *name,
+static int
+check_match(struct ipt_entry_match *m, const char *name,
const struct ipt_ip *ip,
unsigned int hookmask, unsigned int *i)
{
@@ -623,7 +629,7 @@ static inline int check_match(struct ipt_entry_match *m, const char *name,
return ret;
}
-static inline int
+static int
find_check_match(struct ipt_entry_match *m,
const char *name,
const struct ipt_ip *ip,
@@ -652,7 +658,7 @@ err:
return ret;
}
-static inline int check_target(struct ipt_entry *e, const char *name)
+static int check_target(struct ipt_entry *e, const char *name)
{
struct ipt_entry_target *t;
struct xt_target *target;
@@ -673,7 +679,7 @@ static inline int check_target(struct ipt_entry *e, const char *name)
return ret;
}
-static inline int
+static int
find_check_entry(struct ipt_entry *e, const char *name, unsigned int size,
unsigned int *i)
{
@@ -717,7 +723,7 @@ find_check_entry(struct ipt_entry *e, const char *name, unsigned int size,
return ret;
}
-static inline int
+static int
check_entry_size_and_hooks(struct ipt_entry *e,
struct xt_table_info *newinfo,
unsigned char *base,
@@ -760,7 +766,7 @@ check_entry_size_and_hooks(struct ipt_entry *e,
return 0;
}
-static inline int
+static int
cleanup_entry(struct ipt_entry *e, unsigned int *i)
{
struct ipt_entry_target *t;
@@ -916,7 +922,7 @@ get_counters(const struct xt_table_info *t,
}
}
-static inline struct xt_counters * alloc_counters(struct xt_table *table)
+static struct xt_counters * alloc_counters(struct xt_table *table)
{
unsigned int countersize;
struct xt_counters *counters;
@@ -1304,7 +1310,7 @@ do_replace(void __user *user, unsigned int len)
/* We're lazy, and add to the first CPU; overflow works its fey magic
* and everything is OK. */
-static inline int
+static int
add_counter_to_entry(struct ipt_entry *e,
const struct xt_counters addme[],
unsigned int *i)
@@ -1465,7 +1471,7 @@ out:
return ret;
}
-static inline int
+static int
compat_find_calc_match(struct ipt_entry_match *m,
const char *name,
const struct ipt_ip *ip,
@@ -1489,7 +1495,7 @@ compat_find_calc_match(struct ipt_entry_match *m,
return 0;
}
-static inline int
+static int
compat_release_match(struct ipt_entry_match *m, unsigned int *i)
{
if (i && (*i)-- == 0)
@@ -1499,7 +1505,7 @@ compat_release_match(struct ipt_entry_match *m, unsigned int *i)
return 0;
}
-static inline int
+static int
compat_release_entry(struct compat_ipt_entry *e, unsigned int *i)
{
struct ipt_entry_target *t;
@@ -1514,7 +1520,7 @@ compat_release_entry(struct compat_ipt_entry *e, unsigned int *i)
return 0;
}
-static inline int
+static int
check_compat_entry_size_and_hooks(struct compat_ipt_entry *e,
struct xt_table_info *newinfo,
unsigned int *size,
@@ -1637,7 +1643,8 @@ compat_copy_entry_from_user(struct compat_ipt_entry *e, void **dstptr,
return ret;
}
-static inline int compat_check_entry(struct ipt_entry *e, const char *name,
+static int
+compat_check_entry(struct ipt_entry *e, const char *name,
unsigned int *i)
{
int j, ret;
diff --git a/net/ipv6/netfilter/ip6_tables.c b/net/ipv6/netfilter/ip6_tables.c
index 4ed16d2..dd7860f 100644
--- a/net/ipv6/netfilter/ip6_tables.c
+++ b/net/ipv6/netfilter/ip6_tables.c
@@ -90,6 +90,7 @@ ip6t_ext_hdr(u8 nexthdr)
}
/* Returns whether matches rule or not. */
+/* Performance critical - called for every packet */
static inline bool
ip6_packet_match(const struct sk_buff *skb,
const char *indev,
@@ -182,7 +183,7 @@ ip6_packet_match(const struct sk_buff *skb,
}
/* should be ip6 safe */
-static inline bool
+static bool
ip6_checkentry(const struct ip6t_ip6 *ipv6)
{
if (ipv6->flags & ~IP6T_F_MASK) {
@@ -212,8 +213,9 @@ ip6t_error(struct sk_buff *skb,
return NF_DROP;
}
-static inline
-bool do_match(struct ip6t_entry_match *m,
+/* Performance critical - called for every packet */
+static inline bool
+do_match(struct ip6t_entry_match *m,
const struct sk_buff *skb,
const struct net_device *in,
const struct net_device *out,
@@ -236,6 +238,7 @@ get_entry(void *base, unsigned int offset)
}
/* All zeroes == unconditional rule. */
+/* Mildly perf critical (only if packet tracing is on) */
static inline int
unconditional(const struct ip6t_ip6 *ipv6)
{
@@ -251,7 +254,7 @@ unconditional(const struct ip6t_ip6 *ipv6)
#if defined(CONFIG_NETFILTER_XT_TARGET_TRACE) || \
defined(CONFIG_NETFILTER_XT_TARGET_TRACE_MODULE)
/* This cries for unification! */
-static const char *hooknames[] = {
+static const char *const hooknames[] = {
[NF_INET_PRE_ROUTING] = "PREROUTING",
[NF_INET_LOCAL_IN] = "INPUT",
[NF_INET_FORWARD] = "FORWARD",
@@ -265,7 +268,7 @@ enum nf_ip_trace_comments {
NF_IP6_TRACE_COMMENT_POLICY,
};
-static const char *comments[] = {
+static const char *const comments[] = {
[NF_IP6_TRACE_COMMENT_RULE] = "rule",
[NF_IP6_TRACE_COMMENT_RETURN] = "return",
[NF_IP6_TRACE_COMMENT_POLICY] = "policy",
@@ -281,6 +284,7 @@ static struct nf_loginfo trace_loginfo = {
},
};
+/* Mildly perf critical (only if packet tracing is on) */
static inline int
get_chainname_rulenum(struct ip6t_entry *s, struct ip6t_entry *e,
char *hookname, char **chainname,
@@ -595,7 +599,7 @@ mark_source_chains(struct xt_table_info *newinfo,
return 1;
}
-static inline int
+static int
cleanup_match(struct ip6t_entry_match *m, unsigned int *i)
{
if (i && (*i)-- == 0)
@@ -607,7 +611,7 @@ cleanup_match(struct ip6t_entry_match *m, unsigned int *i)
return 0;
}
-static inline int
+static int
check_entry(struct ip6t_entry *e, const char *name)
{
struct ip6t_entry_target *t;
@@ -628,7 +632,7 @@ check_entry(struct ip6t_entry *e, const char *name)
return 0;
}
-static inline int check_match(struct ip6t_entry_match *m, const char *name,
+static int check_match(struct ip6t_entry_match *m, const char *name,
const struct ip6t_ip6 *ipv6,
unsigned int hookmask, unsigned int *i)
{
@@ -651,7 +655,7 @@ static inline int check_match(struct ip6t_entry_match *m, const char *name,
return ret;
}
-static inline int
+static int
find_check_match(struct ip6t_entry_match *m,
const char *name,
const struct ip6t_ip6 *ipv6,
@@ -680,7 +684,7 @@ err:
return ret;
}
-static inline int check_target(struct ip6t_entry *e, const char *name)
+static int check_target(struct ip6t_entry *e, const char *name)
{
struct ip6t_entry_target *t;
struct xt_target *target;
@@ -701,7 +705,7 @@ static inline int check_target(struct ip6t_entry *e, const char *name)
return ret;
}
-static inline int
+static int
find_check_entry(struct ip6t_entry *e, const char *name, unsigned int size,
unsigned int *i)
{
@@ -745,7 +749,7 @@ find_check_entry(struct ip6t_entry *e, const char *name, unsigned int size,
return ret;
}
-static inline int
+static int
check_entry_size_and_hooks(struct ip6t_entry *e,
struct xt_table_info *newinfo,
unsigned char *base,
@@ -788,7 +792,7 @@ check_entry_size_and_hooks(struct ip6t_entry *e,
return 0;
}
-static inline int
+static int
cleanup_entry(struct ip6t_entry *e, unsigned int *i)
{
struct ip6t_entry_target *t;
@@ -944,7 +948,7 @@ get_counters(const struct xt_table_info *t,
}
}
-static inline struct xt_counters *alloc_counters(struct xt_table *table)
+static struct xt_counters *alloc_counters(struct xt_table *table)
{
unsigned int countersize;
struct xt_counters *counters;
@@ -1494,7 +1498,7 @@ out:
return ret;
}
-static inline int
+static int
compat_find_calc_match(struct ip6t_entry_match *m,
const char *name,
const struct ip6t_ip6 *ipv6,
@@ -1518,7 +1522,7 @@ compat_find_calc_match(struct ip6t_entry_match *m,
return 0;
}
-static inline int
+static int
compat_release_match(struct ip6t_entry_match *m, unsigned int *i)
{
if (i && (*i)-- == 0)
@@ -1528,7 +1532,7 @@ compat_release_match(struct ip6t_entry_match *m, unsigned int *i)
return 0;
}
-static inline int
+static int
compat_release_entry(struct compat_ip6t_entry *e, unsigned int *i)
{
struct ip6t_entry_target *t;
@@ -1543,7 +1547,7 @@ compat_release_entry(struct compat_ip6t_entry *e, unsigned int *i)
return 0;
}
-static inline int
+static int
check_compat_entry_size_and_hooks(struct compat_ip6t_entry *e,
struct xt_table_info *newinfo,
unsigned int *size,
@@ -1666,7 +1670,7 @@ compat_copy_entry_from_user(struct compat_ip6t_entry *e, void **dstptr,
return ret;
}
-static inline int compat_check_entry(struct ip6t_entry *e, const char *name,
+static int compat_check_entry(struct ip6t_entry *e, const char *name,
unsigned int *i)
{
int j, ret;
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH] ip[6]_tables.c: remove some inlines
2007-12-31 16:43 ` [PATCH] ip[6]_tables.c: " Denys Vlasenko
@ 2007-12-31 17:00 ` Patrick McHardy
2007-12-31 17:55 ` Denys Vlasenko
0 siblings, 1 reply; 7+ messages in thread
From: Patrick McHardy @ 2007-12-31 17:00 UTC (permalink / raw)
To: Denys Vlasenko
Cc: Harald Welte, netdev, linux-kernel,
Netfilter Development Mailinglist
Denys Vlasenko wrote:
> ip[6]_tables.c seem to abuse inline.
>
> This patch removes inlines except those which are used
> by packet matching code and thus are performance-critical.
>
Some people also consider the ruleset replacement path performance
critical, but overall I agree with your patch. I'm travelling currently
though so it will take a few days until I'll apply it.
Do you have some numbers that show the actual difference these
changes make?
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] ip[6]_tables.c: remove some inlines
2007-12-31 17:00 ` Patrick McHardy
@ 2007-12-31 17:55 ` Denys Vlasenko
2008-01-08 16:36 ` Patrick McHardy
0 siblings, 1 reply; 7+ messages in thread
From: Denys Vlasenko @ 2007-12-31 17:55 UTC (permalink / raw)
To: Patrick McHardy
Cc: Harald Welte, netdev, linux-kernel,
Netfilter Development Mailinglist
On Monday 31 December 2007 17:00, Patrick McHardy wrote:
> Denys Vlasenko wrote:
> > ip[6]_tables.c seem to abuse inline.
> >
> > This patch removes inlines except those which are used
> > by packet matching code and thus are performance-critical.
> >
>
> Some people also consider the ruleset replacement path performance
> critical, but overall I agree with your patch. I'm travelling currently
> though so it will take a few days until I'll apply it.
>
> Do you have some numbers that show the actual difference these
> changes make?
Before:
$ size */*/*/ip*tables*.o
text data bss dec hex filename
6402 500 16 6918 1b06 net/ipv4/netfilter/ip_tables.o
7130 500 16 7646 1dde net/ipv6/netfilter/ip6_tables.o
After:
$ size */*/*/ip*tables*.o
text data bss dec hex filename
6307 500 16 6823 1aa7 net/ipv4/netfilter/ip_tables.o
7010 500 16 7526 1d66 net/ipv6/netfilter/ip6_tables.o
--
vda
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] ip[6]_tables.c: remove some inlines
2007-12-31 17:55 ` Denys Vlasenko
@ 2008-01-08 16:36 ` Patrick McHardy
0 siblings, 0 replies; 7+ messages in thread
From: Patrick McHardy @ 2008-01-08 16:36 UTC (permalink / raw)
To: Denys Vlasenko
Cc: Harald Welte, netdev, linux-kernel,
Netfilter Development Mailinglist
Denys Vlasenko wrote:
> On Monday 31 December 2007 17:00, Patrick McHardy wrote:
>> Denys Vlasenko wrote:
>>> ip[6]_tables.c seem to abuse inline.
>>>
>>> This patch removes inlines except those which are used
>>> by packet matching code and thus are performance-critical.
>>>
>> Some people also consider the ruleset replacement path performance
>> critical, but overall I agree with your patch. I'm travelling currently
>> though so it will take a few days until I'll apply it.
>>
>> Do you have some numbers that show the actual difference these
>> changes make?
>
> Before:
>
> $ size */*/*/ip*tables*.o
> text data bss dec hex filename
> 6402 500 16 6918 1b06 net/ipv4/netfilter/ip_tables.o
> 7130 500 16 7646 1dde net/ipv6/netfilter/ip6_tables.o
>
> After:
>
> $ size */*/*/ip*tables*.o
> text data bss dec hex filename
> 6307 500 16 6823 1aa7 net/ipv4/netfilter/ip_tables.o
> 7010 500 16 7526 1d66 net/ipv6/netfilter/ip6_tables.o
Applied, thanks.
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2008-01-08 16:40 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
[not found] <200712161247.53986.vda.linux@googlemail.com>
2007-12-17 14:47 ` [PATCH] net/ipv4/netfilter/ip_tables.c: remove some inlines Patrick McHardy
2007-12-27 4:01 ` Denys Vlasenko
2007-12-28 15:30 ` Patrick McHardy
2007-12-31 16:43 ` [PATCH] ip[6]_tables.c: " Denys Vlasenko
2007-12-31 17:00 ` Patrick McHardy
2007-12-31 17:55 ` Denys Vlasenko
2008-01-08 16:36 ` Patrick McHardy
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).