* nfbz 621 - 3 kernel patches
@ 2009-12-05 20:26 Jan Engelhardt
2009-12-05 20:26 ` [PATCH 1/3] netfilter: xt_recent: save 8 bytes per htable Jan Engelhardt
` (3 more replies)
0 siblings, 4 replies; 12+ messages in thread
From: Jan Engelhardt @ 2009-12-05 20:26 UTC (permalink / raw)
To: kaber; +Cc: netfilter-devel
Hi,
a few simple commits that turned up while processing NF bug 621.
Thanks for taking care of them :)
The following changes since commit 22763c5cf3690a681551162c15d34d935308c8d7:
Linus Torvalds (1):
Linux 2.6.32
are available in the git repository at:
git://dev.medozas.de/linux master
Jan Engelhardt (3):
netfilter: xt_recent: save 8 bytes per htable
netfilter: xtables: do not grab random bytes at __init
netfilter: xtables: obtain random bytes earlier, in checkentry
net/netfilter/xt_NFQUEUE.c | 6 +++++-
net/netfilter/xt_RATEEST.c | 7 ++++++-
net/netfilter/xt_connlimit.c | 17 ++++++-----------
net/netfilter/xt_hashlimit.c | 8 ++++----
net/netfilter/xt_recent.c | 20 ++++++++------------
5 files changed, 29 insertions(+), 29 deletions(-)
^ permalink raw reply [flat|nested] 12+ messages in thread
* [PATCH 1/3] netfilter: xt_recent: save 8 bytes per htable
2009-12-05 20:26 nfbz 621 - 3 kernel patches Jan Engelhardt
@ 2009-12-05 20:26 ` Jan Engelhardt
2010-01-04 15:26 ` Patrick McHardy
2009-12-05 20:26 ` [PATCH 2/3] netfilter: xtables: do not grab random bytes at __init Jan Engelhardt
` (2 subsequent siblings)
3 siblings, 1 reply; 12+ messages in thread
From: Jan Engelhardt @ 2009-12-05 20:26 UTC (permalink / raw)
To: kaber; +Cc: netfilter-devel
Moving rnd_inited into the hole after the uint8 lets go of the uint32
rnd_inited was using, plus the padding that would follow the int group.
Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
---
net/netfilter/xt_hashlimit.c | 8 ++++----
1 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/net/netfilter/xt_hashlimit.c b/net/netfilter/xt_hashlimit.c
index dd16e40..5bdc1fb 100644
--- a/net/netfilter/xt_hashlimit.c
+++ b/net/netfilter/xt_hashlimit.c
@@ -81,13 +81,13 @@ struct xt_hashlimit_htable {
struct hlist_node node; /* global list of all htables */
atomic_t use;
u_int8_t family;
+ bool rnd_initialized;
struct hashlimit_cfg1 cfg; /* config */
/* used internally */
spinlock_t lock; /* lock for list_head */
u_int32_t rnd; /* random seed for hash */
- int rnd_initialized;
unsigned int count; /* number entries in table */
struct timer_list timer; /* timer for gc */
@@ -150,7 +150,7 @@ dsthash_alloc_init(struct xt_hashlimit_htable *ht,
* the first hashtable entry */
if (!ht->rnd_initialized) {
get_random_bytes(&ht->rnd, sizeof(ht->rnd));
- ht->rnd_initialized = 1;
+ ht->rnd_initialized = true;
}
if (ht->cfg.max && ht->count >= ht->cfg.max) {
@@ -235,7 +235,7 @@ static int htable_create_v0(struct xt_hashlimit_info *minfo, u_int8_t family)
atomic_set(&hinfo->use, 1);
hinfo->count = 0;
hinfo->family = family;
- hinfo->rnd_initialized = 0;
+ hinfo->rnd_initialized = false;
spin_lock_init(&hinfo->lock);
hinfo->pde = proc_create_data(minfo->name, 0,
(family == NFPROTO_IPV4) ?
@@ -296,7 +296,7 @@ static int htable_create(struct xt_hashlimit_mtinfo1 *minfo, u_int8_t family)
atomic_set(&hinfo->use, 1);
hinfo->count = 0;
hinfo->family = family;
- hinfo->rnd_initialized = 0;
+ hinfo->rnd_initialized = false;
spin_lock_init(&hinfo->lock);
hinfo->pde = proc_create_data(minfo->name, 0,
--
1.6.5.3
^ permalink raw reply related [flat|nested] 12+ messages in thread
* [PATCH 2/3] netfilter: xtables: do not grab random bytes at __init
2009-12-05 20:26 nfbz 621 - 3 kernel patches Jan Engelhardt
2009-12-05 20:26 ` [PATCH 1/3] netfilter: xt_recent: save 8 bytes per htable Jan Engelhardt
@ 2009-12-05 20:26 ` Jan Engelhardt
2010-01-04 15:27 ` Patrick McHardy
2009-12-05 20:26 ` [PATCH 3/3] netfilter: xtables: obtain random bytes earlier, in checkentry Jan Engelhardt
2009-12-14 13:51 ` nfbz 621 - 3 kernel patches Patrick McHardy
3 siblings, 1 reply; 12+ messages in thread
From: Jan Engelhardt @ 2009-12-05 20:26 UTC (permalink / raw)
To: kaber; +Cc: netfilter-devel
References: http://bugzilla.netfilter.org/show_bug.cgi?id=621
"It is deliberately not done in the init function, since we might not
have sufficient random while booting."
Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
---
net/netfilter/xt_NFQUEUE.c | 6 +++++-
net/netfilter/xt_RATEEST.c | 7 ++++++-
2 files changed, 11 insertions(+), 2 deletions(-)
diff --git a/net/netfilter/xt_NFQUEUE.c b/net/netfilter/xt_NFQUEUE.c
index f28f6a5..12dcd70 100644
--- a/net/netfilter/xt_NFQUEUE.c
+++ b/net/netfilter/xt_NFQUEUE.c
@@ -28,6 +28,7 @@ MODULE_ALIAS("ip6t_NFQUEUE");
MODULE_ALIAS("arpt_NFQUEUE");
static u32 jhash_initval __read_mostly;
+static bool rnd_inited __read_mostly;
static unsigned int
nfqueue_tg(struct sk_buff *skb, const struct xt_target_param *par)
@@ -90,6 +91,10 @@ static bool nfqueue_tg_v1_check(const struct xt_tgchk_param *par)
const struct xt_NFQ_info_v1 *info = par->targinfo;
u32 maxid;
+ if (unlikely(!rnd_inited)) {
+ get_random_bytes(&jhash_initval, sizeof(jhash_initval));
+ rnd_inited = true;
+ }
if (info->queues_total == 0) {
pr_err("NFQUEUE: number of total queues is 0\n");
return false;
@@ -135,7 +140,6 @@ static struct xt_target nfqueue_tg_reg[] __read_mostly = {
static int __init nfqueue_tg_init(void)
{
- get_random_bytes(&jhash_initval, sizeof(jhash_initval));
return xt_register_targets(nfqueue_tg_reg, ARRAY_SIZE(nfqueue_tg_reg));
}
diff --git a/net/netfilter/xt_RATEEST.c b/net/netfilter/xt_RATEEST.c
index d80b819..87ae97e 100644
--- a/net/netfilter/xt_RATEEST.c
+++ b/net/netfilter/xt_RATEEST.c
@@ -23,6 +23,7 @@ static DEFINE_MUTEX(xt_rateest_mutex);
#define RATEEST_HSIZE 16
static struct hlist_head rateest_hash[RATEEST_HSIZE] __read_mostly;
static unsigned int jhash_rnd __read_mostly;
+static bool rnd_inited __read_mostly;
static unsigned int xt_rateest_hash(const char *name)
{
@@ -93,6 +94,11 @@ static bool xt_rateest_tg_checkentry(const struct xt_tgchk_param *par)
struct gnet_estimator est;
} cfg;
+ if (unlikely(!rnd_inited)) {
+ get_random_bytes(&jhash_rnd, sizeof(jhash_rnd));
+ rnd_inited = true;
+ }
+
est = xt_rateest_lookup(info->name);
if (est) {
/*
@@ -164,7 +170,6 @@ static int __init xt_rateest_tg_init(void)
for (i = 0; i < ARRAY_SIZE(rateest_hash); i++)
INIT_HLIST_HEAD(&rateest_hash[i]);
- get_random_bytes(&jhash_rnd, sizeof(jhash_rnd));
return xt_register_target(&xt_rateest_tg_reg);
}
--
1.6.5.3
^ permalink raw reply related [flat|nested] 12+ messages in thread
* [PATCH 3/3] netfilter: xtables: obtain random bytes earlier, in checkentry
2009-12-05 20:26 nfbz 621 - 3 kernel patches Jan Engelhardt
2009-12-05 20:26 ` [PATCH 1/3] netfilter: xt_recent: save 8 bytes per htable Jan Engelhardt
2009-12-05 20:26 ` [PATCH 2/3] netfilter: xtables: do not grab random bytes at __init Jan Engelhardt
@ 2009-12-05 20:26 ` Jan Engelhardt
2010-01-04 15:29 ` Patrick McHardy
2009-12-14 13:51 ` nfbz 621 - 3 kernel patches Patrick McHardy
3 siblings, 1 reply; 12+ messages in thread
From: Jan Engelhardt @ 2009-12-05 20:26 UTC (permalink / raw)
To: kaber; +Cc: netfilter-devel
We can initialize the random hash bytes on checkentry. This is
preferable since it is outside the hot path.
Reference: http://bugzilla.netfilter.org/show_bug.cgi?id=621
Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
---
net/netfilter/xt_connlimit.c | 17 ++++++-----------
net/netfilter/xt_recent.c | 20 ++++++++------------
2 files changed, 14 insertions(+), 23 deletions(-)
diff --git a/net/netfilter/xt_connlimit.c b/net/netfilter/xt_connlimit.c
index 38f03f7..8103bef 100644
--- a/net/netfilter/xt_connlimit.c
+++ b/net/netfilter/xt_connlimit.c
@@ -40,15 +40,11 @@ struct xt_connlimit_data {
spinlock_t lock;
};
-static u_int32_t connlimit_rnd;
-static bool connlimit_rnd_inited;
+static u_int32_t connlimit_rnd __read_mostly;
+static bool connlimit_rnd_inited __read_mostly;
static inline unsigned int connlimit_iphash(__be32 addr)
{
- if (unlikely(!connlimit_rnd_inited)) {
- get_random_bytes(&connlimit_rnd, sizeof(connlimit_rnd));
- connlimit_rnd_inited = true;
- }
return jhash_1word((__force __u32)addr, connlimit_rnd) & 0xFF;
}
@@ -59,11 +55,6 @@ connlimit_iphash6(const union nf_inet_addr *addr,
union nf_inet_addr res;
unsigned int i;
- if (unlikely(!connlimit_rnd_inited)) {
- get_random_bytes(&connlimit_rnd, sizeof(connlimit_rnd));
- connlimit_rnd_inited = true;
- }
-
for (i = 0; i < ARRAY_SIZE(addr->ip6); ++i)
res.ip6[i] = addr->ip6[i] & mask->ip6[i];
@@ -226,6 +217,10 @@ static bool connlimit_mt_check(const struct xt_mtchk_param *par)
struct xt_connlimit_info *info = par->matchinfo;
unsigned int i;
+ if (unlikely(!connlimit_rnd_inited)) {
+ get_random_bytes(&connlimit_rnd, sizeof(connlimit_rnd));
+ connlimit_rnd_inited = true;
+ }
if (nf_ct_l3proto_try_module_get(par->family) < 0) {
printk(KERN_WARNING "cannot load conntrack support for "
"address family %u\n", par->family);
diff --git a/net/netfilter/xt_recent.c b/net/netfilter/xt_recent.c
index eb0ceb8..2176b1b 100644
--- a/net/netfilter/xt_recent.c
+++ b/net/netfilter/xt_recent.c
@@ -90,25 +90,17 @@ static struct proc_dir_entry *recent_proc_dir;
static const struct file_operations recent_old_fops, recent_mt_fops;
#endif
-static u_int32_t hash_rnd;
-static bool hash_rnd_initted;
+static u_int32_t hash_rnd __read_mostly;
+static bool hash_rnd_inited __read_mostly;
-static unsigned int recent_entry_hash4(const union nf_inet_addr *addr)
+static inline unsigned int recent_entry_hash4(const union nf_inet_addr *addr)
{
- if (!hash_rnd_initted) {
- get_random_bytes(&hash_rnd, sizeof(hash_rnd));
- hash_rnd_initted = true;
- }
return jhash_1word((__force u32)addr->ip, hash_rnd) &
(ip_list_hash_size - 1);
}
-static unsigned int recent_entry_hash6(const union nf_inet_addr *addr)
+static inline unsigned int recent_entry_hash6(const union nf_inet_addr *addr)
{
- if (!hash_rnd_initted) {
- get_random_bytes(&hash_rnd, sizeof(hash_rnd));
- hash_rnd_initted = true;
- }
return jhash2((u32 *)addr->ip6, ARRAY_SIZE(addr->ip6), hash_rnd) &
(ip_list_hash_size - 1);
}
@@ -287,6 +279,10 @@ static bool recent_mt_check(const struct xt_mtchk_param *par)
unsigned i;
bool ret = false;
+ if (unlikely(!hash_rnd_inited)) {
+ get_random_bytes(&hash_rnd, sizeof(hash_rnd));
+ hash_rnd_inited = true;
+ }
if (hweight8(info->check_set &
(XT_RECENT_SET | XT_RECENT_REMOVE |
XT_RECENT_CHECK | XT_RECENT_UPDATE)) != 1)
--
1.6.5.3
^ permalink raw reply related [flat|nested] 12+ messages in thread
* Re: nfbz 621 - 3 kernel patches
2009-12-05 20:26 nfbz 621 - 3 kernel patches Jan Engelhardt
` (2 preceding siblings ...)
2009-12-05 20:26 ` [PATCH 3/3] netfilter: xtables: obtain random bytes earlier, in checkentry Jan Engelhardt
@ 2009-12-14 13:51 ` Patrick McHardy
2010-01-25 11:01 ` Jan Engelhardt
3 siblings, 1 reply; 12+ messages in thread
From: Patrick McHardy @ 2009-12-14 13:51 UTC (permalink / raw)
To: Jan Engelhardt; +Cc: netfilter-devel
Jan Engelhardt wrote:
> a few simple commits that turned up while processing NF bug 621.
> Thanks for taking care of them :)
>
>
> The following changes since commit 22763c5cf3690a681551162c15d34d935308c8d7:
> Linus Torvalds (1):
> Linux 2.6.32
>
> are available in the git repository at:
>
> git://dev.medozas.de/linux master
I'll apply those once the net-next tree opens up.
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 1/3] netfilter: xt_recent: save 8 bytes per htable
2009-12-05 20:26 ` [PATCH 1/3] netfilter: xt_recent: save 8 bytes per htable Jan Engelhardt
@ 2010-01-04 15:26 ` Patrick McHardy
0 siblings, 0 replies; 12+ messages in thread
From: Patrick McHardy @ 2010-01-04 15:26 UTC (permalink / raw)
To: Jan Engelhardt; +Cc: netfilter-devel
Jan Engelhardt wrote:
> Moving rnd_inited into the hole after the uint8 lets go of the uint32
> rnd_inited was using, plus the padding that would follow the int group.
Applied, thanks Jan.
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 2/3] netfilter: xtables: do not grab random bytes at __init
2009-12-05 20:26 ` [PATCH 2/3] netfilter: xtables: do not grab random bytes at __init Jan Engelhardt
@ 2010-01-04 15:27 ` Patrick McHardy
0 siblings, 0 replies; 12+ messages in thread
From: Patrick McHardy @ 2010-01-04 15:27 UTC (permalink / raw)
To: Jan Engelhardt; +Cc: netfilter-devel
Jan Engelhardt wrote:
> References: http://bugzilla.netfilter.org/show_bug.cgi?id=621
>
> "It is deliberately not done in the init function, since we might not
> have sufficient random while booting."
Applied, thanks.
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: [PATCH 3/3] netfilter: xtables: obtain random bytes earlier, in checkentry
2009-12-05 20:26 ` [PATCH 3/3] netfilter: xtables: obtain random bytes earlier, in checkentry Jan Engelhardt
@ 2010-01-04 15:29 ` Patrick McHardy
0 siblings, 0 replies; 12+ messages in thread
From: Patrick McHardy @ 2010-01-04 15:29 UTC (permalink / raw)
To: Jan Engelhardt; +Cc: netfilter-devel
Jan Engelhardt wrote:
> We can initialize the random hash bytes on checkentry. This is
> preferable since it is outside the hot path.
>
> Reference: http://bugzilla.netfilter.org/show_bug.cgi?id=621
Also applied, thanks.
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: nfbz 621 - 3 kernel patches
2009-12-14 13:51 ` nfbz 621 - 3 kernel patches Patrick McHardy
@ 2010-01-25 11:01 ` Jan Engelhardt
2010-01-25 12:09 ` Patrick McHardy
0 siblings, 1 reply; 12+ messages in thread
From: Jan Engelhardt @ 2010-01-25 11:01 UTC (permalink / raw)
To: Patrick McHardy; +Cc: netfilter-devel
On Monday 2009-12-14 14:51, Patrick McHardy wrote:
>Jan Engelhardt wrote:
>> a few simple commits that turned up while processing NF bug 621.
>> Thanks for taking care of them :)
>>
>> The following changes since commit 22763c5cf3690a681551162c15d34d935308c8d7:
>> Linus Torvalds (1):
>> Linux 2.6.32
>>
>> are available in the git repository at:
>>
>> git://dev.medozas.de/linux master
>
>I'll apply those once the net-next tree opens up.
Is it open already? :)
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: nfbz 621 - 3 kernel patches
2010-01-25 11:01 ` Jan Engelhardt
@ 2010-01-25 12:09 ` Patrick McHardy
2010-01-25 23:05 ` Jan Engelhardt
0 siblings, 1 reply; 12+ messages in thread
From: Patrick McHardy @ 2010-01-25 12:09 UTC (permalink / raw)
To: Jan Engelhardt; +Cc: netfilter-devel
Jan Engelhardt wrote:
> On Monday 2009-12-14 14:51, Patrick McHardy wrote:
>> Jan Engelhardt wrote:
>>> a few simple commits that turned up while processing NF bug 621.
>>> Thanks for taking care of them :)
>>>
>>> The following changes since commit 22763c5cf3690a681551162c15d34d935308c8d7:
>>> Linus Torvalds (1):
>>> Linux 2.6.32
>>>
>>> are available in the git repository at:
>>>
>>> git://dev.medozas.de/linux master
>> I'll apply those once the net-next tree opens up.
>
> Is it open already? :)
Yes, but I don't have those patches in my inbox anymore. Please
resend based on nf-next-2.6.git.
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: nfbz 621 - 3 kernel patches
2010-01-25 12:09 ` Patrick McHardy
@ 2010-01-25 23:05 ` Jan Engelhardt
2010-01-26 11:08 ` Patrick McHardy
0 siblings, 1 reply; 12+ messages in thread
From: Jan Engelhardt @ 2010-01-25 23:05 UTC (permalink / raw)
To: Patrick McHardy; +Cc: netfilter-devel
On Monday 2010-01-25 13:09, Patrick McHardy wrote:
>Jan Engelhardt wrote:
>> On Monday 2009-12-14 14:51, Patrick McHardy wrote:
>>> Jan Engelhardt wrote:
>>>> a few simple commits that turned up while processing NF bug 621.
>>>> Thanks for taking care of them :)
>>>>
>>>> The following changes since commit 22763c5cf3690a681551162c15d34d935308c8d7:
>>>> Linus Torvalds (1):
>>>> Linux 2.6.32
>>>>
>>>> are available in the git repository at:
>>>>
>>>> git://dev.medozas.de/linux master
>>> I'll apply those once the net-next tree opens up.
>>
>> Is it open already? :)
>
>Yes, but I don't have those patches in my inbox anymore. Please
>resend based on nf-next-2.6.git.
Seems they've already got merged.
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: nfbz 621 - 3 kernel patches
2010-01-25 23:05 ` Jan Engelhardt
@ 2010-01-26 11:08 ` Patrick McHardy
0 siblings, 0 replies; 12+ messages in thread
From: Patrick McHardy @ 2010-01-26 11:08 UTC (permalink / raw)
To: Jan Engelhardt; +Cc: netfilter-devel
Jan Engelhardt wrote:
> On Monday 2010-01-25 13:09, Patrick McHardy wrote:
>> Jan Engelhardt wrote:
>>> On Monday 2009-12-14 14:51, Patrick McHardy wrote:
>>>> Jan Engelhardt wrote:
>>>>> a few simple commits that turned up while processing NF bug 621.
>>>>> Thanks for taking care of them :)
>>>>>
>>>>> The following changes since commit 22763c5cf3690a681551162c15d34d935308c8d7:
>>>>> Linus Torvalds (1):
>>>>> Linux 2.6.32
>>>>>
>>>>> are available in the git repository at:
>>>>>
>>>>> git://dev.medozas.de/linux master
>>>> I'll apply those once the net-next tree opens up.
>>> Is it open already? :)
>> Yes, but I don't have those patches in my inbox anymore. Please
>> resend based on nf-next-2.6.git.
>
> Seems they've already got merged.
Ah, right, just as I wrote on the 4th :)
^ permalink raw reply [flat|nested] 12+ messages in thread
end of thread, other threads:[~2010-01-26 11:08 UTC | newest]
Thread overview: 12+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2009-12-05 20:26 nfbz 621 - 3 kernel patches Jan Engelhardt
2009-12-05 20:26 ` [PATCH 1/3] netfilter: xt_recent: save 8 bytes per htable Jan Engelhardt
2010-01-04 15:26 ` Patrick McHardy
2009-12-05 20:26 ` [PATCH 2/3] netfilter: xtables: do not grab random bytes at __init Jan Engelhardt
2010-01-04 15:27 ` Patrick McHardy
2009-12-05 20:26 ` [PATCH 3/3] netfilter: xtables: obtain random bytes earlier, in checkentry Jan Engelhardt
2010-01-04 15:29 ` Patrick McHardy
2009-12-14 13:51 ` nfbz 621 - 3 kernel patches Patrick McHardy
2010-01-25 11:01 ` Jan Engelhardt
2010-01-25 12:09 ` Patrick McHardy
2010-01-25 23:05 ` Jan Engelhardt
2010-01-26 11:08 ` 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).