[PATCH 6/8 v2] netfilter: fix string extension for case insensitive pattern matching

[PATCH 6/8 v2] netfilter: fix string extension for case insensitive pattern matching

[Joonwoo Park]

The flag XT_STRING_FLAG_IGNORECASE indicates case insensitive string
matching.
netfilter can find cmd.exe, Cmd.exe, cMd.exe and etc easily.

Signed-off-by: Joonwoo Park <joonwpark81@gmail.com>
---
 include/linux/netfilter/xt_string.h |    3 +++
 net/netfilter/xt_string.c           |    5 ++++-
 2 files changed, 7 insertions(+), 1 deletions(-)

diff --git a/include/linux/netfilter/xt_string.h b/include/linux/netfilter/xt_string.h
index bb21dd1..2de02e9 100644
--- a/include/linux/netfilter/xt_string.h
+++ b/include/linux/netfilter/xt_string.h
@@ -4,6 +4,8 @@
 #define XT_STRING_MAX_PATTERN_SIZE 128
 #define XT_STRING_MAX_ALGO_NAME_SIZE 16
 
+#define XT_STRING_FLAG_IGNORECASE 0x01
+
 struct xt_string_info
 {
 	u_int16_t from_offset;
@@ -12,6 +14,7 @@ struct xt_string_info
 	char 	  pattern[XT_STRING_MAX_PATTERN_SIZE];
 	u_int8_t  patlen;
 	u_int8_t  invert;
+	u_int8_t  flags;
 
 	/* Used internally by the kernel */
 	struct ts_config __attribute__((aligned(8))) *config;
diff --git a/net/netfilter/xt_string.c b/net/netfilter/xt_string.c
index 72f694d..853ca35 100644
--- a/net/netfilter/xt_string.c
+++ b/net/netfilter/xt_string.c
@@ -46,6 +46,7 @@ string_mt_check(const char *tablename, const void *ip,
 {
 	struct xt_string_info *conf = matchinfo;
 	struct ts_config *ts_conf;
+	int flags = TS_AUTOLOAD;
 
 	/* Damn, can't handle this case properly with iptables... */
 	if (conf->from_offset > conf->to_offset)
@@ -54,8 +55,10 @@ string_mt_check(const char *tablename, const void *ip,
 		return false;
 	if (conf->patlen > XT_STRING_MAX_PATTERN_SIZE)
 		return false;
+	if (conf->flags & XT_STRING_FLAG_IGNORECASE)
+		flags |= TS_IGNORECASE;
 	ts_conf = textsearch_prepare(conf->algo, conf->pattern, conf->patlen,
-				     GFP_KERNEL, TS_AUTOLOAD);
+				     GFP_KERNEL, flags);
 	if (IS_ERR(ts_conf))
 		return false;
 
-- 
1.5.4.3

Re: [PATCH 6/8 v2] netfilter: fix string extension for case insensitive pattern matching

[Patrick McHardy]

Joonwoo Park wrote:
> The flag XT_STRING_FLAG_IGNORECASE indicates case insensitive string
> matching.
> netfilter can find cmd.exe, Cmd.exe, cMd.exe and etc easily.

I've asked you to provide some more information about this patch:

<quote>
A few words for the changelog why this won't break compatibility
by changing the data structure would be nice, especially
considering CRIS that doesn't perform any padding itself (you
might be lucky though because of aligned attribute, but please
verify that yourself).
</quote>

Re: [PATCH 6/8 v2] netfilter: fix string extension for case insensitive pattern matching

[Joonwoo Park]

On Wed, Jun 25, 2008 at 01:03:38PM +0200, Patrick McHardy wrote:
> Joonwoo Park wrote:
>> The flag XT_STRING_FLAG_IGNORECASE indicates case insensitive string
>> matching.
>> netfilter can find cmd.exe, Cmd.exe, cMd.exe and etc easily.
>
> I've asked you to provide some more information about this patch:
>
> <quote>
> A few words for the changelog why this won't break compatibility
> by changing the data structure would be nice, especially
> considering CRIS that doesn't perform any padding itself (you
> might be lucky though because of aligned attribute, but please
> verify that yourself).
> </quote>
>

I missed it and added few words for the changelog.

Thanks Patrick!

---
netfilter: fix string extension for case insensitive pattern matching

The flag XT_STRING_FLAG_IGNORECASE indicates case insensitive string
matching.
netfilter can find cmd.exe, Cmd.exe, cMd.exe and etc easily.

Also this patch won't break compatibility since we have 2 bytes space
between invert and config, due to config was 8 bytes aligned.
So we can put a new value 'flags' in xt_string_info between invert and
config safely.
If previous version of iptables interact with this kernel, to kernel
side, flags will be set to zero and to user side, flags will be ignored.

Signed-off-by: Joonwoo Park <joonwpark81@gmail.com>
---
 include/linux/netfilter/xt_string.h |    3 +++
 net/netfilter/xt_string.c           |    5 ++++-
 2 files changed, 7 insertions(+), 1 deletions(-)

diff --git a/include/linux/netfilter/xt_string.h b/include/linux/netfilter/xt_string.h
index bb21dd1..2de02e9 100644
--- a/include/linux/netfilter/xt_string.h
+++ b/include/linux/netfilter/xt_string.h
@@ -4,6 +4,8 @@
 #define XT_STRING_MAX_PATTERN_SIZE 128
 #define XT_STRING_MAX_ALGO_NAME_SIZE 16
 
+#define XT_STRING_FLAG_IGNORECASE 0x01
+
 struct xt_string_info
 {
 	u_int16_t from_offset;
@@ -12,6 +14,7 @@ struct xt_string_info
 	char 	  pattern[XT_STRING_MAX_PATTERN_SIZE];
 	u_int8_t  patlen;
 	u_int8_t  invert;
+	u_int8_t  flags;
 
 	/* Used internally by the kernel */
 	struct ts_config __attribute__((aligned(8))) *config;
diff --git a/net/netfilter/xt_string.c b/net/netfilter/xt_string.c
index 72f694d..853ca35 100644
--- a/net/netfilter/xt_string.c
+++ b/net/netfilter/xt_string.c
@@ -46,6 +46,7 @@ string_mt_check(const char *tablename, const void *ip,
 {
 	struct xt_string_info *conf = matchinfo;
 	struct ts_config *ts_conf;
+	int flags = TS_AUTOLOAD;
 
 	/* Damn, can't handle this case properly with iptables... */
 	if (conf->from_offset > conf->to_offset)
@@ -54,8 +55,10 @@ string_mt_check(const char *tablename, const void *ip,
 		return false;
 	if (conf->patlen > XT_STRING_MAX_PATTERN_SIZE)
 		return false;
+	if (conf->flags & XT_STRING_FLAG_IGNORECASE)
+		flags |= TS_IGNORECASE;
 	ts_conf = textsearch_prepare(conf->algo, conf->pattern, conf->patlen,
-				     GFP_KERNEL, TS_AUTOLOAD);
+				     GFP_KERNEL, flags);
 	if (IS_ERR(ts_conf))
 		return false;
 
-- 
1.5.4.3
---

Re: [PATCH 6/8 v2] netfilter: fix string extension for case insensitive pattern matching

[Patrick McHardy]

Joonwoo Park wrote:
> netfilter: fix string extension for case insensitive pattern matching
> 
> The flag XT_STRING_FLAG_IGNORECASE indicates case insensitive string
> matching.
> netfilter can find cmd.exe, Cmd.exe, cMd.exe and etc easily.
> 
> Also this patch won't break compatibility since we have 2 bytes space
> between invert and config, due to config was 8 bytes aligned.
> So we can put a new value 'flags' in xt_string_info between invert and
> config safely.
> If previous version of iptables interact with this kernel, to kernel
> side, flags will be set to zero and to user side, flags will be ignored.

This makes it hard for userspace to know whether insensitive
search is supported or not, similar for possible future flag
additions.

Please

- add a new revision for userspace to know whether flags are
   supported at all. You don't need to duplicate the checkentry
   function for this since the structure layout didn't change,
   you can simply parse flags dependant on match->revision.
   Userspace can use the existance of the new revision as
   indication that flags are supported.

- check for unknown flags and return an error

Re: [PATCH 6/8 v2] netfilter: fix string extension for case insensitive pattern matching

[Joonwoo Park]

2008/6/26 Patrick McHardy <kaber@trash.net>:
> Joonwoo Park wrote:
>>
>> netfilter: fix string extension for case insensitive pattern matching
>>
>> The flag XT_STRING_FLAG_IGNORECASE indicates case insensitive string
>> matching.
>> netfilter can find cmd.exe, Cmd.exe, cMd.exe and etc easily.
>>
>> Also this patch won't break compatibility since we have 2 bytes space
>> between invert and config, due to config was 8 bytes aligned.
>> So we can put a new value 'flags' in xt_string_info between invert and
>> config safely.
>> If previous version of iptables interact with this kernel, to kernel
>> side, flags will be set to zero and to user side, flags will be ignored.
>
> This makes it hard for userspace to know whether insensitive
> search is supported or not, similar for possible future flag
> additions.
>
> Please
>
> - add a new revision for userspace to know whether flags are
>  supported at all. You don't need to duplicate the checkentry
>  function for this since the structure layout didn't change,
>  you can simply parse flags dependant on match->revision.
>  Userspace can use the existance of the new revision as
>  indication that flags are supported.
>
> - check for unknown flags and return an error
>

Hi Patrick,
Thanks for your advice.
Since I'm traveling America, so maybe I could catch up these things
you've asked next weekend.
I don't need to rush, do I?

Thanks again,
Joonwoo