Forum Replies Created
-
Posts
-
Hmm, this is pretty strange, probably the driver start setting you have used have a problem in this case. As I had mentioned before driver should start after NDIS.SYS and before TCPIP.SYS in order to intercept TCP/IP protocols registration and binding adapter’s process. This is the only required condition. Regretfully I can’t say more without putting hands on these systems…
Probably all these systems have some common software which prevents WinpkFilter from working normally. Without using the kernel debugger it’s usually not easy to detect the conflicting software. What kind of VPN software do you use? It can be also based on NDIS hooking drivers and conflict with WinpkFilter.
WInpkFilter is fully compatible with Windows built-in firewall and ICS.
1. I should to change these values (Group: “Streams Drivers” Start: 1) in the registry?
Current WinpkFilter build installed with the settings above.
3. If I later unistall Kerio, these values will not break work of the driver?
No, there is no actual dependency from the Kerio driver.
Driver can’t to detect network cards RTL8139/810X Family Sad
(Windows XP SP2) – any comments?There are no problems with any network cards because the details of realization or each one are hidden inside NDIS. The problem you have expirienced rather specific to the software installed (firewalls, traffic counting monitoring tools and etc.). Also it may appear that the driver start load order is broken…
Can you give more info about it?
How I must install driver?
What version of Kerio has conflict with WinPKFilter?Actually the only requirement for WinpkFilter driver is starting between NDIS.SYS and TCPIP.SYS, but some hooking drivers(like the case with Kerio) may add additional conditions (like start before or after these drivers). In case with Kerio the conflict was detected when WinpkFilter driver strats before Kerio driver, but the problem gone if WinpkFilter starts after Kerio. Thats why we had changed load order to “Streams Drivers” Start:1. However, the situation may change from release to release (by the date of the post mentioned the problem was with the newest Kerio release, and there were no problems with previous releases).
You can create PASS ALL rule for the interface you want to disable NeT Firewall at. This rule should have highest priority(an example ID=1).
Could you please e-mail the screenshot of the log with blocked packets and the screenshot of the rules list to support@ntkernel.com.
By default only fragmented packets are dropped in low security mode, others are passed.
What the reason is speciifed for blocked packets? Is it a rule or just General Security Policy?
You can start from PASSTHRU sample from WinpkFilter package and extend it to the firewall by adding packet analises routines.
At the NDIS level program context is not available for the packet but you can limit bandwidth by IP-address/IP-protocol/port information. So, an example you can limit HTTP bandwidth and etc… In order to limit bandwidth for certain application you would also need application level filter (LSP, TDI filter) in order to match application name to IP/protocol/port information extracted from the packet at the NDIS level.
Hope it helps…
WinPkFilter can send it ?
Yes
if WinPkFilter can send the packet, How can i send them ?
Using SendPacketToAdapter function.
I am not touching the ACK/SYN at all, I am modifying the ip_len and the payload only. that too, if the packet has payload then only i am modifying the packet.
If you change length of the TCP packet then you should modify SYN/ACK fields.
Is there any way I can drop the old packet? How can I remove it from the local stack?
Actually you already drop the original packet, but system generates it again and again (because your invalid packet is droped by remote system).
Probably you modify TCP packet and do this wrong. Destination system drops your packet and don’t send ACK for it, thats why your local stack sends packet again after some timeout.
You can try using MSTCP_FLAG_LOOPBACK_BLOCK for the adapter you work over. It drops incoming packets if source MAC is the same as local MAC. Regretfully there is no way to prevent packet indication on NT4 at all.
Personally I use this one
//
// Function recalculates IP checksum
//
VOID
RecalculateIPChecksum (
iphdr_ptr pIpHeader
)
{
unsigned short word16;
unsigned int sum = 0;
unsigned int i = 0;
PUCHAR buff;// Initialize checksum to zero
pIpHeader->ip_sum = 0;
buff = (PUCHAR)pIpHeader;// Calculate IP header checksum
for (i = 0; i < pIpHeader->ip_hl*sizeof(DWORD); i=i+2)
{
word16 = ((buff<<8)&0xFF00)+(buff[i+1]&0xFF);
sum = sum+word16;
}// keep only the last 16 bits of the 32 bit calculated sum and add the carries
while (sum>>16)
sum = (sum & 0xFFFF)+(sum >> 16);// Take the one’s complement of sum
sum = ~sum;pIpHeader->ip_sum = htons((unsigned short) sum);
}
