-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, I am using a prio queue with two bands. Packets in queue 0 are send directly whereas packets of queue 1 are analyzed by a kernel module. In order to redirect the packets of queue 1 I modified the prio_dequeue function of sch_prio.c (if prio==1 -> call kernel_module, do processing, send skb). Although doing only an easy check within the kernel module the throughput drops from about 90 MBit/s to 5 MBit/s! In contrary when omitting tc and registering the kernel module directly to the netfilter architecture, the throughput more or less stays constant. Is the dequeue funtion the wrong place to intercept packets? It would be great if I could use tc and the kernel module without a huge performance degradation. Any hint is welcome Thanks Andreas -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iD8DBQFAL50u68eBr8WIgcgRAvYmAJ9EcOmfU9LNaU3LfiuhtxGxMmbxTwCdF13A wcJkXksfVThAWwa+lTgcLX0= =C1Mj -----END PGP SIGNATURE----- _______________________________________________ LARTC mailing list / LARTC@xxxxxxxxxxxxxxx http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
