Hello linux-net, It seems like there's a need for a consistent method of reliably getting a callback when an skb is destructed. So far, it seems like this functionality has just been sort of hacked into the skb for each module that needs it (sock, netfilter). Rather than hack in another callback for our own purposes and changing code everywhere to make sure it gets called, we're thinking we should create a generic system that replaces sk_buff.destructor with a pointer to some sort of 'destructor' structure. This structure could be allocated out of a free pool and contain a stack of pointers to people who want callbacks, allowing for an ordered and consistent chain of events on destruction (hopefully without too much additional overhead). We could then load our network module (doing some packet intercepting stuff between dev and the protocols) and easily set up for callbacks on destruction packets we deem interesting. Anyway, we're playing with the idea... Seems like this could also be a move in the right direction for driver writers who could benefit by managing their own memory somewhat (though we don't have enough experience with drivers or specifics of skb allocation to know for sure who else this could benefit). Comments? Does this violate some 'larger vision' for the stack that we're not seeing? -- -Will :: AD6XL :: http://tyranny.egregious.net/~will/ Orton :: finger will@tyranny.egregious.net for GPG public key - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org
