On 8 October 2013 15:45, Johannes Berg <johannes@xxxxxxxxxxxxxxxx> wrote: > On Tue, 2013-10-08 at 15:45 +0200, Johannes Berg wrote: >> On Tue, 2013-10-08 at 15:41 +0200, Ard Biesheuvel wrote: >> >> > Actually, as the size is always the same, it should be feasible to >> > alloc a couple of request structs at init time. would one for rx and >> > one for tx be sufficient? or is this code more reentrant than that? >> >> TX can run concurrently on multiple (four) queues using the same key. > > And maybe even more with injection ... I wouldn't go there. > OK, clear. If you think this patch has any merit at all, I am happy to modify it so that it kmalloc()s the request struct with GFP_ATOMIC at every invocation. However, personally I don't think this should be necessary and in fact my patch removes a stack allocation of u8[48] (from ieee80211_crypto_ccmp_decrypt() and from ccmp_encrypt_skb() in wpa.c) so it does even out a bit. regards, Ard. -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
