On Mon, Oct 20, 2014 at 10:02:17AM +0100, Martin Townsend wrote: > Hi Alex, > > I've completely forgotten what the original problem was it was so long ago :) , let me dig through the old emails and see if I can jog my memory. > > My intention was to do the kfree_skb patch in my next submission but I'll see what I can do. > yea, the problem a general error handling issue in mainly two parts: - detection errors. Means the NET_RX_FOO conversion to errno. But you put out the netif_rx call outside of lowpan_header_create which is very well. I approve that one. Now it's important that the "lowpan give to upper layer" functions not returning ERRNO's and NET_RX_FOO. Again, errno is a negative value and NET_RX_FOO is 0 or 1. We can't check on a error with if (ret < 0) or (ret == NET_RX_DROP). Also (ret != NET_RX_SUCCESS) will not work because it's "1" and "0" indicate successful or sometimes not successful because NET_RX_DROP was returned and this is "0". Lot of confusing here. For the "lowpan give to upper layer" functions doesn't matter if errno is returned or NET_RX_FOO, but don't a mix of both! The "lowpan_header_create" should return errno's. - Second issue was a complete wrong reaction on error handling and memory managment with "kfree_skb" which needs a fix of the above one at first to introduce a correct error handling. Now I see a little bit the fix of the first one. But not the second one. Or should these things all fine now, otherwise I will take a more careful review. - Alex -- To unsubscribe from this list: send the line "unsubscribe linux-bluetooth" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
