Re: [Lksctp-developers] Increment of SctpChecksumErrors

[Sridhar S. <sr...@us...>]

Yes, This is bug and we are incrementing checksum errors twice.
Currently we don't have a counter to count errors other than checksum errors.
So looks like we used the same counter even for other errors which is wrong.
The correct solution is to remove 
   SCTP_INC_STATS(SctpChecksumErrors) 
from sctp_rcv().
sctp_rcv() will be called only in BH context. So SCTP_INC_STATS_BH is the
right macro to use.

Thanks
Sridhar

On Mon, 10 Nov 2003, Chandrasekhar V wrote:

> Greetings!!
>  
> I am confused to see the below code while going through sctp_rcv()
> function. Could someone please explain if I am missing some thing here?
>  
> -----------------------------
> Code from function sctp_rcv_checksum():
>  
> sctp_rcv_checksum() function  increments checksum stats using the code
>             SCTP_INC_STATS_BH(SctpChecksumErrors);
> ---------------------------------------------
>  
> Code from function sctp_rcv():
>  
> if (sctp_rcv_checksum(skb) < 0)
>                 goto bad_packet;
> 
> There is again code present in this function as below:
> bad_packet:
>         SCTP_INC_STATS(SctpChecksumErrors);
> ----------------------------------
>  
> Are the checksum stats for SctpChecksumErrors being incremented twice? 
> Can in_softirq() value be different each time sctp_rcv() is executed?
>  
> Thanks for your time,
> Chandu
>  
>             
>  
>        
>  
>