| Message ID | 20130323021008.GC939@order.stressinduktion.org |
|---|---|
| State | RFC, archived |
| Delegated to: | David Miller |
| Headers | show |
On 22/03/13 10:10 PM, Hannes Frederic Sowa wrote: > On Fri, Mar 22, 2013 at 08:07:22PM -0400, rebelyouth wrote: >> On 22/03/13 07:54 PM, Eric Dumazet wrote: >>> On Fri, 2013-03-22 at 19:07 -0400, rebelyouth wrote: >>>> Hi, >>>> >>>> I'm writing again for the same issue about : >>>> >>> Your mail is so long that we are lost. >>> >>> Have you done some research to tell if the bug is : >>> >>> 1) Sending corrupt frames >>> >>> 2) receive corrupt frames >>> >>> >> Hi Eric and Hannes, >> >> The server receiver errors when I use the atheros: >> >> RX packets:78125369 errors:58741 dropped:0 overruns:58741 frame:0 >> >> but on the client side there aren't any error but the file is corrupted >> >> I rebooted the server and did the test with the Realtek: >> >> file ok , no errors in both sides >> >> I also reset again the server and did a test on windows using samba and >> the Atheros: >> >> file ok , no errors in both sides > Could you try this patch? It solves the problem with an atl1e nic on my > friends computer. This does not solve the real problem, which I have > not yet understood (wrong packets will be discarded now and will be > hopefully resend). > > [PATCH RFC] atl1e: don't rely on hardware checksumming > > Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org> > --- > drivers/net/ethernet/atheros/atl1e/atl1e_main.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/drivers/net/ethernet/atheros/atl1e/atl1e_main.c b/drivers/net/ethernet/atheros/atl1e/atl1e_main.c > index e1f1b2a..0363e5b 100644 > --- a/drivers/net/ethernet/atheros/atl1e/atl1e_main.c > +++ b/drivers/net/ethernet/atheros/atl1e/atl1e_main.c > @@ -1337,6 +1337,8 @@ static inline void atl1e_rx_checksum(struct atl1e_adapter *adapter, > u16 err_flags; > > skb_checksum_none_assert(skb); > + goto hw_xsum; > + > pkt_flags = prrs->pkt_flag; > err_flags = prrs->err_flag; > if (((pkt_flags & RRS_IS_IPV4) || (pkt_flags & RRS_IS_IPV6)) && Hi Hannes, I applied your patch on my the kernel 3.7.9 and look like the patch do what you said before : "This does not solve the real problem, which I have not yet understood (wrong packets will be discarded now and will be hopefully resend). " There are errors on the RX side of the server and the speed is 5~10% less but the files are ok in nfsv4 , SFTP/FTP and samba. Maybe the Atheros drivers for windows had a patch for the checksum already in the drivers and the chipset have some hardware bug, but for me your patch is already an achievement and I really thank you for your time and consideration. You will upload this patch to be included in the kernel? I hope so, a lot of people had problem with this issue before and will be fantastic if this fix will be included. Thank you Rebelyouth -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Sat, Mar 23, 2013 at 12:54:25AM -0400, rebelyouth wrote: > I applied your patch on my the kernel 3.7.9 and look like the patch do > what you said before : > > "This does not solve the real problem, which I have not yet understood > (wrong packets will be discarded now and will be hopefully resend). " > > There are errors on the RX side of the server and the speed is 5~10% > less but the files are ok in nfsv4 , SFTP/FTP and samba. > > Maybe the Atheros drivers for windows had a patch for the checksum > already in the drivers and the chipset have some hardware bug, but for > me your patch is already an achievement and I really thank you for your > time and consideration. The upstream driver (which is a tarball I found with google AR81Family-linux-v1.0.1.14.tar.gz) has this function ifdefed out. I have a couple of more things I want to test as soon as I have access to the hardware again. If I don't find a better solution I will submit a patch which disables rx checksumming as a whole for inclusion to net-next. Thanks, Hannes -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Hannes Frederic Sowa <hannes@stressinduktion.org> wrote: > The upstream driver (which is a tarball I found with google > AR81Family-linux-v1.0.1.14.tar.gz) has this function ifdefed out. I > have a couple of more things I want to test as soon as I have access > to the hardware again. If I don't find a better solution I will > submit a patch which disables rx checksumming as a whole for inclusion > to net-next. In addition to my other mail: Please have a look at Debian Bug #558426 <http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=558426> which is about TSO and TCP/IPv6 on atl1e which had problems as well. I wouldn't find it surprising if any offloading or hardware checksumming is broken on that hardware. Grüße, Sven.
diff --git a/drivers/net/ethernet/atheros/atl1e/atl1e_main.c b/drivers/net/ethernet/atheros/atl1e/atl1e_main.c index e1f1b2a..0363e5b 100644 --- a/drivers/net/ethernet/atheros/atl1e/atl1e_main.c +++ b/drivers/net/ethernet/atheros/atl1e/atl1e_main.c @@ -1337,6 +1337,8 @@ static inline void atl1e_rx_checksum(struct atl1e_adapter *adapter, u16 err_flags; skb_checksum_none_assert(skb); + goto hw_xsum; + pkt_flags = prrs->pkt_flag; err_flags = prrs->err_flag; if (((pkt_flags & RRS_IS_IPV4) || (pkt_flags & RRS_IS_IPV6)) &&