* gianfar: incorrect TCP checksum when timestamps are enabled
@ 2011-01-10 6:22 Alex Dubov
0 siblings, 0 replies; 2+ messages in thread
From: Alex Dubov @ 2011-01-10 6:22 UTC (permalink / raw)
To: linuxppc-dev
Greetings.=0A=0AI'm working on MPC8548 based board (I started with 2.6.37-r=
c7, now upgraded=0Ato post release 2.6.37 git head).=0A=0AIt so appears, th=
at when tcp timestamps are enabled (default setting)=0ATCP checksums of out=
going SYN packets are calculated incorrectly. This=0Acauses remote machines=
to reject such packets, effectively preventing any=0Auseful network use. T=
he problem is reproducible every time.=0A=0AOther protocols appear to work =
reliably (UDP/DHCP do work).=0A=0AIf tcp timestamps are disabled, checksums=
are calculated correctly.=0A=0ANetdev guys tend to think that the problem =
is with hardware checksum=0Aoffloading, the theory I'm going to test now.=
=0A=0A=0A=0A
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: gianfar: incorrect TCP checksum when timestamps are enabled
@ 2011-01-17 3:10 Alex Dubov
0 siblings, 0 replies; 2+ messages in thread
From: Alex Dubov @ 2011-01-17 3:10 UTC (permalink / raw)
To: linuxppc-dev
Hi.=0A=0AIt seems, to have TCP working at all with the recent kernel git I =
have to=0Aclear the NETIF_F_IP_CSUM feature of the gianfar driver, otherwis=
e checksums=0Aturn out to be incorrect.=0A=0ADoes anybody have a clue on wh=
ere the actual problem may be?=0A=0A=0A=0A
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2011-01-17 3:10 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-01-10 6:22 gianfar: incorrect TCP checksum when timestamps are enabled Alex Dubov
2011-01-17 3:10 Alex Dubov
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.