Date: Fri, 21 Mar 2008 08:54:20 +0900 From: Pyun YongHyeon <pyunyh@gmail.com> To: Giulio Ferro <auryn@zirakzigil.org> Cc: Andrew Thompson <thompsa@freebsd.org>, freebsd-net@freebsd.org Subject: Re: VLAN trunking and fragmentation Message-ID: <20080320235420.GA70708@cdnetworks.co.kr> In-Reply-To: <47E298BD.7050807@zirakzigil.org> References: <20080313033029.GF16972@cdnetworks.co.kr> <47D92F9D.4070701@zirakzigil.org> <20080314001704.GA22788@cdnetworks.co.kr> <47DA4DCD.6050304@zirakzigil.org> <20080314101555.GF22788@cdnetworks.co.kr> <47DA682E.9010504@zirakzigil.org> <20080314122708.GG22788@cdnetworks.co.kr> <47DA9BC0.8030205@zirakzigil.org> <20080317043945.GA2503@cdnetworks.co.kr> <47E298BD.7050807@zirakzigil.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Mar 20, 2008 at 06:02:53PM +0100, Giulio Ferro wrote: > Pyun YongHyeon wrote: > >Please try latest attempt to fix re(4) issues. > > http://people.freebsd.org/~yongari/re/if_re.c > > http://people.freebsd.org/~yongari/re/if_rlreg.h > > > >This one is attempt to fix the following bug reports on re(4). > > - VLAN tagging does not work on multi-fragmented frames. > > - Non-working/instability on re(4) hardwares that uses MSI. > > - Unconditional VLAN tag stripping without respect to ifconfig > > configuration. > > - ENOBUFS on transmitting UDP frames under heavy network loads. > > - DAC does not work. > > > >This one removed previous checksum offload hack for checksum offload > >on PCIe hardwares so I'm not sure checksum offload still work on > >PCIe hardwares. Please test and let me know how it goes. > >Also make sure you have all hardware features enabled.(checksum > >offload, TSO, VLAN hardware tagging etc). > > > > > > Sorry for the slow answer. > > I've tried the new files, and they seem to work: > > # ifconfig re0 > re0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500 > options=19b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,TSO4> > ether 00:14:c1:40:c2:e8 > inet 192.168.60.1 netmask 0xffffff00 broadcast 192.168.60.255 > media: Ethernet autoselect (1000baseTX <full-duplex>) > status: active > > > The only problem I see now is the one I already mention in the last > mail. There are slow packets > in the stream of the communication: > > # ping -s 2000 192.168.100.2 > PING 192.168.100.2 (192.168.100.2): 2000 data bytes > 2008 bytes from 192.168.100.2: icmp_seq=0 ttl=64 time=0.264 ms > 2008 bytes from 192.168.100.2: icmp_seq=1 ttl=64 time=1001.186 ms [...] > 2008 bytes from 192.168.100.2: icmp_seq=41 ttl=64 time=0.181 ms > > > > It would be great if you could understand why this happens... > Try set tunable hw.re.msi_disable and let me know how it goes. (Add hw.re.msi_disable="1" to /boot/loader.conf) -- Regards, Pyun YongHyeon
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20080320235420.GA70708>