I am currently debugging this and had planned to send an email asking for suggestions when code review was nearing completion. This bug has been elusive for a long time. The packet format themselves looked fine as I have found a way to test them and have successfully tested them.
The problem is that ovs-vswitchd crashes and happens only in the automation i.e "make check", or "make dist-check", since user-space code has been successfully tested manually and I couldn't find a way to run gdb during automation it's been actively debugged on and off. I did try "make check-valgrind" as suggested by you but it didn't reveal anything. Addtionally, 1. thought automation didn't support eth_types 0x8847 and 0x8848 and looked for asserts and didn't find it, so this was ruled out. 2. thought ofpbuf handling was not correct (highly unlikely since user-space code was tested), so completely commented "push_mpls" function in lib/packets.c and still got ovs-vswitchd crash. 3. currently assuming netlink attribute handling for mpls is wrong (unlikely same argument as above). Right now I am just doing code read to see if I can spot the bug, if you have any suggestion please let me know. Please note I would have brought this problem once you were done with the code-review. Thanks, Ravi On Tue, Jun 12, 2012 at 11:13 AM, Ben Pfaff <b...@nicira.com> wrote: > Oh, I saw one testsuite failure also: > 448: ofproto-dpif - controller FAILED > (ofproto-dpif.at:323) > Does this pass for you. > > On Tue, Jun 12, 2012 at 11:11 AM, Ben Pfaff <b...@nicira.com> wrote: >> Ravi, I'm more or less happy with the userspace code here. I mostly >> have style kinds of concerns. (I haven't fully scrutinized every line >> though.) >> >> I did notice that compose_dec_mpls_ttl() decrements "ttl" twice (it >> has --ttl in two places). >> >> I believe that Pravin is going to review the kernel portions of this >> code now. Once he's happy, I'll suggest touch-ups for userspace, or >> just do them myself. >> >> On Mon, Jun 11, 2012 at 03:00:45PM -0700, ravi kerur wrote: >>> Attached latest mpls and qinq patch. It is based off latest git master. >>> >>> Incorporates code review comments, takes care of performance issue on >>> offload. >>> >>> Thanks, >>> Ravi >>> >>> On Fri, Jun 1, 2012 at 2:32 PM, ravi kerur <rke...@gmail.com> wrote: >>> > This is based off >>> > >>> > commit 73c0ce349ba8d13a63a249a56aad0bec6e6caf26 >>> > Author: Joe Stringer <j...@wand.net.nz> >>> > Date: Tue May 29 00:38:21 2012 +1200 >>> > >>> > it can wait until your return. I will probably do some performance >>> > testing in the mean time for both mpls + qinq tcp offload. >>> > >>> > Thanks, >>> > Ravi _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
