[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

(usagi-users 03905) Re: Some UMIP fixes

Dear Takamiya-san,

On 2007/07/18, at 19:40, Noriaki TAKAMIYA wrote:
It happened on a Debian Lenny with autoconf 2.61-4, flex 2.5.33-11,
byacc 20050813-1
Could you provide me your autoconf, flex, byacc version? I'll try to
dig this issue a bit more.

In case you'd need the patch, the fix is in makefile-gnu-source.diff

The list below is my environment:

gcc:      version 4.1.2 20061115 (prerelease) (Debian 4.1.1-21)
autoconf: autoconf (GNU Autoconf) 2.61(Debian 2.61-4)
flex    : flex 2.5.33(Debian 2.5.33-11)

I don't use byacc. It may be the difference.

So which yacc parser do you use? bison?
Anyway, I did not manage to reproduce the issue. My guess is that it came from a buggy yacc parser, because the problem only appeared when processing gram.y. So I don't think it's useful to apply this patch at this point. Let's check if in the future someone reports a similar problem.

* src/xfrm.c
- mip6d tries to install a policy that already exists and fails. Thus
we update it instead. See the comment in the patch (starting with

It is OK.

The fix is in xfrm.c.diff (but you may want to remove the commented lines once the fix has been applied).

When I applied this patch, mip6d always fails to send the first BU for the home registration.

After I stopped mip6d and run again, home registration succeeds normally.

  The log is the following:
Thu Jul 12 04:09:38 mh_send: sendmsg: Operation not permitted
Thu Jul 12 04:09:38 mn_send_bu_msg: mh_send failed  ret: -1

  As long as I see, the policy to send BU to the HA is not inserted.
  I'm investigation the reason, and did you see the same problem?

No, I do not have this issue (tested on a patched kernel, and on the GIT kernel). On both kernels it always succeeds to install the policy. Tell me if you need more tests or information from me.


Romain KUNTZ
Louis Pasteur University - Networks and Protocols Team