(usagi-users 03857) Re: Some UMIP fixes
Noriaki TAKAMIYA <[email protected]> Thu, 21 Jun 2007 22:45:40 +0900 (JST)
| Newsgroups | gmane.linux.ipv6.usagi.users |
|---|---|
| Message-ID | <[email protected]> |
Hi, Romain, I'm sorry to be silence too long. >> Thu, 7 Jun 2007 19:25:05 +0200 >> [Subject: (usagi-users 03856) Some UMIP fixes] >> Romain KUNTZ <[email protected]>.... > Hi, > > I've been using UMIP a bit these days an found some issues and > patches I'd like to share here. Take a look at the enclosed patch, > some comments below: First of all, I appreciate your patch, again. Could you find my comments below? > * chkconf_kernel.sh > The chkconf_kernel.sh script is not up-to-date with the 2.6.21 kernel > options. The patch fixes that by fixing a typo for XFRM_SUB_POLICY > and removing the check for the IPV6_ADVANCED_ROUTER option. > > * config.h.in, configure.ac, mh.c > This is actually the fix sent few days ago by Masahide Nakamura in > the MIPL ML for the RH issue. These two fixes are fine. > * src/Makefile.am > I am experiencing an issue on some PC config at compilation time when > the gram.y file is processed: gram.y includes mn.h and complains > about pthread_rwlock_t that is used in mn.h (whereas no other file > including mn.h ever complains). > Adding -D_GNU_SOURCE flag at compilation fixes the problem (although > this should be implicit with the AC_GNU_SOURCE located in the > confiure.ac file, this does not work with this specific gram.y file. > Maybe related to a lex/yacc issue?) Could you show your environment? We don't see such a problem on the following environment: - Debian Etch - SuSE 9.3 - Ubuntu 7.0.4 Feisty Fawn > * src/mn.c > Cosmetic change (fixes a typo) It looks fine. > * src/movement.c > Several things here: > - Cosmetic changes to avoid warnings at compilation time. > - Changed the definition of process_nlmsg to avoid warnings at > compoilation > - There is an issue regarding some of the BLACKHOLE rules installed > by md_block_rule_add/del and mn_block_rule_add/del: when exiting > mip6d, I always have a remaining "from all blackhole" rule. It should be fine, but can I have more time to check? I find the other problem, but it is another issue. > * src/proc_sys.h > - Cosmetic change to avoid warnings at compilation time. It is OK. > * src/rtnl.h > - Added a new priority for the HoA blackhole rule It is related to the fix in src/movement.c, OK. > * 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 RK). It is OK. Romain, could you split your patch regarding as the point to be fixed? Some of them are OK, but the other needs to be checked. Regards, -- Noriaki TAKAMIYA