Re: rtrequest1 error cleanup

On 19/08/15(Wed) 13:06, Alexander Bluhm wrote: Hi, I have converted all calls to rtrequest1() into a common pattern. In the man page I tried to clarify the usage of the returned route. ok? Ok. Your manpage change might lead to confusions because the code is tricky. If you pass a rt

Useless rtfree(9)

Instead of asking for a reference in rtrequest1(9) simply pass NULL and let the function call rtfree(9) for us. rt_missmsg() does not need the route to be alive, so it's fine. Ok? Index: net/route.c === RCS file:

Re: ipip interrupt path and list iteration

On 09/07/15(Thu) 09:00, Martin Pieuchot wrote: Most of the network interrupt paths are now free from global list iterations. The rule I tried to follow is simple. If you already have an ifp pointer and need a per-ifp resource use it, otherwise do a route lookup. Diff below converts one of

Re: Partially unlock the em(4) interrupt handler

On 17/08/15(Mon) 19:55, Mark Kettenis wrote: The diff below is a first step towards running the em(4) interrupt handler without grabbing the kernel lock. It runs the rx completion path without the lock, which is the important bit to be able to test the work that has been going on to make the

Re: rtrequest1 error cleanup

I have followed your previous conversation and this diff looks good to me as well. Ok mikeb

Kill IN6_IFF_NODAD

KAME people chose to add states to the address descriptors in order to implement various IPv6 automagic features. But IN6_IFF_NODAD is not a real state, it's just a hack for some spaghetti code. Diff belows get rids of this flag by shuffling some code around and only using the IN6_IFF_TENTATIVE

Re: v6 autoconf, where have my connected routes gone?

On 18/08/15(Tue) 23:22, Martin Pieuchot wrote: On 18/08/15(Tue) 14:00, Stuart Henderson wrote: On 2015/08/18 14:27, Martin Pieuchot wrote: On 18/08/15(Tue) 13:05, Stuart Henderson wrote: I'm trying to add v6 to my second ISP connection on pppoe1, unlike my first ISP, this one

Re: ipip interrupt path and list iteration

On Wed, Aug 19, 2015 at 02:41:22PM +0200, Martin Pieuchot wrote: Diff below converts one of the few remaining iteration on the global list of interfaces (ifnet). /* Check for local address spoofing. */ - if (((ifp = if_get(m-m_pkthdr.ph_ifidx)) == NULL || - !(ifp-if_flags

[patch] ctags -u

Hi, ctags(1) has some crazy system(3) call I want to remove, that are in -u logic (update 'tags' file). Instead of switch from system(3) to fork(2)/exec(2), I think changing the logic for -u would be better. Currently, when updating, the logic is: - generate tags in memory for passed

Re: Partially unlock the em(4) interrupt handler

On 2015/08/17 19:55, Mark Kettenis wrote: The diff below is a first step towards running the em(4) interrupt handler without grabbing the kernel lock. It runs the rx completion path without the lock, which is the important bit to be able to test the work that has been going on to make the

Re: Useless rtfree(9)

On Wed, Aug 19, 2015 at 01:13:59PM +0200, Martin Pieuchot wrote: Instead of asking for a reference in rtrequest1(9) simply pass NULL and let the function call rtfree(9) for us. rt_missmsg() does not need the route to be alive, so it's fine. Ok? This is the diff that was done before in rev

Re: [patch] ctags -u

On Wed, 19 Aug 2015 20:21:08 +0200, Sebastien Marie wrote: Some minor comments inline. - todd Index: ctags.c === RCS file: /cvs/src/usr.bin/ctags/ctags.c,v retrieving revision 1.15 diff -u -p -r1.15 ctags.c --- ctags.c 8

libm: avoid left-shifting a negative integer

I saw this drifting by on FreeBSD (from Dimitry Andric): In libm's exp2(3), avoid left-shifting a negative integer, which is undefined. Replace it with the intended value, in a defined way. OK? Don't bother? Index: src/s_exp2.c

sockaddr initialize zero

Hi, Struct sockaddr_in and sockaddr_in6 should always be initialized to zero. Most of the kernel does this already, this diff fixes all other places there sin_family is assigned. Do not pass around pointers to uninitialized stack memory. There are some global route variables, I think the

Re: whois(1): fix lookup of XX.network

Stuart Henderson [st...@openbsd.org] wrote: When I added code to use whois.nic.XX for new TLDs I forgot to think about one case, where the TLD is a substring of one of the traditional TLDs who have to use the old whois-servers.net method. Specifically, trying to lookup a .network name will

ssh-keygen.1: consistently use host_key/user_key

In the CERTIFICATES section of ssh-keygen.1, be consistent about using host_key and user_key for the respective key types. Index: ssh-keygen.1 === RCS file: /cvs/src/usr.bin/ssh/ssh-keygen.1,v retrieving revision 1.126 diff -u -p

Re: LibreSSL 2.2.2 release

On Wed, Aug 12, 2015 at 1:29 PM, Mark Kettenis mark.kette...@xs4all.nl wrote: Date: Tue, 11 Aug 2015 09:15:43 -0600 From: Bob Beck b...@obtuse.com I'm wondering out loud if these versions should follow the openbsd shlib major minor numbers. That is where we are careful about semantic

Re: sockaddr initialize zero

Struct sockaddr_in and sockaddr_in6 should always be initialized to zero. In the kernel, for sure. Just in case a bcmp is run by someone, or a whole sockaddr is copied out to userland. In userland, the policy is not quite the same. There (as guenther has explained in various forums) the idea

rtrequest1 error cleanup

Hi, I have converted all calls to rtrequest1() into a common pattern. In the man page I tried to clarify the usage of the returned route. ok? bluhm Index: sys/net/route.c === RCS file: