On 2026 Apr 10 (Fri) at 13:12:54 -0400 (-0400), Sonic wrote: :Not sure what I'm doing wrong as I just started to enable ipv6 access :using an OpenBSD based router, running the latest snapshots. :The issue is that I lose general net access via ipv6 from client :systems after rebooting the router. The client systems correctly :configure themselves using SLAAC and I can ping6 systems on the :Internet, but general net access such as that from a browser fails - :https://test-ipv6.com/ claims no public ipv6 address. :However if I reload the pf rules after the reboot (pfctl -f :/etc/pf.conf) all client systems start to work fine. :Any clues on what I might be missing? : :Thank you, :Chris : You'll need to show your pf.conf for us to properly debug it, but I predict the issue is either you are using dns names in pf.conf (which only resolve at pfctl time...) or you forgot to add ()'s around interfaces, so it only has the addresses that were assigned at pfctl time. Probably sprinkle some (egress) or (em0) in your ruleset, especially if you receive your own addresses from slaac/dhcp. -- Do not sleep in a eucalyptus tree tonight.
Sunday, April 12, 2026
Re: Fix audio/csound on powerpc
On Tue, 7 Apr 2026 11:07:53 +0200 Raphael Graf <r@undefined.ch> wrote: > According to the latest bulk build report, building csound fails on powerpc: > > In file included from /usr/obj/ports/csound-7.0.0beta13/csound-7.0.0-beta.13/OOps/pffft.c:107: > /usr/lib/clang/19/include/altivec.h:13:2: error: "AltiVec support not enabled" > 13 | #error "AltiVec support not enabled" > | ^ > /usr/lib/clang/19/include/altivec.h:50:19: error: unknown type name 'vector' > 50 | static __inline__ vector signed char __ATTRS_o_ai vec_perm( > > > The diff below should "fix" this by disabling altivec on powerpc. > (I have not tested it on powerpc) > > ok? Your diff unbreaks csound on my PowerBook G4 from 2004. ok gkoehler@ By default, #include <altivec.h> works on powerpc64 and gives this #error on powerpc. One can use "cc -maltivec" to enable altivec on powerpc, but the result would crash on a Mac G3. It is better to disable altivec. Some packages (like xenocara/lib/pixman) can check for altivec at runtime, compile one file with -maltivec, and call that file only if the check is true. I see no such check in csound. --gkoehler > Index: Makefile > =================================================================== > RCS file: /cvs/ports/audio/csound/Makefile,v > retrieving revision 1.4 > diff -u -p -u -p -r1.4 Makefile > --- Makefile 2 Apr 2026 05:31:44 -0000 1.4 > +++ Makefile 7 Apr 2026 09:04:52 -0000 > @@ -44,6 +44,10 @@ LIB_DEPENDS = audio/liblo \ > devel/gettext,-runtime \ > net/curl > > +.if ${MACHINE_ARCH} == "powerpc" > +CFLAGS += -DPFFFT_SIMD_DISABLE > +.endif > + > TEST_ENV = OPCODE7DIR64=${WRKBUILD} > PORTHOME = ${WRKDIR} > >