Re: Creating /dev/dk* nodes so as to be persistent
On Fri, 18 Oct 2013, Thomas Mueller wrote: Maybe I could install NetBSD root on a partition above the /dev/dk* range, as long as I have the devpubd line in /etv/rc.conf? I'd be surprised if that works. How will /etc/rc.d/fsck_root work before devpubd has been started, if there is no appropriate /dev/dk* node? Why not just make more /dev/dk* nodes? --apb (Alan Barrett)
Netbsd/amd64 build failure with both MKGCC and MKLLVM set
Hi, On my my main developpement machine running -current NetBSD/amd64, i do see a problem if i try to build both gcc and clang (MKGCC=yes and MKLLVM=yes). The default compiler remaining unchanged (gcc). njoly@lanfeust [NetBSD/src] ./build.sh -Uu -m amd64 -N 1 release [...] create libc/libunwind.d /local/src/NetBSD/dest/amd64/usr/include/g++/bits/c++0x_warning.h:31:2: error: #error This file requires compiler and library support for the up coming ISO C++ standard,C++0x. This support is currently experimental,and must be enabled with the -std=c++0x or -std=gnu++0x compiler options . nbmkdep: compile failed. *** Failed target: libunwind.d *** Failed command: CC=/local/src/NetBSD/tool/bin/x86_64--netbsd-gcc /local/src/NetBSD/tool/bin/nbmkdep -f libunwind.d -- --sysroot=/local/src/NetBSD/dest/amd64 -D_LIBC -DLIBC_SCCS -DSYSLIBC_SCCS -D_REENTRANT -D_DIAGNOSTIC -DMLIBDIR=\\-DHESIOD -DINET6 -DNLS -DYP -I/local/src/NetBSD/src/lib/libc/include -I/local/src/NetBSD/src/lib/libc -I/local/src/NetBSD/src/sys -I/local/src/NetBSD/src/lib/libc/compat/../locale -I/local/src/NetBSD/src/lib/libc/compat/stdlib -I/local/src/NetBSD/src/lib/libc/compat/../stdlib -D__BUILD_LEGACY -I/local/src/NetBSD/src/lib/libc/../../common/lib/libc/quad -I/local/src/NetBSD/src/lib/libc/../../common/lib/libc/string -I/local/src/NetBSD/src/lib/libc/../../common/lib/libc/arch/x86_64/string -D__DBINTERFACE_PRIVATE -I/local/src/NetBSD/src/libexec/ld.elf_so -I/local/src/NetBSD/src/lib/libc/dlfcn -I/local/src/NetBSD/src/lib/libc/gdtoa -I/local/src/NetBSD/src/lib/libc/locale -DNO_FENV_H -I/local/src/NetBSD/src/lib/libc/arch/x86_64/gdtoa -DWITH_RUNE -I/local/src/NetBSD/src/lib/libc -DPOSIX_MISTAKE -DCOMPAT__RES -DUSE_POLL -DPORTMAP -DWIDE_DOUBLE -DALL_STATE -DUSG_COMPAT -D_FORTIFY_SOURCE=2 /local/src/NetBSD/src/sys/lib/libunwind/libunwind.cxx *** Error code 1 njoly@lanfeust [NetBSD/src] make show-params | grep -e GCC -e LLVM -e PCC HAVE_GCC = '45' HAVE_LLVM = (undefined) HAVE_PCC = (undefined) MKGCC = 'yes' MKGCCCMDS = 'yes' MKLLVM = 'yes' MKPCC = 'no' Thanks. -- Nicolas Joly Biology IT Center Institut Pasteur, Paris.
Re: Creating /dev/dk* nodes so as to be persistent
On Fri, 18 Oct 2013, Thomas Mueller wrote: Why not just make more /dev/dk* nodes? Problem is how to make more /dev/dk* nodes so as to show at the next boot. Make them in the underlying disk file system, as has been suggested to you several times. I can run MAKEDEV dk15 dk16 dk17 ... but these don't survive for the next boot. Probably because you are making those device nodes in a tmpfs file system. With the devpubd line in /etc/rc.conf, NO changed to YES, I saw the desired /dev/dk* nodes, but haven't yet had the situation where root partition needed to be fsck'ed since then. Maybe the /dev/dk* nodes would be made before fsck_root? /etc/rc.d/devpubd runs long after /etc/rc.d/fsck_root, as you can see by analysing the PROVIDE/REQUIRE/BEFORE lines in the scripts, or looking at the output from rcorder /etc/rc.d/*, or looking at the log file in /var/run/rc.log. --apb (Alan Barrett)
Re: Netbsd/amd64 build failure with both MKGCC and MKLLVM set
On Fri, Oct 18, 2013 at 11:46:29PM +0200, Joerg Sonnenberger wrote: On Fri, Oct 18, 2013 at 08:35:17PM +0200, Nicolas Joly wrote: On Fri, Oct 18, 2013 at 07:26:45PM +0200, Joerg Sonnenberger wrote: On Fri, Oct 18, 2013 at 07:20:31PM +0200, Nicolas Joly wrote: On my my main developpement machine running -current NetBSD/amd64, i do see a problem if i try to build both gcc and clang (MKGCC=yes and MKLLVM=yes). The default compiler remaining unchanged (gcc). MKLLVM=yes now needs MKLIBCXX=yes too. Hmmm ... Adding it to my mk.conf does not seems to make any difference. A clean build fails the very same way. Oh, MKLLVM=yes without HAVE_LLVM. I haven't adjusted the parts in libc for that. That's it, thanks. I had MKLLVM=yes for a while, to test it on some other projets i'm working on. I'll turn if off for now. -- Nicolas Joly Biology IT Center Institut Pasteur, Paris.
Re: amd64 build break - liblfs/ulfs_quota2.c
On Fri, Oct 18, 2013 at 10:14:51PM +0300, Andreas Gustafsson wrote: Paul Goyette wrote: /test-bed/src/sys/rump/fs/lib/liblfs/../../../../ufs/lfs/ulfs_quota2.c: In function 'getq2h': /test-bed/src/sys/rump/fs/lib/liblfs/../../../../ufs/lfs/ulfs_quota2.c:162:6: error: 'needswap' undeclared (first use in this function) The build is still failing, even with src/sys/ufs/lfs/ulfs_quota2.c 1.14: /tmp/bracket/build/2013.10.18.17.48.44-i386/src/sys/rump/fs/lib/liblfs/../../../../ufs/lfs/ulfs_quota2.c: In function 'dq2clear_callback': /tmp/bracket/build/2013.10.18.17.48.44-i386/src/sys/rump/fs/lib/liblfs/../../../../ufs/lfs/ulfs_quota2.c:662:6: error: 'needswap' undeclared (first use in this function) Christos did it wrong :-) mrg has a patch, not sure when he's intending to commit it. -- David A. Holland dholl...@netbsd.org