On Wed, 2021-06-09 11:35:19 +0200, Martin Husemann <martin%duskware.de@localhost> wrote: > On Wed, Jun 09, 2021 at 10:47:37AM +0200, Jan-Benedict Glaw wrote: > > Building tools worked, but building the release broke as it broke > > when building within the Linux (Debian) docker containers due to not > > finding gcov-iov.h. So I guess something broke in the toolchain? > > > > Logs can be found at: > > > > https://lug-owl.de/~jbglaw/laminar/tools.log > > https://lug-owl.de/~jbglaw/laminar/release.log > > Well, it works for the official builds (which are done from cvs source > trees), so it must be something git specific. That's an interesting idea; I went out and checked it. The CVS and GIT trees are quite in sync (except for sync difference, the GIT tree lags behind a few hours), with the exception of this: $ diff -Nurp netbsd-from-cvs netbsd-from-git diff -Nurp netbsd-from-cvs/sys/arch/x68k/include/cdefs.h netbsd-from-git/sys/arch/x68k/include/cdefs.h --- netbsd-from-cvs/sys/arch/x68k/include/cdefs.h 1996-05-05 14:17:03.000000000 +0200 +++ netbsd-from-git/sys/arch/x68k/include/cdefs.h 2021-06-10 09:43:48.000000000 +0200 @@ -1,4 +1,4 @@ -/* $NetBSD: cdefs.h,v 1.1.1.1 1996/05/05 12:17:03 oki Exp $ */ +/* $NetBSD: cdefs.h,v 1.1 1996/05/05 12:17:15 oki Exp $ */ #ifndef _MACHINE_CDEFS_H_ #define _MACHINE_CDEFS_H_ diff -Nurp netbsd-from-cvs/sys/arch/x68k/include/cpufunc.h netbsd-from-git/sys/arch/x68k/include/cpufunc.h --- netbsd-from-cvs/sys/arch/x68k/include/cpufunc.h 1996-05-05 14:17:03.000000000 +0200 +++ netbsd-from-git/sys/arch/x68k/include/cpufunc.h 2021-06-10 09:43:48.000000000 +0200 @@ -1,4 +1,4 @@ -/* $NetBSD: cpufunc.h,v 1.1.1.1 1996/05/05 12:17:03 oki Exp $ */ +/* $NetBSD: cpufunc.h,v 1.1 1996/05/05 12:17:15 oki Exp $ */ /* * Functions to provide access to special cpu instructions. diff -Nurp netbsd-from-cvs/sys/arch/x68k/include/ieeefp.h netbsd-from-git/sys/arch/x68k/include/ieeefp.h --- netbsd-from-cvs/sys/arch/x68k/include/ieeefp.h 1996-05-05 14:17:03.000000000 +0200 +++ netbsd-from-git/sys/arch/x68k/include/ieeefp.h 2021-06-10 09:43:48.000000000 +0200 @@ -1,4 +1,4 @@ -/* $NetBSD: ieeefp.h,v 1.1.1.1 1996/05/05 12:17:03 oki Exp $ */ +/* $NetBSD: ieeefp.h,v 1.1 1996/05/05 12:17:14 oki Exp $ */ /* Just use the common m68k definition */ #include <m68k/ieeefp.h> diff -Nurp netbsd-from-cvs/sys/arch/x68k/include/profile.h netbsd-from-git/sys/arch/x68k/include/profile.h --- netbsd-from-cvs/sys/arch/x68k/include/profile.h 1996-05-05 14:17:03.000000000 +0200 +++ netbsd-from-git/sys/arch/x68k/include/profile.h 2021-06-10 09:43:48.000000000 +0200 @@ -1,3 +1,3 @@ -/* $NetBSD: profile.h,v 1.1.1.1 1996/05/05 12:17:03 oki Exp $ */ +/* $NetBSD: profile.h,v 1.1 1996/05/05 12:17:14 oki Exp $ */ #include <m68k/profile.h> diff -Nurp netbsd-from-cvs/sys/arch/x68k/include/setjmp.h netbsd-from-git/sys/arch/x68k/include/setjmp.h --- netbsd-from-cvs/sys/arch/x68k/include/setjmp.h 1996-05-05 14:17:03.000000000 +0200 +++ netbsd-from-git/sys/arch/x68k/include/setjmp.h 2021-06-10 09:43:48.000000000 +0200 @@ -1,3 +1,3 @@ -/* $NetBSD: setjmp.h,v 1.1.1.1 1996/05/05 12:17:03 oki Exp $ */ +/* $NetBSD: setjmp.h,v 1.1 1996/05/05 12:17:15 oki Exp $ */ #include <m68k/setjmp.h> > Maybe a timestamp getting set differently, unveiling an unnoticed makefile > bug? Timestamp handling (file-based vs. commit based) may indeed be an issue, but as it turned out, after tar-piping over a NetBSD CVS tree (this VM has no direct Internet connectivity) to the NetBSD VM (the same 9.2 freshly installed as for the previous GIT-based build), it failed as well. My build logs (from `script`) are here: GIT: tools: https://lug-owl.de/~jbglaw/laminar/tools.log release: https://lug-owl.de/~jbglaw/laminar/release.log CVS: tools: https://lug-owl.de/~jbglaw/laminar/tools-cvs.log release: https://lug-owl.de/~jbglaw/laminar/release-cvs.log MfG, JBG --
Attachment:
signature.asc
Description: PGP signature