Re: misc/58762 (arm binary/sets/base32.tgz is bogus)

["Martin Husemann via gnats" <gnats-admin%NetBSD.org@localhost>]

The following reply was made to PR misc/58762; it has been noted by GNATS.

From: Martin Husemann <martin%duskware.de@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: misc/58762 (arm binary/sets/base32.tgz is bogus)
Date: Wed, 7 May 2025 20:27:09 +0200

 I think this is due to a discrepancy between the compat building code
 (which e.g. on arm is only enabled when compiling with clang) and the
 set name code that only trigers on MKCOMPAT != no.
 
 But:
 
 > cd src/compat && /work/tools/bin/nbmake-evbarm -v ARCHDIR_SUBDIR -v MKCOMPAT
 
 yes
 
 
 so we have MKCOMPAT enabled but no subdirs for the lib builds (so we don't
 build anything).
 
 Untested patch below. Note that this is still inconsistent (ACTIVE_CC vs
 HAVE_GCC), but I didn't want to touch the existing aarch64 conditional.
 
 A better option would be to make src/distrib/sets/sets.subr invoke
 src/compat/archdirs.mk directly and extract the compat names from
 there, then build different sets for different compats.
 
 Martin
 
 Index: bsd.own.mk
 ===================================================================
 RCS file: /cvsroot/src/share/mk/bsd.own.mk,v
 retrieving revision 1.1416
 diff -u -r1.1416 bsd.own.mk
 --- bsd.own.mk	27 Apr 2025 03:48:34 -0000	1.1416
 +++ bsd.own.mk	7 May 2025 18:23:19 -0000
 @@ -1115,7 +1115,7 @@
      ${MACHINE_ARCH} == "powerpc64" || \
      (${MACHINE_ARCH} == "aarch64" && ${HAVE_GCC:U0} == 0) || \
      ${MACHINE_ARCH} == "riscv64" || \
 -    ${MACHINE_ARCH:Mearm*}
 +    (${MACHINE_ARCH:Mearm*} && ${HAVE_GCC:U0} == 0)
  MKCOMPAT?=	yes
  .else
  # Don't let this build where it really isn't supported.