a cross-sh4 toolchain, created with "crossdev -t sh4-unknown-linux-gnu" no longer links to libc when cpu subtype and endianness parameters are supplied: sh4-unknown-linux-gnu-gcc -o test -O2 -m4 -ml test.c /usr/libexec/gcc/sh4-unknown-linux-gnu/ld: cannot find crt1.o: No such file or directory /usr/libexec/gcc/sh4-unknown-linux-gnu/ld: cannot find crti.o: No such file or directory Omitting the "-m4 -ml" parameters (optimize for little-endian plain SH4 with FPU) results in a successful link. I've was able to build a complete rootfs with the above CFLAGS when I last played with SH4 (around 2010), so I guess something broke in the meantime. Reproducible: Always
Please post your `(cross-)emerge --info' output in a comment.
binutils-9999 from 20140522, gcc-4.9.0 glibc-2.19 Portage 2.2.10 (embedded, gcc-4.9.0, unavailable, 3.15.0-rc6-00001-gd778825-dirty x86_64) ================================================================= System uname: Linux-3.15.0-rc6-00001-gd778825-dirty-x86_64-AMD_Phenom-tm-_II_X6_1090T_Processor-with-gentoo-2.2 KiB Mem: 8159188 total, 3707012 free KiB Swap: 2097148 total, 2097148 free Timestamp of tree: Mon, 26 May 2014 08:15:01 +0000 ld GNU ld (GNU Binutils) 2.24.51.20140522 distcc 3.1 x86_64-pc-linux-gnu [disabled] ccache version 3.1.9 [disabled] Repositories: gentoo x11 mano ACCEPT_KEYWORDS="sh ~sh" ACCEPT_LICENSE="* -@EULA" CBUILD="x86_64-pc-linux-gnu" CFLAGS="-Os -pipe -fomit-frame-pointer" CHOST="sh4-unknown-linux-gnu" CONFIG_PROTECT="/etc /usr/share/config /usr/share/gnupg/qualified.txt" CONFIG_PROTECT_MASK="/etc/ca-certificates.conf /etc/env.d /etc/fonts/fonts.conf /etc/gconf /etc/gentoo-release /etc/php/apache2-php5.5/ext-active/ /etc/php/cgi-php5.5/ext-active/ /etc/php/cli-php5.5/ext-active/ /etc/revdep-rebuild /etc/sandbox.d /etc/terminfo /etc/texmf/language.dat.d /etc/texmf/language.def.d /etc/texmf/updmap.d /etc/texmf/web2c" CXXFLAGS="-Os -pipe -fomit-frame-pointer" DISTDIR="/usr/portage/distfiles" EMERGE_DEFAULT_OPTS="--quiet-build=n" FCFLAGS="-Os -pipe" FEATURES="assume-digests binpkg-logs buildpkg config-protect-if-modified distlocks ebuild-locks fixlafiles merge-sync news nodoc noinfo noman parallel-fetch preserve-libs protect-owned sandbox sfperms strict unknown-features-warn unmerge-logs unmerge-orphans userfetch userpriv usersandbox usersync" FFLAGS="-Os -pipe" GENTOO_MIRRORS="http://gentoo.inode.at" LANG="C" LDFLAGS="" MAKEOPTS="-j6" PKGDIR="/usr/sh4-unknown-linux-gnu/packages/" PORTAGE_CONFIGROOT="/usr/sh4-unknown-linux-gnu/" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --omit-dir-times --compress --force --whole-file --delete --stats --human-readable --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages" PORTAGE_TMPDIR="/usr/sh4-unknown-linux-gnu/tmp/" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/var/lib/layman/x11 /etc/portage/overlay" USE="bindist kdrive make-symlinks minimal multicall sh zlib" INPUT_DEVICES="evdev mouse keyboard tslib" KERNEL="linux" LINGUAS="en" USERLAND="GNU" VIDEO_CARDS="fbdev" Unset: CPPFLAGS, CTARGET, INSTALL_MASK, LC_ALL, PORTAGE_BUNZIP2_COMMAND, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS, SYNC, USE_PYTHON
using -m4 with a gcc configured for sh4-unknown-linux-gnu is redundant, but it should still work the sysroot_suffix_spec is getting in the way here ...
Still happens. Diff output of $ sh4-unknown-linux-gnu-gcc -o test -O2 a.c -v -Wl,--verbose # ok $ sh4-unknown-linux-gnu-gcc -o test -O2 -m4 a.c -v -Wl,--verbose # bad ... -LIBRARY_PATH=/usr/lib/gcc/sh4-unknown-linux-gnu/7.2.0/:/usr/lib/gcc/sh4-unknown-linux-gnu/7.2.0/../../../../sh4-unknown-linux-gnu/lib/:/usr/sh4-unknown-linux-gnu/lib/:/usr/sh4-unknown-linux-gnu/usr/lib/ +LIBRARY_PATH=/usr/lib/gcc/sh4-unknown-linux-gnu/7.2.0/:/usr/lib/gcc/sh4-unknown-linux-gnu/7.2.0/../../../../sh4-unknown-linux-gnu/lib/ ... -attempt to open /usr/sh4-unknown-linux-gnu/usr/lib/Scrt1.o succeeded -/usr/sh4-unknown-linux-gnu/usr/lib/Scrt1.o -attempt to open /usr/sh4-unknown-linux-gnu/usr/lib/crti.o succeeded -/usr/sh4-unknown-linux-gnu/usr/lib/crti.o +attempt to open Scrt1.o failed +attempt to open crti.o failed $SYSROOT/usr/lib/ got lost from LIBRARY_PATH here.
sh is one of targets that supports multilib: https://github.com/gcc-mirror/gcc/blob/master/gcc/config.gcc#L2829 (sh4-* defaults to single ABI, -m4-nofpu?).
Now I understand Mike's comment about SYSROOT_SUFFIX_SPEC better. sysroot suffix is part of spec generated based on sh4 multilib layout here: gcc/gcc $ rm sysroot-suffix.h gcc/gcc $ make sysroot-suffix.h Makefile:2632: warning: overriding recipe for target 'gt-sh.h' ../../gcc/gcc/config/sh/t-sh:93: warning: ignoring old recipe for target 'gt-sh.h' /bin/sh ../../gcc/gcc/config/print-sysroot-suffix.sh \ "=! m1=!m1 /m1=!/m1 m2a=!m2a /m2a=!/m2a m2a-nofpu=!m2a-nofpu /m2a-nofpu=!/m2a-nofpu m2a-single-only=!m2a-single-only /m2a-single-only=!/m2a-single-only m2a-single=!m2a-single /m2a-single=!/m2a-single m2e=!m2e /m2e=!/m2e m2=!m2 /m2=!/m2 m3e=!m3e /m3e=!/m3e m3=!m3 /m3=!/m3 m4-nofpu=!m4-nofpu /m4-nofpu=!/m4-nofpu m4-single-only=!m4-single-only /m4-single-only=!/m4-single-only m4-single=!m4-single /m4-single=!/m4-single m4=!m4 /m4=!/m4 m4a-nofpu=!m4a-nofpu /m4a-nofpu=!/m4a-nofpu m4a-single-only=!m4a-single-only /m4a-single-only=!/m4a-single-only m4a-single=!m4a-single /m4a-single=!/m4a-single m4a=!m4a /m4a=!/m4a m4al=!m4al /m4al=!/m4al" " m4" \ "" "" > tmp-sysroot-suffix.h mv tmp-sysroot-suffix.h sysroot-suffix.h gcc/gcc $ cat sysroot-suffix.h #undef SYSROOT_SUFFIX_SPEC #define SYSROOT_SUFFIX_SPEC "" \ "%{m4:/m4;" \ ":}" AFAIU this translates to: - if -m4 option is passed explicitly then append /m4 to $SYSROOT - otherwise prepend nothing Thus it's print-sysroot-suffix.sh arguments that are at fault (whitespace in '" m4"'?) ot lack of OTHER_ENDIAN set in gcc/config/sh/t-sh:OTHER_ENDIAN = $(word 2,$(TM_ENDIAN_CONFIG)) Let's see which works first.
Here is the minimal call of how match happens: bash ../../gcc/gcc/config/print-sysroot-suffix.sh 'm4=!m4' 'm4' '' '' #undef SYSROOT_SUFFIX_SPEC #define SYSROOT_SUFFIX_SPEC "" \ "%{m4:/m4;" \ ":}" What has the knowledge that default is 'm4' and it should be filtered out?
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=49d0cca0f886e06d54896ae70b5d24a364ef17e3 commit 49d0cca0f886e06d54896ae70b5d24a364ef17e3 Author: Sergei Trofimovich <slyfox@gentoo.org> AuthorDate: 2018-04-14 12:36:15 +0000 Commit: Sergei Trofimovich <slyfox@gentoo.org> CommitDate: 2018-04-14 12:36:15 +0000 sys-devel/gcc: bump patchset of 7.3.0 up to 1.2, bug #511548 Single new patch: + 92_all_sh-drop-sysroot-suffix.patch : drop sysroot-suffix Reported-by: Manuel Lauss Bug: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=42947 Closes: https://bugs.gentoo.org/511548 Package-Manager: Portage-2.3.28, Repoman-2.3.9 sys-devel/gcc/Manifest | 1 + sys-devel/gcc/gcc-7.3.0-r2.ebuild | 25 +++++++++++++++++++++++++ 2 files changed, 26 insertions(+)
