Created attachment 828095 [details] compressed build log from armv7-cross from the build log: x86_64-pc-linux-gnu-gcc -c -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -march=znver1 -pipe -std=c99 -Wextra -Wno-unused-result -Wno-unused-parameter -Wno-missing-field-initializers -Wstrict-prototypes -Werror=implicit-function-declaration -fvisibility=hidden -I/var/tmp/portage/dev-lang/python-3.10.8_p2/work/Python-3.10.8/Include/internal -IObjects -IInclude -IPython -I. -I/var/tmp/portage/dev-lang/python-3.10.8_p2/work/Python-3.10.8/Include -I/usr/armv7a-unknown-linux-gnueabihf/usr/include/ncursesw -fPIC -DPy_BUILD_CORE -o Objects/bytes_methods.o /var/tmp/portage/dev-lang/python-3.10.8_p2/work/Python-3.10.8/Objects/bytes_methods.c In file included from /var/tmp/portage/dev-lang/python-3.10.8_p2/work/Python-3.10.8/Include/Python.h:50, from /var/tmp/portage/dev-lang/python-3.10.8_p2/work/Python-3.10.8/Parser/string_parser.c:3: /var/tmp/portage/dev-lang/python-3.10.8_p2/work/Python-3.10.8/Include/pyport.h:746:2: error: #error "LONG_BIT definition appears wrong for platform (bad gcc/glibc config?)." 746 | #error "LONG_BIT definition appears wrong for platform (bad gcc/glibc config?)." | ^~~~~ In file included from /var/tmp/portage/dev-lang/python-3.10.8_p2/work/Python-3.10.8/Include/Python.h:50, from /var/tmp/portage/dev-lang/python-3.10.8_p2/work/Python-3.10.8/Parser/peg_api.c:1: /var/tmp/portage/dev-lang/python-3.10.8_p2/work/Python-3.10.8/Include/pyport.h:746:2: error: #error "LONG_BIT definition appears wrong for platform (bad gcc/glibc config?)." 746 | #error "LONG_BIT definition appears wrong for platform (bad gcc/glibc config?)." I don't know what happens here, but emerge -pv dev-lang/python tells me that =dev-lang/python-3.10.6_p2 compiled fine: emerge-armv7a-unknown-linux-gnueabihf -pv dev-lang/python [ebuild U ] dev-lang/python-3.10.8_p2:3.10::gentoo [3.10.6_p2:3.10::gentoo] to /usr/armv7a-unknown-linux-gnueabihf/ USE="ensurepip gdbm ncurses readline sqlite ssl xml -bluetooth -build -examples -hardened -libedit -lto -pgo -test -tk -verify-sig" 0 KiB Total: 1 package (1 upgrade), Size of downloads: 0 KiB Ideas, suggestions?
Created attachment 828097 [details] output from emerge --info (armv7)
working for 3.10 branch are: dev-lang/python-3.10.6_p2 dev-lang/python-3.10.6_p3 dev-lang/python-3.10.6_p4 dev-lang/python-3.10.7 dev-lang/python-3.10.7 with this commit breaks it: https://github.com/gentoo/gentoo/commit/29d2314810bd8e1de73df5c3ebe30c826281c253
I bet this is because I tested with sparc64 and not sparc (bitness).
*** Bug 880137 has been marked as a duplicate of this bug. ***
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=e5cb1ccc95873d55f9f643b043f3f7493fd31eee commit e5cb1ccc95873d55f9f643b043f3f7493fd31eee Author: Sam James <sam@gentoo.org> AuthorDate: 2022-11-07 07:57:49 +0000 Commit: Sam James <sam@gentoo.org> CommitDate: 2022-11-07 07:57:49 +0000 dev-lang/python: append ncurses include path after CBUILD Python is built We don't want to contaminate the mini CBUILD Python cross build with paths in SYSROOT. Bug: https://bugs.gentoo.org/880059 Bug: https://bugs.gentoo.org/880137 Signed-off-by: Sam James <sam@gentoo.org> dev-lang/python/python-3.10.8_p2.ebuild | 10 +++++----- dev-lang/python/python-3.11.0_p1.ebuild | 10 +++++----- dev-lang/python/python-3.12.0_alpha1_p1.ebuild | 10 +++++----- dev-lang/python/python-3.8.15_p2.ebuild | 10 +++++----- dev-lang/python/python-3.9.15_p2.ebuild | 10 +++++----- 5 files changed, 25 insertions(+), 25 deletions(-)
I don't think the ncurses bit is sufficient, just a start. Can you share the config.logs?
Created attachment 828221 [details] compressed build log from armv7-cross new build log with ncurses change
Created attachment 828223 [details] config.log from cross compile attempt this is the only config.log I was able to find inside of work, it does x86_64 things only I'm afraid
Would you mind giving me some crossdev commands from scratch to get to this? I've tried and I really can't reproduce it yet even with the same tuple :(
Created attachment 831685 [details] output from emerge --info (amd64 host) The host is amd64 on a multilib profile, maybe that contributes?
(In reply to tt_1 from comment #10) > Created attachment 831685 [details] > output from emerge --info (amd64 host) > > The host is amd64 on a multilib profile, maybe that contributes? Mine is too. :( Can you try give some instructions from scratch?
its nothing special really: crossdev -S armv7a-unknown-linux-gnueabihf --libc 2.35-r11 when done, adjust make.profile symlinks to use profiles/default/arm/17.0/armv7a/ edit make.conf, adjust ACCEPT_KEYWORDS to arm only, dropping ~arm edit make.conf, add -su to global flags, and -rsync-verify ( the latter is too much hassle for no benefit, but you won't get there anyway ) add =sys-libs/glibc-2.35-r11 ~arm to package.accept_keywords of /usr/armv7a-unknown-linux-gnueabihf and finally populate your userland with: emerge-armv7a-unknown-linux-gnueabihf -auvND world - the process will stop with dev-lang/python and the error above.
Created attachment 831947 [details] compressed build log with gcc-11.3.0 initial bug was with gcc-10.4.0 toolchain, I recompiled it with more recent gcc, but same result
Thanks, reproduced. At first, I thought it was out-of-source build being a bit broken, but it happens when I copy S too.
We suspect https://archives.gentoo.org/gentoo-dev/message/aebd8c41144bcb933a2d91fd15f0598d will fix this.
(In reply to Sam James from comment #15) > We suspect > https://archives.gentoo.org/gentoo-dev/message/ > aebd8c41144bcb933a2d91fd15f0598d will fix this. this looks good to me in terms of compiling, it works now on armv7a
I've merged the fix.
