Hi,

some time in between:
Dec 17 2025 commit 34eed88a70fc8238cd5b623cc96b8bf0e40392ed (OK)
and latest
Oct 16 2026 commit e23dabed154a8495f3ac1ff3cc9d04fa1c207893 (FAIL)

A somewhat severe regression seems to have been introduced,
confirmed with i386-tcc when compiling https://codeberg.org/aggi/linux-tcc
with contained Kbuild.sh script:

1) while compiling with Kbuild.sh/CC|AS=i386-tcc
linux-tcc.git/usr/src/linux/arch/i386/kernel/entry.S:141: error: register 
expected
-> In this case the macro expansion for SAVE_ALL seems to fail, which did work
until Dec 2025 with a functional linux kernel emitted.

2) while compiling with Kbuild.sh/CC=i386-tcc
linux-tcc.git/usr/src/linux/arch/i386/kernel/acpi.c
Segmentation fault (core dumped)
-> In this case compiling mentioned c source crashes tcc

I cannot rule out any side-effects of my local development setup,
but i386-tcc did work flawlessly until Dec 2025 emitting kernel (AS and CC)
and too processing a complete userspace of ~500builds here was fine:
http://tinyfront.mooo.com/downloads/x86-embedded/666666/tinyfront-packages.list

Strangely when compiling such complete userspace with most recent tcc version
no irregularities were spotted and those userspace pieces passed initial
runtime testing today, but all of the sudden Kbuild.sh errors when trying
to re-build kernel. I can only reliably isolate the problem to the recent
git-pull of latest tcc sources.

Bisecting accross a hundred commits might take a while, so i rather ask for
confirmation from tinycc-devel if any issue is known already.
Please let me know if there is an easy fix for this before i begin bisecting
for hours and days.

Question: had there been any issues to run Kbuild.sh yourself for some 
quick testing when pushing commits to mob? I mean, kernel build finishes
quickly and it's too easily booted inside some VM to see if it's working.
A most valuable test-case for i386-tcc at least more than once already.

Side notes:
I too would be curious when this problem was introduced, because i did not
do regular nightly builds and git-pull once a year only since i do
maintain another 500ebuilds for tcc already, and too began re-factoring
https://codeberg.org/aggi/tiny-bootstrap for a complete system-bootstrap
of tinycc mob/HEAD (with plans to add the missing kernel piece until december
this year for a bootstrapped linux-tcc/busybox/vi/i386-tcc live.iso released)
For this i too managed to cut short the dependency graph of bootloader
compilation of some older syslinux/isolinux.asm with i386-tcc/yasm toolchain
already (see tiny-bootstrap/steps-tiny/syslinux-3.86 if you wish)
Furthermore i intended to re-base some pnut-cc related patchset onto latest
mob/HEAD for the bootstrapping stages of tcc, and i would prefer to do
it with a known good working state of tcc itself, hence my worries expressed a
while ago on tinycc-devel tcc might become destablized,
which it seems it slightly did blocking progress on mob branch currently.

Thank You.

-- 
Michael Ackermann

Attachment: signature.asc
Description: Digital signature

_______________________________________________
Tinycc-devel mailing list
[email protected]
https://lists.nongnu.org/mailman/listinfo/tinycc-devel

Reply via email to