[arch-dev-public] [signoff] klibc stuff + mkinitcpio
I put new klibc, klibc-kbd, klibc-m-i-t, klibc-udev, klibc-extras and v86d to testing. Those are no version bumps, but just a rebuild of klibc against 2.6.31 with a small bugfix (see more below) - and the corresponding rebuilds for the new ABI, as with each change of build headers for klibc. mkinitcpio also needed a rebuild due to the klibc-kbd versioned dependency that was wrong - I didn't change the version here either. Signoff procedure: 1) cp /boot/kernel26.img /boot/kernel26.img.working 2) pacman -Syu 3) mkinitcpio -p kernel26 4) reboot i686 packages are completely untested, x86_64 works and boots. About the fixed bug: Recently, ps ax would display: 1 ? Ss 0:01 ini while it should (and does now) display 1 ? Ss 0:01 init [5] This confused and annoyed people. As for uclibc-initramfs: I didn't have time anymore, but last time I tried, I was unable to get the toolchain to work as I would expect. Thus, I keep maintaining klibc crap.
Thomas Bächler schrieb:
I put new klibc, klibc-kbd, klibc-m-i-t, klibc-udev, klibc-extras and v86d to testing. Those are no version bumps, but just a rebuild of klibc against 2.6.31 with a small bugfix (see more below) - and the corresponding rebuilds for the new ABI, as with each change of build headers for klibc.
mkinitcpio also needed a rebuild due to the klibc-kbd versioned dependency that was wrong - I didn't change the version here either.
Signoff procedure: 1) cp /boot/kernel26.img /boot/kernel26.img.working 2) pacman -Syu 3) mkinitcpio -p kernel26 4) reboot
i686 packages are completely untested, x86_64 works and boots.
Okay, Eric says it doesn't work on i686 - I don't know why, but please don't forget step 1) if you are testing it.
Thomas Bächler schrieb:
Signoff procedure: 1) cp /boot/kernel26.img /boot/kernel26.img.working 2) pacman -Syu 3) mkinitcpio -p kernel26 4) reboot
i686 packages are completely untested, x86_64 works and boots.
Okay, Eric says it doesn't work on i686 - I don't know why, but please don't forget step 1) if you are testing it.
I removed the i686 packages from testing for now, they're still in /home/thomas/klibc-testing-i686.
Thomas Bächler wrote:
Thomas Bächler schrieb:
Signoff procedure: 1) cp /boot/kernel26.img /boot/kernel26.img.working 2) pacman -Syu 3) mkinitcpio -p kernel26 4) reboot
i686 packages are completely untested, x86_64 works and boots.
Okay, Eric says it doesn't work on i686 - I don't know why, but please don't forget step 1) if you are testing it.
I removed the i686 packages from testing for now, they're still in /home/thomas/klibc-testing-i686.
I can confirm: ::Loading udev... clocksource tsc unstable (delta = 1720416923ms) Maybe updating klibc-udev will help. Allan
Allan McRae wrote:
Thomas Bächler wrote:
Thomas Bächler schrieb:
Signoff procedure: 1) cp /boot/kernel26.img /boot/kernel26.img.working 2) pacman -Syu 3) mkinitcpio -p kernel26 4) reboot
i686 packages are completely untested, x86_64 works and boots.
Okay, Eric says it doesn't work on i686 - I don't know why, but please don't forget step 1) if you are testing it.
I removed the i686 packages from testing for now, they're still in /home/thomas/klibc-testing-i686.
I can confirm:
::Loading udev... clocksource tsc unstable (delta = 1720416923ms)
Maybe updating klibc-udev will help.
Which of course is never as simple as it sounds... it is not a simple bump pkgver and md5sum rebuild and I do not have time to figure it out now. Allan
Allan McRae schrieb:
I can confirm:
::Loading udev... clocksource tsc unstable (delta = 1720416923ms)
Newer udev doesn't work with klibc, it requires blkid, which we will probably never be able to build with klibc. I want to get rid of klibc - it sucks, it's a nightmare, it's unmaintained, it's crap, should I say more? The reason might be that I did some magic to the i686 version to make it build, which might have been wrong - I have no idea. But then, I don't know why it would hang here, as according to Eric everything seems to work (you can run udevadm, you can run modprobe and so on). I might set up a test system with i686 and see for myself.
Thomas Bächler wrote:
Allan McRae schrieb:
I can confirm:
::Loading udev... clocksource tsc unstable (delta = 1720416923ms)
Newer udev doesn't work with klibc, it requires blkid, which we will probably never be able to build with klibc.
I want to get rid of klibc - it sucks, it's a nightmare, it's unmaintained, it's crap, should I say more?
The reason might be that I did some magic to the i686 version to make it build, which might have been wrong - I have no idea.
But then, I don't know why it would hang here, as according to Eric everything seems to work (you can run udevadm, you can run modprobe and so on). I might set up a test system with i686 and see for myself.
Just adding that I tried heaps of clocksource= options (acpi_pm,htc,jiffies,rtc) and none of them helped this issue. Allan
Allan McRae schrieb:
Just adding that I tried heaps of clocksource= options (acpi_pm,htc,jiffies,rtc) and none of them helped this issue.
Allan
That clocksource message is probably unrelated - I get similar messages after successful boot. Eric didn't get those, but got a message about his keyboard being detected. It is normal that kernel messages appear here after control has been passed to userspace - the problem is that userspace hangs. If you have the time and energy, can you add "echo" calls to /lib/initcpio/hooks/udev to find out if the udevd --daemon, udevadm trigger or udevadm settle is the place where is hangs. I would guess the latter, but I don't know for sure.
On Thu, Oct 22, 2009 at 3:40 AM, Thomas Bächler <thomas@archlinux.org> wrote:
Allan McRae schrieb:
Just adding that I tried heaps of clocksource= options (acpi_pm,htc,jiffies,rtc) and none of them helped this issue.
Allan
That clocksource message is probably unrelated - I get similar messages after successful boot. Eric didn't get those, but got a message about his keyboard being detected. It is normal that kernel messages appear here after control has been passed to userspace - the problem is that userspace hangs.
FTR, my message was: Loading udev... input: AT Translated Set 2 keyboard as ... It stucked there for a few minutes then proceeded with "Alarm clock Done" etc, etc. Only dm_mod was loaded so it failed mounting the root partition and dropped me in a recovery shell. I haven't tried udevadm or modprobe but udevd was running and lsmod was working. I sent my broken image to Thomas and everything was there. Very strange.
If you have the time and energy, can you add "echo" calls to /lib/initcpio/hooks/udev to find out if the udevd --daemon, udevadm trigger or udevadm settle is the place where is hangs. I would guess the latter, but I don't know for sure.
I put new packages to /home/thomas/klibc-testing-i686/new, can anyone try if those change anything?
Thomas Bächler wrote:
I put new packages to /home/thomas/klibc-testing-i686/new, can anyone try if those change anything?
Same issue here. Allan
On Thu, Oct 22, 2009 at 3:35 AM, Thomas Bächler <thomas@archlinux.org> wrote:
I put new packages to /home/thomas/klibc-testing-i686/new, can anyone try if those change anything?
I suspect this stems from the signal stuff that you patched in klibc. As a trial, can you try building without that patch and report the error here? I'm curious if we can fix this without that patch...
Aaron Griffin schrieb:
On Thu, Oct 22, 2009 at 3:35 AM, Thomas Bächler <thomas@archlinux.org> wrote:
I put new packages to /home/thomas/klibc-testing-i686/new, can anyone try if those change anything?
I suspect this stems from the signal stuff that you patched in klibc. As a trial, can you try building without that patch and report the error here? I'm curious if we can fix this without that patch...
KLIBCCC usr/klibc/vsnprintf.o In file included from usr/include/arch/i386/klibc/archsignal.h:99, from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: linux/include/asm-generic/signal.h:7:1: warning: "_NSIG_BPW" redefined In file included from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: usr/include/arch/i386/klibc/archsignal.h:17:1: warning: this is the location of the previous definition In file included from usr/include/arch/i386/klibc/archsignal.h:99, from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: linux/include/asm-generic/signal.h:67:1: warning: "SA_NOCLDSTOP" redefined In file included from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: usr/include/arch/i386/klibc/archsignal.h:76:1: warning: this is the location of the previous definition In file included from usr/include/arch/i386/klibc/archsignal.h:99, from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: linux/include/asm-generic/signal.h:68:1: warning: "SA_NOCLDWAIT" redefined In file included from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: usr/include/arch/i386/klibc/archsignal.h:77:1: warning: this is the location of the previous definition In file included from usr/include/arch/i386/klibc/archsignal.h:99, from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: linux/include/asm-generic/signal.h:69:1: warning: "SA_SIGINFO" redefined In file included from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: usr/include/arch/i386/klibc/archsignal.h:78:1: warning: this is the location of the previous definition In file included from usr/include/arch/i386/klibc/archsignal.h:99, from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: linux/include/asm-generic/signal.h:70:1: warning: "SA_ONSTACK" redefined In file included from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: usr/include/arch/i386/klibc/archsignal.h:79:1: warning: this is the location of the previous definition In file included from usr/include/arch/i386/klibc/archsignal.h:99, from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: linux/include/asm-generic/signal.h:71:1: warning: "SA_RESTART" redefined In file included from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: usr/include/arch/i386/klibc/archsignal.h:80:1: warning: this is the location of the previous definition In file included from usr/include/arch/i386/klibc/archsignal.h:99, from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: linux/include/asm-generic/signal.h:72:1: warning: "SA_NODEFER" redefined In file included from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: usr/include/arch/i386/klibc/archsignal.h:81:1: warning: this is the location of the previous definition In file included from usr/include/arch/i386/klibc/archsignal.h:99, from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: linux/include/asm-generic/signal.h:73:1: warning: "SA_RESETHAND" redefined In file included from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: usr/include/arch/i386/klibc/archsignal.h:82:1: warning: this is the location of the previous definition In file included from usr/include/arch/i386/klibc/archsignal.h:99, from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: linux/include/asm-generic/signal.h:95: error: conflicting types for ‘sigset_t’ usr/include/arch/i386/klibc/archsignal.h:22: note: previous declaration of ‘sigset_t’ was here In file included from usr/klibc/../include/signal.h:14, from usr/klibc/../include/sys/select.h:11, from usr/klibc/../include/unistd.h:12, from usr/klibc/../include/stdio.h:11, from usr/klibc/vsnprintf.c:13: usr/include/arch/i386/klibc/archsignal.h:104: error: redefinition of ‘struct sigaction’ make[2]: *** [usr/klibc/vsnprintf.o] Error 1 make[2]: *** Waiting for unfinished jobs.... make[1]: *** [all] Error 2 make: *** [klibc] Error 2
participants (4)
-
Aaron Griffin
-
Allan McRae
-
Eric Bélanger
-
Thomas Bächler