Not sure what to put in this bug, but it seems that if I halt my system, reboot my system using reboot or init 6 or init 0 or halt or poweroff or shutdown -h now, they all have the same result. After unloading the pcmcia modules the sytem has unmounted the drives and I get an kernel oops. Reproducible: Always Steps to Reproduce: 1. emerge sync on 2005 July 22 2. make new kernel 3. reboot to new kernel 4. get oops on every shutdown, halt, reboot, etc. Expected Results: It should just shutdown, I'm going to test 2.6.10 and see if I get the same thing. I did not have this problem in 2.6.11, but I can't find the ebuild for 2.6.11.
emerge --info missing.
Okay here is my emerge --info Gentoo Base System version 1.6.13 Portage 2.0.51.22-r2 (default-linux/x86/2005.0, gcc-3.3.5-20050130, glibc-2.3.5-r0, 2.6.12-gentoo-r6 i586) ================================================================= System uname: 2.6.12-gentoo-r6 i586 Mobile Pentium MMX dev-lang/python: 2.3.5 sys-apps/sandbox: 1.2.11 sys-devel/autoconf: 2.13, 2.59-r6 sys-devel/automake: 1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.5 sys-devel/binutils: 2.15.92.0.2-r10 sys-devel/libtool: 1.5.18-r1 virtual/os-headers: 2.6.11-r2 ACCEPT_KEYWORDS="x86" AUTOCLEAN="yes" CBUILD="i386-pc-linux-gnu" CFLAGS="-O2 -mcpu=i586 -fomit-frame-pointer" CHOST="i386-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/kde/2/share/config /usr/kde/3/share/config /usr/lib/X11/xkb /usr/share/config /var/qmail/control" CONFIG_PROTECT_MASK="/etc/gconf /etc/rc.d /etc/splash /etc/terminfo /etc/wget /etc/env.d" CXXFLAGS="-O2 -mcpu=i586 -fomit-frame-pointer" DISTDIR="/usr/portage/distfiles" FEATURES="autoconfig distlocks sandbox sfperms strict" GENTOO_MIRRORS="http://distfiles.gentoo.org http://distro.ibiblio.org/pub/Linux/distributions/gentoo" PKGDIR="/usr/portage/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="x86 X alsa apm avi berkdb bitmap-fonts crypt curl directfb emboss encode fbcon foomaticdb fortran gdbm gif gpm gtk gtk2 imlib ipv6 jpeg libg++ libwww mad mikmod mp3 mpeg ncurses ogg oggvorbis oss pam pdflib perl png python qt quicktime readline spell ssl tcltk tcpd tiff truetype truetype-fonts type1-fonts vorbis xml2 xmms zlib userland_GNU kernel_linux elibc_glibc" Unset: ASFLAGS, CTARGET, LANG, LC_ALL, LDFLAGS, LINGUAS, MAKEOPTS, PORTDIR_OVERLAY
Created attachment 64363 [details] 2.6.10 kernel config file I'm uploading my 2.6.10 config file. It seems that 2.6.10 has no problems with poweroff.
Created attachment 64364 [details] 2.6.12 kernel config file This is the 2.6.12 kernel config that I am using. This should be very similar to the 2.6.10, except I think 2.6.12 has more patches applied to it. I'd upload the oops message, but the fs is already unmounted and the system is not in working order at that point. I do have a photo of it that I will try to upload to show more.
Created attachment 64383 [details] image of the kernel oops an image of the kernel oops
Created attachment 64384 [details] kernel oops transcribed text My transcription of the kernel oops image I just uploaded for the download challenged
Okay, I have verified that when running 2.6.10 I do not have this problem. I think it is a patch applied to 2.6.12 that is not applied to 2.6.10.
Is this reproducable with the latest development kernel? (currently vanilla-sources-2.6.13_rc4)
see comment #8
(In reply to comment #9) > see comment #8 Okay I've noticed that there is an rc6 now, but I am going to emerge rc4. Also when I do the emerge its a masked package, and it also requires udev and hotplug, which are not on my system.
No, please use rc6 (rc4 was the latest 2 weeks ago). And yes you need udev, devfs is no more.
(In reply to comment #11) > No, please use rc6 (rc4 was the latest 2 weeks ago). And yes you need udev, > devfs is no more. Okay, I'm building vanilla sources 2.6.13 rc6. It will takea few hours on this old 233mhz.
Okay, just tested the poweroff with 2.6.13-rc6. It works. So does reboot. However none of my sound modules seem to load. I gets lots of unknown symbol in module. snd_pcm_lib_free_pages, snd_device_free, and more. I'll have to double check the config, but I think something else is going on. Then it complains about the wireless extensions for eth0 have been compiled with version 18, but this program supports version 17. My guess is I need to recompile or update cardmgr and pcmcia tools.
Sounds like you made some changes without updating modules or bzimage. Remove /lib/modules/2.6.13-rc6, remake everything, copy over a new bzimage, reinstall all your modules, and try again. You probably need to recompile your wireless stuff.
I did a make clean, and make bzImage and make module and then make modules_install. Then I copy'd the System.map and bzImage to /boot as well as the config file. I also did a modules-update too, just to see if that had any effect. I still get Unknown symbol snd_ctl_add, snd_pcm_new, etc, etc, etc Sound of course does not work..... is there some change in the /etc/modules.d file needed?
Did you "rm -rf /lib/modules/2.6.13-rc6" first as suggested?
(In reply to comment #16) > Did you "rm -rf /lib/modules/2.6.13-rc6" first as suggested? Yes, I did. That was the first thing I did. Did the /etc/modules.d config file format change for alsa or something?
gentoo-sources-2.6.13 now in portage
Okay here is what I found out. 2.6.12-r10 still has the bug. This is the current gentoo-sources version. 2.6.13 that was release does not have this issue. It however has another issue, or mabe something has changed and is not documented. In the /etc/modules.d/alsa file I had options snd device_mode=0666 which is supposed to set the permissions on the devices when they are created. This is not working for me. Once I removed this line, I am able to load the modules. However the problem ends up, that the devices are created with the wrong permissions, so I have to be root to use the devices. They are created with 660 instead of 666. Should I open a new bug for this new issue?
Yes if there is not one reported already
hmm correction, 2.6.13 does still have the shutdown issue, but the 2.6.13-r6 does not have this issue. to many kernel here for me ;-)
Strange. Is it the exact same kernel oops?
The Oops is pretty close to the initial one, only the call trace does not show the functions names backtrace ( I'm guessing that's what that is ). Addresses are different.
You need to use CONFIG_KALLSYMS to see the function names. Please capture a new image of the oops after you have compiled that option in. It would also be useful if you could test 2.6.13-rc7 so we can figure out if this problem (re)appeared between rc6 and rc7 or between rc7 and final
Created attachment 68639 [details] image of oops This attachment is an image of the oops. As the oops is not in the logs, as the filesystem is mounted readonly. I'm not going to transcsibe this, as it looks to me like the other one. I don't have access to 2.6.13-r7 as it is not in my portage tree. Do you have someplace to download it from?
Probably easiest to do it manually. In /usr/src, make sure "linux-2.6.12" does not exist and unpack http://www.kernel.org/pub/linux/kernel/v2.6/linux-2.6.12.tar.bz2 Also download http://www.kernel.org/pub/linux/kernel/v2.6/testing/patch-2.6.13-rc7.bz2 # cd linux-2.6.12 # bzcat /path/to/patch-2.6.13-rc7.bz2 | patch -p1 Now compile/install your kernel as usual. To prevent future confusion, you might also rename /usr/src/linux-2.6.12 to linux-2.6.13-rc7 :)
I think I am getting somewhere with this! In the case of 2.6.13-r7 I have been doing some real QA testing here to find out what the possible cause is. I ams also testing 2.6.12-gentoo-r10. I don't know enough about the kernel to say for sure, but I believe it may have something to do with alsa, in particular the snd-opl3sa2 sound card module, or opl3 module. Not 100% sure which one, but when these modules are loaded it oopses on reboots, shutdown, halt. 1) 2.6.13-rc7: I've setup the system so that the alsa snd module is options snd device-mode=0666 When the device gets created it SHOULD BE with mode 666 so that me as a user can use it without needing to be root. 2.6.13 does not recognize the device_mode option, so sound modules are NOT loaded because during boot the snd module is not recoginizing the device_mode=0666 which is okay in 2.6.10. The I reboot. System reboots with no problem! If I REMOVE the device_mode option in /etc/modules.d/alsa and rerun modules-update, then run /etc/init.d/alsasound restart the sound modules now get loaded. I then type reboot, only this time it gives an oops. It stops local, saves iptables state, firewall, klogd / syslogd, then alsa, then lo, then eth0, unmounting filesystems, then it goes into remounting remaining filesystems readonly... then oops.. system is still half up / mostly down, and lsmod shows snd modules still loaded. I tried this several times, to make sure that I was not just loosing my mind ;-). 2) 2.6.12-gentoo-r10: If I comment out the line below in the /etc/modules.d/alsa file: options snd-opl3sa2 fm_port=0x388 modi_port=0x330 port=0x370 wss_port=0x530 isapnp=0 dma1=1 dma2=0 irq=5 then run boot 2.6.12 it does not load the sound modules and it DOES NOT oops. If I uncomment these lines and do a modules-update and then restart alsasound, and try the reboot it then oopses. It is definately sound related! Both 2.6.13-rc7 and 2.6.12-gentoo-r10 reboot, halt and shutdown if the snd-opl3sa2 module is not loaded. I'm going to retest 2.6.13 to se if this problem exists there too. At this poin, I think that 2.6.13-rc6 may also have had this problem, but I did not notice because maybe I did not test it with sound (device_mode issue).
Yup it exists in 2.6.13 as well.
Ok, thanks for investigating that. Can you please test the latest development kernel (currently vanilla-sources-2.6.14_rc2), so that we can file an upstream bug report.
Hello, I downloaded the 2.6.14_r2 last night and it had finished compiling this morning. My initial test has been good, but I want to do some more testing tonight. It loaded the alsa drivers on the first boot and when I rebooted I did not get an oops. It seems like it is working okay, except alsa sound still does not recgonize device_mode anymore, which I have filed as bug -> 106400. Maybe I can configure this in udev? I'm going to test more after I get home from work and make sure all is okay with 2.6.14_r2, thanks.
Okay after further testing the oops is gone with 2.6.14_r2. I guess this means one bug is fixed, and 2 are created ;-). device_mode option still does not work and now there is no sound. The sound modules do get loaded. The same modules are getting loaded also.
gentoo-sources-2.6.14 now available in portage
