Bug 835340
Summary: | panic during boot: unable to open libgcc_s.so.1 | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | George R. Goffe <grgoffe> | ||||||
Component: | dracut | Assignee: | dracut-maint | ||||||
Status: | CLOSED CANTFIX | QA Contact: | Fedora Extras Quality Assurance <extras-qa> | ||||||
Severity: | high | Docs Contact: | |||||||
Priority: | unspecified | ||||||||
Version: | 16 | CC: | dracut-maint, gansalmon, harald, itamar, jonathan, kernel-maint, madhu.chinakonda | ||||||
Target Milestone: | --- | ||||||||
Target Release: | --- | ||||||||
Hardware: | x86_64 | ||||||||
OS: | Linux | ||||||||
Whiteboard: | |||||||||
Fixed In Version: | Doc Type: | Bug Fix | |||||||
Doc Text: | Story Points: | --- | |||||||
Clone Of: | Environment: | ||||||||
Last Closed: | 2012-09-27 09:30:26 UTC | Type: | Bug | ||||||
Regression: | --- | Mount Type: | --- | ||||||
Documentation: | --- | CRM: | |||||||
Verified Versions: | Category: | --- | |||||||
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||||
Cloudforms Team: | --- | Target Upstream Version: | |||||||
Embargoed: | |||||||||
Attachments: |
|
Description
George R. Goffe
2012-06-26 03:25:37 UTC
That sounds like your initramfs is messed up. Do: lsinitrd /boot/initramfs-3.4.2-1.fc16.x86_64.img | grep libgcc and see if it outputs any files. Also, you say that 3.3.8-1 is the version impacted, but your reproduce steps only mention 3.4.2-1. Josh, Here's what I get: joker bash-4.2 ~# lsinitrd /boot/initramfs-3.4.2-1.fc16.x86_64.img | grep libgcc -rwxr-xr-x 1 root root 87016 Mar 6 09:54 run/initramfs/lib64/libgcc_s-4.6.3-20120306.so.1 lrwxrwxrwx 1 root root 28 Jun 26 12:57 run/initramfs/lib64/libgcc_s.so.1 -> libgcc_s-4.6.3-20120306.so.1 Notice that the "run" above does NOT have "/var/" preceding it. I re-created the initramfs-3.4.2-1.fc16.x86_64.img file in /boot but get the same results sort of. Now it's a panic loop which I'm unable to capture. Argh! The message "udevd... killing /sbin/blkid -o udev -u /dev/sda6 appears frequently then there's a trace dump then it repeats... looping I'm guessing. I let it run for about 20 mins... My partitions are old style (see below). Perhaps someone has made an assumption that everyone has all these partitions in one big partition? /dev/sda2 2093160 1638816 433376 80% / /dev/sda5 3138272 829016 2277800 27% /var /dev/sda1 2093160 241204 1830988 12% /boot /dev/sda7 30334620 19285024 11049596 64% /usr /dev/sda8 668965300 534091292 128177460 81% /export/home /dev/sda6 21953708 21379060 354448 99% /opt Which version of dracut is this? What is the contents of /etc/dracut.conf and /etc/dracut.conf.d/* Harald, At the time of the failure, the version of dracut was dracut-013-22.fc16.noarch. I got a newer version of dracut from the git repository for the purposes of getting system dumps. The newer version is dracut-018-55.git20120606.fc16.noarch. George... (In reply to comment #3) > What is the contents of /etc/dracut.conf and /etc/dracut.conf.d/* you did not answer my question Does it work with dracut-018-55.git20120606.fc16 from the updates? Note, you have recreate the initramfs. # dracut -f /boot/initramfs-<kernel-version>.img <kernel-version> or a shortcut: # dracut -f '' <kernel-version> Created attachment 596675 [details]
gzip'd tar of dracut config fiies
Harald,
It looks like my config file tar ball didn't make it into this bug.
I have just rebuilt the initramfs and will test in a few minutes and post the results here.
Thanks,
George...
there is a fips.conf in /etc/dracut.conf.d . Are you sure, you want a FIPS system? If not, uninstall dracut-fips or remove the config file. Harald, I have not made any config changes with dracut. As far as I know, the FIPS stuff came along with dracut updates and/or the install of this system. Also, I re-recreated the initramfs with dracut. The resulting boot was highly unstable. Without a way to record the messages during boot (new systemd?) I have no way to show just what happened. There are some messages in /var/log/messages though. I see there's a new kernel in town, I'll try booting this and then do the dracut if it appears necessary. THANKS! George... (In reply to comment #8) > Harald, > > I have not made any config changes with dracut. As far as I know, the FIPS > stuff came along with dracut updates and/or the install of this system. > > Also, I re-recreated the initramfs with dracut. The resulting boot was > highly unstable. Without a way to record the messages during boot (new > systemd?) I have no way to show just what happened. There are some messages > in /var/log/messages though. I see there's a new kernel in town, I'll try > booting this and then do the dracut if it appears necessary. > > THANKS! > > George... # yum remove dracut-fips with systemd and udev > 020-57 debugging will be a lot easier... just try rawhide tomorrow # journalctl -a will display _all_ messages Harald, I removed the dracut-fps. What is "systemd and udev > 020-57"? A config file option? Which config file? Anything else that I need to do? journalctl -a gives me a "command not found" msg. What package? Thanks, George... (In reply to comment #10) > Harald, > > I removed the dracut-fps. > > What is "systemd and udev > 020-57"? A config file option? Which config > file? Anything else that I need to do? > > journalctl -a gives me a "command not found" msg. What package? > > Thanks, > > George... Oh ... sorry, you are on F16. Forget that comment. (In reply to comment #8) > Harald, > > I have not made any config changes with dracut. As far as I know, the FIPS > stuff came along with dracut updates and/or the install of this system. > > Also, I re-recreated the initramfs with dracut. The resulting boot was > highly unstable. Without a way to record the messages during boot (new > systemd?) I have no way to show just what happened. There are some messages > in /var/log/messages though. I see there's a new kernel in town, I'll try > booting this and then do the dracut if it appears necessary. > > THANKS! > > George... With F16 the debugging procedure is as following: - add "rd.debug rd.shell" to the kernel command line - if dracut does not boot, you are dropped to a shell - "/run/initramfs/init.log" and the output of dmesg contain the debug information - you can mount /boot or an USB stick by hand and copy over the /run/initramfs/init.log Created attachment 599870 [details]
flat file containing current dmesg buffer
Harald,
I still have a kernel that doesn't want to boot and/or execute properly. Here's the current dmesg buffer from my system. It shows a TON of msgs from dracut.
I'm attempting to capture console messages from the failing kernel but that's not working very well. I'm trying the netconsole procedure now.
I have only one operational kernel now since I have removed the kernel-3.4.4-4.fc16.x86_64 which is still having trouble from my system.
Your thoughts please?
Regards,
George...
[ 49.071088] bootchart-collector started as pid 1769 with 2 args: '--dump' '/tmp/bootchart.JB5pGt7Qfn' seems like bootchart is running... maybe you should disable that first. you could try to regenerate the initramfs. # dracut -f /boot/initramfs-3.3.8-1.fc16.x86_64.img 3.3.8-1.fc16.x86_64 And maybe try and see, if it is dracut that is failing by verifying that you cannot get to a shell with "init=/bin/sh" on the kernel command line. Harald, I ran "yum update" yesterday and got a new kernel (3.4.6-1.fc16.x86_64) and this problem seems to have gone away. We can wait a few days and then close this case if you want to. Regards and THANKS for all your help. George... |