libc segfault after upgrade to -current. upowerd issue?
 

Hi,

i have upgraded machine no. 1 to -current which went reasonably well.

So i thought it safe to upgrade machine no. 2. That one is now stuck and will not boot, with none of my kernels.

upgrade procedure in both cases:

• both machines were on -current as of 30th of June (ie. already on libc-2.15.so)
• both machines are x86_64 with Eric's multilibs (up to date as of 30th of June)
• manually upgraded packages aaa_elflibs, openssl_solibs, openssl, wget, slackpkg
• slackpkg update, install-new, upgrade-all, clean-system with manual review of .new files
• reboot

Machine no. 1 booted fine and got a complete re-install of the nvidia driver plus xwmconfig for the users. Machine no. 1 works fine.

Machine no. 2 fails to boot. Symptoms (taken from pictures taken with digital camera):

• Hardware recognition as per usual, rc.S starts
• "Triggering udev events" (ie. rc.udev starts)
• "mtp-probe[1031]: segfault at 0 ip 00007f73398c1300 sp 00007ffffb53c508 error 4 in libc-2.15.so[7f7339809000+1b3000]"
• hwclock and fsck start
• fsck says "No such file or directory while trying to open /dev/sda5"
• rc.S says "An error occurred..." and "Once you exit the single-user shell, the system will reboot."
• the system reboots immediately (saying "Unmounting filesystems." ... "Rebooting system.")

I have tried commenting out anything related to lvm and modprobe in rc.S and rc.alsa (since i don't use them) but it didn't make a difference.

A google search on the segfault brings up a few hits about mounting iphones, but nothing helpful. There is a patch for libiphone

http://libiphone.lighthouseapp.com/p...ith-home-unset

but i am unsure how that is relevant here. It points towards upowerd, though.

Any ideas?

addendum: prompted by the patch I tried setting
in rc.S but without effect

Hi,
experienced the same issue today after upgrading to -current,
on a machine where I run a custom kernel.

-current brings udev to version 182 and I just discovered that this one needs CONFIG_DEVTMPFS=y and CONFIG_DEVTMPFS_MOUNT=y in kernel, something I had not enabled.
So I'll have to recompile my kernel.

You might want to check whether this applies to your configuration, as well.

sorry i can't report any news, but the plot has thickened. Not knowing about the issue my son started the machine and then switched off power while in BIOS startup, meaning the BIOS settings are gone and I need to obtain a suitable keyboard to get into the BIOS settings to switch on the other keyboard and then redo finicky memory and cpu settings and test the stability over night, before I can turn to the upowerd problem once more. Said the last action hero: I hate it when this happens.

edit: thanks, giacomo, that's precisely what i would have checked/done next

Having restored everything else I made some progress on the original issue: setting CONFIG_DEVTMPFS=y and CONFIG_DEVTMPFS_MOUNT=y in the kernel config is indeed the solution. Marking the thread as [SOLVED].