Anyone have experience with segfaults halting the boot process?

[$trapped]

I have a family member who has a computer running Manjaro. It froze up on the user; I rebooted but I keep getting a segfault error during the boot process and then it just sits there. I've reseated the RAM modules, I've rebooted with only one stick in and every time I get:

[93.475400] systemd[1]: segfault at 7ffc8ecfa188 (pretty sure this changes each time I've tried) ip (another string of changing numbers/letters) sp (more changing number/letters) error 6 in libsystemd-shared-255.2-2.so[more changing numbers/letters] likely on CPU (X) (core X, socket X) X is a single number that changes each time I attempt to boot.


Anyone have any idea what I'm dealing with?

System details:
AMD 2200G
2X 4gb ram
asrock b450m pro4
128gb ssd

 

[learners permit]

Seg fault definition and causes. Also check with multimeter the psu's outputs on 3.3v and 5v outputs and see if they sag below spec. 5v. can be checked at a molex connector red wire 3.3v is found in the 24 pin mainboard plug and can be carefully back probed at the green wires. Software is most likely the cause but memory addressing can't be discounted either. Try a bootable distro either usb or disk to see if the issue remains and proceed from there.

 

[$trapped]

I've successfully booted a usb distro, but I don't have a multimeter so that'll have to wait.

 

[learners permit]

No issues with the usb distro means you can forget the psu testing as it's likely a software or kernel issue memory address violation.

 

[$trapped]

If it was kernel related wouldn't this have happened immediately after changing the kernel?

 

[learners permit]

Yea it likely should have. You can create a bootable memtest flash drive and let it run overnight to see if any errors pop up. I really don't know a lot about linux distros as I have only occasionally dabbled with a few but I found a decent guide for getting to terminal for further troubleshooting and error log analysis by folks in the know on the manajaro forum here or use the bootable usb you made earlier to examine the logs for errors and ask for help with the logs. Sorry I can not be of more help but my ability to help you has reached it's limit.

 

[toast0]

Either your filesystem got corrupted, your SSD is giving junk data, or the software is bad.

Maybe you can check file checksums from the USB live distro.

 

[$trapped]

I appreciate the help. I will work on this more tomorrow. It's been a while since I've had to do any troubleshooting so hopefully it doesn't take me too long to get in the swing of it. Thanks again.

 

[michalrz]

I appreciate the help. I will work on this more tomorrow. It's been a while since I've had to do any troubleshooting so hopefully it doesn't take me too long to get in the swing of it. Thanks again.

If it's not hardware, then the only thing that changes between each boot are log files.
That would explain why the "numbers" change.
Sounds like a corrupted filesystem or corrupted kernel filesystem module, maybe corrupted initrd image.
Restoring the bootloader might help, I think it also builds initrd images (not sure!). Recreate the swap space. Depending on how far along this happens, might help.

 

[$trapped]

Ok. So I've tried rebooting using a different kernel, no dice. Tried changing run level, no dice. I probably should have tried to recover using a live distro but I chose to run memtest86+ next. So far it's gone through 14 passes and there are 10 errors. Attached is a screen shot I took just after the 1st error showed up. Since I've already tried booting with each stick, should I just accept it's time to replace the ram or should I attempt to move slots and see if that slot is bad? Before the error, memtest was showing slots 2 and 3 which confused me because the bios shows the sticks in 2 and 4.

 

[toast0]

Oh. Bad ram makes sense. Not sure why the live usb didn't have problems.

Double check ram voltage in bios? Maybe give it a bit more voltage if it's not already high?

 

[$trapped]

Oh. Bad ram makes sense. Not sure why the live usb didn't have problems.

Double check ram voltage in bios? Maybe give it a bit more voltage if it's not already high?

Yeah, I was wondering why the live usb was ok too. Should I keep going with the memtest or stop?

 

[toast0]

Yeah, I was wondering why the live usb was ok too. Should I keep going with the memtest or stop?

No reason to continue testing at the current settings. Once you have one failure, ram is bad. You can run further tests in different configs if you want...

 

[learners permit]

Memory bit flip potential causes. Memory bit flipped at that address and likely wrote an error to the kernel requiring a reinstall of the OS.

 

[$trapped]

Ok, so new ram needed. Thanks guys, appreciate the guidance/knowledge.

 

[michalrz]

Yeah, I was wondering why the live usb was ok too.

It probably was lucky enough to position itself within the 'good' area of the memory.
It's cool you got it with Memtest, very fortunate actually to have it pop up so quickly.

 

[$trapped]

Well, installed new ram and got the same response. I did a quick couple of passes through memtest to make sure I didn't get a bad batch, but it came out clean. So it looks like this is software related. At least it spat out more information than the previous attempts although I have no idea if it provided anything useful.

 

[SmokeRngs]

Well, installed new ram and got the same response. I did a quick couple of passes through memtest to make sure I didn't get a bad batch, but it came out clean. So it looks like this is software related. At least it spat out more information than the previous attempts although I have no idea if it provided anything useful.

If the memory was bad and caused errors, it's likely the install, OS and otherwise, is completely hosed. Bad memory causes errors in which those errors are written to disk. It was unlikely the install was going to be saved from the beginning and will need to be wiped and redone.