[15:00] <slanck> i got a kernel crash dump and would like to analyze it with GDB but the .crash file on /var/crash doesn't seem to have a CoreDump file when I unpack it using apport-unpack
[15:01] <slanck> and using the .crash file or the file generated on /var/crash/DATE/dump.XYZ with GDB gives incorrect file format
[15:02] <slanck> I would like to have the CoreDump file to analyze it using GDB, apport-retrace gives the error of missing files
[16:04] <slanck> it's working now
[16:05] <gpiccoli> slanck, you need to use the crash tool in order to analyze kernel dumps
[16:05] <gpiccoli> need was a strong statement...it's recomended that you use the crash tool, it's meant to that heheh
[16:06] <gpiccoli> for userspace core dumps, gdb is usually the proper tool =)
[16:07] <slanck> gpiccoli: hi
[16:08] <slanck> yes, I was using the crash command. The issue was that, i don't know yet why, but the dump.XYZ file generated on /var/crash/ was kind of incomplete
[16:08] <slanck> i regenerated the crash using echo c > /proc/sysrq... and now the dump.XYZ file worked as expected using crash
[16:09] <slanck> the first dump is not detected as a kdump file
[16:09] <slanck> sudo file 202007010222/dump.202007010222 
[16:09] <slanck> 202007010222/dump.202007010222: data
[16:10] <slanck> sudo file 202007011558/dump.202007011558 
[16:10] <slanck> 202007011558/dump.202007011558: Kdump compressed dump v6, system Linux,
[16:11] <slanck> i don't know what happened, when the first dump file was generated, i didn't have the linux-crashdump package installed even though there was a crash command available on my system
[16:11] <slanck> i then installed the linux-crashdump and generated a new crash and got the result shown above, i guess is that, without linux-crashdump, it generates a dump but kind of incomplete
[16:12] <slanck> when i installed linux-crashdump, it didn't replace the crash tool i had already on my system, i checked the version and it's the same 
[16:12] <slanck> crash 7.2.8
[16:14] <slanck> gpiccoli: thank you anyways
[16:16] <gpiccoli> slanck, interesting, so you got a bad dump on the first time. The package linux-crashdump is a meta pkg that maps to all usually required packages to have kdump working
[16:16] <gpiccoli> the real packages you need to collect the dump properly (and they all come with linux-crashdump) are kexec-tools, kdump-tools and makedumpfile
[16:17] <gpiccoli> The crash package (also included on the metapkg) brings the crash binary to analyze the vmcore =)
[16:22] <slanck> @gpiccoli thank you very much, the bad dump was generated triggering a kernel bug and the good dump shown above was generated using echo c > /proc/sysrq, it's worth to mention I am working on a custom kernel
[16:23] <slanck> i will trigger the bug again later today and see if it generates a "good" dump now that I've installed linux-crashdump
[16:25] <gpiccoli> awesome, hope it works!
[16:25] <gpiccoli> What version of Ubuntu are you suing ?
[16:25] <gpiccoli> *using
[16:26] <gpiccoli> I suggest to increase a bit the crashkernel memory if it's not working reliably, also worth trying to capture the console output during the kdump kernel boot
[16:26] <slanck> Ubuntu 20.04, it's a custom kernel based on vmlinuz-5.4.0-37-generic
[16:27] <slanck> i had some issues with crashkernel I guess because I am using a KASAN-enabled kernel
[16:28] <slanck> initially the system was only able to boot properly when I configured the VM with more or equal 6GB of RAM, less than that and I got a crash during boot
[16:29] <slanck> with 6GB of RAM, i had to increase crashkernel memory up to the point I got kdump properly configured and working
[16:30] <slanck> with working I mean being able to load the kexec kernel and creating a dump file on /var/crash but I didn't need to investigate the dump file until today
[16:30] <slanck> I had to set crashkernel=2048M to get to that point
[16:31] <slanck> oh, sorry, the VM has 12GB of RAM and crashkernel=2048M
[16:34] <slanck> i triggered the kernel bug again and now it seems it working fine, the issue seems to be that the package linux-crashdump was missing
[16:49] <slanck> it kind of works, i don't have the registers when the system crashed
[16:49] <slanck> crash> i r
[16:49] <slanck> The program has no registers now.
[16:49] <slanck> gdb: gdb request failed: i r
[16:49] <slanck> crash> 
[16:50] <slanck> sudo crash vmlinux_symbols dump.202007011645
[17:41] <gpiccoli> is "i r" a command on crash? I'm not sure
[17:41] <gpiccoli> try checking some structure or some pointer you know the address, like linux_banner or something like this
[21:38] <weskrasko> Hi, #nm sent me here. I followed some instructions on a site to fix my Nvidia drivers, ran "sudo ubuntu-drivers autoinstall", fixed Nvidia but then after reboot, all my network interfaces are gone!
[21:39] <weskrasko> So I ran ip link show as recommended in other channel. Shows only lo.
[21:39] <weskrasko> Ran lspci, shows Ethernet and Wireless cards, but dmesg doesn't seem to show any of that.
[21:40] <weskrasko> They said maybe my kernel is now not loading the firmware/driver/module's needed? I don't know what happened, my wifi was working "out of the box" before this.
[22:25] <jeremy31> weskrasko: was there a recent kernel update?  You might want to see if the linux-modules-extra was also installed for that kernel version
[22:26] <weskrasko> Thanks, how would I do that?
[22:27] <weskrasko> And yes, right after I ran the command above I did a normal update as I always do, I saw kernel and firmware stuff. NEver had an issue with udpating so thought nothing of it.
[22:28] <jeremy31> weskrasko: check in terminal>  dpkg -l | grep linux-modules-extra-$(uname -r)
[22:36] <weskrasko> Nope, no limux-modules-extra for current kernel
[22:36] <weskrasko> How can I fix with NO connection? Download deb?
[22:38] <jeremy31> weskrasko: you can use grub menu to boot into older kernel, then install the linux-modules-extra for the newer kernel
[22:43] <weskrasko> OMG thank you it worked!! Never had that happen before, but at least it's back up. :)
[22:43] <sarnold> jeremy31: nice :D