On Fri, Jun 11, 2021 at 1:57 AM Rich Ercolani <rincebr...@gmail.com> wrote: > > Package: kdump-tools > Version: 1:1.6.5-1 > Severity: important > > Dear Maintainer, > > Installed kdump-tools appear to work, echo 'c' | sudo tee /proc/sysrq-trigger > boots the crashkernel and > takes a dump and reboots, but attempting to use "crash" on the dump in a way > that works > in other configurations (e.g. crash /usr/lib/debug/boot/vmlinux-1234 > /var/crash/1234/dump.1234) reports > (in this example, on buster, with -d1): > > crash: page excluded: kernel virtual address: c19546a4 type: "possible" > WARNING: cannot read cpu_possible_map > crash: page excluded: kernel virtual address: c195469c type: "present" > WARNING: cannot read cpu_present_map > crash: page excluded: kernel virtual address: c19546a0 type: "online" > WARNING: cannot read cpu_online_map > crash: page excluded: kernel virtual address: c1954698 type: "active" > WARNING: cannot read cpu_active_map > crash: page excluded: kernel virtual address: c18c769c type: "pv_init_ops" > crash: page excluded: kernel virtual address: c1a92268 type: > "shadow_timekeeper xtime_sec" > xtime timespec.tv_sec: 9aee2b: Tue Apr 28 08:25:15 1970 > crash: page excluded: kernel virtual address: c18bb1c4 type: "init_uts_ns" > utsname: > sysname: > nodename: > release: > version: > machine: > domainname: > base kernel version: 0.0.0 > crash: page excluded: kernel virtual address: c16a9160 type: "accessible > check" > crash: /usr/lib/debug/vmlinux-4.19.0-16-686-pae and > /var/crash/202106110101/dump.202106110101 do not match! > > And: > # ls -al /var/crash/202106110101/dump.202106110101;file > /var/crash/202106110101/dump.202106110101; > -rw------- 1 root root 21422351 Jun 11 01:01 > /var/crash/202106110101/dump.202106110101 > /var/crash/202106110101/dump.202106110101: Kdump compressed dump v6, system > Linux, node debian32, release 4.19.0-16-686-pae, version #1 SMP Debian > 4.19.181-1 (2021-03-19), machine i686, domain (none) > > crashkernel=512M-:192M, on here, changes the behavior to what #989714 > describes on x86_64 - > no printouts from a crashkernel or anything else, no dump ever saved, just an > indefinite hang. > crashkernel=384M-:192M (or 256M) does not hang, but produces an equally > useless "dump". > > According to the console output when this happens, "makedumpfile Completed" > both times, and > dmesg's output appears intact and complete. > > (I neglected to mention this in #989714, but these are all on > VirtualBox-backed VMs.) > > Please let me know if there's anything further I can do to help debug this.
So it sounds like you were able to get a successful dump after adjusting crashkernel, is that correct? And the issue is that crash doesn't deal with it? Wouldn't that be a bug in the crash tool then? -dann
