kdump: kexec_file_load失败:无法分配请求的地址

Question

问题：

SERVER:~ # systemctl start kdump.service
Job for kdump.service failed because the control process exited with error code. See "systemctl status kdump.service" and "journalctl -xe" for details.
SERVER:~ # systemctl status kdump.service
● kdump.service - Load kdump kernel on startup
   Loaded: loaded (/usr/lib/systemd/system/kdump.service; enabled; vendor preset: disabled)
   Active: failed (Result: exit-code) since Wed 2018-10-17 12:29:34 EDT; 1s ago
  Process: 59804 ExecStart=/lib/kdump/load.sh (code=exited, status=1/FAILURE)
 Main PID: 59804 (code=exited, status=1/FAILURE)

Oct 17 12:29:33 SERVER systemd[1]: Starting Load kdump kernel on startup...
Oct 17 12:29:34 SERVER load.sh[59804]: kexec_file_load failed: Cannot assign requested address
Oct 17 12:29:34 SERVER systemd[1]: kdump.service: Main process exited, code=exited, status=1/FAILURE
Oct 17 12:29:34 SERVER systemd[1]: Failed to start Load kdump kernel on startup.
Oct 17 12:29:34 SERVER systemd[1]: kdump.service: Unit entered failed state.
Oct 17 12:29:34 SERVER systemd[1]: kdump.service: Failed with result 'exit-code'.
SERVER:~ # 

日志：

SERVER:~ # tail /var/log/messages
2018-10-17T12:29:33.980232-04:00 SERVER systemd[1]: Starting Load kdump kernel on startup...
2018-10-17T12:29:34.133151-04:00 SERVER kdump[59974]: FAILED to load kdump kernel: /sbin/kexec -p /boot/vmlinuz-4.4.121-92.80-default --append="quiet console=tty0 console=ttyS0,9600 elevator=noop transparent_hugepage=never numa_balancing=disable intel_idle.max_cstate=1 elevator=deadline sysrq=yes reset_devices acpi_no_memhotplug cgroup_disable=memory irqpoll nr_cpus=1 root=kdump rootflags=bind rd.udev.children-max=8 disable_cpu_apicid=0   panic=1" --initrd=/boot/initrd-4.4.121-92.80-default-kdump  -s, Result: kexec_file_load failed: Cannot assign requested address
2018-10-17T12:29:34.133560-04:00 SERVER load.sh[59804]: kexec_file_load failed: Cannot assign requested address
2018-10-17T12:29:34.133726-04:00 SERVER systemd[1]: kdump.service: Main process exited, code=exited, status=1/FAILURE
2018-10-17T12:29:34.133958-04:00 SERVER systemd[1]: Failed to start Load kdump kernel on startup.
2018-10-17T12:29:34.134105-04:00 SERVER systemd[1]: kdump.service: Unit entered failed state.
2018-10-17T12:29:34.134233-04:00 SERVER systemd[1]: kdump.service: Failed with result 'exit-code'.
SERVER:~ #

版本的信息：

SERVER:~ # rpm -qa|grep -i kdump
yast2-kdump-3.1.44-11.6.15.x86_64
kdump-0.8.15-28.5.x86_64
SERVER:~ # uname -a
Linux SERVER 4.4.121-92.80-default #1 SMP Mon May 21 14:40:10 UTC 2018 (2afdd00) x86_64 x86_64 x86_64 GNU/Linux
SERVER:~ #
SERVER:~ # cat /etc/SuSE-release
SUSE Linux Enterprise Server 12 (x86_64)
VERSION = 12
PATCHLEVEL = 2
# This file is deprecated and will be removed in a future service pack or release.
# Please check /etc/os-release for details about this release.
SERVER:~ #

问题是:为什么kdump.service不能启动？我遗漏了什么？

AFAIK 12不需要内核-kdump包，还是我错了？如果是，从哪里可以得到内核-kdump包？

基于https://distrowatch.com/table-mobile.php?distribution=sle&pkglist=true&version=12-sp2，kdump版本看起来没问题。

2018年12月5日最新情况：

• rpm -V kdump-0.8.15-28.5.x86_64；回波$？->它是0，所以好
• 我找到了一台内核版本相同的机器，但在那里，kdump可以工作！但却找不到健康和坏宿主之间的区别。
• 试图取代initrd但没能帮上忙。
• 试图重新安装kdump，没有帮助: rpm -e yast2-kdump；rpm -e kdump；zypper in kdump
• 试图做一个"systemctl取消掩码kdump；systemctl启用kdump；systemctl重新启动kdump“和"systemctl守护进程重新加载”，没有起到什么作用。

2018年12月07最新情况：

cat /proc/cmdline
BOOT_IMAGE=/vmlinuz-4.4.121-92.80-default root=/dev/mapper/vg00-lv_root splash=silent quiet showopts console=tty0 console=ttyS0,9600 elevator=noop transparent_hugepage=never crashkernel=768M numa_balancing=disable intel_idle.max_cstate=1

2018年12月11日更新:在无法启动kdump的节点上发布/proc/iomem：

SERVER:~ # cat /proc/iomem
00000000-00000fff : reserved
00001000-0009bfff : System RAM
0009c000-0009ffff : reserved
000a0000-000bffff : PCI Bus 0000:00
000c0000-000c7fff : Video ROM
000cd800-000d53ff : Adapter ROM
000e0000-000fffff : reserved
  000f0000-000fffff : System ROM
00100000-5eeb0fff : System RAM
  01000000-015fbb30 : Kernel code
  015fbb31-01d59b7f : Kernel data
  01f6b000-021e8fff : Kernel bss
5eeb1000-66eb8fff : reserved
66eb9000-6a733fff : System RAM
6a734000-6a742fff : reserved
6a743000-6a743fff : System RAM
6a744000-7a7c4fff : reserved
7a7c5000-7cc82fff : System RAM
7cc83000-7ccb4fff : reserved
7ccb5000-a41b7fff : System RAM
a41b8000-b93fefff : reserved
b93ff000-bb3fefff : ACPI Non-volatile Storage
bb3ff000-bb7fefff : ACPI Tables
bb7ff000-bb7fffff : System RAM
bb800000-cfffffff : reserved
  c0000000-cfffffff : PCI MMCONFIG 0000 [bus 00-ff]
d0000000-e7ffbfff : PCI Bus 0000:00
  d0000000-d01fffff : PCI Bus 0000:06
    d0000000-d00fffff : 0000:06:00.0
    d0100000-d01fffff : 0000:06:00.1
  d0200000-d020ffff : 0000:00:11.0
  d03fc000-d03fcfff : 0000:00:05.4
  d03fe000-d03fe3ff : 0000:00:1a.0
    d03fe000-d03fe3ff : ehci_hcd
  d03ff000-d03ff3ff : 0000:00:1d.0
    d03ff000-d03ff3ff : ehci_hcd
  d0400000-d05fffff : PCI Bus 0000:0b
    d04f0000-d04fffff : 0000:0b:00.0
      d04f0000-d04fffff : megasas: LSI
    d0500000-d05fffff : 0000:0b:00.0
  d0600000-d0ffffff : PCI Bus 0000:11
    d0600000-d0ffffff : PCI Bus 0000:12
      d0600000-d06fffff : PCI Bus 0000:15
        d06fe000-d06fefff : 0000:15:00.0
        d06ff000-d06fffff : 0000:15:00.0
      d0700000-d0ffffff : PCI Bus 0000:13
        d0700000-d0ffffff : PCI Bus 0000:14
          d07fc000-d07fffff : 0000:14:00.0
          d07fc000-d07fffff : mgadrmfb_mmio
          d0800000-d0ffffff : 0000:14:00.0
  d1000000-d1ffffff : PCI Bus 0000:11
    d1000000-d1ffffff : PCI Bus 0000:12
      d1000000-d1ffffff : PCI Bus 0000:13
        d1000000-d1ffffff : PCI Bus 0000:14
          d1000000-d1ffffff : 0000:14:00.0
          d1000000-d1ffffff : mgadrmfb_vram
  d2000000-d5ffffff : PCI Bus 0000:06
    d2000000-d3ffffff : 0000:06:00.0
      d2000000-d3ffffff : mlx5_core
    d4000000-d5ffffff : 0000:06:00.1
      d4000000-d5ffffff : mlx5_core
e7ffc000-e7ffcfff : dmar1
e8000000-fbffbfff : PCI Bus 0000:80
  e8000000-e81fffff : PCI Bus 0000:81
    e8000000-e80fffff : 0000:81:00.0
    e8100000-e81fffff : 0000:81:00.1
  e9fff000-e9ffffff : 0000:80:05.4
  ea000000-edffffff : PCI Bus 0000:81
    ea000000-ebffffff : 0000:81:00.0
      ea000000-ebffffff : mlx5_core
    ec000000-edffffff : 0000:81:00.1
      ec000000-edffffff : mlx5_core
fbffc000-fbffcfff : dmar0
fec00000-fecfffff : PNP0003:00
  fec00000-fec003ff : IOAPIC 0
  fec01000-fec013ff : IOAPIC 1
  fec40000-fec403ff : IOAPIC 2
fed00000-fed003ff : HPET 0
  fed00000-fed003ff : PNP0103:00
fed12000-fed1200f : pnp 00:01
fed12010-fed1201f : pnp 00:01
fed1b000-fed1bfff : pnp 00:01
fed1c000-fed1ffff : reserved
  fed1f410-fed1f414 : iTCO_wdt.0.auto
fed45000-fed8bfff : pnp 00:01
fee00000-feefffff : pnp 00:01
  fee00000-fee00fff : Local APIC
ff000000-ffffffff : reserved
  ff000000-ffffffff : pnp 00:01
100000000-1003fffffff : System RAM
38000000000-3bfffffffff : PCI Bus 0000:00
  38000000000-38000000fff : 0000:00:1f.6
  3800000c000-3800000c00f : 0000:00:16.0
  3800000d000-3800000d00f : 0000:00:16.1
  3800000e000-3800000e0ff : 0000:00:1f.3
  38000010000-3800001ffff : 0000:00:14.0
    38000010000-3800001ffff : xhci-hcd
3c000000000-3ffffffffff : PCI Bus 0000:80
SERVER:~ #

Answer 1

让我用所提供的资料尽我所能回答。

首先，SLES 12 (及更高版本)实际上不需要内核-kdump包。这种特殊的内核风格仅在古代才需要，因为恐慌内核必须以与运行中的内核不同的物理地址加载，但是只有在编译时才能更改加载地址(也就是内核是不可重定位的)。

其次，kdump不会启动，因为底层的kexec_file_load系统调用在EADDRNOTAVAIL中失败。如果系统无法分配一个或多个缓冲区来将惊慌内核加载到RAM中，则会发生这种情况。请注意，理论上可能有足够的内存用于恐慌内核，但由于分配具有Linux内核引导代码和/或驱动程序施加的一些附加约束，此RAM可能无法用于加载恐慌内核。由于不同的物理内存布局，另一个系统可能更幸运。

作为第一步，我将尝试增加内核命令行(例如crashkernel=256M)上的保留内存大小，重新启动并查看是否有帮助。