IBM Support

Fadump fails under heavy workload and repeated DLPAR memory operations

News


Abstract

Firmware-assisted memory dump (Fadump) fails to capture the memory dump when the system is under heavy workload with repeated dynamic logical partitioning (DLPAR) memory operations. In SAP HANA environments, during the BWH Phase 2 workload, CPU and memory usage can exceed 70%. When the system crashes under these conditions and fadump is unable to capture the memory dump, then the result is memory dump capture failure.

Content

Linux Releases Affected

SUSE Linux Enterprise Sserver (SLES) 16.0

IBM Systems Affected

IBM Power11 systems that run SLES 16.0

Symptoms

When the system fails, the fadump kernel boot process (which captures the memory dump) gets stuck, and the console shows the following logs:

Loading Linux 6.12.0-160000.6-default ...
appendedsig: successfully verified the signature with a trusted key
Loading initial ramdisk ...
OF stdout device is: /vdevice/vty@30000000
Preparing to boot Linux version 6.12.0-160000.6-default (geeko@buildhost) (gcc-13 (SUSE Linux) 13.4.0, GNU ld (GNU Binutils; SUSE Linux 16) 2.43.1.20241209-160000.2) #1 SMP Fri Oct 17 10:54:40 UTC 2025 (724dacd)
Detected machine type: 0000000000000101
command line: BOOT_IMAGE=/boot/vmlinux-6.12.0-160000.6-default root=UUID=1ac607d3-6259-4830-a4b3-b4dfcb117cc2 splash=silent mitigations=auto quiet security=selinux selinux=1 transparent_hugepage=never crashkernel=1
84320M fadump=on
Max number of cores passed to firmware: 1024 (NR_CPUS = 8192)
Calling ibm,client-architecture-support... done
memory layout at init:
  memory_limit : 0000000000000000 (16 MB aligned)
  alloc_bottom : 0000000012cc0000
  alloc_top    : 0000000030000000
  alloc_top_hi : 0000000100000000
  rmo_top      : 0000000030000000
  ram_top      : 0000000100000000
instantiating rtas at 0x000000002eb90000... done
prom_hold_cpus: skipped
copying OF device tree...
Building dt strings...
Building dt structure...
Device tree strings 0x0000000012cd0000 -> 0x0000000012cd19f9
Device tree struct  0x0000000012ce0000 -> 0x0000000012d60000
Quiescing Open Firmware ...
Booting Linux via __start() @ 0x000000000e6e0000 ...

Workaround

You can use the kdump dump capture mechanism instead of fadump to prevent the issue.

Fix Outlook

The fix for this issue will be included in a later release.

I/O device impacted (if any)

N/A

[{"Type":"MASTER","Line of Business":{"code":"LOB68","label":"Power HW"},"Business Unit":{"code":"BU070","label":"IBM Infrastructure"},"Product":{"code":"SGMV168","label":"IBM Support for SUSE Linux Enterprise Server"},"ARM Category":[{"code":"a8m0z000000GnlCAAS","label":"SUSE Linux Enterprise Server"}],"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"11.0.0;12.0.0;15.0.0"}]

Document Information

Modified date:
03 March 2026

UID

ibm17261263