DENX . DULG . LinuxPostMortemAnalysis

14.3.3. Linux Post Mortem Analysis

You may find yourself in a situation where the Linux kernel crashes or hangs without any output on the console. The first attempt to get more information in such a situation is a Post Mortem dump of the log buffer - often the Linux kernel has already collected useful information in its console I/O buffer which just does not get printed because the kernel does not run until successful initialization of the console port.

Proceed as follows:

  1. Find out the virtual address of the log buffer; For 2.4 Linux kernels search for "log_buf":
    2.4 Linux:
    bash$ grep log_buf
    c0182f54 b log_buf
    Here the virtual address of the buffer is 0xC0182F54
    For 2.6 kernels "__log_buf" must be used:
    bash$ grep __log_buf
    c02124c4 b __log_buf
    Here the virtual address of the buffer is 0xC02124C4  
  2. Convert to physical address: on Power Architecture systems, the kernel is usually configured for a virtual address of kernel base (CONFIG_KERNEL_START) of 0xC0000000. Just subtract this value from the address you found. In our case we get:
    physical address = 0xC0182F54 - 0xC0000000 = 0x00182F54
  3. Reset your board - do not power-cycle it!
  4. Use your boot loader (you're running U-Boot, right?) to print a memory dump of that memory area:
    => md 0x00182F54

This whole operation is based on the assumption that your boot loader does not overwrite the RAM contents - U-Boot will take care not to destroy such valuable information.

----- Revision r1.5 - 07 Aug 2009 - 19:58 - WolfgangDenk
Copyright © 2002-2017 by DENX Software Engineering