docs: admin-guide: for kernel bugs refer to other kernel documentation

The current section 'If something goes wrong' makes a number of suggestions
for debugging, bug hunting and reporting issues, which are quite briefly
described in that section.

However, the suggestions are also well covered in other kernel
documentation or sometimes simply outdated. Here, each suggestion in that
section is summarized, and then followed with its assessment, and the
derived action for each suggestion:

  - use MAINTAINERS and mailing list: covered in 'Reporting issues',
    summarized in the short guide, detailed in its further section.
    Reporting issues even provides some specific examples that guides
    readers well through the needed steps. Refer to 'Reporting issues'.

  - contact Linus Torvalds: probably outdated as currently described.
    nevertheless covered in 'Reporting issues'. Reporting issues points out
    to contact the relevant kernel maintainers first, and after some
    patience and failed attempts with those maintainers, contacting Linus
    Torvalds might be okay. Refer to 'Reporting issues'.

  - tell what kernel, how to duplicate, the setup, if the problem is new
    or old and when did you notice: covered in 'Reporting issues',
    especially in Step-by-step guide how to report issues to the kernel
    maintainers. Refer to 'Reporting issues'.

  - duplicate kernel bug reports exactly: covered in 'Reporting issues',
    especially in Write and send the report. Refer to 'Reporting issues'.

  - read 'Bug hunting': keep this reference. Refer to 'Bug hunting'.

  - compile the kernel with CONFIG_KALLSYMS: covered in 'Reporting issues',
    especially in Decode failure messages. Refer to 'Reporting issues'.

  - alternatively, use ksymoops: ksymoops at the mentioned URL seems not to
    be maintained anymore. It was released roughly once a year until
    version 2.4.11 in 2005, but has not seen a new release since then. The
    information in ./scripts/ksymoops/README is from 1999, and does not
    give more insight on its actual maintenance state either. Ksymoops is
    mentioned as system utility in changes.rst, but also not recommended
    there. Drop the explanation on using ksymoops.

  - alternatively, lookup dump manually with the EIP and nm to determine
    the function in which the kernel crashes: this method seems already a
    quite advanced and low-level debugging method. Even all the further
    references on bug hunting and debugging do not mention it. Drop this
    alternative method and limit mentioning methods explained in the other
    existing kernel documentation.

  - read 'Reporting issues': keep this reference.
    Refer to 'Reporting issues'.

  - use gdb for debugging: some specific details, e.g., edit
    arch/x86/Makefile, are probably outdated or limited to one (historic
    important) setup. Using gdb is covered in 'Bug hunting', 'Debugging
    kernel and modules via gdb' and 'Using kgdb, kdb and the kernel
    debugger internals'. Refer to those three documents.

Overall, it is sufficient to refer to reporting-issues.rst,
bug-hunting.rst, gdb-kernel-debugging.rst and kgdb.rst and this way cover
the existing suggestions.

'Reporting issues' is quite new and probably up to date. 'Bug hunting',
'Debugging kernel and modules via gdb' and 'Using kgdb, kdb and the kernel
debugger internals' might need some revisit and update, but they are
generally in an acceptable state for referring to them.

Replace the existing suggestions by reference to other existing kernel
documentation covering those suggestions---partly even nicely summarized
and then explained in greater detail.

Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Link: https://lore.kernel.org/r/20220720041325.15693-3-lukas.bulwahn@gmail.com
Signed-off-by: Jonathan Corbet <corbet@lwn.net>

0 files changed, 0 insertions, 0 deletions