Re: [PATCH v2 0/3] powerpc/crash: protect kdump from active watchdogs

Sourabh Jain [off-list ref] writes:

On 13/07/26 10:51, Ritesh Harjani (IBM) wrote:

quoted

Sourabh Jain [off-list ref] writes:

quoted

Changelog:
==========

v2:
  - Move H_WATCHDOG definitions to a common header for shared use
    across pseries code. 1/3
  - Added a new patch to handle pseries watchdog device registration
    failure. 2/3
  - Stop active watchdogs in crash hanlder. 3/3 Ritesh
  - Add suggested-by tag 1/3 & 3/3

Reviewed the changes and mostly looks good with some minor nits added to
the individual patches.

Small request -
Could you please also update test results with v3 in your changelog
(since you mentioned we are able to reproduce the issue easily with your
test code).

I tested this fix with the program I shared in cover letter. The watchdog
was successfully stopped even when H_WATCHDOG is called form crash
handler.  I will share my test details in v3 cover letter also.

quoted

aah one other thing I just noticed since you are ccing stable and you
added a Fixes tag in patch-3.
Patch-3 alone cannot be easily backported now due to patch-1 and
patch-2. There must be a way to define the dependencies if you are
looking for backporting the fix patch to stable tree, please check that
and follow that accordingly in v3.

I thought about that as well, but since they are part of the same patch 
series,
I assumed they would be picked together. However, I don't think that 
will work
in all cases.

I checked the older commits and noticed that a backport note was added. 
I think
we can do the same for the fix patch. I'll add a note indicating that the
following patches should be backported first:

powerpc/pseries: Move H_WATCHDOG definitions to a common header
powerpc/pseries: Handle and log pseries-wdt registration failures

Since these patches are not upstream yet, I'll refer to them by their 
commit titles.

Does that look good to you?

Documentation/process/stable-kernel-rules.rst
  Note that for a patch series, you do not have to list as prerequisites the
  patches present in the series itself. For example, if you have the following
  patch series::

    patch1
    patch2

  where patch2 depends on patch1, you do not have to list patch1 as
  prerequisite of patch2 if you have already marked patch1 for stable
  inclusion.


In that case, I think, we should mark all 3 patches for stable inclusion.

  patch 1/3   Cc: stable@vger.kernel.org
  patch 2/3   Cc: stable@vger.kernel.org
  patch 3/3   Cc: stable@vger.kernel.org
              Fixes: 69472ffa6575 ("watchdog/pseries-wdt: initial support for H_WATCHDOG-based watchdog timers")

-ritesh