Linux Documentation
 help / color / mirror / Atom feed
* [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
@ 2026-11-16 20:05 Sasha Levin
  2026-11-17  9:00 ` Greg Kroah-Hartman
                   ` (2 more replies)
  0 siblings, 3 replies; 21+ messages in thread
From: Sasha Levin @ 2026-11-16 20:05 UTC (permalink / raw)
  To: Jonathan Corbet
  Cc: Sasha Levin, Andrew Morton, Willy Tarreau, Shuah Khan,
	Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

"Describe your changes" already asks for the user-visible impact of a
change, but changelogs for fixes still tend to say what the code does,
why that's wrong and what the patch changes, and stop there. Whoever
has to decide if the fix needs backporting is left to work out what
the bug does to a running system.

Say what that means for a bug fix, in two parts. For users who aren't
looking for the bug: what a normal workload is likely to run into, and
separately the worst case and what it takes to get there. For someone
triggering it on purpose: whether an unprivileged user could do it and
what that would take, since a race that no real workload loses can
still be won by a program written to win it. Give the paragraph a
label so people can be pointed at it.

Coding assistants are told to commit a fix with a message that
describes the problem and the solution. Add the user-visible impact to
that list and point them at the section.

Suggested-by: Andrew Morton <akpm@linux-foundation.org>
Link: https://lore.kernel.org/r/20261111171055.527b4bf3da705367bc78a64c@linux-foundation.org
Signed-off-by: Sasha Levin <sashal@kernel.org>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Willy Tarreau <w@1wt.eu>
---
Changes since v1:
- Split the bug-fix paragraph into users who aren't looking for the bug
  and someone triggering it on purpose (Andrew, Willy).

v1: https://lore.kernel.org/r/20261116191436.188098-1-sashal@kernel.org

 Documentation/process/coding-assistants.rst  |  6 +++--
 Documentation/process/submitting-patches.rst | 23 ++++++++++++++++++++
 2 files changed, 27 insertions(+), 2 deletions(-)

diff --git a/Documentation/process/coding-assistants.rst b/Documentation/process/coding-assistants.rst
index 509dbf23a099b..ef01ce8d0ff91 100644
--- a/Documentation/process/coding-assistants.rst
+++ b/Documentation/process/coding-assistants.rst
@@ -63,8 +63,10 @@ these steps:
    re-running a complete analysis; drop any fix that doesn't work and try
    another one. The fix must not add build warnings and must pass the
    checkpatch.pl checks (see submitting-patches.rst).
-6. Commit the working fix with a detailed message describing the problem, the
-   solution and a Fixes tag. Do not add a Signed-off-by tag.
+6. Commit the working fix with a detailed message describing the problem, its
+   user-visible impact (see "Describe your changes" in
+   Documentation/process/submitting-patches.rst), the solution and a Fixes
+   tag. Do not add a Signed-off-by tag.
 7. Identify the maintainers and lists using scripts/get_maintainer.pl.
    Documentation/process/security-bugs.rst shows how to do that.
 8. Indicate what could not be done. If the fix could not be built or tested, or
diff --git a/Documentation/process/submitting-patches.rst b/Documentation/process/submitting-patches.rst
index 9cd19a71e96ec..e5f2181e4d663 100644
--- a/Documentation/process/submitting-patches.rst
+++ b/Documentation/process/submitting-patches.rst
@@ -51,6 +51,8 @@ motivated you to do this work.  Convince the reviewer that there is a
 problem worth fixing and that it makes sense for them to read past the
 first paragraph.
 
+.. _user_visible_impact:
+
 Describe user-visible impact.  Straight up crashes and lockups are
 pretty convincing, but not all bugs are that blatant.  Even if the
 problem was spotted during code review, describe the impact you think
@@ -61,6 +63,27 @@ from upstream, so include anything that could help route your change
 downstream: provoking circumstances, excerpts from dmesg, crash
 descriptions, performance regressions, latency spikes, lockups, etc.
 
+For a bug fix, this means saying what the bug does to a running system.
+An account of what the code does wrong is not a substitute, as it leaves
+every reader to work the consequences out again.  Cover two cases:
+
+- Users who are not looking for the bug.  Say what a normal workload is
+  likely to run into, and separately what the worst case is and what it
+  takes to get there: configuration, hardware, privileges, timing.  If
+  the bug was found by review or by a tool and has never been seen to
+  happen, say so.  If it has no effect that a user could notice, say
+  that too.
+
+- Someone triggering the bug on purpose.  Say whether an unprivileged
+  user could do it and what that would take.  A race that no normal
+  workload loses can still be won by a program that retries it on every
+  CPU for an hour.  If that makes the bug a vulnerability under
+  Documentation/process/threat-model.rst, follow
+  Documentation/process/security-bugs.rst.
+
+Maintainers and the people looking after stable trees decide from this
+whether the fix needs to be backported.
+
 Quantify optimizations and trade-offs.  If you claim improvements in
 performance, memory consumption, stack footprint, or binary size,
 include numbers that back them up.  But also describe non-obvious

base-commit: 2138148bf44fd84d3e0a863f0e8cb3627cdcc606
-- 
2.53.0


^ permalink raw reply related	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-16 20:05 [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact Sasha Levin
@ 2026-11-17  9:00 ` Greg Kroah-Hartman
  2026-11-17  9:27 ` Leon Romanovsky
  2026-11-19 10:12 ` Krzysztof Kozlowski
  2 siblings, 0 replies; 21+ messages in thread
From: Greg Kroah-Hartman @ 2026-11-17  9:00 UTC (permalink / raw)
  To: Sasha Levin
  Cc: Jonathan Corbet, Andrew Morton, Willy Tarreau, Shuah Khan,
	Randy Dunlap, Theodore Ts'o, David Hildenbrand, workflows,
	linux-doc, linux-kernel

On Mon, Nov 16, 2026 at 03:05:15PM -0500, Sasha Levin wrote:
> "Describe your changes" already asks for the user-visible impact of a
> change, but changelogs for fixes still tend to say what the code does,
> why that's wrong and what the patch changes, and stop there. Whoever
> has to decide if the fix needs backporting is left to work out what
> the bug does to a running system.
> 
> Say what that means for a bug fix, in two parts. For users who aren't
> looking for the bug: what a normal workload is likely to run into, and
> separately the worst case and what it takes to get there. For someone
> triggering it on purpose: whether an unprivileged user could do it and
> what that would take, since a race that no real workload loses can
> still be won by a program written to win it. Give the paragraph a
> label so people can be pointed at it.
> 
> Coding assistants are told to commit a fix with a message that
> describes the problem and the solution. Add the user-visible impact to
> that list and point them at the section.
> 
> Suggested-by: Andrew Morton <akpm@linux-foundation.org>
> Link: https://lore.kernel.org/r/20261111171055.527b4bf3da705367bc78a64c@linux-foundation.org
> Signed-off-by: Sasha Levin <sashal@kernel.org>
> Acked-by: Andrew Morton <akpm@linux-foundation.org>
> Reviewed-by: Willy Tarreau <w@1wt.eu>

Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-16 20:05 [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact Sasha Levin
  2026-11-17  9:00 ` Greg Kroah-Hartman
@ 2026-11-17  9:27 ` Leon Romanovsky
  2026-11-17 15:07   ` Sasha Levin
  2026-11-19 10:13   ` Krzysztof Kozlowski
  2026-11-19 10:12 ` Krzysztof Kozlowski
  2 siblings, 2 replies; 21+ messages in thread
From: Leon Romanovsky @ 2026-11-17  9:27 UTC (permalink / raw)
  To: Sasha Levin
  Cc: Jonathan Corbet, Andrew Morton, Willy Tarreau, Shuah Khan,
	Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On Mon, Nov 16, 2026 at 03:05:15PM -0500, Sasha Levin wrote:
> "Describe your changes" already asks for the user-visible impact of a
> change, but changelogs for fixes still tend to say what the code does,
> why that's wrong and what the patch changes, and stop there. Whoever
> has to decide if the fix needs backporting is left to work out what
> the bug does to a running system.
> 
> Say what that means for a bug fix, in two parts. For users who aren't
> looking for the bug: what a normal workload is likely to run into, and
> separately the worst case and what it takes to get there. For someone
> triggering it on purpose: whether an unprivileged user could do it and
> what that would take, since a race that no real workload loses can
> still be won by a program written to win it. Give the paragraph a
> label so people can be pointed at it.
> 
> Coding assistants are told to commit a fix with a message that
> describes the problem and the solution. Add the user-visible impact to
> that list and point them at the section.
> 
> Suggested-by: Andrew Morton <akpm@linux-foundation.org>
> Link: https://lore.kernel.org/r/20261111171055.527b4bf3da705367bc78a64c@linux-foundation.org
> Signed-off-by: Sasha Levin <sashal@kernel.org>
> Acked-by: Andrew Morton <akpm@linux-foundation.org>
> Reviewed-by: Willy Tarreau <w@1wt.eu>
> ---
> Changes since v1:
> - Split the bug-fix paragraph into users who aren't looking for the bug
>   and someone triggering it on purpose (Andrew, Willy).
> 
> v1: https://lore.kernel.org/r/20261116191436.188098-1-sashal@kernel.org
> 
>  Documentation/process/coding-assistants.rst  |  6 +++--
>  Documentation/process/submitting-patches.rst | 23 ++++++++++++++++++++
>  2 files changed, 27 insertions(+), 2 deletions(-)

Sasha,

Since you're already making changes in that area, could you add a sentence
asking reporters to provide the full dmesg stack trace?

Many AI-generated patches reduce it to one or two lines, which gives no
indication of whether the report is genuine or fabricated. Early in the AI
era, I saw some completely fabricated stack traces.

Thanks

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-17  9:27 ` Leon Romanovsky
@ 2026-11-17 15:07   ` Sasha Levin
  2026-11-17 15:27     ` Leon Romanovsky
  2026-11-19 10:13   ` Krzysztof Kozlowski
  1 sibling, 1 reply; 21+ messages in thread
From: Sasha Levin @ 2026-11-17 15:07 UTC (permalink / raw)
  To: Leon Romanovsky
  Cc: Jonathan Corbet, Andrew Morton, Willy Tarreau, Shuah Khan,
	Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On Tue, Nov 17, 2026 at 11:27:23AM +0200, Leon Romanovsky wrote:
>On Mon, Nov 16, 2026 at 03:05:15PM -0500, Sasha Levin wrote:
>> "Describe your changes" already asks for the user-visible impact of a
>> change, but changelogs for fixes still tend to say what the code does,
>> why that's wrong and what the patch changes, and stop there. Whoever
>> has to decide if the fix needs backporting is left to work out what
>> the bug does to a running system.
>>
>> Say what that means for a bug fix, in two parts. For users who aren't
>> looking for the bug: what a normal workload is likely to run into, and
>> separately the worst case and what it takes to get there. For someone
>> triggering it on purpose: whether an unprivileged user could do it and
>> what that would take, since a race that no real workload loses can
>> still be won by a program written to win it. Give the paragraph a
>> label so people can be pointed at it.
>>
>> Coding assistants are told to commit a fix with a message that
>> describes the problem and the solution. Add the user-visible impact to
>> that list and point them at the section.
>>
>> Suggested-by: Andrew Morton <akpm@linux-foundation.org>
>> Link: https://lore.kernel.org/r/20261111171055.527b4bf3da705367bc78a64c@linux-foundation.org
>> Signed-off-by: Sasha Levin <sashal@kernel.org>
>> Acked-by: Andrew Morton <akpm@linux-foundation.org>
>> Reviewed-by: Willy Tarreau <w@1wt.eu>
>> ---
>> Changes since v1:
>> - Split the bug-fix paragraph into users who aren't looking for the bug
>>   and someone triggering it on purpose (Andrew, Willy).
>>
>> v1: https://lore.kernel.org/r/20261116191436.188098-1-sashal@kernel.org
>>
>>  Documentation/process/coding-assistants.rst  |  6 +++--
>>  Documentation/process/submitting-patches.rst | 23 ++++++++++++++++++++
>>  2 files changed, 27 insertions(+), 2 deletions(-)
>
>Sasha,
>
>Since you're already making changes in that area, could you add a sentence
>asking reporters to provide the full dmesg stack trace?

In this doc, we already ask them not to [1] :)

>Many AI-generated patches reduce it to one or two lines, which gives no
>indication of whether the report is genuine or fabricated. Early in the AI
>era, I saw some completely fabricated stack traces.

Is it still an issue? We could add something explicit in coding-assistants.rst,
like:

   Evidence
   ========

   Every claim an AI assistant makes about the kernel's behavior at runtime,
   including backtraces, log excerpts, test results and measurements, **MUST** be
   backed by an observation from a real run.


[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/submitting-patches.rst#n760

-- 
Thanks,
Sasha

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-17 15:07   ` Sasha Levin
@ 2026-11-17 15:27     ` Leon Romanovsky
  2026-11-17 16:03       ` Sasha Levin
  2026-11-19 12:34       ` Willy Tarreau
  0 siblings, 2 replies; 21+ messages in thread
From: Leon Romanovsky @ 2026-11-17 15:27 UTC (permalink / raw)
  To: Sasha Levin
  Cc: Jonathan Corbet, Andrew Morton, Willy Tarreau, Shuah Khan,
	Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On Tue, Nov 17, 2026 at 10:07:49AM -0500, Sasha Levin wrote:
> On Tue, Nov 17, 2026 at 11:27:23AM +0200, Leon Romanovsky wrote:
> > On Mon, Nov 16, 2026 at 03:05:15PM -0500, Sasha Levin wrote:
> > > "Describe your changes" already asks for the user-visible impact of a
> > > change, but changelogs for fixes still tend to say what the code does,
> > > why that's wrong and what the patch changes, and stop there. Whoever
> > > has to decide if the fix needs backporting is left to work out what
> > > the bug does to a running system.
> > > 
> > > Say what that means for a bug fix, in two parts. For users who aren't
> > > looking for the bug: what a normal workload is likely to run into, and
> > > separately the worst case and what it takes to get there. For someone
> > > triggering it on purpose: whether an unprivileged user could do it and
> > > what that would take, since a race that no real workload loses can
> > > still be won by a program written to win it. Give the paragraph a
> > > label so people can be pointed at it.
> > > 
> > > Coding assistants are told to commit a fix with a message that
> > > describes the problem and the solution. Add the user-visible impact to
> > > that list and point them at the section.
> > > 
> > > Suggested-by: Andrew Morton <akpm@linux-foundation.org>
> > > Link: https://lore.kernel.org/r/20261111171055.527b4bf3da705367bc78a64c@linux-foundation.org
> > > Signed-off-by: Sasha Levin <sashal@kernel.org>
> > > Acked-by: Andrew Morton <akpm@linux-foundation.org>
> > > Reviewed-by: Willy Tarreau <w@1wt.eu>
> > > ---
> > > Changes since v1:
> > > - Split the bug-fix paragraph into users who aren't looking for the bug
> > >   and someone triggering it on purpose (Andrew, Willy).
> > > 
> > > v1: https://lore.kernel.org/r/20261116191436.188098-1-sashal@kernel.org
> > > 
> > >  Documentation/process/coding-assistants.rst  |  6 +++--
> > >  Documentation/process/submitting-patches.rst | 23 ++++++++++++++++++++
> > >  2 files changed, 27 insertions(+), 2 deletions(-)
> > 
> > Sasha,
> > 
> > Since you're already making changes in that area, could you add a sentence
> > asking reporters to provide the full dmesg stack trace?
> 
> In this doc, we already ask them not to [1] :)

This is really bad. You assumed they would trim the logs correctly, and I,
as the recipient was expected to trust that they were accurate. They are not.

They trim everything, which makes these logs useless to me. As I said, they
include AI-generated reports, cases with reproductions that are impossible
with upstream code, and reports for completely out-of-tree code. Everything
like that was visible in original dump stack and it is lost now.

> 
> > Many AI-generated patches reduce it to one or two lines, which gives no
> > indication of whether the report is genuine or fabricated. Early in the AI
> > era, I saw some completely fabricated stack traces.
> 
> Is it still an issue?

I don't know about faked, but useless yes, especially in the cases where
in-kernel API is involved.

> We could add something explicit in coding-assistants.rst,
> like:
> 
>   Evidence
>   ========
> 
>   Every claim an AI assistant makes about the kernel's behavior at runtime,
>   including backtraces, log excerpts, test results and measurements, **MUST** be
>   backed by an observation from a real run.

It won't help, they already overcome it by saying "Found by static
analysis tool ...".

Thanks

> 
> 
> [1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/submitting-patches.rst#n760
> 
> -- 
> Thanks,
> Sasha

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-17 15:27     ` Leon Romanovsky
@ 2026-11-17 16:03       ` Sasha Levin
  2026-11-19 12:34       ` Willy Tarreau
  1 sibling, 0 replies; 21+ messages in thread
From: Sasha Levin @ 2026-11-17 16:03 UTC (permalink / raw)
  To: Leon Romanovsky
  Cc: Sasha Levin, Borislav Petkov, Jonathan Corbet, Andrew Morton,
	Willy Tarreau, Shuah Khan, Randy Dunlap, Greg Kroah-Hartman,
	Theodore Ts'o, David Hildenbrand, workflows, linux-doc,
	linux-kernel

On Tue, Nov 17, 2026 at 05:27:21PM +0200, Leon Romanovsky wrote:
>On Tue, Nov 17, 2026 at 10:07:49AM -0500, Sasha Levin wrote:
>> On Tue, Nov 17, 2026 at 11:27:23AM +0200, Leon Romanovsky wrote:
>> > Since you're already making changes in that area, could you add a sentence
>> > asking reporters to provide the full dmesg stack trace?
>>
>> In this doc, we already ask them not to [1] :)
>
>This is really bad. You assumed they would trim the logs correctly, and I,
>as the recipient was expected to trust that they were accurate. They are not.
>
>They trim everything, which makes these logs useless to me. As I said, they
>include AI-generated reports, cases with reproductions that are impossible
>with upstream code, and reports for completely out-of-tree code. Everything
>like that was visible in original dump stack and it is lost now.

That section came in with

  78f101a1b258 ("Documentation/submitting-patches: Add blurb about backtraces in commit messages")

and the tip handbook points at it, so adding Borislav.

Borislav, see above. Leon gets fixes for tool-found bugs with the
backtrace cut down to two or three lines, the way "Backtraces in commit
messages" asks for, and can't vet the report from what's left.

--
Thanks,
Sasha

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-16 20:05 [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact Sasha Levin
  2026-11-17  9:00 ` Greg Kroah-Hartman
  2026-11-17  9:27 ` Leon Romanovsky
@ 2026-11-19 10:12 ` Krzysztof Kozlowski
  2 siblings, 0 replies; 21+ messages in thread
From: Krzysztof Kozlowski @ 2026-11-19 10:12 UTC (permalink / raw)
  To: Sasha Levin, Jonathan Corbet
  Cc: Andrew Morton, Willy Tarreau, Shuah Khan, Randy Dunlap,
	Greg Kroah-Hartman, Theodore Ts'o, David Hildenbrand,
	workflows, linux-doc, linux-kernel

On 16/11/2026 21:05, Sasha Levin wrote:
> "Describe your changes" already asks for the user-visible impact of a
> change, but changelogs for fixes still tend to say what the code does,
> why that's wrong and what the patch changes, and stop there. Whoever
> has to decide if the fix needs backporting is left to work out what
> the bug does to a running system.
> 
> Say what that means for a bug fix, in two parts. For users who aren't
> looking for the bug: what a normal workload is likely to run into, and
> separately the worst case and what it takes to get there. For someone
> triggering it on purpose: whether an unprivileged user could do it and
> what that would take, since a race that no real workload loses can
> still be won by a program written to win it. Give the paragraph a
> label so people can be pointed at it.
> 
> Coding assistants are told to commit a fix with a message that
> describes the problem and the solution. Add the user-visible impact to
> that list and point them at the section.
> 
> Suggested-by: Andrew Morton <akpm@linux-foundation.org>
> Link: https://lore.kernel.org/r/20261111171055.527b4bf3da705367bc78a64c@linux-foundation.org
> Signed-off-by: Sasha Levin <sashal@kernel.org>
> Acked-by: Andrew Morton <akpm@linux-foundation.org>
> Reviewed-by: Willy Tarreau <w@1wt.eu>
> ---


Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>

Best regards,
Krzysztof

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-17  9:27 ` Leon Romanovsky
  2026-11-17 15:07   ` Sasha Levin
@ 2026-11-19 10:13   ` Krzysztof Kozlowski
  2026-11-19 10:29     ` Geert Uytterhoeven
  2026-11-19 11:19     ` Leon Romanovsky
  1 sibling, 2 replies; 21+ messages in thread
From: Krzysztof Kozlowski @ 2026-11-19 10:13 UTC (permalink / raw)
  To: Leon Romanovsky, Sasha Levin
  Cc: Jonathan Corbet, Andrew Morton, Willy Tarreau, Shuah Khan,
	Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On 17/11/2026 10:27, Leon Romanovsky wrote:
> On Mon, Nov 16, 2026 at 03:05:15PM -0500, Sasha Levin wrote:
>> "Describe your changes" already asks for the user-visible impact of a
>> change, but changelogs for fixes still tend to say what the code does,
>> why that's wrong and what the patch changes, and stop there. Whoever
>> has to decide if the fix needs backporting is left to work out what
>> the bug does to a running system.
>>
>> Say what that means for a bug fix, in two parts. For users who aren't
>> looking for the bug: what a normal workload is likely to run into, and
>> separately the worst case and what it takes to get there. For someone
>> triggering it on purpose: whether an unprivileged user could do it and
>> what that would take, since a race that no real workload loses can
>> still be won by a program written to win it. Give the paragraph a
>> label so people can be pointed at it.
>>
>> Coding assistants are told to commit a fix with a message that
>> describes the problem and the solution. Add the user-visible impact to
>> that list and point them at the section.
>>
>> Suggested-by: Andrew Morton <akpm@linux-foundation.org>
>> Link: https://lore.kernel.org/r/20261111171055.527b4bf3da705367bc78a64c@linux-foundation.org
>> Signed-off-by: Sasha Levin <sashal@kernel.org>
>> Acked-by: Andrew Morton <akpm@linux-foundation.org>
>> Reviewed-by: Willy Tarreau <w@1wt.eu>
>> ---
>> Changes since v1:
>> - Split the bug-fix paragraph into users who aren't looking for the bug
>>   and someone triggering it on purpose (Andrew, Willy).
>>
>> v1: https://lore.kernel.org/r/20261116191436.188098-1-sashal@kernel.org
>>
>>  Documentation/process/coding-assistants.rst  |  6 +++--
>>  Documentation/process/submitting-patches.rst | 23 ++++++++++++++++++++
>>  2 files changed, 27 insertions(+), 2 deletions(-)
> 
> Sasha,
> 
> Since you're already making changes in that area, could you add a sentence
> asking reporters to provide the full dmesg stack trace?

If so, then trimmed. We do not need all addresses and registers and the
[time markers], but rather the calltrace from decode_stacktrace.sh.
That's often review feedback, btw, people just paste huge bunch of dmesg
without thinking what is useful there.

Best regards,
Krzysztof

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-19 10:13   ` Krzysztof Kozlowski
@ 2026-11-19 10:29     ` Geert Uytterhoeven
  2026-11-19 11:19       ` Leon Romanovsky
  2026-11-19 11:19     ` Leon Romanovsky
  1 sibling, 1 reply; 21+ messages in thread
From: Geert Uytterhoeven @ 2026-11-19 10:29 UTC (permalink / raw)
  To: Krzysztof Kozlowski
  Cc: Leon Romanovsky, Sasha Levin, Jonathan Corbet, Andrew Morton,
	Willy Tarreau, Shuah Khan, Randy Dunlap, Greg Kroah-Hartman,
	Theodore Ts'o, David Hildenbrand, workflows, linux-doc,
	linux-kernel

On Thu, 19 Nov 2026 at 11:14, Krzysztof Kozlowski <krzk@kernel.org> wrote:
> On 17/11/2026 10:27, Leon Romanovsky wrote:
> > Since you're already making changes in that area, could you add a sentence
> > asking reporters to provide the full dmesg stack trace?
>
> If so, then trimmed. We do not need all addresses and registers and the
> [time markers], but rather the calltrace from decode_stacktrace.sh.
> That's often review feedback, btw, people just paste huge bunch of dmesg
> without thinking what is useful there.

And full build logs, to prove their patch built fine:
https://lore.kernel.org/all/tencent_684DDE3E94F7C2470E5E02BF79AB1A1DB906@qq.com/

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-19 10:13   ` Krzysztof Kozlowski
  2026-11-19 10:29     ` Geert Uytterhoeven
@ 2026-11-19 11:19     ` Leon Romanovsky
  2026-11-19 12:04       ` Krzysztof Kozlowski
  2026-11-19 13:46       ` Mark Brown
  1 sibling, 2 replies; 21+ messages in thread
From: Leon Romanovsky @ 2026-11-19 11:19 UTC (permalink / raw)
  To: Krzysztof Kozlowski
  Cc: Sasha Levin, Jonathan Corbet, Andrew Morton, Willy Tarreau,
	Shuah Khan, Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On Thu, Nov 19, 2026 at 11:13:57AM +0100, Krzysztof Kozlowski wrote:
> On 17/11/2026 10:27, Leon Romanovsky wrote:
> > On Mon, Nov 16, 2026 at 03:05:15PM -0500, Sasha Levin wrote:
> >> "Describe your changes" already asks for the user-visible impact of a
> >> change, but changelogs for fixes still tend to say what the code does,
> >> why that's wrong and what the patch changes, and stop there. Whoever
> >> has to decide if the fix needs backporting is left to work out what
> >> the bug does to a running system.
> >>
> >> Say what that means for a bug fix, in two parts. For users who aren't
> >> looking for the bug: what a normal workload is likely to run into, and
> >> separately the worst case and what it takes to get there. For someone
> >> triggering it on purpose: whether an unprivileged user could do it and
> >> what that would take, since a race that no real workload loses can
> >> still be won by a program written to win it. Give the paragraph a
> >> label so people can be pointed at it.
> >>
> >> Coding assistants are told to commit a fix with a message that
> >> describes the problem and the solution. Add the user-visible impact to
> >> that list and point them at the section.
> >>
> >> Suggested-by: Andrew Morton <akpm@linux-foundation.org>
> >> Link: https://lore.kernel.org/r/20261111171055.527b4bf3da705367bc78a64c@linux-foundation.org
> >> Signed-off-by: Sasha Levin <sashal@kernel.org>
> >> Acked-by: Andrew Morton <akpm@linux-foundation.org>
> >> Reviewed-by: Willy Tarreau <w@1wt.eu>
> >> ---
> >> Changes since v1:
> >> - Split the bug-fix paragraph into users who aren't looking for the bug
> >>   and someone triggering it on purpose (Andrew, Willy).
> >>
> >> v1: https://lore.kernel.org/r/20261116191436.188098-1-sashal@kernel.org
> >>
> >>  Documentation/process/coding-assistants.rst  |  6 +++--
> >>  Documentation/process/submitting-patches.rst | 23 ++++++++++++++++++++
> >>  2 files changed, 27 insertions(+), 2 deletions(-)
> > 
> > Sasha,
> > 
> > Since you're already making changes in that area, could you add a sentence
> > asking reporters to provide the full dmesg stack trace?
> 
> If so, then trimmed. We do not need all addresses and registers and the
> [time markers], but rather the calltrace from decode_stacktrace.sh.

Regarding time markers, I agree. However, addresses are an important part of
debugging information. These bugs are rare, but addresses are still useful for
identifying incorrect pointer arithmetic or pointer assignments.

> That's often review feedback, btw, people just paste huge bunch of dmesg
> without thinking what is useful there.

It’s still better than AI-generated prose peppered with short, two- or
three-line stack traces intended to trick me into merging a patch. I’d
rather trim the description myself than rely on random submitters to
provide exactly the right amount of information—no more, no less.

Thanks

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-19 10:29     ` Geert Uytterhoeven
@ 2026-11-19 11:19       ` Leon Romanovsky
  0 siblings, 0 replies; 21+ messages in thread
From: Leon Romanovsky @ 2026-11-19 11:19 UTC (permalink / raw)
  To: Geert Uytterhoeven
  Cc: Krzysztof Kozlowski, Sasha Levin, Jonathan Corbet, Andrew Morton,
	Willy Tarreau, Shuah Khan, Randy Dunlap, Greg Kroah-Hartman,
	Theodore Ts'o, David Hildenbrand, workflows, linux-doc,
	linux-kernel

On Thu, Nov 19, 2026 at 11:29:36AM +0100, Geert Uytterhoeven wrote:
> On Thu, 19 Nov 2026 at 11:14, Krzysztof Kozlowski <krzk@kernel.org> wrote:
> > On 17/11/2026 10:27, Leon Romanovsky wrote:
> > > Since you're already making changes in that area, could you add a sentence
> > > asking reporters to provide the full dmesg stack trace?
> >
> > If so, then trimmed. We do not need all addresses and registers and the
> > [time markers], but rather the calltrace from decode_stacktrace.sh.
> > That's often review feedback, btw, people just paste huge bunch of dmesg
> > without thinking what is useful there.
> 
> And full build logs, to prove their patch built fine:
> https://lore.kernel.org/all/tencent_684DDE3E94F7C2470E5E02BF79AB1A1DB906@qq.com/

This is more of an anecdote than a real issue we need to worry about.

Thanks

> 
> Gr{oetje,eeting}s,
> 
>                         Geert
> 
> -- 
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
> 
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
>                                 -- Linus Torvalds

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-19 11:19     ` Leon Romanovsky
@ 2026-11-19 12:04       ` Krzysztof Kozlowski
  2026-11-19 12:36         ` Willy Tarreau
  2026-11-19 13:46       ` Mark Brown
  1 sibling, 1 reply; 21+ messages in thread
From: Krzysztof Kozlowski @ 2026-11-19 12:04 UTC (permalink / raw)
  To: Leon Romanovsky
  Cc: Sasha Levin, Jonathan Corbet, Andrew Morton, Willy Tarreau,
	Shuah Khan, Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On 19/11/2026 12:19, Leon Romanovsky wrote:
> On Thu, Nov 19, 2026 at 11:13:57AM +0100, Krzysztof Kozlowski wrote:
>>>> v1: https://lore.kernel.org/r/20261116191436.188098-1-sashal@kernel.org
>>>>
>>>>  Documentation/process/coding-assistants.rst  |  6 +++--
>>>>  Documentation/process/submitting-patches.rst | 23 ++++++++++++++++++++
>>>>  2 files changed, 27 insertions(+), 2 deletions(-)
>>>
>>> Sasha,
>>>
>>> Since you're already making changes in that area, could you add a sentence
>>> asking reporters to provide the full dmesg stack trace?
>>
>> If so, then trimmed. We do not need all addresses and registers and the
>> [time markers], but rather the calltrace from decode_stacktrace.sh.
> 
> Regarding time markers, I agree. However, addresses are an important part of
> debugging information. These bugs are rare, but addresses are still useful for
> identifying incorrect pointer arithmetic or pointer assignments.

Depending on the case, e.g. some driver lockdep reports don't need them.
NULL ptr usually as well. The point is that dmesg should be trimmed to
relevant and useful information. If that includes addresses for that
particular case, then of course these should stay. But 75% of data here
in commit like 5b4bd13e0552926648a46c2c9a12a24dfaee3715 is not really
helping. In the same time undecoded call trace is useless, because these
addresses depends on the build. The dmesg should simply be the decoded
stacktrace ONLY for this case.

Best regards,
Krzysztof

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-17 15:27     ` Leon Romanovsky
  2026-11-17 16:03       ` Sasha Levin
@ 2026-11-19 12:34       ` Willy Tarreau
  2026-11-19 12:59         ` Leon Romanovsky
  1 sibling, 1 reply; 21+ messages in thread
From: Willy Tarreau @ 2026-11-19 12:34 UTC (permalink / raw)
  To: Leon Romanovsky
  Cc: Sasha Levin, Jonathan Corbet, Andrew Morton, Shuah Khan,
	Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On Tue, Nov 17, 2026 at 05:27:21PM +0200, Leon Romanovsky wrote:
> > > Since you're already making changes in that area, could you add a sentence
> > > asking reporters to provide the full dmesg stack trace?
> > 
> > In this doc, we already ask them not to [1] :)
> 
> This is really bad. You assumed they would trim the logs correctly, and I,
> as the recipient was expected to trust that they were accurate. They are not.
> 
> They trim everything, which makes these logs useless to me. As I said, they
> include AI-generated reports, cases with reproductions that are impossible
> with upstream code, and reports for completely out-of-tree code. Everything
> like that was visible in original dump stack and it is lost now.
> 
> > 
> > > Many AI-generated patches reduce it to one or two lines, which gives no
> > > indication of whether the report is genuine or fabricated. Early in the AI
> > > era, I saw some completely fabricated stack traces.
> > 
> > Is it still an issue?
> 
> I don't know about faked, but useless yes, especially in the cases where
> in-kernel API is involved.

The situation completely changed over the last year. LLMs that would
hallucinate bugs and dmesg in single-turn conversations have been
replaced with agents/harnesses calling shell commands and able to
capture contents without inventing them. I agree that we should ask
them to capture such dmesg which nowadays have no reason to be faked
anymore.

Willy

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-19 12:04       ` Krzysztof Kozlowski
@ 2026-11-19 12:36         ` Willy Tarreau
  2026-11-19 12:54           ` Leon Romanovsky
  0 siblings, 1 reply; 21+ messages in thread
From: Willy Tarreau @ 2026-11-19 12:36 UTC (permalink / raw)
  To: Krzysztof Kozlowski
  Cc: Leon Romanovsky, Sasha Levin, Jonathan Corbet, Andrew Morton,
	Shuah Khan, Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On Thu, Nov 19, 2026 at 01:04:32PM +0100, Krzysztof Kozlowski wrote:
> On 19/11/2026 12:19, Leon Romanovsky wrote:
> > On Thu, Nov 19, 2026 at 11:13:57AM +0100, Krzysztof Kozlowski wrote:
> >>>> v1: https://lore.kernel.org/r/20261116191436.188098-1-sashal@kernel.org
> >>>>
> >>>>  Documentation/process/coding-assistants.rst  |  6 +++--
> >>>>  Documentation/process/submitting-patches.rst | 23 ++++++++++++++++++++
> >>>>  2 files changed, 27 insertions(+), 2 deletions(-)
> >>>
> >>> Sasha,
> >>>
> >>> Since you're already making changes in that area, could you add a sentence
> >>> asking reporters to provide the full dmesg stack trace?
> >>
> >> If so, then trimmed. We do not need all addresses and registers and the
> >> [time markers], but rather the calltrace from decode_stacktrace.sh.
> > 
> > Regarding time markers, I agree. However, addresses are an important part of
> > debugging information. These bugs are rare, but addresses are still useful for
> > identifying incorrect pointer arithmetic or pointer assignments.
> 
> Depending on the case, e.g. some driver lockdep reports don't need them.
> NULL ptr usually as well. The point is that dmesg should be trimmed to
> relevant and useful information. If that includes addresses for that
> particular case, then of course these should stay. But 75% of data here
> in commit like 5b4bd13e0552926648a46c2c9a12a24dfaee3715 is not really
> helping. In the same time undecoded call trace is useless, because these
> addresses depends on the build. The dmesg should simply be the decoded
> stacktrace ONLY for this case.

Then just the 2-3 lines capture inline, and the full log attached could
please everyone ?

Willy

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-19 12:36         ` Willy Tarreau
@ 2026-11-19 12:54           ` Leon Romanovsky
  2026-11-19 12:58             ` Willy Tarreau
  2026-11-19 13:16             ` Krzysztof Kozlowski
  0 siblings, 2 replies; 21+ messages in thread
From: Leon Romanovsky @ 2026-11-19 12:54 UTC (permalink / raw)
  To: Willy Tarreau
  Cc: Krzysztof Kozlowski, Sasha Levin, Jonathan Corbet, Andrew Morton,
	Shuah Khan, Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On Thu, Nov 19, 2026 at 01:36:51PM +0100, Willy Tarreau wrote:
> On Thu, Nov 19, 2026 at 01:04:32PM +0100, Krzysztof Kozlowski wrote:
> > On 19/11/2026 12:19, Leon Romanovsky wrote:
> > > On Thu, Nov 19, 2026 at 11:13:57AM +0100, Krzysztof Kozlowski wrote:
> > >>>> v1: https://lore.kernel.org/r/20261116191436.188098-1-sashal@kernel.org
> > >>>>
> > >>>>  Documentation/process/coding-assistants.rst  |  6 +++--
> > >>>>  Documentation/process/submitting-patches.rst | 23 ++++++++++++++++++++
> > >>>>  2 files changed, 27 insertions(+), 2 deletions(-)
> > >>>
> > >>> Sasha,
> > >>>
> > >>> Since you're already making changes in that area, could you add a sentence
> > >>> asking reporters to provide the full dmesg stack trace?
> > >>
> > >> If so, then trimmed. We do not need all addresses and registers and the
> > >> [time markers], but rather the calltrace from decode_stacktrace.sh.
> > > 
> > > Regarding time markers, I agree. However, addresses are an important part of
> > > debugging information. These bugs are rare, but addresses are still useful for
> > > identifying incorrect pointer arithmetic or pointer assignments.
> > 
> > Depending on the case, e.g. some driver lockdep reports don't need them.
> > NULL ptr usually as well. The point is that dmesg should be trimmed to
> > relevant and useful information. If that includes addresses for that
> > particular case, then of course these should stay. But 75% of data here
> > in commit like 5b4bd13e0552926648a46c2c9a12a24dfaee3715 is not really
> > helping. In the same time undecoded call trace is useless, because these
> > addresses depends on the build. The dmesg should simply be the decoded
> > stacktrace ONLY for this case.
> 
> Then just the 2-3 lines capture inline, and the full log attached could
> please everyone ?

I don't think so, because this information is also important to preserve in the
Git history. I received offline feedback that the full stack is useful for
backporting and customer support, too. Sometimes the dmesg output is similar
enough to recognize, but differs in two or three lines.

I welcome AI-generated patches, but they should not be trusted blindly.
I also don't believe AI can reliably trim stack traces.

Thanks

> 
> Willy

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-19 12:54           ` Leon Romanovsky
@ 2026-11-19 12:58             ` Willy Tarreau
  2026-11-19 13:16             ` Krzysztof Kozlowski
  1 sibling, 0 replies; 21+ messages in thread
From: Willy Tarreau @ 2026-11-19 12:58 UTC (permalink / raw)
  To: Leon Romanovsky
  Cc: Krzysztof Kozlowski, Sasha Levin, Jonathan Corbet, Andrew Morton,
	Shuah Khan, Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On Thu, Nov 19, 2026 at 02:54:50PM +0200, Leon Romanovsky wrote:
> On Thu, Nov 19, 2026 at 01:36:51PM +0100, Willy Tarreau wrote:
> > On Thu, Nov 19, 2026 at 01:04:32PM +0100, Krzysztof Kozlowski wrote:
(...)
> > > Depending on the case, e.g. some driver lockdep reports don't need them.
> > > NULL ptr usually as well. The point is that dmesg should be trimmed to
> > > relevant and useful information. If that includes addresses for that
> > > particular case, then of course these should stay. But 75% of data here
> > > in commit like 5b4bd13e0552926648a46c2c9a12a24dfaee3715 is not really
> > > helping. In the same time undecoded call trace is useless, because these
> > > addresses depends on the build. The dmesg should simply be the decoded
> > > stacktrace ONLY for this case.
> > 
> > Then just the 2-3 lines capture inline, and the full log attached could
> > please everyone ?
> 
> I don't think so, because this information is also important to preserve in the
> Git history. I received offline feedback that the full stack is useful for
> backporting and customer support, too. Sometimes the dmesg output is similar
> enough to recognize, but differs in two or three lines.

Good point and several times as a user facing an bugs I've looked for
messages in history as well before asking for help.

> I welcome AI-generated patches, but they should not be trusted blindly.
> I also don't believe AI can reliably trim stack traces.

Never trusted indeed, however running "sed -n first,last,p" is now quite
common and reliable since the LLM generates the extraction/trimming command,
not the contents anymore.

Willy

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-19 12:34       ` Willy Tarreau
@ 2026-11-19 12:59         ` Leon Romanovsky
  2026-11-19 13:01           ` Willy Tarreau
  0 siblings, 1 reply; 21+ messages in thread
From: Leon Romanovsky @ 2026-11-19 12:59 UTC (permalink / raw)
  To: Willy Tarreau
  Cc: Sasha Levin, Jonathan Corbet, Andrew Morton, Shuah Khan,
	Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On Thu, Nov 19, 2026 at 01:34:45PM +0100, Willy Tarreau wrote:
> On Tue, Nov 17, 2026 at 05:27:21PM +0200, Leon Romanovsky wrote:
> > > > Since you're already making changes in that area, could you add a sentence
> > > > asking reporters to provide the full dmesg stack trace?
> > > 
> > > In this doc, we already ask them not to [1] :)
> > 
> > This is really bad. You assumed they would trim the logs correctly, and I,
> > as the recipient was expected to trust that they were accurate. They are not.
> > 
> > They trim everything, which makes these logs useless to me. As I said, they
> > include AI-generated reports, cases with reproductions that are impossible
> > with upstream code, and reports for completely out-of-tree code. Everything
> > like that was visible in original dump stack and it is lost now.
> > 
> > > 
> > > > Many AI-generated patches reduce it to one or two lines, which gives no
> > > > indication of whether the report is genuine or fabricated. Early in the AI
> > > > era, I saw some completely fabricated stack traces.
> > > 
> > > Is it still an issue?
> > 
> > I don't know about faked, but useless yes, especially in the cases where
> > in-kernel API is involved.
> 
> The situation completely changed over the last year. LLMs that would
> hallucinate bugs and dmesg in single-turn conversations have been
> replaced with agents/harnesses calling shell commands and able to
> capture contents without inventing them. I agree that we should ask
> them to capture such dmesg which nowadays have no reason to be faked
> anymore.

The meaning of “faked” has shifted. As you said, it previously referred to a
completely imaginary call stack, now it refers to a custom module that abuses
the in-kernel API. The latter is still “faked” because it cannot be done with
the current code and unlikely will be possible in the future.

Thanks

> 
> Willy

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-19 12:59         ` Leon Romanovsky
@ 2026-11-19 13:01           ` Willy Tarreau
  0 siblings, 0 replies; 21+ messages in thread
From: Willy Tarreau @ 2026-11-19 13:01 UTC (permalink / raw)
  To: Leon Romanovsky
  Cc: Sasha Levin, Jonathan Corbet, Andrew Morton, Shuah Khan,
	Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On Thu, Nov 19, 2026 at 02:59:47PM +0200, Leon Romanovsky wrote:
> On Thu, Nov 19, 2026 at 01:34:45PM +0100, Willy Tarreau wrote:
> > On Tue, Nov 17, 2026 at 05:27:21PM +0200, Leon Romanovsky wrote:
> > > > > Since you're already making changes in that area, could you add a sentence
> > > > > asking reporters to provide the full dmesg stack trace?
> > > > 
> > > > In this doc, we already ask them not to [1] :)
> > > 
> > > This is really bad. You assumed they would trim the logs correctly, and I,
> > > as the recipient was expected to trust that they were accurate. They are not.
> > > 
> > > They trim everything, which makes these logs useless to me. As I said, they
> > > include AI-generated reports, cases with reproductions that are impossible
> > > with upstream code, and reports for completely out-of-tree code. Everything
> > > like that was visible in original dump stack and it is lost now.
> > > 
> > > > 
> > > > > Many AI-generated patches reduce it to one or two lines, which gives no
> > > > > indication of whether the report is genuine or fabricated. Early in the AI
> > > > > era, I saw some completely fabricated stack traces.
> > > > 
> > > > Is it still an issue?
> > > 
> > > I don't know about faked, but useless yes, especially in the cases where
> > > in-kernel API is involved.
> > 
> > The situation completely changed over the last year. LLMs that would
> > hallucinate bugs and dmesg in single-turn conversations have been
> > replaced with agents/harnesses calling shell commands and able to
> > capture contents without inventing them. I agree that we should ask
> > them to capture such dmesg which nowadays have no reason to be faked
> > anymore.
> 
> The meaning of "faked" has shifted. As you said, it previously referred to a
> completely imaginary call stack, now it refers to a custom module that abuses
> the in-kernel API. The latter is still "faked" because it cannot be done with
> the current code and unlikely will be possible in the future.

Ah OK we were indeed not speaking about the same cases.

Willy

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-19 12:54           ` Leon Romanovsky
  2026-11-19 12:58             ` Willy Tarreau
@ 2026-11-19 13:16             ` Krzysztof Kozlowski
  2026-11-19 14:07               ` Leon Romanovsky
  1 sibling, 1 reply; 21+ messages in thread
From: Krzysztof Kozlowski @ 2026-11-19 13:16 UTC (permalink / raw)
  To: Leon Romanovsky, Willy Tarreau
  Cc: Sasha Levin, Jonathan Corbet, Andrew Morton, Shuah Khan,
	Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On 19/11/2026 13:54, Leon Romanovsky wrote:
> On Thu, Nov 19, 2026 at 01:36:51PM +0100, Willy Tarreau wrote:
>> On Thu, Nov 19, 2026 at 01:04:32PM +0100, Krzysztof Kozlowski wrote:
>>> On 19/11/2026 12:19, Leon Romanovsky wrote:
>>>> On Thu, Nov 19, 2026 at 11:13:57AM +0100, Krzysztof Kozlowski wrote:
>>>>>>> v1: https://lore.kernel.org/r/20261116191436.188098-1-sashal@kernel.org
>>>>>>>
>>>>>>>  Documentation/process/coding-assistants.rst  |  6 +++--
>>>>>>>  Documentation/process/submitting-patches.rst | 23 ++++++++++++++++++++
>>>>>>>  2 files changed, 27 insertions(+), 2 deletions(-)
>>>>>>
>>>>>> Sasha,
>>>>>>
>>>>>> Since you're already making changes in that area, could you add a sentence
>>>>>> asking reporters to provide the full dmesg stack trace?
>>>>>
>>>>> If so, then trimmed. We do not need all addresses and registers and the
>>>>> [time markers], but rather the calltrace from decode_stacktrace.sh.
>>>>
>>>> Regarding time markers, I agree. However, addresses are an important part of
>>>> debugging information. These bugs are rare, but addresses are still useful for
>>>> identifying incorrect pointer arithmetic or pointer assignments.
>>>
>>> Depending on the case, e.g. some driver lockdep reports don't need them.
>>> NULL ptr usually as well. The point is that dmesg should be trimmed to
>>> relevant and useful information. If that includes addresses for that
>>> particular case, then of course these should stay. But 75% of data here
>>> in commit like 5b4bd13e0552926648a46c2c9a12a24dfaee3715 is not really
>>> helping. In the same time undecoded call trace is useless, because these
>>> addresses depends on the build. The dmesg should simply be the decoded
>>> stacktrace ONLY for this case.
>>
>> Then just the 2-3 lines capture inline, and the full log attached could
>> please everyone ?
> 
> I don't think so, because this information is also important to preserve in the
> Git history. I received offline feedback that the full stack is useful for
> backporting and customer support, too. Sometimes the dmesg output is similar

Just like we do not mark commits as fixing CVEs/security issues, which
plenty of downstream distros would like us to do, we do not write commit
msgs for backporting or customer support. Just look at that commit I
pointed out - 5b4bd13e0552926648a46c2c9a12a24dfaee3715 - the addresses
are making things less readable and the readability is important.

If you need something for customers, then that is what Link tag is for.
It can point to your bugzilla or email report with everything.



Best regards,
Krzysztof

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-19 11:19     ` Leon Romanovsky
  2026-11-19 12:04       ` Krzysztof Kozlowski
@ 2026-11-19 13:46       ` Mark Brown
  1 sibling, 0 replies; 21+ messages in thread
From: Mark Brown @ 2026-11-19 13:46 UTC (permalink / raw)
  To: Leon Romanovsky
  Cc: Krzysztof Kozlowski, Sasha Levin, Jonathan Corbet, Andrew Morton,
	Willy Tarreau, Shuah Khan, Randy Dunlap, Greg Kroah-Hartman,
	Theodore Ts'o, David Hildenbrand, workflows, linux-doc,
	linux-kernel

[-- Attachment #1: Type: text/plain, Size: 920 bytes --]

On Thu, Nov 19, 2026 at 01:19:02PM +0200, Leon Romanovsky wrote:
> On Thu, Nov 19, 2026 at 11:13:57AM +0100, Krzysztof Kozlowski wrote:

> > That's often review feedback, btw, people just paste huge bunch of dmesg
> > without thinking what is useful there.

> It’s still better than AI-generated prose peppered with short, two- or
> three-line stack traces intended to trick me into merging a patch. I’d
> rather trim the description myself than rely on random submitters to
> provide exactly the right amount of information—no more, no less.

Honestly, it's generally just as bad - it's usually multiple screenfuls
of irrelevant noise which doesn't visually fit the format for a
changelog.  Half the time people put them in the middle of the changelog
which doesn't help anything.  With the generated changelogs at least you
know to skim read the whole thing and not trust a word that's in there.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact
  2026-11-19 13:16             ` Krzysztof Kozlowski
@ 2026-11-19 14:07               ` Leon Romanovsky
  0 siblings, 0 replies; 21+ messages in thread
From: Leon Romanovsky @ 2026-11-19 14:07 UTC (permalink / raw)
  To: Krzysztof Kozlowski
  Cc: Willy Tarreau, Sasha Levin, Jonathan Corbet, Andrew Morton,
	Shuah Khan, Randy Dunlap, Greg Kroah-Hartman, Theodore Ts'o,
	David Hildenbrand, workflows, linux-doc, linux-kernel

On Thu, Nov 19, 2026 at 02:16:30PM +0100, Krzysztof Kozlowski wrote:
> On 19/11/2026 13:54, Leon Romanovsky wrote:
> > On Thu, Nov 19, 2026 at 01:36:51PM +0100, Willy Tarreau wrote:
> >> On Thu, Nov 19, 2026 at 01:04:32PM +0100, Krzysztof Kozlowski wrote:
> >>> On 19/11/2026 12:19, Leon Romanovsky wrote:
> >>>> On Thu, Nov 19, 2026 at 11:13:57AM +0100, Krzysztof Kozlowski wrote:
> >>>>>>> v1: https://lore.kernel.org/r/20261116191436.188098-1-sashal@kernel.org
> >>>>>>>
> >>>>>>>  Documentation/process/coding-assistants.rst  |  6 +++--
> >>>>>>>  Documentation/process/submitting-patches.rst | 23 ++++++++++++++++++++
> >>>>>>>  2 files changed, 27 insertions(+), 2 deletions(-)
> >>>>>>
> >>>>>> Sasha,
> >>>>>>
> >>>>>> Since you're already making changes in that area, could you add a sentence
> >>>>>> asking reporters to provide the full dmesg stack trace?
> >>>>>
> >>>>> If so, then trimmed. We do not need all addresses and registers and the
> >>>>> [time markers], but rather the calltrace from decode_stacktrace.sh.
> >>>>
> >>>> Regarding time markers, I agree. However, addresses are an important part of
> >>>> debugging information. These bugs are rare, but addresses are still useful for
> >>>> identifying incorrect pointer arithmetic or pointer assignments.
> >>>
> >>> Depending on the case, e.g. some driver lockdep reports don't need them.
> >>> NULL ptr usually as well. The point is that dmesg should be trimmed to
> >>> relevant and useful information. If that includes addresses for that
> >>> particular case, then of course these should stay. But 75% of data here
> >>> in commit like 5b4bd13e0552926648a46c2c9a12a24dfaee3715 is not really
> >>> helping. In the same time undecoded call trace is useless, because these
> >>> addresses depends on the build. The dmesg should simply be the decoded
> >>> stacktrace ONLY for this case.
> >>
> >> Then just the 2-3 lines capture inline, and the full log attached could
> >> please everyone ?
> > 
> > I don't think so, because this information is also important to preserve in the
> > Git history. I received offline feedback that the full stack is useful for
> > backporting and customer support, too. Sometimes the dmesg output is similar
> 
> Just like we do not mark commits as fixing CVEs/security issues, which
> plenty of downstream distros would like us to do, we do not write commit
> msgs for backporting or customer support. Just look at that commit I
> pointed out - 5b4bd13e0552926648a46c2c9a12a24dfaee3715 - the addresses
> are making things less readable and the readability is important.

I looked, but nothing stood out as distracting. I've been reading these
stacks for nearly 20 years.

> 
> If you need something for customers, then that is what Link tag is for.
> It can point to your bugzilla or email report with everything.

You missed my point, I'm not asking customers to dig into the kernel. An FAE
or customer comes to me with a crash, and I need to determine whether it's
a known issue (and point them to the relevant patch) or a different one.

Thanks

> 
> 
> 
> Best regards,
> Krzysztof
> 

^ permalink raw reply	[flat|nested] 21+ messages in thread

end of thread, other threads:[~2026-11-19 14:07 UTC | newest]

Thread overview: 21+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-16 20:05 [PATCH v2] docs: submitting-patches: ask bug fixes for their user-visible impact Sasha Levin
2026-11-17  9:00 ` Greg Kroah-Hartman
2026-11-17  9:27 ` Leon Romanovsky
2026-11-17 15:07   ` Sasha Levin
2026-11-17 15:27     ` Leon Romanovsky
2026-11-17 16:03       ` Sasha Levin
2026-11-19 12:34       ` Willy Tarreau
2026-11-19 12:59         ` Leon Romanovsky
2026-11-19 13:01           ` Willy Tarreau
2026-11-19 10:13   ` Krzysztof Kozlowski
2026-11-19 10:29     ` Geert Uytterhoeven
2026-11-19 11:19       ` Leon Romanovsky
2026-11-19 11:19     ` Leon Romanovsky
2026-11-19 12:04       ` Krzysztof Kozlowski
2026-11-19 12:36         ` Willy Tarreau
2026-11-19 12:54           ` Leon Romanovsky
2026-11-19 12:58             ` Willy Tarreau
2026-11-19 13:16             ` Krzysztof Kozlowski
2026-11-19 14:07               ` Leon Romanovsky
2026-11-19 13:46       ` Mark Brown
2026-11-19 10:12 ` Krzysztof Kozlowski

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox