In commit b5325b2a270f ("coredump: hand a pidfd to the usermode coredump
helper") a new core_pattern specifier, %F, was added to provide a pidfs
to the usermode helper process referring to the crashed process.

Update the documentation to include the new core_pattern specifier.

Link: https://github.com/systemd/systemd/pull/37125
Link: https://lwn.net/Articles/1024160/
Signed-off-by: Salvatore Bonaccorso <carnil@debian.org>
---
 Documentation/admin-guide/sysctl/kernel.rst | 1 +
 1 file changed, 1 insertion(+)

diff --git a/Documentation/admin-guide/sysctl/kernel.rst b/Documentation/admin-guide/sysctl/kernel.rst
index dd49a89a62d3..f1b2ab219a08 100644
--- a/Documentation/admin-guide/sysctl/kernel.rst
+++ b/Documentation/admin-guide/sysctl/kernel.rst
@@ -177,6 +177,7 @@ core_pattern
 	%E		executable path
 	%c		maximum size of core file by resource limit RLIMIT_CORE
 	%C		CPU the task ran on
+	%F		pidfd number
 	%<OTHER>	both are dropped
 	========	==========================================
 
-- 
2.49.0

On Thu, Jun 12, 2025 at 08:02:04AM +0200, Salvatore Bonaccorso wrote:
> In commit b5325b2a270f ("coredump: hand a pidfd to the usermode coredump
> helper") a new core_pattern specifier, %F, was added to provide a pidfs
> to the usermode helper process referring to the crashed process.
> 
> Update the documentation to include the new core_pattern specifier.
> 
> Link: https://github.com/systemd/systemd/pull/37125
> Link: https://lwn.net/Articles/1024160/
> Signed-off-by: Salvatore Bonaccorso <carnil@debian.org>
> ---

Thanks!

Just throwing a random thought out there: It would be nice if we could
optionally create a connection between a source code path/file and the
documentation directory/file in Documentation. For example, coredump
stuff is documented in Documentation/admin-guide/sysctl/kernel.rst so
there could be a line to that effect in fs/coredump.c and it creates a
warning if that file is removed.

Maybe that's dumb but it certainly would help to not miss such updates
in the future.

Christian

>  Documentation/admin-guide/sysctl/kernel.rst | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/Documentation/admin-guide/sysctl/kernel.rst b/Documentation/admin-guide/sysctl/kernel.rst
> index dd49a89a62d3..f1b2ab219a08 100644
> --- a/Documentation/admin-guide/sysctl/kernel.rst
> +++ b/Documentation/admin-guide/sysctl/kernel.rst
> @@ -177,6 +177,7 @@ core_pattern
>  	%E		executable path
>  	%c		maximum size of core file by resource limit RLIMIT_CORE
>  	%C		CPU the task ran on
> +	%F		pidfd number
>  	%<OTHER>	both are dropped
>  	========	==========================================
>  
> -- 
> 2.49.0
>

Salvatore Bonaccorso <carnil@debian.org> writes:

> In commit b5325b2a270f ("coredump: hand a pidfd to the usermode coredump
> helper") a new core_pattern specifier, %F, was added to provide a pidfs
> to the usermode helper process referring to the crashed process.
>
> Update the documentation to include the new core_pattern specifier.
>
> Link: https://github.com/systemd/systemd/pull/37125
> Link: https://lwn.net/Articles/1024160/
> Signed-off-by: Salvatore Bonaccorso <carnil@debian.org>
> ---
>  Documentation/admin-guide/sysctl/kernel.rst | 1 +
>  1 file changed, 1 insertion(+)

Applied, thanks.

> FWIW, I was not sure if I should add a 'Fixes: b5325b2a270f
> ("coredump: hand a pidfd to the usermode coredump helper") to make
> sure that it get backported as well to the series where a backport of
> b5325b2a270f landed.

A Fixes tag is not sufficient to ensure that a patch is backported to
previous kernels - you must CC stable@vger.kernel.org to be sure.  In
this case, it's a documentation tweak for a 6.16 change, so I don't
think there is a lot of urgency here.

Thanks,

jon

Hi Jonathan,

On Sat, Jun 21, 2025 at 02:18:38PM -0600, Jonathan Corbet wrote:
> Salvatore Bonaccorso <carnil@debian.org> writes:
> 
> > In commit b5325b2a270f ("coredump: hand a pidfd to the usermode coredump
> > helper") a new core_pattern specifier, %F, was added to provide a pidfs
> > to the usermode helper process referring to the crashed process.
> >
> > Update the documentation to include the new core_pattern specifier.
> >
> > Link: https://github.com/systemd/systemd/pull/37125
> > Link: https://lwn.net/Articles/1024160/
> > Signed-off-by: Salvatore Bonaccorso <carnil@debian.org>
> > ---
> >  Documentation/admin-guide/sysctl/kernel.rst | 1 +
> >  1 file changed, 1 insertion(+)
> 
> Applied, thanks.

Thank you!

> > FWIW, I was not sure if I should add a 'Fixes: b5325b2a270f
> > ("coredump: hand a pidfd to the usermode coredump helper") to make
> > sure that it get backported as well to the series where a backport of
> > b5325b2a270f landed.
> 
> A Fixes tag is not sufficient to ensure that a patch is backported to
> previous kernels - you must CC stable@vger.kernel.org to be sure.  In
> this case, it's a documentation tweak for a 6.16 change, so I don't
> think there is a lot of urgency here.

Ack (apart the change did land in variou stable series, i.e.
b5325b2a270f ("coredump: hand a pidfd to the usermode coredump
helper") did land in 6.16-rc1 *and* the stable versions v6.15.1,
v6.14.10, v6.12.32, v6.6.93, v6.1.141, v5.15.185, v5.10.238 and
v5.4.294.

You are right it is not urgent, would be more to have documentation
consistent as well in those versions.

Regards,
Salvatore

Hi,

On Thu, Jun 12, 2025 at 08:02:04AM +0200, Salvatore Bonaccorso wrote:
> In commit b5325b2a270f ("coredump: hand a pidfd to the usermode coredump
> helper") a new core_pattern specifier, %F, was added to provide a pidfs
> to the usermode helper process referring to the crashed process.
> 
> Update the documentation to include the new core_pattern specifier.
> 
> Link: https://github.com/systemd/systemd/pull/37125
> Link: https://lwn.net/Articles/1024160/
> Signed-off-by: Salvatore Bonaccorso <carnil@debian.org>
> ---
>  Documentation/admin-guide/sysctl/kernel.rst | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/Documentation/admin-guide/sysctl/kernel.rst b/Documentation/admin-guide/sysctl/kernel.rst
> index dd49a89a62d3..f1b2ab219a08 100644
> --- a/Documentation/admin-guide/sysctl/kernel.rst
> +++ b/Documentation/admin-guide/sysctl/kernel.rst
> @@ -177,6 +177,7 @@ core_pattern
>  	%E		executable path
>  	%c		maximum size of core file by resource limit RLIMIT_CORE
>  	%C		CPU the task ran on
> +	%F		pidfd number
>  	%<OTHER>	both are dropped
>  	========	==========================================

FWIW, I was not sure if I should add a 'Fixes: b5325b2a270f
("coredump: hand a pidfd to the usermode coredump helper") to make
sure that it get backported as well to the series where a backport of
b5325b2a270f landed.

Regards,
Salvatore