Message-ID: <20250425-erbschaft-nummer-8ddbe420ae22@brauner>
Date: Fri, 25 Apr 2025 18:49:19 +0200
From: Christian Brauner <brauner@...nel.org>
To: Benjamin Drung <benjamin.drung@...onical.com>
Cc: linux-fsdevel@...r.kernel.org, Oleg Nesterov <oleg@...hat.com>, 
	Luca Boccassi <luca.boccassi@...il.com>, Lennart Poettering <lennart@...ttering.net>, 
	Daan De Meyer <daan.j.demeyer@...il.com>, Mike Yuan <me@...dnzj.com>, 
	Zbigniew Jędrzejewski-Szmek <zbyszek@...waw.pl>, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2 3/3] coredump: hand a pidfd to the usermode coredump
 helper

On Fri, Apr 25, 2025 at 02:03:34PM +0200, Benjamin Drung wrote:
> On Fri, 2025-04-25 at 13:57 +0200, Christian Brauner wrote:
> > On Fri, Apr 25, 2025 at 01:31:56PM +0200, Benjamin Drung wrote:
> > > Hi,
> > > 
> > > On Mon, 2025-04-14 at 15:55 +0200, Christian Brauner wrote:
> > > > Give userspace a way to instruct the kernel to install a pidfd into the
> > > > usermode helper process. This makes coredump handling a lot more
> > > > reliable for userspace. In parallel with this commit we already have
> > > > systemd adding support for this in [1].
> > > > 
> > > > We create a pidfs file for the coredumping process when we process the
> > > > corename pattern. When the usermode helper process is forked we then
> > > > install the pidfs file as file descriptor three into the usermode
> > > > helpers file descriptor table so it's available to the exec'd program.
> > > > 
> > > > Since usermode helpers are either children of the system_unbound_wq
> > > > workqueue or kthreadd we know that the file descriptor table is empty
> > > > and can thus always use three as the file descriptor number.
> > > > 
> > > > Note, that we'll install a pidfd for the thread-group leader even if a
> > > > subthread is calling do_coredump(). We know that task linkage hasn't
> > > > been removed due to delay_group_leader() and even if this @current isn't
> > > > the actual thread-group leader we know that the thread-group leader
> > > > cannot be reaped until @current has exited.
> > > > 
> > > > Link: https://github.com/systemd/systemd/pull/37125 [1]
> > > > Tested-by: Luca Boccassi <luca.boccassi@...il.com>
> > > > Signed-off-by: Christian Brauner <brauner@...nel.org>
> > > > ---
> > > >  fs/coredump.c            | 59 ++++++++++++++++++++++++++++++++++++++++++++----
> > > >  include/linux/coredump.h |  1 +
> > > >  2 files changed, 56 insertions(+), 4 deletions(-)
> > > > 
> > > > diff --git a/fs/coredump.c b/fs/coredump.c
> > > > index 9da592aa8f16..403be0ff780e 100644
> > > > --- a/fs/coredump.c
> > > > +++ b/fs/coredump.c
> > > > @@ -43,6 +43,9 @@
> > > >  #include <linux/timekeeping.h>
> > > >  #include <linux/sysctl.h>
> > > >  #include <linux/elf.h>
> > > > +#include <linux/pidfs.h>
> > > > +#include <uapi/linux/pidfd.h>
> > > > +#include <linux/vfsdebug.h>
> > > >  
> > > >  #include <linux/uaccess.h>
> > > >  #include <asm/mmu_context.h>
> > > > @@ -60,6 +63,12 @@ static void free_vma_snapshot(struct coredump_params *cprm);
> > > >  #define CORE_FILE_NOTE_SIZE_DEFAULT (4*1024*1024)
> > > >  /* Define a reasonable max cap */
> > > >  #define CORE_FILE_NOTE_SIZE_MAX (16*1024*1024)
> > > > +/*
> > > > + * File descriptor number for the pidfd for the thread-group leader of
> > > > + * the coredumping task installed into the usermode helper's file
> > > > + * descriptor table.
> > > > + */
> > > > +#define COREDUMP_PIDFD_NUMBER 3
> > > >  
> > > >  static int core_uses_pid;
> > > >  static unsigned int core_pipe_limit;
> > > > @@ -339,6 +348,27 @@ static int format_corename(struct core_name *cn, struct coredump_params *cprm,
> > > >  			case 'C':
> > > >  				err = cn_printf(cn, "%d", cprm->cpu);
> > > >  				break;
> > > > +			/* pidfd number */
> > > > +			case 'F': {
> > > > +				/*
> > > > +				 * Installing a pidfd only makes sense if
> > > > +				 * we actually spawn a usermode helper.
> > > > +				 */
> > > > +				if (!ispipe)
> > > > +					break;
> > > > +
> > > > +				/*
> > > > +				 * Note that we'll install a pidfd for the
> > > > +				 * thread-group leader. We know that task
> > > > +				 * linkage hasn't been removed yet and even if
> > > > +				 * this @current isn't the actual thread-group
> > > > +				 * leader we know that the thread-group leader
> > > > +				 * cannot be reaped until @current has exited.
> > > > +				 */
> > > > +				cprm->pid = task_tgid(current);
> > > > +				err = cn_printf(cn, "%d", COREDUMP_PIDFD_NUMBER);
> > > > +				break;
> > > > +			}
> > > >  			default:
> > > >  				break;
> > > >  			}
> > > > 
> > > 
> > > I tried this change with Apport: I took the Ubuntu mainline kernel build
> > > https://kernel.ubuntu.com/mainline/daily/2025-04-24/ (that refers to
> > > mainline commit e54f9b0410347c49b7ffdd495578811e70d7a407) and applied
> > > these three patches on top. Then I modified Apport to take the
> > > additional `-F%F` and tested that on Ubuntu 25.04 (plucky). The result
> > > is the coredump failed as long as there was `-F%F` on
> > 
> > I have no clue what -F%F is and whether that leading -F is something
> > specific to Apport but the specifier is %F not -F%F. For example:
> > 
> >         > cat /proc/sys/kernel/core_pattern
> >         |/usr/lib/systemd/systemd-coredump %P %u %g %s %t %c %h %F
> > 
> > And note that this requires the pipe logic to be used, aka "|" needs to
> > be specified. Without it this doesn't make sense.
> 
> Apport takes short option parameters. They match the kernel template
> specifiers. The failing pattern:
> 
> $ cat /proc/sys/kernel/core_pattern 
> |/usr/share/apport/apport -p%p -s%s -c%c -d%d -P%P -u%u -g%g -F%F -- %E
> 
> Once I drop %F Apport is called without issues:
> 
> $ cat /proc/sys/kernel/core_pattern 
> |/usr/share/apport/apport -p%p -s%s -c%c -d%d -P%P -u%u -g%g -- %E

Youm must have CONFIG_DEBUG_VFS=y enabled where we trample the pidfs
file we just allocated. It's a debug only assert. I've removed it now
and pushed it to vfs-6.16.coredump. Can you either try with that or
simply unset CONFIG_DEBUG_VFS and retest.

Powered by blists - more mailing lists