Linux filesystem development
 help / color / mirror / Atom feed
From: Nathan Chancellor <nathan@kernel.org>
To: Christian Brauner <brauner@kernel.org>
Cc: linux-fsdevel@vger.kernel.org,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Alexander Viro <viro@zeniv.linux.org.uk>,
	Seth Forshee <sforshee@kernel.org>,
	Tycho Andersen <tycho@tycho.pizza>
Subject: Re: [PATCH 2/2] pidfd: add pidfdfs
Date: Thu, 22 Feb 2024 12:03:34 -0700	[thread overview]
Message-ID: <20240222190334.GA412503@dev-arch.thelio-3990X> (raw)
In-Reply-To: <20240213-vfs-pidfd_fs-v1-2-f863f58cfce1@kernel.org>

Hi Christian,

On Tue, Feb 13, 2024 at 05:45:47PM +0100, Christian Brauner wrote:
> This moves pidfds from the anonymous inode infrastructure to a tiny
> pseudo filesystem. This has been on my todo for quite a while as it will
> unblock further work that we weren't able to do simply because of the
> very justified limitations of anonymous inodes. Moving pidfds to a tiny
> pseudo filesystem allows:
> 
> * statx() on pidfds becomes useful for the first time.
> * pidfds can be compared simply via statx() and then comparing inode
>   numbers.
> * pidfds have unique inode numbers for the system lifetime.
> * struct pid is now stashed in inode->i_private instead of
>   file->private_data. This means it is now possible to introduce
>   concepts that operate on a process once all file descriptors have been
>   closed. A concrete example is kill-on-last-close.
> * file->private_data is freed up for per-file options for pidfds.
> * Each struct pid will refer to a different inode but the same struct
>   pid will refer to the same inode if it's opened multiple times. In
>   contrast to now where each struct pid refers to the same inode. Even
>   if we were to move to anon_inode_create_getfile() which creates new
>   inodes we'd still be associating the same struct pid with multiple
>   different inodes.
> * Pidfds now go through the regular dentry_open() path which means that
>   all security hooks are called unblocking proper LSM management for
>   pidfds. In addition fsnotify hooks are called and allow for listening
>   to open events on pidfds.
> 
> The tiny pseudo filesystem is not visible anywhere in userspace exactly
> like e.g., pipefs and sockfs. There's no lookup, there's no complex
> inode operations, nothing. Dentries and inodes are always deleted when
> the last pidfd is closed.
> 
> The code is entirely optional and fairly small. If it's not selected we
> fallback to anonymous inodes. Heavily inspired by nsfs which uses a
> similar stashing mechanism just for namespaces.
> 
> Signed-off-by: Christian Brauner <brauner@kernel.org>

Apologies if this has already been reported or fixed but I did not see
anything on the mailing list.

On next-20240221 and next-20240222, with CONFIG_FS_PID=y, some of my
services such as abrtd, dbus, and polkit fail to start on my Fedora
machines, which causes further isssues like failing to start network
interfaces with NetworkManager. I can easily reproduce this in a Fedora
39 QEMU virtual machine, which has:

  # systemctl --version
  systemd 254 (254.9-1.fc39)
  +PAM +AUDIT +SELINUX -APPARMOR +IMA +SMACK +SECCOMP -GCRYPT +GNUTLS +OPENSSL +ACL +BLKID +CURL +ELFUTILS +FIDO2 +IDN2 -IDN -IPTC +KMOD +LIBCRYPTSETUP +LIBFDISK +PCRE2 +PWQUALITY +P11KIT +QRENCODE +TPM2 +BZIP2 +LZ4 +XZ +ZLIB +ZSTD +BPF_FRAMEWORK +XKBCOMMON +UTMP +SYSVINIT
   default-hierarchy=unified

Unfortunately, there does not really appear to be much information to
provide off bat but I am more than happy to try and gather whatever
information would be helpful if you are not able to reproduce locally.

  # uname -r
  6.8.0-rc1-00017-ga1a466d5af6c

  # zgrep CONFIG_FS_PID /proc/config.gz
  CONFIG_FS_PID=y

  # systemctl status polkit.service
  × polkit.service - Authorization Manager
       Loaded: loaded (/usr/lib/systemd/system/polkit.service; static)
      Drop-In: /usr/lib/systemd/system/service.d
               └─10-timeout-abort.conf
       Active: failed (Result: timeout) since Thu 2024-02-22 11:35:52 MST; 11min ago
         Docs: man:polkit(8)
      Process: 844 ExecStart=/usr/lib/polkit-1/polkitd --no-debug (code=killed, signal=TERM)
     Main PID: 844 (code=killed, signal=TERM)
          CPU: 116ms

  Feb 22 11:34:22 qemu systemd[1]: Starting polkit.service - Authorization Manager...
  Feb 22 11:34:22 qemu polkitd[844]: Started polkitd version 123
  Feb 22 11:34:22 qemu polkitd[844]: Loading rules from directory /etc/polkit-1/rules.d
  Feb 22 11:34:22 qemu polkitd[844]: Loading rules from directory /usr/share/polkit-1/rules.d
  Feb 22 11:34:22 qemu polkitd[844]: Finished loading, compiling and executing 5 rules
  Feb 22 11:34:22 qemu polkitd[844]: Acquired the name org.freedesktop.PolicyKit1 on the system bus
  Feb 22 11:35:52 qemu systemd[1]: polkit.service: start operation timed out. Terminating.
  Feb 22 11:35:52 qemu systemd[1]: polkit.service: Failed with result 'timeout'.
  Feb 22 11:35:52 qemu systemd[1]: Failed to start polkit.service - Authorization Manager.

vs.

  # uname -r
  6.8.0-rc1-00016-gd68c1231c030

  # systemctl status polkit.service
  ● polkit.service - Authorization Manager
       Loaded: loaded (/usr/lib/systemd/system/polkit.service; static)
      Drop-In: /usr/lib/systemd/system/service.d
               └─10-timeout-abort.conf
       Active: active (running) since Thu 2024-02-22 11:30:38 MST; 21s ago
         Docs: man:polkit(8)
     Main PID: 843 (polkitd)
        Tasks: 4 (limit: 19010)
       Memory: 5.0M
          CPU: 169ms
       CGroup: /system.slice/polkit.service
               └─843 /usr/lib/polkit-1/polkitd --no-debug

  Feb 22 11:30:38 qemu systemd[1]: Starting polkit.service - Authorization Manager...
  Feb 22 11:30:38 qemu polkitd[843]: Started polkitd version 123
  Feb 22 11:30:38 qemu polkitd[843]: Loading rules from directory /etc/polkit-1/rules.d
  Feb 22 11:30:38 qemu polkitd[843]: Loading rules from directory /usr/share/polkit-1/rules.d
  Feb 22 11:30:38 qemu polkitd[843]: Finished loading, compiling and executing 5 rules
  Feb 22 11:30:38 qemu polkitd[843]: Acquired the name org.freedesktop.PolicyKit1 on the system bus
  Feb 22 11:30:38 qemu systemd[1]: Started polkit.service - Authorization Manager.

or

  # uname -r
  6.8.0-rc1-00017-ga1a466d5af6c

  # zgrep CONFIG_FS_PID /proc/config.gz
  # CONFIG_FS_PID is not set

  # systemctl status polkit.service
  ● polkit.service - Authorization Manager
       Loaded: loaded (/usr/lib/systemd/system/polkit.service; static)
      Drop-In: /usr/lib/systemd/system/service.d
               └─10-timeout-abort.conf
       Active: active (running) since Thu 2024-02-22 11:52:41 MST; 5min ago
         Docs: man:polkit(8)
     Main PID: 845 (polkitd)
        Tasks: 4 (limit: 19010)
       Memory: 5.0M
          CPU: 177ms
       CGroup: /system.slice/polkit.service
               └─845 /usr/lib/polkit-1/polkitd --no-debug

  Feb 22 11:52:41 qemu systemd[1]: Starting polkit.service - Authorization Manager...
  Feb 22 11:52:41 qemu polkitd[845]: Started polkitd version 123
  Feb 22 11:52:41 qemu polkitd[845]: Loading rules from directory /etc/polkit-1/rules.d
  Feb 22 11:52:41 qemu polkitd[845]: Loading rules from directory /usr/share/polkit-1/rules.d
  Feb 22 11:52:41 qemu polkitd[845]: Finished loading, compiling and executing 5 rules
  Feb 22 11:52:41 qemu polkitd[845]: Acquired the name org.freedesktop.PolicyKit1 on the system bus
  Feb 22 11:52:41 qemu systemd[1]: Started polkit.service - Authorization Manager.

I looked your most recent push of vfs.pidfd but I did not see anything
that would have appeared to fix this, so I did not test it.

Cheers,
Nathan

  parent reply	other threads:[~2024-02-22 19:03 UTC|newest]

Thread overview: 61+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-02-13 16:45 [PATCH 0/2] Move pidfd to tiny pseudo fs Christian Brauner
2024-02-13 16:45 ` [PATCH 1/2] pidfd: move struct pidfd_fops Christian Brauner
2024-02-13 16:45 ` [PATCH 2/2] pidfd: add pidfdfs Christian Brauner
2024-02-13 17:17   ` Linus Torvalds
2024-02-14 14:40     ` Christian Brauner
2024-02-14 18:27       ` Christian Brauner
2024-02-14 18:37         ` Linus Torvalds
2024-02-15 16:11           ` Christian Brauner
2024-02-16 11:50             ` Christian Brauner
2024-02-16 16:41               ` Christian Brauner
2024-02-17 13:59               ` Oleg Nesterov
2024-02-17 17:30                 ` Linus Torvalds
2024-02-17 17:38                   ` Linus Torvalds
2024-02-18 11:15                   ` Christian Brauner
2024-02-18 11:33                     ` Christian Brauner
2024-02-18 17:54                       ` Christian Brauner
2024-02-18 18:08                         ` Linus Torvalds
2024-02-18 18:57                           ` Linus Torvalds
2024-02-19 18:05                             ` Christian Brauner
2024-02-19 18:34                               ` Linus Torvalds
2024-02-19 21:18                                 ` Christian Brauner
2024-02-19 23:24                                   ` Linus Torvalds
2024-02-18 14:27                     ` Oleg Nesterov
2024-02-18  9:30                 ` Christian Brauner
2024-02-22 19:03   ` Nathan Chancellor [this message]
2024-02-23 10:18     ` Heiko Carstens
2024-02-23 11:56       ` Christian Brauner
2024-02-23 11:55     ` Christian Brauner
2024-02-23 12:57       ` Heiko Carstens
2024-02-23 13:27         ` Christian Brauner
2024-02-23 13:35           ` Heiko Carstens
2024-02-23 13:41       ` Christian Brauner
2024-02-23 21:26       ` Christian Brauner
2024-02-23 21:58         ` Linus Torvalds
2024-02-24  5:52           ` Christian Brauner
2024-02-24  6:05             ` Christian Brauner
2024-02-24 18:48             ` Linus Torvalds
2024-02-24 19:15               ` Christian Brauner
2024-02-24 19:19                 ` Christian Brauner
2024-02-24 19:21                 ` Linus Torvalds
2024-02-27 19:26                 ` Nathan Chancellor
2024-02-27 22:13                   ` Christian Brauner
2024-03-12 10:35   ` Geert Uytterhoeven
2024-03-12 14:09     ` Christian Brauner
2024-05-15 11:10   ` Jiri Slaby
2024-05-15 16:39     ` Christian Brauner
2024-05-16  5:28       ` Jiri Slaby
2024-05-17  7:09         ` Jiri Slaby
2024-05-17  7:54           ` Jiri Slaby
2024-05-17 20:07             ` Linus Torvalds
2024-05-20  8:23               ` Jiri Slaby
2024-05-20 19:01                 ` Linus Torvalds
2024-05-20 19:15                   ` Linus Torvalds
2024-05-21  6:07                     ` Jiri Slaby
2024-05-21  6:13                       ` Jiri Slaby
2024-05-21 12:33                         ` Christian Brauner
2024-05-21 12:40                           ` Christian Brauner
2024-05-21 15:10                             ` Linus Torvalds
2024-05-25 11:57                               ` Christian Brauner
2024-05-21 12:16               ` Christian Brauner
2024-02-13 17:02 ` [PATCH 0/2] Move pidfd to tiny pseudo fs Christian Brauner

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20240222190334.GA412503@dev-arch.thelio-3990X \
    --to=nathan@kernel.org \
    --cc=brauner@kernel.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=sforshee@kernel.org \
    --cc=torvalds@linux-foundation.org \
    --cc=tycho@tycho.pizza \
    --cc=viro@zeniv.linux.org.uk \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox