[PATCH v1 1/2] landlock: Handle weird files

[PATCH v1 1/2] landlock: Handle weird files

[Mickaël Salaün]

A corrupted filesystem (e.g. bcachefs) might return weird files.
Instead of throwing a warning and allowing access to such file, treat
them as regular files.

Cc: Dave Chinner <david@fromorbit.com>
Cc: Günther Noack <gnoack@google.com>
Cc: Kent Overstreet <kent.overstreet@linux.dev>
Cc: Paul Moore <paul@paul-moore.com>
Reported-by: syzbot+34b68f850391452207df@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/r/000000000000a65b35061cffca61@google.com
Reported-by: syzbot+360866a59e3c80510a62@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/r/67379b3f.050a0220.85a0.0001.GAE@google.com
Reported-by: Ubisectech Sirius <bugreport@ubisectech.com>
Closes: https://lore.kernel.org/r/c426821d-8380-46c4-a494-7008bbd7dd13.bugreport@ubisectech.com
Fixes: cb2c7d1a1776 ("landlock: Support filesystem access-control")
Signed-off-by: Mickaël Salaün <mic@digikod.net>
---
 security/landlock/fs.c | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)

diff --git a/security/landlock/fs.c b/security/landlock/fs.c
index e31b97a9f175..7adb25150488 100644
--- a/security/landlock/fs.c
+++ b/security/landlock/fs.c
@@ -937,10 +937,6 @@ static access_mask_t get_mode_access(const umode_t mode)
 	switch (mode & S_IFMT) {
 	case S_IFLNK:
 		return LANDLOCK_ACCESS_FS_MAKE_SYM;
-	case 0:
-		/* A zero mode translates to S_IFREG. */
-	case S_IFREG:
-		return LANDLOCK_ACCESS_FS_MAKE_REG;
 	case S_IFDIR:
 		return LANDLOCK_ACCESS_FS_MAKE_DIR;
 	case S_IFCHR:
@@ -951,9 +947,12 @@ static access_mask_t get_mode_access(const umode_t mode)
 		return LANDLOCK_ACCESS_FS_MAKE_FIFO;
 	case S_IFSOCK:
 		return LANDLOCK_ACCESS_FS_MAKE_SOCK;
+	case S_IFREG:
+	case 0:
+		/* A zero mode translates to S_IFREG. */
 	default:
-		WARN_ON_ONCE(1);
-		return 0;
+		/* Treats weird files as regular files. */
+		return LANDLOCK_ACCESS_FS_MAKE_REG;
 	}
 }
 
-- 
2.47.1

[PATCH v1 2/2] landlock: Constify get_mode_access()

[Mickaël Salaün]

Use __attribute_const__ for get_mode_access().

Cc: Günther Noack <gnoack@google.com>
Signed-off-by: Mickaël Salaün <mic@digikod.net>
---
 security/landlock/fs.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/security/landlock/fs.c b/security/landlock/fs.c
index 7adb25150488..f81d0335b825 100644
--- a/security/landlock/fs.c
+++ b/security/landlock/fs.c
@@ -932,7 +932,7 @@ static int current_check_access_path(const struct path *const path,
 	return check_access_path(dom, path, access_request);
 }
 
-static access_mask_t get_mode_access(const umode_t mode)
+static __attribute_const__ access_mask_t get_mode_access(const umode_t mode)
 {
 	switch (mode & S_IFMT) {
 	case S_IFLNK:
-- 
2.47.1

Re: [PATCH v1 1/2] landlock: Handle weird files

[Günther Noack]

On Fri, Jan 10, 2025 at 04:39:13PM +0100, Mickaël Salaün wrote:
> A corrupted filesystem (e.g. bcachefs) might return weird files.
> Instead of throwing a warning and allowing access to such file, treat
> them as regular files.
> 
> Cc: Dave Chinner <david@fromorbit.com>
> Cc: Günther Noack <gnoack@google.com>
> Cc: Kent Overstreet <kent.overstreet@linux.dev>
> Cc: Paul Moore <paul@paul-moore.com>
> Reported-by: syzbot+34b68f850391452207df@syzkaller.appspotmail.com
> Closes: https://lore.kernel.org/r/000000000000a65b35061cffca61@google.com
> Reported-by: syzbot+360866a59e3c80510a62@syzkaller.appspotmail.com
> Closes: https://lore.kernel.org/r/67379b3f.050a0220.85a0.0001.GAE@google.com
> Reported-by: Ubisectech Sirius <bugreport@ubisectech.com>
> Closes: https://lore.kernel.org/r/c426821d-8380-46c4-a494-7008bbd7dd13.bugreport@ubisectech.com
> Fixes: cb2c7d1a1776 ("landlock: Support filesystem access-control")
> Signed-off-by: Mickaël Salaün <mic@digikod.net>
> ---
>  security/landlock/fs.c | 11 +++++------
>  1 file changed, 5 insertions(+), 6 deletions(-)
> 
> diff --git a/security/landlock/fs.c b/security/landlock/fs.c
> index e31b97a9f175..7adb25150488 100644
> --- a/security/landlock/fs.c
> +++ b/security/landlock/fs.c
> @@ -937,10 +937,6 @@ static access_mask_t get_mode_access(const umode_t mode)
>  	switch (mode & S_IFMT) {
>  	case S_IFLNK:
>  		return LANDLOCK_ACCESS_FS_MAKE_SYM;
> -	case 0:
> -		/* A zero mode translates to S_IFREG. */
> -	case S_IFREG:
> -		return LANDLOCK_ACCESS_FS_MAKE_REG;
>  	case S_IFDIR:
>  		return LANDLOCK_ACCESS_FS_MAKE_DIR;
>  	case S_IFCHR:
> @@ -951,9 +947,12 @@ static access_mask_t get_mode_access(const umode_t mode)
>  		return LANDLOCK_ACCESS_FS_MAKE_FIFO;
>  	case S_IFSOCK:
>  		return LANDLOCK_ACCESS_FS_MAKE_SOCK;
> +	case S_IFREG:
> +	case 0:
> +		/* A zero mode translates to S_IFREG. */
>  	default:
> -		WARN_ON_ONCE(1);
> -		return 0;
> +		/* Treats weird files as regular files. */
> +		return LANDLOCK_ACCESS_FS_MAKE_REG;
>  	}
>  }
>  
> -- 
> 2.47.1
> 

Reviewed-by: Günther Noack <gnoack3000@gmail.com>

Makes sense to me, since this is enforcing a stronger check than before
and can only happen in the case of corruption.

I do not have a good intuition about what happens afterwards when the
file system is in such a state.  I imagine that this will usually give
an error shortly afterwards, as the opening of the file continues?  Is
that right?

–Günther

Re: [PATCH v1 2/2] landlock: Constify get_mode_access()

[Günther Noack]

On Fri, Jan 10, 2025 at 04:39:14PM +0100, Mickaël Salaün wrote:
> Use __attribute_const__ for get_mode_access().
> 
> Cc: Günther Noack <gnoack@google.com>
> Signed-off-by: Mickaël Salaün <mic@digikod.net>
> ---
>  security/landlock/fs.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/security/landlock/fs.c b/security/landlock/fs.c
> index 7adb25150488..f81d0335b825 100644
> --- a/security/landlock/fs.c
> +++ b/security/landlock/fs.c
> @@ -932,7 +932,7 @@ static int current_check_access_path(const struct path *const path,
>  	return check_access_path(dom, path, access_request);
>  }
>  
> -static access_mask_t get_mode_access(const umode_t mode)
> +static __attribute_const__ access_mask_t get_mode_access(const umode_t mode)
>  {
>  	switch (mode & S_IFMT) {
>  	case S_IFLNK:
> -- 
> 2.47.1
> 

Reviewed-by: Günther Noack <gnoack3000@gmail.com>

Re: [PATCH v1 1/2] landlock: Handle weird files

[Mickaël Salaün]

On Fri, Jan 10, 2025 at 05:37:26PM +0100, Günther Noack wrote:
> On Fri, Jan 10, 2025 at 04:39:13PM +0100, Mickaël Salaün wrote:
> > A corrupted filesystem (e.g. bcachefs) might return weird files.
> > Instead of throwing a warning and allowing access to such file, treat
> > them as regular files.
> > 
> > Cc: Dave Chinner <david@fromorbit.com>
> > Cc: Günther Noack <gnoack@google.com>
> > Cc: Kent Overstreet <kent.overstreet@linux.dev>
> > Cc: Paul Moore <paul@paul-moore.com>
> > Reported-by: syzbot+34b68f850391452207df@syzkaller.appspotmail.com
> > Closes: https://lore.kernel.org/r/000000000000a65b35061cffca61@google.com
> > Reported-by: syzbot+360866a59e3c80510a62@syzkaller.appspotmail.com
> > Closes: https://lore.kernel.org/r/67379b3f.050a0220.85a0.0001.GAE@google.com
> > Reported-by: Ubisectech Sirius <bugreport@ubisectech.com>
> > Closes: https://lore.kernel.org/r/c426821d-8380-46c4-a494-7008bbd7dd13.bugreport@ubisectech.com
> > Fixes: cb2c7d1a1776 ("landlock: Support filesystem access-control")
> > Signed-off-by: Mickaël Salaün <mic@digikod.net>
> > ---
> >  security/landlock/fs.c | 11 +++++------
> >  1 file changed, 5 insertions(+), 6 deletions(-)
> > 
> > diff --git a/security/landlock/fs.c b/security/landlock/fs.c
> > index e31b97a9f175..7adb25150488 100644
> > --- a/security/landlock/fs.c
> > +++ b/security/landlock/fs.c
> > @@ -937,10 +937,6 @@ static access_mask_t get_mode_access(const umode_t mode)
> >  	switch (mode & S_IFMT) {
> >  	case S_IFLNK:
> >  		return LANDLOCK_ACCESS_FS_MAKE_SYM;
> > -	case 0:
> > -		/* A zero mode translates to S_IFREG. */
> > -	case S_IFREG:
> > -		return LANDLOCK_ACCESS_FS_MAKE_REG;
> >  	case S_IFDIR:
> >  		return LANDLOCK_ACCESS_FS_MAKE_DIR;
> >  	case S_IFCHR:
> > @@ -951,9 +947,12 @@ static access_mask_t get_mode_access(const umode_t mode)
> >  		return LANDLOCK_ACCESS_FS_MAKE_FIFO;
> >  	case S_IFSOCK:
> >  		return LANDLOCK_ACCESS_FS_MAKE_SOCK;
> > +	case S_IFREG:
> > +	case 0:
> > +		/* A zero mode translates to S_IFREG. */
> >  	default:
> > -		WARN_ON_ONCE(1);
> > -		return 0;
> > +		/* Treats weird files as regular files. */
> > +		return LANDLOCK_ACCESS_FS_MAKE_REG;
> >  	}
> >  }
> >  
> > -- 
> > 2.47.1
> > 
> 
> Reviewed-by: Günther Noack <gnoack3000@gmail.com>
> 
> Makes sense to me, since this is enforcing a stronger check than before
> and can only happen in the case of corruption.
> 
> I do not have a good intuition about what happens afterwards when the
> file system is in such a state.  I imagine that this will usually give
> an error shortly afterwards, as the opening of the file continues?  Is
> that right?

I guess it depends on the filesystem implementation.  For instance, XFS
returns an error if a weird file is detected [1], whereas bcachefs
ignores it (which is considered a bug, but not fixed yet) [2].

[1] https://lore.kernel.org/all/Zpc46HEacI%2Fwd7Rg@dread.disaster.area/
[2] https://lore.kernel.org/all/4hohnthh54adx35lnxzedop3oxpntpmtygxso4iraiexfdlt4d@6m7ssepvjyar/

> 
> –Günther
> 

Re: [PATCH v1 1/2] landlock: Handle weird files

[Christoph Hellwig]

On Sat, Jan 11, 2025 at 04:38:56PM +0100, Mickaël Salaün wrote:
> I guess it depends on the filesystem implementation.  For instance, XFS
> returns an error if a weird file is detected [1], whereas bcachefs
> ignores it (which is considered a bug, but not fixed yet) [2].

If a filesyste, returns an invalid mode that's a file system bug and
needs to be fixed there.  Warning in a consumer is perfectly fine.
But the right action in that case is indeed not to grant the access.

Re: [PATCH v1 1/2] landlock: Handle weird files

[Christian Brauner]

On Tue, Jan 14, 2025 at 11:15:42PM -0800, Christoph Hellwig wrote:
> On Sat, Jan 11, 2025 at 04:38:56PM +0100, Mickaël Salaün wrote:
> > I guess it depends on the filesystem implementation.  For instance, XFS
> > returns an error if a weird file is detected [1], whereas bcachefs
> > ignores it (which is considered a bug, but not fixed yet) [2].
> 
> If a filesyste, returns an invalid mode that's a file system bug and
> needs to be fixed there.  Warning in a consumer is perfectly fine.
> But the right action in that case is indeed not to grant the access.

Fyi, anonymous inodes traditionally set the mode to 0 which is
really annoying:

lrwx------ 1 root root  64 15. Jan 11:52 94 -> anon_inode:bpf-prog

> sudo stat -L /proc/1/fd/94
  File: /proc/1/fd/94
  Size: 0               Blocks: 0          IO Block: 4096   weird file
Device: 0,15    Inode: 4120        Links: 1
Access: (0600/?rw-------)  Uid: (    0/    root)   Gid: (    0/    root)
Access: 2024-11-05 17:15:54.404000000 +0100
Modify: 2024-11-05 17:15:54.404000000 +0100
Change: 2024-11-05 17:15:54.404000000 +0100
 Birth: -