On 09/05/2022 11:06, Miguel Ojeda wrote:

Hi Mickaël,

(Answering in v1 since I want to quote something in this cover letter)

There is only a v1 for this specific series, but other series (with 
higher versions) are now relying on this one. ;)

On Fri, May 6, 2022 at 6:03 PM Mickaël Salaün [off-list ref] wrote:

quoted

I used a simple coding style for the initial Landlock code.  However,
this may be subject to different interpretations.  To avoid relying
on tacit knowledge or text editors for these kind of rules, let's
automate it as much as possible thanks to clang-format.  This makes the
code formatting simple, consistent and impersonal.

Thanks for moving your subsystem to `clang-format`. I am glad you
found it useful.

quoted

Several versions of clang-format can be use but they may have (small)
different behaviors for undefined/new configuration parts.  After
testing different versions, I picked clang-format-14 which is relatively
new and fixes a bug present in version 11 to 13 (visible in the Landlock
formatted code).

Which was the bug?

You can check it by running clang-format-X on security/landlock/fs.c [1] 
and you'll get different results for this hunk (only):

--- a/security/landlock/fs.c
+++ b/security/landlock/fs.c

@@ -688,7 +688,7 @@ static inline access_mask_t maybe_remove(const 

struct dentry *const dentry)
  	if (d_is_negative(dentry))
  		return 0;
  	return d_is_dir(dentry) ? LANDLOCK_ACCESS_FS_REMOVE_DIR :
-				  LANDLOCK_ACCESS_FS_REMOVE_FILE;
+					LANDLOCK_ACCESS_FS_REMOVE_FILE;
  }

  /**


[1] 
https://git.kernel.org/pub/scm/linux/kernel/git/mic/linux.git/log/?h=landlock-wip