On Mon, Aug 26, 2024 at 3:57 PM Alejandro Colomar [off-list ref] wrote:

Hi Elliott,

On Mon, Aug 26, 2024 at 12:37:46PM GMT, enh wrote:

quoted

looks like Linux 6.6...

Thanks for researching that!

quoted

commit 475d4df82719225510625b4263baa1105665f4b3
Merge: 511fb5bafed1 712143795327
Author: Linus Torvalds [off-list ref]
Date:   Mon Aug 28 11:25:27 2023 -0700

    Merge tag 'v6.6-vfs.fchmodat2' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs

    Pull fchmodat2 system call from Christian Brauner:
     "This adds the fchmodat2() system call. It is a revised version of the
      fchmodat() system call, adding a missing flag argument. Support for
      both AT_SYMLINK_NOFOLLOW and AT_EMPTY_PATH are included.

Maybe we need to add a "C library/kernel differences" section?

what's your timeline for "currently interesting" vs "just historical"?
(though i guess whatever that is, Linux 6.6 is pretty new at less than
12 months old!)

        $ grep -rh '^\.SS ' man | sort | uniq -c | sort | tail
              5 .SS glibc
              7 .SS Authors and copyright conditions
              8 .SS Address format
             11 .SS Ioctls
             11 .SS glibc notes
             12 .SS Socket options
             13 .SS /proc interfaces
             15 .SS ISO/IEC\~8859 alphabets
             51 .SS C library/kernel differences
             99 .SS Program source

Would you mind preparing a patch?

if you just want the trivial removal of the incorrect sentence, sure
... if you want a whole new "C library/kernel differences" section
you're probably able to write that ... which is probably a good
juncture to point out that there's also a whole missing function ---
there's no mention of lchmod() which seems to have been added in
GLIBC_2.3.2 (yes, 22 years ago; not to be confused with 2.32 :-) ...
except i don't think it _worked_ on Linux until glibc sha
6b89c385d8bd0700b25bac2c2d0bebe68d5cc05d in 2020? i don't know how you
want to document that!).

Have a lovely day!
Alex

quoted

      Adding this system call revision has been a longstanding request but
      so far has always fallen through the cracks. While the kernel
      implementation of fchmodat() does not have a flag argument the libc
      provided POSIX-compliant fchmodat(3) version does. Both glibc and musl
      have to implement a workaround in order to support AT_SYMLINK_NOFOLLOW
      (see [1] and [2]).

      The workaround is brittle because it relies not just on O_PATH and
      O_NOFOLLOW semantics and procfs magic links but also on our rather
      inconsistent symlink semantics.

      This gives userspace a proper fchmodat2() system call that libcs can
      use to properly implement fchmodat(3) and allows them to get rid of
      their hacks. In this case it will immediately benefit them as the
      current workaround is already defunct because of aformentioned
      inconsistencies.

      In addition to AT_SYMLINK_NOFOLLOW, give userspace the ability to use
      AT_EMPTY_PATH with fchmodat2(). This is already possible with
      fchownat() so there's no reason to not also support it for
      fchmodat2().

      The implementation is simple and comes with selftests. Implementation
      of the system call and wiring up the system call are done as separate
      patches even though they could arguably be one patch. But in case
      there are merge conflicts from other system call additions it can be
      beneficial to have separate patches"

    Link: https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/unix/sysv/linux/fchmodat.c;h=17eca54051ee28ba1ec3f9aed170a62630959143;hb=a492b1e5ef7ab50c6fdd4e4e9879ea5569ab0a6c#l35
[1]
    Link: https://git.musl-libc.org/cgit/musl/tree/src/stat/fchmodat.c?id=718f363bc2067b6487900eddc9180c84e7739f80#n28
[2]

    * tag 'v6.6-vfs.fchmodat2' of
git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs:
      selftests: fchmodat2: remove duplicate unneeded defines
      fchmodat2: add support for AT_EMPTY_PATH
      selftests: Add fchmodat2 selftest
      arch: Register fchmodat2, usually as syscall 452
      fs: Add fchmodat2()
      Non-functional cleanup of a "__user * filename"

On Sun, Aug 25, 2024 at 5:52 AM Alejandro Colomar [off-list ref] wrote:

quoted

Hi Elliott,

On Fri, Aug 23, 2024 at 02:40:16PM GMT, enh wrote:

quoted

subject says it all... the "This flag is not currently implemented."
on the man page is wrong, i think?

I guess it was right when the text was written in 2006.

$ git blame --follow -- man/man2/chmod.2 | grep currently
92f114db82 man2/chmod.2     (Michael Kerrisk   2014-02-21 08:35:27 +0100 207) This flag is not currently implemented.
$ git blame 92f114db82^ -- man2/fchmodat.2 | grep currently
a53b8cb2a0 (Michael Kerrisk  2006-05-02 00:05:06 +0000  99) This flag is not currently implemented.

That might perfectly have changed in the last 18 years.  :)

Would you mind writing a small program and shell session that
demonstrates it?  It could be interesting for an EXAMPLES section.


Have a lovely day!
Alex

--
<https://www.alejandro-colomar.es/>

--
<https://www.alejandro-colomar.es/>