linux-kernel - Re: [PATCH 1/2] fs: move getname and putname handlers into namei.h

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20251201075440.GZ3538@ZenIV>
Date: Mon, 1 Dec 2025 07:54:40 +0000
From: Al Viro <viro@...iv.linux.org.uk>
To: Mateusz Guzik <mjguzik@...il.com>
Cc: brauner@...nel.org, jack@...e.cz, linux-kernel@...r.kernel.org,
	linux-fsdevel@...r.kernel.org
Subject: Re: [PATCH 1/2] fs: move getname and putname handlers into namei.h

On Mon, Dec 01, 2025 at 07:27:05AM +0100, Mateusz Guzik wrote:
> self-NAK, i'm going to rebase on top of
> https://lore.kernel.org/linux-fsdevel/20251129170142.150639-1-viro@zeniv.linux.org.uk/
> when the time comes

	FWIW, I'm putting together struct filename-related
branches for the next cycle (trying to figure out what's the
best way to linearize the mess I've got).

	If you throw a patch on top of the series I've posted
on Friday (and it should shrink a whole lot - the damn thing
is static in fs/namei.c now, with cache initialization done
in the same place) I would be glad to apply it.

	Other series around struct filename for the next
cycle:

* DEFINE_CLASS(filename, struct filename *, putname(_T), getname(p), const char __user *p)
EXTEND_CLASS(filename, _kernel, getname_kernel(p), const char *p)
EXTEND_CLASS(filename, _flags, getname_flags(p, f), const char *p, unsigned int f)
EXTEND_CLASS(filename, _uflags, getname_uflags(p, f), const char *p, unsigned int f)
EXTEND_CLASS(filename, _maybe_null, getname_maybe_null(p, f), const char *p, unsigned int f)
EXTEND_CLASS(filename, _consume, no_free_ptr(p), struct filename *p)
and an bunch of conversions making use of that.  Generally I dislike RAII
patterns, but uses of struct filename make a reasonably good fit.

* change of calling conventions for do_filp_open() - let it DTRT when
it's given ERR_PTR() for name.  Surprising amount of error handling
in the callers (and callers of callers, etc.) folds into that.
One thing that is very tempting is s/do_filp_open/do_file_open/,
while we are at it...

* killing pointless checks for IS_ERR() before calling filename_lookup()
and vfs_path_parent_lookup() - those already treat ERR_PTR() for name
as "bail out immediately".

* [currently very tentative] dealing with __audit_inode().  It's
a surprisingly convoluted series, and it's not quite finished yet.
Not sure if that one won't end up slipping past 6.20...

Linearizing that stuff into something that would not be a hell wrt
merges is... interesting.