linux-kernel - Re: [PATCH 03/11] fs: add new read_uptr and write

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

Date:   Wed, 24 Jun 2020 10:19:16 -0700
From:   Linus Torvalds <torvalds@...ux-foundation.org>
To:     Christoph Hellwig <hch@....de>
Cc:     Al Viro <viro@...iv.linux.org.uk>,
        Luis Chamberlain <mcgrof@...nel.org>,
        Kees Cook <keescook@...omium.org>,
        Iurii Zaikin <yzaikin@...gle.com>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        linux-fsdevel <linux-fsdevel@...r.kernel.org>
Subject: Re: [PATCH 03/11] fs: add new read_uptr and write_uptr file operations

On Wed, Jun 24, 2020 at 9:29 AM Christoph Hellwig <hch@....de> wrote:
>
> Add two new file operations that are identical to ->read and ->write
> except that they can also safely take kernel pointers using the uptr_t
> type.

Honestly, I think this is the wrong way to go.

All of this new complexity and messiness, just to remove a few
unimportant final cases?

If somebody can't be bothered to convert a driver to
iter_read/iter_write, why would they be bothered to convert it to
read_uptr/write_uptr?

And this messiness will stay around for decades.

So let's not go down that path.

If you want to do "splice() and kernel_read() requires read_iter"
(with a warning so that we find any cases), then that's fine. But
let's not add yet _another_ read type.

Why did you care so much about sysctl, and why couldn't they use the iter ops?

                    Linus