linux-kernel - Re: [PATCH v2] epoll: Use user_write_access_begin() and unsafe_put_user() in epoll_put

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]

Message-ID: <CAHk-=wiYvRqYYCi3Wcq==Y7Zg_i9u-04sk4BtXrP03txZ0yzww@mail.gmail.com>
Date: Tue, 28 Oct 2025 12:12:38 -0700
From: Linus Torvalds <torvalds@...ux-foundation.org>
To: Kuniyuki Iwashima <kuniyu@...gle.com>
Cc: Christian Brauner <brauner@...nel.org>, Jens Axboe <axboe@...nel.dk>, 
	Dave Hansen <dave.hansen@...ux.intel.com>, David Laight <david.laight.linux@...il.com>, 
	Eric Dumazet <edumazet@...gle.com>, Kuniyuki Iwashima <kuni1840@...il.com>, linux-kernel@...r.kernel.org, 
	Dave Hansen <dave.hansen@...el.com>
Subject: Re: [PATCH v2] epoll: Use user_write_access_begin() and
 unsafe_put_user() in epoll_put_uevent().

 On Tue, 28 Oct 2025 at 12:02, Linus Torvalds
<torvalds@...ux-foundation.org> wrote:
>
> Because I think the proper patch should look something like the
> attached.. For me, that generates
>
>         movabs $0x123456789abcdef,%rcx
>         cmp    %rcx,%r15
>         cmova  %rcx,%r15
>         stac
>         mov    %r12d,(%r15)
>         mov    %rax,0x4(%r15)
>         clac
>
> which honestly should be pretty much optimal.

Side note: when I say "pretty much optimal", the truly optimal
situation would be to do the address masking entirely outside the loop
in ep_send_events(), but we don't expose that interface, and we have
absolutely horrible experiences with doing user address checking
separately from the actual code that does the access, so I'm really
loathe to add that kind of logic.

So the above is *not* optimal in the sense that the

        movabs $0x123456789abcdef,%rcx
        cmp    %rcx,%r15
        cmova  %rcx,%r15

part certainly *could* be done outside the list_for_each_entry_safe()
loop in ep_send_events(), but I do not think that three simple
ALU-only instructions should make a difference here.

              Linus