linux-kernel - Re: [PATCH v5 3/3] userfaultfd: use per-vma locks in userfaultfd operations

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: Tue, 13 Feb 2024 14:51:25 -0500
From: "Liam R. Howlett" <Liam.Howlett@...cle.com>
To: Lokesh Gidra <lokeshgidra@...gle.com>
Cc: Suren Baghdasaryan <surenb@...gle.com>, akpm@...ux-foundation.org,
        linux-fsdevel@...r.kernel.org, linux-mm@...ck.org,
        linux-kernel@...r.kernel.org, selinux@...r.kernel.org,
        kernel-team@...roid.com, aarcange@...hat.com, peterx@...hat.com,
        david@...hat.com, axelrasmussen@...gle.com, bgeffon@...gle.com,
        willy@...radead.org, jannh@...gle.com, kaleshsingh@...gle.com,
        ngeoffray@...gle.com, timmurray@...gle.com, rppt@...nel.org
Subject: Re: [PATCH v5 3/3] userfaultfd: use per-vma locks in userfaultfd
 operations

* Lokesh Gidra <lokeshgidra@...gle.com> [240213 14:37]:
..

> Asking to avoid any more iterations: these functions should call the
> currently defined ones or should replace them. For instance, should I
> do the following:
> 
> #ifdef CONFIG_PER_VMA_LOCK
> ... uffd_mfill_lock()
> {
>         return find_and_lock_dst_vma(...);
> }
> #else
> ...uffd_mfill_lock()
> {
>        return lock_mm_and_find_dst_vma(...);
> }
> #endif
> 
> or have the function replace
> find_and_lock_dst_vma()/lock_mm_and_find_dst_vma() ?

Since the two have the same prototype, then you can replace the function
names directly.

The other side should take the vma and use vma->vm_mm to get the mm to
unlock the mmap_lock in the !CONFIG_PER_VMA_LOCK.  That way those
prototypes also match and can use the same names directly.

move_pages() requires unlocking two VMAs or one, so pass both VMAs
through and do the check in there.  This, unfortunately means that one
of the VMAs will not be used in the !CONFIG_PER_VMA_LOCK case.  You
could add an assert to ensure src_vma is locked prior to using dst_vma
to unlock the mmap_lock(), to avoid potential bot emails.

Thanks,
Liam