Re: [PATCH v10 2/8] mm: rust: add vm_area_struct methods that require read access
From: Lorenzo Stoakes
Date: Thu Dec 05 2024 - 05:47:57 EST
On Thu, Dec 05, 2024 at 11:40:03AM +0100, Alice Ryhl wrote:
> On Fri, Nov 29, 2024 at 5:32 PM Alice Ryhl <aliceryhl@xxxxxxxxxx> wrote:
> >
> > This adds a type called VmAreaRef which is used when referencing a vma
> > that you have read access to. Here, read access means that you hold
> > either the mmap read lock or the vma read lock (or stronger).
> >
> > Additionally, a vma_lookup method is added to the mmap read guard, which
> > enables you to obtain a &VmAreaRef in safe Rust code.
> >
> > This patch only provides a way to lock the mmap read lock, but a
> > follow-up patch also provides a way to just lock the vma read lock.
> >
> > Acked-by: Lorenzo Stoakes <lorenzo.stoakes@xxxxxxxxxx> (for mm bits)
> > Reviewed-by: Jann Horn <jannh@xxxxxxxxxx>
> > Signed-off-by: Alice Ryhl <aliceryhl@xxxxxxxxxx>
>
> It looks like binder needs a way to check whether a given vma is
> associated with a given mm. I guess we can do that by adding a method
> to VmAreaRef that returns a &MmWithUser.
Presumably this would be with a lock held to ensure the VMA doesn't
disappear from under us?
I guess that's implied by possessing a VmAreaRef in the first place.
I also suppoes that the mm having users is implied by you having the VMA
that implies a lock held on it :)
So that's probably fine then, as long as you can implement some sensible
means of comparison between a known 'given mm' vs. the &MmWithUser.
>
> This indicates that vma->vm_mm references an mm with non-zero mm_users
> whenever it's legal to have a reference to the vma.
>
> Alice