[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZSQO4fHaAxDkbGyz@google.com>
Date: Mon, 9 Oct 2023 07:32:48 -0700
From: Sean Christopherson <seanjc@...gle.com>
To: Al Viro <viro@...iv.linux.org.uk>
Cc: linux-kernel@...r.kernel.org, kvm@...r.kernel.org,
Paolo Bonzini <pbonzini@...hat.com>
Subject: Re: [PATCH gmem FIXUP] kvm: guestmem: do not use a file system
On Mon, Oct 09, 2023, Al Viro wrote:
> On Thu, Sep 28, 2023 at 07:22:16PM -0700, Sean Christopherson wrote:
> > On Thu, 28 Sep 2023 14:06:51 -0400, Paolo Bonzini wrote:
> > > Use a run-of-the-mill anonymous inode, there is nothing useful
> > > being provided by kvm_gmem_fs.
> > >
> > >
> >
> > Applied to kvm-x86 guest_memfd, thanks!
> >
> > [1/1] kvm: guestmem: do not use a file system
> > https://github.com/kvm-x86/linux/commit/0f7e60a5f42a
>
> Please, revert; this is completely broken. anon_inode_getfile()
> yields a file with the same ->f_inode every time it is called.
>
> Again, ->f_inode of those things is shared to hell and back,
> very much by design. You can't modify its ->i_op or anything
> other field, for that matter. No information can be stored
> in that thing - you are only allowed to use the object you've
> passed via 'priv' argument.
Yeah, we found that out the hard way. Is using the "secure" variant to get a
per-file inode a sane approach, or is that abuse that's going to bite us too?
/*
* Use the so called "secure" variant, which creates a unique inode
* instead of reusing a single inode. Each guest_memfd instance needs
* its own inode to track the size, flags, etc.
*/
file = anon_inode_getfile_secure(anon_name, &kvm_gmem_fops, gmem,
O_RDWR, NULL);
Powered by blists - more mailing lists