Date:   Wed, 8 Dec 2021 15:34:51 -0500
From:   "Michael S. Tsirkin" <mst@redhat.com>
To:     Mike Christie <michael.christie@oracle.com>
Cc:     geert@linux-m68k.org, vverma@digitalocean.com, hdanton@sina.com,
        hch@infradead.org, stefanha@redhat.com, jasowang@redhat.com,
        sgarzare@redhat.com, virtualization@lists.linux-foundation.org,
        christian.brauner@ubuntu.com, axboe@kernel.dk,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH V6 01/10] Use copy_process in vhost layer
Message-ID: <20211208153420-mutt-send-email-mst@kernel.org>
References: <20211129194707.5863-1-michael.christie@oracle.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20211129194707.5863-1-michael.christie@oracle.com>
Precedence: bulk

On Mon, Nov 29, 2021 at 01:46:57PM -0600, Mike Christie wrote:
> The following patches made over Linus's tree, allow the vhost layer to do
> a copy_process on the thread that does the VHOST_SET_OWNER ioctl like how
> io_uring does a copy_process against its userspace app. This allows the
> vhost layer's worker threads to inherit cgroups, namespaces, address
> space, etc and this worker thread will also be accounted for against that
> owner/parent process's RLIMIT_NPROC limit.
> 
> If you are not familiar with qemu and vhost here is more detailed
> problem description:
> 
> Qemu will create vhost devices in the kernel which perform network, SCSI,
> etc IO and management operations from worker threads created by the
> kthread API. Because the kthread API does a copy_process on the kthreadd
> thread, the vhost layer has to use kthread_use_mm to access the Qemu
> thread's memory and cgroup_attach_task_all to add itself to the Qemu
> thread's cgroups.
> 
> The problem with this approach is that we then have to add new functions/
> args/functionality for every thing we want to inherit. I started doing
> that here:
> 
> https://lkml.org/lkml/2021/6/23/1233
> 
> for the RLIMIT_NPROC check, but it seems it might be easier to just
> inherit everything from the beginning, becuase I'd need to do something
> like that patch several times.


So who's merging this? Me? Did all patches get acked by appropriate
maintainers?

> V6:
> - Rename kernel_worker to user_worker and fix prefixes.
> - Add better patch descriptions.
> V5:
> - Handle kbuild errors by building patchset against current kernel that
>   has all deps merged. Also add patch to remove create_io_thread code as
>   it's not used anymore.
> - Rebase patchset against current kernel and handle a new vm PF_IO_WORKER
>   case added in 5.16-rc1.
> - Add PF_USER_WORKER flag so we can check it later after the initial
>   thread creation for the wake up, vm and singal cses.
> - Added patch to auto reap the worker thread.
> V4:
> - Drop NO_SIG patch and replaced with Christian's SIG_IGN patch.
> - Merged Christian's kernel_worker_flags_valid helpers into patch 5 that
>   added the new kernel worker functions.
> - Fixed extra "i" issue.
> - Added PF_USER_WORKER flag and added check that kernel_worker_start users
>   had that flag set. Also dropped patches that passed worker flags to
>   copy_thread and replaced with PF_USER_WORKER check.
> V3:
> - Add parentheses in p->flag and work_flags check in copy_thread.
> - Fix check in arm/arm64 which was doing the reverse of other archs
>   where it did likely(!flags) instead of unlikely(flags).
> V2:
> - Rename kernel_copy_process to kernel_worker.
> - Instead of exporting functions, make kernel_worker() a proper
>   function/API that does common work for the caller.
> - Instead of adding new fields to kernel_clone_args for each option
>   make it flag based similar to CLONE_*.
> - Drop unused completion struct in vhost.
> - Fix compile warnings by merging vhost cgroup cleanup patch and
>   vhost conversion patch.
> ~                              
>