Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230109213809.418135-1-tjmercier@google.com> <20230109213809.418135-4-tjmercier@google.com>
 <20230110014720.281-1-hdanton@sina.com>
In-Reply-To: <20230110014720.281-1-hdanton@sina.com>
From:   "T.J. Mercier" <tjmercier@google.com>
Date:   Tue, 10 Jan 2023 13:20:42 -0800
Message-ID: <CABdmKX11WP-ijLbU34Y7GG21NtqsCyMVyhnkxMMnL_hG7+TV6g@mail.gmail.com>
Subject: Re: [PATCH 3/4] binder: Add flags to relinquish ownership of fds
To:     Hillf Danton <hdanton@sina.com>
Cc:     Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        linux-kernel@vger.kernel.org, dri-devel@lists.freedesktop.org,
        daniel.vetter@ffwll.ch, Tejun Heo <tj@kernel.org>,
        linux-media@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, Jan 9, 2023 at 6:07 PM Hillf Danton <hdanton@sina.com> wrote:
>
> On 9 Jan 2023 21:38:06 +0000 T.J. Mercier <tjmercier@google.com>
> >
> > @@ -2275,6 +2276,26 @@ static int binder_translate_fd(u32 fd, binder_size_t fd_offset,
> >               goto err_security;
> >       }
> >
> > +     if (IS_ENABLED(CONFIG_MEMCG) && (flags & BINDER_FD_FLAG_XFER_CHARGE)) {
> > +             struct dma_buf *dmabuf;
> > +
> > +             if (unlikely(!is_dma_buf_file(file))) {
> > +                     binder_user_error(
> > +                             "%d:%d got transaction with XFER_CHARGE for non-dmabuf fd, %d\n",
> > +                             proc->pid, thread->pid, fd);
> > +                     ret = -EINVAL;
> > +                     goto err_dmabuf;
> > +             }
>
> It barely makes sense to expose is_dma_buf_file() only for this.
> > +
> > +             dmabuf = file->private_data;
> > +             ret = dma_buf_transfer_charge(dmabuf, target_proc->tsk);
> > +             if (ret) {
> > +                     pr_warn("%d:%d Unable to transfer DMA-BUF fd charge to %d\n",
> > +                             proc->pid, thread->pid, target_proc->pid);
> > +                     goto err_xfer;
> > +             }
> > +     }
> > +
>
> This whole hunk should go to dma-buf instead by adding change to
> dma_buf_transfer_charge() for instance.

Fair enough, will change this for v2. I think we'd still want to
distinguish between the two failure modes for logging purposes, so
I'll use the return value of dma_buf_transfer_charge to do that.