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;
Date:   Wed, 23 Mar 2022 09:23:27 -0400
From:   "Michael S. Tsirkin" <mst@redhat.com>
To:     Keir Fraser <keirf@google.com>
Cc:     Jason Wang <jasowang@redhat.com>, kernel-team@android.com,
        virtualization <virtualization@lists.linux-foundation.org>,
        linux-kernel <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v2] virtio: pci: sanity check bar indexes
Message-ID: <20220323092254-mutt-send-email-mst@kernel.org>
References: <20220322151952.2950143-1-keirf@google.com>
 <CACGkMEubcU4rFVem7neKYb-qT3TQUN802bbLNq7vh+y8gdD5AA@mail.gmail.com>
 <20220323075030-mutt-send-email-mst@kernel.org>
 <Yjse870sgpcMGsT6@google.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <Yjse870sgpcMGsT6@google.com>
Precedence: bulk

On Wed, Mar 23, 2022 at 01:21:55PM +0000, Keir Fraser wrote:
> On Wed, Mar 23, 2022 at 08:01:42AM -0400, Michael S. Tsirkin wrote:
> > On Wed, Mar 23, 2022 at 03:57:59PM +0800, Jason Wang wrote:
> > > On Tue, Mar 22, 2022 at 11:20 PM Keir Fraser <keirf@google.com> wrote:
> > > >
> > > > The bar index is used as an index into the device's resource list
> > > > and should be checked as within range for a standard bar.
> > > >
> > > > Also clean up an existing check to consistently use PCI_STD_NUM_BARS.
> > > >
> > > > Signed-off-by: Keir Fraser <keirf@google.com>
> > > > ---
> > > >  drivers/virtio/virtio_pci_modern.c     | 10 ++++++++--
> > > >  drivers/virtio/virtio_pci_modern_dev.c |  8 +++++++-
> > > >  2 files changed, 15 insertions(+), 3 deletions(-)
> > > >
> > > > diff --git a/drivers/virtio/virtio_pci_modern.c b/drivers/virtio/virtio_pci_modern.c
> > > > index 5455bc041fb6..84bace98dff5 100644
> > > > --- a/drivers/virtio/virtio_pci_modern.c
> > > > +++ b/drivers/virtio/virtio_pci_modern.c
> > > > @@ -293,7 +293,7 @@ static int virtio_pci_find_shm_cap(struct pci_dev *dev, u8 required_id,
> > > >
> > > >         for (pos = pci_find_capability(dev, PCI_CAP_ID_VNDR); pos > 0;
> > > >              pos = pci_find_next_capability(dev, pos, PCI_CAP_ID_VNDR)) {
> > > > -               u8 type, cap_len, id;
> > > > +               u8 type, cap_len, id, res_bar;
> > > >                 u32 tmp32;
> > > >                 u64 res_offset, res_length;
> > > >
> > > > @@ -317,7 +317,12 @@ static int virtio_pci_find_shm_cap(struct pci_dev *dev, u8 required_id,
> > > >
> > > >                 /* Type, and ID match, looks good */
> > > >                 pci_read_config_byte(dev, pos + offsetof(struct virtio_pci_cap,
> > > > -                                                        bar), bar);
> > > > +                                                        bar), &res_bar);
> > > > +               if (res_bar >= PCI_STD_NUM_BARS) {
> > > > +                       dev_err(&dev->dev, "%s: shm cap with bad bar: %d\n",
> > > > +                               __func__, res_bar);
> > > > +                       continue;
> > > > +               }
> > > >
> > > >                 /* Read the lower 32bit of length and offset */
> > > >                 pci_read_config_dword(dev, pos + offsetof(struct virtio_pci_cap,
> > 
> > In fact, the spec says such BAR values are reserved, not bad, so
> > the capabiluty should be ignored, they should not cause the driver to error out
> > or print errors.
> 
> Ah yes, so I see. It makes sense then to silently ignore the capability and print nothing.
> I will fix it.
> 
> > > > @@ -337,6 +342,7 @@ static int virtio_pci_find_shm_cap(struct pci_dev *dev, u8 required_id,
> > > >                                                      length_hi), &tmp32);
> > > >                 res_length |= ((u64)tmp32) << 32;
> > > >
> > > > +               *bar = res_bar;
> > > >                 *offset = res_offset;
> > > >                 *len = res_length;
> > > >
> > > > diff --git a/drivers/virtio/virtio_pci_modern_dev.c b/drivers/virtio/virtio_pci_modern_dev.c
> > > > index e8b3ff2b9fbc..a6911d1e212a 100644
> > > > --- a/drivers/virtio/virtio_pci_modern_dev.c
> > > > +++ b/drivers/virtio/virtio_pci_modern_dev.c
> > > > @@ -35,6 +35,12 @@ vp_modern_map_capability(struct virtio_pci_modern_device *mdev, int off,
> > > >         pci_read_config_dword(dev, off + offsetof(struct virtio_pci_cap, length),
> > > >                               &length);
> > > >
> > > > +       if (bar >= PCI_STD_NUM_BARS) {
> > > > +               dev_err(&dev->dev,
> > > > +                       "virtio_pci: bad capability bar %u\n", bar);
> > 
> > In fact, I would say the issue is less that bar is reserved.
> > The real issue is that the value apparently changed since
> > we read it the first time. I think it's a good idea to
> > reflect that in the message. Maybe find_capability should return
> > the capability structure so we don't need to re-read it from
> > the device?
> 
> I will have a look and fix it up one way or the other, and respin
> this patch.
> 
>  Thanks,
>   Keir

BTW avoiding extra reads is good for start up speed. This is slow path,
but still.

> > > > +               return NULL;
> > > > +       }
> > > > +
> > > >         if (length <= start) {
> > > >                 dev_err(&dev->dev,
> > > >                         "virtio_pci: bad capability len %u (>%u expected)\n",
> > > > @@ -120,7 +126,7 @@ static inline int virtio_pci_find_capability(struct pci_dev *dev, u8 cfg_type,
> > > >                                      &bar);
> > > >
> > > >                 /* Ignore structures with reserved BAR values */
> > > > -               if (bar > 0x5)
> > > > +               if (bar >= PCI_STD_NUM_BARS)
> > > >                         continue;
> > > 
> > > Just notice that the spec said:
> > > 
> > > "
> > > values 0x0 to 0x5 specify a Base Address register (BAR) belonging to
> > > the function located beginning at 10h in PCI Configuration Space and
> > > used to map the structure into Memory or I/O Space. The BAR is
> > > permitted to be either 32-bit or 64-bit, it can map Memory Space or
> > > I/O Space.
> > > 
> > > Any other value is reserved for future use.
> > > "
> > > So we probably need to stick 0x5 instead of 0x6 (PCI_STD_NUM_BARS) for
> > > this and other places.
> > > 
> > > Thanks
> > 
> > It does not matter much IMHO, the reason spec uses 0 to 0x5 is precisely
> > because that's the standard number of BARs. Both ways work as long as we
> > are consistent, and I guess PCI_STD_NUM_BARS might be preferable since
> > people tend to copy paste values.
> > 
> > > >
> > > >                 if (type == cfg_type) {
> > > > --
> > > > 2.35.1.894.gb6a874cedc-goog
> > > >
> >