Re: [PATCH v4 2/2] virtio: fix PCI config err handling

[Luca Boccassi <bluca@debian.org>]

On Mon, 2018-08-27 at 13:29 +0800, Tiwei Bie wrote:
> On Fri, Aug 24, 2018 at 06:14:20PM +0100, Luca Boccassi wrote:
> > From: Brian Russell <brussell@brocade.com>
> > 
> > In virtio_read_caps and vtpci_msix_detect, rte_pci_read_config
> > returns
> > the number of bytes read from PCI config or < 0 on error.
> > If less than the expected number of bytes are read then log the
> > failure and return rather than carrying on with garbage.
> > 
> > Fixes: 6ba1f63b5ab0 ("virtio: support specification 1.0")
> > 
> > Signed-off-by: Brian Russell <brussell@brocade.com>
> > Signed-off-by: Luca Boccassi <bluca@debian.org>
> > ---
> > v2: handle additional rte_pci_read_config incomplete reads
> > v3: do not handle rte_pci_read_config of virtio cap, added in v2,
> >     as it's less clear what the right thing to do there is
> > v4: do a more robust check - first check what the vendor is, and
> >     skip the cap entirely if it's not what we are looking for.
> > 
> >  drivers/net/virtio/virtio_pci.c | 57 ++++++++++++++++++++++++-----
> > ----
> >  1 file changed, 42 insertions(+), 15 deletions(-)
> > 
> > diff --git a/drivers/net/virtio/virtio_pci.c
> > b/drivers/net/virtio/virtio_pci.c
> > index 6bd22e54a6..cfefa9789b 100644
> > --- a/drivers/net/virtio/virtio_pci.c
> > +++ b/drivers/net/virtio/virtio_pci.c
> > @@ -567,16 +567,30 @@ virtio_read_caps(struct rte_pci_device *dev,
> > struct virtio_hw *hw)
> >  	}
> >  
> >  	ret = rte_pci_read_config(dev, &pos, 1,
> > PCI_CAPABILITY_LIST);
> > -	if (ret < 0) {
> > -		PMD_INIT_LOG(DEBUG, "failed to read pci capability
> > list");
> > +	if (ret != 1) {
> > +		PMD_INIT_LOG(DEBUG,
> > +			     "failed to read pci capability list,
> > ret %d", ret);
> >  		return -1;
> >  	}
> >  
> >  	while (pos) {
> > +		ret = rte_pci_read_config(dev, &cap, 2, pos);
> > +		if (ret != 2) {
> > +			PMD_INIT_LOG(DEBUG,
> > +				     "failed to read pci cap at
> > pos: %x ret %d",
> > +				     pos, ret);
> > +			break;
> > +		}
> > +		if (cap.cap_vndr != PCI_CAP_ID_MSIX &&
> > +				cap.cap_vndr != PCI_CAP_ID_VNDR) {
> > +			goto next;
> > +		}
> > +
> >  		ret = rte_pci_read_config(dev, &cap, sizeof(cap),
> > pos);
> > -		if (ret < 0) {
> > -			PMD_INIT_LOG(ERR,
> > -				"failed to read pci cap at pos:
> > %x", pos);
> > +		if (ret != sizeof(cap)) {
> > +			PMD_INIT_LOG(DEBUG,
> > +				     "failed to read pci cap at
> > pos: %x ret %d",
> > +				     pos, ret);
> >  			break;
> >  		}
> >  
> 
> It seems that I didn't make myself clear in my previous
> comments. I mean it's better to handle MSIX cap and virtio
> cap respectively in this function. Currently we're always
> reading them as virtio caps. As we are strictly requiring
> that _read_config() should return the required number of
> bytes, it's not perfect to require it to return "virtio
> cap size" of bytes while we're trying to read a MSIX cap.
> So please change the code to something similar to this:

Sorry, though you meant in the vtpci_msix_detect function, which I
changed. Fixed in v5.

> > @@ -689,25 +703,38 @@ enum virtio_msix_status
> >  vtpci_msix_detect(struct rte_pci_device *dev)
> >  {
> >  	uint8_t pos;
> > -	struct virtio_pci_cap cap;
> >  	int ret;
> >  
> >  	ret = rte_pci_read_config(dev, &pos, 1,
> > PCI_CAPABILITY_LIST);
> > -	if (ret < 0) {
> > -		PMD_INIT_LOG(DEBUG, "failed to read pci capability
> > list");
> > +	if (ret != 1) {
> > +		PMD_INIT_LOG(DEBUG,
> > +			     "failed to read pci capability list,
> > ret %d", ret);
> >  		return VIRTIO_MSIX_NONE;
> >  	}
> >  
> >  	while (pos) {
> > -		ret = rte_pci_read_config(dev, &cap, sizeof(cap),
> > pos);
> > -		if (ret < 0) {
> > -			PMD_INIT_LOG(ERR,
> > -				"failed to read pci cap at pos:
> > %x", pos);
> > +		uint8_t cap[2];
> > +
> > +		ret = rte_pci_read_config(dev, cap, sizeof(cap),
> > pos);
> > +		if (ret != sizeof(cap)) {
> > +			PMD_INIT_LOG(DEBUG,
> > +				     "failed to read pci cap at
> > pos: %x ret %d",
> > +				     pos, ret);
> >  			break;
> >  		}
> >  
> > -		if (cap.cap_vndr == PCI_CAP_ID_MSIX) {
> > -			uint16_t flags = ((uint16_t *)&cap)[1];
> > +		if (cap[0] == PCI_CAP_ID_MSIX) {
> > +			uint16_t flags;
> > +
> > +			ret = rte_pci_read_config(dev, &flags,
> > sizeof(flags),
> > +					pos + sizeof(cap));
> > +			if (ret != sizeof(flags)) {
> > +				PMD_INIT_LOG(DEBUG,
> > +					     "failed to read pci
> > cap at pos:"
> > +					     " %x ret %d", pos +
> > sizeof(cap),
> 
> There is a build error:
> 
> In file included from drivers/net/virtio/virtio_pci.c:15:
> drivers/net/virtio/virtio_pci.c: In function ‘vtpci_msix_detect’:
> drivers/net/virtio/virtio_logs.h:13:3: error: format ‘%x’ expects
> argument of type ‘unsigned int’, but argument 5 has type ‘long
> unsigned int’ [-Werror=format=]
>    "%s(): " fmt "\n", __func__, ##args)
>    ^~~~~~~~
> drivers/net/virtio/virtio_pci.c:732:5: note: in expansion of macro
> ‘PMD_INIT_LOG’
>      PMD_INIT_LOG(DEBUG,
>      ^~~~~~~~~~~~
> drivers/net/virtio/virtio_pci.c:734:14: note: format string is
> defined here
>            " %x ret %d", pos + sizeof(cap),
>              ~^
>              %lx

Ok, changed to %lx - which GCC version was that with? Didn't get any
warning with 6.3 on Debian 9.

-- 
Kind regards,
Luca Boccassi