Re: [PATCH] qed: fix uninit pointer read in qed_mcp_nvm_info_populate()

[Simon Horman <horms@kernel.org>]

On Fri, Dec 13, 2024 at 01:13:12PM +0100, Gianfranco Trad wrote:
> On 12/12/24 18:04, Simon Horman wrote:
> > On Wed, Dec 11, 2024 at 02:40:42PM +0100, Gianfranco Trad wrote:
> > > Coverity reports an uninit pointer read in qed_mcp_nvm_info_populate().
> > > If qed_mcp_bist_nvm_get_num_images() returns -EOPNOTSUPP, this leads to
> > > jump to label out with nvm_info.image_att being uninit while assigning it
> > > to p_hwfn->nvm_info.image_att.
> > > Add check on rc against -EOPNOTSUPP to avoid such uninit pointer read.
> > > 
> > > Closes: https://scan5.scan.coverity.com/#/project-view/63204/10063?selectedIssue=1636666
> > > Signed-off-by: Gianfranco Trad <gianf.trad@gmail.com>
> > > ---
> > > Note:
> > > - Fixes: tag should be "7a0ea70da56e net/qed: allow old cards not supporting "num_images" to work" ?
> > >   drivers/net/ethernet/qlogic/qed/qed_mcp.c | 2 +-
> > >   1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/drivers/net/ethernet/qlogic/qed/qed_mcp.c b/drivers/net/ethernet/qlogic/qed/qed_mcp.c
> > > index b45efc272fdb..127943b39f61 100644
> > > --- a/drivers/net/ethernet/qlogic/qed/qed_mcp.c
> > > +++ b/drivers/net/ethernet/qlogic/qed/qed_mcp.c
> > > @@ -3387,7 +3387,7 @@ int qed_mcp_nvm_info_populate(struct qed_hwfn *p_hwfn)
> > >   	}
> > >   out:
> > >   	/* Update hwfn's nvm_info */
> > > -	if (nvm_info.num_images) {
> > > +	if (nvm_info.num_images && rc != -EOPNOTSUPP) {
> > >   		p_hwfn->nvm_info.num_images = nvm_info.num_images;
> > >   		kfree(p_hwfn->nvm_info.image_att);
> > >   		p_hwfn->nvm_info.image_att = nvm_info.image_att;
> > 
> 
> Hi Simon,
> 
> > Are you sure that nvm_info.num_images can be non-zero if rc == -EOPNOTSUPP?
> > 
> 
> In the coverity report, the static analyzer is able to take the true branch
> on nvm_info.num_images. I didn't physically reproduce this logical state as
> I don't possess the matching hardware.
> 
> > The cited commit state:
> > 
> >      Commit 43645ce03e00 ("qed: Populate nvm image attribute shadow.")
> >      added support for populating flash image attributes, notably
> >      "num_images". However, some cards were not able to return this
> >      information. In such cases, the driver would return EINVAL, causing the
> >      driver to exit.
> > 
> >      Add check to return EOPNOTSUPP instead of EINVAL when the card is not
> >      able to return these information. The caller function already handles
> >      EOPNOTSUPP without error.
> > 
> > So I would expect that nvm_info.num_images is 0.
> > 
> > If not, perhaps an alternate fix is to make that so, either by setting
> > it in qed_mcp_bist_nvm_get_num_images, or where the return value of
> > qed_mcp_bist_nvm_get_num_images is checked (just before the goto out).
> > 
> 
> Makes sense, so something like this I suppose:
> 
> --- a/drivers/net/ethernet/qlogic/qed/qed_mcp.c
> +++ b/drivers/net/ethernet/qlogic/qed/qed_mcp.c
> @@ -3301,8 +3301,10 @@ int qed_mcp_bist_nvm_get_num_images(struct qed_hwfn
> *p_hwfn,
>   	if (rc)
>   		return rc;
> 
> -	if (((rsp & FW_MSG_CODE_MASK) == FW_MSG_CODE_UNSUPPORTED))
> +	if (((rsp & FW_MSG_CODE_MASK) == FW_MSG_CODE_UNSUPPORTED)) {
> +		*num_images = 0;
>   		rc = -EOPNOTSUPP;
> +	}
> 
> Or the second option you stated.

Yes, that is what I was thinking.
But as it is a side effect, and thus hidden slightly,
on reflection perhaps the second option is better. IDK.

> > And, in any case I think some testing is in order.
> 
> I strongly agree. Let me know if I can help more with this.
> 
> Thanks for your time,
> --Gian
>