Re: [PATCH net v4] net: nfc: nci: Fix parameter validation for packet data

[Michael Thalmeier]

Am 04.01.26 um 19:13 schrieb Jakub Kicinski:
> On Tue, 23 Dec 2025 08:25:52 +0100 Michael Thalmeier wrote:
> > diff --git a/net/nfc/nci/ntf.c b/net/nfc/nci/ntf.c
> > index 418b84e2b260..a5cafcd10cc3 100644
> > --- a/net/nfc/nci/ntf.c
> > +++ b/net/nfc/nci/ntf.c
>
> > @@ -380,6 +384,10 @@ static int nci_rf_discover_ntf_packet(struct nci_dev *ndev,
> >   	pr_debug("rf_tech_specific_params_len %d\n",
> >   		 ntf.rf_tech_specific_params_len);
> >   
> > +	if (skb->len < (data - skb->data) +
> > +			ntf.rf_tech_specific_params_len + sizeof(ntf.ntf_type))
> > +		return -EINVAL;
>
> Are we validating ntf.rf_tech_specific_params_len against the
> extraction logic in nci_extract_rf_params_nfca_passive_poll()
> and friends?

You are right. The current patch is only validating that the received 
packet is consistent in the way that the rf_tech_specific_params_len 
number of bytes is also contained in the buffer.

There is currently no code that validates that 
nci_extract_rf_params_nfca_passive_poll and friends only access the 
given number of bytes in their logic.
And to be frank, I do not know how to implement this without either 
cluttering the code with validation logic or re-implementing half the 
parsing logic for length validation.