[dpdk-dev] bonding: fix crash when no slave devices

  If a bonded device is created when there are no slave devices
there is loop in bond_ethdev_promiscous_enable() which results
in a segmentation fault.
I have applied a similar fix to bond_ethdev_promiscous_disable()
where a similar loop could occur.

Fixes: 2efb58cbab6e ("bond: new link bonding library")
Signed-off-by: Bernard Iremonger <bernard.iremonger@intel.com>
---
 drivers/net/bonding/rte_eth_bond_pmd.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
  

On 2/23/2016 12:13 PM, Bernard Iremonger wrote:
> If a bonded device is created when there are no slave devices
> there is loop in bond_ethdev_promiscous_enable() which results
> in a segmentation fault.
> I have applied a similar fix to bond_ethdev_promiscous_disable()
> where a similar loop could occur.
> 
> Fixes: 2efb58cbab6e ("bond: new link bonding library")
> Signed-off-by: Bernard Iremonger <bernard.iremonger@intel.com>
> ---
>  drivers/net/bonding/rte_eth_bond_pmd.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/bonding/rte_eth_bond_pmd.c b/drivers/net/bonding/rte_eth_bond_pmd.c
> index b63c886..78972fc 100644
> --- a/drivers/net/bonding/rte_eth_bond_pmd.c
> +++ b/drivers/net/bonding/rte_eth_bond_pmd.c
> @@ -1870,7 +1870,8 @@ bond_ethdev_promiscuous_enable(struct rte_eth_dev *eth_dev)
>  	case BONDING_MODE_TLB:
>  	case BONDING_MODE_ALB:
>  	default:
> -		rte_eth_promiscuous_enable(internals->current_primary_port);
> +		if (internals->slave_count > 0)
> +			rte_eth_promiscuous_enable(internals->current_primary_port);
>  	}
>  }
>  
> @@ -1898,7 +1899,8 @@ bond_ethdev_promiscuous_disable(struct rte_eth_dev *dev)
>  	case BONDING_MODE_TLB:
>  	case BONDING_MODE_ALB:
>  	default:
> -		rte_eth_promiscuous_disable(internals->current_primary_port);
> +		if (internals->slave_count > 0)
> +			rte_eth_promiscuous_disable(internals->current_primary_port);
>  	}
>  }
>  
> 
Hi Bernard,

The reason of this crash is when there is no slave, the value of
current_primary_port is 0, which is valid port_id, is this correct?

Does it make sense, instead of slave_count check, to make default
current_primary_port value a non valid port_id, like -1, so
is_valid_port() check catches it to prevents crash? For this and any
other cases.

Thanks,
ferruh
  

Hi Ferruh,
> -----Original Message-----
> From: Yigit, Ferruh
> Sent: Friday, March 4, 2016 5:14 PM
> To: Iremonger, Bernard <bernard.iremonger@intel.com>; dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH] bonding: fix crash when no slave devices
> 
> On 2/23/2016 12:13 PM, Bernard Iremonger wrote:
> > If a bonded device is created when there are no slave devices there is
> > loop in bond_ethdev_promiscous_enable() which results in a
> > segmentation fault.
> > I have applied a similar fix to bond_ethdev_promiscous_disable() where
> > a similar loop could occur.
> >
> > Fixes: 2efb58cbab6e ("bond: new link bonding library")
> > Signed-off-by: Bernard Iremonger <bernard.iremonger@intel.com>
> > ---
> >  drivers/net/bonding/rte_eth_bond_pmd.c | 6 ++++--
> >  1 file changed, 4 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/net/bonding/rte_eth_bond_pmd.c
> > b/drivers/net/bonding/rte_eth_bond_pmd.c
> > index b63c886..78972fc 100644
> > --- a/drivers/net/bonding/rte_eth_bond_pmd.c
> > +++ b/drivers/net/bonding/rte_eth_bond_pmd.c
> > @@ -1870,7 +1870,8 @@ bond_ethdev_promiscuous_enable(struct
> rte_eth_dev *eth_dev)
> >  	case BONDING_MODE_TLB:
> >  	case BONDING_MODE_ALB:
> >  	default:
> > -		rte_eth_promiscuous_enable(internals-
> >current_primary_port);
> > +		if (internals->slave_count > 0)
> > +			rte_eth_promiscuous_enable(internals-
> >current_primary_port);
> >  	}
> >  }
> >
> > @@ -1898,7 +1899,8 @@ bond_ethdev_promiscuous_disable(struct
> rte_eth_dev *dev)
> >  	case BONDING_MODE_TLB:
> >  	case BONDING_MODE_ALB:
> >  	default:
> > -		rte_eth_promiscuous_disable(internals-
> >current_primary_port);
> > +		if (internals->slave_count > 0)
> > +			rte_eth_promiscuous_disable(internals-
> >current_primary_port);
> >  	}
> >  }
> >
> >
> Hi Bernard,
> 
> The reason of this crash is when there is no slave, the value of
> current_primary_port is 0, which is valid port_id, is this correct?

Yes.

 
> Does it make sense, instead of slave_count check, to make default
> current_primary_port value a non valid port_id, like -1, so
> is_valid_port() check catches it to prevents crash? For this and any other
> cases.
> 
> Thanks,
> Ferruh

I will try initializing current_primary_port to -1 and see if this helps.
Thanks for reviewing.

Regards,

Bernard.