netdev - Re: [PATCH v4 bpf-next 01/10] ice: compress branches in ice_set

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 20 Jun 2022 11:48:02 +0200
From:   Maciej Fijalkowski <maciej.fijalkowski@...el.com>
To:     John Fastabend <john.fastabend@...il.com>
Cc:     bpf@...r.kernel.org, ast@...nel.org, daniel@...earbox.net,
        netdev@...r.kernel.org, magnus.karlsson@...el.com,
        bjorn@...nel.org, kuba@...nel.org,
        Alexandr Lobakin <alexandr.lobakin@...el.com>
Subject: Re: [PATCH v4 bpf-next 01/10] ice: compress branches in
 ice_set_features()

On Fri, Jun 17, 2022 at 06:46:35PM -0700, John Fastabend wrote:
> Maciej Fijalkowski wrote:
> > Instead of rather verbose comparison of current netdev->features bits vs
> > the incoming ones from user, let us compress them by a helper features
> > set that will be the result of netdev->features XOR features. This way,
> > current, extensive branches:
> > 
> > 	if (features & NETIF_F_BIT && !(netdev->features & NETIF_F_BIT))
> > 		set_feature(true);
> > 	else if (!(features & NETIF_F_BIT) && netdev->features & NETIF_F_BIT)
> > 		set_feature(false);
> > 
> > can become:
> > 
> > 	netdev_features_t changed = netdev->features ^ features;
> > 
> > 	if (changed & NETIF_F_BIT)
> > 		set_feature(!!(features & NETIF_F_BIT));
> > 
> > This is nothing new as currently several other drivers use this
> > approach, which I find much more convenient.
> 
> Looks good couple nits below. Up to you if you want to follow through
> on them or not I don't have a strong opinion. For what its worth the
> other intel drivers also do the 'netdev->features ^ features'
> assignment.
> 
> Acked-by: John Fastabend <john.fastabend@...il.com>
> 
> > 
> > CC: Alexandr Lobakin <alexandr.lobakin@...el.com>
> > Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@...el.com>
> > ---
> >  drivers/net/ethernet/intel/ice/ice_main.c | 40 +++++++++++------------
> >  1 file changed, 19 insertions(+), 21 deletions(-)
> > 
> > diff --git a/drivers/net/ethernet/intel/ice/ice_main.c b/drivers/net/ethernet/intel/ice/ice_main.c
> > index e1cae253412c..23d1b1fc39fb 100644
> > --- a/drivers/net/ethernet/intel/ice/ice_main.c
> > +++ b/drivers/net/ethernet/intel/ice/ice_main.c
> > @@ -5910,44 +5910,41 @@ ice_set_vlan_features(struct net_device *netdev, netdev_features_t features)
> >  static int
> >  ice_set_features(struct net_device *netdev, netdev_features_t features)
> >  {
> > +	netdev_features_t changed = netdev->features ^ features;
> >  	struct ice_netdev_priv *np = netdev_priv(netdev);
> >  	struct ice_vsi *vsi = np->vsi;
> >  	struct ice_pf *pf = vsi->back;
> >  	int ret = 0;
> >  
> >  	/* Don't set any netdev advanced features with device in Safe Mode */
> > -	if (ice_is_safe_mode(vsi->back)) {
> > -		dev_err(ice_pf_to_dev(vsi->back), "Device is in Safe Mode - not enabling advanced netdev features\n");
> > +	if (ice_is_safe_mode(pf)) {
> > +		dev_err(ice_pf_to_dev(vsi->back),
> 
> bit of nitpicking but if you use pf in the 'if' above why not use it here
> as well and save a few keys. Also matches below then.

Hey John thanks for all of the input!
You're right above, I just forgot to change it to use 'pf' in
ice_pf_to_dev(). Will fix.

> 
> > +			"Device is in Safe Mode - not enabling advanced netdev features\n");
> >  		return ret;
> >  	}
> >  
> >  	/* Do not change setting during reset */
> >  	if (ice_is_reset_in_progress(pf->state)) {
> > -		dev_err(ice_pf_to_dev(vsi->back), "Device is resetting, changing advanced netdev features temporarily unavailable.\n");
> > +		dev_err(ice_pf_to_dev(pf),
> > +			"Device is resetting, changing advanced netdev features temporarily unavailable.\n");
> >  		return -EBUSY;
> >  	}
> >  
> 
> [...]
> 
> > @@ -5956,11 +5953,12 @@ ice_set_features(struct net_device *netdev, netdev_features_t features)
> >  		return -EACCES;
> >  	}
> >  
> > -	if ((features & NETIF_F_HW_TC) &&
> > -	    !(netdev->features & NETIF_F_HW_TC))
> > -		set_bit(ICE_FLAG_CLS_FLOWER, pf->flags);
> > -	else
> > -		clear_bit(ICE_FLAG_CLS_FLOWER, pf->flags);
> > +	if (changed & NETIF_F_HW_TC) {
> > +		bool ena = !!(features & NETIF_F_HW_TC);
> > +
> > +		ena ? set_bit(ICE_FLAG_CLS_FLOWER, pf->flags) :
> > +		      clear_bit(ICE_FLAG_CLS_FLOWER, pf->flags);
> > +	}
> 
> Just a note you changed the logic slightly here. Above you always
> clear the bit. But, it looks like it doesn't matter caveat being
> I don't know what might happen in hardware.

I believe that previous logic was wrong due to the fact that in case
NETIF_F_HW_TC was set but in current ice_set_features() was not modified
then this setting would be lost.

> 
> >  
> >  	return 0;
> >  }
> > -- 
> > 2.27.0
> > 
> 
>