Re: [PATCH net-next] bridge: Fix setting a flag in br_fill_ifvlaninfo_range().

[roopa <roopa@cumulusnetworks.com>]

On 5/15/15, 11:45 AM, Arad, Ronen wrote:
>
>> -----Original Message-----
>> From: Rosen, Rami
>> Sent: Friday, May 15, 2015 7:27 AM
>> To: davem@davemloft.net
>> Cc: netdev@vger.kernel.org; roopa@cumulusnetworks.com; Arad, Ronen; Rosen,
>> Rami
>> Subject: [PATCH net-next] bridge: Fix setting a flag in
>> br_fill_ifvlaninfo_range().
>>
>> This patch fixes setting of vinfo.flags in the br_fill_ifvlaninfo_range(). The
>> assignment of vinfo.flags &= ~BRIDGE_VLAN_INFO_RANGE_BEGIN has no effect
>> without
>> this patch, as vinfo.flags value is overriden by the
>> vinfo.flags = flags | BRIDGE_VLAN_INFO_RANGE_END assignement,  which follows
>> it immedialtely.
>>
>> Signed-off-by: Rami Rosen <rami.rosen@intel.com>
>> ---
>> net/bridge/br_netlink.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/net/bridge/br_netlink.c b/net/bridge/br_netlink.c
>> index 6b67ed3..fec65bb 100644
>> --- a/net/bridge/br_netlink.c
>> +++ b/net/bridge/br_netlink.c
>> @@ -167,7 +167,7 @@ static int br_fill_ifvlaninfo_range(struct sk_buff *skb,
>> u16 vid_start,
>> 		vinfo.flags &= ~BRIDGE_VLAN_INFO_RANGE_BEGIN;
>>
>> 		vinfo.vid = vid_end;
>> -		vinfo.flags = flags | BRIDGE_VLAN_INFO_RANGE_END;
>> +		vinfo.flags &= flags | BRIDGE_VLAN_INFO_RANGE_END;
>> 		if (nla_put(skb, IFLA_BRIDGE_VLAN_INFO,
>> 			    sizeof(vinfo), &vinfo))
>> 			goto nla_put_failure;
>
> The first part of " &= flags " has no impact as it reverts vinfo.flags to
> just the bits set in the flags argument. This is exactly what
> "vinfo.flags &= ~BRIDGE_VLAN_INFO_RANGE_BEGIN" does.
>
> If original code was intended to indicate that the same flags are used for
> both start and end attributes except for the RANGE_BEGIN/RANGE_END flags
> it would be sufficient to keep the line that clears RANGE_BEGIN bit and
> OR vinfo.flags with RANGE_END for the end attribute:
>
> vinfo.flags |= BRIDGE_VLAN_INFO_RANGE_END;
>
> If we want the code to show what is actually been written into each
> attribute, the fix should be to just delete the line which clears
> RANGE_BEGIN flag.
I would prefer the fix to just delete the line that clears RANGE_BEGIN 
because as you and this patch points out that it is a no-op.

>
> In either case the code as exists today is working as expected. A fix is
> only needed to improve readability.
agreed.

In any case Acked-by: Roopa Prabhu <roopa@cumulusnetworks.com>

thanks!