Re: [dpdk-dev] [PATCH] i40e: Fix a vlan bug

["Qiu, Michael" <michael.qiu@intel.com>]

Hi Mark,

I think Huawei (huawei.xie@intel.com) has one patch set to fix this issue.

If your patch is totally different with him:

[dpdk-dev] [PATCH v4 0/2] lib/librte_pmd_i40e: set vlan filter fix

please ignore my comments :)

But you both calculation are different.

Thanks,
Michael
On 12/4/2014 5:51 PM, Chen Jing D(Mark) wrote:
> From: "Chen Jing D(Mark)" <jing.d.chen@intel.com>
>
> i40e uses an bitmap array to store those vlan tags that are set by
> application. In function i40e_set_vlan_filter, it stores vlan tag
> to wrong place. This change will fix it.
>
> Signed-off-by: Chen Jing D(Mark) <jing.d.chen@intel.com>
> ---
>  lib/librte_pmd_i40e/i40e_ethdev.c |   11 ++++-------
>  1 files changed, 4 insertions(+), 7 deletions(-)
>
> diff --git a/lib/librte_pmd_i40e/i40e_ethdev.c b/lib/librte_pmd_i40e/i40e_ethdev.c
> index 87e750a..cebc21d 100644
> --- a/lib/librte_pmd_i40e/i40e_ethdev.c
> +++ b/lib/librte_pmd_i40e/i40e_ethdev.c
> @@ -4163,8 +4163,8 @@ i40e_find_vlan_filter(struct i40e_vsi *vsi,
>  {
>  	uint32_t vid_idx, vid_bit;
>  
> -	vid_idx = (uint32_t) ((vlan_id >> 5) & 0x7F);
> -	vid_bit = (uint32_t) (1 << (vlan_id & 0x1F));
> +	vid_idx = (uint32_t)(vlan_id / I40E_UINT32_BIT_SIZE);
> +	vid_bit = (uint32_t)(1 << (vlan_id % I40E_UINT32_BIT_SIZE));
>  
>  	if (vsi->vfta[vid_idx] & vid_bit)
>  		return 1;
> @@ -4178,14 +4178,11 @@ i40e_set_vlan_filter(struct i40e_vsi *vsi,
>  {
>  	uint32_t vid_idx, vid_bit;
>  
> -#define UINT32_BIT_MASK      0x1F
> -#define VALID_VLAN_BIT_MASK  0xFFF
>  	/* VFTA is 32-bits size array, each element contains 32 vlan bits, Find the
>  	 *  element first, then find the bits it belongs to
>  	 */
> -	vid_idx = (uint32_t) ((vlan_id & VALID_VLAN_BIT_MASK) >>
> -		  sizeof(uint32_t));
> -	vid_bit = (uint32_t) (1 << (vlan_id & UINT32_BIT_MASK));
> +	vid_idx = (uint32_t)(vlan_id / I40E_UINT32_BIT_SIZE);
> +	vid_bit = (uint32_t)(1 << (vlan_id % I40E_UINT32_BIT_SIZE));
>  
>  	if (on)
>  		vsi->vfta[vid_idx] |= vid_bit;