netdev - Re: [PATCH net-next 1/4] net: hns3: refine the tcam key convert handle

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, 15 May 2023 17:51:03 +0200
From: Simon Horman <simon.horman@...igine.com>
To: Hao Lan <lanhao@...wei.com>
Cc: netdev@...r.kernel.org, yisen.zhuang@...wei.com, salil.mehta@...wei.com,
	davem@...emloft.net, edumazet@...gle.com, kuba@...nel.org,
	pabeni@...hat.com, richardcochran@...il.com,
	wangpeiyang1@...wei.com, shenjian15@...wei.com,
	chenhao418@...wei.com, wangjie125@...wei.com, yuanjilin@...rlc.com,
	cai.huoqing@...ux.dev, xiujianfeng@...wei.com
Subject: Re: [PATCH net-next 1/4] net: hns3: refine the tcam key convert
 handle

On Mon, May 15, 2023 at 09:46:40PM +0800, Hao Lan wrote:
> From: Jian Shen <shenjian15@...wei.com>
> 
> The expression '(k ^ ~v)' is exaclty '(k & v)', and

This part doesn't seem correct to me.

Suppose both k and v are 0.
For simplicity, consider only one bit.

Then: (k ^ ~v) == (0 ^ ~0) == (0 ^ 1) = 1
But   (k & v)  == (0 & 0)  == 0

> '(k & v) & k' is exaclty 'k & v'. So simplify the
> expression for tcam key convert.

This I follow.

> (k ^ ~v) & k ==  (k & v) & k ==  k & k & v == k & v

Looking at the truth table (in non table form), this seems correct.

k == 0, v == 0:

  (k ^ ~v) & k == (0 ^ ~0) & 0 == (0 ^ 1) & 0 == 1 & 0 == 0
  k & v == 0 & 0 == 0
  Good!

k == 0, v == 1:

  (k ^ ~v) & k == (0 ^ ~1) & 0 == (0 ^ 0) & 0 == 1 & 0 == 0
  k & v == 0 & 1 == 0
  Good!

k == 1, v == 0:

  (k ^ ~v) & k == (1 ^ ~0) & 1 == (1 ^ 1) & 1 == 0 & 1 == 0
  k & v == 1 & 0 == 0
  Good!

k == 1, v == 1:

  (k ^ ~v) & k == (1 ^ ~1) & 1 == (1 ^ 0) & 1 == 1 & 1 == 1
  k & v == 1 & 1 == 1
  Good!

> 
> It also add necessary brackets for them.
> 
> Signed-off-by: Jian Shen <shenjian15@...wei.com>
> Signed-off-by: Hao Lan <lanhao@...wei.com>
> ---
>  .../net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h   | 11 +++--------
>  1 file changed, 3 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
> index 81aa6b0facf5..6a43d1515585 100644
> --- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
> +++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.h
> @@ -835,15 +835,10 @@ struct hclge_vf_vlan_cfg {
>   * Then for input key(k) and mask(v), we can calculate the value by
>   * the formulae:
>   *	x = (~k) & v
> - *	y = (k ^ ~v) & k
> + *	y = k & v
>   */
> -#define calc_x(x, k, v) (x = ~(k) & (v))
> -#define calc_y(y, k, v) \
> -	do { \
> -		const typeof(k) _k_ = (k); \
> -		const typeof(v) _v_ = (v); \
> -		(y) = (_k_ ^ ~_v_) & (_k_); \
> -	} while (0)
> +#define calc_x(x, k, v) ((x) = ~(k) & (v))
> +#define calc_y(y, k, v) ((y) = (k) & (v))

This also looks good to me.

>  
>  #define HCLGE_MAC_STATS_FIELD_OFF(f) (offsetof(struct hclge_mac_stats, f))
>  #define HCLGE_STATS_READ(p, offset) (*(u64 *)((u8 *)(p) + (offset)))