From: Claudiu Manoil <claudiu.manoil@nxp.com>
To: Zumeng Chen <zumeng.chen@gmail.com>,
        "netdev@vger.kernel.org" <netdev@vger.kernel.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
CC: "davem@davemloft.net" <davem@davemloft.net>
Subject: RE: [PATCH 1/1] gianfar: fix a flooded alignment reports because of
 padding issue.
Thread-Topic: [PATCH 1/1] gianfar: fix a flooded alignment reports because of
 padding issue.
Thread-Index: AQHTbK3JlkX1F5k68EKvMuES9+QQnKMzTx6A
Date: Mon, 4 Dec 2017 16:06:09 +0000
Message-ID: <AM5PR0401MB256141EAA45FA0BBB301D8B7963C0@AM5PR0401MB2561.eurprd04.prod.outlook.com>
References: <1512357722-12124-1-git-send-email-zumeng.chen@gmail.com>
In-Reply-To: <1512357722-12124-1-git-send-email-zumeng.chen@gmail.com>
Accept-Language: en-US
Content-Language: en-US
spamdiagnosticoutput: 1:99
spamdiagnosticmetadata: NSPM
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
X-MS-Exchange-CrossTenant-Network-Message-Id: 2fee7dcc-5f42-42e1-ab15-08d53b30ef37
X-MS-Exchange-CrossTenant-originalarrivaltime: 04 Dec 2017 16:06:09.9727
 (UTC)
X-MS-Exchange-CrossTenant-fromentityheader: Hosted
X-MS-Exchange-CrossTenant-id: 686ea1d3-bc2b-4c6f-a92c-d99c5c301635
X-MS-Exchange-Transport-CrossTenantHeadersStamped: AM5PR0401MB2564
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by nfs id vB4G6ieN022126
Content-Length: 2471
Lines: 72

>-----Original Message-----
>From: Zumeng Chen [mailto:zumeng.chen@gmail.com]
>Sent: Monday, December 04, 2017 5:22 AM
>To: netdev@vger.kernel.org; linux-kernel@vger.kernel.org
>Cc: Claudiu Manoil <claudiu.manoil@nxp.com>; davem@davemloft.net
>Subject: [PATCH 1/1] gianfar: fix a flooded alignment reports because of padding
>issue.
>
>According to LS1021A RM, the value of PAL can be set so that the start of the
>IP header in the receive data buffer is aligned to a 32-bit boundary. Normally,
>setting PAL = 2 provides minimal padding to ensure such alignment of the IP
>header.
>
>However every incoming packet's 8-byte time stamp will be inserted into the
>packet data buffer as padding alignment bytes when hardware time stamping is
>enabled.
>
>So we set the padding 8+2 here to avoid the flooded alignment faults:
>
>root@128:~# cat /proc/cpu/alignment
>User:           0
>System:         17539 (inet_gro_receive+0x114/0x2c0)
>Skipped:        0
>Half:           0
>Word:           0
>DWord:          0
>Multi:          17539
>User faults:    2 (fixup)
>
[...]
>
> drivers/net/ethernet/freescale/gianfar.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
>diff --git a/drivers/net/ethernet/freescale/gianfar.c
>b/drivers/net/ethernet/freescale/gianfar.c
>index e616b71..e47945f 100644
>--- a/drivers/net/ethernet/freescale/gianfar.c
>+++ b/drivers/net/ethernet/freescale/gianfar.c
>@@ -1413,9 +1413,11 @@ static int gfar_probe(struct platform_device *ofdev)
>
> 	gfar_init_addr_hash_table(priv);
>
>-	/* Insert receive time stamps into padding alignment bytes */
>+	/* Insert receive time stamps into padding alignment bytes, and
>+	 * plus 2 bytes padding to ensure the cpu alignment.
>+	 */
> 	if (priv->device_flags & FSL_GIANFAR_DEV_HAS_TIMER)
>-		priv->padding = 8;
>+		priv->padding = 8 + DEFAULT_PADDING;
>
> 	if (dev->features & NETIF_F_IP_CSUM ||
> 	    priv->device_flags & FSL_GIANFAR_DEV_HAS_TIMER)
>--
>2.5.0

Why handle only the rx timestamp path (HAS_TIMER) when the issue, as presented 
by you, should be applicable to the default path as well?

The code change according to the patch description should be more likely:

+       priv->padding = DEFAULT_PADDING;
        /* Insert receive time stamps into padding alignment bytes */
        if (priv->device_flags & FSL_GIANFAR_DEV_HAS_TIMER)
-               priv->padding = 8;
+               priv->padding += 8;


Do you have any performance numbers for this change?

Thanks,
Claudiu