Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1311386 > unrolled thread

[PATCH net-next] net: hns: bug fix about hisilicon TSO BD mode

Started byDaode Huang <huangdaode@hisilicon.com>
First post2016-01-18 10:10 +0100
Last post2016-01-19 03:00 +0100
Articles 5 — 4 participants

Back to article view | Back to linux.kernel

Contents

  [PATCH net-next] net: hns: bug fix about hisilicon TSO BD mode Daode Huang <huangdaode@hisilicon.com> - 2016-01-18 10:10 +0100
    Re: [PATCH net-next] net: hns: bug fix about hisilicon TSO BD mode Andy Shevchenko <andriy.shevchenko@linux.intel.com> - 2016-01-18 10:40 +0100
      Re: [PATCH net-next] net: hns: bug fix about hisilicon TSO BD mode huangdaode <huangdaode@hisilicon.com> - 2016-01-18 11:00 +0100
    Re: [PATCH net-next] net: hns: bug fix about hisilicon TSO BD mode David Miller <davem@davemloft.net> - 2016-01-18 18:00 +0100
      Re: [PATCH net-next] net: hns: bug fix about hisilicon TSO BD mode huangdaode <huangdaode@hisilicon.com> - 2016-01-19 03:00 +0100

#1311386 — [PATCH net-next] net: hns: bug fix about hisilicon TSO BD mode

FromDaode Huang <huangdaode@hisilicon.com>
Date2016-01-18 10:10 +0100
Subject[PATCH net-next] net: hns: bug fix about hisilicon TSO BD mode
Message-ID<qSdZ0-6VF-9@gated-at.bofh.it>
The current upstreaming code fails to set the tso_mode register
when initilizes, when processes large size packets, the default 4 bd is
not enough, so this patch initilizes it and set the default value to 8 bds

Signed-off-by: Daode Huang <huangdaode@hisilicon.com>
---
 drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c | 13 +++++++++++--
 drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h |  3 +++
 drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h |  5 +++++
 3 files changed, 19 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c
index d2263c7..1218880 100644
--- a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c
+++ b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c
@@ -369,8 +369,17 @@ int hns_rcb_common_init_hw(struct rcb_common_cb *rcb_common)
 	dsaf_write_dev(rcb_common, RCB_COM_CFG_ENDIAN_REG,
 		       HNS_RCB_COMMON_ENDIAN);
 
-	dsaf_write_dev(rcb_common, RCB_COM_CFG_FNA_REG, 0x0);
-	dsaf_write_dev(rcb_common, RCB_COM_CFG_FA_REG, 0x1);
+	if (AE_IS_VER1(rcb_common->dsaf_dev->dsaf_ver)) {
+		dsaf_write_dev(rcb_common, RCB_COM_CFG_FNA_REG, 0x0);
+		dsaf_write_dev(rcb_common, RCB_COM_CFG_FA_REG, 0x1);
+	} else {
+		dsaf_set_dev_bit(rcb_common, RCBV2_COM_CFG_USER_REG,
+				 RCB_COM_CFG_FNA_B, false);
+		dsaf_set_dev_bit(rcb_common, RCBV2_COM_CFG_USER_REG,
+				 RCB_COM_CFG_FA_B, true);
+		dsaf_set_dev_bit(rcb_common, RCBV2_COM_CFG_TSO_MODE_REG,
+				 RCB_COM_TSO_MODE_B, HNS_TSO_MODE_8BD_32K);
+	}
 
 	return 0;
 }
diff --git a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h
index 29041b1..81fe9f8 100644
--- a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h
+++ b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h
@@ -54,6 +54,9 @@ struct rcb_common_cb;
 #define HNS_DUMP_REG_NUM			500
 #define HNS_STATIC_REG_NUM			12
 
+#define HNS_TSO_MODE_8BD_32K			1
+#define HNS_TSO_MDOE_4BD_16K			0
+
 enum rcb_int_flag {
 	RCB_INT_FLAG_TX = 0x1,
 	RCB_INT_FLAG_RX = (0x1 << 1),
diff --git a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h
index 5d1b746..f0c4f9b 100644
--- a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h
+++ b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h
@@ -363,6 +363,8 @@
 #define RCB_COM_CFG_FA_REG			0x3C
 #define RCB_COM_CFG_PKT_TC_BP_REG		0x40
 #define RCB_COM_CFG_PPE_TNL_CLKEN_REG		0x44
+#define RCBV2_COM_CFG_USER_REG			0x30
+#define RCBV2_COM_CFG_TSO_MODE_REG		0x50
 
 #define RCB_COM_INTMSK_TX_PKT_REG		0x3A0
 #define RCB_COM_RINT_TX_PKT_REG			0x3A8
@@ -860,6 +862,9 @@
 
 #define PPE_COMMON_CNT_CLR_CE_B	0
 #define PPE_COMMON_CNT_CLR_SNAP_EN_B	1
+#define RCB_COM_TSO_MODE_B	0
+#define RCB_COM_CFG_FNA_B	1
+#define RCB_COM_CFG_FA_B	0
 
 #define GMAC_DUPLEX_TYPE_B 0
 
-- 
1.9.1

[toc] | [next] | [standalone]

#1311420

FromAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Date2016-01-18 10:40 +0100
Message-ID<qSes2-78G-25@gated-at.bofh.it>
In reply to#1311386
On Mon, 2016-01-18 at 17:24 +0800, Daode Huang wrote:
> The current upstreaming code fails to set the tso_mode register
> when initilizes, when processes large size packets, the default 4 bd
> is
> not enough, so this patch initilizes it and set the default value to
> 8 bds

Please, next time try to thin out the Cc list. I have no intention to
be in it.

> 
> Signed-off-by: Daode Huang <huangdaode@hisilicon.com>
> ---
>  drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c | 13 +++++++++++--
>  drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h |  3 +++
>  drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h |  5 +++++
>  3 files changed, 19 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c
> b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c
> index d2263c7..1218880 100644
> --- a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c
> +++ b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c
> @@ -369,8 +369,17 @@ int hns_rcb_common_init_hw(struct rcb_common_cb
> *rcb_common)
>  	dsaf_write_dev(rcb_common, RCB_COM_CFG_ENDIAN_REG,
>  		       HNS_RCB_COMMON_ENDIAN);
>  
> -	dsaf_write_dev(rcb_common, RCB_COM_CFG_FNA_REG, 0x0);
> -	dsaf_write_dev(rcb_common, RCB_COM_CFG_FA_REG, 0x1);
> +	if (AE_IS_VER1(rcb_common->dsaf_dev->dsaf_ver)) {
> +		dsaf_write_dev(rcb_common, RCB_COM_CFG_FNA_REG,
> 0x0);
> +		dsaf_write_dev(rcb_common, RCB_COM_CFG_FA_REG, 0x1);
> +	} else {
> +		dsaf_set_dev_bit(rcb_common, RCBV2_COM_CFG_USER_REG,
> +				 RCB_COM_CFG_FNA_B, false);
> +		dsaf_set_dev_bit(rcb_common, RCBV2_COM_CFG_USER_REG,
> +				 RCB_COM_CFG_FA_B, true);
> +		dsaf_set_dev_bit(rcb_common,
> RCBV2_COM_CFG_TSO_MODE_REG,
> +				 RCB_COM_TSO_MODE_B,
> HNS_TSO_MODE_8BD_32K);
> +	}
>  
>  	return 0;
>  }
> diff --git a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h
> b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h
> index 29041b1..81fe9f8 100644
> --- a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h
> +++ b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h
> @@ -54,6 +54,9 @@ struct rcb_common_cb;
>  #define HNS_DUMP_REG_NUM			500
>  #define HNS_STATIC_REG_NUM			12
>  
> +#define HNS_TSO_MODE_8BD_32K			1
> +#define HNS_TSO_MDOE_4BD_16K			0

Typo: MDOE

> +
>  enum rcb_int_flag {
>  	RCB_INT_FLAG_TX = 0x1,
>  	RCB_INT_FLAG_RX = (0x1 << 1),
> diff --git a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h
> b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h
> index 5d1b746..f0c4f9b 100644
> --- a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h
> +++ b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h
> @@ -363,6 +363,8 @@
>  #define RCB_COM_CFG_FA_REG			0x3C
>  #define RCB_COM_CFG_PKT_TC_BP_REG		0x40
>  #define RCB_COM_CFG_PPE_TNL_CLKEN_REG		0x44
> +#define RCBV2_COM_CFG_USER_REG			0x30
> +#define RCBV2_COM_CFG_TSO_MODE_REG		0x50
>  
>  #define RCB_COM_INTMSK_TX_PKT_REG		0x3A0
>  #define RCB_COM_RINT_TX_PKT_REG			0x3A8
> @@ -860,6 +862,9 @@
>  
>  #define PPE_COMMON_CNT_CLR_CE_B	0
>  #define PPE_COMMON_CNT_CLR_SNAP_EN_B	1
> +#define RCB_COM_TSO_MODE_B	0
> +#define RCB_COM_CFG_FNA_B	1
> +#define RCB_COM_CFG_FA_B	0
>  
>  #define GMAC_DUPLEX_TYPE_B 0
>  

-- 
Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Intel Finland Oy

[toc] | [prev] | [next] | [standalone]

#1311431

Fromhuangdaode <huangdaode@hisilicon.com>
Date2016-01-18 11:00 +0100
Message-ID<qSeLo-7gM-3@gated-at.bofh.it>
In reply to#1311420

On 2016/1/18 17:36, Andy Shevchenko wrote:
> On Mon, 2016-01-18 at 17:24 +0800, Daode Huang wrote:
>> The current upstreaming code fails to set the tso_mode register
>> when initilizes, when processes large size packets, the default 4 bd
>> is
>> not enough, so this patch initilizes it and set the default value to
>> 8 bds
> Please, next time try to thin out the Cc list. I have no intention to
> be in it.

OK, sorry for interrupt!
thanks.

>> Signed-off-by: Daode Huang <huangdaode@hisilicon.com>
>> ---
>>   drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c | 13 +++++++++++--
>>   drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h |  3 +++
>>   drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h |  5 +++++
>>   3 files changed, 19 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c
>> b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c
>> index d2263c7..1218880 100644
>> --- a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c
>> +++ b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.c
>> @@ -369,8 +369,17 @@ int hns_rcb_common_init_hw(struct rcb_common_cb
>> *rcb_common)
>>   	dsaf_write_dev(rcb_common, RCB_COM_CFG_ENDIAN_REG,
>>   		       HNS_RCB_COMMON_ENDIAN);
>>   
>> -	dsaf_write_dev(rcb_common, RCB_COM_CFG_FNA_REG, 0x0);
>> -	dsaf_write_dev(rcb_common, RCB_COM_CFG_FA_REG, 0x1);
>> +	if (AE_IS_VER1(rcb_common->dsaf_dev->dsaf_ver)) {
>> +		dsaf_write_dev(rcb_common, RCB_COM_CFG_FNA_REG,
>> 0x0);
>> +		dsaf_write_dev(rcb_common, RCB_COM_CFG_FA_REG, 0x1);
>> +	} else {
>> +		dsaf_set_dev_bit(rcb_common, RCBV2_COM_CFG_USER_REG,
>> +				 RCB_COM_CFG_FNA_B, false);
>> +		dsaf_set_dev_bit(rcb_common, RCBV2_COM_CFG_USER_REG,
>> +				 RCB_COM_CFG_FA_B, true);
>> +		dsaf_set_dev_bit(rcb_common,
>> RCBV2_COM_CFG_TSO_MODE_REG,
>> +				 RCB_COM_TSO_MODE_B,
>> HNS_TSO_MODE_8BD_32K);
>> +	}
>>   
>>   	return 0;
>>   }
>> diff --git a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h
>> b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h
>> index 29041b1..81fe9f8 100644
>> --- a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h
>> +++ b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_rcb.h
>> @@ -54,6 +54,9 @@ struct rcb_common_cb;
>>   #define HNS_DUMP_REG_NUM			500
>>   #define HNS_STATIC_REG_NUM			12
>>   
>> +#define HNS_TSO_MODE_8BD_32K			1
>> +#define HNS_TSO_MDOE_4BD_16K			0
> Typo: MDOE
>
>> +
>>   enum rcb_int_flag {
>>   	RCB_INT_FLAG_TX = 0x1,
>>   	RCB_INT_FLAG_RX = (0x1 << 1),
>> diff --git a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h
>> b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h
>> index 5d1b746..f0c4f9b 100644
>> --- a/drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h
>> +++ b/drivers/net/ethernet/hisilicon/hns/hns_dsaf_reg.h
>> @@ -363,6 +363,8 @@
>>   #define RCB_COM_CFG_FA_REG			0x3C
>>   #define RCB_COM_CFG_PKT_TC_BP_REG		0x40
>>   #define RCB_COM_CFG_PPE_TNL_CLKEN_REG		0x44
>> +#define RCBV2_COM_CFG_USER_REG			0x30
>> +#define RCBV2_COM_CFG_TSO_MODE_REG		0x50
>>   
>>   #define RCB_COM_INTMSK_TX_PKT_REG		0x3A0
>>   #define RCB_COM_RINT_TX_PKT_REG			0x3A8
>> @@ -860,6 +862,9 @@
>>   
>>   #define PPE_COMMON_CNT_CLR_CE_B	0
>>   #define PPE_COMMON_CNT_CLR_SNAP_EN_B	1
>> +#define RCB_COM_TSO_MODE_B	0
>> +#define RCB_COM_CFG_FNA_B	1
>> +#define RCB_COM_CFG_FA_B	0
>>   
>>   #define GMAC_DUPLEX_TYPE_B 0
>>

[toc] | [prev] | [next] | [standalone]

#1311678

FromDavid Miller <davem@davemloft.net>
Date2016-01-18 18:00 +0100
Message-ID<qSljT-3pb-61@gated-at.bofh.it>
In reply to#1311386
From: Daode Huang <huangdaode@hisilicon.com>
Date: Mon, 18 Jan 2016 17:24:16 +0800

> The current upstreaming code fails to set the tso_mode register
> when initilizes, when processes large size packets, the default 4 bd is
> not enough, so this patch initilizes it and set the default value to 8 bds
> 
> Signed-off-by: Daode Huang <huangdaode@hisilicon.com>

Your insistance of continually using "net-next" in your subject lines
shows you have no idea what GIT trees are active and in what trees
development and bug fixing is actually occurring in.

The 'net' tree is the only tree active now and your driver has been in
there for over a week.

And this is the place where all appropriate bug fixes go.

Therefore if your code is present in the 'net' tree and you are fixing
things in that codebase, then that is the tree you target.

I've been silently applying your patches there instead of net-next,
but I'm not going to do it any more and I will just reject patches not
submitted properly from this point forward.

Thanks.

[toc] | [prev] | [next] | [standalone]

#1311889

Fromhuangdaode <huangdaode@hisilicon.com>
Date2016-01-19 03:00 +0100
Message-ID<qStKq-Kn-1@gated-at.bofh.it>
In reply to#1311678

On 2016/1/19 0:55, David Miller wrote:
> From: Daode Huang <huangdaode@hisilicon.com>
> Date: Mon, 18 Jan 2016 17:24:16 +0800
>
>> The current upstreaming code fails to set the tso_mode register
>> when initilizes, when processes large size packets, the default 4 bd is
>> not enough, so this patch initilizes it and set the default value to 8 bds
>>
>> Signed-off-by: Daode Huang <huangdaode@hisilicon.com>
> Your insistance of continually using "net-next" in your subject lines
> shows you have no idea what GIT trees are active and in what trees
> development and bug fixing is actually occurring in.
>
> The 'net' tree is the only tree active now and your driver has been in
> there for over a week.
>
> And this is the place where all appropriate bug fixes go.
>
> Therefore if your code is present in the 'net' tree and you are fixing
> things in that codebase, then that is the tree you target.
>
> I've been silently applying your patches there instead of net-next,
> but I'm not going to do it any more and I will just reject patches not
> submitted properly from this point forward.
>
> Thanks.
>
> .

Hi David,
Thanks for your comments, will pay more attention to it.

Best Regards
Daode.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web