ixgb: driver update (upstream)
Subsystems (2)
cover ixgb: driver update (upstream)
The following patches were diff's against Garzik's latest "upstream" branch and can be pulled from the following location:
git://lost.foo-projects.org/linux-2.6.git ixgb-upstream
The following series implements...
01. Fix compilation errors by initializing variables
02. Fix the use of dprintk rather than printk
03. Fix hard coded numbers
04. Fix duplicate code
05. Fix timeout code
06. Fix TSO
07. Fixed flow control parameters
08. Add support for copper 10GbE
09. Add performance enhancements to the buffer_info struct
10. Add whitespace changes
11. Add prefetch
--
Cheers,
Jeff
PATCH 1/11 ixgb: Fix compilation errors by initializing variables
- initialized varaibles
- rev'd driver version
---
drivers/net/ixgb/ixgb_main.c | 4 ++-- 1 files changed, 2 insertions(+), 2 deletions(-)
drivers/net/ixgb/ixgb_main.c +2 -2
@@ -52,7 +52,7 @@ static char ixgb_driver_string[] = "Inte#else #define DRIVERNAPI "-NAPI"#endif -#define DRV_VERSION "1.0.100-k2"DRIVERNAPI +#define DRV_VERSION "1.0.104-k2"DRIVERNAPIchar ixgb_driver_version[] =DRV_VERSION ;static char ixgb_copyright[] ="Copyright (c) 1999-2005 Intel Corporation." ;
@@ -346,7 +346,7 @@ ixgb_probe(struct pci_dev *pdev, const struct pci_device_id *ent) {struct net_device *netdev =NULL ; -struct ixgb_adapter *adapter; +struct ixgb_adapter *adapter =NULL ;static int cards_found =0 ;unsigned long mmio_start;int mmio_len;
Re: [PATCH 01/11] ixgb: Fix compilation errors by initializing variables
Jeff Kirsher <[email protected]> :
Quoted 4 lines
drivers/net/ixgb/ixgb_main.c +0
-0
[...]
drivers/net/ixgb/ixgb_main.c ixgb_probe(struct pci_dev *pdev, · line 349
Quoted 6 lines
drivers/net/ixgb/ixgb_main.c +1
-1@@ -346,7 +346,7 @@ ixgb_probe(struct pci_dev *pdev, const struct pci_device_id *ent) {struct net_device *netdev =NULL ; -struct ixgb_adapter *adapter; +struct ixgb_adapter *adapter =NULL ;
Afaiks the warning/error from the compiler is bogus.
netdev initialization is useless as well.
Are we supposed to bloat the kernel code on a large scale just to please a compiler which emits bogus warning or is the option left to the maintainer ?
Signature · 1 line
Ueimor
PATCH 2/11 ixgb: Fix the use of dprintk rather than printk
In reply to Jeff Kirsher- use DPRINTK and msglvl instead of printk
- allow ethtool control of msglvl
---
drivers/net/ixgb/ixgb.h | 8 +++++- drivers/net/ixgb/ixgb_ethtool.c | 16 ++++++++++++ drivers/net/ixgb/ixgb_main.c | 52 +++++++++++++++++++++++++-------------- 3 files changed, 56 insertions(+), 20 deletions(-)
drivers/net/ixgb/ixgb.h +7 -1
@@ -84,7 +84,12 @@ struct ixgb_adapter; #define IXGB_DBG(args...)#endif -#define IXGB_ERR(args...) printk(KERN_ERR "ixgb: " args) +#define PFX "ixgb: " +#define DPRINTK(nlevel, klevel, fmt, args...) \ + (void)((NETIF_MSG_##nlevel & adapter->msg_enable) && \ + printk(KERN_##klevel PFX "%s: %s: " fmt, adapter->netdev->name, \ + __FUNCTION__ , ## args)) +/* TX/RX descriptor defines */ #define DEFAULT_TXD 256
@@ -192,6 +197,7 @@ struct ixgb_adapter {/* structs defined in ixgb_hw.h */ struct ixgb_hw hw; +u16 msg_enable;struct ixgb_hw_stats stats;#ifdef CONFIG_PCI_MSI boolean_t have_msi;
drivers/net/ixgb/ixgb_ethtool.c +16 -0
@@ -251,6 +251,20 @@ ixgb_set_tso(struct net_device *netdev, }#endif /* NETIF_F_TSO */ +static uint32_t +ixgb_get_msglevel (struct net_device *netdev) +{ +struct ixgb_adapter *adapter = netdev->priv; +return adapter->msg_enable; +} + +static void +ixgb_set_msglevel (struct net_device *netdev,uint32_t data) +{ +struct ixgb_adapter *adapter = netdev->priv; + adapter->msg_enable = data; +} +#define IXGB_GET_STAT (_A_, _R_) _A_->stats._R_ static int
@@ -714,6 +728,8 @@ static struct ethtool_ops ixgb_ethtool_o .set_tx_csum = ixgb_set_tx_csum, .get_sg = ethtool_op_get_sg, .set_sg = ethtool_op_set_sg, + .get_msglevel = ixgb_get_msglevel, + .set_msglevel = ixgb_set_msglevel, #ifdef NETIF_F_TSO .get_tso = ethtool_op_get_tso, .set_tso = ixgb_set_tso,
drivers/net/ixgb/ixgb_main.c +33 -19
161 lines not on this page · raw message
Re: [PATCH 02/11] ixgb: Fix the use of dprintk rather than printk
Jeff Kirsher <[email protected]> :
[...]
Quoted 4 lines
drivers/net/ixgb/ixgb.h +0
-0
[...]
Quoted 20 lines
drivers/net/ixgb/ixgb.h +1
-0@@ -192,6 +197,7 @@ struct ixgb_adapter {/* structs defined in ixgb_hw.h */ struct ixgb_hw hw; +u16 msg_enable;struct ixgb_hw_stats stats;#ifdef CONFIG_PCI_MSI boolean_t have_msi;drivers/net/ixgb/ixgb_ethtool.c +4
-0@@ -251,6 +251,20 @@ ixgb_set_tso(struct net_device *netdev, }#endif /* NETIF_F_TSO */ +static uint32_t +ixgb_get_msglevel (struct net_device *netdev) +{ +struct ixgb_adapter *adapter = netdev->priv;
drivers/net/ixgb/ixgb_ethtool.c ixgb_set_tso(struct net_device *netdev, · line 267
Quoted 10 lines
drivers/net/ixgb/ixgb_ethtool.c +10
-0+return adapter->msg_enable; +} + +static void +ixgb_set_msglevel (struct net_device *netdev,uint32_t data) +{ +struct ixgb_adapter *adapter = netdev->priv; + adapter->msg_enable = data; +} +
Minor nits:
- you may consider removing the u{8/16/32} in drivers/net/ixgb for consistency sake in a different patch (there is a strong majority of uint_something in the driver).
- s/netdev->priv/netdev_priv(netdev)/ ?
[...]
drivers/net/ixgb/ixgb_ethtool.c ixgb_probe(struct pci_dev *pdev, · line 498
Quoted 7 lines
drivers/net/ixgb/ixgb_ethtool.c +1
-2@@ -486,8 +495,7 @@ ixgb_probe(struct pci_dev *pdev,netif_carrier_off (netdev);netif_stop_queue (netdev); -printk (KERN_INFO "%s: Intel(R) PRO/10GbE Network Connection\n" , - netdev->name); +DPRINTK (PROBE ,INFO ,"Intel(R) PRO/10GbE Network Connection\n" );
It could probably be factored out with ixgb_driver_string.
Signature · 1 line
Ueimor
Re: [PATCH 02/11] ixgb: Fix the use of dprintk rather than printk
Quoted 34 lines · On Sat, 22 Apr 2006 11:03:01 +0200 Francois Romieu <[email protected]> wrote:
Jeff Kirsher <[email protected]> :
[...]Quoted 4 lines
drivers/net/ixgb/ixgb.h +0
-0[...]
Quoted 27 lines
drivers/net/ixgb/ixgb.h +1
-0@@ -192,6 +197,7 @@ struct ixgb_adapter {/* structs defined in ixgb_hw.h */ struct ixgb_hw hw; +u16 msg_enable;struct ixgb_hw_stats stats;#ifdef CONFIG_PCI_MSI boolean_t have_msi;drivers/net/ixgb/ixgb_ethtool.c +11
-0@@ -251,6 +251,20 @@ ixgb_set_tso(struct net_device *netdev, }#endif /* NETIF_F_TSO */ +static uint32_t +ixgb_get_msglevel (struct net_device *netdev) +{ +struct ixgb_adapter *adapter = netdev->priv; +return adapter->msg_enable; +} + +static void +ixgb_set_msglevel (struct net_device *netdev,uint32_t data) +{ +struct ixgb_adapter *adapter = netdev->priv;
Quoted 8 lines
drivers/net/ixgb/ixgb_ethtool.c ixgb_set_tso(struct net_device *netdev, · line 267
Quoted 3 lines
drivers/net/ixgb/ixgb_ethtool.c +3
-0+ adapter->msg_enable = data; +} +Minor nits:
- you may consider removing the u{8/16/32} indrivers/net/ixgbfor consistency sake in a different patch (there is a strong majority ofuint_somethingin the driver).
All the uint32_t should be removed. Kernel style is u32.
Re: [PATCH 02/11] ixgb: Fix the use of dprintk rather than printk
In reply to Francois RomieuQuoted 10 lines · Francois Romieu
Jeff Kirsher <[email protected]> :
drivers/net/ixgb/ixgb_main.c ixgb_probe(struct pci_dev *pdev, · line 498
Quoted 7 lines
drivers/net/ixgb/ixgb_main.c +1
-2@@ -486,8 +495,7 @@ ixgb_probe(struct pci_dev *pdev,netif_carrier_off (netdev);netif_stop_queue (netdev); -printk (KERN_INFO "%s: Intel(R) PRO/10GbE Network Connection\n" , - netdev->name); +DPRINTK (PROBE ,INFO ,"Intel(R) PRO/10GbE Network Connection\n" );It could probably be factored out with
ixgb_driver_string.
the driver name is different from the interface/device that it drives - Hence the different name. The 'Connection' name as above is the branding name as well, and that's why we print it here as such. e1000 does the same thing for the same reason.
Auke
PATCH 3/11 ixgb: Fix hard coded numbers
In reply to Jeff Kirsher- removed hard coded numbers and used constants instead
---
drivers/net/ixgb/ixgb_ethtool.c | 4 +++- drivers/net/ixgb/ixgb_main.c | 8 ++++---- 2 files changed, 7 insertions(+), 5 deletions(-)
drivers/net/ixgb/ixgb_ethtool.c +3 -1
@@ -44,6 +44,8 @@ extern void ixgb_free_rx_resources(strucextern void ixgb_free_tx_resources (struct ixgb_adapter *adapter);extern void ixgb_update_stats (struct ixgb_adapter *adapter); +#define IXGB_ALL_RAR_ENTRIES 16 +struct ixgb_stats {char stat_string[ETH_GSTRING_LEN ];int sizeof_stat;
@@ -317,7 +319,7 @@ ixgb_get_regs(struct net_device *netdev, *reg++ =IXGB_READ_REG (hw,RXCSUM );/* 20 */ /* there are 16 RAR entries in hardware, we only use 3 */ -for (i =0 ; i <16 ; i++) { +for (i =0 ; i <IXGB_ALL_RAR_ENTRIES ; i++) { *reg++ =IXGB_READ_REG_ARRAY (hw,RAL , (i <<1 ));/*21,...,51 */ *reg++ =IXGB_READ_REG_ARRAY (hw,RAH , (i <<1 ));/*22,...,52 */ }
drivers/net/ixgb/ixgb_main.c +4 -4
@@ -1227,11 +1227,11 @@ ixgb_tso(struct ixgb_adapter *adapter, sif (++i == adapter->tx_ring.count) i =0 ; adapter->tx_ring.next_to_use = i; -return 1 ; +return TRUE; }#endif -return 0 ; +return FALSE; } static inline boolean_t
@@ -1413,7 +1413,7 @@ ixgb_xmit_frame(struct sk_buff *skb, strif (unlikely (IXGB_DESC_UNUSED (&adapter->tx_ring) <DESC_NEEDED )) {netif_stop_queue (netdev);spin_unlock_irqrestore (&adapter->tx_lock, flags); -return 1 ; +return NETDEV_TX_BUSY ; }spin_unlock_irqrestore (&adapter->tx_lock, flags);
@@ -1440,7 +1440,7 @@ ixgb_xmit_frame(struct sk_buff *skb, str netdev->trans_start = jiffies; -return 0 ; +return NETDEV_TX_OK ; }/**
Re: [PATCH 03/11] ixgb: Fix hard coded numbers
Jeff Kirsher <[email protected]> :
[...]
Quoted 4 lines
drivers/net/ixgb/ixgb_ethtool.c +0
-0
[...]
Quoted 6 lines
drivers/net/ixgb/ixgb_main.c +0
-0@@ -1227,11 +1227,11 @@ ixgb_tso(struct ixgb_adapter *adapter, sif (++i == adapter->tx_ring.count) i =0 ;
drivers/net/ixgb/ixgb_main.c ixgb_tso(struct ixgb_adapter *adapter, s · line 1235
Quoted 10 lines
drivers/net/ixgb/ixgb_main.c +2
-2adapter->tx_ring.next_to_use = i; - return 1; + return TRUE; }#endif -return 0 ; +return FALSE; }
Grmbl... TRUE/FALSE
More importantly, it seems bogus. See below:
drivers/net/ixgb/ixgb_main.c::ixgb_xmit_frame
[...]
tso = ixgb_tso(adapter, skb);
if (tso < 0) {
^^^
dev_kfree_skb_any(skb);
return N[...]
drivers/net/ixgb/ixgb_main.c ixgb_xmit_frame(struct sk_buff *skb, str · line 1417
Quoted 7 lines
drivers/net/ixgb/ixgb_main.c +1
-1@@ -1413,7 +1413,7 @@ ixgb_xmit_frame(struct sk_buff *skb, strif (unlikely (IXGB_DESC_UNUSED (&adapter->tx_ring) <DESC_NEEDED )) {netif_stop_queue (netdev);spin_unlock_irqrestore (&adapter->tx_lock, flags); -return 1 ; +return NETDEV_TX_BUSY ; }
It is considered a bug. You can check for this situation on the current skb but the driver should check if there is enough room for the *next* packet before it leaves ixgb_xmit_frame().
Signature · 1 line
Ueimor
PATCH 4/11 ixgb: Fix duplicate code
In reply to Jeff Kirsher- created function ixgb_set_speed_duplex to remove duplicate code
---
drivers/net/ixgb/ixgb_ethtool.c | 35 ++++++++++++++--------------------- 1 files changed, 14 insertions(+), 21 deletions(-)
drivers/net/ixgb/ixgb_ethtool.c +14 -21
@@ -119,6 +119,16 @@ ixgb_get_settings(struct net_device *netreturn 0 ; } +static void ixgb_set_speed_duplex (struct net_device *netdev) +{ +struct ixgb_adapter *adapter =netdev_priv (netdev); +/* be optimistic about our link, since we were up before */ + adapter->link_speed =10000 ; + adapter->link_duplex =FULL_DUPLEX ; +netif_carrier_on (netdev); +netif_wake_queue (netdev); +} +static int ixgb_set_settings (struct net_device *netdev,struct ethtool_cmd *ecmd) {
@@ -132,12 +142,7 @@ ixgb_set_settings(struct net_device *netixgb_down (adapter, TRUE);ixgb_reset (adapter);ixgb_up (adapter); -/* be optimistic about our link, since we were up before */ - adapter->link_speed =10000 ; - adapter->link_duplex =FULL_DUPLEX ; -netif_carrier_on (netdev); -netif_wake_queue (netdev); - +ixgb_set_speed_duplex (netdev); }else ixgb_reset (adapter);
@@ -185,11 +190,7 @@ ixgb_set_pauseparam(struct net_device *nif (netif_running (adapter->netdev)) {ixgb_down (adapter, TRUE);ixgb_up (adapter); -/* be optimistic about our link, since we were up before */ - adapter->link_speed =10000 ; - adapter->link_duplex =FULL_DUPLEX ; -netif_carrier_on (netdev); -netif_wake_queue (netdev); +ixgb_set_speed_duplex (netdev); }else ixgb_reset (adapter);
@@ -214,11 +215,7 @@ ixgb_set_rx_csum(struct net_device *netdif (netif_running (netdev)) {ixgb_down (adapter,TRUE);ixgb_up (adapter); -/* be optimistic about our link, since we were up before */ - adapter->link_speed =10000 ; - adapter->link_duplex =FULL_DUPLEX ; -netif_carrier_on (netdev); -netif_wake_queue (netdev); +ixgb_set_speed_duplex (netdev); }else ixgb_reset (adapter);return 0 ;
@@ -609,11 +606,7 @@ ixgb_set_ringparam(struct net_device *ne adapter->tx_ring = tx_new;if ((err =ixgb_up (adapter)))return err; -/* be optimistic about our link, since we were up before */ - adapter->link_speed =10000 ; - adapter->link_duplex =FULL_DUPLEX ; -netif_carrier_on (netdev); -netif_wake_queue (netdev); +ixgb_set_speed_duplex (netdev); }return 0 ;
PATCH 5/11 ixgb: Fix timeout code
In reply to Jeff Kirsher- aligned timeout code to match e1000
---
drivers/net/ixgb/ixgb.h | 1 + drivers/net/ixgb/ixgb_ethtool.c | 1 + drivers/net/ixgb/ixgb_main.c | 29 +++++++++++++++++++++++++---- 3 files changed, 27 insertions(+), 4 deletions(-)
drivers/net/ixgb/ixgb.h +1 -0
@@ -180,6 +180,7 @@ struct ixgb_adapter {uint64_t hw_csum_tx_good;uint64_t hw_csum_tx_error;uint32_t tx_int_delay; +uint32_t tx_timeout_count;boolean_t tx_int_delay_enable;boolean_t detect_tx_hung;
drivers/net/ixgb/ixgb_ethtool.c +1 -0
@@ -78,6 +78,7 @@ static struct ixgb_stats ixgb_gstrings_s {"tx_heartbeat_errors", IXGB_STAT(net_stats.tx_heartbeat_errors)}, {"tx_window_errors", IXGB_STAT(net_stats.tx_window_errors)}, {"tx_deferred_ok", IXGB_STAT(stats.dc)}, + {"tx_timeout_count", IXGB_STAT(tx_timeout_count) }, {"rx_long_length_errors", IXGB_STAT(stats.roc)}, {"rx_short_length_errors", IXGB_STAT(stats.ruc)}, #ifdef NETIF_F_TSO
drivers/net/ixgb/ixgb_main.c +25 -4
@@ -423,7 +423,7 @@ ixgb_probe(struct pci_dev *pdev, netdev->change_mtu = &ixgb_change_mtu;ixgb_set_ethtool_ops (netdev); netdev->tx_timeout = &ixgb_tx_timeout; - netdev->watchdog_timeo =HZ ; + netdev->watchdog_timeo =5 *HZ ;#ifdef CONFIG_IXGB_NAPI netdev->poll = &ixgb_clean; netdev->weight =64 ;
@@ -1462,6 +1462,7 @@ ixgb_tx_timeout_task(struct net_device * {struct ixgb_adapter *adapter =netdev_priv (netdev); + adapter->tx_timeout_count++;ixgb_down (adapter, TRUE);ixgb_up (adapter); }
@@ -1791,11 +1792,31 @@ ixgb_clean_tx_irq(struct ixgb_adapter *a/* detect a transmit hang in hardware, this serializes the * check with the clearing of time_stamp and movement of i */ adapter->detect_tx_hung = FALSE; -if (tx_ring->buffer_info[i].dma && -time_after (jiffies, tx_ring->buffer_info[i].time_stamp +HZ ) +if (tx_ring->buffer_info[eop].dma && +time_after (jiffies, tx_ring->buffer_info[eop].time_stamp +HZ ) && !(IXGB_READ_REG (&adapter->hw,STATUS ) & -IXGB_STATUS_TXOFF )) +IXGB_STATUS_TXOFF )) { +/* detected Tx unit hang */ +DPRINTK (DRV ,ERR ,"Detected Tx Unit Hang\n" +" TDH <%x>\n" +" TDT <%x>\n" +" next_to_use <%x>\n" +" next_to_clean <%x>\n" +"buffer_info[next_to_clean]\n" +" time_stamp <%lx>\n" +" next_to_watch <%x>\n" +" jiffies <%lx>\n" +" next_to_watch.status <%x>\n" , +IXGB_READ_REG (&adapter->hw,TDH ), +IXGB_READ_REG (&adapter->hw,TDT ), + tx_ring->next_to_use, + tx_ring->next_to_clean, + tx_ring->buffer_info[eop].time_stamp, + eop, + jiffies, + eop_desc->status);netif_stop_queue (netdev); + } }return cleaned;
PATCH 6/11 ixgb: Fix TSO
In reply to Jeff Kirsher- this fixes an issue of premature desc writeback by hardware
---
drivers/net/ixgb/ixgb_main.c | 17 +++++++++++++++-- 1 files changed, 15 insertions(+), 2 deletions(-)
drivers/net/ixgb/ixgb_main.c +15 -2
@@ -1280,6 +1280,7 @@ ixgb_tx_map(struct ixgb_adapter *adapterstruct ixgb_buffer *buffer_info;int len = skb->len;unsigned int offset =0 , size, count =0 , i; +unsigned int mss =skb_shinfo (skb)->tso_size;unsigned int nr_frags =skb_shinfo (skb)->nr_frags;unsigned int f;
@@ -1291,6 +1292,11 @@ ixgb_tx_map(struct ixgb_adapter *adapterwhile (len) { buffer_info = &tx_ring->buffer_info[i]; size =min (len,IXGB_MAX_JUMBO_FRAME_SIZE ); +/* Workaround for premature desc write-backs +* in TSO mode. Append 4-byte sentinel desc */ +if (unlikely (mss && !nr_frags && size == len && size >8 )) + size -=4 ; + buffer_info->length = size; buffer_info->dma = pci_map_single(adapter->pdev,
@@ -1315,6 +1321,12 @@ ixgb_tx_map(struct ixgb_adapter *adapterwhile (len) { buffer_info = &tx_ring->buffer_info[i]; size =min (len,IXGB_MAX_JUMBO_FRAME_SIZE ); +/* Workaround for premature desc write-backs +* in TSO mode. Append 4-byte sentinel desc */ +if (unlikely (mss && (f == (nr_frags-1 )) && (size == len) + && (size >8 ))) + size -=4 ; + buffer_info->length = size; buffer_info->dma = pci_map_page(adapter->pdev,
@@ -1392,7 +1404,8 @@ ixgb_tx_queue(struct ixgb_adapter *adapt#define TXD_USE_COUNT (S ) (((S) >> IXGB_MAX_TXD_PWR) + \ (((S) & (IXGB_MAX_DATA_PER_TXD - 1)) ? 1 : 0))#define DESC_NEEDED TXD_USE_COUNT(IXGB_MAX_DATA_PER_TXD) + \ - MAX_SKB_FRAGS * TXD_USE_COUNT(PAGE_SIZE) + 1 + MAX_SKB_FRAGS * TXD_USE_COUNT(PAGE_SIZE) + 1 \ + /* one more for TSO workaround */ + 1static int ixgb_xmit_frame (struct sk_buff *skb,struct net_device *netdev)
@@ -1430,7 +1443,7 @@ ixgb_xmit_frame(struct sk_buff *skb, strreturn NETDEV_TX_OK ; } -if (tso) +if (likely (tso)) tx_flags |=IXGB_TX_FLAGS_TSO ;else if (ixgb_tx_csum (adapter, skb)) tx_flags |=IXGB_TX_FLAGS_CSUM ;
PATCH 7/11 ixgb: Fixed flow control parameters
In reply to Jeff Kirsher- make default flow control only have *sending* of flow control packets enabled
- fix to disable / enable flow control correctly
---
drivers/net/ixgb/ixgb_param.c | 22 ++++++++++------------ 1 files changed, 10 insertions(+), 12 deletions(-)
drivers/net/ixgb/ixgb_param.c +10 -12
@@ -137,7 +137,7 @@ IXGB_PARAM(RxFCLowThresh, "Receive Flow* * Valid Range: 1 - 65535 * -* Default Value: 256 (0x100) +* Default Value: 65535 (0xffff) (we'll send an xon if we recover) */ IXGB_PARAM (FCReqTimeout,"Flow Control Request Timeout" );
@@ -165,8 +165,6 @@ IXGB_PARAM(IntDelayEnable, "Transmit Int#define XSUMRX_DEFAULT OPTION_ENABLED -#define FLOW_CONTROL_FULL ixgb_fc_full -#define FLOW_CONTROL_DEFAULT FLOW_CONTROL_FULL#define DEFAULT_FCRTL 0x28000#define DEFAULT_FCRTH 0x30000#define MIN_FCRTL 0
@@ -174,9 +172,9 @@ IXGB_PARAM(IntDelayEnable, "Transmit Int#define MIN_FCRTH 8#define MAX_FCRTH 0x3FFF0 -#define DEFAULT_FCPAUSE 0x100/* this may be too long */ #define MIN_FCPAUSE 1#define MAX_FCPAUSE 0xffff +#define DEFAULT_FCPAUSE 0xFFFF/* this may be too long */ struct ixgb_option {enum { enable_option, range_option, list_option } type;
@@ -336,7 +334,7 @@ ixgb_check_options(struct ixgb_adapter * .type = list_option, .name = "Flow Control", .err = "reading default settings from EEPROM", - .def = ixgb_fc_full, + .def = ixgb_fc_tx_pause, .arg = { .l = { .nr = LIST_LEN(fc_list), .p = fc_list }} };
@@ -365,8 +363,8 @@ ixgb_check_options(struct ixgb_adapter * }else { adapter->hw.fc.high_water = opt.def; } -if (!(adapter->hw.fc.type & ixgb_fc_rx_pause) ) -printk (KERN_INFO +if ( !(adapter->hw.fc.type & ixgb_fc_tx_pause) ) +printk (KERN_INFO "Ignoring RxFCHighThresh when no RxFC\n" ); } {/* Receive Flow Control Low Threshold */
@@ -385,8 +383,8 @@ ixgb_check_options(struct ixgb_adapter * }else { adapter->hw.fc.low_water = opt.def; } -if (!(adapter->hw.fc.type & ixgb_fc_rx_pause) ) -printk (KERN_INFO +if ( !(adapter->hw.fc.type & ixgb_fc_tx_pause) ) +printk (KERN_INFO "Ignoring RxFCLowThresh when no RxFC\n" ); } {/* Flow Control Pause Time Request*/
@@ -406,12 +404,12 @@ ixgb_check_options(struct ixgb_adapter * }else { adapter->hw.fc.pause_time = opt.def; } -if (!(adapter->hw.fc.type & ixgb_fc_rx_pause) ) -printk (KERN_INFO +if ( !(adapter->hw.fc.type & ixgb_fc_tx_pause) ) +printk (KERN_INFO "Ignoring FCReqTimeout when no RxFC\n" ); }/* high low and spacing check for rx flow control thresholds */ -if (adapter->hw.fc.type & ixgb_fc_rx_pause) { +if (adapter->hw.fc.type & ixgb_fc_tx_pause) {/* high must be greater than low */ if (adapter->hw.fc.high_water < (adapter->hw.fc.low_water +8 )) {/* set defaults */
PATCH 8/11 ixgb: Add support for copper 10GbE
In reply to Jeff Kirsher- Add support for Copper 10GbE device ID 109E
---
drivers/net/ixgb/ixgb_hw.h | 1 + drivers/net/ixgb/ixgb_ids.h | 4 +++- drivers/net/ixgb/ixgb_main.c | 7 +++++-- 3 files changed, 9 insertions(+), 3 deletions(-)
drivers/net/ixgb/ixgb_hw.h +1 -0
@@ -57,6 +57,7 @@ typedef enum {typedef enum { ixgb_media_type_unknown =0 , ixgb_media_type_fiber =1 , + ixgb_media_type_copper =2 , ixgb_num_media_types }ixgb_media_type ;
drivers/net/ixgb/ixgb_ids.h +3 -1
@@ -43,6 +43,8 @@#define IXGB_SUBDEVICE_ID_A11F 0xA11F#define IXGB_SUBDEVICE_ID_A01F 0xA01F -#endif /* #ifndef _IXGB_IDS_H_ */ +#define IXGB_DEVICE_ID_82597EX_CX4 0x109E +#define IXGB_SUBDEVICE_ID_A00C 0xA00C +#endif /* #ifndef _IXGB_IDS_H_ */ /* End of File */
drivers/net/ixgb/ixgb_main.c +5 -2
@@ -67,6 +67,8 @@ static char ixgb_copyright[] = "Copyrighstatic struct pci_device_id ixgb_pci_tbl[] = { {INTEL_VENDOR_ID ,IXGB_DEVICE_ID_82597EX ,PCI_ANY_ID ,PCI_ANY_ID ,0 ,0 ,0 }, + {INTEL_VENDOR_ID ,IXGB_DEVICE_ID_82597EX_CX4 , +PCI_ANY_ID ,PCI_ANY_ID ,0 ,0 ,0 }, {INTEL_VENDOR_ID ,IXGB_DEVICE_ID_82597EX_SR ,PCI_ANY_ID ,PCI_ANY_ID ,0 ,0 ,0 }, {INTEL_VENDOR_ID ,IXGB_DEVICE_ID_82597EX_LR ,
@@ -570,8 +572,9 @@ ixgb_sw_init(struct ixgb_adapter *adapte hw->max_frame_size = netdev->mtu +ENET_HEADER_SIZE +ENET_FCS_LENGTH ;if ((hw->device_id ==IXGB_DEVICE_ID_82597EX ) - ||(hw->device_id ==IXGB_DEVICE_ID_82597EX_LR ) - ||(hw->device_id ==IXGB_DEVICE_ID_82597EX_SR )) + || (hw->device_id ==IXGB_DEVICE_ID_82597EX_CX4 ) + || (hw->device_id ==IXGB_DEVICE_ID_82597EX_LR ) + || (hw->device_id ==IXGB_DEVICE_ID_82597EX_SR )) hw->mac_type = ixgb_82597;else {/* should never have loaded on this device */
PATCH 9/11 ixgb: Add performance enhancements to the buffer_info struct
In reply to Jeff Kirsher---
drivers/net/ixgb/ixgb_main.c | 36 ++++++++++++++++++++---------------- 1 files changed, 20 insertions(+), 16 deletions(-)
drivers/net/ixgb/ixgb_main.c +20 -16
@@ -922,17 +922,20 @@ ixgb_unmap_and_free_tx_resource(struct i struct ixgb_buffer *buffer_info) {struct pci_dev *pdev = adapter->pdev; -if (buffer_info->dma) { -pci_unmap_page (pdev, - buffer_info->dma, - buffer_info->length, -PCI_DMA_TODEVICE ); - buffer_info->dma =0 ; - } -if (buffer_info->skb) { + +if (buffer_info->dma) +pci_unmap_page (pdev, buffer_info->dma, buffer_info->length, +PCI_DMA_TODEVICE ); + +if (buffer_info->skb)dev_kfree_skb_any (buffer_info->skb); - buffer_info->skb =NULL ; - } + + buffer_info->skb =NULL ; + buffer_info->dma =0 ; + buffer_info->time_stamp =0 ; +/* these fields must always be initialized in tx +* buffer_info->length = 0; +* buffer_info->next_to_watch = 0; */ }/**
@@ -1307,6 +1310,7 @@ ixgb_tx_map(struct ixgb_adapter *adapter size, PCI_DMA_TODEVICE); buffer_info->time_stamp = jiffies; + buffer_info->next_to_watch =0 ; len -= size; offset += size;
@@ -1338,6 +1342,7 @@ ixgb_tx_map(struct ixgb_adapter *adapter size, PCI_DMA_TODEVICE); buffer_info->time_stamp = jiffies; + buffer_info->next_to_watch =0 ; len -= size; offset += size;
@@ -1909,6 +1914,7 @@ ixgb_clean_rx_irq(struct ixgb_adapter *a#endif status = rx_desc->status; skb = buffer_info->skb; + buffer_info->skb =NULL ;prefetch (skb->data);
@@ -1982,7 +1988,6 @@ ixgb_clean_rx_irq(struct ixgb_adapter *a rxdesc_done:/* clean up descriptor, might be written over by hw */ rx_desc->status =0 ; - buffer_info->skb =NULL ;/* use prefetched values */ rx_desc = next_rxd;
@@ -2041,11 +2046,10 @@ ixgb_alloc_rx_buffers(struct ixgb_adapte buffer_info->skb = skb; buffer_info->length = adapter->rx_buffer_len; - buffer_info->dma = -pci_map_single (pdev, - skb->data, - adapter->rx_buffer_len, -PCI_DMA_FROMDEVICE ); + buffer_info->dma =pci_map_single (pdev, + skb->data, + adapter->rx_buffer_len, +PCI_DMA_FROMDEVICE ); rx_desc->buff_addr =cpu_to_le64 (buffer_info->dma);/* guarantee DD bit not set now before h/w gets descriptor
PATCH 10/11 ixgb: clean up whitespace
In reply to Jeff Kirsher- remove trailing whitespace
---
drivers/net/ixgb/ixgb.h | 28 ++++++++-------- drivers/net/ixgb/ixgb_ethtool.c | 68 ++++++++++++++++++++------------------- drivers/net/ixgb/ixgb_main.c | 62 ++++++++++++++++++------------------ drivers/net/ixgb/ixgb_osdep.h | 26 +++++++-------- 4 files changed, 92 insertions(+), 92 deletions(-)
drivers/net/ixgb/ixgb.h +14 -14
@@ -1,25 +1,25 @@/******************************************************************************* -+ Copyright(c) 1999 - 2005 Intel Corporation. All rights reserved. -- This program is free software; you can redistribute it and/or modify it -under the terms of the GNU General Public License as published by the Free -Software Foundation; either version 2 of the License, or (at your option) + +This program is free software; you can redistribute it and/or modify it +under the terms of the GNU General Public License as published by the Free +Software Foundation; either version 2 of the License, or (at your option) any later version. -- This program is distributed in the hope that it will be useful, but WITHOUT -ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or -FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + +This program is distributed in the hope that it will be useful, but WITHOUT +ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or +FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. -+ You should have received a copy of the GNU General Public License along with -this program; if not, write to the Free Software Foundation, Inc., 59 +this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. -+ The full GNU General Public License is included in this distribution in the file called LICENSE. -+ Contact Information: Linux NICS <[email protected]> Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
@@ -97,7 +97,7 @@ struct ixgb_adapter;#define MIN_TXD 64/* hardware cannot reliably support more than 512 descriptors owned by -* hardware descrioptor cache otherwise an unreliable ring under heavy +* hardware descrioptor cache otherwise an unreliable ring under heavy * recieve load may result */ /* #define DEFAULT_RXD 1024 */ /* #define MAX_RXD 4096 */
drivers/net/ixgb/ixgb_ethtool.c +34 -34
183 lines not on this page · raw message
drivers/net/ixgb/ixgb_main.c +31 -31
165 lines not on this page · raw message
drivers/net/ixgb/ixgb_osdep.h +13 -13
@@ -1,25 +1,25 @@/******************************************************************************* -+ Copyright(c) 1999 - 2005 Intel Corporation. All rights reserved. -- This program is free software; you can redistribute it and/or modify it -under the terms of the GNU General Public License as published by the Free -Software Foundation; either version 2 of the License, or (at your option) + +This program is free software; you can redistribute it and/or modify it +under the terms of the GNU General Public License as published by the Free +Software Foundation; either version 2 of the License, or (at your option) any later version. -- This program is distributed in the hope that it will be useful, but WITHOUT -ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or -FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + +This program is distributed in the hope that it will be useful, but WITHOUT +ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or +FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. -+ You should have received a copy of the GNU General Public License along with -this program; if not, write to the Free Software Foundation, Inc., 59 +this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. -+ The full GNU General Public License is included in this distribution in the file called LICENSE. -+ Contact Information: Linux NICS <[email protected]> Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
PATCH 11/11 ixgb: Add prefetch
In reply to Jeff Kirsher- This patch is to improve performance by adding prefetch to the ixgb driver
- Add driver comments
---
drivers/net/ixgb/ixgb_main.c | 10 +++++++++- 1 files changed, 9 insertions(+), 1 deletions(-)
drivers/net/ixgb/ixgb_main.c +9 -1
@@ -29,6 +29,13 @@#include "ixgb.h" /* Change Log +* 1.0.104 10-Jan-2006 +* - fix for copybreak/recycle +* 1.0.103 Oct-3 +* - suck in some e1000 changes, including copybreak and LLTX +* - support for CX4 adapters +* 1.0.102 June-20-2005 +* - add a workaround for a hardware issue when using TSO * 1.0.96 04/19/05 * - Make needlessly global code static -- [email protected] * - ethtool cleanup -- [email protected]
@@ -1916,7 +1923,7 @@ ixgb_clean_rx_irq(struct ixgb_adapter *a skb = buffer_info->skb; buffer_info->skb =NULL ; -prefetch (skb->data); +prefetch (skb->data -NET_IP_ALIGN );if (++i == rx_ring->count) i =0 ; next_rxd =IXGB_RX_DESC (*rx_ring, i);
@@ -1929,6 +1936,7 @@ ixgb_clean_rx_irq(struct ixgb_adapter *a next_buffer = &rx_ring->buffer_info[i]; next_skb = next_buffer->skb;prefetch (next_skb); +prefetch (next_skb->data -NET_IP_ALIGN ); cleaned = TRUE;
Re: [PATCH 11/11] ixgb: Add prefetch
Quoted 30 lines · Jeff Kirsher
- This patch is to improve performance by adding prefetch to the ixgb driver
- Add driver commentsSigned-off-by Jeff Kirsher <[email protected]>Signed-off-by Jesse Brandeburg <[email protected]>Signed-off-by John Ronciak <[email protected]>---
drivers/net/ixgb/ixgb_main.c | 10 +++++++++- 1 files changed, 9 insertions(+), 1 deletions(-)drivers/net/ixgb/ixgb_main.c +7
-0@@ -29,6 +29,13 @@#include "ixgb.h" /* Change Log +* 1.0.104 10-Jan-2006 +* - fix for copybreak/recycle +* 1.0.103 Oct-3 +* - suck in some e1000 changes, including copybreak and LLTX +* - support for CX4 adapters +* 1.0.102 June-20-2005 +* - add a workaround for a hardware issue when using TSO * 1.0.96 04/19/05 * - Make needlessly global code static -- [email protected] * - ethtool cleanup -- [email protected]
drivers/net/ixgb/ixgb_main.c ixgb_clean_rx_irq(struct ixgb_adapter *a · line 1926
Quoted 5 lines
drivers/net/ixgb/ixgb_main.c +1
-1@@ -1916,7 +1923,7 @@ ixgb_clean_rx_irq(struct ixgb_adapter *a skb = buffer_info->skb; buffer_info->skb =NULL ; -prefetch (skb->data); +prefetch (skb->data -NET_IP_ALIGN );
I doubt this change is usefull.
skb->data and dkb->dta - NET_IP_ALIGN are on the same cache line.
So prefetch(skb->data) is cheaper for he compiler and has the same effect on the memory prefetch that is eventually done by the cpu.
drivers/net/ixgb/ixgb_main.c ixgb_clean_rx_irq(struct ixgb_adapter *a · line 1939
Quoted 8 lines
drivers/net/ixgb/ixgb_main.c +1
-0if (++i == rx_ring->count) i =0 ; next_rxd =IXGB_RX_DESC (*rx_ring, i);@@ -1929,6 +1936,7 @@ ixgb_clean_rx_irq(struct ixgb_adapter *a next_buffer = &rx_ring->buffer_info[i]; next_skb = next_buffer->skb;prefetch (next_skb); +prefetch (next_skb->data -NET_IP_ALIGN );
I doubt that next->skb_data is available for free just after a prefetch(next_skb). This second prefetch has a hidden cost : The memory location (&next_skb->data) must be in L1 cache. So basically the prefetch(next_skb) is useless...
prefetch are not magic things. They have a cost (they increase the code size), and should be used carefully.
Eric
Quoted 3 lines · Jeff Kirsher
The following patches were diff's against Garzik's latest "upstream" branch and can be pulled from the following location:
git://lost.foo-projects.org/linux-2.6.gitixgb-upstream
pull failed:
[jgarzik@pretzel netdev-2.6]$ git pull git://lost.foo-projects.org/linux-2.6.git ixgb-upstream
error: no such remote ref refs/heads/ixgb-upstream
Fetch failure: git://lost.foo-projects.org/linux-2.6.git