| 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
| ||
|
Date: Tue, 2 Dec 2008 17:40:32 +0800
From: Yu Zhao <yu.zhao@...el.com>
To: "linux-pci@...r.kernel.org" <linux-pci@...r.kernel.org>
Cc: "greg@...ah.com" <greg@...ah.com>,
"matthew@....cx" <matthew@....cx>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"kvm@...r.kernel.org" <kvm@...r.kernel.org>,
"xen-devel@...ts.xensource.com" <xen-devel@...ts.xensource.com>,
"virtualization@...ts.linux-foundation.org"
<virtualization@...ts.linux-foundation.org>,
"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
"Nakajima, Jun" <jun.nakajima@...el.com>,
"Rose, Gregory V" <gregory.v.rose@...el.com>
Subject: [SR-IOV driver example 1/3 resend] PF driver: hardware specific
operations
This patch makes the IGB driver allocate hardware resource (rx/tx queues)
for Virtual Functions. All operations in this patch are hardware specific.
From: Intel Corporation, LAN Access Division <e1000-devel@...ts.sourceforge.net>
Signed-off-by: Yu Zhao <yu.zhao@...el.com>
---
drivers/net/igb/Makefile | 2 +-
drivers/net/igb/e1000_82575.c | 1 +
drivers/net/igb/e1000_82575.h | 61 +++++
drivers/net/igb/e1000_defines.h | 7 +
drivers/net/igb/e1000_hw.h | 2 +
drivers/net/igb/e1000_regs.h | 13 +
drivers/net/igb/igb.h | 8 +
drivers/net/igb/igb_main.c | 567 +++++++++++++++++++++++++++++++++++++-
drivers/pci/iov.c | 6 +-
9 files changed, 649 insertions(+), 18 deletions(-)
diff --git a/drivers/net/igb/Makefile b/drivers/net/igb/Makefile
index 1927b3f..ab3944c 100644
--- a/drivers/net/igb/Makefile
+++ b/drivers/net/igb/Makefile
@@ -33,5 +33,5 @@
obj-$(CONFIG_IGB) += igb.o
igb-objs := igb_main.o igb_ethtool.o e1000_82575.o \
- e1000_mac.o e1000_nvm.o e1000_phy.o
+ e1000_mac.o e1000_nvm.o e1000_phy.o e1000_vf.o
diff --git a/drivers/net/igb/e1000_82575.c b/drivers/net/igb/e1000_82575.c
index f5e2e72..bb823ac 100644
--- a/drivers/net/igb/e1000_82575.c
+++ b/drivers/net/igb/e1000_82575.c
@@ -87,6 +87,7 @@ static s32 igb_get_invariants_82575(struct e1000_hw *hw)
case E1000_DEV_ID_82576:
case E1000_DEV_ID_82576_FIBER:
case E1000_DEV_ID_82576_SERDES:
+ case E1000_DEV_ID_82576_QUAD_COPPER:
mac->type = e1000_82576;
break;
default:
diff --git a/drivers/net/igb/e1000_82575.h b/drivers/net/igb/e1000_82575.h
index c1928b5..8c488ab 100644
--- a/drivers/net/igb/e1000_82575.h
+++ b/drivers/net/igb/e1000_82575.h
@@ -170,4 +170,65 @@ struct e1000_adv_tx_context_desc {
#define E1000_DCA_TXCTRL_CPUID_SHIFT 24 /* Tx CPUID now in the last byte */
#define E1000_DCA_RXCTRL_CPUID_SHIFT 24 /* Rx CPUID now in the last byte */
+#define MAX_NUM_VFS 8
+
+#define E1000_DTXSWC_VMDQ_LOOPBACK_EN (1 << 31) /* global VF LB enable */
+
+/* Easy defines for setting default pool, would normally be left a zero */
+#define E1000_VT_CTL_DEFAULT_POOL_SHIFT 7
+#define E1000_VT_CTL_DEFAULT_POOL_MASK (0x7 << E1000_VT_CTL_DEFAULT_POOL_SHIFT)
+
+/* Other useful VMD_CTL register defines */
+#define E1000_VT_CTL_DISABLE_DEF_POOL (1 << 29)
+#define E1000_VT_CTL_VM_REPL_EN (1 << 30)
+
+/* Per VM Offload register setup */
+#define E1000_VMOLR_LPE 0x00010000 /* Accept Long packet */
+#define E1000_VMOLR_AUPE 0x01000000 /* Accept untagged packets */
+#define E1000_VMOLR_BAM 0x08000000 /* Accept Broadcast packets */
+#define E1000_VMOLR_MPME 0x10000000 /* Multicast promiscuous mode */
+#define E1000_VMOLR_STRVLAN 0x40000000 /* Vlan stripping enable */
+
+#define E1000_P2VMAILBOX_STS 0x00000001 /* Initiate message send to VF */
+#define E1000_P2VMAILBOX_ACK 0x00000002 /* Ack message recv'd from VF */
+#define E1000_P2VMAILBOX_VFU 0x00000004 /* VF owns the mailbox buffer */
+#define E1000_P2VMAILBOX_PFU 0x00000008 /* PF owns the mailbox buffer */
+
+#define E1000_VLVF_ARRAY_SIZE 32
+#define E1000_VLVF_VLANID_MASK 0x00000FFF
+#define E1000_VLVF_POOLSEL_SHIFT 12
+#define E1000_VLVF_POOLSEL_MASK (0xFF << E1000_VLVF_POOLSEL_SHIFT)
+#define E1000_VLVF_VLANID_ENABLE 0x80000000
+
+#define E1000_VFMAILBOX_SIZE 16 /* 16 32 bit words - 64 bytes */
+
+/* If it's a E1000_VF_* msg then it originates in the VF and is sent to the
+ * PF. The reverse is true if it is E1000_PF_*.
+ * Message ACK's are the value or'd with 0xF0000000
+ */
+#define E1000_VT_MSGTYPE_ACK 0xF0000000 /* Messages below or'd with
+ * this are the ACK */
+#define E1000_VT_MSGTYPE_NACK 0xFF000000 /* Messages below or'd with
+ * this are the NACK */
+#define E1000_VT_MSGINFO_SHIFT 16
+/* bits 23:16 are used for exra info for certain messages */
+#define E1000_VT_MSGINFO_MASK (0xFF << E1000_VT_MSGINFO_SHIFT)
+
+#define E1000_VF_MSGTYPE_REQ_MAC 1 /* VF needs to know its MAC */
+#define E1000_VF_MSGTYPE_VFLR 2 /* VF notifies VFLR to PF */
+#define E1000_VF_SET_MULTICAST 3 /* VF requests PF to set MC addr */
+#define E1000_VF_SET_VLAN 4 /* VF requests PF to set VLAN */
+#define E1000_VF_SET_LPE 5 /* VF requests PF to set VMOLR.LPE */
+
+s32 e1000_send_mail_to_vf(struct e1000_hw *hw, u32 *msg,
+ u32 vf_number, s16 size);
+s32 e1000_receive_mail_from_vf(struct e1000_hw *hw, u32 *msg,
+ u32 vf_number, s16 size);
+void e1000_vmdq_loopback_enable_vf(struct e1000_hw *hw);
+void e1000_vmdq_loopback_disable_vf(struct e1000_hw *hw);
+void e1000_vmdq_replication_enable_vf(struct e1000_hw *hw, u32 enables);
+void e1000_vmdq_replication_disable_vf(struct e1000_hw *hw);
+bool e1000_check_for_pf_ack_vf(struct e1000_hw *hw);
+bool e1000_check_for_pf_mail_vf(struct e1000_hw *hw, u32*);
+
#endif
diff --git a/drivers/net/igb/e1000_defines.h b/drivers/net/igb/e1000_defines.h
index ce70068..08f9db0 100644
--- a/drivers/net/igb/e1000_defines.h
+++ b/drivers/net/igb/e1000_defines.h
@@ -389,6 +389,7 @@
#define E1000_ICR_RXDMT0 0x00000010 /* rx desc min. threshold (0) */
#define E1000_ICR_RXO 0x00000040 /* rx overrun */
#define E1000_ICR_RXT0 0x00000080 /* rx timer intr (ring 0) */
+#define E1000_ICR_VMMB 0x00000100 /* VM MB event */
#define E1000_ICR_MDAC 0x00000200 /* MDIO access complete */
#define E1000_ICR_RXCFG 0x00000400 /* Rx /c/ ordered set */
#define E1000_ICR_GPI_EN0 0x00000800 /* GP Int 0 */
@@ -451,6 +452,7 @@
/* Interrupt Mask Set */
#define E1000_IMS_TXDW E1000_ICR_TXDW /* Transmit desc written back */
#define E1000_IMS_LSC E1000_ICR_LSC /* Link Status Change */
+#define E1000_IMS_VMMB E1000_ICR_VMMB /* Mail box activity */
#define E1000_IMS_RXSEQ E1000_ICR_RXSEQ /* rx sequence error */
#define E1000_IMS_RXDMT0 E1000_ICR_RXDMT0 /* rx desc min. threshold */
#define E1000_IMS_RXT0 E1000_ICR_RXT0 /* rx timer intr */
@@ -768,4 +770,9 @@
#define E1000_GEN_CTL_ADDRESS_SHIFT 8
#define E1000_GEN_POLL_TIMEOUT 640
+#define E1000_WRITE_FLUSH(a) (readl((a)->hw_addr + E1000_STATUS))
+#define E1000_MRQC_ENABLE_MASK 0x00000007
+#define E1000_MRQC_ENABLE_VMDQ 0x00000003
+#define E1000_CTRL_EXT_PFRSTD 0x00004000
+
#endif
diff --git a/drivers/net/igb/e1000_hw.h b/drivers/net/igb/e1000_hw.h
index 99504a6..b57ecfd 100644
--- a/drivers/net/igb/e1000_hw.h
+++ b/drivers/net/igb/e1000_hw.h
@@ -41,6 +41,7 @@ struct e1000_hw;
#define E1000_DEV_ID_82576 0x10C9
#define E1000_DEV_ID_82576_FIBER 0x10E6
#define E1000_DEV_ID_82576_SERDES 0x10E7
+#define E1000_DEV_ID_82576_QUAD_COPPER 0x10E8
#define E1000_DEV_ID_82575EB_COPPER 0x10A7
#define E1000_DEV_ID_82575EB_FIBER_SERDES 0x10A9
#define E1000_DEV_ID_82575GB_QUAD_COPPER 0x10D6
@@ -91,6 +92,7 @@ enum e1000_phy_type {
e1000_phy_gg82563,
e1000_phy_igp_3,
e1000_phy_ife,
+ e1000_phy_vf,
};
enum e1000_bus_type {
diff --git a/drivers/net/igb/e1000_regs.h b/drivers/net/igb/e1000_regs.h
index 95523af..8a39bbc 100644
--- a/drivers/net/igb/e1000_regs.h
+++ b/drivers/net/igb/e1000_regs.h
@@ -262,6 +262,19 @@
#define E1000_RETA(_i) (0x05C00 + ((_i) * 4))
#define E1000_RSSRK(_i) (0x05C80 + ((_i) * 4)) /* RSS Random Key - RW Array */
+/* VT Registers */
+#define E1000_MBVFICR 0x00C80 /* Mailbox VF Cause - RWC */
+#define E1000_MBVFIMR 0x00C84 /* Mailbox VF int Mask - RW */
+#define E1000_VFLRE 0x00C88 /* VF Register Events - RWC */
+#define E1000_VFRE 0x00C8C /* VF Receive Enables */
+#define E1000_VFTE 0x00C90 /* VF Transmit Enables */
+#define E1000_DTXSWC 0x03500 /* DMA Tx Switch Control - RW */
+/* These act per VF so an array friendly macro is used */
+#define E1000_P2VMAILBOX(_n) (0x00C00 + (4 * (_n)))
+#define E1000_VMBMEM(_n) (0x00800 + (64 * (_n)))
+#define E1000_VMOLR(_n) (0x05AD0 + (4 * (_n)))
+#define E1000_VLVF(_n) (0x05D00 + (4 * (_n))) /* VLAN Virtual Machine */
+
#define wr32(reg, value) (writel(value, hw->hw_addr + reg))
#define rd32(reg) (readl(hw->hw_addr + reg))
#define wrfl() ((void)rd32(E1000_STATUS))
diff --git a/drivers/net/igb/igb.h b/drivers/net/igb/igb.h
index 4ff6f05..47d474e 100644
--- a/drivers/net/igb/igb.h
+++ b/drivers/net/igb/igb.h
@@ -294,6 +294,14 @@ struct igb_adapter {
unsigned int lro_flushed;
unsigned int lro_no_desc;
#endif
+ unsigned int vfs_allocated_count;
+ struct work_struct msg_task;
+ u32 vf_icr;
+ u32 vflre;
+ unsigned char vf_mac_addresses[8][6];
+ u8 vfta_tracking_entry[128];
+ int int0counter;
+ int int1counter;
};
#define IGB_FLAG_HAS_MSI (1 << 0)
diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c
index 1cbae85..f0361ef 100644
--- a/drivers/net/igb/igb_main.c
+++ b/drivers/net/igb/igb_main.c
@@ -62,6 +62,7 @@ static struct pci_device_id igb_pci_tbl[] = {
{ PCI_VDEVICE(INTEL, E1000_DEV_ID_82576), board_82575 },
{ PCI_VDEVICE(INTEL, E1000_DEV_ID_82576_FIBER), board_82575 },
{ PCI_VDEVICE(INTEL, E1000_DEV_ID_82576_SERDES), board_82575 },
+ { PCI_VDEVICE(INTEL, E1000_DEV_ID_82576_QUAD_COPPER), board_82575 },
{ PCI_VDEVICE(INTEL, E1000_DEV_ID_82575EB_COPPER), board_82575 },
{ PCI_VDEVICE(INTEL, E1000_DEV_ID_82575EB_FIBER_SERDES), board_82575 },
{ PCI_VDEVICE(INTEL, E1000_DEV_ID_82575GB_QUAD_COPPER), board_82575 },
@@ -126,6 +127,17 @@ static void igb_vlan_rx_register(struct net_device *, struct vlan_group *);
static void igb_vlan_rx_add_vid(struct net_device *, u16);
static void igb_vlan_rx_kill_vid(struct net_device *, u16);
static void igb_restore_vlan(struct igb_adapter *);
+static void igb_msg_task(struct work_struct *);
+int igb_send_msg_to_vf(struct igb_adapter *, u32 *, u32);
+static int igb_get_vf_msg_ack(struct igb_adapter *, u32);
+static int igb_rcv_msg_from_vf(struct igb_adapter *, u32);
+static int igb_set_pf_mac(struct net_device *, int, u8*);
+static void igb_enable_pf_queues(struct igb_adapter *adapter);
+static void igb_set_vf_vmolr(struct igb_adapter *adapter, int vfn);
+void igb_set_mc_list_pools(struct igb_adapter *, struct e1000_hw *, int, u16);
+static int igb_vmm_control(struct igb_adapter *, bool);
+static int igb_set_vf_mac(struct net_device *, int, u8*);
+static void igb_mbox_handler(struct igb_adapter *);
static int igb_suspend(struct pci_dev *, pm_message_t);
#ifdef CONFIG_PM
@@ -169,7 +181,7 @@ static struct pci_driver igb_driver = {
.resume = igb_resume,
#endif
.shutdown = igb_shutdown,
- .err_handler = &igb_err_handler
+ .err_handler = &igb_err_handler,
};
static int global_quad_port_a; /* global quad port a indication */
@@ -292,6 +304,7 @@ static void igb_assign_vector(struct igb_adapter *adapter, int rx_queue,
u32 msixbm = 0;
struct e1000_hw *hw = &adapter->hw;
u32 ivar, index;
+ u32 rbase_offset = adapter->vfs_allocated_count;
switch (hw->mac.type) {
case e1000_82575:
@@ -316,9 +329,9 @@ static void igb_assign_vector(struct igb_adapter *adapter, int rx_queue,
a vector number along with a "valid" bit. Sadly, the layout
of the table is somewhat counterintuitive. */
if (rx_queue > IGB_N0_QUEUE) {
- index = (rx_queue & 0x7);
+ index = ((rx_queue + rbase_offset) & 0x7);
ivar = array_rd32(E1000_IVAR0, index);
- if (rx_queue < 8) {
+ if ((rx_queue + rbase_offset) < 8) {
/* vector goes into low byte of register */
ivar = ivar & 0xFFFFFF00;
ivar |= msix_vector | E1000_IVAR_VALID;
@@ -331,9 +344,9 @@ static void igb_assign_vector(struct igb_adapter *adapter, int rx_queue,
array_wr32(E1000_IVAR0, index, ivar);
}
if (tx_queue > IGB_N0_QUEUE) {
- index = (tx_queue & 0x7);
+ index = ((tx_queue + rbase_offset) & 0x7);
ivar = array_rd32(E1000_IVAR0, index);
- if (tx_queue < 8) {
+ if ((tx_queue + rbase_offset) < 8) {
/* vector goes into second byte of register */
ivar = ivar & 0xFFFF00FF;
ivar |= (msix_vector | E1000_IVAR_VALID) << 8;
@@ -419,6 +432,8 @@ static void igb_configure_msix(struct igb_adapter *adapter)
case e1000_82576:
tmp = (vector++ | E1000_IVAR_VALID) << 8;
wr32(E1000_IVAR_MISC, tmp);
+ if (adapter->vfs_allocated_count > 0)
+ wr32(E1000_MBVFIMR, 0xFF);
adapter->eims_enable_mask = (1 << (vector)) - 1;
adapter->eims_other = 1 << (vector - 1);
@@ -440,6 +455,7 @@ static int igb_request_msix(struct igb_adapter *adapter)
{
struct net_device *netdev = adapter->netdev;
int i, err = 0, vector = 0;
+ u32 rbase_offset = adapter->vfs_allocated_count;
vector = 0;
@@ -451,7 +467,7 @@ static int igb_request_msix(struct igb_adapter *adapter)
&(adapter->tx_ring[i]));
if (err)
goto out;
- ring->itr_register = E1000_EITR(0) + (vector << 2);
+ ring->itr_register = E1000_EITR(0 + rbase_offset) + (vector << 2);
ring->itr_val = 976; /* ~4000 ints/sec */
vector++;
}
@@ -466,7 +482,7 @@ static int igb_request_msix(struct igb_adapter *adapter)
&(adapter->rx_ring[i]));
if (err)
goto out;
- ring->itr_register = E1000_EITR(0) + (vector << 2);
+ ring->itr_register = E1000_EITR(0 + rbase_offset) + (vector << 2);
ring->itr_val = adapter->itr;
/* overwrite the poll routine for MSIX, we've already done
* netif_napi_add */
@@ -649,7 +665,11 @@ static void igb_irq_enable(struct igb_adapter *adapter)
wr32(E1000_EIAC, adapter->eims_enable_mask);
wr32(E1000_EIAM, adapter->eims_enable_mask);
wr32(E1000_EIMS, adapter->eims_enable_mask);
+#ifdef CONFIG_PCI_IOV
+ wr32(E1000_IMS, (E1000_IMS_LSC | E1000_IMS_VMMB));
+#else
wr32(E1000_IMS, E1000_IMS_LSC);
+#endif
} else {
wr32(E1000_IMS, IMS_ENABLE_MASK);
wr32(E1000_IAM, IMS_ENABLE_MASK);
@@ -773,6 +793,14 @@ int igb_up(struct igb_adapter *adapter)
if (adapter->msix_entries)
igb_configure_msix(adapter);
+ if (adapter->vfs_allocated_count > 0) {
+ igb_vmm_control(adapter, true);
+ igb_set_pf_mac(adapter->netdev,
+ adapter->vfs_allocated_count,
+ hw->mac.addr);
+ igb_enable_pf_queues(adapter);
+ }
+
/* Clear any pending interrupts. */
rd32(E1000_ICR);
igb_irq_enable(adapter);
@@ -1189,6 +1217,7 @@ static int __devinit igb_probe(struct pci_dev *pdev,
INIT_WORK(&adapter->reset_task, igb_reset_task);
INIT_WORK(&adapter->watchdog_task, igb_watchdog_task);
+ INIT_WORK(&adapter->msg_task, igb_msg_task);
/* Initialize link & ring properties that are user-changeable */
adapter->tx_ring->count = 256;
@@ -1404,8 +1433,13 @@ static int __devinit igb_sw_init(struct igb_adapter *adapter)
/* Number of supported queues. */
/* Having more queues than CPUs doesn't make sense. */
+#ifdef CONFIG_PCI_IOV
+ adapter->num_rx_queues = 1;
+ adapter->num_tx_queues = 1;
+#else
adapter->num_rx_queues = min((u32)IGB_MAX_RX_QUEUES, (u32)num_online_cpus());
adapter->num_tx_queues = min(IGB_MAX_TX_QUEUES, num_online_cpus());
+#endif
/* This call may decrease the number of queues depending on
* interrupt mode. */
@@ -1469,6 +1503,14 @@ static int igb_open(struct net_device *netdev)
* clean_rx handler before we do so. */
igb_configure(adapter);
+ if (adapter->vfs_allocated_count > 0) {
+ igb_vmm_control(adapter, true);
+ igb_set_pf_mac(netdev,
+ adapter->vfs_allocated_count,
+ hw->mac.addr);
+ igb_enable_pf_queues(adapter);
+ }
+
err = igb_request_irq(adapter);
if (err)
goto err_req_irq;
@@ -1623,9 +1665,10 @@ static void igb_configure_tx(struct igb_adapter *adapter)
u32 tctl;
u32 txdctl, txctrl;
int i;
+ u32 rbase_offset = adapter->vfs_allocated_count;
- for (i = 0; i < adapter->num_tx_queues; i++) {
- struct igb_ring *ring = &(adapter->tx_ring[i]);
+ for (i = rbase_offset; i < (adapter->num_tx_queues + rbase_offset); i++) {
+ struct igb_ring *ring = &(adapter->tx_ring[i - rbase_offset]);
wr32(E1000_TDLEN(i),
ring->count * sizeof(struct e1000_tx_desc));
@@ -1772,6 +1815,8 @@ static void igb_setup_rctl(struct igb_adapter *adapter)
u32 rctl;
u32 srrctl = 0;
int i;
+ u32 rbase_offset = adapter->vfs_allocated_count;
+ u32 vmolr;
rctl = rd32(E1000_RCTL);
@@ -1794,6 +1839,7 @@ static void igb_setup_rctl(struct igb_adapter *adapter)
rctl &= ~E1000_RCTL_LPE;
else
rctl |= E1000_RCTL_LPE;
+#ifndef CONFIG_PCI_IOV
if (adapter->rx_buffer_len <= IGB_RXBUFFER_2048) {
/* Setup buffer sizes */
rctl &= ~E1000_RCTL_SZ_4096;
@@ -1818,9 +1864,12 @@ static void igb_setup_rctl(struct igb_adapter *adapter)
break;
}
} else {
+#endif
rctl &= ~E1000_RCTL_BSEX;
srrctl = adapter->rx_buffer_len >> E1000_SRRCTL_BSIZEPKT_SHIFT;
+#ifndef CONFIG_PCI_IOV
}
+#endif
/* 82575 and greater support packet-split where the protocol
* header is placed in skb->data and the packet data is
@@ -1836,13 +1885,32 @@ static void igb_setup_rctl(struct igb_adapter *adapter)
srrctl |= adapter->rx_ps_hdr_size <<
E1000_SRRCTL_BSIZEHDRSIZE_SHIFT;
srrctl |= E1000_SRRCTL_DESCTYPE_HDR_SPLIT_ALWAYS;
+#ifdef CONFIG_PCI_IOV
+ srrctl |= 0x80000000;
+#endif
} else {
adapter->rx_ps_hdr_size = 0;
srrctl |= E1000_SRRCTL_DESCTYPE_ADV_ONEBUF;
}
- for (i = 0; i < adapter->num_rx_queues; i++)
+ for (i = rbase_offset; i < (adapter->num_rx_queues + rbase_offset); i++) {
wr32(E1000_SRRCTL(i), srrctl);
+ if ((rctl & E1000_RCTL_LPE) && adapter->vfs_allocated_count > 0 ) {
+ vmolr = rd32(E1000_VMOLR(i));
+ vmolr |= E1000_VMOLR_LPE;
+ wr32(E1000_VMOLR(i), vmolr);
+ }
+ }
+
+ /* Attention!!! For SR-IOV PF driver operations you must enable
+ * queue drop for the queue 0 or the PF driver will *never* receive
+ * any traffic on it's own default queue, which will be equal to the
+ * number of VFs enabled.
+ */
+ if (adapter->vfs_allocated_count > 0) {
+ srrctl = rd32(E1000_SRRCTL(0));
+ wr32(E1000_SRRCTL(0), (srrctl | 0x80000000));
+ }
wr32(E1000_RCTL, rctl);
}
@@ -1860,6 +1928,7 @@ static void igb_configure_rx(struct igb_adapter *adapter)
u32 rctl, rxcsum;
u32 rxdctl;
int i;
+ u32 rbase_offset = adapter->vfs_allocated_count;
/* disable receives while setting up the descriptors */
rctl = rd32(E1000_RCTL);
@@ -1872,8 +1941,8 @@ static void igb_configure_rx(struct igb_adapter *adapter)
/* Setup the HW Rx Head and Tail Descriptor Pointers and
* the Base and Length of the Rx Descriptor Ring */
- for (i = 0; i < adapter->num_rx_queues; i++) {
- struct igb_ring *ring = &(adapter->rx_ring[i]);
+ for (i = rbase_offset; i < (adapter->num_rx_queues + rbase_offset); i++) {
+ struct igb_ring *ring = &(adapter->rx_ring[i - rbase_offset]);
rdba = ring->dma;
wr32(E1000_RDBAL(i),
rdba & 0x00000000ffffffffULL);
@@ -2268,8 +2337,20 @@ static void igb_set_multi(struct net_device *netdev)
memcpy(mta_list + (i*ETH_ALEN), mc_ptr->dmi_addr, ETH_ALEN);
mc_ptr = mc_ptr->next;
}
- igb_update_mc_addr_list_82575(hw, mta_list, i, 1,
- mac->rar_entry_count);
+ if (adapter->vfs_allocated_count > 0) {
+ igb_update_mc_addr_list_82575(hw, mta_list, i,
+ adapter->vfs_allocated_count + 1,
+ mac->rar_entry_count);
+ igb_set_mc_list_pools(adapter, hw, i, mac->rar_entry_count);
+ /* TODO - if this is done after VF's are loaded and have their MC
+ * addresses set then we need to restore their entries in the MTA.
+ * This means we have to save them in the adapter structure somewhere
+ * so that we can retrieve them when this particular event occurs
+ */
+ } else
+ igb_update_mc_addr_list_82575(hw, mta_list, i, 1,
+ mac->rar_entry_count);
+
kfree(mta_list);
}
@@ -3274,6 +3355,22 @@ static irqreturn_t igb_msix_other(int irq, void *data)
struct e1000_hw *hw = &adapter->hw;
u32 icr = rd32(E1000_ICR);
+#ifdef CONFIG_PCI_IOV
+ adapter->int0counter++;
+
+ /* Check for a mailbox event */
+ if (icr & E1000_ICR_VMMB) {
+ adapter->vf_icr = rd32(E1000_MBVFICR);
+ /* Clear the bits */
+ wr32(E1000_MBVFICR, adapter->vf_icr);
+ E1000_WRITE_FLUSH(hw);
+ adapter->vflre = rd32(E1000_VFLRE);
+ wr32(E1000_VFLRE, adapter->vflre);
+ E1000_WRITE_FLUSH(hw);
+ igb_mbox_handler(adapter);
+ }
+#endif
+
/* reading ICR causes bit 31 of EICR to be cleared */
if (!(icr & E1000_ICR_LSC))
goto no_link_interrupt;
@@ -3283,7 +3380,10 @@ static irqreturn_t igb_msix_other(int irq, void *data)
mod_timer(&adapter->watchdog_timer, jiffies + 1);
no_link_interrupt:
- wr32(E1000_IMS, E1000_IMS_LSC);
+ if (adapter->vfs_allocated_count)
+ wr32(E1000_IMS, E1000_IMS_LSC | E1000_IMS_VMMB);
+ else
+ wr32(E1000_IMS, E1000_IMS_LSC);
wr32(E1000_EIMS, adapter->eims_other);
return IRQ_HANDLED;
@@ -3342,6 +3442,10 @@ static irqreturn_t igb_msix_rx(int irq, void *data)
* previous interrupt.
*/
+#ifdef CONFIG_PCI_IOV
+ adapter->int1counter++;
+#endif
+
igb_write_itr(rx_ring);
if (netif_rx_schedule_prep(adapter->netdev, &rx_ring->napi))
@@ -4192,6 +4296,9 @@ static void igb_vlan_rx_add_vid(struct net_device *netdev, u16 vid)
vfta = array_rd32(E1000_VFTA, index);
vfta |= (1 << (vid & 0x1F));
igb_write_vfta(&adapter->hw, index, vfta);
+#ifdef CONFIG_PCI_IOV
+ adapter->vfta_tracking_entry[index] = (u8)vfta;
+#endif
}
static void igb_vlan_rx_kill_vid(struct net_device *netdev, u16 vid)
@@ -4219,6 +4326,9 @@ static void igb_vlan_rx_kill_vid(struct net_device *netdev, u16 vid)
vfta = array_rd32(E1000_VFTA, index);
vfta &= ~(1 << (vid & 0x1F));
igb_write_vfta(&adapter->hw, index, vfta);
+#ifdef CONFIG_PCI_IOV
+ adapter->vfta_tracking_entry[index] = (u8)vfta;
+#endif
}
static void igb_restore_vlan(struct igb_adapter *adapter)
@@ -4529,4 +4639,431 @@ static void igb_io_resume(struct pci_dev *pdev)
}
+static void igb_set_vf_multicasts(struct igb_adapter *adapter,
+ u32 *msgbuf, u32 vf)
+{
+ struct e1000_hw *hw = &adapter->hw;
+ int n = (msgbuf[0] & E1000_VT_MSGINFO_MASK) >> E1000_VT_MSGINFO_SHIFT;
+ int i;
+ u32 hash_value;
+ u8 *p = (u8 *)&msgbuf[1];
+
+ /* VFs are limited to using the MTA hash table for their multicast
+ * addresses */
+ for (i = 0; i < n; i++) {
+ hash_value = igb_hash_mc_addr(hw, p);
+ printk("Adding MC Addr: %2.2X:%2.2X:%2.2X:%2.2X:%2.2X:%2.2X\n"
+ "for VF %d\n",
+ p[0],
+ p[1],
+ p[2],
+ p[3],
+ p[4],
+ p[5],
+ vf);
+ printk("Hash value = 0x%03X\n", hash_value);
+ igb_mta_set(hw, hash_value);
+ p += ETH_ALEN;
+ }
+}
+
+static void igb_set_vf_vlan(struct igb_adapter *adapter,
+ u32 *msgbuf, u32 vf)
+{
+ struct e1000_hw *hw = &adapter->hw;
+ int add = (msgbuf[0] & E1000_VT_MSGINFO_MASK) >> E1000_VT_MSGINFO_SHIFT;
+ int vid = (msgbuf[1] & E1000_VLVF_VLANID_MASK);
+ u32 reg, index, vfta;
+ int i;
+
+ if (add) {
+ /* See if a vlan filter for this id is already
+ * set and enabled */
+ for(i = 0; i < E1000_VLVF_ARRAY_SIZE; i++) {
+ reg = rd32(E1000_VLVF(i));
+ if ((reg & E1000_VLVF_VLANID_ENABLE) &&
+ vid == (reg & E1000_VLVF_VLANID_MASK))
+ break;
+ }
+ if (i < E1000_VLVF_ARRAY_SIZE) {
+ /* Found an enabled entry with the same VLAN
+ * ID. Just enable the pool select bit for
+ * this requesting VF
+ */
+ reg |= 1 << (E1000_VLVF_POOLSEL_SHIFT + vf);
+ wr32(E1000_VLVF(i), reg);
+ msgbuf[0] |= E1000_VT_MSGTYPE_ACK;
+ } else {
+ /* Did not find a matching VLAN ID filter entry
+ * that was also enabled. Search for a free
+ * filter entry, i.e. one without the enable
+ * bit set
+ */
+ for(i = 0; i < E1000_VLVF_ARRAY_SIZE; i++) {
+ reg = rd32(E1000_VLVF(i));
+ if (!(reg & E1000_VLVF_VLANID_ENABLE))
+ break;
+ }
+ if (i == E1000_VLVF_ARRAY_SIZE) {
+ /* oops, no free entry, send nack */
+ msgbuf[0] |= E1000_VT_MSGTYPE_NACK;
+ } else {
+ /* add VID to filter table */
+ index = (vid >> 5) & 0x7F;
+ vfta = array_rd32(E1000_VFTA, index);
+ vfta |= (1 << (vid & 0x1F));
+ igb_write_vfta(hw, index, vfta);
+ reg |= vid;
+ reg |= 1 << (E1000_VLVF_POOLSEL_SHIFT + vf);
+ reg |= E1000_VLVF_VLANID_ENABLE;
+ wr32(E1000_VLVF(i), reg);
+ msgbuf[0] |= E1000_VT_MSGTYPE_ACK;
+ }
+ }
+ } else {
+ /* Find the vlan filter for this id */
+ for(i = 0; i < E1000_VLVF_ARRAY_SIZE; i++) {
+ reg = rd32(E1000_VLVF(i));
+ if ((reg & E1000_VLVF_VLANID_ENABLE) &&
+ vid == (reg & E1000_VLVF_VLANID_MASK))
+ break;
+ }
+ if (i == E1000_VLVF_ARRAY_SIZE) {
+ /* oops, not found. send nack */
+ msgbuf[0] |= E1000_VT_MSGTYPE_NACK;
+ } else {
+ u32 pool_sel;
+ /* Check to see if the entry belongs to more than one
+ * pool. If so just reset this VF's pool select bit
+ */
+ /* mask off the pool select bits */
+ pool_sel = (reg & E1000_VLVF_POOLSEL_MASK) >>
+ E1000_VLVF_POOLSEL_SHIFT;
+ /* reset this VF's pool select bit */
+ pool_sel &= ~(1 << vf);
+ /* check if other pools are set */
+ if (pool_sel != 0) {
+ reg &= ~(E1000_VLVF_POOLSEL_MASK);
+ reg |= pool_sel;
+ } else {
+ /* just disable the whole entry */
+ reg = 0;
+ /* remove VID from filter table *IF AND
+ * ONLY IF!!!* this entry was enabled for
+ * VFs only through a write to the VFTA
+ * table a few lines above here in this
+ * function. If this VFTA entry was added
+ * through the rx_add_vid function then
+ * we can't delete it here. */
+ index = (vid >> 5) & 0x7F;
+ if (adapter->vfta_tracking_entry[index] == 0) {
+ vfta = array_rd32(E1000_VFTA, index);
+ vfta &= ~(1 << (vid & 0x1F));
+ igb_write_vfta(hw, index, vfta);
+ }
+ }
+ wr32(E1000_VLVF(i), reg);
+ msgbuf[0] |= E1000_VT_MSGTYPE_ACK;
+ }
+ }
+}
+
+static void igb_msg_task(struct work_struct *work)
+{
+ struct igb_adapter *adapter;
+ struct e1000_hw *hw;
+ u32 bit, vf, vfr;
+ u32 vflre;
+ u32 vf_icr;
+
+ adapter = container_of(work, struct igb_adapter, msg_task);
+ hw = &adapter->hw;
+
+ vflre = adapter->vflre;
+ vf_icr = adapter->vf_icr;
+
+ /* Now that we have salted away local values of these events
+ * for processing we can enable the interrupt so more events
+ * can be captured
+ */
+
+ wr32(E1000_IMS, E1000_IMS_VMMB);
+
+ if (vflre & 0xFF) {
+ printk("VFLR Event %2.2X\n", vflre);
+ vfr = rd32(E1000_VFRE);
+ wr32(E1000_VFRE, vfr | vflre);
+ E1000_WRITE_FLUSH(hw);
+ vfr = rd32(E1000_VFTE);
+ wr32(E1000_VFTE, vfr | vflre);
+ E1000_WRITE_FLUSH(hw);
+ }
+
+ if (!vf_icr)
+ return;
+
+ /* Check for message acks from VF first as that may affect
+ * pending messages to the VF
+ */
+ for (bit = 1, vf = 0; bit < 0x100; bit <<= 1, vf++) {
+ if ((bit << 16) & vf_icr)
+ igb_get_vf_msg_ack(adapter, vf);
+ }
+
+ /* Check for message sent from a VF */
+ for (bit = 1, vf = 0; bit < 0x100; bit <<= 1, vf++) {
+ if (bit & vf_icr)
+ igb_rcv_msg_from_vf(adapter, vf);
+ }
+}
+
+int igb_send_msg_to_vf(struct igb_adapter *adapter, u32 *msg, u32 vfn)
+{
+ struct e1000_hw *hw = &adapter->hw;
+
+ return e1000_send_mail_to_vf(hw, msg, vfn, 16);
+}
+
+static int igb_get_vf_msg_ack(struct igb_adapter *adapter, u32 vf)
+{
+ return 0;
+}
+
+static int igb_rcv_msg_from_vf(struct igb_adapter *adapter, u32 vf)
+{
+ u32 msgbuf[E1000_VFMAILBOX_SIZE];
+ struct net_device *netdev = adapter->netdev;
+ struct e1000_hw *hw = &adapter->hw;
+ u32 reg;
+ s32 retval;
+ int err = 0;
+
+ retval = e1000_receive_mail_from_vf(hw, msgbuf, vf, 16);
+
+ switch ((msgbuf[0] & 0xFFFF)) {
+ case E1000_VF_MSGTYPE_REQ_MAC:
+ {
+ unsigned char *p;
+ msgbuf[0] |= E1000_VT_MSGTYPE_ACK;
+ p = (char *)&msgbuf[1];
+ memcpy(p, adapter->vf_mac_addresses[vf], ETH_ALEN);
+ if ((err = igb_send_msg_to_vf(adapter, msgbuf, vf)
+ == 0)) {
+ printk(KERN_INFO "Sending MAC Address %2.2x:%2.2x:"
+ "%2.2x:%2.2x:%2.2x:%2.2x to VF %d\n",
+ p[0], p[1], p[2], p[3], p[4], p[5], vf);
+ igb_set_vf_mac(netdev,
+ vf,
+ adapter->vf_mac_addresses[vf]);
+ igb_set_vf_vmolr(adapter, vf);
+ }
+ else {
+ printk(KERN_ERR "Error %d Sending MAC Address to VF\n",
+ err);
+ }
+ }
+ break;
+ case E1000_VF_MSGTYPE_VFLR:
+ {
+ u32 vfe = rd32(E1000_VFTE);
+ vfe |= (1 << vf);
+ wr32(E1000_VFTE, vfe);
+ vfe = rd32(E1000_VFRE);
+ vfe |= (1 << vf);
+ wr32(E1000_VFRE, vfe);
+ printk(KERN_INFO "Enabling VFTE and VFRE for vf %d\n",
+ vf);
+ msgbuf[0] |= E1000_VT_MSGTYPE_ACK;
+ if ((err = igb_send_msg_to_vf(adapter, msgbuf, vf)
+ != 0))
+ printk(KERN_ERR "Error %d Sending VFLR Ack"
+ "to VF\n", err);
+ }
+ break;
+ case E1000_VF_SET_MULTICAST:
+ igb_set_vf_multicasts(adapter, msgbuf, vf);
+ break;
+ case E1000_VF_SET_LPE:
+ /* Make sure global LPE is set */
+ reg = rd32(E1000_RCTL);
+ reg |= E1000_RCTL_LPE;
+ wr32(E1000_RCTL, reg);
+ /* Set per VM LPE */
+ reg = rd32(E1000_VMOLR(vf));
+ reg |= E1000_VMOLR_LPE;
+ wr32(E1000_VMOLR(vf), reg);
+ msgbuf[0] |= E1000_VT_MSGTYPE_ACK;
+ if ((err = igb_send_msg_to_vf(adapter, msgbuf, vf) != 0))
+ printk(KERN_ERR "Error %d Sending set VMOLR LPE Ack"
+ "to VF\n", err);
+ break;
+ case E1000_VF_SET_VLAN:
+ igb_set_vf_vlan(adapter, msgbuf, vf);
+ if ((err = igb_send_msg_to_vf(adapter, msgbuf, vf) != 0))
+ printk(KERN_ERR "Error %d Sending set VLAN ID Ack"
+ "to VF\n", err);
+ break;
+ default:
+ if ((msgbuf[0] & 0xFF000000) != E1000_VT_MSGTYPE_ACK &&
+ (msgbuf[0] & 0xFF000000) != E1000_VT_MSGTYPE_NACK)
+ printk(KERN_ERR "Unhandled Msg %8.8x\n", msgbuf[0]);
+ break;
+ }
+
+ return retval;
+}
+
+static void igb_mbox_handler(struct igb_adapter *adapter)
+{
+ schedule_work(&adapter->msg_task);
+}
+
+#define E1000_RAH(_i) (((_i) <= 15) ? (0x05404 + ((_i) * 8)) : (0x054E4 + ((_i - 16) * 8)))
+
+static int igb_set_pf_mac(struct net_device *netdev, int queue, u8*mac_addr)
+{
+ struct igb_adapter *adapter;
+ struct e1000_hw *hw;
+ u32 reg_data;
+
+ adapter = netdev_priv(netdev);
+ hw = &adapter->hw;
+
+ /* point the pool selector for our default MAC entry to
+ * the right pool, which is equal to the number of vfs enabled.
+ */
+ reg_data = rd32(E1000_RAH(0));
+ reg_data |= (1 << (18 + queue));
+ wr32(E1000_RAH(0), reg_data);
+
+ return 0;
+}
+
+static void igb_set_vf_vmolr(struct igb_adapter *adapter, int vfn)
+{
+ struct e1000_hw *hw = &adapter->hw;
+ u32 reg_data;
+
+ reg_data = rd32(E1000_VMOLR(vfn));
+ reg_data |= 0xF << 24; /* aupe, rompe, rope, bam */
+ reg_data |= E1000_VMOLR_STRVLAN; /* Strip vlan tags */
+ wr32(E1000_VMOLR(vfn), reg_data);
+}
+
+static int igb_set_vf_mac(struct net_device *netdev,
+ int vf,
+ unsigned char *mac_addr)
+{
+ struct igb_adapter *adapter;
+ struct e1000_hw *hw;
+ u32 reg_data;
+ int rar_entry = vf + 1; /* VF MAC addresses start at entry 1 */
+
+ adapter = netdev_priv(netdev);
+ hw = &adapter->hw;
+
+ igb_rar_set(hw, mac_addr, rar_entry);
+
+ memcpy(adapter->vf_mac_addresses[vf], mac_addr, 6);
+
+ reg_data = rd32(E1000_RAH(rar_entry));
+ reg_data |= (1 << (18 + vf));
+ wr32(E1000_RAH(rar_entry), reg_data);
+
+ return 0;
+}
+
+static int igb_vmm_control(struct igb_adapter *adapter, bool enable)
+{
+ struct e1000_hw *hw;
+ u32 reg_data;
+
+ hw = &adapter->hw;
+
+ if (enable) {
+ /* Enable multi-queue */
+ reg_data = rd32(E1000_MRQC);
+ reg_data &= E1000_MRQC_ENABLE_MASK;
+ reg_data |= E1000_MRQC_ENABLE_VMDQ;
+ wr32(E1000_MRQC, reg_data);
+ /* VF's need PF reset indication before they
+ * can send/receive mail */
+ reg_data = rd32(E1000_CTRL_EXT);
+ reg_data |= E1000_CTRL_EXT_PFRSTD;
+ wr32(E1000_CTRL_EXT, reg_data);
+
+ /* Set the default pool for the PF's first queue */
+ reg_data = rd32(E1000_VMD_CTL);
+ reg_data &= ~(E1000_VMD_CTL | E1000_VT_CTL_DISABLE_DEF_POOL);
+ reg_data |= adapter->vfs_allocated_count <<
+ E1000_VT_CTL_DEFAULT_POOL_SHIFT;
+ wr32(E1000_VMD_CTL, reg_data);
+
+ e1000_vmdq_loopback_enable_vf(hw);
+ e1000_vmdq_replication_enable_vf(hw, 0xFF);
+ } else {
+ e1000_vmdq_loopback_disable_vf(hw);
+ e1000_vmdq_replication_disable_vf(hw);
+ }
+
+ return 0;
+}
+
+static void igb_enable_pf_queues(struct igb_adapter *adapter)
+{
+ u64 rdba;
+ int i;
+ u32 rbase_offset = adapter->vfs_allocated_count;
+ struct e1000_hw *hw = &adapter->hw;
+ u32 rxdctl;
+
+ for (i = rbase_offset;
+ i < (adapter->num_rx_queues + rbase_offset); i++) {
+ struct igb_ring *ring = &adapter->rx_ring[i - rbase_offset];
+ rdba = ring->dma;
+
+ rxdctl = rd32(E1000_RXDCTL(i));
+ rxdctl |= E1000_RXDCTL_QUEUE_ENABLE;
+ rxdctl &= 0xFFF00000;
+ rxdctl |= IGB_RX_PTHRESH;
+ rxdctl |= IGB_RX_HTHRESH << 8;
+ rxdctl |= IGB_RX_WTHRESH << 16;
+ wr32(E1000_RXDCTL(i), rxdctl);
+ printk("RXDCTL%d == %8.8x\n", i, rxdctl);
+
+ wr32(E1000_RDBAL(i),
+ rdba & 0x00000000ffffffffULL);
+ wr32(E1000_RDBAH(i), rdba >> 32);
+ wr32(E1000_RDLEN(i),
+ ring->count * sizeof(union e1000_adv_rx_desc));
+
+ writel(ring->next_to_use, adapter->hw.hw_addr + ring->tail);
+ writel(ring->next_to_clean, adapter->hw.hw_addr + ring->head);
+ }
+}
+
+void igb_set_mc_list_pools(struct igb_adapter *adapter,
+ struct e1000_hw *hw,
+ int entry_count, u16 total_rar_filters)
+{
+ u32 reg_data;
+ int i;
+ int pool = adapter->vfs_allocated_count;
+
+ for (i = adapter->vfs_allocated_count + 1; i < total_rar_filters; i++) {
+ reg_data = rd32(E1000_RAH(i));
+ reg_data |= (1 << (18 + pool));
+ wr32(E1000_RAH(i), reg_data);
+ entry_count--;
+ if (!entry_count)
+ break;
+ }
+
+ reg_data = rd32(E1000_VMOLR(pool));
+ /* Set bit 25 for this pool in the VM Offload register so that
+ * it can accept packets that match the MTA table */
+ reg_data |= (1 << 25);
+ wr32(E1000_VMOLR(pool), reg_data);
+}
+
/* igb_main.c */
diff --git a/drivers/pci/iov.c b/drivers/pci/iov.c
index b4c1b5a..79b49e5 100644
--- a/drivers/pci/iov.c
+++ b/drivers/pci/iov.c
@@ -487,9 +487,11 @@ void pci_iov_unregister(struct pci_dev *dev)
sysfs_remove_group(&dev->dev.kobj, &iov_attr_group);
- mutex_lock(&pdev->iov->physfn->iov->lock);
+ mutex_lock(&dev->iov->physfn->iov->lock);
+
iov_disable(dev);
- mutex_unlock(&pdev->iov->physfn->iov->lock);
+
+ mutex_unlock(&dev->iov->physfn->iov->lock);
kobject_uevent(&dev->dev.kobj, KOBJ_CHANGE);
}
--
1.5.6.4
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists