[net-next 00/13] bna: Update the Driver to v3.2.23.0

[net-next 00/13] bna: Update the Driver to v3.2.23.0

[Rasesh Mody]

Hi Dave,

This patch set consists of feature additions like s/w ptp support, multi-buffer
RX, firmware patch simplification, enhancements for RX filters, RX processing,
bug fixes and updates the firmware version to v3.2.3.0.

This patch set updates the BNA driver to v3.2.23.0 and was tested against
net-next 3.12.0-rc6 kernel.

Thanks
Rasesh

Rasesh Mody (13):
  bna: Add software PTP support
  bna: Set Get IOC fw State
  bna: Fix Filter Add Del
  bna: RX Filter Enhancements
  bna: Enable Multi Buffer RX
  bna: UDP RX Processing Improvements
  bna: Fix Stale MTU used to Configure Multi-buffer RX
  bna: CQ Read Fix
  bna: Add NULL Check Before Dereferencing TCB
  bna: Handle the TX Setup Failures
  bna: Embed SKB Length in TX Vector
  bna: Firmware Patch Simplification
  bna: Update the Driver Version to 3.2.23.0

 drivers/net/ethernet/brocade/bna/bfa_ioc.c      | 650 ++++++++++++++++++++++--
 drivers/net/ethernet/brocade/bna/bfa_ioc.h      |   8 +
 drivers/net/ethernet/brocade/bna/bfa_ioc_ct.c   |  40 ++
 drivers/net/ethernet/brocade/bna/bfi.h          |  33 +-
 drivers/net/ethernet/brocade/bna/bfi_enet.h     |   3 +-
 drivers/net/ethernet/brocade/bna/bna.h          |  24 +-
 drivers/net/ethernet/brocade/bna/bna_enet.c     |  58 ++-
 drivers/net/ethernet/brocade/bna/bna_hw_defs.h  |   4 +
 drivers/net/ethernet/brocade/bna/bna_tx_rx.c    | 251 +++++++--
 drivers/net/ethernet/brocade/bna/bna_types.h    |  57 ++-
 drivers/net/ethernet/brocade/bna/bnad.c         | 543 ++++++++++++++------
 drivers/net/ethernet/brocade/bna/bnad.h         |  29 +-
 drivers/net/ethernet/brocade/bna/bnad_ethtool.c |   1 +
 drivers/net/ethernet/brocade/bna/cna.h          |   4 +-
 14 files changed, 1412 insertions(+), 293 deletions(-)

-- 
1.8.2.3

[net-next 01/13] bna: Add software PTP support

[Rasesh Mody]

 - Invoke skb_tx_timestamp() API just before invoking txq_doorbell()
 - Add ethtool (-T) support

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bnad.c         | 2 ++
 drivers/net/ethernet/brocade/bna/bnad_ethtool.c | 1 +
 2 files changed, 3 insertions(+)

diff --git a/drivers/net/ethernet/brocade/bna/bnad.c b/drivers/net/ethernet/brocade/bna/bnad.c
index f276433..9b1c550 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.c
+++ b/drivers/net/ethernet/brocade/bna/bnad.c
@@ -2911,6 +2911,8 @@ bnad_start_xmit(struct sk_buff *skb, struct net_device *netdev)
 	if (unlikely(!test_bit(BNAD_TXQ_TX_STARTED, &tcb->flags)))
 		return NETDEV_TX_OK;
 
+	skb_tx_timestamp(skb);
+
 	bna_txq_prod_indx_doorbell(tcb);
 	smp_mb();
 
diff --git a/drivers/net/ethernet/brocade/bna/bnad_ethtool.c b/drivers/net/ethernet/brocade/bna/bnad_ethtool.c
index 455b5a2..f9e1508 100644
--- a/drivers/net/ethernet/brocade/bna/bnad_ethtool.c
+++ b/drivers/net/ethernet/brocade/bna/bnad_ethtool.c
@@ -1131,6 +1131,7 @@ static const struct ethtool_ops bnad_ethtool_ops = {
 	.get_eeprom = bnad_get_eeprom,
 	.set_eeprom = bnad_set_eeprom,
 	.flash_device = bnad_flash_device,
+	.get_ts_info = ethtool_op_get_ts_info,
 };
 
 void
-- 
1.8.2.3

[net-next 02/13] bna: Set Get IOC fw State

[Rasesh Mody]

Add APIs to set and get IOC currnet fw state and alt IOC fw state
 - bfa_ioc_ct_set_cur_ioc_fwstate()
 - bfa_ioc_ct_get_cur_ioc_fwstate()
 - bfa_ioc_ct_set_alt_ioc_fwstate()
 - bfa_ioc_ct_get_alt_ioc_fwstate()

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bfa_ioc.c    | 32 +++++++++++++--------
 drivers/net/ethernet/brocade/bna/bfa_ioc.h    |  7 +++++
 drivers/net/ethernet/brocade/bna/bfa_ioc_ct.c | 40 +++++++++++++++++++++++++++
 3 files changed, 67 insertions(+), 12 deletions(-)

diff --git a/drivers/net/ethernet/brocade/bna/bfa_ioc.c b/drivers/net/ethernet/brocade/bna/bfa_ioc.c
index 6f3cac0..b81132a 100644
--- a/drivers/net/ethernet/brocade/bna/bfa_ioc.c
+++ b/drivers/net/ethernet/brocade/bna/bfa_ioc.c
@@ -42,6 +42,14 @@
 			((__ioc)->ioc_hwif->ioc_sync_ack(__ioc))
 #define bfa_ioc_sync_complete(__ioc)			\
 			((__ioc)->ioc_hwif->ioc_sync_complete(__ioc))
+#define bfa_ioc_set_cur_ioc_fwstate(__ioc, __fwstate)		\
+			((__ioc)->ioc_hwif->ioc_set_fwstate(__ioc, __fwstate))
+#define bfa_ioc_get_cur_ioc_fwstate(__ioc)		\
+			((__ioc)->ioc_hwif->ioc_get_fwstate(__ioc))
+#define bfa_ioc_set_alt_ioc_fwstate(__ioc, __fwstate)		\
+		((__ioc)->ioc_hwif->ioc_set_alt_fwstate(__ioc, __fwstate))
+#define bfa_ioc_get_alt_ioc_fwstate(__ioc)		\
+			((__ioc)->ioc_hwif->ioc_get_alt_fwstate(__ioc))
 
 #define bfa_ioc_mbox_cmd_pending(__ioc)		\
 			(!list_empty(&((__ioc)->mbox_mod.cmd_q)) || \
@@ -860,7 +868,7 @@ bfa_iocpf_sm_disabling(struct bfa_iocpf *iocpf, enum iocpf_event event)
 		 */
 
 	case IOCPF_E_TIMEOUT:
-		writel(BFI_IOC_FAIL, ioc->ioc_regs.ioc_fwstate);
+		bfa_ioc_set_cur_ioc_fwstate(ioc, BFI_IOC_FAIL);
 		bfa_fsm_set_state(iocpf, bfa_iocpf_sm_disabling_sync);
 		break;
 
@@ -949,7 +957,7 @@ bfa_iocpf_sm_initfail_sync(struct bfa_iocpf *iocpf, enum iocpf_event event)
 	case IOCPF_E_SEMLOCKED:
 		bfa_ioc_notify_fail(ioc);
 		bfa_ioc_sync_leave(ioc);
-		writel(BFI_IOC_FAIL, ioc->ioc_regs.ioc_fwstate);
+		bfa_ioc_set_cur_ioc_fwstate(ioc, BFI_IOC_FAIL);
 		bfa_nw_ioc_hw_sem_release(ioc);
 		bfa_fsm_set_state(iocpf, bfa_iocpf_sm_initfail);
 		break;
@@ -1031,7 +1039,7 @@ bfa_iocpf_sm_fail_sync(struct bfa_iocpf *iocpf, enum iocpf_event event)
 		bfa_ioc_notify_fail(ioc);
 		if (!iocpf->auto_recover) {
 			bfa_ioc_sync_leave(ioc);
-			writel(BFI_IOC_FAIL, ioc->ioc_regs.ioc_fwstate);
+			bfa_ioc_set_cur_ioc_fwstate(ioc, BFI_IOC_FAIL);
 			bfa_nw_ioc_hw_sem_release(ioc);
 			bfa_fsm_set_state(iocpf, bfa_iocpf_sm_fail);
 		} else {
@@ -1162,7 +1170,7 @@ bfa_ioc_hw_sem_init(struct bfa_ioc *ioc)
 		r32 = readl(ioc->ioc_regs.ioc_init_sem_reg);
 	}
 
-	fwstate = readl(ioc->ioc_regs.ioc_fwstate);
+	fwstate = bfa_ioc_get_cur_ioc_fwstate(ioc);
 	if (fwstate == BFI_IOC_UNINIT) {
 		writel(1, ioc->ioc_regs.ioc_init_sem_reg);
 		return;
@@ -1176,8 +1184,8 @@ bfa_ioc_hw_sem_init(struct bfa_ioc *ioc)
 	}
 
 	bfa_ioc_fwver_clear(ioc);
-	writel(BFI_IOC_UNINIT, ioc->ioc_regs.ioc_fwstate);
-	writel(BFI_IOC_UNINIT, ioc->ioc_regs.alt_ioc_fwstate);
+	bfa_ioc_set_cur_ioc_fwstate(ioc, BFI_IOC_UNINIT);
+	bfa_ioc_set_alt_ioc_fwstate(ioc, BFI_IOC_UNINIT);
 
 	/*
 	 * Try to lock and then unlock the semaphore.
@@ -1366,7 +1374,7 @@ bfa_ioc_hwinit(struct bfa_ioc *ioc, bool force)
 	bool fwvalid;
 	u32 boot_env;
 
-	ioc_fwstate = readl(ioc->ioc_regs.ioc_fwstate);
+	ioc_fwstate = bfa_ioc_get_cur_ioc_fwstate(ioc);
 
 	if (force)
 		ioc_fwstate = BFI_IOC_UNINIT;
@@ -1859,11 +1867,11 @@ bfa_ioc_boot(struct bfa_ioc *ioc, enum bfi_fwboot_type boot_type,
 	 * Initialize IOC state of all functions on a chip reset.
 	 */
 	if (boot_type == BFI_FWBOOT_TYPE_MEMTEST) {
-		writel(BFI_IOC_MEMTEST, ioc->ioc_regs.ioc_fwstate);
-		writel(BFI_IOC_MEMTEST, ioc->ioc_regs.alt_ioc_fwstate);
+		bfa_ioc_set_cur_ioc_fwstate(ioc, BFI_IOC_MEMTEST);
+		bfa_ioc_set_alt_ioc_fwstate(ioc, BFI_IOC_MEMTEST);
 	} else {
-		writel(BFI_IOC_INITING, ioc->ioc_regs.ioc_fwstate);
-		writel(BFI_IOC_INITING, ioc->ioc_regs.alt_ioc_fwstate);
+		bfa_ioc_set_cur_ioc_fwstate(ioc, BFI_IOC_INITING);
+		bfa_ioc_set_alt_ioc_fwstate(ioc, BFI_IOC_INITING);
 	}
 
 	bfa_ioc_msgflush(ioc);
@@ -2473,7 +2481,7 @@ bfa_nw_iocpf_sem_timeout(void *ioc_arg)
 static void
 bfa_ioc_poll_fwinit(struct bfa_ioc *ioc)
 {
-	u32 fwstate = readl(ioc->ioc_regs.ioc_fwstate);
+	u32 fwstate = bfa_ioc_get_cur_ioc_fwstate(ioc);
 
 	if (fwstate == BFI_IOC_DISABLED) {
 		bfa_fsm_send_event(&ioc->iocpf, IOCPF_E_FWREADY);
diff --git a/drivers/net/ethernet/brocade/bna/bfa_ioc.h b/drivers/net/ethernet/brocade/bna/bfa_ioc.h
index f04e0aa..9521cbf 100644
--- a/drivers/net/ethernet/brocade/bna/bfa_ioc.h
+++ b/drivers/net/ethernet/brocade/bna/bfa_ioc.h
@@ -215,6 +215,13 @@ struct bfa_ioc_hwif {
 	void		(*ioc_sync_ack)		(struct bfa_ioc *ioc);
 	bool		(*ioc_sync_complete)	(struct bfa_ioc *ioc);
 	bool		(*ioc_lpu_read_stat)	(struct bfa_ioc *ioc);
+	void		(*ioc_set_fwstate)	(struct bfa_ioc *ioc,
+					enum bfi_ioc_state fwstate);
+	enum bfi_ioc_state (*ioc_get_fwstate) (struct bfa_ioc *ioc);
+	void		(*ioc_set_alt_fwstate)	(struct bfa_ioc *ioc,
+					enum bfi_ioc_state fwstate);
+	enum bfi_ioc_state (*ioc_get_alt_fwstate) (struct bfa_ioc *ioc);
+
 };
 
 #define bfa_ioc_pcifn(__ioc)		((__ioc)->pcidev.pci_func)
diff --git a/drivers/net/ethernet/brocade/bna/bfa_ioc_ct.c b/drivers/net/ethernet/brocade/bna/bfa_ioc_ct.c
index 5df0b0c..d639558 100644
--- a/drivers/net/ethernet/brocade/bna/bfa_ioc_ct.c
+++ b/drivers/net/ethernet/brocade/bna/bfa_ioc_ct.c
@@ -48,6 +48,12 @@ static void bfa_ioc_ct_sync_join(struct bfa_ioc *ioc);
 static void bfa_ioc_ct_sync_leave(struct bfa_ioc *ioc);
 static void bfa_ioc_ct_sync_ack(struct bfa_ioc *ioc);
 static bool bfa_ioc_ct_sync_complete(struct bfa_ioc *ioc);
+static void bfa_ioc_ct_set_cur_ioc_fwstate(
+			struct bfa_ioc *ioc, enum bfi_ioc_state fwstate);
+static enum bfi_ioc_state bfa_ioc_ct_get_cur_ioc_fwstate(struct bfa_ioc *ioc);
+static void bfa_ioc_ct_set_alt_ioc_fwstate(
+			struct bfa_ioc *ioc, enum bfi_ioc_state fwstate);
+static enum bfi_ioc_state bfa_ioc_ct_get_alt_ioc_fwstate(struct bfa_ioc *ioc);
 static enum bfa_status bfa_ioc_ct_pll_init(void __iomem *rb,
 				enum bfi_asic_mode asic_mode);
 static enum bfa_status bfa_ioc_ct2_pll_init(void __iomem *rb,
@@ -68,6 +74,10 @@ static const struct bfa_ioc_hwif nw_hwif_ct = {
 	.ioc_sync_leave	     = bfa_ioc_ct_sync_leave,
 	.ioc_sync_ack	     = bfa_ioc_ct_sync_ack,
 	.ioc_sync_complete   = bfa_ioc_ct_sync_complete,
+	.ioc_set_fwstate     = bfa_ioc_ct_set_cur_ioc_fwstate,
+	.ioc_get_fwstate     = bfa_ioc_ct_get_cur_ioc_fwstate,
+	.ioc_set_alt_fwstate     = bfa_ioc_ct_set_alt_ioc_fwstate,
+	.ioc_get_alt_fwstate     = bfa_ioc_ct_get_alt_ioc_fwstate,
 };
 
 static const struct bfa_ioc_hwif nw_hwif_ct2 = {
@@ -85,6 +95,10 @@ static const struct bfa_ioc_hwif nw_hwif_ct2 = {
 	.ioc_sync_leave	     = bfa_ioc_ct_sync_leave,
 	.ioc_sync_ack	     = bfa_ioc_ct_sync_ack,
 	.ioc_sync_complete   = bfa_ioc_ct_sync_complete,
+	.ioc_set_fwstate     = bfa_ioc_ct_set_cur_ioc_fwstate,
+	.ioc_get_fwstate     = bfa_ioc_ct_get_cur_ioc_fwstate,
+	.ioc_set_alt_fwstate     = bfa_ioc_ct_set_alt_ioc_fwstate,
+	.ioc_get_alt_fwstate     = bfa_ioc_ct_get_alt_ioc_fwstate,
 };
 
 /* Called from bfa_ioc_attach() to map asic specific calls. */
@@ -565,6 +579,32 @@ bfa_ioc_ct_sync_complete(struct bfa_ioc *ioc)
 	return false;
 }
 
+static void
+bfa_ioc_ct_set_cur_ioc_fwstate(struct bfa_ioc *ioc,
+			       enum bfi_ioc_state fwstate)
+{
+	writel(fwstate, ioc->ioc_regs.ioc_fwstate);
+}
+
+static enum bfi_ioc_state
+bfa_ioc_ct_get_cur_ioc_fwstate(struct bfa_ioc *ioc)
+{
+	return (enum bfi_ioc_state)readl(ioc->ioc_regs.ioc_fwstate);
+}
+
+static void
+bfa_ioc_ct_set_alt_ioc_fwstate(struct bfa_ioc *ioc,
+			       enum bfi_ioc_state fwstate)
+{
+	writel(fwstate, ioc->ioc_regs.alt_ioc_fwstate);
+}
+
+static enum bfi_ioc_state
+bfa_ioc_ct_get_alt_ioc_fwstate(struct bfa_ioc *ioc)
+{
+	return (enum bfi_ioc_state)readl(ioc->ioc_regs.alt_ioc_fwstate);
+}
+
 static enum bfa_status
 bfa_ioc_ct_pll_init(void __iomem *rb, enum bfi_asic_mode asic_mode)
 {
-- 
1.8.2.3

[net-next 03/13] bna: Fix Filter Add Del

[Rasesh Mody]

Change Details:
 - bna_rx_mcast_listset() API first looks at free_q only and not at other
   pending Qs rendering it non-deterministic of giving an upper limit.
   Modify bna_rx_mcast_listset() implementation to not use only half of the
   limit.
 - Allocate and initialize queue for deleting
 - Segregate the adding and deleting process by using separate queues.
 - The filter framework in bna does not let adding addresses to its max capacity
   due to asynchronous operations involved.
   Provide a synchronous option to set a given list.

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bna.h       | 16 ++++---
 drivers/net/ethernet/brocade/bna/bna_enet.c  | 48 +++++++++-----------
 drivers/net/ethernet/brocade/bna/bna_tx_rx.c | 67 ++++++++++++++++------------
 drivers/net/ethernet/brocade/bna/bna_types.h |  8 ++--
 4 files changed, 74 insertions(+), 65 deletions(-)

diff --git a/drivers/net/ethernet/brocade/bna/bna.h b/drivers/net/ethernet/brocade/bna/bna.h
index f1eafc4..b11fa6c 100644
--- a/drivers/net/ethernet/brocade/bna/bna.h
+++ b/drivers/net/ethernet/brocade/bna/bna.h
@@ -354,6 +354,14 @@ do {									\
 	}								\
 } while (0)
 
+#define bna_mcam_mod_free_q(_bna) (&(_bna)->mcam_mod.free_q)
+
+#define bna_mcam_mod_del_q(_bna) (&(_bna)->mcam_mod.del_q)
+
+#define bna_ucam_mod_free_q(_bna) (&(_bna)->ucam_mod.free_q)
+
+#define bna_ucam_mod_del_q(_bna) (&(_bna)->ucam_mod.del_q)
+
 /*  Inline functions  */
 
 static inline struct bna_mac *bna_mac_find(struct list_head *q, u8 *addr)
@@ -391,12 +399,8 @@ int bna_num_rxp_set(struct bna *bna, int num_rxp);
 void bna_hw_stats_get(struct bna *bna);
 
 /* APIs for RxF */
-struct bna_mac *bna_ucam_mod_mac_get(struct bna_ucam_mod *ucam_mod);
-void bna_ucam_mod_mac_put(struct bna_ucam_mod *ucam_mod,
-			  struct bna_mac *mac);
-struct bna_mac *bna_mcam_mod_mac_get(struct bna_mcam_mod *mcam_mod);
-void bna_mcam_mod_mac_put(struct bna_mcam_mod *mcam_mod,
-			  struct bna_mac *mac);
+struct bna_mac *bna_cam_mod_mac_get(struct list_head *head);
+void bna_cam_mod_mac_put(struct list_head *tail, struct bna_mac *mac);
 struct bna_mcam_handle *bna_mcam_mod_handle_get(struct bna_mcam_mod *mod);
 void bna_mcam_mod_handle_put(struct bna_mcam_mod *mcam_mod,
 			  struct bna_mcam_handle *handle);
diff --git a/drivers/net/ethernet/brocade/bna/bna_enet.c b/drivers/net/ethernet/brocade/bna/bna_enet.c
index 3ca77fa..79afa11 100644
--- a/drivers/net/ethernet/brocade/bna/bna_enet.c
+++ b/drivers/net/ethernet/brocade/bna/bna_enet.c
@@ -1811,6 +1811,13 @@ bna_ucam_mod_init(struct bna_ucam_mod *ucam_mod, struct bna *bna,
 		list_add_tail(&ucam_mod->ucmac[i].qe, &ucam_mod->free_q);
 	}
 
+	/* A separate queue to allow synchronous setting of a list of MACs */
+	INIT_LIST_HEAD(&ucam_mod->del_q);
+	for (i = i; i < (bna->ioceth.attr.num_ucmac * 2); i++) {
+		bfa_q_qe_init(&ucam_mod->ucmac[i].qe);
+		list_add_tail(&ucam_mod->ucmac[i].qe, &ucam_mod->del_q);
+	}
+
 	ucam_mod->bna = bna;
 }
 
@@ -1851,6 +1858,13 @@ bna_mcam_mod_init(struct bna_mcam_mod *mcam_mod, struct bna *bna,
 				&mcam_mod->free_handle_q);
 	}
 
+	/* A separate queue to allow synchronous setting of a list of MACs */
+	INIT_LIST_HEAD(&mcam_mod->del_q);
+	for (i = i; i < (bna->ioceth.attr.num_mcmac * 2); i++) {
+		bfa_q_qe_init(&mcam_mod->mcmac[i].qe);
+		list_add_tail(&mcam_mod->mcmac[i].qe, &mcam_mod->del_q);
+	}
+
 	mcam_mod->bna = bna;
 }
 
@@ -1976,7 +1990,7 @@ bna_mod_res_req(struct bna *bna, struct bna_res_info *res_info)
 		BNA_MEM_T_KVA;
 	res_info[BNA_MOD_RES_MEM_T_UCMAC_ARRAY].res_u.mem_info.num = 1;
 	res_info[BNA_MOD_RES_MEM_T_UCMAC_ARRAY].res_u.mem_info.len =
-		attr->num_ucmac * sizeof(struct bna_mac);
+		(attr->num_ucmac * 2) * sizeof(struct bna_mac);
 
 	/* Virtual memory for Multicast MAC address - stored by mcam module */
 	res_info[BNA_MOD_RES_MEM_T_MCMAC_ARRAY].res_type = BNA_RES_T_MEM;
@@ -1984,7 +1998,7 @@ bna_mod_res_req(struct bna *bna, struct bna_res_info *res_info)
 		BNA_MEM_T_KVA;
 	res_info[BNA_MOD_RES_MEM_T_MCMAC_ARRAY].res_u.mem_info.num = 1;
 	res_info[BNA_MOD_RES_MEM_T_MCMAC_ARRAY].res_u.mem_info.len =
-		attr->num_mcmac * sizeof(struct bna_mac);
+		(attr->num_mcmac * 2) * sizeof(struct bna_mac);
 
 	/* Virtual memory for Multicast handle - stored by mcam module */
 	res_info[BNA_MOD_RES_MEM_T_MCHANDLE_ARRAY].res_type = BNA_RES_T_MEM;
@@ -2080,41 +2094,21 @@ bna_num_rxp_set(struct bna *bna, int num_rxp)
 }
 
 struct bna_mac *
-bna_ucam_mod_mac_get(struct bna_ucam_mod *ucam_mod)
+bna_cam_mod_mac_get(struct list_head *head)
 {
 	struct list_head *qe;
 
-	if (list_empty(&ucam_mod->free_q))
+	if (list_empty(head))
 		return NULL;
 
-	bfa_q_deq(&ucam_mod->free_q, &qe);
-
-	return (struct bna_mac *)qe;
-}
-
-void
-bna_ucam_mod_mac_put(struct bna_ucam_mod *ucam_mod, struct bna_mac *mac)
-{
-	list_add_tail(&mac->qe, &ucam_mod->free_q);
-}
-
-struct bna_mac *
-bna_mcam_mod_mac_get(struct bna_mcam_mod *mcam_mod)
-{
-	struct list_head *qe;
-
-	if (list_empty(&mcam_mod->free_q))
-		return NULL;
-
-	bfa_q_deq(&mcam_mod->free_q, &qe);
-
+	bfa_q_deq(head, &qe);
 	return (struct bna_mac *)qe;
 }
 
 void
-bna_mcam_mod_mac_put(struct bna_mcam_mod *mcam_mod, struct bna_mac *mac)
+bna_cam_mod_mac_put(struct list_head *tail, struct bna_mac *mac)
 {
-	list_add_tail(&mac->qe, &mcam_mod->free_q);
+	list_add_tail(&mac->qe, tail);
 }
 
 struct bna_mcam_handle *
diff --git a/drivers/net/ethernet/brocade/bna/bna_tx_rx.c b/drivers/net/ethernet/brocade/bna/bna_tx_rx.c
index 3c07064..40558b6 100644
--- a/drivers/net/ethernet/brocade/bna/bna_tx_rx.c
+++ b/drivers/net/ethernet/brocade/bna/bna_tx_rx.c
@@ -529,13 +529,13 @@ bna_rxf_mcast_cfg_apply(struct bna_rxf *rxf)
 	struct list_head *qe;
 	int ret;
 
-	/* Delete multicast entries previousely added */
+	/* First delete multicast entries to maintain the count */
 	while (!list_empty(&rxf->mcast_pending_del_q)) {
 		bfa_q_deq(&rxf->mcast_pending_del_q, &qe);
 		bfa_q_qe_init(qe);
 		mac = (struct bna_mac *)qe;
 		ret = bna_rxf_mcast_del(rxf, mac, BNA_HARD_CLEANUP);
-		bna_mcam_mod_mac_put(&rxf->rx->bna->mcam_mod, mac);
+		bna_cam_mod_mac_put(bna_mcam_mod_del_q(rxf->rx->bna), mac);
 		if (ret)
 			return ret;
 	}
@@ -586,7 +586,7 @@ bna_rxf_mcast_cfg_reset(struct bna_rxf *rxf, enum bna_cleanup_type cleanup)
 		bfa_q_qe_init(qe);
 		mac = (struct bna_mac *)qe;
 		ret = bna_rxf_mcast_del(rxf, mac, cleanup);
-		bna_mcam_mod_mac_put(&rxf->rx->bna->mcam_mod, mac);
+		bna_cam_mod_mac_put(bna_mcam_mod_del_q(rxf->rx->bna), mac);
 		if (ret)
 			return ret;
 	}
@@ -796,20 +796,20 @@ bna_rxf_uninit(struct bna_rxf *rxf)
 	while (!list_empty(&rxf->ucast_pending_add_q)) {
 		bfa_q_deq(&rxf->ucast_pending_add_q, &mac);
 		bfa_q_qe_init(&mac->qe);
-		bna_ucam_mod_mac_put(&rxf->rx->bna->ucam_mod, mac);
+		bna_cam_mod_mac_put(bna_ucam_mod_free_q(rxf->rx->bna), mac);
 	}
 
 	if (rxf->ucast_pending_mac) {
 		bfa_q_qe_init(&rxf->ucast_pending_mac->qe);
-		bna_ucam_mod_mac_put(&rxf->rx->bna->ucam_mod,
-			rxf->ucast_pending_mac);
+		bna_cam_mod_mac_put(bna_ucam_mod_free_q(rxf->rx->bna),
+				    rxf->ucast_pending_mac);
 		rxf->ucast_pending_mac = NULL;
 	}
 
 	while (!list_empty(&rxf->mcast_pending_add_q)) {
 		bfa_q_deq(&rxf->mcast_pending_add_q, &mac);
 		bfa_q_qe_init(&mac->qe);
-		bna_mcam_mod_mac_put(&rxf->rx->bna->mcam_mod, mac);
+		bna_cam_mod_mac_put(bna_mcam_mod_free_q(rxf->rx->bna), mac);
 	}
 
 	rxf->rxmode_pending = 0;
@@ -869,7 +869,7 @@ bna_rx_ucast_set(struct bna_rx *rx, u8 *ucmac,
 
 	if (rxf->ucast_pending_mac == NULL) {
 		rxf->ucast_pending_mac =
-				bna_ucam_mod_mac_get(&rxf->rx->bna->ucam_mod);
+			bna_cam_mod_mac_get(bna_ucam_mod_free_q(rxf->rx->bna));
 		if (rxf->ucast_pending_mac == NULL)
 			return BNA_CB_UCAST_CAM_FULL;
 		bfa_q_qe_init(&rxf->ucast_pending_mac->qe);
@@ -900,7 +900,7 @@ bna_rx_mcast_add(struct bna_rx *rx, u8 *addr,
 		return BNA_CB_SUCCESS;
 	}
 
-	mac = bna_mcam_mod_mac_get(&rxf->rx->bna->mcam_mod);
+	mac = bna_cam_mod_mac_get(bna_mcam_mod_free_q(rxf->rx->bna));
 	if (mac == NULL)
 		return BNA_CB_MCAST_LIST_FULL;
 	bfa_q_qe_init(&mac->qe);
@@ -919,32 +919,20 @@ enum bna_cb_status
 bna_rx_mcast_listset(struct bna_rx *rx, int count, u8 *mclist,
 		     void (*cbfn)(struct bnad *, struct bna_rx *))
 {
+	struct bna_mcam_mod *mcam_mod = &rx->bna->mcam_mod;
 	struct bna_rxf *rxf = &rx->rxf;
 	struct list_head list_head;
 	struct list_head *qe;
 	u8 *mcaddr;
-	struct bna_mac *mac;
+	struct bna_mac *mac, *del_mac;
 	int i;
 
-	/* Allocate nodes */
-	INIT_LIST_HEAD(&list_head);
-	for (i = 0, mcaddr = mclist; i < count; i++) {
-		mac = bna_mcam_mod_mac_get(&rxf->rx->bna->mcam_mod);
-		if (mac == NULL)
-			goto err_return;
-		bfa_q_qe_init(&mac->qe);
-		memcpy(mac->addr, mcaddr, ETH_ALEN);
-		list_add_tail(&mac->qe, &list_head);
-
-		mcaddr += ETH_ALEN;
-	}
-
 	/* Purge the pending_add_q */
 	while (!list_empty(&rxf->mcast_pending_add_q)) {
 		bfa_q_deq(&rxf->mcast_pending_add_q, &qe);
 		bfa_q_qe_init(qe);
 		mac = (struct bna_mac *)qe;
-		bna_mcam_mod_mac_put(&rxf->rx->bna->mcam_mod, mac);
+		bna_cam_mod_mac_put(&mcam_mod->free_q, mac);
 	}
 
 	/* Schedule active_q entries for deletion */
@@ -952,7 +940,26 @@ bna_rx_mcast_listset(struct bna_rx *rx, int count, u8 *mclist,
 		bfa_q_deq(&rxf->mcast_active_q, &qe);
 		mac = (struct bna_mac *)qe;
 		bfa_q_qe_init(&mac->qe);
-		list_add_tail(&mac->qe, &rxf->mcast_pending_del_q);
+
+		del_mac = bna_cam_mod_mac_get(&mcam_mod->del_q);
+
+		memcpy(del_mac, mac, sizeof(*mac));
+		list_add_tail(&del_mac->qe, &rxf->mcast_pending_del_q);
+		mac->handle = NULL;
+		bna_cam_mod_mac_put(&mcam_mod->free_q, mac);
+	}
+
+	/* Allocate nodes */
+	INIT_LIST_HEAD(&list_head);
+	for (i = 0, mcaddr = mclist; i < count; i++) {
+		mac = bna_cam_mod_mac_get(&mcam_mod->free_q);
+		if (mac == NULL)
+			goto err_return;
+		bfa_q_qe_init(&mac->qe);
+		memcpy(mac->addr, mcaddr, ETH_ALEN);
+		list_add_tail(&mac->qe, &list_head);
+
+		mcaddr += ETH_ALEN;
 	}
 
 	/* Add the new entries */
@@ -974,7 +981,7 @@ err_return:
 		bfa_q_deq(&list_head, &qe);
 		mac = (struct bna_mac *)qe;
 		bfa_q_qe_init(&mac->qe);
-		bna_mcam_mod_mac_put(&rxf->rx->bna->mcam_mod, mac);
+		bna_cam_mod_mac_put(&mcam_mod->free_q, mac);
 	}
 
 	return BNA_CB_MCAST_LIST_FULL;
@@ -1022,7 +1029,7 @@ bna_rxf_ucast_cfg_apply(struct bna_rxf *rxf)
 		bfa_q_qe_init(qe);
 		mac = (struct bna_mac *)qe;
 		bna_bfi_ucast_req(rxf, mac, BFI_ENET_H2I_MAC_UCAST_DEL_REQ);
-		bna_ucam_mod_mac_put(&rxf->rx->bna->ucam_mod, mac);
+		bna_cam_mod_mac_put(bna_ucam_mod_del_q(rxf->rx->bna), mac);
 		return 1;
 	}
 
@@ -1062,11 +1069,13 @@ bna_rxf_ucast_cfg_reset(struct bna_rxf *rxf, enum bna_cleanup_type cleanup)
 		bfa_q_qe_init(qe);
 		mac = (struct bna_mac *)qe;
 		if (cleanup == BNA_SOFT_CLEANUP)
-			bna_ucam_mod_mac_put(&rxf->rx->bna->ucam_mod, mac);
+			bna_cam_mod_mac_put(bna_ucam_mod_del_q(rxf->rx->bna),
+					    mac);
 		else {
 			bna_bfi_ucast_req(rxf, mac,
 				BFI_ENET_H2I_MAC_UCAST_DEL_REQ);
-			bna_ucam_mod_mac_put(&rxf->rx->bna->ucam_mod, mac);
+			bna_cam_mod_mac_put(bna_ucam_mod_del_q(rxf->rx->bna),
+					    mac);
 			return 1;
 		}
 	}
diff --git a/drivers/net/ethernet/brocade/bna/bna_types.h b/drivers/net/ethernet/brocade/bna/bna_types.h
index dc50f78..acedac2 100644
--- a/drivers/net/ethernet/brocade/bna/bna_types.h
+++ b/drivers/net/ethernet/brocade/bna/bna_types.h
@@ -866,8 +866,9 @@ struct bna_rx_mod {
 /* CAM */
 
 struct bna_ucam_mod {
-	struct bna_mac *ucmac;		/* BFI_MAX_UCMAC entries */
+	struct bna_mac *ucmac;		/* num_ucmac * 2 entries */
 	struct list_head			free_q;
+	struct list_head			del_q;
 
 	struct bna *bna;
 };
@@ -880,9 +881,10 @@ struct bna_mcam_handle {
 };
 
 struct bna_mcam_mod {
-	struct bna_mac *mcmac;		/* BFI_MAX_MCMAC entries */
-	struct bna_mcam_handle *mchandle;	/* BFI_MAX_MCMAC entries */
+	struct bna_mac *mcmac;		/* num_mcmac * 2 entries */
+	struct bna_mcam_handle *mchandle;	/* num_mcmac entries */
 	struct list_head			free_q;
+	struct list_head			del_q;
 	struct list_head			free_handle_q;
 
 	struct bna *bna;
-- 
1.8.2.3

[net-next 04/13] bna: RX Filter Enhancements

[Rasesh Mody]

Change Details:
 - Added bna_rx_ucast_listset() for synchronous ucast listadd operation.
 - Clear mac->handle before adding it to free_q.
 - bnad_set_rx_mode() rewritten. bnad_set_rx_mode() adds the MACs in uc_list
   to UCAM. If it exceeds the max supported, DEFAULT mode is turned on. If
   MCAM limit is exceeded, ALLMULTI mode is turned on.
 - Clear CF flags, check for the new mode and reprogram the Rx approach.
 - Added bnad_set_rx_ucast_fltr() and bnad_set_rx_mcast_fltr().
 - Check for IFF_PROMISC to set the correct mode.

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bna.h       |   8 ++
 drivers/net/ethernet/brocade/bna/bna_enet.c  |  10 +-
 drivers/net/ethernet/brocade/bna/bna_tx_rx.c | 138 ++++++++++++++++++++++-
 drivers/net/ethernet/brocade/bna/bnad.c      | 163 ++++++++++++++++++---------
 drivers/net/ethernet/brocade/bna/bnad.h      |   3 +-
 5 files changed, 266 insertions(+), 56 deletions(-)

diff --git a/drivers/net/ethernet/brocade/bna/bna.h b/drivers/net/ethernet/brocade/bna/bna.h
index b11fa6c..1f51219 100644
--- a/drivers/net/ethernet/brocade/bna/bna.h
+++ b/drivers/net/ethernet/brocade/bna/bna.h
@@ -497,11 +497,17 @@ enum bna_cb_status
 bna_rx_ucast_del(struct bna_rx *rx, u8 *ucmac,
 		 void (*cbfn)(struct bnad *, struct bna_rx *));
 enum bna_cb_status
+bna_rx_ucast_listset(struct bna_rx *rx, int count, u8 *uclist,
+		     void (*cbfn)(struct bnad *, struct bna_rx *));
+enum bna_cb_status
 bna_rx_mcast_add(struct bna_rx *rx, u8 *mcmac,
 		 void (*cbfn)(struct bnad *, struct bna_rx *));
 enum bna_cb_status
 bna_rx_mcast_listset(struct bna_rx *rx, int count, u8 *mcmac,
 		     void (*cbfn)(struct bnad *, struct bna_rx *));
+void
+bna_rx_mcast_delall(struct bna_rx *rx,
+		    void (*cbfn)(struct bnad *, struct bna_rx *));
 enum bna_cb_status
 bna_rx_mode_set(struct bna_rx *rx, enum bna_rxmode rxmode,
 		enum bna_rxmode bitmask,
@@ -509,6 +515,8 @@ bna_rx_mode_set(struct bna_rx *rx, enum bna_rxmode rxmode,
 void bna_rx_vlan_add(struct bna_rx *rx, int vlan_id);
 void bna_rx_vlan_del(struct bna_rx *rx, int vlan_id);
 void bna_rx_vlanfilter_enable(struct bna_rx *rx);
+void bna_rx_vlan_strip_enable(struct bna_rx *rx);
+void bna_rx_vlan_strip_disable(struct bna_rx *rx);
 /* ENET */
 
 /* API for RX */
diff --git a/drivers/net/ethernet/brocade/bna/bna_enet.c b/drivers/net/ethernet/brocade/bna/bna_enet.c
index 79afa11..13f9636 100644
--- a/drivers/net/ethernet/brocade/bna/bna_enet.c
+++ b/drivers/net/ethernet/brocade/bna/bna_enet.c
@@ -1825,11 +1825,16 @@ static void
 bna_ucam_mod_uninit(struct bna_ucam_mod *ucam_mod)
 {
 	struct list_head *qe;
-	int i = 0;
+	int i;
 
+	i = 0;
 	list_for_each(qe, &ucam_mod->free_q)
 		i++;
 
+	i = 0;
+	list_for_each(qe, &ucam_mod->del_q)
+		i++;
+
 	ucam_mod->bna = NULL;
 }
 
@@ -1878,6 +1883,9 @@ bna_mcam_mod_uninit(struct bna_mcam_mod *mcam_mod)
 	list_for_each(qe, &mcam_mod->free_q) i++;
 
 	i = 0;
+	list_for_each(qe, &mcam_mod->del_q) i++;
+
+	i = 0;
 	list_for_each(qe, &mcam_mod->free_handle_q) i++;
 
 	mcam_mod->bna = NULL;
diff --git a/drivers/net/ethernet/brocade/bna/bna_tx_rx.c b/drivers/net/ethernet/brocade/bna/bna_tx_rx.c
index 40558b6..e02c126 100644
--- a/drivers/net/ethernet/brocade/bna/bna_tx_rx.c
+++ b/drivers/net/ethernet/brocade/bna/bna_tx_rx.c
@@ -916,6 +916,75 @@ bna_rx_mcast_add(struct bna_rx *rx, u8 *addr,
 }
 
 enum bna_cb_status
+bna_rx_ucast_listset(struct bna_rx *rx, int count, u8 *uclist,
+		     void (*cbfn)(struct bnad *, struct bna_rx *))
+{
+	struct bna_ucam_mod *ucam_mod = &rx->bna->ucam_mod;
+	struct bna_rxf *rxf = &rx->rxf;
+	struct list_head list_head;
+	struct list_head *qe;
+	u8 *mcaddr;
+	struct bna_mac *mac, *del_mac;
+	int i;
+
+	/* Purge the pending_add_q */
+	while (!list_empty(&rxf->ucast_pending_add_q)) {
+		bfa_q_deq(&rxf->ucast_pending_add_q, &qe);
+		bfa_q_qe_init(qe);
+		mac = (struct bna_mac *)qe;
+		bna_cam_mod_mac_put(&ucam_mod->free_q, mac);
+	}
+
+	/* Schedule active_q entries for deletion */
+	while (!list_empty(&rxf->ucast_active_q)) {
+		bfa_q_deq(&rxf->ucast_active_q, &qe);
+		mac = (struct bna_mac *)qe;
+		bfa_q_qe_init(&mac->qe);
+
+		del_mac = bna_cam_mod_mac_get(&ucam_mod->del_q);
+		memcpy(del_mac, mac, sizeof(*del_mac));
+		list_add_tail(&del_mac->qe, &rxf->ucast_pending_del_q);
+		bna_cam_mod_mac_put(&ucam_mod->free_q, mac);
+	}
+
+	/* Allocate nodes */
+	INIT_LIST_HEAD(&list_head);
+	for (i = 0, mcaddr = uclist; i < count; i++) {
+		mac = bna_cam_mod_mac_get(&ucam_mod->free_q);
+		if (mac == NULL)
+			goto err_return;
+		bfa_q_qe_init(&mac->qe);
+		memcpy(mac->addr, mcaddr, ETH_ALEN);
+		list_add_tail(&mac->qe, &list_head);
+		mcaddr += ETH_ALEN;
+	}
+
+	/* Add the new entries */
+	while (!list_empty(&list_head)) {
+		bfa_q_deq(&list_head, &qe);
+		mac = (struct bna_mac *)qe;
+		bfa_q_qe_init(&mac->qe);
+		list_add_tail(&mac->qe, &rxf->ucast_pending_add_q);
+	}
+
+	rxf->cam_fltr_cbfn = cbfn;
+	rxf->cam_fltr_cbarg = rx->bna->bnad;
+	bfa_fsm_send_event(rxf, RXF_E_CONFIG);
+
+	return BNA_CB_SUCCESS;
+
+err_return:
+	while (!list_empty(&list_head)) {
+		bfa_q_deq(&list_head, &qe);
+		mac = (struct bna_mac *)qe;
+		bfa_q_qe_init(&mac->qe);
+		bna_cam_mod_mac_put(&ucam_mod->free_q, mac);
+	}
+
+	return BNA_CB_UCAST_CAM_FULL;
+}
+
+enum bna_cb_status
 bna_rx_mcast_listset(struct bna_rx *rx, int count, u8 *mclist,
 		     void (*cbfn)(struct bnad *, struct bna_rx *))
 {
@@ -943,7 +1012,7 @@ bna_rx_mcast_listset(struct bna_rx *rx, int count, u8 *mclist,
 
 		del_mac = bna_cam_mod_mac_get(&mcam_mod->del_q);
 
-		memcpy(del_mac, mac, sizeof(*mac));
+		memcpy(del_mac, mac, sizeof(*del_mac));
 		list_add_tail(&del_mac->qe, &rxf->mcast_pending_del_q);
 		mac->handle = NULL;
 		bna_cam_mod_mac_put(&mcam_mod->free_q, mac);
@@ -988,6 +1057,49 @@ err_return:
 }
 
 void
+bna_rx_mcast_delall(struct bna_rx *rx,
+		    void (*cbfn)(struct bnad *, struct bna_rx *))
+{
+	struct bna_rxf *rxf = &rx->rxf;
+	struct list_head *qe;
+	struct bna_mac *mac, *del_mac;
+	int need_hw_config = 0;
+
+	/* Purge all entries from pending_add_q */
+	while (!list_empty(&rxf->mcast_pending_add_q)) {
+		bfa_q_deq(&rxf->mcast_pending_add_q, &qe);
+		mac = (struct bna_mac *)qe;
+		bfa_q_qe_init(&mac->qe);
+		bna_cam_mod_mac_put(bna_mcam_mod_free_q(rxf->rx->bna), mac);
+	}
+
+	/* Schedule all entries in active_q for deletion */
+	while (!list_empty(&rxf->mcast_active_q)) {
+		bfa_q_deq(&rxf->mcast_active_q, &qe);
+		mac = (struct bna_mac *)qe;
+		bfa_q_qe_init(&mac->qe);
+
+		del_mac = bna_cam_mod_mac_get(bna_mcam_mod_del_q(rxf->rx->bna));
+
+		memcpy(del_mac, mac, sizeof(*del_mac));
+		list_add_tail(&del_mac->qe, &rxf->mcast_pending_del_q);
+		mac->handle = NULL;
+		bna_cam_mod_mac_put(bna_mcam_mod_free_q(rxf->rx->bna), mac);
+		need_hw_config = 1;
+	}
+
+	if (need_hw_config) {
+		rxf->cam_fltr_cbfn = cbfn;
+		rxf->cam_fltr_cbarg = rx->bna->bnad;
+		bfa_fsm_send_event(rxf, RXF_E_CONFIG);
+		return;
+	}
+
+	if (cbfn)
+		(*cbfn)(rx->bna->bnad, rx);
+}
+
+void
 bna_rx_vlan_add(struct bna_rx *rx, int vlan_id)
 {
 	struct bna_rxf *rxf = &rx->rxf;
@@ -2679,6 +2791,30 @@ bna_rx_cleanup_complete(struct bna_rx *rx)
 	bfa_fsm_send_event(rx, RX_E_CLEANUP_DONE);
 }
 
+void
+bna_rx_vlan_strip_enable(struct bna_rx *rx)
+{
+	struct bna_rxf *rxf = &rx->rxf;
+
+	if (rxf->vlan_strip_status == BNA_STATUS_T_DISABLED) {
+		rxf->vlan_strip_status = BNA_STATUS_T_ENABLED;
+		rxf->vlan_strip_pending = true;
+		bfa_fsm_send_event(rxf, RXF_E_CONFIG);
+	}
+}
+
+void
+bna_rx_vlan_strip_disable(struct bna_rx *rx)
+{
+	struct bna_rxf *rxf = &rx->rxf;
+
+	if (rxf->vlan_strip_status != BNA_STATUS_T_DISABLED) {
+		rxf->vlan_strip_status = BNA_STATUS_T_DISABLED;
+		rxf->vlan_strip_pending = true;
+		bfa_fsm_send_event(rxf, RXF_E_CONFIG);
+	}
+}
+
 enum bna_cb_status
 bna_rx_mode_set(struct bna_rx *rx, enum bna_rxmode new_mode,
 		enum bna_rxmode bitmask,
diff --git a/drivers/net/ethernet/brocade/bna/bnad.c b/drivers/net/ethernet/brocade/bna/bnad.c
index 9b1c550..225939c 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.c
+++ b/drivers/net/ethernet/brocade/bna/bnad.c
@@ -2624,9 +2624,6 @@ bnad_stop(struct net_device *netdev)
 	bnad_destroy_tx(bnad, 0);
 	bnad_destroy_rx(bnad, 0);
 
-	/* These config flags are cleared in the hardware */
-	bnad->cfg_flags &= ~(BNAD_CF_ALLMULTI | BNAD_CF_PROMISC);
-
 	/* Synchronize mailbox IRQ */
 	bnad_mbox_irq_sync(bnad);
 
@@ -2939,73 +2936,133 @@ bnad_get_stats64(struct net_device *netdev, struct rtnl_link_stats64 *stats)
 	return stats;
 }
 
+static void
+bnad_set_rx_ucast_fltr(struct bnad *bnad)
+{
+	struct net_device *netdev = bnad->netdev;
+	int uc_count = netdev_uc_count(netdev);
+	enum bna_cb_status ret;
+	u8 *mac_list;
+	struct netdev_hw_addr *ha;
+	int entry;
+
+	if (netdev_uc_empty(bnad->netdev)) {
+		bna_rx_ucast_listset(bnad->rx_info[0].rx, 0, NULL, NULL);
+		return;
+	}
+
+	if (uc_count > bna_attr(&bnad->bna)->num_ucmac)
+		goto mode_default;
+
+	mac_list = kzalloc(uc_count * ETH_ALEN, GFP_ATOMIC);
+	if (mac_list == NULL)
+		goto mode_default;
+
+	entry = 0;
+	netdev_for_each_uc_addr(ha, netdev) {
+		memcpy(&mac_list[entry * ETH_ALEN],
+		       &ha->addr[0], ETH_ALEN);
+		entry++;
+	}
+
+	ret = bna_rx_ucast_listset(bnad->rx_info[0].rx, entry,
+			mac_list, NULL);
+	kfree(mac_list);
+
+	if (ret != BNA_CB_SUCCESS)
+		goto mode_default;
+
+	return;
+
+	/* ucast packets not in UCAM are routed to default function */
+mode_default:
+	bnad->cfg_flags |= BNAD_CF_DEFAULT;
+	bna_rx_ucast_listset(bnad->rx_info[0].rx, 0, NULL, NULL);
+}
+
+static void
+bnad_set_rx_mcast_fltr(struct bnad *bnad)
+{
+	struct net_device *netdev = bnad->netdev;
+	int mc_count = netdev_mc_count(netdev);
+	enum bna_cb_status ret;
+	u8 *mac_list;
+
+	if (netdev->flags & IFF_ALLMULTI)
+		goto mode_allmulti;
+
+	if (netdev_mc_empty(netdev))
+		return;
+
+	if (mc_count > bna_attr(&bnad->bna)->num_mcmac)
+		goto mode_allmulti;
+
+	mac_list = kzalloc((mc_count + 1) * ETH_ALEN, GFP_ATOMIC);
+
+	if (mac_list == NULL)
+		goto mode_allmulti;
+
+	memcpy(&mac_list[0], &bnad_bcast_addr[0], ETH_ALEN);
+
+	/* copy rest of the MCAST addresses */
+	bnad_netdev_mc_list_get(netdev, mac_list);
+	ret = bna_rx_mcast_listset(bnad->rx_info[0].rx, mc_count + 1,
+			mac_list, NULL);
+	kfree(mac_list);
+
+	if (ret != BNA_CB_SUCCESS)
+		goto mode_allmulti;
+
+	return;
+
+mode_allmulti:
+	bnad->cfg_flags |= BNAD_CF_ALLMULTI;
+	bna_rx_mcast_delall(bnad->rx_info[0].rx, NULL);
+}
+
 void
 bnad_set_rx_mode(struct net_device *netdev)
 {
 	struct bnad *bnad = netdev_priv(netdev);
-	u32	new_mask, valid_mask;
+	enum bna_rxmode new_mode, mode_mask;
 	unsigned long flags;
 
 	spin_lock_irqsave(&bnad->bna_lock, flags);
 
-	new_mask = valid_mask = 0;
-
-	if (netdev->flags & IFF_PROMISC) {
-		if (!(bnad->cfg_flags & BNAD_CF_PROMISC)) {
-			new_mask = BNAD_RXMODE_PROMISC_DEFAULT;
-			valid_mask = BNAD_RXMODE_PROMISC_DEFAULT;
-			bnad->cfg_flags |= BNAD_CF_PROMISC;
-		}
-	} else {
-		if (bnad->cfg_flags & BNAD_CF_PROMISC) {
-			new_mask = ~BNAD_RXMODE_PROMISC_DEFAULT;
-			valid_mask = BNAD_RXMODE_PROMISC_DEFAULT;
-			bnad->cfg_flags &= ~BNAD_CF_PROMISC;
-		}
-	}
-
-	if (netdev->flags & IFF_ALLMULTI) {
-		if (!(bnad->cfg_flags & BNAD_CF_ALLMULTI)) {
-			new_mask |= BNA_RXMODE_ALLMULTI;
-			valid_mask |= BNA_RXMODE_ALLMULTI;
-			bnad->cfg_flags |= BNAD_CF_ALLMULTI;
-		}
-	} else {
-		if (bnad->cfg_flags & BNAD_CF_ALLMULTI) {
-			new_mask &= ~BNA_RXMODE_ALLMULTI;
-			valid_mask |= BNA_RXMODE_ALLMULTI;
-			bnad->cfg_flags &= ~BNAD_CF_ALLMULTI;
-		}
+	if (bnad->rx_info[0].rx == NULL) {
+		spin_unlock_irqrestore(&bnad->bna_lock, flags);
+		return;
 	}
 
-	if (bnad->rx_info[0].rx == NULL)
-		goto unlock;
+	/* clear bnad flags to update it with new settings */
+	bnad->cfg_flags &= ~(BNAD_CF_PROMISC | BNAD_CF_DEFAULT |
+			BNAD_CF_ALLMULTI);
 
-	bna_rx_mode_set(bnad->rx_info[0].rx, new_mask, valid_mask, NULL);
+	new_mode = 0;
+	if (netdev->flags & IFF_PROMISC) {
+		new_mode |= BNAD_RXMODE_PROMISC_DEFAULT;
+		bnad->cfg_flags |= BNAD_CF_PROMISC;
+	} else {
+		bnad_set_rx_mcast_fltr(bnad);
 
-	if (!netdev_mc_empty(netdev)) {
-		u8 *mcaddr_list;
-		int mc_count = netdev_mc_count(netdev);
+		if (bnad->cfg_flags & BNAD_CF_ALLMULTI)
+			new_mode |= BNA_RXMODE_ALLMULTI;
 
-		/* Index 0 holds the broadcast address */
-		mcaddr_list =
-			kzalloc((mc_count + 1) * ETH_ALEN,
-				GFP_ATOMIC);
-		if (!mcaddr_list)
-			goto unlock;
+		bnad_set_rx_ucast_fltr(bnad);
 
-		memcpy(&mcaddr_list[0], &bnad_bcast_addr[0], ETH_ALEN);
+		if (bnad->cfg_flags & BNAD_CF_DEFAULT)
+			new_mode |= BNA_RXMODE_DEFAULT;
+	}
 
-		/* Copy rest of the MC addresses */
-		bnad_netdev_mc_list_get(netdev, mcaddr_list);
+	mode_mask = BNA_RXMODE_PROMISC | BNA_RXMODE_DEFAULT |
+			BNA_RXMODE_ALLMULTI;
+	bna_rx_mode_set(bnad->rx_info[0].rx, new_mode, mode_mask, NULL);
 
-		bna_rx_mcast_listset(bnad->rx_info[0].rx, mc_count + 1,
-					mcaddr_list, NULL);
+	if (bnad->cfg_flags & BNAD_CF_PROMISC)
+		bna_rx_vlan_strip_disable(bnad->rx_info[0].rx);
+	else
+		bna_rx_vlan_strip_enable(bnad->rx_info[0].rx);
 
-		/* Should we enable BNAD_CF_ALLMULTI for err != 0 ? */
-		kfree(mcaddr_list);
-	}
-unlock:
 	spin_unlock_irqrestore(&bnad->bna_lock, flags);
 }
 
diff --git a/drivers/net/ethernet/brocade/bna/bnad.h b/drivers/net/ethernet/brocade/bna/bnad.h
index f7e033f..7055432 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.h
+++ b/drivers/net/ethernet/brocade/bna/bnad.h
@@ -260,7 +260,8 @@ struct bnad_rx_unmap_q {
 #define	BNAD_CF_DIM_ENABLED		0x01	/* DIM */
 #define	BNAD_CF_PROMISC			0x02
 #define BNAD_CF_ALLMULTI		0x04
-#define	BNAD_CF_MSIX			0x08	/* If in MSIx mode */
+#define	BNAD_CF_DEFAULT			0x08
+#define	BNAD_CF_MSIX			0x10	/* If in MSIx mode */
 
 /* Defines for run_flags bit-mask */
 /* Set, tested & cleared using xxx_bit() functions */
-- 
1.8.2.3

[net-next 05/13] bna: Enable Multi Buffer RX

[Rasesh Mody]

The CT2 HW supports multi-buffer Rx. This patch provides the necessary changes
for bnad to use multi-buffer Rx feature. For BNAD, multi-buffer Rx is by
default enabled when MTU is > 4096. Using module parameter bnad_multi_buffer_rx
this behavior can be reverted. For >4096 MTU, q0 data/large buffers are of 2048
size. As the resource requirements of multi-buffer Rx are different new Rx needs
to be created to use this feature. ASIC posts multiple completions if frame
exceeds buffer size. The last completion is marked with EOP flag.
 - Separate HQ and DQ enums for resource allocations and configurations.
 - rx_config and rxq structure changes to pass the correct info from bnad.
 - DQ depth need not be same as HQ depth. So CQ depth is adjusted accordingly.
 - Rx CFG frame size is taken from configured MTU.
 - Rx q0 buffer size is configured from bnad s rx_config when multi-buffer is
   enabled.
 - Poll for entire frame completion.
 - Once EOP completion is received gather the number of vectors used by the
   frame to submit it to the stack.
 - Changed MTU to frame size wherever necessary.

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bfi_enet.h    |   3 +-
 drivers/net/ethernet/brocade/bna/bna_hw_defs.h |   4 +
 drivers/net/ethernet/brocade/bna/bna_tx_rx.c   |  66 +++--
 drivers/net/ethernet/brocade/bna/bna_types.h   |  49 ++--
 drivers/net/ethernet/brocade/bna/bnad.c        | 320 ++++++++++++++++++-------
 drivers/net/ethernet/brocade/bna/bnad.h        |  19 +-
 6 files changed, 336 insertions(+), 125 deletions(-)

diff --git a/drivers/net/ethernet/brocade/bna/bfi_enet.h b/drivers/net/ethernet/brocade/bna/bfi_enet.h
index 7d10e33..ae072dc 100644
--- a/drivers/net/ethernet/brocade/bna/bfi_enet.h
+++ b/drivers/net/ethernet/brocade/bna/bfi_enet.h
@@ -472,7 +472,8 @@ enum bfi_enet_hds_type {
 
 struct bfi_enet_rx_cfg {
 	u8		rxq_type;
-	u8		rsvd[3];
+	u8		rsvd[1];
+	u16		frame_size;
 
 	struct {
 		u8			max_header_size;
diff --git a/drivers/net/ethernet/brocade/bna/bna_hw_defs.h b/drivers/net/ethernet/brocade/bna/bna_hw_defs.h
index af3f7bb..2702d02 100644
--- a/drivers/net/ethernet/brocade/bna/bna_hw_defs.h
+++ b/drivers/net/ethernet/brocade/bna/bna_hw_defs.h
@@ -322,6 +322,10 @@ do {									\
 #define	BNA_CQ_EF_REMOTE	(1 << 19)
 
 #define	BNA_CQ_EF_LOCAL		(1 << 20)
+/* CAT2 ASIC does not use bit 21 as per the SPEC.
+ * Bit 31 is set in every end of frame completion
+ */
+#define BNA_CQ_EF_EOP		(1 << 31)
 
 /* Data structures */
 
diff --git a/drivers/net/ethernet/brocade/bna/bna_tx_rx.c b/drivers/net/ethernet/brocade/bna/bna_tx_rx.c
index e02c126..85e6354 100644
--- a/drivers/net/ethernet/brocade/bna/bna_tx_rx.c
+++ b/drivers/net/ethernet/brocade/bna/bna_tx_rx.c
@@ -1811,6 +1811,7 @@ bna_bfi_rx_enet_start(struct bna_rx *rx)
 	cfg_req->mh.num_entries = htons(
 		bfi_msgq_num_cmd_entries(sizeof(struct bfi_enet_rx_cfg_req)));
 
+	cfg_req->rx_cfg.frame_size = bna_enet_mtu_get(&rx->bna->enet);
 	cfg_req->num_queue_sets = rx->num_paths;
 	for (i = 0, rxp_qe = bfa_q_first(&rx->rxp_q);
 		i < rx->num_paths;
@@ -1832,8 +1833,17 @@ bna_bfi_rx_enet_start(struct bna_rx *rx)
 			/* Large/Single RxQ */
 			bfi_enet_datapath_q_init(&cfg_req->q_cfg[i].ql.q,
 						&q0->qpt);
-			q0->buffer_size =
-				bna_enet_mtu_get(&rx->bna->enet);
+			if (q0->multi_buffer)
+				/* multi-buffer is enabled by allocating
+				 * a new rx with new set of resources.
+				 * q0->buffer_size should be initialized to
+				 * fragment size.
+				 */
+				cfg_req->rx_cfg.multi_buffer =
+					BNA_STATUS_T_ENABLED;
+			else
+				q0->buffer_size =
+					bna_enet_mtu_get(&rx->bna->enet);
 			cfg_req->q_cfg[i].ql.rx_buffer_size =
 				htons((u16)q0->buffer_size);
 			break;
@@ -2383,8 +2393,8 @@ bna_rx_res_req(struct bna_rx_config *q_cfg, struct bna_res_info *res_info)
 	u32 hq_depth;
 	u32 dq_depth;
 
-	dq_depth = q_cfg->q_depth;
-	hq_depth = ((q_cfg->rxp_type == BNA_RXP_SINGLE) ? 0 : q_cfg->q_depth);
+	dq_depth = q_cfg->q0_depth;
+	hq_depth = ((q_cfg->rxp_type == BNA_RXP_SINGLE) ? 0 : q_cfg->q1_depth);
 	cq_depth = dq_depth + hq_depth;
 
 	BNA_TO_POWER_OF_2_HIGH(cq_depth);
@@ -2501,10 +2511,10 @@ bna_rx_create(struct bna *bna, struct bnad *bnad,
 	struct bna_rxq *q0;
 	struct bna_rxq *q1;
 	struct bna_intr_info *intr_info;
-	u32 page_count;
+	struct bna_mem_descr *hqunmap_mem;
+	struct bna_mem_descr *dqunmap_mem;
 	struct bna_mem_descr *ccb_mem;
 	struct bna_mem_descr *rcb_mem;
-	struct bna_mem_descr *unmapq_mem;
 	struct bna_mem_descr *cqpt_mem;
 	struct bna_mem_descr *cswqpt_mem;
 	struct bna_mem_descr *cpage_mem;
@@ -2514,8 +2524,10 @@ bna_rx_create(struct bna *bna, struct bnad *bnad,
 	struct bna_mem_descr *dsqpt_mem;
 	struct bna_mem_descr *hpage_mem;
 	struct bna_mem_descr *dpage_mem;
-	int i;
-	int dpage_count, hpage_count, rcb_idx;
+	u32 dpage_count, hpage_count;
+	u32 hq_idx, dq_idx, rcb_idx;
+	u32 cq_depth, i;
+	u32 page_count;
 
 	if (!bna_rx_res_check(rx_mod, rx_cfg))
 		return NULL;
@@ -2523,7 +2535,8 @@ bna_rx_create(struct bna *bna, struct bnad *bnad,
 	intr_info = &res_info[BNA_RX_RES_T_INTR].res_u.intr_info;
 	ccb_mem = &res_info[BNA_RX_RES_MEM_T_CCB].res_u.mem_info.mdl[0];
 	rcb_mem = &res_info[BNA_RX_RES_MEM_T_RCB].res_u.mem_info.mdl[0];
-	unmapq_mem = &res_info[BNA_RX_RES_MEM_T_UNMAPQ].res_u.mem_info.mdl[0];
+	dqunmap_mem = &res_info[BNA_RX_RES_MEM_T_UNMAPDQ].res_u.mem_info.mdl[0];
+	hqunmap_mem = &res_info[BNA_RX_RES_MEM_T_UNMAPHQ].res_u.mem_info.mdl[0];
 	cqpt_mem = &res_info[BNA_RX_RES_MEM_T_CQPT].res_u.mem_info.mdl[0];
 	cswqpt_mem = &res_info[BNA_RX_RES_MEM_T_CSWQPT].res_u.mem_info.mdl[0];
 	cpage_mem = &res_info[BNA_RX_RES_MEM_T_CQPT_PAGE].res_u.mem_info.mdl[0];
@@ -2575,7 +2588,8 @@ bna_rx_create(struct bna *bna, struct bnad *bnad,
 	}
 
 	rx->num_paths = rx_cfg->num_paths;
-	for (i = 0, rcb_idx = 0; i < rx->num_paths; i++) {
+	for (i = 0, hq_idx = 0, dq_idx = 0, rcb_idx = 0;
+			i < rx->num_paths; i++) {
 		rxp = bna_rxp_get(rx_mod);
 		list_add_tail(&rxp->qe, &rx->rxp_q);
 		rxp->type = rx_cfg->rxp_type;
@@ -2618,9 +2632,13 @@ bna_rx_create(struct bna *bna, struct bnad *bnad,
 		q0->rxp = rxp;
 
 		q0->rcb = (struct bna_rcb *) rcb_mem[rcb_idx].kva;
-		q0->rcb->unmap_q = (void *)unmapq_mem[rcb_idx].kva;
-		rcb_idx++;
-		q0->rcb->q_depth = rx_cfg->q_depth;
+		q0->rcb->unmap_q = (void *)dqunmap_mem[dq_idx].kva;
+		rcb_idx++; dq_idx++;
+		q0->rcb->q_depth = rx_cfg->q0_depth;
+		q0->q_depth = rx_cfg->q0_depth;
+		q0->multi_buffer = rx_cfg->q0_multi_buf;
+		q0->buffer_size = rx_cfg->q0_buf_size;
+		q0->num_vecs = rx_cfg->q0_num_vecs;
 		q0->rcb->rxq = q0;
 		q0->rcb->bnad = bna->bnad;
 		q0->rcb->id = 0;
@@ -2640,15 +2658,18 @@ bna_rx_create(struct bna *bna, struct bnad *bnad,
 			q1->rxp = rxp;
 
 			q1->rcb = (struct bna_rcb *) rcb_mem[rcb_idx].kva;
-			q1->rcb->unmap_q = (void *)unmapq_mem[rcb_idx].kva;
-			rcb_idx++;
-			q1->rcb->q_depth = rx_cfg->q_depth;
+			q1->rcb->unmap_q = (void *)hqunmap_mem[hq_idx].kva;
+			rcb_idx++; hq_idx++;
+			q1->rcb->q_depth = rx_cfg->q1_depth;
+			q1->q_depth = rx_cfg->q1_depth;
+			q1->multi_buffer = BNA_STATUS_T_DISABLED;
+			q1->num_vecs = 1;
 			q1->rcb->rxq = q1;
 			q1->rcb->bnad = bna->bnad;
 			q1->rcb->id = 1;
 			q1->buffer_size = (rx_cfg->rxp_type == BNA_RXP_HDS) ?
 					rx_cfg->hds_config.forced_offset
-					: rx_cfg->small_buff_size;
+					: rx_cfg->q1_buf_size;
 			q1->rx_packets = q1->rx_bytes = 0;
 			q1->rx_packets_with_error = q1->rxbuf_alloc_failed = 0;
 
@@ -2663,9 +2684,14 @@ bna_rx_create(struct bna *bna, struct bnad *bnad,
 		/* Setup CQ */
 
 		rxp->cq.ccb = (struct bna_ccb *) ccb_mem[i].kva;
-		rxp->cq.ccb->q_depth =	rx_cfg->q_depth +
-					((rx_cfg->rxp_type == BNA_RXP_SINGLE) ?
-					0 : rx_cfg->q_depth);
+		cq_depth = rx_cfg->q0_depth +
+			((rx_cfg->rxp_type == BNA_RXP_SINGLE) ?
+			 0 : rx_cfg->q1_depth);
+		/* if multi-buffer is enabled sum of q0_depth
+		 * and q1_depth need not be a power of 2
+		 */
+		BNA_TO_POWER_OF_2_HIGH(cq_depth);
+		rxp->cq.ccb->q_depth = cq_depth;
 		rxp->cq.ccb->cq = &rxp->cq;
 		rxp->cq.ccb->rcb[0] = q0->rcb;
 		q0->rcb->ccb = rxp->cq.ccb;
diff --git a/drivers/net/ethernet/brocade/bna/bna_types.h b/drivers/net/ethernet/brocade/bna/bna_types.h
index acedac2..621547c 100644
--- a/drivers/net/ethernet/brocade/bna/bna_types.h
+++ b/drivers/net/ethernet/brocade/bna/bna_types.h
@@ -109,20 +109,21 @@ enum bna_tx_res_req_type {
 enum bna_rx_mem_type {
 	BNA_RX_RES_MEM_T_CCB		= 0,	/* CQ context */
 	BNA_RX_RES_MEM_T_RCB		= 1,	/* CQ context */
-	BNA_RX_RES_MEM_T_UNMAPQ		= 2,	/* UnmapQ for RxQs */
-	BNA_RX_RES_MEM_T_CQPT		= 3,	/* CQ QPT */
-	BNA_RX_RES_MEM_T_CSWQPT		= 4,	/* S/W QPT */
-	BNA_RX_RES_MEM_T_CQPT_PAGE	= 5,	/* CQPT page */
-	BNA_RX_RES_MEM_T_HQPT		= 6,	/* RX QPT */
-	BNA_RX_RES_MEM_T_DQPT		= 7,	/* RX QPT */
-	BNA_RX_RES_MEM_T_HSWQPT		= 8,	/* RX s/w QPT */
-	BNA_RX_RES_MEM_T_DSWQPT		= 9,	/* RX s/w QPT */
-	BNA_RX_RES_MEM_T_DPAGE		= 10,	/* RX s/w QPT */
-	BNA_RX_RES_MEM_T_HPAGE		= 11,	/* RX s/w QPT */
-	BNA_RX_RES_MEM_T_IBIDX		= 12,
-	BNA_RX_RES_MEM_T_RIT		= 13,
-	BNA_RX_RES_T_INTR		= 14,	/* Rx interrupts */
-	BNA_RX_RES_T_MAX		= 15
+	BNA_RX_RES_MEM_T_UNMAPHQ	= 2,
+	BNA_RX_RES_MEM_T_UNMAPDQ	= 3,
+	BNA_RX_RES_MEM_T_CQPT		= 4,
+	BNA_RX_RES_MEM_T_CSWQPT		= 5,
+	BNA_RX_RES_MEM_T_CQPT_PAGE	= 6,
+	BNA_RX_RES_MEM_T_HQPT		= 7,
+	BNA_RX_RES_MEM_T_DQPT		= 8,
+	BNA_RX_RES_MEM_T_HSWQPT		= 9,
+	BNA_RX_RES_MEM_T_DSWQPT		= 10,
+	BNA_RX_RES_MEM_T_DPAGE		= 11,
+	BNA_RX_RES_MEM_T_HPAGE		= 12,
+	BNA_RX_RES_MEM_T_IBIDX		= 13,
+	BNA_RX_RES_MEM_T_RIT		= 14,
+	BNA_RX_RES_T_INTR		= 15,
+	BNA_RX_RES_T_MAX		= 16
 };
 
 enum bna_tx_type {
@@ -583,6 +584,8 @@ struct bna_rxq {
 
 	int			buffer_size;
 	int			q_depth;
+	u32			num_vecs;
+	enum bna_status		multi_buffer;
 
 	struct bna_qpt qpt;
 	struct bna_rcb *rcb;
@@ -632,6 +635,8 @@ struct bna_ccb {
 	struct bna_rcb *rcb[2];
 	void			*ctrl; /* For bnad */
 	struct bna_pkt_rate pkt_rate;
+	u32			pkts_una;
+	u32			bytes_per_intr;
 
 	/* Control path */
 	struct bna_cq *cq;
@@ -671,14 +676,22 @@ struct bna_rx_config {
 	int			num_paths;
 	enum bna_rxp_type rxp_type;
 	int			paused;
-	int			q_depth;
 	int			coalescing_timeo;
 	/*
 	 * Small/Large (or Header/Data) buffer size to be configured
-	 * for SLR and HDS queue type. Large buffer size comes from
-	 * enet->mtu.
+	 * for SLR and HDS queue type.
 	 */
-	int			small_buff_size;
+	u32			frame_size;
+
+	/* header or small queue */
+	u32			q1_depth;
+	u32			q1_buf_size;
+
+	/* data or large queue */
+	u32			q0_depth;
+	u32			q0_buf_size;
+	u32			q0_num_vecs;
+	enum bna_status		q0_multi_buf;
 
 	enum bna_status rss_status;
 	struct bna_rss_config rss_config;
diff --git a/drivers/net/ethernet/brocade/bna/bnad.c b/drivers/net/ethernet/brocade/bna/bnad.c
index 225939c..33aeff2 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.c
+++ b/drivers/net/ethernet/brocade/bna/bnad.c
@@ -49,6 +49,10 @@ module_param(bna_debugfs_enable, uint, S_IRUGO | S_IWUSR);
 MODULE_PARM_DESC(bna_debugfs_enable, "Enables debugfs feature, default=1,"
 		 " Range[false:0|true:1]");
 
+static uint bnad_multi_buffer_rx = 1;
+module_param(bnad_multi_buffer_rx, uint, 0444);
+MODULE_PARM_DESC(bnad_multi_buffer_rx, "Multi-buffer Rx; 1 - en(default), 0 - dis");
+
 /*
  * Global variables
  */
@@ -282,27 +286,32 @@ static int
 bnad_rxq_alloc_init(struct bnad *bnad, struct bna_rcb *rcb)
 {
 	struct bnad_rx_unmap_q *unmap_q = rcb->unmap_q;
-	int mtu, order;
+	int order;
 
 	bnad_rxq_alloc_uninit(bnad, rcb);
 
-	mtu = bna_enet_mtu_get(&bnad->bna.enet);
-	order = get_order(mtu);
+	order = get_order(rcb->rxq->buffer_size);
+
+	unmap_q->type = BNAD_RXBUF_PAGE;
 
 	if (bna_is_small_rxq(rcb->id)) {
 		unmap_q->alloc_order = 0;
 		unmap_q->map_size = rcb->rxq->buffer_size;
 	} else {
-		unmap_q->alloc_order = order;
-		unmap_q->map_size =
-			(rcb->rxq->buffer_size > 2048) ?
-			PAGE_SIZE << order : 2048;
+		if (rcb->rxq->multi_buffer) {
+			unmap_q->alloc_order = 0;
+			unmap_q->map_size = rcb->rxq->buffer_size;
+			unmap_q->type = BNAD_RXBUF_MULTI_BUFF;
+		} else {
+			unmap_q->alloc_order = order;
+			unmap_q->map_size =
+				(rcb->rxq->buffer_size > 2048) ?
+				PAGE_SIZE << order : 2048;
+		}
 	}
 
 	BUG_ON(((PAGE_SIZE << order) % unmap_q->map_size));
 
-	unmap_q->type = BNAD_RXBUF_PAGE;
-
 	return 0;
 }
 
@@ -345,10 +354,10 @@ bnad_rxq_cleanup(struct bnad *bnad, struct bna_rcb *rcb)
 	for (i = 0; i < rcb->q_depth; i++) {
 		struct bnad_rx_unmap *unmap = &unmap_q->unmap[i];
 
-		if (BNAD_RXBUF_IS_PAGE(unmap_q->type))
-			bnad_rxq_cleanup_page(bnad, unmap);
-		else
+		if (BNAD_RXBUF_IS_SK_BUFF(unmap_q->type))
 			bnad_rxq_cleanup_skb(bnad, unmap);
+		else
+			bnad_rxq_cleanup_page(bnad, unmap);
 	}
 	bnad_rxq_alloc_uninit(bnad, rcb);
 }
@@ -480,10 +489,10 @@ bnad_rxq_post(struct bnad *bnad, struct bna_rcb *rcb)
 	if (!(to_alloc >> BNAD_RXQ_REFILL_THRESHOLD_SHIFT))
 		return;
 
-	if (BNAD_RXBUF_IS_PAGE(unmap_q->type))
-		bnad_rxq_refill_page(bnad, rcb, to_alloc);
-	else
+	if (BNAD_RXBUF_IS_SK_BUFF(unmap_q->type))
 		bnad_rxq_refill_skb(bnad, rcb, to_alloc);
+	else
+		bnad_rxq_refill_page(bnad, rcb, to_alloc);
 }
 
 #define flags_cksum_prot_mask (BNA_CQ_EF_IPV4 | BNA_CQ_EF_L3_CKSUM_OK | \
@@ -500,62 +509,91 @@ bnad_rxq_post(struct bnad *bnad, struct bna_rcb *rcb)
 #define flags_udp6 (BNA_CQ_EF_IPV6 | \
 				BNA_CQ_EF_UDP | BNA_CQ_EF_L4_CKSUM_OK)
 
-static inline struct sk_buff *
-bnad_cq_prepare_skb(struct bnad_rx_ctrl *rx_ctrl,
-		struct bnad_rx_unmap_q *unmap_q,
-		struct bnad_rx_unmap *unmap,
-		u32 length, u32 flags)
+static void
+bnad_cq_drop_packet(struct bnad *bnad, struct bna_rcb *rcb,
+		    u32 sop_ci, u32 nvecs)
 {
-	struct bnad *bnad = rx_ctrl->bnad;
-	struct sk_buff *skb;
+	struct bnad_rx_unmap_q *unmap_q;
+	struct bnad_rx_unmap *unmap;
+	u32 ci, vec;
+
+	unmap_q = rcb->unmap_q;
+	for (vec = 0, ci = sop_ci; vec < nvecs; vec++) {
+		unmap = &unmap_q->unmap[ci];
+		BNA_QE_INDX_INC(ci, rcb->q_depth);
+
+		if (BNAD_RXBUF_IS_SK_BUFF(unmap_q->type))
+			bnad_rxq_cleanup_skb(bnad, unmap);
+		else
+			bnad_rxq_cleanup_page(bnad, unmap);
+	}
+}
 
-	if (BNAD_RXBUF_IS_PAGE(unmap_q->type)) {
-		skb = napi_get_frags(&rx_ctrl->napi);
-		if (unlikely(!skb))
-			return NULL;
+static void
+bnad_cq_setup_skb_frags(struct bna_rcb *rcb, struct sk_buff *skb,
+			u32 sop_ci, u32 nvecs, u32 last_fraglen)
+{
+	struct bnad *bnad;
+	u32 ci, vec, len, totlen = 0;
+	struct bnad_rx_unmap_q *unmap_q;
+	struct bnad_rx_unmap *unmap;
+
+	unmap_q = rcb->unmap_q;
+	bnad = rcb->bnad;
+	for (vec = 1, ci = sop_ci; vec <= nvecs; vec++) {
+		unmap = &unmap_q->unmap[ci];
+		BNA_QE_INDX_INC(ci, rcb->q_depth);
 
 		dma_unmap_page(&bnad->pcidev->dev,
 				dma_unmap_addr(&unmap->vector, dma_addr),
 				unmap->vector.len, DMA_FROM_DEVICE);
+
+		len = (vec == nvecs) ?
+			last_fraglen : unmap->vector.len;
+		totlen += len;
+
 		skb_fill_page_desc(skb, skb_shinfo(skb)->nr_frags,
-				unmap->page, unmap->page_offset, length);
-		skb->len += length;
-		skb->data_len += length;
-		skb->truesize += length;
+				unmap->page, unmap->page_offset, len);
 
 		unmap->page = NULL;
 		unmap->vector.len = 0;
-
-		return skb;
 	}
 
-	skb = unmap->skb;
-	BUG_ON(!skb);
+	skb->len += totlen;
+	skb->data_len += totlen;
+	skb->truesize += totlen;
+}
+
+static inline void
+bnad_cq_setup_skb(struct bnad *bnad, struct sk_buff *skb,
+		  struct bnad_rx_unmap *unmap, u32 len)
+{
+	prefetch(skb->data);
 
 	dma_unmap_single(&bnad->pcidev->dev,
 			dma_unmap_addr(&unmap->vector, dma_addr),
 			unmap->vector.len, DMA_FROM_DEVICE);
 
-	skb_put(skb, length);
-
+	skb_put(skb, len);
 	skb->protocol = eth_type_trans(skb, bnad->netdev);
 
 	unmap->skb = NULL;
 	unmap->vector.len = 0;
-	return skb;
 }
 
 static u32
 bnad_cq_process(struct bnad *bnad, struct bna_ccb *ccb, int budget)
 {
-	struct bna_cq_entry *cq, *cmpl;
+	struct bna_cq_entry *cq, *cmpl, *next_cmpl;
 	struct bna_rcb *rcb = NULL;
 	struct bnad_rx_unmap_q *unmap_q;
-	struct bnad_rx_unmap *unmap;
-	struct sk_buff *skb;
+	struct bnad_rx_unmap *unmap = NULL;
+	struct sk_buff *skb = NULL;
 	struct bna_pkt_rate *pkt_rt = &ccb->pkt_rate;
 	struct bnad_rx_ctrl *rx_ctrl = ccb->ctrl;
-	u32 packets = 0, length = 0, flags, masked_flags;
+	u32 packets = 0, len = 0, totlen = 0;
+	u32 pi, vec, sop_ci = 0, nvecs = 0;
+	u32 flags, masked_flags;
 
 	prefetch(bnad->netdev);
 
@@ -563,9 +601,6 @@ bnad_cq_process(struct bnad *bnad, struct bna_ccb *ccb, int budget)
 	cmpl = &cq[ccb->producer_index];
 
 	while (cmpl->valid && (packets < budget)) {
-		packets++;
-		flags = ntohl(cmpl->flags);
-		length = ntohs(cmpl->length);
 		BNA_UPDATE_PKT_CNT(pkt_rt, ntohs(cmpl->length));
 
 		if (bna_is_small_rxq(cmpl->rxq_id))
@@ -574,25 +609,68 @@ bnad_cq_process(struct bnad *bnad, struct bna_ccb *ccb, int budget)
 			rcb = ccb->rcb[0];
 
 		unmap_q = rcb->unmap_q;
-		unmap = &unmap_q->unmap[rcb->consumer_index];
 
-		if (unlikely(flags & (BNA_CQ_EF_MAC_ERROR |
-					BNA_CQ_EF_FCS_ERROR |
-					BNA_CQ_EF_TOO_LONG))) {
-			if (BNAD_RXBUF_IS_PAGE(unmap_q->type))
-				bnad_rxq_cleanup_page(bnad, unmap);
-			else
-				bnad_rxq_cleanup_skb(bnad, unmap);
+		/* start of packet ci */
+		sop_ci = rcb->consumer_index;
+
+		if (BNAD_RXBUF_IS_SK_BUFF(unmap_q->type)) {
+			unmap = &unmap_q->unmap[sop_ci];
+			skb = unmap->skb;
+		} else {
+			skb = napi_get_frags(&rx_ctrl->napi);
+			if (unlikely(!skb))
+				break;
+		}
+		prefetch(skb);
 
+		flags = ntohl(cmpl->flags);
+		len = ntohs(cmpl->length);
+		totlen = len;
+		nvecs = 1;
+
+		/* Check all the completions for this frame.
+		 * busy-wait doesn't help much, break here.
+		 */
+		if (BNAD_RXBUF_IS_MULTI_BUFF(unmap_q->type) &&
+		    (flags & BNA_CQ_EF_EOP) == 0) {
+			pi = ccb->producer_index;
+			do {
+				BNA_QE_INDX_INC(pi, ccb->q_depth);
+				next_cmpl = &cq[pi];
+
+				if (!next_cmpl->valid)
+					break;
+
+				len = ntohs(next_cmpl->length);
+				flags = ntohl(next_cmpl->flags);
+
+				nvecs++;
+				totlen += len;
+			} while ((flags & BNA_CQ_EF_EOP) == 0);
+
+			if (!next_cmpl->valid)
+				break;
+		}
+
+		/* TODO: BNA_CQ_EF_LOCAL ? */
+		if (unlikely(flags & (BNA_CQ_EF_MAC_ERROR |
+						BNA_CQ_EF_FCS_ERROR |
+						BNA_CQ_EF_TOO_LONG))) {
+			bnad_cq_drop_packet(bnad, rcb, sop_ci, nvecs);
 			rcb->rxq->rx_packets_with_error++;
+
 			goto next;
 		}
 
-		skb = bnad_cq_prepare_skb(ccb->ctrl, unmap_q, unmap,
-				length, flags);
+		if (BNAD_RXBUF_IS_SK_BUFF(unmap_q->type))
+			bnad_cq_setup_skb(bnad, skb, unmap, len);
+		else
+			bnad_cq_setup_skb_frags(rcb, skb, sop_ci, nvecs, len);
 
-		if (unlikely(!skb))
-			break;
+		packets++;
+		rcb->rxq->rx_packets++;
+		rcb->rxq->rx_bytes += totlen;
+		ccb->bytes_per_intr += totlen;
 
 		masked_flags = flags & flags_cksum_prot_mask;
 
@@ -606,21 +684,21 @@ bnad_cq_process(struct bnad *bnad, struct bna_ccb *ccb, int budget)
 		else
 			skb_checksum_none_assert(skb);
 
-		rcb->rxq->rx_packets++;
-		rcb->rxq->rx_bytes += length;
-
 		if (flags & BNA_CQ_EF_VLAN)
 			__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), ntohs(cmpl->vlan_tag));
 
-		if (BNAD_RXBUF_IS_PAGE(unmap_q->type))
-			napi_gro_frags(&rx_ctrl->napi);
-		else
+		if (BNAD_RXBUF_IS_SK_BUFF(unmap_q->type))
 			netif_receive_skb(skb);
+		else
+			napi_gro_frags(&rx_ctrl->napi);
 
 next:
-		cmpl->valid = 0;
-		BNA_QE_INDX_INC(rcb->consumer_index, rcb->q_depth);
-		BNA_QE_INDX_INC(ccb->producer_index, ccb->q_depth);
+		BNA_QE_INDX_ADD(rcb->consumer_index, nvecs, rcb->q_depth);
+		for (vec = 0; vec < nvecs; vec++) {
+			cmpl = &cq[ccb->producer_index];
+			cmpl->valid = 0;
+			BNA_QE_INDX_INC(ccb->producer_index, ccb->q_depth);
+		}
 		cmpl = &cq[ccb->producer_index];
 	}
 
@@ -1950,10 +2028,38 @@ bnad_init_rx_config(struct bnad *bnad, struct bna_rx_config *rx_config)
 		memset(&rx_config->rss_config, 0,
 		       sizeof(rx_config->rss_config));
 	}
+
+	rx_config->frame_size = BNAD_FRAME_SIZE(bnad->netdev->mtu);
+
+	/* BNA_RXP_SINGLE - one data-buffer queue
+	 * BNA_RXP_SLR - one small-buffer and one large-buffer queues
+	 * BNA_RXP_HDS - one header-buffer and one data-buffer queues
+	 */
+	/* TODO: configurable param for queue type */
 	rx_config->rxp_type = BNA_RXP_SLR;
-	rx_config->q_depth = bnad->rxq_depth;
 
-	rx_config->small_buff_size = BFI_SMALL_RXBUF_SIZE;
+	if (BNAD_PCI_DEV_IS_CAT2(bnad) &&
+	    bnad_enable_multi_buffer(rx_config->frame_size)) {
+		/* though size_routing_enable is set in SLR,
+		 * small packets may get routed to same rxq.
+		 * set buf_size to 2048 instead of PAGE_SIZE.
+		 */
+		rx_config->q0_buf_size = 2048;
+		/* this should be in multiples of 2 */
+		rx_config->q0_num_vecs = 4;
+		rx_config->q0_depth = bnad->rxq_depth * rx_config->q0_num_vecs;
+		rx_config->q0_multi_buf = BNA_STATUS_T_ENABLED;
+	} else {
+		rx_config->q0_buf_size = rx_config->frame_size;
+		rx_config->q0_num_vecs = 1;
+		rx_config->q0_depth = bnad->rxq_depth;
+	}
+
+	/* initialize for q1 for BNA_RXP_SLR/BNA_RXP_HDS */
+	if (rx_config->rxp_type == BNA_RXP_SLR) {
+		rx_config->q1_depth = bnad->rxq_depth;
+		rx_config->q1_buf_size = BFI_SMALL_RXBUF_SIZE;
+	}
 
 	rx_config->vlan_strip_status = BNA_STATUS_T_ENABLED;
 }
@@ -1969,6 +2075,41 @@ bnad_rx_ctrl_init(struct bnad *bnad, u32 rx_id)
 }
 
 /* Called with mutex_lock(&bnad->conf_mutex) held */
+u32
+bnad_reinit_rx(struct bnad *bnad)
+{
+	struct net_device *netdev = bnad->netdev;
+	u32 err = 0, current_err = 0;
+	u32 rx_id = 0, count = 0;
+	unsigned long flags;
+
+	/* destroy and create new rx objects */
+	for (rx_id = 0; rx_id < bnad->num_rx; rx_id++) {
+		if (!bnad->rx_info[rx_id].rx)
+			continue;
+		count++;
+		bnad_destroy_rx(bnad, rx_id);
+		current_err = bnad_setup_rx(bnad, rx_id);
+		if (current_err && !err) {
+			err = current_err;
+			pr_err("RXQ:%u setup failed\n", rx_id);
+		}
+	}
+
+	/* restore rx configuration */
+	if (bnad->rx_info[0].rx && !err) {
+		bnad_restore_vlans(bnad, 0);
+		bnad_enable_default_bcast(bnad);
+		spin_lock_irqsave(&bnad->bna_lock, flags);
+		bnad_mac_addr_set_locked(bnad, netdev->dev_addr);
+		spin_unlock_irqrestore(&bnad->bna_lock, flags);
+		bnad_set_rx_mode(netdev);
+	}
+
+	return count;
+}
+
+/* Called with bnad_conf_lock() held */
 void
 bnad_destroy_rx(struct bnad *bnad, u32 rx_id)
 {
@@ -2047,13 +2188,19 @@ bnad_setup_rx(struct bnad *bnad, u32 rx_id)
 	spin_unlock_irqrestore(&bnad->bna_lock, flags);
 
 	/* Fill Unmap Q memory requirements */
-	BNAD_FILL_UNMAPQ_MEM_REQ(&res_info[BNA_RX_RES_MEM_T_UNMAPQ],
-			rx_config->num_paths +
-			((rx_config->rxp_type == BNA_RXP_SINGLE) ?
-			 0 : rx_config->num_paths),
-			((bnad->rxq_depth * sizeof(struct bnad_rx_unmap)) +
-			 sizeof(struct bnad_rx_unmap_q)));
-
+	BNAD_FILL_UNMAPQ_MEM_REQ(&res_info[BNA_RX_RES_MEM_T_UNMAPDQ],
+				 rx_config->num_paths,
+			(rx_config->q0_depth *
+			 sizeof(struct bnad_rx_unmap)) +
+			 sizeof(struct bnad_rx_unmap_q));
+
+	if (rx_config->rxp_type != BNA_RXP_SINGLE) {
+		BNAD_FILL_UNMAPQ_MEM_REQ(&res_info[BNA_RX_RES_MEM_T_UNMAPHQ],
+					 rx_config->num_paths,
+				(rx_config->q1_depth *
+				 sizeof(struct bnad_rx_unmap) +
+				 sizeof(struct bnad_rx_unmap_q)));
+	}
 	/* Allocate resource */
 	err = bnad_rx_res_alloc(bnad, res_info, rx_id);
 	if (err)
@@ -2548,7 +2695,6 @@ bnad_open(struct net_device *netdev)
 	int err;
 	struct bnad *bnad = netdev_priv(netdev);
 	struct bna_pause_config pause_config;
-	int mtu;
 	unsigned long flags;
 
 	mutex_lock(&bnad->conf_mutex);
@@ -2567,10 +2713,9 @@ bnad_open(struct net_device *netdev)
 	pause_config.tx_pause = 0;
 	pause_config.rx_pause = 0;
 
-	mtu = ETH_HLEN + VLAN_HLEN + bnad->netdev->mtu + ETH_FCS_LEN;
-
 	spin_lock_irqsave(&bnad->bna_lock, flags);
-	bna_enet_mtu_set(&bnad->bna.enet, mtu, NULL);
+	bna_enet_mtu_set(&bnad->bna.enet,
+			 BNAD_FRAME_SIZE(bnad->netdev->mtu), NULL);
 	bna_enet_pause_config(&bnad->bna.enet, &pause_config, NULL);
 	bna_enet_enable(&bnad->bna.enet);
 	spin_unlock_irqrestore(&bnad->bna_lock, flags);
@@ -3092,14 +3237,14 @@ bnad_set_mac_address(struct net_device *netdev, void *mac_addr)
 }
 
 static int
-bnad_mtu_set(struct bnad *bnad, int mtu)
+bnad_mtu_set(struct bnad *bnad, int frame_size)
 {
 	unsigned long flags;
 
 	init_completion(&bnad->bnad_completions.mtu_comp);
 
 	spin_lock_irqsave(&bnad->bna_lock, flags);
-	bna_enet_mtu_set(&bnad->bna.enet, mtu, bnad_cb_enet_mtu_set);
+	bna_enet_mtu_set(&bnad->bna.enet, frame_size, bnad_cb_enet_mtu_set);
 	spin_unlock_irqrestore(&bnad->bna_lock, flags);
 
 	wait_for_completion(&bnad->bnad_completions.mtu_comp);
@@ -3110,18 +3255,27 @@ bnad_mtu_set(struct bnad *bnad, int mtu)
 static int
 bnad_change_mtu(struct net_device *netdev, int new_mtu)
 {
-	int err, mtu = netdev->mtu;
+	int err, mtu;
 	struct bnad *bnad = netdev_priv(netdev);
+	u32 rx_count = 0;
 
 	if (new_mtu + ETH_HLEN < ETH_ZLEN || new_mtu > BNAD_JUMBO_MTU)
 		return -EINVAL;
 
 	mutex_lock(&bnad->conf_mutex);
 
+	mtu = netdev->mtu;
 	netdev->mtu = new_mtu;
 
-	mtu = ETH_HLEN + VLAN_HLEN + new_mtu + ETH_FCS_LEN;
-	err = bnad_mtu_set(bnad, mtu);
+	/* check if multi-buffer needs to be enabled */
+	if (BNAD_PCI_DEV_IS_CAT2(bnad) &&
+	    bnad_reinit_rx_needed(BNAD_FRAME_SIZE(new_mtu)))
+		rx_count = bnad_reinit_rx(bnad);
+
+	/* rx_count > 0 - new rx created
+	 *	- Linux set err = 0 and return
+	 */
+	err = bnad_mtu_set(bnad, new_mtu);
 	if (err)
 		err = -EBUSY;
 
diff --git a/drivers/net/ethernet/brocade/bna/bnad.h b/drivers/net/ethernet/brocade/bna/bnad.h
index 7055432..03de9cc 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.h
+++ b/drivers/net/ethernet/brocade/bna/bnad.h
@@ -105,6 +105,9 @@ struct bnad_rx_ctrl {
 #define BNAD_NUM_TXQ			(bnad->num_tx * bnad->num_txq_per_tx)
 #define BNAD_NUM_RXP			(bnad->num_rx * bnad->num_rxp_per_rx)
 
+#define BNAD_FRAME_SIZE(_mtu) \
+	(ETH_HLEN + VLAN_HLEN + (_mtu) + ETH_FCS_LEN)
+
 /*
  * DATA STRUCTURES
  */
@@ -241,12 +244,13 @@ struct bnad_rx_unmap {
 
 enum bnad_rxbuf_type {
 	BNAD_RXBUF_NONE		= 0,
-	BNAD_RXBUF_SKB		= 1,
+	BNAD_RXBUF_SK_BUFF	= 1,
 	BNAD_RXBUF_PAGE		= 2,
-	BNAD_RXBUF_MULTI	= 3
+	BNAD_RXBUF_MULTI_BUFF	= 3
 };
 
-#define BNAD_RXBUF_IS_PAGE(_type)	((_type) == BNAD_RXBUF_PAGE)
+#define BNAD_RXBUF_IS_SK_BUFF(_type)	((_type) == BNAD_RXBUF_SK_BUFF)
+#define BNAD_RXBUF_IS_MULTI_BUFF(_type)	((_type) == BNAD_RXBUF_MULTI_BUFF)
 
 struct bnad_rx_unmap_q {
 	int			reuse_pi;
@@ -256,6 +260,15 @@ struct bnad_rx_unmap_q {
 	struct bnad_rx_unmap	unmap[0];
 };
 
+#define BNAD_PCI_DEV_IS_CAT2(_bnad) \
+	((_bnad)->pcidev->device == BFA_PCI_DEVICE_ID_CT2)
+
+#define bnad_enable_multi_buffer(_mtu) \
+		bnad_reinit_rx_needed((_mtu))
+
+#define bnad_reinit_rx_needed(_frame) \
+	(bnad_multi_buffer_rx && (_frame) > 4096)
+
 /* Bit mask values for bnad->cfg_flags */
 #define	BNAD_CF_DIM_ENABLED		0x01	/* DIM */
 #define	BNAD_CF_PROMISC			0x02
-- 
1.8.2.3

[net-next 06/13] bna: UDP RX Processing Improvements

[Rasesh Mody]

Change Details:
 - Prefetch header in GRO path. This reduces napi_frags_skb time from 9% to 5%.
 - Changed the configurable limit of RxQ depth to 16384 (was 2048).
 - Rx unmap structure of 40 bytes is packed to 32 bytes reducing unmap_q size by 16K.
 - bnad_rx_unmap_q elements are cachealigned.

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bnad.c | 5 +++++
 drivers/net/ethernet/brocade/bna/bnad.h | 8 ++++----
 2 files changed, 9 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/brocade/bna/bnad.c b/drivers/net/ethernet/brocade/bna/bnad.c
index 33aeff2..849de5a 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.c
+++ b/drivers/net/ethernet/brocade/bna/bnad.c
@@ -540,6 +540,11 @@ bnad_cq_setup_skb_frags(struct bna_rcb *rcb, struct sk_buff *skb,
 
 	unmap_q = rcb->unmap_q;
 	bnad = rcb->bnad;
+
+	/* prefetch header */
+	prefetch(page_address(unmap_q->unmap[sop_ci].page) +
+			unmap_q->unmap[sop_ci].page_offset);
+
 	for (vec = 1, ci = sop_ci; vec <= nvecs; vec++) {
 		unmap = &unmap_q->unmap[ci];
 		BNA_QE_INDX_INC(ci, rcb->q_depth);
diff --git a/drivers/net/ethernet/brocade/bna/bnad.h b/drivers/net/ethernet/brocade/bna/bnad.h
index 03de9cc..d1dc930 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.h
+++ b/drivers/net/ethernet/brocade/bna/bnad.h
@@ -84,7 +84,7 @@ struct bnad_rx_ctrl {
 #define BNAD_IOCETH_TIMEOUT	     10000
 
 #define BNAD_MIN_Q_DEPTH		512
-#define BNAD_MAX_RXQ_DEPTH		2048
+#define BNAD_MAX_RXQ_DEPTH		16384
 #define BNAD_MAX_TXQ_DEPTH		2048
 
 #define BNAD_JUMBO_MTU			9000
@@ -233,13 +233,13 @@ struct bnad_tx_unmap {
 struct bnad_rx_vector {
 	DEFINE_DMA_UNMAP_ADDR(dma_addr);
 	u32			len;
-};
+} __packed;
 
 struct bnad_rx_unmap {
 	struct page		*page;
-	u32			page_offset;
 	struct sk_buff		*skb;
 	struct bnad_rx_vector	vector;
+	u32			page_offset;
 };
 
 enum bnad_rxbuf_type {
@@ -257,7 +257,7 @@ struct bnad_rx_unmap_q {
 	int			alloc_order;
 	u32			map_size;
 	enum bnad_rxbuf_type	type;
-	struct bnad_rx_unmap	unmap[0];
+	struct bnad_rx_unmap	unmap[0] ____cacheline_aligned;
 };
 
 #define BNAD_PCI_DEV_IS_CAT2(_bnad) \
-- 
1.8.2.3

Re: [net-next 06/13] bna: UDP RX Processing Improvements

[David Miller]

From: Rasesh Mody <rmody@brocade.com>
Date: Sat, 7 Dec 2013 21:48:37 -0800

> @@ -233,13 +233,13 @@ struct bnad_tx_unmap {
>  struct bnad_rx_vector {
>  	DEFINE_DMA_UNMAP_ADDR(dma_addr);
>  	u32			len;
> -};
> +} __packed;

This is absolutely unacceptable, and at best not even mentioned in your
commit message.

There is no reason whatsoever to mark this structure as __packed.

I'm not applying this patch series, sorry.

RE: [net-next 06/13] bna: UDP RX Processing Improvements

[Rasesh Mody]

Hi Dave,

>From: David Miller [mailto:davem@davemloft.net]
>Sent: Monday, December 09, 2013 4:18 PM
>
>From: Rasesh Mody <rmody@brocade.com>
>Date: Sat, 7 Dec 2013 21:48:37 -0800
>
>> @@ -233,13 +233,13 @@ struct bnad_tx_unmap {  struct bnad_rx_vector {
>>  	DEFINE_DMA_UNMAP_ADDR(dma_addr);
>>  	u32			len;
>> -};
>> +} __packed;
>
>This is absolutely unacceptable, and at best not even mentioned in your
>commit message.
>
>There is no reason whatsoever to mark this structure as __packed.

By having the bnad_rx_unmap structure marked as __packed we were trying to
reduce the unmap_q size by 16K, __packed would ensure the RX unmap
structure of 40 bytes is packed to 32 bytes. This change was also captured
as a part of the commit message of patch #6 under change details.

We'll address your concern and resubmit the patches.

Thanks,
--Rasesh

Re: [net-next 06/13] bna: UDP RX Processing Improvements

[David Miller]

From: Rasesh Mody <rmody@brocade.com>
Date: Tue, 10 Dec 2013 16:05:46 -0800

> By having the bnad_rx_unmap structure marked as __packed we were trying to
> reduce the unmap_q size by 16K, __packed would ensure the RX unmap
> structure of 40 bytes is packed to 32 bytes. This change was also captured
> as a part of the commit message of patch #6 under change details.

Do you have any idea what the costs are on certain architectures
for adding this attribute?

When you mark something __packed, the compiler can no longer assume
anything about the alignment of any instance of that structure, nor
nor any structure which is contained within that type either.

So the compiler will, for example, emit byte by byte operations
to load multi-byte quantities on architectures where unaligned
data accesses can trap.

You should only use __packed as the extreme, very last resort method
to achieve a certain goal.  Frankly I think it should never be used
except in cases where layout is dictated by a network protocol or
hardware descriptor _AND_ the layout cannot be achieved using portable
fixed sized types.

[net-next 07/13] bna: Fix Stale MTU used to Configure Multi-buffer RX

[Rasesh Mody]

Change Details:
 - Rx initialization uses MTU. ENET MTU set should be called before setting up
   Rx. Added checks to make sure that reinit is called only when required.

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bnad.c | 32 +++++++++++++++++++++++++-------
 drivers/net/ethernet/brocade/bna/bnad.h |  6 +-----
 2 files changed, 26 insertions(+), 12 deletions(-)

diff --git a/drivers/net/ethernet/brocade/bna/bnad.c b/drivers/net/ethernet/brocade/bna/bnad.c
index 849de5a..508f5fc 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.c
+++ b/drivers/net/ethernet/brocade/bna/bnad.c
@@ -2013,6 +2013,7 @@ err_return:
 static void
 bnad_init_rx_config(struct bnad *bnad, struct bna_rx_config *rx_config)
 {
+	memset(rx_config, 0, sizeof(*rx_config));
 	rx_config->rx_type = BNA_RX_T_REGULAR;
 	rx_config->num_paths = bnad->num_rxp_per_rx;
 	rx_config->coalescing_timeo = bnad->rx_coalescing_timeo;
@@ -2035,6 +2036,7 @@ bnad_init_rx_config(struct bnad *bnad, struct bna_rx_config *rx_config)
 	}
 
 	rx_config->frame_size = BNAD_FRAME_SIZE(bnad->netdev->mtu);
+	rx_config->q0_multi_buf = BNA_STATUS_T_DISABLED;
 
 	/* BNA_RXP_SINGLE - one data-buffer queue
 	 * BNA_RXP_SLR - one small-buffer and one large-buffer queues
@@ -2044,7 +2046,8 @@ bnad_init_rx_config(struct bnad *bnad, struct bna_rx_config *rx_config)
 	rx_config->rxp_type = BNA_RXP_SLR;
 
 	if (BNAD_PCI_DEV_IS_CAT2(bnad) &&
-	    bnad_enable_multi_buffer(rx_config->frame_size)) {
+			bnad_enable_multi_buffer &&
+			rx_config->frame_size > 4096) {
 		/* though size_routing_enable is set in SLR,
 		 * small packets may get routed to same rxq.
 		 * set buf_size to 2048 instead of PAGE_SIZE.
@@ -2092,8 +2095,16 @@ bnad_reinit_rx(struct bnad *bnad)
 	for (rx_id = 0; rx_id < bnad->num_rx; rx_id++) {
 		if (!bnad->rx_info[rx_id].rx)
 			continue;
-		count++;
 		bnad_destroy_rx(bnad, rx_id);
+	}
+
+	spin_lock_irqsave(&bnad->bna_lock, flags);
+	bna_enet_mtu_set(&bnad->bna.enet,
+			 BNAD_FRAME_SIZE(bnad->netdev->mtu), NULL);
+	spin_unlock_irqrestore(&bnad->bna_lock, flags);
+
+	for (rx_id = 0; rx_id < bnad->num_rx; rx_id++) {
+		count++;
 		current_err = bnad_setup_rx(bnad, rx_id);
 		if (current_err && !err) {
 			err = current_err;
@@ -3262,7 +3273,7 @@ bnad_change_mtu(struct net_device *netdev, int new_mtu)
 {
 	int err, mtu;
 	struct bnad *bnad = netdev_priv(netdev);
-	u32 rx_count = 0;
+	u32 rx_count = 0, frame, new_frame;
 
 	if (new_mtu + ETH_HLEN < ETH_ZLEN || new_mtu > BNAD_JUMBO_MTU)
 		return -EINVAL;
@@ -3272,15 +3283,22 @@ bnad_change_mtu(struct net_device *netdev, int new_mtu)
 	mtu = netdev->mtu;
 	netdev->mtu = new_mtu;
 
+	frame = BNAD_FRAME_SIZE(mtu);
+	new_frame = BNAD_FRAME_SIZE(new_mtu);
+
 	/* check if multi-buffer needs to be enabled */
-	if (BNAD_PCI_DEV_IS_CAT2(bnad) &&
-	    bnad_reinit_rx_needed(BNAD_FRAME_SIZE(new_mtu)))
-		rx_count = bnad_reinit_rx(bnad);
+	if (BNAD_PCI_DEV_IS_CAT2(bnad) && bnad_enable_multi_buffer &&
+	    netif_running(bnad->netdev)) {
+		/* only when transition is over 4K */
+		if ((frame <= 4096 && new_frame > 4096) ||
+		    (frame > 4096 && new_frame <= 4096))
+			rx_count = bnad_reinit_rx(bnad);
+	}
 
 	/* rx_count > 0 - new rx created
 	 *	- Linux set err = 0 and return
 	 */
-	err = bnad_mtu_set(bnad, new_mtu);
+	err = bnad_mtu_set(bnad, new_frame);
 	if (err)
 		err = -EBUSY;
 
diff --git a/drivers/net/ethernet/brocade/bna/bnad.h b/drivers/net/ethernet/brocade/bna/bnad.h
index d1dc930..f2b50d9 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.h
+++ b/drivers/net/ethernet/brocade/bna/bnad.h
@@ -263,11 +263,7 @@ struct bnad_rx_unmap_q {
 #define BNAD_PCI_DEV_IS_CAT2(_bnad) \
 	((_bnad)->pcidev->device == BFA_PCI_DEVICE_ID_CT2)
 
-#define bnad_enable_multi_buffer(_mtu) \
-		bnad_reinit_rx_needed((_mtu))
-
-#define bnad_reinit_rx_needed(_frame) \
-	(bnad_multi_buffer_rx && (_frame) > 4096)
+#define bnad_enable_multi_buffer bnad_multi_buffer_rx
 
 /* Bit mask values for bnad->cfg_flags */
 #define	BNAD_CF_DIM_ENABLED		0x01	/* DIM */
-- 
1.8.2.3

[net-next 08/13] bna: CQ Read Fix

[Rasesh Mody]

Valid bit check for completion needs read fence, so that it does not get
reordered with other loads.

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bnad.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/brocade/bna/bnad.c b/drivers/net/ethernet/brocade/bna/bnad.c
index 508f5fc..cd29069 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.c
+++ b/drivers/net/ethernet/brocade/bna/bnad.c
@@ -605,7 +605,11 @@ bnad_cq_process(struct bnad *bnad, struct bna_ccb *ccb, int budget)
 	cq = ccb->sw_q;
 	cmpl = &cq[ccb->producer_index];
 
-	while (cmpl->valid && (packets < budget)) {
+	while (packets < budget) {
+		if (!cmpl->valid)
+			break;
+		rmb();
+
 		BNA_UPDATE_PKT_CNT(pkt_rt, ntohs(cmpl->length));
 
 		if (bna_is_small_rxq(cmpl->rxq_id))
@@ -645,6 +649,7 @@ bnad_cq_process(struct bnad *bnad, struct bna_ccb *ccb, int budget)
 
 				if (!next_cmpl->valid)
 					break;
+				mb();
 
 				len = ntohs(next_cmpl->length);
 				flags = ntohl(next_cmpl->flags);
-- 
1.8.2.3

[net-next 09/13] bna: Add NULL Check Before Dereferencing TCB

[Rasesh Mody]

Currently we already check to see whether the BNAD_TXQ_TX_STARTED cleared.
But if the tcb structure which contains this flag is also already freed by that
time, we would dereference the NULL pointer. This patch is to check tcb for NULL
pointer, before dereferencing it.

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bnad.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/brocade/bna/bnad.c b/drivers/net/ethernet/brocade/bna/bnad.c
index cd29069..27bdd6c 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.c
+++ b/drivers/net/ethernet/brocade/bna/bnad.c
@@ -2947,21 +2947,21 @@ bnad_start_xmit(struct sk_buff *skb, struct net_device *netdev)
 	}
 
 	tcb = bnad->tx_info[0].tcb[txq_id];
-	q_depth = tcb->q_depth;
-	prod = tcb->producer_index;
-
-	unmap_q = tcb->unmap_q;
 
 	/*
 	 * Takes care of the Tx that is scheduled between clearing the flag
 	 * and the netif_tx_stop_all_queues() call.
 	 */
-	if (unlikely(!test_bit(BNAD_TXQ_TX_STARTED, &tcb->flags))) {
+	if (unlikely(!tcb || !test_bit(BNAD_TXQ_TX_STARTED, &tcb->flags))) {
 		dev_kfree_skb(skb);
 		BNAD_UPDATE_CTR(bnad, tx_skb_stopping);
 		return NETDEV_TX_OK;
 	}
 
+	q_depth = tcb->q_depth;
+	prod = tcb->producer_index;
+	unmap_q = tcb->unmap_q;
+
 	vectors = 1 + skb_shinfo(skb)->nr_frags;
 	wis = BNA_TXQ_WI_NEEDED(vectors);	/* 4 vectors per work item */
 
-- 
1.8.2.3

[net-next 10/13] bna: Handle the TX Setup Failures

[Rasesh Mody]

Change details:
 - When bnad_setup_tx() returns NULL, the error is NOT returned to the caller.
   The caller will incorrectly assume success. So Return ENOMEM when bna_tx_create()
   fails.
 - If bnad_tx_msix_register() fails, call bna_tx_destroy() to free tx & to NULL
   the bnad reference to tcb.

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bnad.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/brocade/bna/bnad.c b/drivers/net/ethernet/brocade/bna/bnad.c
index 27bdd6c..2db645a 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.c
+++ b/drivers/net/ethernet/brocade/bna/bnad.c
@@ -1987,8 +1987,10 @@ bnad_setup_tx(struct bnad *bnad, u32 tx_id)
 	tx = bna_tx_create(&bnad->bna, bnad, tx_config, &tx_cbfn, res_info,
 			tx_info);
 	spin_unlock_irqrestore(&bnad->bna_lock, flags);
-	if (!tx)
+	if (!tx) {
+		err = -ENOMEM;
 		goto err_return;
+	}
 	tx_info->tx = tx;
 
 	INIT_DELAYED_WORK(&tx_info->tx_cleanup_work,
@@ -1999,7 +2001,7 @@ bnad_setup_tx(struct bnad *bnad, u32 tx_id)
 		err = bnad_tx_msix_register(bnad, tx_info,
 			tx_id, bnad->num_txq_per_tx);
 		if (err)
-			goto err_return;
+			goto cleanup_tx;
 	}
 
 	spin_lock_irqsave(&bnad->bna_lock, flags);
@@ -2008,6 +2010,12 @@ bnad_setup_tx(struct bnad *bnad, u32 tx_id)
 
 	return 0;
 
+cleanup_tx:
+	spin_lock_irqsave(&bnad->bna_lock, flags);
+	bna_tx_destroy(tx_info->tx);
+	spin_unlock_irqrestore(&bnad->bna_lock, flags);
+	tx_info->tx = NULL;
+	tx_info->tx_id = 0;
 err_return:
 	bnad_tx_res_free(bnad, res_info);
 	return err;
-- 
1.8.2.3

[net-next 11/13] bna: Embed SKB Length in TX Vector

[Rasesh Mody]

 - Store the length of the skb buffer mapped along with the handle and use it
   while unmapping the buffer.

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bnad.c | 6 ++++--
 drivers/net/ethernet/brocade/bna/bnad.h | 1 +
 2 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/brocade/bna/bnad.c b/drivers/net/ethernet/brocade/bna/bnad.c
index 2db645a..5f95663 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.c
+++ b/drivers/net/ethernet/brocade/bna/bnad.c
@@ -146,7 +146,8 @@ bnad_tx_buff_unmap(struct bnad *bnad,
 
 		dma_unmap_page(&bnad->pcidev->dev,
 			dma_unmap_addr(&unmap->vectors[vector], dma_addr),
-			skb_shinfo(skb)->frags[nvecs].size, DMA_TO_DEVICE);
+			dma_unmap_len(&unmap->vectors[vector], dma_len),
+			DMA_TO_DEVICE);
 		dma_unmap_addr_set(&unmap->vectors[vector], dma_addr, 0);
 		nvecs--;
 	}
@@ -3034,7 +3035,7 @@ bnad_start_xmit(struct sk_buff *skb, struct net_device *netdev)
 
 	for (i = 0, vect_id = 0; i < vectors - 1; i++) {
 		const struct skb_frag_struct *frag = &skb_shinfo(skb)->frags[i];
-		u16		size = skb_frag_size(frag);
+		u32		size = skb_frag_size(frag);
 
 		if (unlikely(size == 0)) {
 			/* Undo the changes starting at tcb->producer_index */
@@ -3059,6 +3060,7 @@ bnad_start_xmit(struct sk_buff *skb, struct net_device *netdev)
 
 		dma_addr = skb_frag_dma_map(&bnad->pcidev->dev, frag,
 					    0, size, DMA_TO_DEVICE);
+		unmap->vectors[vect_id].dma_len = size;
 		BNA_SET_DMA_ADDR(dma_addr, &txqent->vector[vect_id].host_addr);
 		txqent->vector[vect_id].length = htons(size);
 		dma_unmap_addr_set(&unmap->vectors[vect_id], dma_addr,
diff --git a/drivers/net/ethernet/brocade/bna/bnad.h b/drivers/net/ethernet/brocade/bna/bnad.h
index f2b50d9..474a597 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.h
+++ b/drivers/net/ethernet/brocade/bna/bnad.h
@@ -222,6 +222,7 @@ struct bnad_rx_info {
 
 struct bnad_tx_vector {
 	DEFINE_DMA_UNMAP_ADDR(dma_addr);
+	DEFINE_DMA_UNMAP_LEN(dma_len);
 };
 
 struct bnad_tx_unmap {
-- 
1.8.2.3

[net-next 12/13] bna: Firmware Patch Simplification

[Rasesh Mody]

This patch includes change to enable firmware patch simplication feature.
This feature is targeted to address the requirement to have independent patch
release for firmware. Prior to the 3.2.3.0 firmware, releasing a patch fix for
firmware would require changes to bna driver, to use new firmware images.
However with these changes, if the new firmware is flashed on to the Adapter,
the driver will use the new firmware after checking the patch release byte in
the firmware version.

Update the f/w version to 3.2.3.0

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bfa_ioc.c | 618 +++++++++++++++++++++++++++--
 drivers/net/ethernet/brocade/bna/bfa_ioc.h |   1 +
 drivers/net/ethernet/brocade/bna/bfi.h     |  33 +-
 drivers/net/ethernet/brocade/bna/cna.h     |   4 +-
 4 files changed, 626 insertions(+), 30 deletions(-)

diff --git a/drivers/net/ethernet/brocade/bna/bfa_ioc.c b/drivers/net/ethernet/brocade/bna/bfa_ioc.c
index b81132a..537bba1 100644
--- a/drivers/net/ethernet/brocade/bna/bfa_ioc.c
+++ b/drivers/net/ethernet/brocade/bna/bfa_ioc.c
@@ -22,6 +22,14 @@
 
 /* IOC local definitions */
 
+#define bfa_ioc_state_disabled(__sm)			\
+	(((__sm) == BFI_IOC_UNINIT) ||			\
+	 ((__sm) == BFI_IOC_INITING) ||			\
+	 ((__sm) == BFI_IOC_HWINIT) ||			\
+	 ((__sm) == BFI_IOC_DISABLED) ||		\
+	 ((__sm) == BFI_IOC_FAIL) ||			\
+	 ((__sm) == BFI_IOC_CFG_DISABLED))
+
 /* Asic specific macros : see bfa_hw_cb.c and bfa_hw_ct.c for details. */
 
 #define bfa_ioc_firmware_lock(__ioc)			\
@@ -84,8 +92,8 @@ static void bfa_ioc_pf_disabled(struct bfa_ioc *ioc);
 static void bfa_ioc_pf_failed(struct bfa_ioc *ioc);
 static void bfa_ioc_pf_hwfailed(struct bfa_ioc *ioc);
 static void bfa_ioc_pf_fwmismatch(struct bfa_ioc *ioc);
-static void bfa_ioc_boot(struct bfa_ioc *ioc, enum bfi_fwboot_type boot_type,
-			 u32 boot_param);
+static enum bfa_status bfa_ioc_boot(struct bfa_ioc *ioc,
+			enum bfi_fwboot_type boot_type, u32 boot_param);
 static u32 bfa_ioc_smem_pgnum(struct bfa_ioc *ioc, u32 fmaddr);
 static void bfa_ioc_get_adapter_serial_num(struct bfa_ioc *ioc,
 						char *serial_num);
@@ -1139,6 +1147,25 @@ bfa_nw_ioc_sem_release(void __iomem *sem_reg)
 	writel(1, sem_reg);
 }
 
+/* Invalidate fwver signature */
+enum bfa_status
+bfa_nw_ioc_fwsig_invalidate(struct bfa_ioc *ioc)
+{
+	u32	pgnum, pgoff;
+	u32	loff = 0;
+	enum bfi_ioc_state ioc_fwstate;
+
+	ioc_fwstate = bfa_ioc_get_cur_ioc_fwstate(ioc);
+	if (!bfa_ioc_state_disabled(ioc_fwstate))
+		return BFA_STATUS_ADAPTER_ENABLED;
+
+	pgnum = bfa_ioc_smem_pgnum(ioc, loff);
+	pgoff = PSS_SMEM_PGOFF(loff);
+	writel(pgnum, ioc->ioc_regs.host_page_num_fn);
+	writel(BFI_IOC_FW_INV_SIGN, ioc->ioc_regs.smem_page_start + loff);
+	return BFA_STATUS_OK;
+}
+
 /* Clear fwver hdr */
 static void
 bfa_ioc_fwver_clear(struct bfa_ioc *ioc)
@@ -1317,22 +1344,510 @@ bfa_nw_ioc_fwver_get(struct bfa_ioc *ioc, struct bfi_ioc_image_hdr *fwhdr)
 	}
 }
 
-/* Returns TRUE if same. */
+static bool
+bfa_ioc_fwver_md5_check(struct bfi_ioc_image_hdr *fwhdr_1,
+			struct bfi_ioc_image_hdr *fwhdr_2)
+{
+	int i;
+
+	for (i = 0; i < BFI_IOC_MD5SUM_SZ; i++) {
+		if (fwhdr_1->md5sum[i] != fwhdr_2->md5sum[i])
+			return false;
+	}
+
+	return true;
+}
+
+/* Returns TRUE if major minor and maintainence are same.
+ * If patch version are same, check for MD5 Checksum to be same.
+ */
+static bool
+bfa_ioc_fw_ver_compatible(struct bfi_ioc_image_hdr *drv_fwhdr,
+			  struct bfi_ioc_image_hdr *fwhdr_to_cmp)
+{
+	if (drv_fwhdr->signature != fwhdr_to_cmp->signature)
+		return false;
+	if (drv_fwhdr->fwver.major != fwhdr_to_cmp->fwver.major)
+		return false;
+	if (drv_fwhdr->fwver.minor != fwhdr_to_cmp->fwver.minor)
+		return false;
+	if (drv_fwhdr->fwver.maint != fwhdr_to_cmp->fwver.maint)
+		return false;
+	if (drv_fwhdr->fwver.patch == fwhdr_to_cmp->fwver.patch &&
+	    drv_fwhdr->fwver.phase == fwhdr_to_cmp->fwver.phase &&
+	    drv_fwhdr->fwver.build == fwhdr_to_cmp->fwver.build)
+		return bfa_ioc_fwver_md5_check(drv_fwhdr, fwhdr_to_cmp);
+
+	return true;
+}
+
+static bool
+bfa_ioc_flash_fwver_valid(struct bfi_ioc_image_hdr *flash_fwhdr)
+{
+	if (flash_fwhdr->fwver.major == 0 || flash_fwhdr->fwver.major == 0xFF)
+		return false;
+
+	return true;
+}
+
+static bool
+fwhdr_is_ga(struct bfi_ioc_image_hdr *fwhdr)
+{
+	if (fwhdr->fwver.phase == 0 &&
+	    fwhdr->fwver.build == 0)
+		return false;
+
+	return true;
+}
+
+/* Returns TRUE if both are compatible and patch of fwhdr_to_cmp is better. */
+static enum bfi_ioc_img_ver_cmp
+bfa_ioc_fw_ver_patch_cmp(struct bfi_ioc_image_hdr *base_fwhdr,
+			 struct bfi_ioc_image_hdr *fwhdr_to_cmp)
+{
+	if (bfa_ioc_fw_ver_compatible(base_fwhdr, fwhdr_to_cmp) == false)
+		return BFI_IOC_IMG_VER_INCOMP;
+
+	if (fwhdr_to_cmp->fwver.patch > base_fwhdr->fwver.patch)
+		return BFI_IOC_IMG_VER_BETTER;
+	else if (fwhdr_to_cmp->fwver.patch < base_fwhdr->fwver.patch)
+		return BFI_IOC_IMG_VER_OLD;
+
+	/* GA takes priority over internal builds of the same patch stream.
+	 * At this point major minor maint and patch numbers are same.
+	 */
+	if (fwhdr_is_ga(base_fwhdr) == true)
+		if (fwhdr_is_ga(fwhdr_to_cmp))
+			return BFI_IOC_IMG_VER_SAME;
+		else
+			return BFI_IOC_IMG_VER_OLD;
+	else
+		if (fwhdr_is_ga(fwhdr_to_cmp))
+			return BFI_IOC_IMG_VER_BETTER;
+
+	if (fwhdr_to_cmp->fwver.phase > base_fwhdr->fwver.phase)
+		return BFI_IOC_IMG_VER_BETTER;
+	else if (fwhdr_to_cmp->fwver.phase < base_fwhdr->fwver.phase)
+		return BFI_IOC_IMG_VER_OLD;
+
+	if (fwhdr_to_cmp->fwver.build > base_fwhdr->fwver.build)
+		return BFI_IOC_IMG_VER_BETTER;
+	else if (fwhdr_to_cmp->fwver.build < base_fwhdr->fwver.build)
+		return BFI_IOC_IMG_VER_OLD;
+
+	/* All Version Numbers are equal.
+	 * Md5 check to be done as a part of compatibility check.
+	 */
+	return BFI_IOC_IMG_VER_SAME;
+}
+
+/* register definitions */
+#define FLI_CMD_REG			0x0001d000
+#define FLI_WRDATA_REG			0x0001d00c
+#define FLI_RDDATA_REG			0x0001d010
+#define FLI_ADDR_REG			0x0001d004
+#define FLI_DEV_STATUS_REG		0x0001d014
+
+#define BFA_FLASH_FIFO_SIZE		128	/* fifo size */
+#define BFA_FLASH_CHECK_MAX		10000	/* max # of status check */
+#define BFA_FLASH_BLOCKING_OP_MAX	1000000	/* max # of blocking op check */
+#define BFA_FLASH_WIP_MASK		0x01	/* write in progress bit mask */
+
+#define NFC_STATE_RUNNING		0x20000001
+#define NFC_STATE_PAUSED		0x00004560
+#define NFC_VER_VALID			0x147
+
+enum bfa_flash_cmd {
+	BFA_FLASH_FAST_READ	= 0x0b,	/* fast read */
+	BFA_FLASH_WRITE_ENABLE	= 0x06,	/* write enable */
+	BFA_FLASH_SECTOR_ERASE	= 0xd8,	/* sector erase */
+	BFA_FLASH_WRITE		= 0x02,	/* write */
+	BFA_FLASH_READ_STATUS	= 0x05,	/* read status */
+};
+
+/* hardware error definition */
+enum bfa_flash_err {
+	BFA_FLASH_NOT_PRESENT	= -1,	/*!< flash not present */
+	BFA_FLASH_UNINIT	= -2,	/*!< flash not initialized */
+	BFA_FLASH_BAD		= -3,	/*!< flash bad */
+	BFA_FLASH_BUSY		= -4,	/*!< flash busy */
+	BFA_FLASH_ERR_CMD_ACT	= -5,	/*!< command active never cleared */
+	BFA_FLASH_ERR_FIFO_CNT	= -6,	/*!< fifo count never cleared */
+	BFA_FLASH_ERR_WIP	= -7,	/*!< write-in-progress never cleared */
+	BFA_FLASH_ERR_TIMEOUT	= -8,	/*!< fli timeout */
+	BFA_FLASH_ERR_LEN	= -9,	/*!< invalid length */
+};
+
+/* flash command register data structure */
+union bfa_flash_cmd_reg {
+	struct {
+#ifdef __BIG_ENDIAN
+		u32	act:1;
+		u32	rsv:1;
+		u32	write_cnt:9;
+		u32	read_cnt:9;
+		u32	addr_cnt:4;
+		u32	cmd:8;
+#else
+		u32	cmd:8;
+		u32	addr_cnt:4;
+		u32	read_cnt:9;
+		u32	write_cnt:9;
+		u32	rsv:1;
+		u32	act:1;
+#endif
+	} r;
+	u32	i;
+};
+
+/* flash device status register data structure */
+union bfa_flash_dev_status_reg {
+	struct {
+#ifdef __BIG_ENDIAN
+		u32	rsv:21;
+		u32	fifo_cnt:6;
+		u32	busy:1;
+		u32	init_status:1;
+		u32	present:1;
+		u32	bad:1;
+		u32	good:1;
+#else
+		u32	good:1;
+		u32	bad:1;
+		u32	present:1;
+		u32	init_status:1;
+		u32	busy:1;
+		u32	fifo_cnt:6;
+		u32	rsv:21;
+#endif
+	} r;
+	u32	i;
+};
+
+/* flash address register data structure */
+union bfa_flash_addr_reg {
+	struct {
+#ifdef __BIG_ENDIAN
+		u32	addr:24;
+		u32	dummy:8;
+#else
+		u32	dummy:8;
+		u32	addr:24;
+#endif
+	} r;
+	u32	i;
+};
+
+/* Flash raw private functions */
+static void
+bfa_flash_set_cmd(void __iomem *pci_bar, u8 wr_cnt,
+		  u8 rd_cnt, u8 ad_cnt, u8 op)
+{
+	union bfa_flash_cmd_reg cmd;
+
+	cmd.i = 0;
+	cmd.r.act = 1;
+	cmd.r.write_cnt = wr_cnt;
+	cmd.r.read_cnt = rd_cnt;
+	cmd.r.addr_cnt = ad_cnt;
+	cmd.r.cmd = op;
+	writel(cmd.i, (pci_bar + FLI_CMD_REG));
+}
+
+static void
+bfa_flash_set_addr(void __iomem *pci_bar, u32 address)
+{
+	union bfa_flash_addr_reg addr;
+
+	addr.r.addr = address & 0x00ffffff;
+	addr.r.dummy = 0;
+	writel(addr.i, (pci_bar + FLI_ADDR_REG));
+}
+
+static int
+bfa_flash_cmd_act_check(void __iomem *pci_bar)
+{
+	union bfa_flash_cmd_reg cmd;
+
+	cmd.i = readl(pci_bar + FLI_CMD_REG);
+
+	if (cmd.r.act)
+		return BFA_FLASH_ERR_CMD_ACT;
+
+	return 0;
+}
+
+/* Flush FLI data fifo. */
+static u32
+bfa_flash_fifo_flush(void __iomem *pci_bar)
+{
+	u32 i;
+	u32 t;
+	union bfa_flash_dev_status_reg dev_status;
+
+	dev_status.i = readl(pci_bar + FLI_DEV_STATUS_REG);
+
+	if (!dev_status.r.fifo_cnt)
+		return 0;
+
+	/* fifo counter in terms of words */
+	for (i = 0; i < dev_status.r.fifo_cnt; i++)
+		t = readl(pci_bar + FLI_RDDATA_REG);
+
+	/* Check the device status. It may take some time. */
+	for (i = 0; i < BFA_FLASH_CHECK_MAX; i++) {
+		dev_status.i = readl(pci_bar + FLI_DEV_STATUS_REG);
+		if (!dev_status.r.fifo_cnt)
+			break;
+	}
+
+	if (dev_status.r.fifo_cnt)
+		return BFA_FLASH_ERR_FIFO_CNT;
+
+	return 0;
+}
+
+/* Read flash status. */
+static u32
+bfa_flash_status_read(void __iomem *pci_bar)
+{
+	union bfa_flash_dev_status_reg	dev_status;
+	u32				status;
+	u32			ret_status;
+	int				i;
+
+	status = bfa_flash_fifo_flush(pci_bar);
+	if (status < 0)
+		return status;
+
+	bfa_flash_set_cmd(pci_bar, 0, 4, 0, BFA_FLASH_READ_STATUS);
+
+	for (i = 0; i < BFA_FLASH_CHECK_MAX; i++) {
+		status = bfa_flash_cmd_act_check(pci_bar);
+		if (!status)
+			break;
+	}
+
+	if (status)
+		return status;
+
+	dev_status.i = readl(pci_bar + FLI_DEV_STATUS_REG);
+	if (!dev_status.r.fifo_cnt)
+		return BFA_FLASH_BUSY;
+
+	ret_status = readl(pci_bar + FLI_RDDATA_REG);
+	ret_status >>= 24;
+
+	status = bfa_flash_fifo_flush(pci_bar);
+	if (status < 0)
+		return status;
+
+	return ret_status;
+}
+
+/* Start flash read operation. */
+static u32
+bfa_flash_read_start(void __iomem *pci_bar, u32 offset, u32 len,
+		     char *buf)
+{
+	u32 status;
+
+	/* len must be mutiple of 4 and not exceeding fifo size */
+	if (len == 0 || len > BFA_FLASH_FIFO_SIZE || (len & 0x03) != 0)
+		return BFA_FLASH_ERR_LEN;
+
+	/* check status */
+	status = bfa_flash_status_read(pci_bar);
+	if (status == BFA_FLASH_BUSY)
+		status = bfa_flash_status_read(pci_bar);
+
+	if (status < 0)
+		return status;
+
+	/* check if write-in-progress bit is cleared */
+	if (status & BFA_FLASH_WIP_MASK)
+		return BFA_FLASH_ERR_WIP;
+
+	bfa_flash_set_addr(pci_bar, offset);
+
+	bfa_flash_set_cmd(pci_bar, 0, (u8)len, 4, BFA_FLASH_FAST_READ);
+
+	return 0;
+}
+
+/* Check flash read operation. */
+static u32
+bfa_flash_read_check(void __iomem *pci_bar)
+{
+	if (bfa_flash_cmd_act_check(pci_bar))
+		return 1;
+
+	return 0;
+}
+
+/* End flash read operation. */
+static void
+bfa_flash_read_end(void __iomem *pci_bar, u32 len, char *buf)
+{
+	u32 i;
+
+	/* read data fifo up to 32 words */
+	for (i = 0; i < len; i += 4) {
+		u32 w = readl(pci_bar + FLI_RDDATA_REG);
+		*((u32 *)(buf + i)) = swab32(w);
+	}
+
+	bfa_flash_fifo_flush(pci_bar);
+}
+
+/* Perform flash raw read. */
+
+#define FLASH_BLOCKING_OP_MAX   500
+#define FLASH_SEM_LOCK_REG	0x18820
+
+static int
+bfa_raw_sem_get(void __iomem *bar)
+{
+	int	locked;
+
+	locked = readl((bar + FLASH_SEM_LOCK_REG));
+
+	return !locked;
+}
+
+static enum bfa_status
+bfa_flash_sem_get(void __iomem *bar)
+{
+	u32 n = FLASH_BLOCKING_OP_MAX;
+
+	while (!bfa_raw_sem_get(bar)) {
+		if (--n <= 0)
+			return BFA_STATUS_BADFLASH;
+		udelay(10000);
+	}
+	return BFA_STATUS_OK;
+}
+
+static void
+bfa_flash_sem_put(void __iomem *bar)
+{
+	writel(0, (bar + FLASH_SEM_LOCK_REG));
+}
+
+static enum bfa_status
+bfa_flash_raw_read(void __iomem *pci_bar, u32 offset, char *buf,
+		   u32 len)
+{
+	u32 n, status;
+	u32 off, l, s, residue, fifo_sz;
+
+	residue = len;
+	off = 0;
+	fifo_sz = BFA_FLASH_FIFO_SIZE;
+	status = bfa_flash_sem_get(pci_bar);
+	if (status != BFA_STATUS_OK)
+		return status;
+
+	while (residue) {
+		s = offset + off;
+		n = s / fifo_sz;
+		l = (n + 1) * fifo_sz - s;
+		if (l > residue)
+			l = residue;
+
+		status = bfa_flash_read_start(pci_bar, offset + off, l,
+								&buf[off]);
+		if (status < 0) {
+			bfa_flash_sem_put(pci_bar);
+			return BFA_STATUS_FAILED;
+		}
+
+		n = BFA_FLASH_BLOCKING_OP_MAX;
+		while (bfa_flash_read_check(pci_bar)) {
+			if (--n <= 0) {
+				bfa_flash_sem_put(pci_bar);
+				return BFA_STATUS_FAILED;
+			}
+		}
+
+		bfa_flash_read_end(pci_bar, l, &buf[off]);
+
+		residue -= l;
+		off += l;
+	}
+	bfa_flash_sem_put(pci_bar);
+
+	return BFA_STATUS_OK;
+}
+
+u32
+bfa_nw_ioc_flash_img_get_size(struct bfa_ioc *ioc)
+{
+	return BFI_FLASH_IMAGE_SZ/sizeof(u32);
+}
+
+#define BFA_FLASH_PART_FWIMG_ADDR	0x100000 /* fw image address */
+
+enum bfa_status
+bfa_nw_ioc_flash_img_get_chnk(struct bfa_ioc *ioc, u32 off,
+			      u32 *fwimg)
+{
+	return bfa_flash_raw_read(ioc->pcidev.pci_bar_kva,
+			BFA_FLASH_PART_FWIMG_ADDR + (off * sizeof(u32)),
+			(char *)fwimg, BFI_FLASH_CHUNK_SZ);
+}
+
+static enum bfi_ioc_img_ver_cmp
+bfa_ioc_flash_fwver_cmp(struct bfa_ioc *ioc,
+			struct bfi_ioc_image_hdr *base_fwhdr)
+{
+	struct bfi_ioc_image_hdr *flash_fwhdr;
+	enum bfa_status status;
+	u32 fwimg[BFI_FLASH_CHUNK_SZ_WORDS];
+
+	status = bfa_nw_ioc_flash_img_get_chnk(ioc, 0, fwimg);
+	if (status != BFA_STATUS_OK)
+		return BFI_IOC_IMG_VER_INCOMP;
+
+	flash_fwhdr = (struct bfi_ioc_image_hdr *)fwimg;
+	if (bfa_ioc_flash_fwver_valid(flash_fwhdr))
+		return bfa_ioc_fw_ver_patch_cmp(base_fwhdr, flash_fwhdr);
+	else
+		return BFI_IOC_IMG_VER_INCOMP;
+}
+
+/**
+ * Returns TRUE if driver is willing to work with current smem f/w version.
+ */
 bool
 bfa_nw_ioc_fwver_cmp(struct bfa_ioc *ioc, struct bfi_ioc_image_hdr *fwhdr)
 {
 	struct bfi_ioc_image_hdr *drv_fwhdr;
-	int i;
+	enum bfi_ioc_img_ver_cmp smem_flash_cmp, drv_smem_cmp;
 
 	drv_fwhdr = (struct bfi_ioc_image_hdr *)
 		bfa_cb_image_get_chunk(bfa_ioc_asic_gen(ioc), 0);
 
-	for (i = 0; i < BFI_IOC_MD5SUM_SZ; i++) {
-		if (fwhdr->md5sum[i] != drv_fwhdr->md5sum[i])
-			return false;
+	/* If smem is incompatible or old, driver should not work with it. */
+	drv_smem_cmp = bfa_ioc_fw_ver_patch_cmp(drv_fwhdr, fwhdr);
+	if (drv_smem_cmp == BFI_IOC_IMG_VER_INCOMP ||
+	    drv_smem_cmp == BFI_IOC_IMG_VER_OLD) {
+		return false;
 	}
 
-	return true;
+	/* IF Flash has a better F/W than smem do not work with smem.
+	 * If smem f/w == flash f/w, as smem f/w not old | incmp, work with it.
+	 * If Flash is old or incomp work with smem iff smem f/w == drv f/w.
+	 */
+	smem_flash_cmp = bfa_ioc_flash_fwver_cmp(ioc, fwhdr);
+
+	if (smem_flash_cmp == BFI_IOC_IMG_VER_BETTER)
+		return false;
+	else if (smem_flash_cmp == BFI_IOC_IMG_VER_SAME)
+		return true;
+	else
+		return (drv_smem_cmp == BFI_IOC_IMG_VER_SAME) ?
+			true : false;
 }
 
 /* Return true if current running version is valid. Firmware signature and
@@ -1341,15 +1856,9 @@ bfa_nw_ioc_fwver_cmp(struct bfa_ioc *ioc, struct bfi_ioc_image_hdr *fwhdr)
 static bool
 bfa_ioc_fwver_valid(struct bfa_ioc *ioc, u32 boot_env)
 {
-	struct bfi_ioc_image_hdr fwhdr, *drv_fwhdr;
+	struct bfi_ioc_image_hdr fwhdr;
 
 	bfa_nw_ioc_fwver_get(ioc, &fwhdr);
-	drv_fwhdr = (struct bfi_ioc_image_hdr *)
-		bfa_cb_image_get_chunk(bfa_ioc_asic_gen(ioc), 0);
-
-	if (fwhdr.signature != drv_fwhdr->signature)
-		return false;
-
 	if (swab32(fwhdr.bootenv) != boot_env)
 		return false;
 
@@ -1388,8 +1897,10 @@ bfa_ioc_hwinit(struct bfa_ioc *ioc, bool force)
 		false : bfa_ioc_fwver_valid(ioc, boot_env);
 
 	if (!fwvalid) {
-		bfa_ioc_boot(ioc, BFI_FWBOOT_TYPE_NORMAL, boot_env);
-		bfa_ioc_poll_fwinit(ioc);
+		if (bfa_ioc_boot(ioc, BFI_FWBOOT_TYPE_NORMAL, boot_env) ==
+								BFA_STATUS_OK)
+			bfa_ioc_poll_fwinit(ioc);
+
 		return;
 	}
 
@@ -1419,8 +1930,9 @@ bfa_ioc_hwinit(struct bfa_ioc *ioc, bool force)
 	/**
 	 * Initialize the h/w for any other states.
 	 */
-	bfa_ioc_boot(ioc, BFI_FWBOOT_TYPE_NORMAL, boot_env);
-	bfa_ioc_poll_fwinit(ioc);
+	if (bfa_ioc_boot(ioc, BFI_FWBOOT_TYPE_NORMAL, boot_env) ==
+							BFA_STATUS_OK)
+		bfa_ioc_poll_fwinit(ioc);
 }
 
 void
@@ -1525,7 +2037,7 @@ bfa_ioc_hb_stop(struct bfa_ioc *ioc)
 }
 
 /* Initiate a full firmware download. */
-static void
+static enum bfa_status
 bfa_ioc_download_fw(struct bfa_ioc *ioc, u32 boot_type,
 		    u32 boot_env)
 {
@@ -1535,18 +2047,47 @@ bfa_ioc_download_fw(struct bfa_ioc *ioc, u32 boot_type,
 	u32 chunkno = 0;
 	u32 i;
 	u32 asicmode;
+	u32 fwimg_size;
+	u32 fwimg_buf[BFI_FLASH_CHUNK_SZ_WORDS];
+	enum bfa_status status;
+
+	if (boot_env == BFI_FWBOOT_ENV_OS &&
+	    boot_type == BFI_FWBOOT_TYPE_FLASH) {
+		fwimg_size = BFI_FLASH_IMAGE_SZ/sizeof(u32);
+
+		status = bfa_nw_ioc_flash_img_get_chnk(ioc,
+			BFA_IOC_FLASH_CHUNK_ADDR(chunkno), fwimg_buf);
+		if (status != BFA_STATUS_OK)
+			return status;
 
-	fwimg = bfa_cb_image_get_chunk(bfa_ioc_asic_gen(ioc), chunkno);
+		fwimg = fwimg_buf;
+	} else {
+		fwimg_size = bfa_cb_image_get_size(bfa_ioc_asic_gen(ioc));
+		fwimg = bfa_cb_image_get_chunk(bfa_ioc_asic_gen(ioc),
+					BFA_IOC_FLASH_CHUNK_ADDR(chunkno));
+	}
 
 	pgnum = bfa_ioc_smem_pgnum(ioc, loff);
 
 	writel(pgnum, ioc->ioc_regs.host_page_num_fn);
 
-	for (i = 0; i < bfa_cb_image_get_size(bfa_ioc_asic_gen(ioc)); i++) {
+	for (i = 0; i < fwimg_size; i++) {
 		if (BFA_IOC_FLASH_CHUNK_NO(i) != chunkno) {
 			chunkno = BFA_IOC_FLASH_CHUNK_NO(i);
-			fwimg = bfa_cb_image_get_chunk(bfa_ioc_asic_gen(ioc),
+			if (boot_env == BFI_FWBOOT_ENV_OS &&
+			    boot_type == BFI_FWBOOT_TYPE_FLASH) {
+				status = bfa_nw_ioc_flash_img_get_chnk(ioc,
+					BFA_IOC_FLASH_CHUNK_ADDR(chunkno),
+					fwimg_buf);
+				if (status != BFA_STATUS_OK)
+					return status;
+
+				fwimg = fwimg_buf;
+			} else {
+				fwimg = bfa_cb_image_get_chunk(
+					bfa_ioc_asic_gen(ioc),
 					BFA_IOC_FLASH_CHUNK_ADDR(chunkno));
+			}
 		}
 
 		/**
@@ -1574,6 +2115,10 @@ bfa_ioc_download_fw(struct bfa_ioc *ioc, u32 boot_type,
 	/*
 	 * Set boot type, env and device mode at the end.
 	*/
+	if (boot_env == BFI_FWBOOT_ENV_OS &&
+	    boot_type == BFI_FWBOOT_TYPE_FLASH) {
+		boot_type = BFI_FWBOOT_TYPE_NORMAL;
+	}
 	asicmode = BFI_FWBOOT_DEVMODE(ioc->asic_gen, ioc->asic_mode,
 					ioc->port0_mode, ioc->port1_mode);
 	writel(asicmode, ((ioc->ioc_regs.smem_page_start)
@@ -1582,6 +2127,7 @@ bfa_ioc_download_fw(struct bfa_ioc *ioc, u32 boot_type,
 			+ (BFI_FWBOOT_TYPE_OFF)));
 	writel(boot_env, ((ioc->ioc_regs.smem_page_start)
 			+ (BFI_FWBOOT_ENV_OFF)));
+	return BFA_STATUS_OK;
 }
 
 static void
@@ -1854,14 +2400,27 @@ bfa_ioc_pll_init(struct bfa_ioc *ioc)
 /* Interface used by diag module to do firmware boot with memory test
  * as the entry vector.
  */
-static void
+static enum bfa_status
 bfa_ioc_boot(struct bfa_ioc *ioc, enum bfi_fwboot_type boot_type,
 		u32 boot_env)
 {
+	struct bfi_ioc_image_hdr *drv_fwhdr;
+	enum bfa_status status;
 	bfa_ioc_stats(ioc, ioc_boots);
 
 	if (bfa_ioc_pll_init(ioc) != BFA_STATUS_OK)
-		return;
+		return BFA_STATUS_FAILED;
+	if (boot_env == BFI_FWBOOT_ENV_OS &&
+	    boot_type == BFI_FWBOOT_TYPE_NORMAL) {
+		drv_fwhdr = (struct bfi_ioc_image_hdr *)
+			bfa_cb_image_get_chunk(bfa_ioc_asic_gen(ioc), 0);
+		/* Work with Flash iff flash f/w is better than driver f/w.
+		 * Otherwise push drivers firmware.
+		 */
+		if (bfa_ioc_flash_fwver_cmp(ioc, drv_fwhdr) ==
+			BFI_IOC_IMG_VER_BETTER)
+			boot_type = BFI_FWBOOT_TYPE_FLASH;
+	}
 
 	/**
 	 * Initialize IOC state of all functions on a chip reset.
@@ -1875,8 +2434,13 @@ bfa_ioc_boot(struct bfa_ioc *ioc, enum bfi_fwboot_type boot_type,
 	}
 
 	bfa_ioc_msgflush(ioc);
-	bfa_ioc_download_fw(ioc, boot_type, boot_env);
-	bfa_ioc_lpu_start(ioc);
+	status = bfa_ioc_download_fw(ioc, boot_type, boot_env);
+	if (status == BFA_STATUS_OK)
+		bfa_ioc_lpu_start(ioc);
+	else
+		bfa_nw_iocpf_timeout(ioc);
+
+	return status;
 }
 
 /* Enable/disable IOC failure auto recovery. */
diff --git a/drivers/net/ethernet/brocade/bna/bfa_ioc.h b/drivers/net/ethernet/brocade/bna/bfa_ioc.h
index 9521cbf..20cff7d 100644
--- a/drivers/net/ethernet/brocade/bna/bfa_ioc.h
+++ b/drivers/net/ethernet/brocade/bna/bfa_ioc.h
@@ -298,6 +298,7 @@ void bfa_nw_ioc_error_isr(struct bfa_ioc *ioc);
 bool bfa_nw_ioc_is_disabled(struct bfa_ioc *ioc);
 bool bfa_nw_ioc_is_operational(struct bfa_ioc *ioc);
 void bfa_nw_ioc_get_attr(struct bfa_ioc *ioc, struct bfa_ioc_attr *ioc_attr);
+enum bfa_status bfa_nw_ioc_fwsig_invalidate(struct bfa_ioc *ioc);
 void bfa_nw_ioc_notify_register(struct bfa_ioc *ioc,
 	struct bfa_ioc_notify *notify);
 bool bfa_nw_ioc_sem_get(void __iomem *sem_reg);
diff --git a/drivers/net/ethernet/brocade/bna/bfi.h b/drivers/net/ethernet/brocade/bna/bfi.h
index 1f24c23..8c563a7 100644
--- a/drivers/net/ethernet/brocade/bna/bfi.h
+++ b/drivers/net/ethernet/brocade/bna/bfi.h
@@ -25,6 +25,7 @@
 /* BFI FW image type */
 #define	BFI_FLASH_CHUNK_SZ			256	/*!< Flash chunk size */
 #define	BFI_FLASH_CHUNK_SZ_WORDS	(BFI_FLASH_CHUNK_SZ/sizeof(u32))
+#define BFI_FLASH_IMAGE_SZ		0x100000
 
 /* Msg header common to all msgs */
 struct bfi_mhdr {
@@ -233,7 +234,29 @@ struct bfi_ioc_getattr_reply {
 #define BFI_IOC_TRC_HDR_SZ	32
 
 #define BFI_IOC_FW_SIGNATURE	(0xbfadbfad)
+#define BFI_IOC_FW_INV_SIGN	(0xdeaddead)
 #define BFI_IOC_MD5SUM_SZ	4
+
+struct bfi_ioc_fwver {
+#ifdef __BIG_ENDIAN
+	u8 patch;
+	u8 maint;
+	u8 minor;
+	u8 major;
+	u8 rsvd[2];
+	u8 build;
+	u8 phase;
+#else
+	u8 major;
+	u8 minor;
+	u8 maint;
+	u8 patch;
+	u8 phase;
+	u8 build;
+	u8 rsvd[2];
+#endif
+};
+
 struct bfi_ioc_image_hdr {
 	u32	signature;	/*!< constant signature */
 	u8	asic_gen;	/*!< asic generation */
@@ -242,10 +265,18 @@ struct bfi_ioc_image_hdr {
 	u8	port1_mode;	/*!< device mode for port 1 */
 	u32	exec;		/*!< exec vector	*/
 	u32	bootenv;	/*!< firmware boot env */
-	u32	rsvd_b[4];
+	u32	rsvd_b[2];
+	struct bfi_ioc_fwver fwver;
 	u32	md5sum[BFI_IOC_MD5SUM_SZ];
 };
 
+enum bfi_ioc_img_ver_cmp {
+	BFI_IOC_IMG_VER_INCOMP,
+	BFI_IOC_IMG_VER_OLD,
+	BFI_IOC_IMG_VER_SAME,
+	BFI_IOC_IMG_VER_BETTER
+};
+
 #define BFI_FWBOOT_DEVMODE_OFF		4
 #define BFI_FWBOOT_TYPE_OFF		8
 #define BFI_FWBOOT_ENV_OFF		12
diff --git a/drivers/net/ethernet/brocade/bna/cna.h b/drivers/net/ethernet/brocade/bna/cna.h
index 43405f6..b3ff6d5 100644
--- a/drivers/net/ethernet/brocade/bna/cna.h
+++ b/drivers/net/ethernet/brocade/bna/cna.h
@@ -37,8 +37,8 @@
 
 extern char bfa_version[];
 
-#define CNA_FW_FILE_CT	"ctfw-3.2.1.1.bin"
-#define CNA_FW_FILE_CT2	"ct2fw-3.2.1.1.bin"
+#define CNA_FW_FILE_CT	"ctfw-3.2.3.0.bin"
+#define CNA_FW_FILE_CT2	"ct2fw-3.2.3.0.bin"
 #define FC_SYMNAME_MAX	256	/*!< max name server symbolic name size */
 
 #pragma pack(1)
-- 
1.8.2.3

[net-next 13/13] bna: Update the Driver Version to 3.2.23.0

[Rasesh Mody]

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bnad.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/brocade/bna/bnad.h b/drivers/net/ethernet/brocade/bna/bnad.h
index 474a597..ea8f1d6 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.h
+++ b/drivers/net/ethernet/brocade/bna/bnad.h
@@ -71,7 +71,7 @@ struct bnad_rx_ctrl {
 #define BNAD_NAME			"bna"
 #define BNAD_NAME_LEN			64
 
-#define BNAD_VERSION			"3.2.21.1"
+#define BNAD_VERSION			"3.2.23.0"
 
 #define BNAD_MAILBOX_MSIX_INDEX		0
 #define BNAD_MAILBOX_MSIX_VECTORS	1
-- 
1.8.2.3

[net-next 00/13] bna: Update the Driver to v3.2.23.0

[Rasesh Mody]

Hi Dave,

This patch set consists of feature additions like s/w ptp support, multi-buffer
RX, firmware patch simplification, enhancements for RX filters, RX processing,
Bug fixes and updates the firmware version to v3.2.3.0.

This patch set updates the BNA driver to v3.2.23.0 and was tested against
net-next 3.12.0-rc6 kernel.

Thanks
Rasesh

Rasesh Mody (13):
  bna: Add software PTP support
  bna: Set Get IOC fw State
  bna: Fix Filter Add Del
  bna: RX Filter Enhancements
  bna: Enable Multi Buffer RX
  bna: UDP RX Processing Improvements
  bna: Fix Stale MTU used to Configure Multi-buffer RX
  bna: CQ Read Fix
  bna: Add NULL Check Before Dereferencing TCB
  bna: Handle the TX Setup Failures
  bna: Embed SKB Length in TX Vector
  bna: Firmware Patch Simplification
  bna: Update the Driver Version to 3.2.23.0

 drivers/net/ethernet/brocade/bna/bfa_ioc.c      | 653 ++++++++++++++++++++++--
 drivers/net/ethernet/brocade/bna/bfa_ioc.h      |   8 +
 drivers/net/ethernet/brocade/bna/bfa_ioc_ct.c   |  40 ++
 drivers/net/ethernet/brocade/bna/bfi.h          |  33 +-
 drivers/net/ethernet/brocade/bna/bfi_enet.h     |   3 +-
 drivers/net/ethernet/brocade/bna/bna.h          |  24 +-
 drivers/net/ethernet/brocade/bna/bna_enet.c     |  58 ++-
 drivers/net/ethernet/brocade/bna/bna_hw_defs.h  |   4 +
 drivers/net/ethernet/brocade/bna/bna_tx_rx.c    | 249 +++++++--
 drivers/net/ethernet/brocade/bna/bna_types.h    |  57 ++-
 drivers/net/ethernet/brocade/bna/bnad.c         | 543 ++++++++++++++------
 drivers/net/ethernet/brocade/bna/bnad.h         |  29 +-
 drivers/net/ethernet/brocade/bna/bnad_ethtool.c |   1 +
 drivers/net/ethernet/brocade/bna/cna.h          |   4 +-
 14 files changed, 1414 insertions(+), 292 deletions(-)

-- 
1.8.2.3

[net-next 13/13] bna: Update the Driver Version to 3.2.23.0

[Rasesh Mody]

Signed-off-by: Rasesh Mody <rmody@brocade.com>
---
 drivers/net/ethernet/brocade/bna/bnad.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/brocade/bna/bnad.h b/drivers/net/ethernet/brocade/bna/bnad.h
index 474a597..ea8f1d6 100644
--- a/drivers/net/ethernet/brocade/bna/bnad.h
+++ b/drivers/net/ethernet/brocade/bna/bnad.h
@@ -71,7 +71,7 @@ struct bnad_rx_ctrl {
 #define BNAD_NAME			"bna"
 #define BNAD_NAME_LEN			64
 
-#define BNAD_VERSION			"3.2.21.1"
+#define BNAD_VERSION			"3.2.23.0"
 
 #define BNAD_MAILBOX_MSIX_INDEX		0
 #define BNAD_MAILBOX_MSIX_VECTORS	1
-- 
1.8.2.3