* Re: NULL pointer dereference in xt_register_target()
From: Eric Dumazet @ 2012-09-05 15:55 UTC (permalink / raw)
To: Cong Wang; +Cc: netfilter-devel, Linux Kernel Network Developers
In-Reply-To: <CAM_iQpVUZ50RYLUgEHbcm7ua996hgE7Gj-P-bWNJutuvm1nUoQ@mail.gmail.com>
On Wed, 2012-09-05 at 23:43 +0800, Cong Wang wrote:
> Hi, folks,
>
> The latest net-next tree can't boot due to a NULL ptr def
> bug in the kernel, the full backtrack is:
>
> http://img1.douban.com/view/photo/photo/public/p1697139550.jpg
>
> the kernel .config file is:
>
> http://pastebin.com/9YTnkqKN
>
> I don't have time to look into the issue. If you need other info,
> please let me know.
It seems xt_nat_init() is called before xt_init(), so xt array is not
yet setup.
^ permalink raw reply
* Re: [net-next.git 2/7] stmmac: manage tx clean out of rx_poll
From: Ben Hutchings @ 2012-09-05 15:51 UTC (permalink / raw)
To: Giuseppe CAVALLARO; +Cc: netdev, davem
In-Reply-To: <1346857432-24657-3-git-send-email-peppe.cavallaro@st.com>
On Wed, 2012-09-05 at 17:03 +0200, Giuseppe CAVALLARO wrote:
> This patch is to invoke the stmmac_tx (tx handler)
> out of the NAPI poll method.
> This will make easier the next step to add the new
> mitigation schema.
> Also the patch enhances the ethtool to report some
> stats for normal TX and RX IRQs.
[...]
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> @@ -824,16 +824,27 @@ static void stmmac_tx_err(struct stmmac_priv *priv)
> netif_wake_queue(priv->dev);
> }
>
> +static inline void stmmac_rx_schedule(struct stmmac_priv *priv)
> +{
> + if (likely(napi_schedule_prep(&priv->napi))) {
> + stmmac_disable_irq(priv);
> + __napi_schedule(&priv->napi);
> + }
> +}
>
> static void stmmac_dma_interrupt(struct stmmac_priv *priv)
> {
> int status;
>
> status = priv->hw->dma->dma_interrupt(priv->ioaddr, &priv->xstats);
> - if (likely(status == handle_tx_rx))
> - _stmmac_schedule(priv);
> -
> - else if (unlikely(status == tx_hard_error_bump_tc)) {
> + if (likely(status == handle_rx)) {
Surely 'status & handle_rx'?
> + priv->xstats.rx_normal_irq_n++;
> + stmmac_rx_schedule(priv);
> + }
> + if (likely(status == handle_tx)) {
'status & handle_tx'?
> + priv->xstats.tx_normal_irq_n++;
> + stmmac_tx(priv);
> + } else if (unlikely(status == tx_hard_error_bump_tc)) {
> /* Try to bump up the dma threshold on this failure */
> if (unlikely(tc != SF_DMA_MODE) && (tc <= 256)) {
> tc += 64;
> @@ -1443,8 +1454,7 @@ static int stmmac_poll(struct napi_struct *napi, int budget)
> struct stmmac_priv *priv = container_of(napi, struct stmmac_priv, napi);
> int work_done = 0;
>
> - priv->xstats.poll_n++;
> - stmmac_tx(priv);
> + priv->xstats.rx_napi_poll++;
> work_done = stmmac_rx(priv, budget);
>
> if (work_done < budget) {
--
Ben Hutchings, Staff Engineer, Solarflare
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
^ permalink raw reply
* Re: [PATCH] ceph: [PATCH] ceph: use list_move_tail instead of list_del/list_add_tail
From: Sage Weil @ 2012-09-05 15:48 UTC (permalink / raw)
To: Wei Yongjun; +Cc: davem, yongjun_wei, ceph-devel, netdev
In-Reply-To: <CAPgLHd-+oONn90yq4yy6-NKDcSwTduqu44ZUFbSCDP8bci84qw@mail.gmail.com>
Applied to the ceph queue. Thanks!
sage
On Wed, 5 Sep 2012, Wei Yongjun wrote:
> From: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
>
> Using list_move_tail() instead of list_del() + list_add_tail().
>
> Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
> ---
> net/ceph/pagelist.c | 5 ++---
> 1 file changed, 2 insertions(+), 3 deletions(-)
>
> diff --git a/net/ceph/pagelist.c b/net/ceph/pagelist.c
> index 665cd23..92866be 100644
> --- a/net/ceph/pagelist.c
> +++ b/net/ceph/pagelist.c
> @@ -1,4 +1,3 @@
> -
> #include <linux/module.h>
> #include <linux/gfp.h>
> #include <linux/pagemap.h>
> @@ -134,8 +133,8 @@ int ceph_pagelist_truncate(struct ceph_pagelist *pl,
> ceph_pagelist_unmap_tail(pl);
> while (pl->head.prev != c->page_lru) {
> page = list_entry(pl->head.prev, struct page, lru);
> - list_del(&page->lru); /* remove from pagelist */
> - list_add_tail(&page->lru, &pl->free_list); /* add to reserve */
> + /* move from pagelist to reserve */
> + list_move_tail(&page->lru, &pl->free_list);
> ++pl->num_pages_free;
> }
> pl->room = c->room;
>
>
^ permalink raw reply
* NULL pointer dereference in xt_register_target()
From: Cong Wang @ 2012-09-05 15:43 UTC (permalink / raw)
To: netfilter-devel; +Cc: Linux Kernel Network Developers
Hi, folks,
The latest net-next tree can't boot due to a NULL ptr def
bug in the kernel, the full backtrack is:
http://img1.douban.com/view/photo/photo/public/p1697139550.jpg
the kernel .config file is:
http://pastebin.com/9YTnkqKN
I don't have time to look into the issue. If you need other info,
please let me know.
Thanks.
^ permalink raw reply
* Re: sctp_close/sk_free: kernel BUG at arch/x86/mm/physaddr.c:18!
From: Eric Dumazet @ 2012-09-05 15:40 UTC (permalink / raw)
To: Marc Kleine-Budde
Cc: Fengguang Wu, H.K. Jerry Chu, Eric W. Biederman, networking,
linux-can
In-Reply-To: <1346859046.13121.144.camel@edumazet-glaptop>
On Wed, 2012-09-05 at 17:30 +0200, Eric Dumazet wrote:
> On Wed, 2012-09-05 at 17:01 +0200, Marc Kleine-Budde wrote:
> > On 09/05/2012 04:55 PM, Fengguang Wu wrote:
> > >>> This in turn means the problem doesn't come from the CAN patches, as
> > >>> both trees have different CAN patches. I'm adding Eric W. Biederman on
> > >>> Cc as he contributed some sctp patches between v3.6 and net-next/master.
> > >>
> > >> Anything is possible, but this seems unlikely as I don't think I touched
> > >> anything close to that part of the code.
> > >
> > > You are both right. The bad commit turns out to be one of:
> > >
> > > 1bed966cc3bd4042110129f0fc51aeeb59c5b200 Merge branch 'tcp_fastopen_server'
> > > 168a8f58059a22feb9e9a2dcc1b8053dbbbc12ef tcp: TCP Fast Open Server - main code path
> > > 8336886f786fdacbc19b719c1f7ea91eb70706d4 tcp: TCP Fast Open Server - support TFO listeners
> > >
> > > Thanks,
> > > Fengguang
> >
> > Thanks for your work Fengguang.
> >
> > Marc
> >
>
> OK I have a good idea how to fix the bug, I will send a patch ASAP
Could you test the following patch please ?
(Not sure why sctp doesnt memset/bzero its whole socket by the way...)
Thanks
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index 4f70ef0..845372b 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -149,11 +149,8 @@ void inet_sock_destruct(struct sock *sk)
pr_err("Attempt to release alive inet socket %p\n", sk);
return;
}
- if (sk->sk_type == SOCK_STREAM) {
- struct fastopen_queue *fastopenq =
- inet_csk(sk)->icsk_accept_queue.fastopenq;
- kfree(fastopenq);
- }
+ if (sk->sk_protocol == IPPROTO_TCP)
+ kfree(inet_csk(sk)->icsk_accept_queue.fastopenq);
WARN_ON(atomic_read(&sk->sk_rmem_alloc));
WARN_ON(atomic_read(&sk->sk_wmem_alloc));
^ permalink raw reply related
* Re: sctp_close/sk_free: kernel BUG at arch/x86/mm/physaddr.c:18!
From: Eric Dumazet @ 2012-09-05 15:30 UTC (permalink / raw)
To: Marc Kleine-Budde
Cc: Fengguang Wu, H.K. Jerry Chu, Eric W. Biederman, networking,
linux-can
In-Reply-To: <50476931.20100@pengutronix.de>
On Wed, 2012-09-05 at 17:01 +0200, Marc Kleine-Budde wrote:
> On 09/05/2012 04:55 PM, Fengguang Wu wrote:
> >>> This in turn means the problem doesn't come from the CAN patches, as
> >>> both trees have different CAN patches. I'm adding Eric W. Biederman on
> >>> Cc as he contributed some sctp patches between v3.6 and net-next/master.
> >>
> >> Anything is possible, but this seems unlikely as I don't think I touched
> >> anything close to that part of the code.
> >
> > You are both right. The bad commit turns out to be one of:
> >
> > 1bed966cc3bd4042110129f0fc51aeeb59c5b200 Merge branch 'tcp_fastopen_server'
> > 168a8f58059a22feb9e9a2dcc1b8053dbbbc12ef tcp: TCP Fast Open Server - main code path
> > 8336886f786fdacbc19b719c1f7ea91eb70706d4 tcp: TCP Fast Open Server - support TFO listeners
> >
> > Thanks,
> > Fengguang
>
> Thanks for your work Fengguang.
>
> Marc
>
OK I have a good idea how to fix the bug, I will send a patch ASAP
Thanks
^ permalink raw reply
* Re: [PATCH 1/4] net: mvneta: driver for Marvell Armada 370/XP network unit
From: Florian Fainelli @ 2012-09-05 15:25 UTC (permalink / raw)
To: linux-arm-kernel
Cc: Lior Amsalem, Andrew Lunn, Ike Pan, Albert Stone, Nadav Haklai,
Ian Molton, Lennert Buytenhek, David Marlin, Rami Rosen,
Yehuda Yitschak, Jani Monoses, Tawfik Bayouk, Dan Frazier,
Eran Ben-Avi, Li Li, Leif Lindholm, Sebastian Hesselbarth,
Jason Cooper, Arnd Bergmann, Jon Masters, Ben Dooks,
Gregory Clement, Thomas Petazzoni
In-Reply-To: <1346764004-16332-2-git-send-email-thomas.petazzoni@free-electrons.com>
Hello Thomas,
The overall driver looks very nice, my biggest concern with this driver being
that it does not implement phylib and therefore reimplements a bit of existing
code. I am not commentin on how to represent this MDIO/PHY devices using
Device Tree since this has been addressed already.
Once you register a MDIO bus for your interface, please make sure that you
give it a unique name in the system (<pdev->name>-<pdev->id>).
Other comments inline.
On Tuesday 04 September 2012 15:06:41 Thomas Petazzoni wrote:
[snip]
> +
> +/* Increment txq get counter */
> +static void mvneta_inc_get(struct mvneta_tx_queue *txq)
> +{
> + txq->txq_get_index++;
> + if (txq->txq_get_index == txq->size)
> + txq->txq_get_index = 0;
> +}
> +
> +/* Increment txq put counter */
> +static void mvneta_inc_put(struct mvneta_tx_queue *txq)
> +{
> + txq->txq_put_index++;
> + if (txq->txq_put_index == txq->size)
> + txq->txq_put_index = 0;
> +}
I would make it clear that these helpers operate on the txq, and suffix it with
_txq.
> +
> +
> +/* Clear all MIB counters */
> +static void mvneta_mib_counters_clear(struct mvneta_port *pp)
> +{
> + int i;
> + u32 dummy;
> +
> + /* Perform dummy reads from MIB counters */
> + for (i = 0; i < MVNETA_MIB_LATE_COLLISION; i += 4)
> + dummy = mvreg_read(pp, (MVNETA_MIB_COUNTERS_BASE + i));
> +}
> +
> +/* Read speed, duplex, and flow control from port status register */
> +static int mvneta_link_status(struct mvneta_port *pp,
> + struct mvneta_lnk_status *status)
> +{
> + u32 val;
> +
> + val = mvreg_read(pp, MVNETA_GMAC_STATUS);
> +
> + if (val & MVNETA_GMAC_SPEED_1000_MASK)
> + status->speed = MVNETA_SPEED_1000;
> + else if (val & MVNETA_GMAC_SPEED_100_MASK)
> + status->speed = MVNETA_SPEED_100;
> + else
> + status->speed = MVNETA_SPEED_10;
> +
> + if (val & MVNETA_GMAC_LINK_UP_MASK)
> + status->linkup = 1;
> + else
> + status->linkup = 0;
> +
> + if (val & MVNETA_GMAC_FULL_DUPLEX_MASK)
> + status->duplex = MVNETA_DUPLEX_FULL;
> + else
> + status->duplex = MVNETA_DUPLEX_HALF;
> +
> + if (val & MVNETA_GMAC_TX_FLOW_CTRL_ACTIVE_MASK)
> + status->tx_fc = MVNETA_FC_ACTIVE;
> + else if (val & MVNETA_GMAC_TX_FLOW_CTRL_ENABLE_MASK)
> + status->tx_fc = MVNETA_FC_ENABLE;
> + else
> + status->tx_fc = MVNETA_FC_DISABLE;
> +
> + if (val & MVNETA_GMAC_RX_FLOW_CTRL_ACTIVE_MASK)
> + status->rx_fc = MVNETA_FC_ACTIVE;
> + else if (val & MVNETA_GMAC_RX_FLOW_CTRL_ENABLE_MASK)
> + status->rx_fc = MVNETA_FC_ENABLE;
> + else
> + status->rx_fc = MVNETA_FC_DISABLE;
> +
I would rather see you use a struct phy_device and update its properties
instead of keeping a local copy of it. This would allow you to have consistent
reporting through ethtool, I have more comments on this later on as well.
> +static void mvneta_rxq_non_occup_desc_add(struct mvneta_port *pp,
> + struct mvneta_rx_queue *rxq,
> + int rx_desc)
> +{
> + u32 val;
> +
> + /* Only 255 descriptors can be added at once */
> + while (rx_desc > 0xff) {
> + val = (0xff << MVNETA_RXQ_ADD_NON_OCCUPIED_OFFS);
> + mvreg_write(pp, MVNETA_RXQ_STATUS_UPDATE_REG(rxq->id), val);
> + rx_desc = rx_desc - 0xff;
> + }
You could probably use a define here for 255/0xff.
[snip]
> + m_delay = 0;
This does not look like an useful name, count would be better
> + do {
> + if (m_delay >= MVNETA_RX_DISABLE_TIMEOUT_MSEC) {
> + netdev_info(pp->dev,
> + "TIMEOUT for RX stopped ! rx_queue_cmd: 0x08%x\n",
> + val);
Please use a different logging level such as netdev_err() or netdev_warn() for
instance.
> + break;
> + }
> + mdelay(1);
> + m_delay++;
What about using msleep() instead here?
> +
> + val = mvreg_read(pp, MVNETA_RXQ_CMD);
> + } while (val & 0xff);
> +
> + /* Stop Tx port activity. Check port Tx activity. Issue stop
> + command for active channels only */
> + val = (mvreg_read(pp, MVNETA_TXQ_CMD)) & MVNETA_TXQ_ENABLE_MASK;
> +
> + if (val != 0)
> + mvreg_write(pp, MVNETA_TXQ_CMD,
> + (val << MVNETA_TXQ_DISABLE_OFFS));
> +
> + /* Wait for all Tx activity to terminate. */
> + m_delay = 0;
> + do {
> + if (m_delay >= MVNETA_TX_DISABLE_TIMEOUT_MSEC) {
> + netdev_info(pp->dev,
> + "TIMEOUT for TX stopped tx_queue_cmd - 0x%08x\n",
> + val);
> + break;
> + }
> + mdelay(1);
> + m_delay++;
> +
> + /* Check TX Command reg that all Txqs are stopped */
> + val = mvreg_read(pp, MVNETA_TXQ_CMD);
Ditto
> +
> + } while (val & 0xff);
> + tx_fifo_empty_mask |= MVNETA_TX_FIFO_EMPTY_MASK;
> + tx_in_prog_mask |= MVNETA_TX_IN_PRGRS_MASK;
> +
> + /* Double check to verify that TX FIFO is empty */
> + m_delay = 0;
> + while (1) {
> + do {
> + if (m_delay >= MVNETA_TX_FIFO_EMPTY_TIMEOUT) {
> + netdev_info(pp->dev,
> + "TX FIFO empty timeout status=0x08%x,
empty=%x, in_prog=%x",
> + val, tx_fifo_empty_mask,
> + tx_in_prog_mask);
> + break;
> + }
> + mdelay(1);
> + m_delay++;
Ditto
> +
> + val = mvreg_read(pp, MVNETA_PORT_STATUS);
> + } while (((val & tx_fifo_empty_mask) != tx_fifo_empty_mask)
> + || ((val & tx_in_prog_mask) != 0));
> +
> + if (m_delay >= MVNETA_TX_FIFO_EMPTY_TIMEOUT)
> + break;
> +
> + val = mvreg_read(pp, MVNETA_PORT_STATUS);
> + if (((val & tx_fifo_empty_mask) == tx_fifo_empty_mask) &&
> + ((val & tx_in_prog_mask) == 0))
> + break;
> + else
> + netdev_info(pp->dev, "TX FIFO Empty double check failed. %d
msec status=0x%x, empty=0x%x, in_prog=0x%x\n",
> + m_delay, val, tx_fifo_empty_mask,
> + tx_in_prog_mask);
> + }
> +
> + udelay(200);
> +}
[snip]
> +
> +/* This method sets defaults to the NETA port:
> + * Clears interrupt Cause and Mask registers.
> + * Clears all MAC tables.
> + * Sets defaults to all registers.
> + * Resets RX and TX descriptor rings.
> + * Resets PHY.
> + * This method can be called after mvneta_port_down() to return the port
> + * settings to defaults.
> + */
Please use standard kernel-doc style comments.
[snip]
> +/* Read the Link Up bit (LinkUp) in port MAC control register */
> +static int mvneta_link_is_up(struct mvneta_port *pp)
> +{
> + u32 val;
> + val = mvreg_read(pp, MVNETA_GMAC_STATUS);
> + if (val & MVNETA_GMAC_LINK_UP_MASK)
> + return 1;
return mvreg_read(pp, MVNETA_GMAC_STATUS) & MVNETA_GMAC_LINK_UP_MASK;
and make it static inline.
> +
> + return 0;
> +}
> +
> +/* Get phy address */
> +static int mvneta_phy_addr_get(struct mvneta_port *pp)
> +{
> + unsigned int val;
> +
> + val = mvreg_read(pp, MVNETA_PHY_ADDR);
> + val &= 0x1f;
Use PHY_MAX_ADDR - 1 instead here.
> + return val;
> +}
> +
[snip]
> +/* Display status (link, duplex, speed) of the port */
> +void mvneta_link_status_print(struct mvneta_port *pp)
> +{
> + struct mvneta_lnk_status link;
> + char *speedstr, *duplexstr;
> +
> + mvneta_link_status(pp, &link);
> +
> + if (link.linkup) {
> + if (link.speed == MVNETA_SPEED_1000)
> + speedstr = "1 Gbps";
> + else if (link.speed == MVNETA_SPEED_100)
> + speedstr = "100 Mbps";
> + else
> + speedstr = "10 Mbps";
> +
> + if (link.duplex == MVNETA_DUPLEX_FULL)
> + duplexstr = "full";
> + else
> + duplexstr = "half";
> +
> + netdev_info(pp->dev,
> + "link up, %s duplex, speed %s\n",
> + duplexstr, speedstr);
> + } else
> + netdev_info(pp->dev, "link down\n");
> +}
You should rather define a phylib adjust_link callback to do this. Otherwise
please reduce the indentation by handling the case when the link is down first.
> +
> +/* Display more error info */
> +static void mvneta_rx_error(struct mvneta_port *pp,
> + struct mvneta_rx_desc *rx_desc)
> +{
> + u32 status = rx_desc->status;
> +
> + if (pp->dev)
> + pp->dev->stats.rx_errors++;
Please do this outside of this function and just let it print the error.
> +
> + if ((status & MVNETA_RXD_FIRST_LAST_DESC_MASK)
> + != MVNETA_RXD_FIRST_LAST_DESC_MASK) {
> + netdev_err(pp->dev,
> + "bad rx status %08x (buffer oversize), size=%d\n",
> + rx_desc->status, rx_desc->data_size);
> + return;
> + }
[snip]
> +
> +/* Refill processing */
> +static int mvneta_rx_refill(struct mvneta_port *pp,
> + struct mvneta_rx_desc *rx_desc)
> +
> +{
> + unsigned long phys_addr;
> + struct sk_buff *skb;
> +
> + skb = netdev_alloc_skb(pp->dev, pp->pkt_size);
> + if (!skb) {
> + mvneta_add_cleanup_timer(pp);
> + return 1;
> + }
> +
> + phys_addr = dma_map_single(pp->dev->dev.parent, skb->head,
> + MVNETA_RX_BUF_SIZE(pp->pkt_size),
> + DMA_FROM_DEVICE);
Check that your phys_addr cookie has been successfully mapped using
dma_mapping_error().
[snip]
> +/* Main tx processing */
> +static int mvneta_tx(struct sk_buff *skb, struct net_device *dev)
> +{
> + struct mvneta_port *pp = netdev_priv(dev);
> +
> + int frags = 0;
> + int res = NETDEV_TX_OK;
> + u32 tx_cmd;
> + struct mvneta_tx_queue *txq = NULL;
> + struct mvneta_tx_desc *tx_desc;
> +
> + if (!test_bit(MVNETA_F_STARTED_BIT, &pp->flags))
> + goto out;
Is not this equivalent to !netif_running(dev)? At least print some message so
we know that this is not supposed to happen.
> +
> + txq = &pp->txqs[mvneta_txq_def];
> +
> + frags = skb_shinfo(skb)->nr_frags + 1;
> +
> + tx_desc = mvneta_tx_desc_get(pp, txq, frags);
> + if (tx_desc == NULL) {
> + frags = 0;
> + dev->stats.tx_dropped++;
> + res = NETDEV_TX_BUSY;
> + goto out;
> + }
> +
> + tx_cmd = mvneta_skb_tx_csum(pp, skb);
> +
> + tx_desc->data_size = skb_headlen(skb);
> +
> + tx_desc->buf_phys_addr = dma_map_single(dev->dev.parent, skb->data,
> + tx_desc->data_size,
> + DMA_TO_DEVICE);
Please check this dma_map_single() return value too.
[snip]
> +static int mvneta_addr_crc(unsigned char *addr)
> +{
> + int crc = 0;
> + int i;
> +
> + for (i = 0; i < 6; i++) {
ETH_ALEN instead of 6 to make it clear it operates on addresses.
> + int j;
> +
> + crc = (crc ^ addr[i]) << 8;
> + for (j = 7; j >= 0; j--) {
> + if (crc & (0x100 << j))
> + crc ^= 0x107 << j;
> + }
> + }
> +
> + return crc;
> +}
[snip]
> +
> +/* Interrupt handling - the callback for request_irq() */
> +static irqreturn_t mvneta_isr(int irq, void *dev_id)
> +{
> + struct mvneta_port *pp = (struct mvneta_port *)dev_id;
> +
> + /* Mask all interrupts */
> + mvreg_write(pp, MVNETA_INTR_NEW_MASK, 0);
> +
> + /* Verify that the device not already on the polling list */
> + if (napi_schedule_prep(&pp->napi))
> + __napi_schedule(&pp->napi);
Does not the hardware generate interrupts for tx completion, PHY interrupts
etc ...?
> +
> + return IRQ_HANDLED;
> +}
> +
> +/* Handle link event */
> +static void mvneta_link_event(struct mvneta_port *pp)
> +{
> + struct net_device *dev = pp->dev;
> +
> + /* Check Link status on ethernet port */
> +
> + if (mvneta_link_is_up(pp)) {
> + mvneta_port_up(pp);
> + set_bit(MVNETA_F_LINK_UP_BIT, &pp->flags);
> +
> + if (dev) {
> + netif_carrier_on(dev);
> + netif_tx_wake_all_queues(dev);
> + }
> + } else {
> + if (dev) {
> + netif_carrier_off(dev);
> + netif_tx_stop_all_queues(dev);
> + }
> + mvneta_port_down(pp);
> + clear_bit(MVNETA_F_LINK_UP_BIT, &pp->flags);
> + }
> +
> + mvneta_link_status_print(pp);
Again, this is taken care of by phylib nicely, and does not require you to
have this F_LINK_UP_BIT.
[snip]
> +
> +/* Handle rxq fill: allocates rxq skbs; called when initializing a port */
> +static int mvneta_rxq_fill(struct mvneta_port *pp, struct mvneta_rx_queue
*rxq,
> + int num)
> +{
> + int i;
> + struct sk_buff *skb;
> + struct mvneta_rx_desc *rx_desc;
> + unsigned long phys_addr;
> + struct net_device *dev = pp->dev;
> +
> + for (i = 0; i < num; i++) {
> + skb = dev_alloc_skb(pp->pkt_size);
> + if (!skb) {
> + netdev_err(pp->dev, "%s:rxq %d, %d of %d buffs filled\n",
> + __func__, rxq->id, i, num);
> + break;
> + }
> +
> + rx_desc = rxq->descs + i;
> + memset(rx_desc, 0, sizeof(struct mvneta_rx_desc));
> + phys_addr = dma_map_single(dev->dev.parent, skb->head,
> + MVNETA_RX_BUF_SIZE(pp->pkt_size),
> + DMA_FROM_DEVICE);
Here again, check phys_addr.
> + mvneta_rx_desc_fill(rx_desc, phys_addr, (u32)skb);
> + }
> +
> + /* add this num of RX descriptors as non occupied (ready to get pkts) */
> + mvneta_rxq_non_occup_desc_add(pp, rxq, i);
> +
> + return i;
> +}
> +
[snip]
> +
> +/* Create a specified RX queue */
> +static int mvneta_rxq_init(struct mvneta_port *pp,
> + struct mvneta_rx_queue *rxq)
> +
> +{
> + rxq->size = pp->rx_ring_size;
> +
> + /* Allocate DMA descriptors array */
> + rxq->descs_orig = dma_alloc_coherent(pp->dev->dev.parent,
> + MVNETA_RX_TOTAL_DESCS_SIZE(rxq),
> + &rxq->descs_phys_orig,
> + GFP_KERNEL);
> + if (rxq->descs_orig == NULL) {
Use dma_mapping_error() instead.
> + netdev_err(pp->dev, "rxQ=%d: Can't allocate %d bytes for %d RX
descr\n",
> + rxq->id, MVNETA_RX_TOTAL_DESCS_SIZE(rxq), rxq->size);
> + return -ENOMEM;
> + }
> +
> + /* Make sure descriptor address is cache line size aligned */
> + rxq->descs = PTR_ALIGN(rxq->descs_orig, MVNETA_CPU_D_CACHE_LINE_SIZE);
> + rxq->descs_phys = ALIGN(rxq->descs_phys_orig,
> + MVNETA_CPU_D_CACHE_LINE_SIZE);
> +
> + rxq->last_desc = rxq->size - 1;
Don't you need some kind of barrier here? I do not know exactly how coherent
your peripherals and memory are, just wondering.
> +
> + /* Set Rx descriptors queue starting address */
> + mvreg_write(pp, MVNETA_RXQ_BASE_ADDR_REG(rxq->id), rxq->descs_phys);
> + mvreg_write(pp, MVNETA_RXQ_SIZE_REG(rxq->id), rxq->size);
> +
> + /* Set Offset */
> + mvneta_rxq_offset_set(pp, rxq, NET_SKB_PAD);
> +
> + /* Set coalescing pkts and time */
> + mvneta_rx_pkts_coal_set(pp, rxq, rxq->pkts_coal);
> + mvneta_rx_time_coal_set(pp, rxq, rxq->time_coal);
> +
> + /* Fill RXQ with buffers from RX pool */
> + mvneta_rxq_buf_size_set(pp, rxq, MVNETA_RX_BUF_SIZE(pp->pkt_size));
> + mvneta_rxq_bm_disable(pp, rxq);
> + mvneta_rxq_fill(pp, rxq, rxq->size);
> +
> + return 0;
> +}
> +
[snip]
> +static int mvneta_txq_init(struct mvneta_port *pp,
> + struct mvneta_tx_queue *txq)
> +{
> + txq->size = pp->tx_ring_size;
> +
> + /* Allocate DMA descriptors array */
> + txq->descs_orig = dma_alloc_coherent(pp->dev->dev.parent,
> + MVNETA_TX_TOTAL_DESCS_SIZE(txq),
> + &txq->descs_phys_orig,
> + GFP_KERNEL);
> + if (txq->descs_orig == NULL) {
Use dma_mapping_error().
> + netdev_err(pp->dev, "txQ=%d: Can't allocate %d bytes for %d TX
descr\n",
> + txq->id, MVNETA_TX_TOTAL_DESCS_SIZE(txq), txq->size);
> + return -ENOMEM;
> + }
> +
[snip]
> +/* Fill rx buffers, start Rx/Tx activity, set coalesing,
> +* clear and unmask interrupt bits
> +*/
> +static int mvneta_start_internals(struct mvneta_port *pp, int mtu)
> +{
> + int err = 0;
> +
> + pp->pkt_size = MVNETA_RX_PKT_SIZE(mtu);
> + if (test_bit(MVNETA_F_STARTED_BIT, &pp->flags))
> + return -EINVAL;
You probably mean -EBUSY here instead?
> +
> + if (mvneta_max_rx_size_set(pp, MVNETA_RX_PKT_SIZE(mtu))) {
> + netdev_err(pp->dev,
> + "%s: can't set maxRxSize=%d mtu=%d\n",
> + __func__, MVNETA_RX_PKT_SIZE(mtu), mtu);
> + return -EINVAL;
> + }
> +
> + err = mvneta_setup_rxqs(pp);
> + if (unlikely(err))
> + return err;
> +
> + err = mvneta_setup_txqs(pp);
> + if (unlikely(err)) {
> + mvneta_cleanup_rxqs(pp);
> + return err;
> + }
> +
> + mvneta_txq_max_tx_size_set(pp, MVNETA_RX_PKT_SIZE(mtu));
> +
> + /* start the Rx/Tx activity */
> + mvneta_port_enable(pp);
> +
> + set_bit(MVNETA_F_LINK_UP_BIT, &pp->flags);
> + set_bit(MVNETA_F_STARTED_BIT, &pp->flags);
> +
> + return 0;
> +}
> +
> +/* Stop port Rx/Tx activity, free skb's from Rx/Tx rings */
> +static int mvneta_stop_internals(struct mvneta_port *pp)
> +{
> + clear_bit(MVNETA_F_STARTED_BIT, &pp->flags);
> +
> + /* Stop the port activity */
> + mvneta_port_disable(pp);
> +
> + /* Clear all ethernet port interrupts */
> + mvreg_write(pp, MVNETA_INTR_MISC_CAUSE, 0);
> + mvreg_write(pp, MVNETA_INTR_OLD_CAUSE, 0);
> +
> + /* Mask all interrupts */
> + mvneta_interrupts_mask(pp);
> + smp_call_function_many(cpu_online_mask, mvneta_interrupts_mask,
> + pp, 1);
> +
> + /* Reset TX port here. */
> + mvneta_tx_reset(pp);
> +
> + mvneta_cleanup_rxqs(pp);
> + mvneta_cleanup_txqs(pp);
> +
> + return 0;
> +
> +}
> +
> +/* Start the port, connect to port interrupt line, unmask interrupts */
> +static int mvneta_start(struct net_device *dev)
> +{
> + struct mvneta_port *pp = netdev_priv(dev);
> +
> + /* In default link is down */
> + netif_carrier_off(dev);
> + netif_tx_stop_all_queues(dev);
> +
> + /* Fill rx buffers, start Rx/Tx activity, set coalescing */
> + if (mvneta_start_internals(pp, dev->mtu) != 0) {
> + netdev_err(dev, "start internals failed\n");
> + return -ENODEV;
> + }
> +
> + /* Enable polling on the port, must be used after netif_poll_disable */
> + napi_enable(&pp->napi);
> +
> + if (pp->flags & MVNETA_F_LINK_UP) {
> + netif_carrier_on(dev);
> + netif_tx_wake_all_queues(dev);
> + } else {
> + netdev_info(dev, "%s: NOT MVNETA_F_LINK_UP\n", __func__);
> + }
Remove this message as well.
> +
> + /* Connect to port interrupt line */
> + if (request_irq(dev->irq, mvneta_isr, (IRQF_DISABLED), "mv_eth", pp)) {
> + netdev_err(dev, "cannot request irq %d\n", dev->irq);
> + napi_disable(&pp->napi);
> + goto error;
> + }
You should probably request the interrupt prior to calling napi_enable()
> +
> + /* Unmask interrupts */
> + mvneta_interrupts_unmask(pp);
> + smp_call_function_many(cpu_online_mask,
> + mvneta_interrupts_unmask,
> + pp, 1);
> +
> + netdev_info(dev, "started\n");
Remove this please.
> + return 0;
> +
> +error:
> + netdev_err(dev, "start failed\n");
> + mvneta_cleanup_rxqs(pp);
> + mvneta_cleanup_txqs(pp);
> +
> + return -ENODEV;
> +}
> +
> + if (dev->irq != 0)
> + free_irq(dev->irq, pp);
This looks superfluous, you refuse to bring up the interface if the interrupt
requesting fails.
> +
> + netdev_info(dev, "stopped\n");
> +
> + return 0;
> +}
> +
> +
> +/* tx timeout callback - display a message and stop/start the network
device */
> +static void mvneta_tx_timeout(struct net_device *dev)
> +{
> + netdev_info(dev, "tx timeout\n");
> + if (netif_running(dev)) {
> + mvneta_stop(dev);
> + mvneta_start(dev);
> + }
You should never end-up with the case where the interface is not running and
you face a transmit timeout.
[snip]
> +/* Change the device mtu */
> +static int mvneta_change_mtu(struct net_device *dev, int mtu)
> +{
> + int old_mtu = dev->mtu;
> +
> + mtu = mvneta_check_mtu_valid(dev, mtu);
> + if (mtu < 0)
> + return -EINVAL;
> +
> + dev->mtu = mtu;
> +
> + if (!netif_running(dev)) {
> + netdev_info(dev, "change mtu %d (buffer-size %d) to %d (buffer-size
%d)\n",
> + old_mtu, MVNETA_RX_PKT_SIZE(old_mtu),
> + dev->mtu, MVNETA_RX_PKT_SIZE(dev->mtu));
> + return 0;
Remove this message.
> + }
> +
> + if (mvneta_stop(dev)) {
> + netdev_err(dev, "stop interface failed\n");
> + goto error;
> + }
> +
> + if (mvneta_start(dev)) {
> + netdev_err(dev, "start interface failed\n");
> + goto error;
> + }
Propagate the returned error codes back to the caller.
> +
> + netdev_info(dev, "change mtu %d (buffer-size %d) to %d (buffer-size %d)\n",
> + old_mtu, MVNETA_RX_PKT_SIZE(old_mtu),
> + dev->mtu, MVNETA_RX_PKT_SIZE(dev->mtu));
Remove this message too.
> +
> + return 0;
> +
> +error:
> + netdev_info(dev, "change mtu failed\n");
> + return -EINVAL;
> +}
> +
> +/* Handle setting mac address (low level) */
> +static int mvneta_set_mac_addr_internals(struct net_device *dev, void
*addr)
> +{
> + struct mvneta_port *pp = netdev_priv(dev);
> + u8 *mac = addr + 2;
> + int i;
> +
> + /* Remove previous address table entry */
> + if (mvneta_mac_addr_set(pp, dev->dev_addr, -1) != 0) {
> + netdev_err(dev, "mvneta_mac_addr_set failed\n");
> + return -EINVAL;
> + }
> +
> + /* Set new addr in hw */
> + if (mvneta_mac_addr_set(pp, mac, mvneta_rxq_def) != 0) {
> + netdev_err(dev, "mvneta_mac_addr_set failed\n");
> + return -EINVAL;
> + }
> +
> + /* Set addr in the device */
> + for (i = 0; i < MVNETA_MAC_ADDR_SIZE; i++)
> + dev->dev_addr[i] = mac[i];
ETH_ALEN.
> +
> + netdev_info(dev, "mac address changed\n");
Remove this please.
> +
> + return 0;
> +}
> +
> +/* Handle setting mac address */
> +static int mvneta_set_mac_addr(struct net_device *dev, void *addr)
> +{
> + if (!netif_running(dev)) {
> + if (mvneta_set_mac_addr_internals(dev, addr) == -1)
> + goto error;
> + return 0;
> + }
Usually you just check if the interface is running, and if it is return
something like -EBUSY.
> +
> + if (mvneta_stop(dev)) {
> + netdev_err(dev, "stop interface failed\n");
> + goto error;
> + }
> +
> + if (mvneta_set_mac_addr_internals(dev, addr) == -1)
> + goto error;
> +
> + if (mvneta_start(dev)) {
> + netdev_err(dev, "start interface failed\n");
> + goto error;
> + }
Propagate error codes here too please.
> +
> + return 0;
> +
> +error:
> + netdev_err(dev, "set mac addr failed\n");
> + return -EINVAL;
> +}
> +
> +/*
> + * Called when a network interface is made active.
> + * Returns 0 on success, -EINVAL or =ENODEV on failure
> + * mvneta_open() is called when a network interface is made
> + * active by the system (IFF_UP). We set the mac address and
> + * invoke mvneta_start() to start the device.
> + */
> +static int mvneta_open(struct net_device *dev)
> +{
> + struct mvneta_port *pp = netdev_priv(dev);
> + int queue = mvneta_rxq_def;
> +
> + if (mvneta_mac_addr_set(pp, dev->dev_addr, queue) != 0) {
> + netdev_err(dev, "mvneta_mac_addr_set failed\n");
> + return -EINVAL;
> + }
> +
> + if (mvneta_start(dev)) {
> + netdev_err(dev, "start interface failed\n");
> + return -ENODEV;
> + }
> +
> + return 0;
Propagate the error code here too.
[snip]
> +static void mvneta_ethtool_get_drvinfo(struct net_device *dev,
> + struct ethtool_drvinfo *drvinfo)
> +{
> + strlcpy(drvinfo->driver, mvneta_driver_name,
> + sizeof(drvinfo->driver));
> + strlcpy(drvinfo->version, mvneta_driver_version,
> + sizeof(drvinfo->version));
You can probably also provide informations about the firmware version, bus_info
at least.
[snip]
> +/* Device initialization routine */
> +static int __devinit mvneta_probe(struct platform_device *pdev)
> +{
> + int err = -EINVAL;
> + struct mvneta_port *pp;
> + struct net_device *dev;
> + u32 phy_addr, clk;
> + int phy_mode;
> + const char *mac_addr;
> + const struct mbus_dram_target_info *dram_target_info;
> + struct device_node *dn = pdev->dev.of_node;
> +
> + dev = alloc_etherdev_mq(sizeof(struct mvneta_port), 8);
> + if (!dev)
> + return -ENOMEM;
> +
> + dev->irq = irq_of_parse_and_map(dn, 0);
> + if (dev->irq == 0) {
> + err = -EINVAL;
> + goto err_irq;
> + }
> +
> + if (of_property_read_u32(dn, "phy-addr", &phy_addr) != 0) {
> + dev_err(&pdev->dev, "could not read phy_addr\n");
> + err = -ENODEV;
> + goto err_node;
> + }
> +
> + phy_mode = of_get_phy_mode(dn);
> + if (phy_mode < 0) {
> + dev_err(&pdev->dev, "wrong phy-mode\n");
> + err = -EINVAL;
> + goto err_node;
> + }
> +
> + if (of_property_read_u32(dn, "clock-frequency", &clk) != 0) {
> + dev_err(&pdev->dev, "could not read clock-frequency\n");
> + err = -EINVAL;
> + goto err_node;
> + }
> +
> + mac_addr = of_get_mac_address(dn);
> +
> + if (!mac_addr || !is_valid_ether_addr(mac_addr))
> + eth_hw_addr_random(dev);
> + else
> + memcpy(dev->dev_addr, mac_addr, 6);
> +
> + dev->tx_queue_len = MVNETA_MAX_TXD;
> + dev->watchdog_timeo = 5 * HZ;
> + dev->netdev_ops = &mvneta_netdev_ops;
> +
> + SET_ETHTOOL_OPS(dev, &mvneta_eth_tool_ops);
> +
> + pp = netdev_priv(dev);
> +
> + pp->tx_done_timer.function = mvneta_tx_done_timer_callback;
> + init_timer(&pp->tx_done_timer);
> + clear_bit(MVNETA_F_TX_DONE_TIMER_BIT, &pp->flags);
> + pp->cleanup_timer.function = mvneta_cleanup_timer_callback;
> + init_timer(&pp->cleanup_timer);
> + clear_bit(MVNETA_F_CLEANUP_TIMER_BIT, &pp->flags);
> +
> + pp->weight = MVNETA_RX_POLL_WEIGHT;
> + pp->clk = clk;
Rename this clk_freq so make it less ambiguous, because this is not a proper
struct clk pointer.
> +
> + pp->base = of_iomap(dn, 0);
> + if (pp->base == NULL) {
> + err = -ENOMEM;
> + goto err_node;
> + }
> +
> + pp->tx_done_timer.data = (unsigned long)dev;
> + pp->cleanup_timer.data = (unsigned long)dev;
> +
> + pp->tx_ring_size = MVNETA_MAX_TXD;
> + pp->rx_ring_size = MVNETA_MAX_RXD;
> +
> + pp->dev = dev;
> +
> + if (mvneta_init(pp, phy_addr)) {
> + dev_err(&pdev->dev, "can't init eth hal\n");
> + err = -ENODEV;
> + goto err_base;
> + }
> + mvneta_port_power_up(pp, phy_mode);
> +
> + dram_target_info = mv_mbus_dram_info();
> + if (dram_target_info)
> + mvneta_conf_mbus_windows(pp, dram_target_info);
> +
> + netif_napi_add(dev, &pp->napi, mvneta_poll, pp->weight);
> +
> + SET_NETDEV_DEV(dev, &pdev->dev);
> +
> + if (register_netdev(dev)) {
> + dev_err(&pdev->dev, "failed to register\n");
> + err = ENOMEM;
> + goto err_base;
> + }
> +
> + dev->features = NETIF_F_SG;
> + dev->hw_features = NETIF_F_SG;
> + dev->priv_flags |= IFF_UNICAST_FLT;
> +
> + if (dev->mtu <= MVNETA_TX_CSUM_MAX_SIZE) {
> + dev->features |= NETIF_F_IP_CSUM;
> + dev->hw_features |= NETIF_F_IP_CSUM;
> + }
At this point, the condition is always true, so just set these features and
update them when the MTU changes.
> +
> + dev_info(&pdev->dev, "%s, mac: %pM pp->base=%p\n", dev->name,
> + dev->dev_addr, pp->base);
> +
> + platform_set_drvdata(pdev, pp->dev);
> +
> + return 0;
> +err_base:
> + iounmap(pp->base);
> +err_node:
> + irq_dispose_mapping(dev->irq);
> +err_irq:
> + free_netdev(dev);
> + return err;
> +}
> +
> +/* Device removal routine */
> +static int __devexit mvneta_remove(struct platform_device *pdev)
> +{
> + struct net_device *dev = platform_get_drvdata(pdev);
> + struct mvneta_port *pp = netdev_priv(dev);
> +
> + dev_info(&pdev->dev, "Removing Marvell Ethernet Driver\n");
I would remove this message.
^ permalink raw reply
* [net-next.git 0/7 (V2)] stmmac: remove dead code for STMMAC_TIMER and add new mitigation schema.
From: Giuseppe CAVALLARO @ 2012-09-05 15:03 UTC (permalink / raw)
To: netdev; +Cc: bhutchings, davem, Giuseppe Cavallaro
These patch series remove the STMMAC_TIMER option no longer updated
and never used and add a new mitigation schema.
Having removed the Timer opt, this has made the driver slim.
On top of this work, it has been easier to introduce the new
mitigation schema based on HW RX-watchdog (available in new cores).
In fact, 3.50 and newer cores have an HW RX-Watchdog that can be used for
mitigating the Rx-interrupts and first results look promising.
Running n-u-t-t-c-p with the following parameters:
Throughput: 500Mbps
UDP Buffer size: 1328bytes
TCP Buffer size: 65536bytes
for example, I got on ST box (arm-based) these improvements:
--------------------------------------------------------------------
Original | With New Mitigation patch
--------------------------------------------------------------------
Test CPU usage pkt/loss | CPU usage pkt/loss
Type Mbps % % |Mbps % %
--------------------------------------------------------------------
UDP-RX 395.5065 95 20.89 |499.9966 25 0.00%
UDP-TX 499.5578 100 0.08915 |499.7156 100 0.06029%
TCP-RX 499.9221 77 |499.8648 41
TCP-TX 389.5719 99 |499.2802 79
--------------------------------------------------------------------
... no regression on ST boxes (SH based) I always test.
This is a brief explanation of the new mitigation schema although there
is a patch that updates the driver's documentation.
o On Rx-side I have:
New GMACs will use the RX-watchdog timer; old ones will continue to
use NAPI to mitigate the RX DMA interrupts.
For the RX-watchdog, there is a parameter that is the RI Watchdog
Timer count. It indicates the number of system clock cycles and can be
set via *ethtool*.
o On Tx-side, the mitigation schema is based on a SW timer
that calls the tx function (stmmac_tx) to reclaim the resource after
transmitting the frames.
Also there is another parameter (a threshold) used to program
the descriptors avoiding to set the interrupt on completion bit in
when the frame is sent (xmit). This means that the stmmac_tx can be
called by the ISR too. Also this parameter can be tuned via ethtool.
V2: the new version of the patches add the ethtool support to get/set
coalesce parameters and totally removed the sysFS support.
Giuseppe Cavallaro (7):
stmmac: remove dead code for TIMER
stmmac: manage tx clean out of rx_poll
stmmac: add the initial tx coalesce schema
stmmac: add Rx watchdog optimization to mitigate the DMA irqs
stmmac: get/set coalesce parameters via ethtool
stmmac: update the doc with new IRQ mitigation
stmmac: update the driver version to Sept_2012
Documentation/networking/stmmac.txt | 28 ++-
drivers/net/ethernet/stmicro/stmmac/Kconfig | 25 --
drivers/net/ethernet/stmicro/stmmac/Makefile | 1 -
drivers/net/ethernet/stmicro/stmmac/common.h | 30 ++-
drivers/net/ethernet/stmicro/stmmac/dwmac1000.h | 3 -
.../net/ethernet/stmicro/stmmac/dwmac1000_dma.c | 6 +
drivers/net/ethernet/stmicro/stmmac/dwmac_dma.h | 3 +-
drivers/net/ethernet/stmicro/stmmac/dwmac_lib.c | 7 +-
drivers/net/ethernet/stmicro/stmmac/stmmac.h | 14 +-
.../net/ethernet/stmicro/stmmac/stmmac_ethtool.c | 83 +++++++-
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 245 ++++++++------------
drivers/net/ethernet/stmicro/stmmac/stmmac_timer.c | 134 -----------
drivers/net/ethernet/stmicro/stmmac/stmmac_timer.h | 46 ----
include/linux/stmmac.h | 1 +
14 files changed, 239 insertions(+), 387 deletions(-)
delete mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_timer.c
delete mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_timer.h
--
1.7.4.4
^ permalink raw reply
* [net-next.git 7/7] stmmac: update the driver version to Sept_2012
From: Giuseppe CAVALLARO @ 2012-09-05 15:03 UTC (permalink / raw)
To: netdev; +Cc: bhutchings, davem, Giuseppe Cavallaro
In-Reply-To: <1346857432-24657-1-git-send-email-peppe.cavallaro@st.com>
Many new feauture have been introduced in the driver:
sysFS, Rx HW watchdog... so this patch updates the
driver's version.
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
---
drivers/net/ethernet/stmicro/stmmac/stmmac.h | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
index 577dc13..eda34c4 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
@@ -24,7 +24,7 @@
#define __STMMAC_H__
#define STMMAC_RESOURCE_NAME "stmmaceth"
-#define DRV_MODULE_VERSION "March_2012"
+#define DRV_MODULE_VERSION "Sept_2012"
#include <linux/clk.h>
#include <linux/stmmac.h>
--
1.7.4.4
^ permalink raw reply related
* [net-next.git 6/7] stmmac: update the doc with new IRQ mitigation
From: Giuseppe CAVALLARO @ 2012-09-05 15:03 UTC (permalink / raw)
To: netdev; +Cc: bhutchings, davem, Giuseppe Cavallaro
In-Reply-To: <1346857432-24657-1-git-send-email-peppe.cavallaro@st.com>
This patch updates the stmmac.txt adding some information
about the new rx/tx mitigation schema adopted in the driver.
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
---
Documentation/networking/stmmac.txt | 28 +++++++++++++++-------------
1 files changed, 15 insertions(+), 13 deletions(-)
diff --git a/Documentation/networking/stmmac.txt b/Documentation/networking/stmmac.txt
index ef9ee71..f9fa6db 100644
--- a/Documentation/networking/stmmac.txt
+++ b/Documentation/networking/stmmac.txt
@@ -29,11 +29,9 @@ The kernel configuration option is STMMAC_ETH:
dma_txsize: DMA tx ring size;
buf_sz: DMA buffer size;
tc: control the HW FIFO threshold;
- tx_coe: Enable/Disable Tx Checksum Offload engine;
watchdog: transmit timeout (in milliseconds);
flow_ctrl: Flow control ability [on/off];
pause: Flow Control Pause Time;
- tmrate: timer period (only if timer optimisation is configured).
3) Command line options
Driver parameters can be also passed in command line by using:
@@ -60,17 +58,19 @@ Then the poll method will be scheduled at some future point.
The incoming packets are stored, by the DMA, in a list of pre-allocated socket
buffers in order to avoid the memcpy (Zero-copy).
-4.3) Timer-Driver Interrupt
-Instead of having the device that asynchronously notifies the frame receptions,
-the driver configures a timer to generate an interrupt at regular intervals.
-Based on the granularity of the timer, the frames that are received by the
-device will experience different levels of latency. Some NICs have dedicated
-timer device to perform this task. STMMAC can use either the RTC device or the
-TMU channel 2 on STLinux platforms.
-The timers frequency can be passed to the driver as parameter; when change it,
-take care of both hardware capability and network stability/performance impact.
-Several performance tests on STM platforms showed this optimisation allows to
-spare the CPU while having the maximum throughput.
+4.3) Interrupt Mitigation
+The driver is able to mitigate the number of its DMA interrupts
+using NAPI for the reception on chips older than the 3.50.
+New chips have an HW RX-Watchdog used for this mitigation.
+
+On Tx-side, the mitigation schema is based on a SW timer that calls the
+tx function (stmmac_tx) to reclaim the resource after transmitting the
+frames.
+Also there is another parameter (like a threshold) used to program
+the descriptors avoiding to set the interrupt on completion bit in
+when the frame is sent (xmit).
+
+Mitigation parameters can be tuned by ethtool.
4.4) WOL
Wake up on Lan feature through Magic and Unicast frames are supported for the
@@ -121,6 +121,7 @@ struct plat_stmmacenet_data {
int bugged_jumbo;
int pmt;
int force_sf_dma_mode;
+ int riwt_off;
void (*fix_mac_speed)(void *priv, unsigned int speed);
void (*bus_setup)(void __iomem *ioaddr);
int (*init)(struct platform_device *pdev);
@@ -156,6 +157,7 @@ Where:
o pmt: core has the embedded power module (optional).
o force_sf_dma_mode: force DMA to use the Store and Forward mode
instead of the Threshold.
+ o riwt_off: force to disable the RX watchdog feature and switch to NAPI mode.
o fix_mac_speed: this callback is used for modifying some syscfg registers
(on ST SoCs) according to the link speed negotiated by the
physical layer .
--
1.7.4.4
^ permalink raw reply related
* [net-next.git 5/7] stmmac: get/set coalesce parameters via ethtool
From: Giuseppe CAVALLARO @ 2012-09-05 15:03 UTC (permalink / raw)
To: netdev; +Cc: bhutchings, davem, Giuseppe Cavallaro
In-Reply-To: <1346857432-24657-1-git-send-email-peppe.cavallaro@st.com>
This patch is to get/set the tx/rx coalesce parameters
via ethtool interface.
Tests have been done on several platform with
different GMAC chips w/o w/ RX watchdog feature.
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
---
drivers/net/ethernet/stmicro/stmmac/common.h | 8 ++-
drivers/net/ethernet/stmicro/stmmac/stmmac.h | 1 +
.../net/ethernet/stmicro/stmmac/stmmac_ethtool.c | 74 ++++++++++++++++++++
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 56 +++++----------
4 files changed, 100 insertions(+), 39 deletions(-)
diff --git a/drivers/net/ethernet/stmicro/stmmac/common.h b/drivers/net/ethernet/stmicro/stmmac/common.h
index 63d4bad..ee9cd8f 100644
--- a/drivers/net/ethernet/stmicro/stmmac/common.h
+++ b/drivers/net/ethernet/stmicro/stmmac/common.h
@@ -169,7 +169,13 @@ struct stmmac_extra_stats {
#define DMA_HW_FEAT_SAVLANINS 0x08000000 /* Source Addr or VLAN Insertion */
#define DMA_HW_FEAT_ACTPHYIF 0x70000000 /* Active/selected PHY interface */
#define DEFAULT_DMA_PBL 8
-#define DEFAULT_DMA_RIWT 0xff /* Max RI Watchdog Timer count */
+
+/* Coalesce defines */
+#define MAX_DMA_RIWT 0xff /* Max RI Watchdog Timer count */
+#define MIN_DMA_RIWT 0x20
+#define STMMAC_COAL_TX_TIMER 40000
+#define STMMAC_MAX_COAL_TX_TICK 100000
+#define STMMAC_TX_MAX_FRAMES(x) (x/4)
enum rx_frame_status { /* IPC status */
good_frame = 0,
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
index c113f28..577dc13 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
@@ -93,6 +93,7 @@ struct stmmac_priv {
u32 tx_count_frames;
u32 tx_coal_frames;
u32 tx_coal_timer;
+ u32 rx_riwt;
};
extern int phyaddr;
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c
index 48ad0bc..1428489 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c
@@ -522,6 +522,78 @@ static int stmmac_ethtool_op_set_eee(struct net_device *dev,
return phy_ethtool_set_eee(priv->phydev, edata);
}
+static u32 stmmac_usec2riwt(u32 usec, struct stmmac_priv *priv)
+{
+ unsigned long clk = clk_get_rate(priv->stmmac_clk);
+
+ if (!clk)
+ return 0;
+
+ return (usec * (clk / 1000000)) / 256;
+}
+
+static u32 stmmac_riwt2usec(u32 riwt, struct stmmac_priv *priv)
+{
+ unsigned long clk = clk_get_rate(priv->stmmac_clk);
+
+ if (!clk)
+ return 0;
+
+ return (riwt * 256) / (clk / 1000000);
+}
+
+static int stmmac_get_coalesce(struct net_device *dev,
+ struct ethtool_coalesce *ec)
+{
+ struct stmmac_priv *priv = netdev_priv(dev);
+
+ ec->tx_coalesce_usecs = priv->tx_coal_timer;
+ ec->tx_max_coalesced_frames = priv->tx_coal_frames;
+
+ if (!priv->napi_mode)
+ ec->rx_coalesce_usecs = stmmac_riwt2usec(priv->rx_riwt, priv);
+
+ return 0;
+}
+
+static int stmmac_set_coalesce(struct net_device *dev,
+ struct ethtool_coalesce *ec)
+{
+ struct stmmac_priv *priv = netdev_priv(dev);
+ unsigned int max_tx_coal_frames;
+ unsigned int rx_riwt;
+
+ /* No rx interrupts will be generated if both are zero */
+ if (ec->rx_coalesce_usecs == 0)
+ return -EINVAL;
+
+ /* No tx interrupts will be generated if both are zero */
+ if ((ec->tx_coalesce_usecs == 0) &&
+ (ec->tx_max_coalesced_frames == 0))
+ return -EINVAL;
+
+ max_tx_coal_frames = STMMAC_TX_MAX_FRAMES(priv->dma_tx_size);
+
+ if ((ec->tx_coalesce_usecs > STMMAC_COAL_TX_TIMER) ||
+ (ec->tx_max_coalesced_frames > max_tx_coal_frames))
+ return -EINVAL;
+
+ rx_riwt = stmmac_usec2riwt(ec->rx_coalesce_usecs, priv);
+
+ if ((rx_riwt > MAX_DMA_RIWT) || (rx_riwt < MIN_DMA_RIWT))
+ return -EINVAL;
+ else if (priv->napi_mode)
+ return -EOPNOTSUPP;
+
+ /* Only copy relevant parameters, ignore all others. */
+ priv->tx_coal_frames = ec->tx_max_coalesced_frames;
+ priv->tx_coal_timer = ec->tx_coalesce_usecs;
+ priv->rx_riwt = rx_riwt;
+ priv->hw->dma->rx_watchdog(priv->ioaddr, priv->rx_riwt);
+
+ return 0;
+}
+
static const struct ethtool_ops stmmac_ethtool_ops = {
.begin = stmmac_check_if_running,
.get_drvinfo = stmmac_ethtool_getdrvinfo,
@@ -542,6 +614,8 @@ static const struct ethtool_ops stmmac_ethtool_ops = {
.set_eee = stmmac_ethtool_op_set_eee,
.get_sset_count = stmmac_get_sset_count,
.get_ts_info = ethtool_op_get_ts_info,
+ .get_coalesce = stmmac_get_coalesce,
+ .set_coalesce = stmmac_set_coalesce,
};
void stmmac_set_ethtool_ops(struct net_device *netdev)
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index 55bb3c9..cea4196 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -77,8 +77,6 @@
#define STMMAC_ALIGN(x) L1_CACHE_ALIGN(x)
#define JUMBO_LEN 9000
-#define STMMAC_TX_TM 40000
-#define STMMAC_TX_MAX_FRAMES 64 /* Max coalesced frame */
/* Module parameters */
#define TX_TIMEO 5000 /* default 5 seconds */
@@ -140,6 +138,8 @@ static int stmmac_init_fs(struct net_device *dev);
static void stmmac_exit_fs(void);
#endif
+#define STMMAC_COAL_TIMER(x) (jiffies + usecs_to_jiffies(x))
+
/**
* stmmac_verify_args - verify the driver parameters.
* Description: it verifies if some wrong parameter is passed to the driver.
@@ -999,34 +999,16 @@ static int stmmac_init_dma_engine(struct stmmac_priv *priv)
priv->dma_rx_phy);
}
-static int stmmac_check_coal(int size, int max_coal_frames)
-{
- int ret = 0;
-
- if (max_coal_frames >= size)
- return ret;
-
- return max_coal_frames;
-}
-
-static int stmmac_init_tx_coalesce(struct stmmac_priv *priv)
+static void stmmac_init_tx_coalesce(struct stmmac_priv *priv)
{
- int ret = -EOPNOTSUPP;
-
- priv->tx_coal_frames = stmmac_check_coal(priv->dma_tx_size,
- STMMAC_TX_MAX_FRAMES);
- if (priv->tx_coal_frames) {
- /* Set Tx coalesce parameters and timers */
- priv->tx_coal_timer = jiffies + usecs_to_jiffies(STMMAC_TX_TM);
- init_timer(&priv->txtimer);
- priv->txtimer.expires = priv->tx_coal_timer;
- priv->txtimer.data = (unsigned long)priv;
- priv->txtimer.function = stmmac_txtimer;
-
- ret = 0;
- }
-
- return ret;
+ /* Set Tx coalesce parameters and timers */
+ priv->tx_coal_frames = STMMAC_TX_MAX_FRAMES(priv->dma_tx_size);
+ priv->tx_coal_timer = STMMAC_COAL_TX_TIMER;
+ init_timer(&priv->txtimer);
+ priv->txtimer.expires = STMMAC_COAL_TIMER(priv->tx_coal_timer);
+ priv->txtimer.data = (unsigned long)priv;
+ priv->txtimer.function = stmmac_txtimer;
+ add_timer(&priv->txtimer);
}
/**
@@ -1141,20 +1123,17 @@ static int stmmac_open(struct net_device *dev)
priv->tx_lpi_timer = STMMAC_DEFAULT_TWT_LS_TIMER;
priv->eee_enabled = stmmac_eee_init(priv);
- ret = stmmac_init_tx_coalesce(priv);
- if (!ret)
- add_timer(&priv->txtimer);
+ stmmac_init_tx_coalesce(priv);
/* Enable NAPI on chip older than the 3.50 where the Rx watchdog
* is not supported.
*/
if (priv->napi_mode)
napi_enable(&priv->napi);
- else if (priv->hw->dma->rx_watchdog)
- /* Program RX Watchdog register to the default values
- * FIXME: provide user value for RIWT
- */
- priv->hw->dma->rx_watchdog(priv->ioaddr, DEFAULT_DMA_RIWT);
+ else if (priv->hw->dma->rx_watchdog) {
+ priv->rx_riwt = MAX_DMA_RIWT;
+ priv->hw->dma->rx_watchdog(priv->ioaddr, MAX_DMA_RIWT);
+ }
skb_queue_head_init(&priv->rx_recycle);
netif_start_queue(dev);
@@ -1318,7 +1297,8 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev)
if (priv->tx_coal_frames > priv->tx_count_frames) {
priv->hw->desc->clear_tx_ic(desc);
priv->xstats.tx_reset_ic_bit++;
- mod_timer(&priv->txtimer, priv->tx_coal_timer);
+ mod_timer(&priv->txtimer,
+ STMMAC_COAL_TIMER(priv->tx_coal_timer));
} else
priv->tx_count_frames = 0;
--
1.7.4.4
^ permalink raw reply related
* [net-next.git 4/7] stmmac: add Rx watchdog optimization to mitigate the DMA irqs
From: Giuseppe CAVALLARO @ 2012-09-05 15:03 UTC (permalink / raw)
To: netdev; +Cc: bhutchings, davem, Giuseppe Cavallaro
In-Reply-To: <1346857432-24657-1-git-send-email-peppe.cavallaro@st.com>
GMAC devices newer than databook 3.50 has an embedded timer
that can be used for mitigating the number of interrupts.
So this patch adds this optimizations.
Old MAC will continue to use NAPI.
At any rate, the Rx watchdog can be disable (on bugged HW) by
passing from the platform the riwt_off field.
In this implementation the rx timer stored in the Reg9 is fixed
to the max value.
V2: added a platform parameter to force to disable the rx-watchdog
for example on new core where it is bugged.
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
---
drivers/net/ethernet/stmicro/stmmac/common.h | 7 ++
drivers/net/ethernet/stmicro/stmmac/dwmac1000.h | 3 -
.../net/ethernet/stmicro/stmmac/dwmac1000_dma.c | 6 ++
drivers/net/ethernet/stmicro/stmmac/dwmac_dma.h | 3 +-
drivers/net/ethernet/stmicro/stmmac/stmmac.h | 1 +
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 72 ++++++++++++++------
include/linux/stmmac.h | 1 +
7 files changed, 68 insertions(+), 25 deletions(-)
diff --git a/drivers/net/ethernet/stmicro/stmmac/common.h b/drivers/net/ethernet/stmicro/stmmac/common.h
index 1d6bd3e..63d4bad 100644
--- a/drivers/net/ethernet/stmicro/stmmac/common.h
+++ b/drivers/net/ethernet/stmicro/stmmac/common.h
@@ -48,6 +48,10 @@
#define CHIP_DBG(fmt, args...) do { } while (0)
#endif
+/* Synopsys Core versions */
+#define DWMAC_CORE_3_40 0x34
+#define DWMAC_CORE_3_50 0x35
+
#undef FRAME_FILTER_DEBUG
/* #define FRAME_FILTER_DEBUG */
@@ -165,6 +169,7 @@ struct stmmac_extra_stats {
#define DMA_HW_FEAT_SAVLANINS 0x08000000 /* Source Addr or VLAN Insertion */
#define DMA_HW_FEAT_ACTPHYIF 0x70000000 /* Active/selected PHY interface */
#define DEFAULT_DMA_PBL 8
+#define DEFAULT_DMA_RIWT 0xff /* Max RI Watchdog Timer count */
enum rx_frame_status { /* IPC status */
good_frame = 0,
@@ -301,6 +306,8 @@ struct stmmac_dma_ops {
struct stmmac_extra_stats *x);
/* If supported then get the optional core features */
unsigned int (*get_hw_feature) (void __iomem *ioaddr);
+ /* Manage HW RX Watchdog*/
+ void (*rx_watchdog) (void __iomem *ioaddr, u8 timer);
};
struct stmmac_ops {
diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac1000.h b/drivers/net/ethernet/stmicro/stmmac/dwmac1000.h
index 0e4cace..7ad56af 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac1000.h
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac1000.h
@@ -230,8 +230,5 @@ enum rtc_control {
#define GMAC_MMC_TX_INTR 0x108
#define GMAC_MMC_RX_CSUM_OFFLOAD 0x208
-/* Synopsys Core versions */
-#define DWMAC_CORE_3_40 0x34
-
extern const struct stmmac_dma_ops dwmac1000_dma_ops;
#endif /* __DWMAC1000_H__ */
diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac1000_dma.c b/drivers/net/ethernet/stmicro/stmmac/dwmac1000_dma.c
index 0335000..e2c9431 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac1000_dma.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac1000_dma.c
@@ -174,6 +174,11 @@ static unsigned int dwmac1000_get_hw_feature(void __iomem *ioaddr)
return readl(ioaddr + DMA_HW_FEATURE);
}
+static void dwmac1000_rx_watchdog(void __iomem *ioaddr, u8 timer)
+{
+ writel(timer, ioaddr + DMA_RX_WATCHDOG);
+}
+
const struct stmmac_dma_ops dwmac1000_dma_ops = {
.init = dwmac1000_dma_init,
.dump_regs = dwmac1000_dump_dma_regs,
@@ -187,4 +192,5 @@ const struct stmmac_dma_ops dwmac1000_dma_ops = {
.stop_rx = dwmac_dma_stop_rx,
.dma_interrupt = dwmac_dma_interrupt,
.get_hw_feature = dwmac1000_get_hw_feature,
+ .rx_watchdog = dwmac1000_rx_watchdog,
};
diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac_dma.h b/drivers/net/ethernet/stmicro/stmmac/dwmac_dma.h
index e49c9a0..4eeff5d 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac_dma.h
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac_dma.h
@@ -35,7 +35,8 @@
#define DMA_CONTROL 0x00001018 /* Ctrl (Operational Mode) */
#define DMA_INTR_ENA 0x0000101c /* Interrupt Enable */
#define DMA_MISSED_FRAME_CTR 0x00001020 /* Missed Frame Counter */
-#define DMA_AXI_BUS_MODE 0x00001028 /* AXI Bus Mode */
+#define DMA_RX_WATCHDOG 0x00001024 /* Receive Int Watchdog Timer */
+#define DMA_AXI_BUS_MODE 0x00001028 /* AXI Bus Mode */
#define DMA_CUR_TX_BUF_ADDR 0x00001050 /* Current Host Tx Buffer */
#define DMA_CUR_RX_BUF_ADDR 0x00001054 /* Current Host Rx Buffer */
#define DMA_HW_FEATURE 0x00001058 /* HW Feature Register */
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
index 0f5ab28..c113f28 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
@@ -89,6 +89,7 @@ struct stmmac_priv {
int eee_active;
int tx_lpi_timer;
struct timer_list txtimer;
+ int napi_mode;
u32 tx_count_frames;
u32 tx_coal_frames;
u32 tx_coal_timer;
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index d7f5482..55bb3c9 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -133,6 +133,7 @@ MODULE_PARM_DESC(eee_timer, "LPI tx expiration time in msec");
#define STMMAC_LPI_TIMER(x) (jiffies + msecs_to_jiffies(x))
static irqreturn_t stmmac_interrupt(int irq, void *dev_id);
+static int stmmac_rx(struct stmmac_priv *priv, int limit);
#ifdef CONFIG_STMMAC_DEBUG_FS
static int stmmac_init_fs(struct net_device *dev);
@@ -516,7 +517,7 @@ static void init_dma_desc_rings(struct net_device *dev)
unsigned int txsize = priv->dma_tx_size;
unsigned int rxsize = priv->dma_rx_size;
unsigned int bfsize;
- int dis_ic = 0;
+ int dis_ic = 1;
int des3_as_data_buf = 0;
/* Set the max buffer size according to the DESC mode
@@ -603,6 +604,8 @@ static void init_dma_desc_rings(struct net_device *dev)
priv->dirty_tx = 0;
priv->cur_tx = 0;
+ if (priv->napi_mode)
+ dis_ic = 0;
/* Clear the Rx/Tx descriptors */
priv->hw->desc->init_rx_desc(priv->dma_rx, rxsize, dis_ic);
priv->hw->desc->init_tx_desc(priv->dma_tx, txsize);
@@ -746,7 +749,7 @@ static void stmmac_tx(struct stmmac_priv *priv)
skb_recycle_check(skb, priv->dma_buf_sz))
__skb_queue_head(&priv->rx_recycle, skb);
else
- dev_kfree_skb(skb);
+ dev_kfree_skb_any(skb);
priv->tx_skbuff[entry] = NULL;
}
@@ -816,12 +819,15 @@ static void stmmac_tx_err(struct stmmac_priv *priv)
netif_wake_queue(priv->dev);
}
-static void stmmac_rx_schedule(struct stmmac_priv *priv)
+static void stmmac_rx_work(struct stmmac_priv *priv)
{
- if (likely(napi_schedule_prep(&priv->napi))) {
- stmmac_disable_irq(priv);
- __napi_schedule(&priv->napi);
- }
+ if (priv->napi_mode) {
+ if (likely(napi_schedule_prep(&priv->napi))) {
+ stmmac_disable_irq(priv);
+ __napi_schedule(&priv->napi);
+ }
+ } else
+ stmmac_rx(priv, priv->dma_rx_size);
}
static void stmmac_dma_interrupt(struct stmmac_priv *priv)
@@ -831,7 +837,7 @@ static void stmmac_dma_interrupt(struct stmmac_priv *priv)
status = priv->hw->dma->dma_interrupt(priv->ioaddr, &priv->xstats);
if (likely(status == handle_rx)) {
priv->xstats.rx_normal_irq_n++;
- stmmac_rx_schedule(priv);
+ stmmac_rx_work(priv);
}
if (likely(status == handle_tx)) {
priv->xstats.tx_normal_irq_n++;
@@ -1139,7 +1145,17 @@ static int stmmac_open(struct net_device *dev)
if (!ret)
add_timer(&priv->txtimer);
- napi_enable(&priv->napi);
+ /* Enable NAPI on chip older than the 3.50 where the Rx watchdog
+ * is not supported.
+ */
+ if (priv->napi_mode)
+ napi_enable(&priv->napi);
+ else if (priv->hw->dma->rx_watchdog)
+ /* Program RX Watchdog register to the default values
+ * FIXME: provide user value for RIWT
+ */
+ priv->hw->dma->rx_watchdog(priv->ioaddr, DEFAULT_DMA_RIWT);
+
skb_queue_head_init(&priv->rx_recycle);
netif_start_queue(dev);
@@ -1183,7 +1199,8 @@ static int stmmac_release(struct net_device *dev)
netif_stop_queue(dev);
- napi_disable(&priv->napi);
+ if (priv->napi_mode)
+ napi_disable(&priv->napi);
skb_queue_purge(&priv->rx_recycle);
/* Free the IRQ lines */
@@ -1448,14 +1465,15 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit)
#endif
skb->protocol = eth_type_trans(skb, priv->dev);
- if (unlikely(!priv->plat->rx_coe)) {
- /* No RX COE for old mac10/100 devices */
+ if (unlikely(!priv->plat->rx_coe))
skb_checksum_none_assert(skb);
- netif_receive_skb(skb);
- } else {
+ else
skb->ip_summed = CHECKSUM_UNNECESSARY;
+
+ if (priv->napi_mode)
napi_gro_receive(&priv->napi, skb);
- }
+ else
+ netif_rx(skb);
priv->dev->stats.rx_packets++;
priv->dev->stats.rx_bytes += frame_len;
@@ -2025,7 +2043,15 @@ struct stmmac_priv *stmmac_dvr_probe(struct device *device,
if (flow_ctrl)
priv->flow_ctrl = FLOW_AUTO; /* RX/TX pause on */
- netif_napi_add(ndev, &priv->napi, stmmac_poll, 64);
+ /* Rx Watchdog is available in MAC newer than the 3.40.
+ * In some case, for example on bugged HW, this feature
+ * has to be disable and this can be done by passing the
+ * riwt_off field from the platform. In this case we will use NAPI. */
+ if ((priv->synopsys_id < DWMAC_CORE_3_50) || (priv->plat->riwt_off)) {
+ priv->napi_mode = 1;
+ netif_napi_add(ndev, &priv->napi, stmmac_poll, 64);
+ } else
+ pr_info(" Enable RX mitigation via HW Watchdog Timer\n");
spin_lock_init(&priv->lock);
spin_lock_init(&priv->tx_lock);
@@ -2068,7 +2094,8 @@ error_mdio_register:
error_clk_get:
unregister_netdev(ndev);
error_netdev_register:
- netif_napi_del(&priv->napi);
+ if (priv->napi_mode)
+ netif_napi_del(&priv->napi);
free_netdev(ndev);
return NULL;
@@ -2102,7 +2129,7 @@ int stmmac_dvr_remove(struct net_device *ndev)
int stmmac_suspend(struct net_device *ndev)
{
struct stmmac_priv *priv = netdev_priv(ndev);
- int dis_ic = 0;
+ int dis_ic = 1;
unsigned long flags;
if (!ndev || !netif_running(ndev))
@@ -2116,8 +2143,10 @@ int stmmac_suspend(struct net_device *ndev)
netif_device_detach(ndev);
netif_stop_queue(ndev);
- napi_disable(&priv->napi);
-
+ if (priv->napi_mode) {
+ dis_ic = 0;
+ napi_disable(&priv->napi);
+ }
/* Stop TX/RX DMA */
priv->hw->dma->stop_tx(priv->ioaddr);
priv->hw->dma->stop_rx(priv->ioaddr);
@@ -2166,7 +2195,8 @@ int stmmac_resume(struct net_device *ndev)
priv->hw->dma->start_tx(priv->ioaddr);
priv->hw->dma->start_rx(priv->ioaddr);
- napi_enable(&priv->napi);
+ if (priv->napi_mode)
+ napi_enable(&priv->napi);
netif_start_queue(ndev);
diff --git a/include/linux/stmmac.h b/include/linux/stmmac.h
index a1547ea..de5b2f8 100644
--- a/include/linux/stmmac.h
+++ b/include/linux/stmmac.h
@@ -104,6 +104,7 @@ struct plat_stmmacenet_data {
int bugged_jumbo;
int pmt;
int force_sf_dma_mode;
+ int riwt_off;
void (*fix_mac_speed)(void *priv, unsigned int speed);
void (*bus_setup)(void __iomem *ioaddr);
int (*init)(struct platform_device *pdev);
--
1.7.4.4
^ permalink raw reply related
* [net-next.git 3/7] stmmac: add the initial tx coalesce schema
From: Giuseppe CAVALLARO @ 2012-09-05 15:03 UTC (permalink / raw)
To: netdev; +Cc: bhutchings, davem, Giuseppe Cavallaro
In-Reply-To: <1346857432-24657-1-git-send-email-peppe.cavallaro@st.com>
This patch adds a new schema used for mitigating the
number of transmit interrupts.
It is based on a sw timer and a threshold value.
The timer is used to periodically call the stmmac_tx
function that can be invoked by the ISR but only for
the descriptors where the interrupt on completion
field has been set. This is tuned by a threshold.
Next step is to add the ability to tune these coalesce
values by ethtool.
Till now I have put a default that showed a real gain
on all the platforms ARM/SH4 where I performed benchmarks.
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
---
drivers/net/ethernet/stmicro/stmmac/common.h | 8 +-
drivers/net/ethernet/stmicro/stmmac/stmmac.h | 4 +
.../net/ethernet/stmicro/stmmac/stmmac_ethtool.c | 9 +-
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 86 +++++++++++++-------
4 files changed, 72 insertions(+), 35 deletions(-)
diff --git a/drivers/net/ethernet/stmicro/stmmac/common.h b/drivers/net/ethernet/stmicro/stmmac/common.h
index bd32fe6..1d6bd3e 100644
--- a/drivers/net/ethernet/stmicro/stmmac/common.h
+++ b/drivers/net/ethernet/stmicro/stmmac/common.h
@@ -95,11 +95,13 @@ struct stmmac_extra_stats {
unsigned long threshold;
unsigned long tx_pkt_n;
unsigned long rx_pkt_n;
- unsigned long rx_napi_poll;
+ unsigned long normal_irq_n;
unsigned long rx_normal_irq_n;
+ unsigned long rx_napi_poll;
unsigned long tx_normal_irq_n;
- unsigned long sched_timer_n;
- unsigned long normal_irq_n;
+ unsigned long txtimer;
+ unsigned long tx_clean;
+ unsigned long tx_reset_ic_bit;
unsigned long mmc_tx_irq_n;
unsigned long mmc_rx_irq_n;
unsigned long mmc_rx_csum_offload_irq_n;
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
index 9f35769..0f5ab28 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
@@ -88,6 +88,10 @@ struct stmmac_priv {
int eee_enabled;
int eee_active;
int tx_lpi_timer;
+ struct timer_list txtimer;
+ u32 tx_count_frames;
+ u32 tx_coal_frames;
+ u32 tx_coal_timer;
};
extern int phyaddr;
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c
index 505fe71..48ad0bc 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c
@@ -90,12 +90,13 @@ static const struct stmmac_stats stmmac_gstrings_stats[] = {
STMMAC_STAT(threshold),
STMMAC_STAT(tx_pkt_n),
STMMAC_STAT(rx_pkt_n),
- STMMAC_STAT(rx_napi_poll),
+ STMMAC_STAT(normal_irq_n),
STMMAC_STAT(rx_normal_irq_n),
+ STMMAC_STAT(rx_napi_poll),
STMMAC_STAT(tx_normal_irq_n),
- STMMAC_STAT(sched_timer_n),
- STMMAC_STAT(normal_irq_n),
- STMMAC_STAT(normal_irq_n),
+ STMMAC_STAT(txtimer),
+ STMMAC_STAT(tx_clean),
+ STMMAC_STAT(tx_reset_ic_bit),
STMMAC_STAT(mmc_tx_irq_n),
STMMAC_STAT(mmc_rx_irq_n),
STMMAC_STAT(mmc_rx_csum_offload_irq_n),
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index b247c39..d7f5482 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -77,6 +77,8 @@
#define STMMAC_ALIGN(x) L1_CACHE_ALIGN(x)
#define JUMBO_LEN 9000
+#define STMMAC_TX_TM 40000
+#define STMMAC_TX_MAX_FRAMES 64 /* Max coalesced frame */
/* Module parameters */
#define TX_TIMEO 5000 /* default 5 seconds */
@@ -695,8 +697,11 @@ static void stmmac_dma_operation_mode(struct stmmac_priv *priv)
static void stmmac_tx(struct stmmac_priv *priv)
{
unsigned int txsize = priv->dma_tx_size;
+ unsigned long flags;
+
+ spin_lock_irqsave(&priv->tx_lock, flags);
- spin_lock(&priv->tx_lock);
+ priv->xstats.tx_clean++;
while (priv->dirty_tx != priv->cur_tx) {
int last;
@@ -765,7 +770,7 @@ static void stmmac_tx(struct stmmac_priv *priv)
stmmac_enable_eee_mode(priv);
mod_timer(&priv->eee_ctrl_timer, STMMAC_LPI_TIMER(eee_timer));
}
- spin_unlock(&priv->tx_lock);
+ spin_unlock_irqrestore(&priv->tx_lock, flags);
}
static inline void stmmac_enable_irq(struct stmmac_priv *priv)
@@ -778,29 +783,16 @@ static inline void stmmac_disable_irq(struct stmmac_priv *priv)
priv->hw->dma->disable_dma_irq(priv->ioaddr);
}
-static int stmmac_has_work(struct stmmac_priv *priv)
+static void stmmac_txtimer(unsigned long data)
{
- unsigned int has_work = 0;
- int rxret, tx_work = 0;
+ struct stmmac_priv *priv = (struct stmmac_priv *)data;
- rxret = priv->hw->desc->get_rx_owner(priv->dma_rx +
- (priv->cur_rx % priv->dma_rx_size));
+ priv->xstats.txtimer++;
if (priv->dirty_tx != priv->cur_tx)
- tx_work = 1;
-
- if (likely(!rxret || tx_work))
- has_work = 1;
+ stmmac_tx(priv);
- return has_work;
-}
-
-static inline void _stmmac_schedule(struct stmmac_priv *priv)
-{
- if (likely(stmmac_has_work(priv))) {
- stmmac_disable_irq(priv);
- napi_schedule(&priv->napi);
- }
+ return;
}
/**
@@ -824,7 +816,7 @@ static void stmmac_tx_err(struct stmmac_priv *priv)
netif_wake_queue(priv->dev);
}
-static inline void stmmac_rx_schedule(struct stmmac_priv *priv)
+static void stmmac_rx_schedule(struct stmmac_priv *priv)
{
if (likely(napi_schedule_prep(&priv->napi))) {
stmmac_disable_irq(priv);
@@ -1001,6 +993,36 @@ static int stmmac_init_dma_engine(struct stmmac_priv *priv)
priv->dma_rx_phy);
}
+static int stmmac_check_coal(int size, int max_coal_frames)
+{
+ int ret = 0;
+
+ if (max_coal_frames >= size)
+ return ret;
+
+ return max_coal_frames;
+}
+
+static int stmmac_init_tx_coalesce(struct stmmac_priv *priv)
+{
+ int ret = -EOPNOTSUPP;
+
+ priv->tx_coal_frames = stmmac_check_coal(priv->dma_tx_size,
+ STMMAC_TX_MAX_FRAMES);
+ if (priv->tx_coal_frames) {
+ /* Set Tx coalesce parameters and timers */
+ priv->tx_coal_timer = jiffies + usecs_to_jiffies(STMMAC_TX_TM);
+ init_timer(&priv->txtimer);
+ priv->txtimer.expires = priv->tx_coal_timer;
+ priv->txtimer.data = (unsigned long)priv;
+ priv->txtimer.function = stmmac_txtimer;
+
+ ret = 0;
+ }
+
+ return ret;
+}
+
/**
* stmmac_open - open entry point of the driver
* @dev : pointer to the device structure.
@@ -1113,6 +1135,10 @@ static int stmmac_open(struct net_device *dev)
priv->tx_lpi_timer = STMMAC_DEFAULT_TWT_LS_TIMER;
priv->eee_enabled = stmmac_eee_init(priv);
+ ret = stmmac_init_tx_coalesce(priv);
+ if (!ret)
+ add_timer(&priv->txtimer);
+
napi_enable(&priv->napi);
skb_queue_head_init(&priv->rx_recycle);
netif_start_queue(dev);
@@ -1202,6 +1228,7 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev)
int nfrags = skb_shinfo(skb)->nr_frags;
struct dma_desc *desc, *first;
unsigned int nopaged_len = skb_headlen(skb);
+ unsigned long flags;
if (unlikely(stmmac_tx_avail(priv) < nfrags + 1)) {
if (!netif_queue_stopped(dev)) {
@@ -1213,10 +1240,7 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev)
return NETDEV_TX_BUSY;
}
- spin_lock(&priv->tx_lock);
-
- if (priv->tx_path_in_lpi_mode)
- stmmac_disable_eee_mode(priv);
+ spin_lock_irqsave(&priv->tx_lock, flags);
entry = priv->cur_tx % txsize;
@@ -1272,7 +1296,14 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev)
/* Interrupt on completition only for the latest segment */
priv->hw->desc->close_tx_desc(desc);
- wmb();
+ /* Do not set the IC according to the coalesce patameters */
+ priv->tx_count_frames++;
+ if (priv->tx_coal_frames > priv->tx_count_frames) {
+ priv->hw->desc->clear_tx_ic(desc);
+ priv->xstats.tx_reset_ic_bit++;
+ mod_timer(&priv->txtimer, priv->tx_coal_timer);
+ } else
+ priv->tx_count_frames = 0;
/* To avoid raise condition */
priv->hw->desc->set_tx_owner(first);
@@ -1302,7 +1333,7 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev)
priv->hw->dma->enable_dma_transmission(priv->ioaddr);
- spin_unlock(&priv->tx_lock);
+ spin_unlock_irqrestore(&priv->tx_lock, flags);
return NETDEV_TX_OK;
}
@@ -1447,7 +1478,6 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit)
* all interfaces.
* Description :
* This function implements the the reception process.
- * Also it runs the TX completion thread
*/
static int stmmac_poll(struct napi_struct *napi, int budget)
{
--
1.7.4.4
^ permalink raw reply related
* [net-next.git 1/7] stmmac: remove dead code for TIMER
From: Giuseppe CAVALLARO @ 2012-09-05 15:03 UTC (permalink / raw)
To: netdev; +Cc: bhutchings, davem, Giuseppe Cavallaro
In-Reply-To: <1346857432-24657-1-git-send-email-peppe.cavallaro@st.com>
TIMER option is not longer supported and this
code can be considered dead for this driver in
the new kernel series.
In fact, It was not updated at all and never used.
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
---
drivers/net/ethernet/stmicro/stmmac/Kconfig | 25 ----
drivers/net/ethernet/stmicro/stmmac/Makefile | 1 -
drivers/net/ethernet/stmicro/stmmac/stmmac.h | 6 -
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 101 +--------------
drivers/net/ethernet/stmicro/stmmac/stmmac_timer.c | 134 --------------------
drivers/net/ethernet/stmicro/stmmac/stmmac_timer.h | 46 -------
6 files changed, 3 insertions(+), 310 deletions(-)
delete mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_timer.c
delete mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_timer.h
diff --git a/drivers/net/ethernet/stmicro/stmmac/Kconfig b/drivers/net/ethernet/stmicro/stmmac/Kconfig
index 9f44827..1164930 100644
--- a/drivers/net/ethernet/stmicro/stmmac/Kconfig
+++ b/drivers/net/ethernet/stmicro/stmmac/Kconfig
@@ -54,31 +54,6 @@ config STMMAC_DA
By default, the DMA arbitration scheme is based on Round-robin
(rx:tx priority is 1:1).
-config STMMAC_TIMER
- bool "STMMAC Timer optimisation"
- default n
- depends on RTC_HCTOSYS_DEVICE
- ---help---
- Use an external timer for mitigating the number of network
- interrupts. Currently, for SH architectures, it is possible
- to use the TMU channel 2 and the SH-RTC device.
-
-choice
- prompt "Select Timer device"
- depends on STMMAC_TIMER
-
-config STMMAC_TMU_TIMER
- bool "TMU channel 2"
- depends on CPU_SH4
- ---help---
-
-config STMMAC_RTC_TIMER
- bool "Real time clock"
- depends on RTC_CLASS
- ---help---
-
-endchoice
-
choice
prompt "Select the DMA TX/RX descriptor operating modes"
depends on STMMAC_ETH
diff --git a/drivers/net/ethernet/stmicro/stmmac/Makefile b/drivers/net/ethernet/stmicro/stmmac/Makefile
index bc965ac..c8e8ea6 100644
--- a/drivers/net/ethernet/stmicro/stmmac/Makefile
+++ b/drivers/net/ethernet/stmicro/stmmac/Makefile
@@ -1,5 +1,4 @@
obj-$(CONFIG_STMMAC_ETH) += stmmac.o
-stmmac-$(CONFIG_STMMAC_TIMER) += stmmac_timer.o
stmmac-$(CONFIG_STMMAC_RING) += ring_mode.o
stmmac-$(CONFIG_STMMAC_CHAINED) += chain_mode.o
stmmac-$(CONFIG_STMMAC_PLATFORM) += stmmac_platform.o
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
index e872e1d..9f35769 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
@@ -31,9 +31,6 @@
#include <linux/phy.h>
#include <linux/pci.h>
#include "common.h"
-#ifdef CONFIG_STMMAC_TIMER
-#include "stmmac_timer.h"
-#endif
struct stmmac_priv {
/* Frequently used values are kept adjacent for cache effect */
@@ -78,9 +75,6 @@ struct stmmac_priv {
spinlock_t tx_lock;
int wolopts;
int wol_irq;
-#ifdef CONFIG_STMMAC_TIMER
- struct stmmac_timer *tm;
-#endif
struct plat_stmmacenet_data *plat;
struct stmmac_counters mmc;
struct dma_features dma_cap;
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index c136162..c8985f3 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -115,16 +115,6 @@ static int tc = TC_DEFAULT;
module_param(tc, int, S_IRUGO | S_IWUSR);
MODULE_PARM_DESC(tc, "DMA threshold control value");
-/* Pay attention to tune this parameter; take care of both
- * hardware capability and network stabitily/performance impact.
- * Many tests showed that ~4ms latency seems to be good enough. */
-#ifdef CONFIG_STMMAC_TIMER
-#define DEFAULT_PERIODIC_RATE 256
-static int tmrate = DEFAULT_PERIODIC_RATE;
-module_param(tmrate, int, S_IRUGO | S_IWUSR);
-MODULE_PARM_DESC(tmrate, "External timer freq. (default: 256Hz)");
-#endif
-
#define DMA_BUFFER_SIZE BUF_SIZE_2KiB
static int buf_sz = DMA_BUFFER_SIZE;
module_param(buf_sz, int, S_IRUGO | S_IWUSR);
@@ -536,12 +526,6 @@ static void init_dma_desc_rings(struct net_device *dev)
else
bfsize = stmmac_set_bfsize(dev->mtu, priv->dma_buf_sz);
-#ifdef CONFIG_STMMAC_TIMER
- /* Disable interrupts on completion for the reception if timer is on */
- if (likely(priv->tm->enable))
- dis_ic = 1;
-#endif
-
DBG(probe, INFO, "stmmac: txsize %d, rxsize %d, bfsize %d\n",
txsize, rxsize, bfsize);
@@ -786,22 +770,12 @@ static void stmmac_tx(struct stmmac_priv *priv)
static inline void stmmac_enable_irq(struct stmmac_priv *priv)
{
-#ifdef CONFIG_STMMAC_TIMER
- if (likely(priv->tm->enable))
- priv->tm->timer_start(tmrate);
- else
-#endif
- priv->hw->dma->enable_dma_irq(priv->ioaddr);
+ priv->hw->dma->enable_dma_irq(priv->ioaddr);
}
static inline void stmmac_disable_irq(struct stmmac_priv *priv)
{
-#ifdef CONFIG_STMMAC_TIMER
- if (likely(priv->tm->enable))
- priv->tm->timer_stop();
- else
-#endif
- priv->hw->dma->disable_dma_irq(priv->ioaddr);
+ priv->hw->dma->disable_dma_irq(priv->ioaddr);
}
static int stmmac_has_work(struct stmmac_priv *priv)
@@ -829,25 +803,6 @@ static inline void _stmmac_schedule(struct stmmac_priv *priv)
}
}
-#ifdef CONFIG_STMMAC_TIMER
-void stmmac_schedule(struct net_device *dev)
-{
- struct stmmac_priv *priv = netdev_priv(dev);
-
- priv->xstats.sched_timer_n++;
-
- _stmmac_schedule(priv);
-}
-
-static void stmmac_no_timer_started(unsigned int x)
-{;
-};
-
-static void stmmac_no_timer_stopped(void)
-{;
-};
-#endif
-
/**
* stmmac_tx_err:
* @priv: pointer to the private device structure
@@ -1049,23 +1004,6 @@ static int stmmac_open(struct net_device *dev)
struct stmmac_priv *priv = netdev_priv(dev);
int ret;
-#ifdef CONFIG_STMMAC_TIMER
- priv->tm = kzalloc(sizeof(struct stmmac_timer *), GFP_KERNEL);
- if (unlikely(priv->tm == NULL))
- return -ENOMEM;
-
- priv->tm->freq = tmrate;
-
- /* Test if the external timer can be actually used.
- * In case of failure continue without timer. */
- if (unlikely((stmmac_open_ext_timer(dev, priv->tm)) < 0)) {
- pr_warning("stmmaceth: cannot attach the external timer.\n");
- priv->tm->freq = 0;
- priv->tm->timer_start = stmmac_no_timer_started;
- priv->tm->timer_stop = stmmac_no_timer_stopped;
- } else
- priv->tm->enable = 1;
-#endif
clk_enable(priv->stmmac_clk);
stmmac_check_ether_addr(priv);
@@ -1152,10 +1090,6 @@ static int stmmac_open(struct net_device *dev)
priv->hw->dma->start_tx(priv->ioaddr);
priv->hw->dma->start_rx(priv->ioaddr);
-#ifdef CONFIG_STMMAC_TIMER
- priv->tm->timer_start(tmrate);
-#endif
-
/* Dump DMA/MAC registers */
if (netif_msg_hw(priv)) {
priv->hw->mac->dump_regs(priv->ioaddr);
@@ -1182,9 +1116,6 @@ open_error_wolirq:
free_irq(dev->irq, dev);
open_error:
-#ifdef CONFIG_STMMAC_TIMER
- kfree(priv->tm);
-#endif
if (priv->phydev)
phy_disconnect(priv->phydev);
@@ -1215,12 +1146,6 @@ static int stmmac_release(struct net_device *dev)
netif_stop_queue(dev);
-#ifdef CONFIG_STMMAC_TIMER
- /* Stop and release the timer */
- stmmac_close_ext_timer();
- if (priv->tm != NULL)
- kfree(priv->tm);
-#endif
napi_disable(&priv->napi);
skb_queue_purge(&priv->rx_recycle);
@@ -1336,12 +1261,6 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev)
/* Interrupt on completition only for the latest segment */
priv->hw->desc->close_tx_desc(desc);
-#ifdef CONFIG_STMMAC_TIMER
- /* Clean IC while using timer */
- if (likely(priv->tm->enable))
- priv->hw->desc->clear_tx_ic(desc);
-#endif
-
wmb();
/* To avoid raise condition */
@@ -1539,7 +1458,7 @@ static int stmmac_poll(struct napi_struct *napi, int budget)
* stmmac_tx_timeout
* @dev : Pointer to net device structure
* Description: this function is called when a packet transmission fails to
- * complete within a reasonable tmrate. The driver will mark the error in the
+ * complete within a reasonable time. The driver will mark the error in the
* netdev structure and arrange for the device to be reset to a sane state
* in order to transmit a new packet.
*/
@@ -2157,11 +2076,6 @@ int stmmac_suspend(struct net_device *ndev)
netif_device_detach(ndev);
netif_stop_queue(ndev);
-#ifdef CONFIG_STMMAC_TIMER
- priv->tm->timer_stop();
- if (likely(priv->tm->enable))
- dis_ic = 1;
-#endif
napi_disable(&priv->napi);
/* Stop TX/RX DMA */
@@ -2212,10 +2126,6 @@ int stmmac_resume(struct net_device *ndev)
priv->hw->dma->start_tx(priv->ioaddr);
priv->hw->dma->start_rx(priv->ioaddr);
-#ifdef CONFIG_STMMAC_TIMER
- if (likely(priv->tm->enable))
- priv->tm->timer_start(tmrate);
-#endif
napi_enable(&priv->napi);
netif_start_queue(ndev);
@@ -2311,11 +2221,6 @@ static int __init stmmac_cmdline_opt(char *str)
} else if (!strncmp(opt, "eee_timer:", 6)) {
if (kstrtoint(opt + 10, 0, &eee_timer))
goto err;
-#ifdef CONFIG_STMMAC_TIMER
- } else if (!strncmp(opt, "tmrate:", 7)) {
- if (kstrtoint(opt + 7, 0, &tmrate))
- goto err;
-#endif
}
}
return 0;
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_timer.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_timer.c
deleted file mode 100644
index 2a0e1ab..0000000
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_timer.c
+++ /dev/null
@@ -1,134 +0,0 @@
-/*******************************************************************************
- STMMAC external timer support.
-
- Copyright (C) 2007-2009 STMicroelectronics Ltd
-
- This program is free software; you can redistribute it and/or modify it
- under the terms and conditions of the GNU General Public License,
- version 2, as published by the Free Software Foundation.
-
- This program is distributed in the hope it will be useful, but WITHOUT
- ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
- more details.
-
- You should have received a copy of the GNU General Public License along with
- this program; if not, write to the Free Software Foundation, Inc.,
- 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
-
- The full GNU General Public License is included in this distribution in
- the file called "COPYING".
-
- Author: Giuseppe Cavallaro <peppe.cavallaro@st.com>
-*******************************************************************************/
-
-#include <linux/kernel.h>
-#include <linux/etherdevice.h>
-#include "stmmac_timer.h"
-
-static void stmmac_timer_handler(void *data)
-{
- struct net_device *dev = (struct net_device *)data;
-
- stmmac_schedule(dev);
-}
-
-#define STMMAC_TIMER_MSG(timer, freq) \
-printk(KERN_INFO "stmmac_timer: %s Timer ON (freq %dHz)\n", timer, freq);
-
-#if defined(CONFIG_STMMAC_RTC_TIMER)
-#include <linux/rtc.h>
-static struct rtc_device *stmmac_rtc;
-static rtc_task_t stmmac_task;
-
-static void stmmac_rtc_start(unsigned int new_freq)
-{
- rtc_irq_set_freq(stmmac_rtc, &stmmac_task, new_freq);
- rtc_irq_set_state(stmmac_rtc, &stmmac_task, 1);
-}
-
-static void stmmac_rtc_stop(void)
-{
- rtc_irq_set_state(stmmac_rtc, &stmmac_task, 0);
-}
-
-int stmmac_open_ext_timer(struct net_device *dev, struct stmmac_timer *tm)
-{
- stmmac_task.private_data = dev;
- stmmac_task.func = stmmac_timer_handler;
-
- stmmac_rtc = rtc_class_open(CONFIG_RTC_HCTOSYS_DEVICE);
- if (stmmac_rtc == NULL) {
- pr_err("open rtc device failed\n");
- return -ENODEV;
- }
-
- rtc_irq_register(stmmac_rtc, &stmmac_task);
-
- /* Periodic mode is not supported */
- if ((rtc_irq_set_freq(stmmac_rtc, &stmmac_task, tm->freq) < 0)) {
- pr_err("set periodic failed\n");
- rtc_irq_unregister(stmmac_rtc, &stmmac_task);
- rtc_class_close(stmmac_rtc);
- return -1;
- }
-
- STMMAC_TIMER_MSG(CONFIG_RTC_HCTOSYS_DEVICE, tm->freq);
-
- tm->timer_start = stmmac_rtc_start;
- tm->timer_stop = stmmac_rtc_stop;
-
- return 0;
-}
-
-int stmmac_close_ext_timer(void)
-{
- rtc_irq_set_state(stmmac_rtc, &stmmac_task, 0);
- rtc_irq_unregister(stmmac_rtc, &stmmac_task);
- rtc_class_close(stmmac_rtc);
- return 0;
-}
-
-#elif defined(CONFIG_STMMAC_TMU_TIMER)
-#include <linux/clk.h>
-#define TMU_CHANNEL "tmu2_clk"
-static struct clk *timer_clock;
-
-static void stmmac_tmu_start(unsigned int new_freq)
-{
- clk_set_rate(timer_clock, new_freq);
- clk_enable(timer_clock);
-}
-
-static void stmmac_tmu_stop(void)
-{
- clk_disable(timer_clock);
-}
-
-int stmmac_open_ext_timer(struct net_device *dev, struct stmmac_timer *tm)
-{
- timer_clock = clk_get(NULL, TMU_CHANNEL);
-
- if (timer_clock == NULL)
- return -1;
-
- if (tmu2_register_user(stmmac_timer_handler, (void *)dev) < 0) {
- timer_clock = NULL;
- return -1;
- }
-
- STMMAC_TIMER_MSG("TMU2", tm->freq);
- tm->timer_start = stmmac_tmu_start;
- tm->timer_stop = stmmac_tmu_stop;
-
- return 0;
-}
-
-int stmmac_close_ext_timer(void)
-{
- clk_disable(timer_clock);
- tmu2_unregister_user();
- clk_put(timer_clock);
- return 0;
-}
-#endif
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_timer.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_timer.h
deleted file mode 100644
index aea9b14..0000000
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_timer.h
+++ /dev/null
@@ -1,46 +0,0 @@
-/*******************************************************************************
- STMMAC external timer Header File.
-
- Copyright (C) 2007-2009 STMicroelectronics Ltd
-
- This program is free software; you can redistribute it and/or modify it
- under the terms and conditions of the GNU General Public License,
- version 2, as published by the Free Software Foundation.
-
- This program is distributed in the hope it will be useful, but WITHOUT
- ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
- more details.
-
- You should have received a copy of the GNU General Public License along with
- this program; if not, write to the Free Software Foundation, Inc.,
- 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
-
- The full GNU General Public License is included in this distribution in
- the file called "COPYING".
-
- Author: Giuseppe Cavallaro <peppe.cavallaro@st.com>
-*******************************************************************************/
-#ifndef __STMMAC_TIMER_H__
-#define __STMMAC_TIMER_H__
-
-struct stmmac_timer {
- void (*timer_start) (unsigned int new_freq);
- void (*timer_stop) (void);
- unsigned int freq;
- unsigned int enable;
-};
-
-/* Open the HW timer device and return 0 in case of success */
-int stmmac_open_ext_timer(struct net_device *dev, struct stmmac_timer *tm);
-/* Stop the timer and release it */
-int stmmac_close_ext_timer(void);
-/* Function used for scheduling task within the stmmac */
-void stmmac_schedule(struct net_device *dev);
-
-#if defined(CONFIG_STMMAC_TMU_TIMER)
-extern int tmu2_register_user(void *fnt, void *data);
-extern void tmu2_unregister_user(void);
-#endif
-
-#endif /* __STMMAC_TIMER_H__ */
--
1.7.4.4
^ permalink raw reply related
* [net-next.git 2/7] stmmac: manage tx clean out of rx_poll
From: Giuseppe CAVALLARO @ 2012-09-05 15:03 UTC (permalink / raw)
To: netdev; +Cc: bhutchings, davem, Giuseppe Cavallaro
In-Reply-To: <1346857432-24657-1-git-send-email-peppe.cavallaro@st.com>
This patch is to invoke the stmmac_tx (tx handler)
out of the NAPI poll method.
This will make easier the next step to add the new
mitigation schema.
Also the patch enhances the ethtool to report some
stats for normal TX and RX IRQs.
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
---
drivers/net/ethernet/stmicro/stmmac/common.h | 13 +++++++----
drivers/net/ethernet/stmicro/stmmac/dwmac_lib.c | 7 +++--
.../net/ethernet/stmicro/stmmac/stmmac_ethtool.c | 4 ++-
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 22 ++++++++++++++-----
4 files changed, 31 insertions(+), 15 deletions(-)
diff --git a/drivers/net/ethernet/stmicro/stmmac/common.h b/drivers/net/ethernet/stmicro/stmmac/common.h
index 719be39..bd32fe6 100644
--- a/drivers/net/ethernet/stmicro/stmmac/common.h
+++ b/drivers/net/ethernet/stmicro/stmmac/common.h
@@ -95,7 +95,9 @@ struct stmmac_extra_stats {
unsigned long threshold;
unsigned long tx_pkt_n;
unsigned long rx_pkt_n;
- unsigned long poll_n;
+ unsigned long rx_napi_poll;
+ unsigned long rx_normal_irq_n;
+ unsigned long tx_normal_irq_n;
unsigned long sched_timer_n;
unsigned long normal_irq_n;
unsigned long mmc_tx_irq_n;
@@ -169,10 +171,11 @@ enum rx_frame_status { /* IPC status */
llc_snap = 4,
};
-enum tx_dma_irq_status {
- tx_hard_error = 1,
- tx_hard_error_bump_tc = 2,
- handle_tx_rx = 3,
+enum dma_irq_status {
+ tx_hard_error = 0x1,
+ tx_hard_error_bump_tc = 0x2,
+ handle_rx = 0x4,
+ handle_tx = 0x8,
};
enum core_specific_irq_mask {
diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac_lib.c b/drivers/net/ethernet/stmicro/stmmac/dwmac_lib.c
index 4e0e18a..73766e6 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac_lib.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac_lib.c
@@ -206,9 +206,10 @@ int dwmac_dma_interrupt(void __iomem *ioaddr,
/* TX/RX NORMAL interrupts */
if (intr_status & DMA_STATUS_NIS) {
x->normal_irq_n++;
- if (likely((intr_status & DMA_STATUS_RI) ||
- (intr_status & (DMA_STATUS_TI))))
- ret = handle_tx_rx;
+ if (likely(intr_status & DMA_STATUS_RI))
+ ret |= handle_rx;
+ if (intr_status & (DMA_STATUS_TI))
+ ret |= handle_tx;
}
/* Optional hardware blocks, interrupts should be disabled */
if (unlikely(intr_status &
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c
index 76fd61a..505fe71 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_ethtool.c
@@ -90,7 +90,9 @@ static const struct stmmac_stats stmmac_gstrings_stats[] = {
STMMAC_STAT(threshold),
STMMAC_STAT(tx_pkt_n),
STMMAC_STAT(rx_pkt_n),
- STMMAC_STAT(poll_n),
+ STMMAC_STAT(rx_napi_poll),
+ STMMAC_STAT(rx_normal_irq_n),
+ STMMAC_STAT(tx_normal_irq_n),
STMMAC_STAT(sched_timer_n),
STMMAC_STAT(normal_irq_n),
STMMAC_STAT(normal_irq_n),
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index c8985f3..b247c39 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -824,16 +824,27 @@ static void stmmac_tx_err(struct stmmac_priv *priv)
netif_wake_queue(priv->dev);
}
+static inline void stmmac_rx_schedule(struct stmmac_priv *priv)
+{
+ if (likely(napi_schedule_prep(&priv->napi))) {
+ stmmac_disable_irq(priv);
+ __napi_schedule(&priv->napi);
+ }
+}
static void stmmac_dma_interrupt(struct stmmac_priv *priv)
{
int status;
status = priv->hw->dma->dma_interrupt(priv->ioaddr, &priv->xstats);
- if (likely(status == handle_tx_rx))
- _stmmac_schedule(priv);
-
- else if (unlikely(status == tx_hard_error_bump_tc)) {
+ if (likely(status == handle_rx)) {
+ priv->xstats.rx_normal_irq_n++;
+ stmmac_rx_schedule(priv);
+ }
+ if (likely(status == handle_tx)) {
+ priv->xstats.tx_normal_irq_n++;
+ stmmac_tx(priv);
+ } else if (unlikely(status == tx_hard_error_bump_tc)) {
/* Try to bump up the dma threshold on this failure */
if (unlikely(tc != SF_DMA_MODE) && (tc <= 256)) {
tc += 64;
@@ -1443,8 +1454,7 @@ static int stmmac_poll(struct napi_struct *napi, int budget)
struct stmmac_priv *priv = container_of(napi, struct stmmac_priv, napi);
int work_done = 0;
- priv->xstats.poll_n++;
- stmmac_tx(priv);
+ priv->xstats.rx_napi_poll++;
work_done = stmmac_rx(priv, budget);
if (work_done < budget) {
--
1.7.4.4
^ permalink raw reply related
* Re: sctp_close/sk_free: kernel BUG at arch/x86/mm/physaddr.c:18!
From: Marc Kleine-Budde @ 2012-09-05 15:01 UTC (permalink / raw)
To: Fengguang Wu; +Cc: H.K. Jerry Chu, Eric W. Biederman, networking, linux-can
In-Reply-To: <20120905145508.GA9450@localhost>
[-- Attachment #1: Type: text/plain, Size: 1078 bytes --]
On 09/05/2012 04:55 PM, Fengguang Wu wrote:
>>> This in turn means the problem doesn't come from the CAN patches, as
>>> both trees have different CAN patches. I'm adding Eric W. Biederman on
>>> Cc as he contributed some sctp patches between v3.6 and net-next/master.
>>
>> Anything is possible, but this seems unlikely as I don't think I touched
>> anything close to that part of the code.
>
> You are both right. The bad commit turns out to be one of:
>
> 1bed966cc3bd4042110129f0fc51aeeb59c5b200 Merge branch 'tcp_fastopen_server'
> 168a8f58059a22feb9e9a2dcc1b8053dbbbc12ef tcp: TCP Fast Open Server - main code path
> 8336886f786fdacbc19b719c1f7ea91eb70706d4 tcp: TCP Fast Open Server - support TFO listeners
>
> Thanks,
> Fengguang
Thanks for your work Fengguang.
Marc
--
Pengutronix e.K. | Marc Kleine-Budde |
Industrial Linux Solutions | Phone: +49-231-2826-924 |
Vertretung West/Dortmund | Fax: +49-5121-206917-5555 |
Amtsgericht Hildesheim, HRA 2686 | http://www.pengutronix.de |
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 259 bytes --]
^ permalink raw reply
* Re: sctp_close/sk_free: kernel BUG at arch/x86/mm/physaddr.c:18!
From: Fengguang Wu @ 2012-09-05 14:55 UTC (permalink / raw)
To: H.K. Jerry Chu
Cc: Eric W. Biederman, Marc Kleine-Budde, networking, linux-can
In-Reply-To: <87mx15zfze.fsf@xmission.com>
On Tue, Sep 04, 2012 at 01:32:21PM -0700, Eric W. Biederman wrote:
> Marc Kleine-Budde <mkl@pengutronix.de> writes:
>
> > On 09/04/2012 04:04 PM, Fengguang Wu wrote:
> >> FYI, another kconfig triggering a slightly different oops on tree
> >>
> >> git://gitorious.org/linux-can/linux-can-next led-trigger
> >
> > This in turn means the problem doesn't come from the CAN patches, as
> > both trees have different CAN patches. I'm adding Eric W. Biederman on
> > Cc as he contributed some sctp patches between v3.6 and net-next/master.
>
> Anything is possible, but this seems unlikely as I don't think I touched
> anything close to that part of the code.
You are both right. The bad commit turns out to be one of:
1bed966cc3bd4042110129f0fc51aeeb59c5b200 Merge branch 'tcp_fastopen_server'
168a8f58059a22feb9e9a2dcc1b8053dbbbc12ef tcp: TCP Fast Open Server - main code path
8336886f786fdacbc19b719c1f7ea91eb70706d4 tcp: TCP Fast Open Server - support TFO listeners
Thanks,
Fengguang
> This most definitely looks like a memory stomp somewhere.
>
> sk->inet_sk->inet_opt has a bad value.
>
> I am puzzled though what are we doing with both ipv4 and ipv6 release
> state doing on the same socket path? Is this some crazy ipv6 socket
> doing sctp with only ipv4 addresses?
>
> >> [ 96.267311] ------------[ cut here ]------------
> >> [ 96.268294] kernel BUG at /c/kernel-tests/src/stable/arch/x86/mm/physaddr.c:18!
> >> [ 96.269988] invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC
> >> [ 96.270636] Modules linked in:
> >> [ 96.270636] CPU 0
> >> [ 96.270636] Pid: 2116, comm: trinity Not tainted 3.6.0-rc3+ #2679 Bochs Bochs
> >> [ 96.270636] RIP: 0010:[<ffffffff8102b22b>] [<ffffffff8102b22b>] __phys_addr+0x46/0x6b
> >> [ 96.270636] RSP: 0018:ffff880019585c98 EFLAGS: 00010213
> >> [ 96.270636] RAX: ffff87ffffffffff RBX: 0000ea6000000bb8 RCX: 0000000000000000
> >> [ 96.270636] RDX: 0000000000000000 RSI: 0000000000000296 RDI: 0000ea6000000bb8
> >> [ 96.270636] RBP: ffff880019585c98 R08: 0000000000000058 R09: 0000000000000008
> >> [ 96.270636] R10: 000000000000000a R11: 0000000000000058 R12: ffff8800195f7718
> >> [ 96.270636] R13: ffffffff816521cf R14: ffffea0000000000 R15: 0000000000000000
> >> [ 96.270636] FS: 00007fa19b534700(0000) GS:ffff88001f200000(0000) knlGS:0000000000000000
> >> [ 96.270636] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> >> [ 96.270636] CR2: 00007fa19b03eba0 CR3: 000000001957b000 CR4: 00000000000006f0
> >> [ 96.270636] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> >> [ 96.270636] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
> >> [ 96.270636] Process trinity (pid: 2116, threadinfo ffff880019584000, task ffff88001af2c680)
> >> [ 96.270636] Stack:
> >> [ 96.270636] ffff880019585cd8 ffffffff811091d7 0000000000000000 ffff88001b1ef200
> >> [ 96.270636] ffff88001b1ef4d0 0000000000000000 ffff88001b1617b0 0000000000000000
> >> [ 96.270636] ffff880019585cf8 ffffffff816521cf ffff88001b1ef200 ffff88001b1ef248
> >> [ 96.270636] Call Trace:
> >> [ 96.270636] [<ffffffff811091d7>] kfree+0x63/0x162
> >> [ 96.270636] [<ffffffff816521cf>] inet_sock_destruct+0x112/0x1ca
> >> [ 96.270636] [<ffffffff815f6fa4>] __sk_free+0x1d/0x114
> >> [ 96.270636] [<ffffffff815f710b>] sk_free+0x1c/0x1e
> >> [ 96.270636] [<ffffffff816d59d5>] sctp_close+0x21a/0x229
> >> [ 96.270636] [<ffffffff810810f6>] ? lock_release_holdtime.part.6+0xb2/0xb7
> >> [ 96.270636] [<ffffffff81651b3e>] ? inet_release+0x65/0xc3
> >> [ 96.270636] [<ffffffff81651b93>] inet_release+0xba/0xc3
> >> [ 96.270636] [<ffffffff81651af9>] ? inet_release+0x20/0xc3
> >> [ 96.270636] [<ffffffff81674134>] inet6_release+0x30/0x3c
> >> [ 96.270636] [<ffffffff815f2317>] sock_release+0x1f/0x77
> >> [ 96.270636] [<ffffffff815f2396>] sock_close+0x27/0x2b
> >> [ 96.270636] [<ffffffff8110ec22>] __fput+0xf0/0x24b
> >> [ 96.270636] [<ffffffff8110ed8b>] ____fput+0xe/0x10
> >> [ 96.270636] [<ffffffff8104f370>] task_work_run+0x5d/0x75
> >> [ 96.270636] [<ffffffff81038a66>] do_exit+0x26b/0x7d7
> >> [ 96.270636] [<ffffffff81725a95>] ? retint_swapgs+0x13/0x1b
> >> [ 96.270636] [<ffffffff8103925b>] do_group_exit+0x7b/0xba
> >> [ 96.270636] [<ffffffff810392b1>] sys_exit_group+0x17/0x17
> >> [ 96.270636] [<ffffffff8172c78e>] tracesys+0xd0/0xd5
> >> [ 96.270636] Code: 00 80 48 01 c7 48 81 ff ff ff ff 1f 76 02 0f 0b 48 89 f8 48 03 05 f6 bd ae 00 eb 32 48 b8 ff ff ff ff ff 87 ff ff 48 39 c7 77 02 <0f> 0b 0f b6 0d 55 57 ba 00 48 b8 00 00 00 00 00 78 00 00 48 01
> >> [ 96.270636] RIP [<ffffffff8102b22b>] __phys_addr+0x46/0x6b
> >> [ 96.270636] RSP <ffff880019585c98>
^ permalink raw reply
* [PATCH] net: add unknown state to sysfs NIC duplex export
From: Nikolay Aleksandrov @ 2012-09-05 14:11 UTC (permalink / raw)
To: netdev
Currently when the NIC duplex state is DUPLEX_UNKNOWN it is exported as
full through sysfs, this patch adds support for DUPLEX_UNKNOWN. It is
handled the same way as in ethtool.
Signed-off-by: Nikolay Aleksandrov <naleksan@redhat.com>
---
net/core/net-sysfs.c | 18 +++++++++++++++---
1 file changed, 15 insertions(+), 3 deletions(-)
diff --git a/net/core/net-sysfs.c b/net/core/net-sysfs.c
index 7260717..7720fb1 100644
--- a/net/core/net-sysfs.c
+++ b/net/core/net-sysfs.c
@@ -166,9 +166,21 @@ static ssize_t show_duplex(struct device *dev,
if (netif_running(netdev)) {
struct ethtool_cmd cmd;
- if (!__ethtool_get_settings(netdev, &cmd))
- ret = sprintf(buf, "%s\n",
- cmd.duplex ? "full" : "half");
+ if (!__ethtool_get_settings(netdev, &cmd)) {
+ char *duplex;
+ switch(cmd.duplex) {
+ case DUPLEX_HALF:
+ duplex = "half";
+ break;
+ case DUPLEX_FULL:
+ duplex = "full";
+ break;
+ default:
+ duplex = "unknown";
+ break;
+ }
+ ret = sprintf(buf, "%s\n", duplex);
+ }
}
rtnl_unlock();
return ret;
--
1.7.11.4
^ permalink raw reply related
* kernel BUG at kernel/timer.c:748!
From: Dave Jones @ 2012-09-05 4:35 UTC (permalink / raw)
To: netdev
Just hit this bug on 3.6-rc4.
The BUG is..
BUG_ON(!timer->function);
Not much to go on... Any thoughts on what I could add to get
more debug info on which protocol etc this was ?
Dave
kernel BUG at kernel/timer.c:748!
invalid opcode: 0000 [#1] SMP
Modules linked in: tun fuse ipt_ULOG binfmt_misc nfnetlink nfc caif_socket caif phonet can llc2 pppoe pppox ppp_generic slhc irda crc_ccitt rds af_key decnet rose x25 atm netrom appletalk ipx p8023 psnap p8022 llc ax25 nfsv3 nfs_acl nfs fscache lockd sunrpc bluetooth rfkill ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 ip6table_filter ip6_tables nf_conntrack_ipv4 nf_defrag_ipv4 xt_state nf_conntrack kvm_intel kvm crc32c_intel ghash_clmulni_intel microcode pcspkr i2c_i801 e1000e uinput i915 video i2c_algo_bit drm_kms_helper drm i2c_core
CPU 3
Pid: 12330, comm: trinity-child3 Not tainted 3.6.0-rc4+ #36
RIP: 0010:[<ffffffff810813f5>] [<ffffffff810813f5>] mod_timer+0x2c5/0x2f0
RSP: 0018:ffff88000dfd7e08 EFLAGS: 00010246
RAX: 000000000000001a RBX: ffff880122d62948 RCX: 000000000000001a
RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff88000dfd7e10
RBP: ffff88000dfd7e48 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000001517000 R11: 0000000000000246 R12: 000000016c000000
R13: 000000016c12bcb1 R14: ffff8801236cee00 R15: 00000000ffffff01
FS: 00007fa96745f740(0000) GS:ffff880148200000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00000000100ff000 CR3: 0000000099344000 CR4: 00000000001407e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process trinity-child3 (pid: 12330, threadinfo ffff88000dfd6000, task ffff880090890000)
Stack:
ffffffff8154cb6d 0000000007b5edf7 ffff88000dfd7e28 ffff880122d62520
0000000000000009 0000000000000004 ffff8801236cee00 00000000ffffff01
ffff88000dfd7e68 ffffffff8154c79c ffffffff81550e6c ffff880122d62520
Call Trace:
[<ffffffff8154cb6d>] ? lock_sock_nested+0x8d/0xa0
[<ffffffff8154c79c>] sk_reset_timer+0x1c/0x30
[<ffffffff81550e6c>] ? sock_setsockopt+0x8c/0x960
[<ffffffff815a84a0>] inet_csk_reset_keepalive_timer+0x20/0x30
[<ffffffff815c018d>] tcp_set_keepalive+0x3d/0x50
[<ffffffff81551703>] sock_setsockopt+0x923/0x960
[<ffffffff810ddf76>] ? trace_hardirqs_on_caller+0x16/0x1e0
[<ffffffff811db0ac>] ? fget_light+0x24c/0x520
[<ffffffff8154af86>] sys_setsockopt+0xc6/0xe0
[<ffffffff816a50ed>] system_call_fastpath+0x1a/0x1f
Code: 00 74 43 9c 58 0f 1f 44 00 00 f6 c4 02 0f 84 14 ff ff ff eb 93 48 c7 c7 20 48 c3 81 e8 f5 70 05 00 85 c0 0f 85 fe fe ff ff eb b7 <0f> 0b 48 8b 75 08 48 89 df e8 3d f6 ff ff e9 b2 fd ff ff 4d 89
RIP [<ffffffff810813f5>] mod_timer+0x2c5/0x2f0
RSP <ffff88000dfd7e08>
---[ end trace 7e7b5910138e49a3 ]---
^ permalink raw reply
* Re: when the MTU interface is modified, the promiscuous mode is reset in gianfar driver
From: Claudiu Manoil @ 2012-09-05 13:46 UTC (permalink / raw)
To: David Miller; +Cc: chikazawa.akifu, netdev, linux-kernel
In-Reply-To: <20120903.155332.1814926007061594344.davem@davemloft.net>
On 09/03/2012 10:53 PM, David Miller wrote:
> From: chikazawa.akifu@jp.fujitsu.com (近沢 哲史)
> Date: Mon, 27 Aug 2012 17:38:34 +0900
>
>> I am using the gianfar ethernet driver. I am having a problem with the
>> interface settings.
>> Under promiscuous mode, when the MTU interface is modified, the promiscuous
>> mode setting is turned off in gianfar driver when it should not be.
>> The details are as follows:
>> After changing MTU with ifconfig, I could see that the interface flag of
>> eth0 is still PROMISC.
>> However, when I checked value of RCTL register with ethtool, PROM bit of
>> RTCL register is cleared.
>> It seems to be cause that is the gfar_init_mac() function, it doesn't set
>> the PROM bit after the interface MTU is changed.
>> This problem was detected on linux-2.6.32.2, but it seems to same on
>> linux-3.6.0-rc3.
>> Is this behavior on purpose?
>>
>> I also attach the amended file,I think it would be so.
>>
>> Signed-off-by: Akifumi Chikazawa <chikazawa.akifu@jp.fujitsu.com>
> It seems like we also lose all of the multicast configurations as
> well.
>
> Therefore, the thing to do is to simply call gfar_set_multi() at the
> appropriate location. That will take care of both the promiscuous
> bit, as well as the multicast addresses.
>
> You can then remove some of the code in gfar_init_mac() that does
> things like gfar_clear_exact_match(), because gfar_set_multi() will
> take care of that if necessary.
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
> .
>
Hello David,
Though apparently it would make sense to change init_mac as shown in the
patch below, I must say I could *not* reproduce the issue with the
current net-next code (just for reference:
uname -a
Linux p1020rdb 3.6.0-rc3-22396-g600e177 #8 SMP Wed Sep 5 ... 2012 ppc
GNU/Linux
)
So, having eth1 up, I configured promisc mode - then checked that this
is correctly
reflected by rctlr (with ethtool -d eth1):
300: 00 08 37 ca 00 00 00 00 00 00 00 00 00 00 00 00
then, after changing mtu (while eth1 up): ifconfig eth1 mtu 500
further checks on rctlr show that the promiscuity setting is preserved
(i.e. rctlr
has same value as above).
Though changing the mtu with the interface up results indeed in a call to
gfar_init_mac(), I found out that the ndo_set_rx_mode hook (set to
gfar_set_multi()) is being also invoked in the process, to update
the multicast and promiscuity settings. So unless there's some kind
of concurrency/timing issue(?), I don't see how the promisc mode would
be reset by changing the mtu. Maybe more details on the issue reproduction
would help (or maybe a usecase involving multicast settings?)
regards,
Claudiu
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
---
drivers/net/ethernet/freescale/gianfar.c | 5 +++--
1 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/freescale/gianfar.c
b/drivers/net/ethernet/freescale/gianfar.c
index 4d5b58c..ede7efe 100644
--- a/drivers/net/ethernet/freescale/gianfar.c
+++ b/drivers/net/ethernet/freescale/gianfar.c
@@ -353,6 +353,9 @@ static void gfar_init_mac(struct net_device *ndev)
/* Configure the coalescing support */
gfar_configure_coalescing(priv, 0xFF, 0xFF);
+ /* update multicast configs and promiscuity of the device */
+ gfar_set_multi(ndev);
+
if (priv->rx_filer_enable) {
rctrl |= RCTRL_FILREN;
/* Program the RIR0 reg with the required distribution */
@@ -364,8 +367,6 @@ static void gfar_init_mac(struct net_device *ndev)
if (priv->extended_hash) {
rctrl |= RCTRL_EXTHASH;
-
- gfar_clear_exact_match(ndev);
rctrl |= RCTRL_EMEN;
}
--
1.7.6.5
^ permalink raw reply related
* Re: [PATCH] cfg80211: use list_move_tail instead of list_del/list_add_tail
From: Johannes Berg @ 2012-09-05 13:40 UTC (permalink / raw)
To: Wei Yongjun; +Cc: linville, davem, yongjun_wei, linux-wireless, netdev
In-Reply-To: <CAPgLHd-7DpM_OQ7RTVip6d-zb8DZS9SiscrMugm2-D992ajjsg@mail.gmail.com>
On Wed, 2012-09-05 at 14:34 +0800, Wei Yongjun wrote:
> From: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
>
> Using list_move_tail() instead of list_del() + list_add_tail().
>
> Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Applied, thanks.
johannes
^ permalink raw reply
* [V2 PATCH 8/9] cxgb4: Chelsio FCoE offload driver submission (cxgb4 common header updates).
From: Naresh Kumar Inna @ 2012-09-05 12:34 UTC (permalink / raw)
To: JBottomley, linux-scsi, dm, leedom; +Cc: netdev, naresh, chethan
In-Reply-To: <1346848442-4573-1-git-send-email-naresh@chelsio.com>
This patch contains updates to firmware/hardware header files shared
between csiostor and cxgb4/cxgb4vf, and the resulting changes to the cxgb4
source files.
Signed-off-by: Naresh Kumar Inna <naresh@chelsio.com>
---
drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c | 2 +-
drivers/net/ethernet/chelsio/cxgb4/sge.c | 10 +-
drivers/net/ethernet/chelsio/cxgb4/t4_hw.c | 16 ++--
drivers/net/ethernet/chelsio/cxgb4/t4_msg.h | 1 +
drivers/net/ethernet/chelsio/cxgb4/t4_regs.h | 69 ++++++++++++++-
drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h | 104 +++++++++++++++++++---
6 files changed, 170 insertions(+), 32 deletions(-)
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
index 5ed49af..b56d96c 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
@@ -3102,7 +3102,7 @@ static int adap_init1(struct adapter *adap, struct fw_caps_config_cmd *c)
memset(c, 0, sizeof(*c));
c->op_to_write = htonl(FW_CMD_OP(FW_CAPS_CONFIG_CMD) |
FW_CMD_REQUEST | FW_CMD_READ);
- c->retval_len16 = htonl(FW_LEN16(*c));
+ c->cfvalid_to_len16 = htonl(FW_LEN16(*c));
ret = t4_wr_mbox(adap, adap->fn, c, sizeof(*c), c);
if (ret < 0)
return ret;
diff --git a/drivers/net/ethernet/chelsio/cxgb4/sge.c b/drivers/net/ethernet/chelsio/cxgb4/sge.c
index d49933e..121b1e9 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/sge.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/sge.c
@@ -455,7 +455,7 @@ static inline void ring_fl_db(struct adapter *adap, struct sge_fl *q)
{
if (q->pend_cred >= 8) {
wmb();
- t4_write_reg(adap, MYPF_REG(SGE_PF_KDOORBELL), DBPRIO |
+ t4_write_reg(adap, MYPF_REG(SGE_PF_KDOORBELL), DBPRIO(1) |
QID(q->cntxt_id) | PIDX(q->pend_cred / 8));
q->pend_cred &= 7;
}
@@ -2020,10 +2020,10 @@ int t4_sge_alloc_rxq(struct adapter *adap, struct sge_rspq *iq, bool fwevtq,
goto fl_nomem;
flsz = fl->size / 8 + STAT_LEN / sizeof(struct tx_desc);
- c.iqns_to_fl0congen = htonl(FW_IQ_CMD_FL0PACKEN |
+ c.iqns_to_fl0congen = htonl(FW_IQ_CMD_FL0PACKEN(1) |
FW_IQ_CMD_FL0FETCHRO(1) |
FW_IQ_CMD_FL0DATARO(1) |
- FW_IQ_CMD_FL0PADEN);
+ FW_IQ_CMD_FL0PADEN(1));
c.fl0dcaen_to_fl0cidxfthresh = htons(FW_IQ_CMD_FL0FBMIN(2) |
FW_IQ_CMD_FL0FBMAX(3));
c.fl0size = htons(flsz);
@@ -2416,10 +2416,10 @@ void t4_sge_init(struct adapter *adap)
unsigned int fl_align_log = ilog2(FL_ALIGN);
t4_set_reg_field(adap, SGE_CONTROL, PKTSHIFT_MASK |
- INGPADBOUNDARY_MASK | EGRSTATUSPAGESIZE,
+ INGPADBOUNDARY_MASK | EGRSTATUSPAGESIZE(1),
INGPADBOUNDARY(fl_align_log - 5) | PKTSHIFT(2) |
RXPKTCPLMODE |
- (STAT_LEN == 128 ? EGRSTATUSPAGESIZE : 0));
+ (STAT_LEN == 128 ? EGRSTATUSPAGESIZE(1) : 0));
/*
* Set up to drop DOORBELL writes when the DOORBELL FIFO overflows
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
index fa947df..a943faa 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
@@ -456,12 +456,12 @@ static int sf1_read(struct adapter *adapter, unsigned int byte_cnt, int cont,
if (!byte_cnt || byte_cnt > 4)
return -EINVAL;
- if (t4_read_reg(adapter, SF_OP) & BUSY)
+ if (t4_read_reg(adapter, SF_OP) & SF_BUSY)
return -EBUSY;
cont = cont ? SF_CONT : 0;
lock = lock ? SF_LOCK : 0;
t4_write_reg(adapter, SF_OP, lock | cont | BYTECNT(byte_cnt - 1));
- ret = t4_wait_op_done(adapter, SF_OP, BUSY, 0, SF_ATTEMPTS, 5);
+ ret = t4_wait_op_done(adapter, SF_OP, SF_BUSY, 0, SF_ATTEMPTS, 5);
if (!ret)
*valp = t4_read_reg(adapter, SF_DATA);
return ret;
@@ -484,14 +484,14 @@ static int sf1_write(struct adapter *adapter, unsigned int byte_cnt, int cont,
{
if (!byte_cnt || byte_cnt > 4)
return -EINVAL;
- if (t4_read_reg(adapter, SF_OP) & BUSY)
+ if (t4_read_reg(adapter, SF_OP) & SF_BUSY)
return -EBUSY;
cont = cont ? SF_CONT : 0;
lock = lock ? SF_LOCK : 0;
t4_write_reg(adapter, SF_DATA, val);
t4_write_reg(adapter, SF_OP, lock |
cont | BYTECNT(byte_cnt - 1) | OP_WR);
- return t4_wait_op_done(adapter, SF_OP, BUSY, 0, SF_ATTEMPTS, 5);
+ return t4_wait_op_done(adapter, SF_OP, SF_BUSY, 0, SF_ATTEMPTS, 5);
}
/**
@@ -1972,14 +1972,14 @@ int t4_wol_pat_enable(struct adapter *adap, unsigned int port, unsigned int map,
t4_write_reg(adap, EPIO_REG(DATA0), mask0);
t4_write_reg(adap, EPIO_REG(OP), ADDRESS(i) | EPIOWR);
t4_read_reg(adap, EPIO_REG(OP)); /* flush */
- if (t4_read_reg(adap, EPIO_REG(OP)) & BUSY)
+ if (t4_read_reg(adap, EPIO_REG(OP)) & SF_BUSY)
return -ETIMEDOUT;
/* write CRC */
t4_write_reg(adap, EPIO_REG(DATA0), crc);
t4_write_reg(adap, EPIO_REG(OP), ADDRESS(i + 32) | EPIOWR);
t4_read_reg(adap, EPIO_REG(OP)); /* flush */
- if (t4_read_reg(adap, EPIO_REG(OP)) & BUSY)
+ if (t4_read_reg(adap, EPIO_REG(OP)) & SF_BUSY)
return -ETIMEDOUT;
}
#undef EPIO_REG
@@ -2118,7 +2118,7 @@ int t4_fw_hello(struct adapter *adap, unsigned int mbox, unsigned int evt_mbox,
struct fw_hello_cmd c;
INIT_CMD(c, HELLO, WRITE);
- c.err_to_mbasyncnot = htonl(
+ c.err_to_clearinit = htonl(
FW_HELLO_CMD_MASTERDIS(master == MASTER_CANT) |
FW_HELLO_CMD_MASTERFORCE(master == MASTER_MUST) |
FW_HELLO_CMD_MBMASTER(master == MASTER_MUST ? mbox : 0xff) |
@@ -2126,7 +2126,7 @@ int t4_fw_hello(struct adapter *adap, unsigned int mbox, unsigned int evt_mbox,
ret = t4_wr_mbox(adap, mbox, &c, sizeof(c), &c);
if (ret == 0 && state) {
- u32 v = ntohl(c.err_to_mbasyncnot);
+ u32 v = ntohl(c.err_to_clearinit);
if (v & FW_HELLO_CMD_INIT)
*state = DEV_STATE_INIT;
else if (v & FW_HELLO_CMD_ERR)
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_msg.h b/drivers/net/ethernet/chelsio/cxgb4/t4_msg.h
index eb71b82..b760808 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_msg.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_msg.h
@@ -658,6 +658,7 @@ struct ulptx_sgl {
__be32 cmd_nsge;
#define ULPTX_CMD(x) ((x) << 24)
#define ULPTX_NSGE(x) ((x) << 0)
+#define ULPTX_MORE (1U << 23)
__be32 len0;
__be64 addr0;
struct ulptx_sge_pair sge[0];
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h b/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
index 111fc32..6bace75 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
@@ -67,7 +67,7 @@
#define QID_MASK 0xffff8000U
#define QID_SHIFT 15
#define QID(x) ((x) << QID_SHIFT)
-#define DBPRIO 0x00004000U
+#define DBPRIO(x) ((x) << 14)
#define PIDX_MASK 0x00003fffU
#define PIDX_SHIFT 0
#define PIDX(x) ((x) << PIDX_SHIFT)
@@ -89,7 +89,7 @@
#define SGE_CONTROL 0x1008
#define DCASYSTYPE 0x00080000U
#define RXPKTCPLMODE 0x00040000U
-#define EGRSTATUSPAGESIZE 0x00020000U
+#define EGRSTATUSPAGESIZE(x) ((x) << 17)
#define PKTSHIFT_MASK 0x00001c00U
#define PKTSHIFT_SHIFT 10
#define PKTSHIFT(x) ((x) << PKTSHIFT_SHIFT)
@@ -111,6 +111,13 @@
#define HOSTPAGESIZEPF0_MASK 0x0000000fU
#define HOSTPAGESIZEPF0_SHIFT 0
#define HOSTPAGESIZEPF0(x) ((x) << HOSTPAGESIZEPF0_SHIFT)
+#define HOSTPAGESIZEPF1(x) ((x) << 4)
+#define HOSTPAGESIZEPF2(x) ((x) << 8)
+#define HOSTPAGESIZEPF3(x) ((x) << 12)
+#define HOSTPAGESIZEPF4(x) ((x) << 16)
+#define HOSTPAGESIZEPF5(x) ((x) << 20)
+#define HOSTPAGESIZEPF6(x) ((x) << 24)
+#define HOSTPAGESIZEPF7(x) ((x) << 28)
#define SGE_EGRESS_QUEUES_PER_PAGE_PF 0x1010
#define QUEUESPERPAGEPF0_MASK 0x0000000fU
@@ -155,6 +162,14 @@
#define SGE_INT_ENABLE3 0x1040
#define SGE_FL_BUFFER_SIZE0 0x1044
#define SGE_FL_BUFFER_SIZE1 0x1048
+#define SGE_FL_BUFFER_SIZE2 0x104c
+#define SGE_FL_BUFFER_SIZE3 0x1050
+#define SGE_FL_BUFFER_SIZE4 0x1054
+#define SGE_FL_BUFFER_SIZE5 0x1058
+#define SGE_FL_BUFFER_SIZE6 0x105c
+#define SGE_FL_BUFFER_SIZE7 0x1060
+#define SGE_FL_BUFFER_SIZE8 0x1064
+
#define SGE_INGRESS_RX_THRESHOLD 0x10a0
#define THRESHOLD_0_MASK 0x3f000000U
#define THRESHOLD_0_SHIFT 24
@@ -173,6 +188,17 @@
#define THRESHOLD_3(x) ((x) << THRESHOLD_3_SHIFT)
#define THRESHOLD_3_GET(x) (((x) & THRESHOLD_3_MASK) >> THRESHOLD_3_SHIFT)
+#define SGE_DBFIFO_STATUS 0x10a4
+#define HP_INT_THRESH_SHIFT 28
+#define HP_INT_THRESH_MASK 0xfU
+#define HP_INT_THRESH(x) ((x) << HP_INT_THRESH_SHIFT)
+#define LP_INT_THRESH_SHIFT 12
+#define LP_INT_THRESH_MASK 0xfU
+#define LP_INT_THRESH(x) ((x) << LP_INT_THRESH_SHIFT)
+
+#define SGE_DOORBELL_CONTROL 0x10a8
+#define ENABLE_DROP (1 << 13)
+
#define SGE_TIMER_VALUE_0_AND_1 0x10b8
#define TIMERVALUE0_MASK 0xffff0000U
#define TIMERVALUE0_SHIFT 16
@@ -184,7 +210,25 @@
#define TIMERVALUE1_GET(x) (((x) & TIMERVALUE1_MASK) >> TIMERVALUE1_SHIFT)
#define SGE_TIMER_VALUE_2_AND_3 0x10bc
+#define TIMERVALUE2_MASK 0xffff0000U
+#define TIMERVALUE2_SHIFT 16
+#define TIMERVALUE2(x) ((x) << TIMERVALUE0_SHIFT)
+#define TIMERVALUE2_GET(x) (((x) & TIMERVALUE0_MASK) >> TIMERVALUE0_SHIFT)
+#define TIMERVALUE3_MASK 0x0000ffffU
+#define TIMERVALUE3_SHIFT 0
+#define TIMERVALUE3(x) ((x) << TIMERVALUE1_SHIFT)
+#define TIMERVALUE3_GET(x) (((x) & TIMERVALUE1_MASK) >> TIMERVALUE1_SHIFT)
+
#define SGE_TIMER_VALUE_4_AND_5 0x10c0
+#define TIMERVALUE4_MASK 0xffff0000U
+#define TIMERVALUE4_SHIFT 16
+#define TIMERVALUE4(x) ((x) << TIMERVALUE0_SHIFT)
+#define TIMERVALUE4_GET(x) (((x) & TIMERVALUE0_MASK) >> TIMERVALUE0_SHIFT)
+#define TIMERVALUE5_MASK 0x0000ffffU
+#define TIMERVALUE5_SHIFT 0
+#define TIMERVALUE5(x) ((x) << TIMERVALUE1_SHIFT)
+#define TIMERVALUE5_GET(x) (((x) & TIMERVALUE1_MASK) >> TIMERVALUE1_SHIFT)
+
#define SGE_DEBUG_INDEX 0x10cc
#define SGE_DEBUG_DATA_HIGH 0x10d0
#define SGE_DEBUG_DATA_LOW 0x10d4
@@ -243,6 +287,10 @@
#define M_HP_INT_THRESH 0xfU
#define M_LP_INT_THRESH 0xfU
+#define PCIE_PF_CFG 0x40
+#define AIVEC(x) ((x) << 4)
+#define AIVEC_MASK 0x3ffU
+
#define PCIE_PF_CLI 0x44
#define PCIE_INT_CAUSE 0x3004
#define UNXSPLCPLERR 0x20000000U
@@ -287,6 +335,15 @@
#define WINDOW(x) ((x) << WINDOW_SHIFT)
#define PCIE_MEM_ACCESS_OFFSET 0x306c
+#define PCIE_FW 0x30b8
+#define PCIE_FW_ERR 0x80000000U
+#define PCIE_FW_INIT 0x40000000U
+#define PCIE_FW_HALT 0x20000000U
+#define PCIE_FW_MASTER_VLD 0x00008000U
+#define PCIE_FW_MASTER(x) ((x) << 12)
+#define PCIE_FW_MASTER_MASK 0x7
+#define PCIE_FW_MASTER_GET(x) (((x) >> 12) & PCIE_FW_MASTER_MASK)
+
#define PCIE_CORE_UTL_SYSTEM_BUS_AGENT_STATUS 0x5908
#define RNPP 0x80000000U
#define RPCP 0x20000000U
@@ -384,6 +441,8 @@
#define EDC_1_BASE_ADDR 0x7980
#define CIM_BOOT_CFG 0x7b00
+#define UPCRST 0x00000001U
+
#define BOOTADDR_MASK 0xffffff00U
#define CIM_PF_MAILBOX_DATA 0x240
@@ -395,6 +454,9 @@
#define MBOWNER(x) ((x) << MBOWNER_SHIFT)
#define MBOWNER_GET(x) (((x) & MBOWNER_MASK) >> MBOWNER_SHIFT)
+#define CIM_PF_HOST_INT_ENABLE 0x288
+#define MBMSGRDYINTEN(x) ((x) << 19)
+
#define CIM_PF_HOST_INT_CAUSE 0x28c
#define MBMSGRDYINT 0x00080000U
@@ -825,7 +887,7 @@
#define SF_DATA 0x193f8
#define SF_OP 0x193fc
-#define BUSY 0x80000000U
+#define SF_BUSY 0x80000000U
#define SF_LOCK 0x00000010U
#define SF_CONT 0x00000008U
#define BYTECNT_MASK 0x00000006U
@@ -884,6 +946,7 @@
#define I2CM 0x00000002U
#define CIM 0x00000001U
+#define PL_INT_ENABLE 0x19410
#define PL_INT_MAP0 0x19414
#define PL_RST 0x19428
#define PIORST 0x00000002U
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h b/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h
index ad53f79..034ca39 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h
@@ -68,6 +68,7 @@ struct fw_wr_hdr {
};
#define FW_WR_OP(x) ((x) << 24)
+#define FW_WR_OP_GET(x) (((x) >> 24) & 0xff)
#define FW_WR_ATOMIC(x) ((x) << 23)
#define FW_WR_FLUSH(x) ((x) << 22)
#define FW_WR_COMPL(x) ((x) << 21)
@@ -155,6 +156,9 @@ struct fw_eth_tx_pkt_vm_wr {
#define FW_CMD_MAX_TIMEOUT 3000
+#define FW_CMD_HELLO_TIMEOUT (3 * FW_CMD_MAX_TIMEOUT)
+#define FW_CMD_HELLO_RETRIES 3
+
enum fw_cmd_opcodes {
FW_LDST_CMD = 0x01,
FW_RESET_CMD = 0x03,
@@ -209,6 +213,7 @@ struct fw_cmd_hdr {
#define FW_CMD_OP(x) ((x) << 24)
#define FW_CMD_OP_GET(x) (((x) >> 24) & 0xff)
#define FW_CMD_REQUEST (1U << 23)
+#define FW_CMD_REQUEST_GET(x) (((x) >> 23) & 0x1)
#define FW_CMD_READ (1U << 22)
#define FW_CMD_WRITE (1U << 21)
#define FW_CMD_EXEC (1U << 20)
@@ -216,6 +221,7 @@ struct fw_cmd_hdr {
#define FW_CMD_RETVAL(x) ((x) << 8)
#define FW_CMD_RETVAL_GET(x) (((x) >> 8) & 0xff)
#define FW_CMD_LEN16(x) ((x) << 0)
+#define FW_LEN16(fw_struct) FW_CMD_LEN16(sizeof(fw_struct) / 16)
enum fw_ldst_addrspc {
FW_LDST_ADDRSPC_FIRMWARE = 0x0001,
@@ -228,7 +234,8 @@ enum fw_ldst_addrspc {
FW_LDST_ADDRSPC_TP_MIB = 0x0012,
FW_LDST_ADDRSPC_MDIO = 0x0018,
FW_LDST_ADDRSPC_MPS = 0x0020,
- FW_LDST_ADDRSPC_FUNC = 0x0028
+ FW_LDST_ADDRSPC_FUNC = 0x0028,
+ FW_LDST_ADDRSPC_FUNC_PCIE = 0x0029,
};
enum fw_ldst_mps_fid {
@@ -290,6 +297,16 @@ struct fw_ldst_cmd {
__be64 data0;
__be64 data1;
} func;
+ struct fw_ldst_pcie {
+ u8 ctrl_to_fn;
+ u8 bnum;
+ u8 r;
+ u8 ext_r;
+ u8 select_naccess;
+ u8 pcie_fn;
+ __be16 nset_pkd;
+ __be32 data[12];
+ } pcie;
} u;
};
@@ -299,24 +316,42 @@ struct fw_ldst_cmd {
#define FW_LDST_CMD_FID(x) ((x) << 15)
#define FW_LDST_CMD_CTL(x) ((x) << 0)
#define FW_LDST_CMD_RPLCPF(x) ((x) << 0)
+#define FW_LDST_CMD_LC (1U << 4)
+#define FW_LDST_CMD_NACCESS(x) ((x) << 0)
+#define FW_LDST_CMD_FN(x) ((x) << 0)
struct fw_reset_cmd {
__be32 op_to_write;
__be32 retval_len16;
__be32 val;
- __be32 r3;
+ __be32 halt_pkd;
+};
+
+#define FW_RESET_CMD_HALT (1U << 31)
+
+enum {
+ FW_HELLO_CMD_STAGE_OS = 0,
+ FW_HELLO_CMD_STAGE_PREOS0 = 1,
+ FW_HELLO_CMD_STAGE_PREOS1 = 2,
+ FW_HELLO_CMD_STAGE_POSTOS = 3,
};
struct fw_hello_cmd {
__be32 op_to_write;
__be32 retval_len16;
- __be32 err_to_mbasyncnot;
-#define FW_HELLO_CMD_ERR (1U << 31)
-#define FW_HELLO_CMD_INIT (1U << 30)
-#define FW_HELLO_CMD_MASTERDIS(x) ((x) << 29)
-#define FW_HELLO_CMD_MASTERFORCE(x) ((x) << 28)
-#define FW_HELLO_CMD_MBMASTER(x) ((x) << 24)
-#define FW_HELLO_CMD_MBASYNCNOT(x) ((x) << 20)
+ __be32 err_to_clearinit;
+#define FW_HELLO_CMD_ERR (1U << 31)
+#define FW_HELLO_CMD_INIT (1U << 30)
+#define FW_HELLO_CMD_MASTERDIS(x) ((x) << 29)
+#define FW_HELLO_CMD_MASTERFORCE(x) ((x) << 28)
+#define FW_HELLO_CMD_MBMASTER_MASK 0xf
+#define FW_HELLO_CMD_MBMASTER(x) ((x) << 24)
+#define FW_HELLO_CMD_MBMASTER_GET(x) \
+ (((x) >> 24) & FW_HELLO_CMD_MBMASTER_MASK)
+#define FW_HELLO_CMD_MBASYNCNOTINT(x) ((x) << 23)
+#define FW_HELLO_CMD_MBASYNCNOT(x) ((x) << 20)
+#define FW_HELLO_CMD_STAGE(x) ((x) << 17)
+#define FW_HELLO_CMD_CLEARINIT (1U << 16)
__be32 fwrev;
};
@@ -399,11 +434,20 @@ enum fw_caps_config_iscsi {
enum fw_caps_config_fcoe {
FW_CAPS_CONFIG_FCOE_INITIATOR = 0x00000001,
FW_CAPS_CONFIG_FCOE_TARGET = 0x00000002,
+ FW_CAPS_CONFIG_FCOE_CTRL_OFLD = 0x00000004,
+};
+
+enum fw_memtype_cf {
+ FW_MEMTYPE_CF_EDC0 = 0x0,
+ FW_MEMTYPE_CF_EDC1 = 0x1,
+ FW_MEMTYPE_CF_EXTMEM = 0x2,
+ FW_MEMTYPE_CF_FLASH = 0x4,
+ FW_MEMTYPE_CF_INTERNAL = 0x5,
};
struct fw_caps_config_cmd {
__be32 op_to_write;
- __be32 retval_len16;
+ __be32 cfvalid_to_len16;
__be32 r2;
__be32 hwmbitmap;
__be16 nbmcaps;
@@ -416,10 +460,15 @@ struct fw_caps_config_cmd {
__be16 r4;
__be16 iscsicaps;
__be16 fcoecaps;
- __be32 r5;
- __be64 r6;
+ __be32 cfcsum;
+ __be32 finiver;
+ __be32 finicsum;
};
+#define FW_CAPS_CONFIG_CMD_CFVALID (1U << 27)
+#define FW_CAPS_CONFIG_CMD_MEMTYPE_CF(x) ((x) << 24)
+#define FW_CAPS_CONFIG_CMD_MEMADDR64K_CF(x) ((x) << 16)
+
/*
* params command mnemonics
*/
@@ -451,6 +500,7 @@ enum fw_params_param_dev {
FW_PARAMS_PARAM_DEV_INTVER_FCOE = 0x0A,
FW_PARAMS_PARAM_DEV_FWREV = 0x0B,
FW_PARAMS_PARAM_DEV_TPREV = 0x0C,
+ FW_PARAMS_PARAM_DEV_CF = 0x0D,
};
/*
@@ -511,6 +561,9 @@ enum fw_params_param_dmaq {
#define FW_PARAMS_PARAM_Z(x) ((x) << 0)
#define FW_PARAMS_PARAM_XYZ(x) ((x) << 0)
#define FW_PARAMS_PARAM_YZ(x) ((x) << 0)
+#define FW_PARAMS_PARAM_X_GET(x) (((x) >> 16) & 0xff)
+#define FW_PARAMS_PARAM_Y_GET(x) (((x) >> 8) & 0xff)
+#define FW_PARAMS_PARAM_Z_GET(x) (((x) >> 0) & 0xff)
struct fw_params_cmd {
__be32 op_to_vfn;
@@ -648,8 +701,8 @@ struct fw_iq_cmd {
#define FW_IQ_CMD_FL0FETCHRO(x) ((x) << 6)
#define FW_IQ_CMD_FL0HOSTFCMODE(x) ((x) << 4)
#define FW_IQ_CMD_FL0CPRIO(x) ((x) << 3)
-#define FW_IQ_CMD_FL0PADEN (1U << 2)
-#define FW_IQ_CMD_FL0PACKEN (1U << 1)
+#define FW_IQ_CMD_FL0PADEN(x) ((x) << 2)
+#define FW_IQ_CMD_FL0PACKEN(x) ((x) << 1)
#define FW_IQ_CMD_FL0CONGEN (1U << 0)
#define FW_IQ_CMD_FL0DCAEN(x) ((x) << 15)
@@ -1137,6 +1190,14 @@ enum fw_port_dcb_cfg_rc {
FW_PORT_DCB_CFG_ERROR = 0x1
};
+enum fw_port_dcb_type {
+ FW_PORT_DCB_TYPE_PGID = 0x00,
+ FW_PORT_DCB_TYPE_PGRATE = 0x01,
+ FW_PORT_DCB_TYPE_PRIORATE = 0x02,
+ FW_PORT_DCB_TYPE_PFC = 0x03,
+ FW_PORT_DCB_TYPE_APP_ID = 0x04,
+};
+
struct fw_port_cmd {
__be32 op_to_portid;
__be32 action_to_len16;
@@ -1204,6 +1265,7 @@ struct fw_port_cmd {
#define FW_PORT_CMD_TXIPG(x) ((x) << 19)
#define FW_PORT_CMD_LSTATUS (1U << 31)
+#define FW_PORT_CMD_LSTATUS_GET(x) (((x) >> 31) & 0x1)
#define FW_PORT_CMD_LSPEED(x) ((x) << 24)
#define FW_PORT_CMD_LSPEED_GET(x) (((x) >> 24) & 0x3f)
#define FW_PORT_CMD_TXPAUSE (1U << 23)
@@ -1252,6 +1314,9 @@ enum fw_port_module_type {
FW_PORT_MOD_TYPE_TWINAX_PASSIVE,
FW_PORT_MOD_TYPE_TWINAX_ACTIVE,
FW_PORT_MOD_TYPE_LRM,
+ FW_PORT_MOD_TYPE_ERROR = FW_PORT_CMD_MODTYPE_MASK - 3,
+ FW_PORT_MOD_TYPE_UNKNOWN = FW_PORT_CMD_MODTYPE_MASK - 2,
+ FW_PORT_MOD_TYPE_NOTSUPPORTED = FW_PORT_CMD_MODTYPE_MASK - 1,
FW_PORT_MOD_TYPE_NONE = FW_PORT_CMD_MODTYPE_MASK
};
@@ -1613,7 +1678,11 @@ struct fw_hdr {
u8 intfver_iscsi;
u8 intfver_fcoe;
u8 reserved2;
- __be32 reserved3[27];
+ __u32 reserved3;
+ __u32 reserved4;
+ __u32 reserved5;
+ __be32 flags;
+ __be32 reserved6[23];
};
#define FW_HDR_FW_VER_MAJOR_GET(x) (((x) >> 24) & 0xff)
@@ -1621,6 +1690,11 @@ struct fw_hdr {
#define FW_HDR_FW_VER_MICRO_GET(x) (((x) >> 8) & 0xff)
#define FW_HDR_FW_VER_BUILD_GET(x) (((x) >> 0) & 0xff)
+enum fw_hdr_flags {
+ FW_HDR_FLAGS_RESET_HALT = 0x00000001,
+};
+
+
#define S_FW_CMD_OP 24
#define V_FW_CMD_OP(x) ((x) << S_FW_CMD_OP)
--
1.7.1
^ permalink raw reply related
* [V2 PATCH 7/9] csiostor: Chelsio FCoE offload driver submission (headers part 2).
From: Naresh Kumar Inna @ 2012-09-05 12:34 UTC (permalink / raw)
To: JBottomley, linux-scsi, dm, leedom; +Cc: netdev, naresh, chethan
In-Reply-To: <1346848442-4573-1-git-send-email-naresh@chelsio.com>
This patch contains the second set of the header files for csiostor driver.
Signed-off-by: Naresh Kumar Inna <naresh@chelsio.com>
---
drivers/scsi/csiostor/csio_lnode.h | 255 ++++++++++++++++++
drivers/scsi/csiostor/csio_mb.h | 278 +++++++++++++++++++
drivers/scsi/csiostor/csio_rnode.h | 141 ++++++++++
drivers/scsi/csiostor/csio_scsi.h | 342 ++++++++++++++++++++++++
drivers/scsi/csiostor/csio_wr.h | 512 ++++++++++++++++++++++++++++++++++++
5 files changed, 1528 insertions(+), 0 deletions(-)
create mode 100644 drivers/scsi/csiostor/csio_lnode.h
create mode 100644 drivers/scsi/csiostor/csio_mb.h
create mode 100644 drivers/scsi/csiostor/csio_rnode.h
create mode 100644 drivers/scsi/csiostor/csio_scsi.h
create mode 100644 drivers/scsi/csiostor/csio_wr.h
diff --git a/drivers/scsi/csiostor/csio_lnode.h b/drivers/scsi/csiostor/csio_lnode.h
new file mode 100644
index 0000000..8d84988
--- /dev/null
+++ b/drivers/scsi/csiostor/csio_lnode.h
@@ -0,0 +1,255 @@
+/*
+ * This file is part of the Chelsio FCoE driver for Linux.
+ *
+ * Copyright (c) 2008-2012 Chelsio Communications, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __CSIO_LNODE_H__
+#define __CSIO_LNODE_H__
+
+#include <linux/kref.h>
+#include <linux/timer.h>
+#include <linux/workqueue.h>
+#include <scsi/fc/fc_els.h>
+
+
+#include "csio_defs.h"
+#include "csio_hw.h"
+
+#define CSIO_FCOE_MAX_NPIV 128
+#define CSIO_FCOE_MAX_RNODES 2048
+
+/* FDMI port attribute unknown speed */
+#define CSIO_HBA_PORTSPEED_UNKNOWN 0x8000
+
+extern int csio_fcoe_rnodes;
+extern int csio_fdmi_enable;
+
+/* State machine evets */
+enum csio_ln_ev {
+ CSIO_LNE_NONE = (uint32_t)0,
+ CSIO_LNE_LINKUP,
+ CSIO_LNE_FAB_INIT_DONE,
+ CSIO_LNE_LINK_DOWN,
+ CSIO_LNE_DOWN_LINK,
+ CSIO_LNE_LOGO,
+ CSIO_LNE_CLOSE,
+ CSIO_LNE_MAX_EVENT,
+};
+
+
+struct csio_fcf_info {
+ struct list_head list;
+ uint8_t priority;
+ uint8_t mac[6];
+ uint8_t name_id[8];
+ uint8_t fabric[8];
+ uint16_t vf_id;
+ uint8_t vlan_id;
+ uint16_t max_fcoe_size;
+ uint8_t fc_map[3];
+ uint32_t fka_adv;
+ uint32_t fcfi;
+ uint8_t get_next:1;
+ uint8_t link_aff:1;
+ uint8_t fpma:1;
+ uint8_t spma:1;
+ uint8_t login:1;
+ uint8_t portid;
+ uint8_t spma_mac[6];
+ struct kref kref;
+};
+
+/* Defines for flags */
+#define CSIO_LNF_FIPSUPP 0x00000001 /* Fip Supported */
+#define CSIO_LNF_NPIVSUPP 0x00000002 /* NPIV supported */
+#define CSIO_LNF_LINK_ENABLE 0x00000004 /* Link enabled */
+#define CSIO_LNF_FDMI_ENABLE 0x00000008 /* FDMI support */
+
+/* Transport events */
+enum csio_ln_fc_evt {
+ CSIO_LN_FC_LINKUP = 1,
+ CSIO_LN_FC_LINKDOWN,
+ CSIO_LN_FC_RSCN,
+ CSIO_LN_FC_ATTRIB_UPDATE,
+};
+
+/* Lnode stats */
+struct csio_lnode_stats {
+ uint32_t n_link_up; /* Link down */
+ uint32_t n_link_down; /* Link up */
+ uint32_t n_err; /* error */
+ uint32_t n_err_nomem; /* memory not available */
+ uint32_t n_inval_parm; /* Invalid parameters */
+ uint32_t n_evt_unexp; /* unexpected event */
+ uint32_t n_evt_drop; /* dropped event */
+ uint32_t n_rnode_match; /* matched rnode */
+ uint32_t n_dev_loss_tmo; /* Device loss timeout */
+ uint32_t n_fdmi_err; /* fdmi err */
+ uint32_t n_evt_fw[RSCN_DEV_LOST]; /* fw events */
+ enum csio_ln_ev n_evt_sm[CSIO_LNE_MAX_EVENT]; /* State m/c events */
+ uint32_t n_rnode_alloc; /* rnode allocated */
+ uint32_t n_rnode_free; /* rnode freed */
+ uint32_t n_rnode_nomem; /* rnode alloc failure */
+ uint32_t n_input_requests; /* Input Requests */
+ uint32_t n_output_requests; /* Output Requests */
+ uint32_t n_control_requests; /* Control Requests */
+ uint32_t n_input_bytes; /* Input Bytes */
+ uint32_t n_output_bytes; /* Output Bytes */
+ uint32_t rsvd1;
+};
+
+/* Common Lnode params */
+struct csio_lnode_params {
+ uint32_t ra_tov;
+ uint32_t fcfi;
+ uint32_t log_level; /* Module level for debugging */
+};
+
+struct csio_service_parms {
+ struct fc_els_csp csp; /* Common service parms */
+ uint8_t wwpn[8]; /* WWPN */
+ uint8_t wwnn[8]; /* WWNN */
+ struct fc_els_cssp clsp[4]; /* Class service params */
+ uint8_t vvl[16]; /* Vendor version level */
+};
+
+/* Lnode */
+struct csio_lnode {
+ struct csio_sm sm; /* State machine + sibling
+ * lnode list.
+ */
+ struct csio_hw *hwp; /* Pointer to the HW module */
+ uint8_t portid; /* Port ID */
+ uint8_t rsvd1;
+ uint16_t rsvd2;
+ uint32_t dev_num; /* Device number */
+ uint32_t flags; /* Flags */
+ struct list_head fcf_lsthead; /* FCF entries */
+ struct csio_fcf_info *fcfinfo; /* FCF in use */
+ struct csio_ioreq *mgmt_req; /* MGMT request */
+
+ /* FCoE identifiers */
+ uint8_t mac[6];
+ uint32_t nport_id;
+ struct csio_service_parms ln_sparm; /* Service parms */
+
+ /* Firmware identifiers */
+ uint32_t fcf_flowid; /*fcf flowid */
+ uint32_t vnp_flowid;
+ uint16_t ssn_cnt; /* Registered Session */
+ uint8_t cur_evt; /* Current event */
+ uint8_t prev_evt; /* Previous event */
+
+ /* Children */
+ struct list_head cln_head; /* Head of the children lnode
+ * list.
+ */
+ uint32_t num_vports; /* Total NPIV/children LNodes*/
+ struct csio_lnode *pln; /* Parent lnode of child
+ * lnodes.
+ */
+ struct list_head cmpl_q; /* Pending I/Os on this lnode */
+
+ /* Remote node information */
+ struct list_head rnhead; /* Head of rnode list */
+ uint32_t num_reg_rnodes; /* Number of rnodes registered
+ * with the host.
+ */
+ uint32_t n_scsi_tgts; /* Number of scsi targets
+ * found
+ */
+ uint32_t last_scan_ntgts;/* Number of scsi targets
+ * found per last scan.
+ */
+ uint32_t tgt_scan_tick; /* timer started after
+ * new tgt found
+ */
+ /* FC transport data */
+ struct fc_vport *fc_vport;
+ struct fc_host_statistics fch_stats;
+
+ struct csio_lnode_stats stats; /* Common lnode stats */
+ struct csio_lnode_params params; /* Common lnode params */
+};
+
+#define csio_lnode_to_hw(ln) ((ln)->hwp)
+#define csio_root_lnode(ln) (csio_lnode_to_hw((ln))->rln)
+#define csio_parent_lnode(ln) ((ln)->pln)
+#define csio_ln_flowid(ln) ((ln)->vnp_flowid)
+#define csio_ln_wwpn(ln) ((ln)->ln_sparm.wwpn)
+#define csio_ln_wwnn(ln) ((ln)->ln_sparm.wwnn)
+
+#define csio_is_root_ln(ln) (((ln) == csio_root_lnode((ln))) ? 1 : 0)
+#define csio_is_phys_ln(ln) (((ln)->pln == NULL) ? 1 : 0)
+#define csio_is_npiv_ln(ln) (((ln)->pln != NULL) ? 1 : 0)
+
+
+#define csio_ln_dbg(_ln, _fmt, ...) \
+ csio_dbg(_ln->hwp, "%x:%x "_fmt, CSIO_DEVID_HI(_ln), \
+ CSIO_DEVID_LO(_ln), ##__VA_ARGS__);
+
+#define csio_ln_err(_ln, _fmt, ...) \
+ csio_err(_ln->hwp, "%x:%x "_fmt, CSIO_DEVID_HI(_ln), \
+ CSIO_DEVID_LO(_ln), ##__VA_ARGS__);
+
+#define csio_ln_warn(_ln, _fmt, ...) \
+ csio_warn(_ln->hwp, "%x:%x "_fmt, CSIO_DEVID_HI(_ln), \
+ CSIO_DEVID_LO(_ln), ##__VA_ARGS__);
+
+/* HW->Lnode notifications */
+enum csio_ln_notify {
+ CSIO_LN_NOTIFY_HWREADY = 1,
+ CSIO_LN_NOTIFY_HWSTOP,
+ CSIO_LN_NOTIFY_HWREMOVE,
+ CSIO_LN_NOTIFY_HWRESET,
+};
+
+void csio_fcoe_fwevt_handler(struct csio_hw *, __u8 cpl_op, __be64 *);
+int csio_is_lnode_ready(struct csio_lnode *);
+void csio_lnode_state_to_str(struct csio_lnode *ln, int8_t *str);
+struct csio_lnode *csio_lnode_lookup_by_wwpn(struct csio_hw *, uint8_t *);
+int csio_get_phy_port_stats(struct csio_hw *, uint8_t ,
+ struct fw_fcoe_port_stats *);
+int csio_scan_done(struct csio_lnode *, unsigned long, unsigned long,
+ unsigned long, unsigned long);
+void csio_notify_lnodes(struct csio_hw *, enum csio_ln_notify);
+void csio_disable_lnodes(struct csio_hw *, uint8_t, bool);
+void csio_lnode_async_event(struct csio_lnode *, enum csio_ln_fc_evt);
+int csio_ln_fdmi_start(struct csio_lnode *, void *);
+int csio_lnode_start(struct csio_lnode *);
+void csio_lnode_stop(struct csio_lnode *);
+void csio_lnode_close(struct csio_lnode *);
+int csio_lnode_init(struct csio_lnode *, struct csio_hw *,
+ struct csio_lnode *);
+void csio_lnode_exit(struct csio_lnode *);
+
+#endif /* ifndef __CSIO_LNODE_H__ */
diff --git a/drivers/scsi/csiostor/csio_mb.h b/drivers/scsi/csiostor/csio_mb.h
new file mode 100644
index 0000000..1788ea5
--- /dev/null
+++ b/drivers/scsi/csiostor/csio_mb.h
@@ -0,0 +1,278 @@
+/*
+ * This file is part of the Chelsio FCoE driver for Linux.
+ *
+ * Copyright (c) 2008-2012 Chelsio Communications, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __CSIO_MB_H__
+#define __CSIO_MB_H__
+
+#include <linux/timer.h>
+#include <linux/completion.h>
+
+#include "t4fw_api.h"
+#include "t4fw_api_stor.h"
+#include "csio_defs.h"
+
+#define CSIO_STATS_OFFSET (2)
+#define CSIO_NUM_STATS_PER_MB (6)
+
+struct fw_fcoe_port_cmd_params {
+ uint8_t portid;
+ uint8_t idx;
+ uint8_t nstats;
+};
+
+#define CSIO_DUMP_MB(__hw, __num, __mb) \
+ csio_dbg(__hw, "\t%llx %llx %llx %llx %llx %llx %llx %llx\n", \
+ (unsigned long long)csio_rd_reg64(__hw, __mb), \
+ (unsigned long long)csio_rd_reg64(__hw, __mb + 8), \
+ (unsigned long long)csio_rd_reg64(__hw, __mb + 16), \
+ (unsigned long long)csio_rd_reg64(__hw, __mb + 24), \
+ (unsigned long long)csio_rd_reg64(__hw, __mb + 32), \
+ (unsigned long long)csio_rd_reg64(__hw, __mb + 40), \
+ (unsigned long long)csio_rd_reg64(__hw, __mb + 48), \
+ (unsigned long long)csio_rd_reg64(__hw, __mb + 56))
+
+#define CSIO_MB_MAX_REGS 8
+#define CSIO_MAX_MB_SIZE 64
+#define CSIO_MB_POLL_FREQ 5 /* 5 ms */
+#define CSIO_MB_DEFAULT_TMO FW_CMD_MAX_TIMEOUT
+
+/* Device master in HELLO command */
+enum csio_dev_master { CSIO_MASTER_CANT, CSIO_MASTER_MAY, CSIO_MASTER_MUST };
+
+enum csio_mb_owner { CSIO_MBOWNER_NONE, CSIO_MBOWNER_FW, CSIO_MBOWNER_PL };
+
+enum csio_dev_state {
+ CSIO_DEV_STATE_UNINIT,
+ CSIO_DEV_STATE_INIT,
+ CSIO_DEV_STATE_ERR
+};
+
+#define FW_PARAM_DEV(param) \
+ (FW_PARAMS_MNEM(FW_PARAMS_MNEM_DEV) | \
+ FW_PARAMS_PARAM_X(FW_PARAMS_PARAM_DEV_##param))
+
+#define FW_PARAM_PFVF(param) \
+ (FW_PARAMS_MNEM(FW_PARAMS_MNEM_PFVF) | \
+ FW_PARAMS_PARAM_X(FW_PARAMS_PARAM_PFVF_##param)| \
+ FW_PARAMS_PARAM_Y(0) | \
+ FW_PARAMS_PARAM_Z(0))
+
+enum {
+ PAUSE_RX = 1 << 0,
+ PAUSE_TX = 1 << 1,
+ PAUSE_AUTONEG = 1 << 2
+};
+
+#define CSIO_INIT_MBP(__mbp, __cp, __tmo, __priv, __fn, __clear) \
+do { \
+ if (__clear) \
+ memset((__cp), 0, \
+ CSIO_MB_MAX_REGS * sizeof(__be64)); \
+ INIT_LIST_HEAD(&(__mbp)->list); \
+ (__mbp)->tmo = (__tmo); \
+ (__mbp)->priv = (void *)(__priv); \
+ (__mbp)->mb_cbfn = (__fn); \
+ (__mbp)->mb_size = sizeof(*(__cp)); \
+} while (0)
+
+struct csio_mbm_stats {
+ uint32_t n_req; /* number of mbox req */
+ uint32_t n_rsp; /* number of mbox rsp */
+ uint32_t n_activeq; /* number of mbox req active Q */
+ uint32_t n_cbfnq; /* number of mbox req cbfn Q */
+ uint32_t n_tmo; /* number of mbox timeout */
+ uint32_t n_cancel; /* number of mbox cancel */
+ uint32_t n_err; /* number of mbox error */
+};
+
+/* Driver version of Mailbox */
+struct csio_mb {
+ struct list_head list; /* for req/resp */
+ /* queue in driver */
+ __be64 mb[CSIO_MB_MAX_REGS]; /* MB in HW format */
+ int mb_size; /* Size of this
+ * mailbox.
+ */
+ uint32_t tmo; /* Timeout */
+ struct completion cmplobj; /* MB Completion
+ * object
+ */
+ void (*mb_cbfn) (struct csio_hw *, struct csio_mb *);
+ /* Callback fn */
+ void *priv; /* Owner private ptr */
+};
+
+struct csio_mbm {
+ uint32_t a_mbox; /* Async mbox num */
+ uint32_t intr_idx; /* Interrupt index */
+ struct timer_list timer; /* Mbox timer */
+ struct list_head req_q; /* Mbox request queue */
+ struct list_head cbfn_q; /* Mbox completion q */
+ struct csio_mb *mcurrent; /* Current mailbox */
+ uint32_t req_q_cnt; /* Outstanding mbox
+ * cmds
+ */
+ struct csio_mbm_stats stats; /* Statistics */
+};
+
+#define csio_set_mb_intr_idx(_m, _i) ((_m)->intr_idx = (_i))
+#define csio_get_mb_intr_idx(_m) ((_m)->intr_idx)
+
+struct csio_iq_params;
+struct csio_eq_params;
+
+enum fw_retval csio_mb_fw_retval(struct csio_mb *);
+
+/* MB helpers */
+void csio_mb_hello(struct csio_hw *, struct csio_mb *, uint32_t,
+ uint32_t, uint32_t, enum csio_dev_master,
+ void (*)(struct csio_hw *, struct csio_mb *));
+
+void csio_mb_process_hello_rsp(struct csio_hw *, struct csio_mb *,
+ enum fw_retval *, enum csio_dev_state *,
+ uint8_t *);
+
+void csio_mb_bye(struct csio_hw *, struct csio_mb *, uint32_t,
+ void (*)(struct csio_hw *, struct csio_mb *));
+
+void csio_mb_reset(struct csio_hw *, struct csio_mb *, uint32_t, int, int,
+ void (*)(struct csio_hw *, struct csio_mb *));
+
+void csio_mb_params(struct csio_hw *, struct csio_mb *, uint32_t, unsigned int,
+ unsigned int, unsigned int, const u32 *, u32 *, bool,
+ void (*)(struct csio_hw *, struct csio_mb *));
+
+void csio_mb_process_read_params_rsp(struct csio_hw *, struct csio_mb *,
+ enum fw_retval *, unsigned int , u32 *);
+
+void csio_mb_ldst(struct csio_hw *hw, struct csio_mb *mbp, uint32_t tmo,
+ int reg);
+
+void csio_mb_caps_config(struct csio_hw *, struct csio_mb *, uint32_t,
+ bool, bool, bool, bool,
+ void (*)(struct csio_hw *, struct csio_mb *));
+
+void csio_rss_glb_config(struct csio_hw *, struct csio_mb *,
+ uint32_t, uint8_t, unsigned int,
+ void (*)(struct csio_hw *, struct csio_mb *));
+
+void csio_mb_pfvf(struct csio_hw *, struct csio_mb *, uint32_t,
+ unsigned int, unsigned int, unsigned int,
+ unsigned int, unsigned int, unsigned int,
+ unsigned int, unsigned int, unsigned int,
+ unsigned int, unsigned int, unsigned int,
+ unsigned int, void (*) (struct csio_hw *, struct csio_mb *));
+
+void csio_mb_port(struct csio_hw *, struct csio_mb *, uint32_t,
+ uint8_t, bool, uint32_t, uint16_t,
+ void (*) (struct csio_hw *, struct csio_mb *));
+
+void csio_mb_process_read_port_rsp(struct csio_hw *, struct csio_mb *,
+ enum fw_retval *, uint16_t *);
+
+void csio_mb_initialize(struct csio_hw *, struct csio_mb *, uint32_t,
+ void (*)(struct csio_hw *, struct csio_mb *));
+
+void csio_mb_iq_alloc_write(struct csio_hw *, struct csio_mb *, void *,
+ uint32_t, struct csio_iq_params *,
+ void (*) (struct csio_hw *, struct csio_mb *));
+
+void csio_mb_iq_alloc_write_rsp(struct csio_hw *, struct csio_mb *,
+ enum fw_retval *, struct csio_iq_params *);
+
+void csio_mb_iq_free(struct csio_hw *, struct csio_mb *, void *,
+ uint32_t, struct csio_iq_params *,
+ void (*) (struct csio_hw *, struct csio_mb *));
+
+void csio_mb_eq_ofld_alloc_write(struct csio_hw *, struct csio_mb *, void *,
+ uint32_t, struct csio_eq_params *,
+ void (*) (struct csio_hw *, struct csio_mb *));
+
+void csio_mb_eq_ofld_alloc_write_rsp(struct csio_hw *, struct csio_mb *,
+ enum fw_retval *, struct csio_eq_params *);
+
+void csio_mb_eq_ofld_free(struct csio_hw *, struct csio_mb *, void *,
+ uint32_t , struct csio_eq_params *,
+ void (*) (struct csio_hw *, struct csio_mb *));
+
+void csio_fcoe_read_res_info_init_mb(struct csio_hw *, struct csio_mb *,
+ uint32_t,
+ void (*) (struct csio_hw *, struct csio_mb *));
+
+void csio_write_fcoe_link_cond_init_mb(struct csio_lnode *, struct csio_mb *,
+ uint32_t, uint8_t, uint32_t, uint8_t, bool, uint32_t,
+ void (*) (struct csio_hw *, struct csio_mb *));
+
+void csio_fcoe_vnp_alloc_init_mb(struct csio_lnode *, struct csio_mb *,
+ uint32_t, uint32_t , uint32_t , uint16_t,
+ uint8_t [8], uint8_t [8],
+ void (*) (struct csio_hw *, struct csio_mb *));
+
+void csio_fcoe_vnp_read_init_mb(struct csio_lnode *, struct csio_mb *,
+ uint32_t, uint32_t , uint32_t ,
+ void (*) (struct csio_hw *, struct csio_mb *));
+
+void csio_fcoe_vnp_free_init_mb(struct csio_lnode *, struct csio_mb *,
+ uint32_t , uint32_t, uint32_t ,
+ void (*) (struct csio_hw *, struct csio_mb *));
+
+void csio_fcoe_read_fcf_init_mb(struct csio_lnode *, struct csio_mb *,
+ uint32_t, uint32_t, uint32_t,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *));
+
+void csio_fcoe_read_portparams_init_mb(struct csio_hw *hw,
+ struct csio_mb *mbp, uint32_t mb_tmo,
+ struct fw_fcoe_port_cmd_params *portparams,
+ void (*cbfn)(struct csio_hw *, struct csio_mb *));
+
+void csio_mb_process_portparams_rsp(struct csio_hw *hw, struct csio_mb *mbp,
+ enum fw_retval *retval,
+ struct fw_fcoe_port_cmd_params *portparams,
+ struct fw_fcoe_port_stats *portstats);
+
+/* MB module functions */
+int csio_mbm_init(struct csio_mbm *, struct csio_hw *,
+ void (*)(uintptr_t));
+void csio_mbm_exit(struct csio_mbm *);
+void csio_mb_intr_enable(struct csio_hw *);
+void csio_mb_intr_disable(struct csio_hw *);
+
+int csio_mb_issue(struct csio_hw *, struct csio_mb *);
+void csio_mb_completions(struct csio_hw *, struct list_head *);
+int csio_mb_fwevt_handler(struct csio_hw *, __be64 *);
+int csio_mb_isr_handler(struct csio_hw *);
+struct csio_mb *csio_mb_tmo_handler(struct csio_hw *);
+void csio_mb_cancel_all(struct csio_hw *, struct list_head *);
+
+#endif /* ifndef __CSIO_MB_H__ */
diff --git a/drivers/scsi/csiostor/csio_rnode.h b/drivers/scsi/csiostor/csio_rnode.h
new file mode 100644
index 0000000..a3b434c
--- /dev/null
+++ b/drivers/scsi/csiostor/csio_rnode.h
@@ -0,0 +1,141 @@
+/*
+ * This file is part of the Chelsio FCoE driver for Linux.
+ *
+ * Copyright (c) 2008-2012 Chelsio Communications, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __CSIO_RNODE_H__
+#define __CSIO_RNODE_H__
+
+#include "csio_defs.h"
+
+/* State machine evets */
+enum csio_rn_ev {
+ CSIO_RNFE_NONE = (uint32_t)0, /* None */
+ CSIO_RNFE_LOGGED_IN, /* [N/F]Port login
+ * complete.
+ */
+ CSIO_RNFE_PRLI_DONE, /* PRLI completed */
+ CSIO_RNFE_PLOGI_RECV, /* Received PLOGI */
+ CSIO_RNFE_PRLI_RECV, /* Received PLOGI */
+ CSIO_RNFE_LOGO_RECV, /* Received LOGO */
+ CSIO_RNFE_PRLO_RECV, /* Received PRLO */
+ CSIO_RNFE_DOWN, /* Rnode is down */
+ CSIO_RNFE_CLOSE, /* Close rnode */
+ CSIO_RNFE_NAME_MISSING, /* Rnode name missing
+ * in name server.
+ */
+ CSIO_RNFE_MAX_EVENT,
+};
+
+/* rnode stats */
+struct csio_rnode_stats {
+ uint32_t n_err; /* error */
+ uint32_t n_err_inval; /* invalid parameter */
+ uint32_t n_err_nomem; /* error nomem */
+ uint32_t n_evt_unexp; /* unexpected event */
+ uint32_t n_evt_drop; /* unexpected event */
+ uint32_t n_evt_fw[RSCN_DEV_LOST]; /* fw events */
+ enum csio_rn_ev n_evt_sm[CSIO_RNFE_MAX_EVENT]; /* State m/c events */
+ uint32_t n_lun_rst; /* Number of resets of
+ * of LUNs under this
+ * target
+ */
+ uint32_t n_lun_rst_fail; /* Number of LUN reset
+ * failures.
+ */
+ uint32_t n_tgt_rst; /* Number of target resets */
+ uint32_t n_tgt_rst_fail; /* Number of target reset
+ * failures.
+ */
+};
+
+/* Defines for rnode role */
+#define CSIO_RNFR_INITIATOR 0x1
+#define CSIO_RNFR_TARGET 0x2
+#define CSIO_RNFR_FABRIC 0x4
+#define CSIO_RNFR_NS 0x8
+#define CSIO_RNFR_NPORT 0x10
+
+struct csio_rnode {
+ struct csio_sm sm; /* State machine -
+ * should be the
+ * 1st member
+ */
+ struct csio_lnode *lnp; /* Pointer to owning
+ * Lnode */
+ uint32_t flowid; /* Firmware ID */
+ struct list_head host_cmpl_q; /* SCSI IOs
+ * pending to completed
+ * to Mid-layer.
+ */
+ /* FC identifiers for remote node */
+ uint32_t nport_id;
+ uint16_t fcp_flags; /* FCP Flags */
+ uint8_t cur_evt; /* Current event */
+ uint8_t prev_evt; /* Previous event */
+ uint32_t role; /* Fabric/Target/
+ * Initiator/NS
+ */
+ struct fcoe_rdev_entry *rdev_entry; /* Rdev entry */
+ struct csio_service_parms rn_sparm;
+
+ /* FC transport attributes */
+ struct fc_rport *rport; /* FC transport rport */
+ uint32_t supp_classes; /* Supported FC classes */
+ uint32_t maxframe_size; /* Max Frame size */
+ uint32_t scsi_id; /* Transport given SCSI id */
+
+ struct csio_rnode_stats stats; /* Common rnode stats */
+};
+
+#define csio_rn_flowid(rn) ((rn)->flowid)
+#define csio_rn_wwpn(rn) ((rn)->rn_sparm.wwpn)
+#define csio_rn_wwnn(rn) ((rn)->rn_sparm.wwnn)
+#define csio_rnode_to_lnode(rn) ((rn)->lnp)
+
+int csio_is_rnode_ready(struct csio_rnode *rn);
+void csio_rnode_state_to_str(struct csio_rnode *rn, int8_t *str);
+
+struct csio_rnode *csio_rnode_lookup_portid(struct csio_lnode *, uint32_t);
+struct csio_rnode *csio_confirm_rnode(struct csio_lnode *,
+ uint32_t, struct fcoe_rdev_entry *);
+
+void csio_rnode_fwevt_handler(struct csio_rnode *rn, uint8_t fwevt);
+
+void csio_put_rnode(struct csio_lnode *ln, struct csio_rnode *rn);
+
+void csio_reg_rnode(struct csio_rnode *);
+void csio_unreg_rnode(struct csio_rnode *);
+
+void csio_rnode_devloss_handler(struct csio_rnode *);
+
+#endif /* ifndef __CSIO_RNODE_H__ */
diff --git a/drivers/scsi/csiostor/csio_scsi.h b/drivers/scsi/csiostor/csio_scsi.h
new file mode 100644
index 0000000..2257c3d
--- /dev/null
+++ b/drivers/scsi/csiostor/csio_scsi.h
@@ -0,0 +1,342 @@
+/*
+ * This file is part of the Chelsio FCoE driver for Linux.
+ *
+ * Copyright (c) 2008-2012 Chelsio Communications, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __CSIO_SCSI_H__
+#define __CSIO_SCSI_H__
+
+#include <linux/spinlock_types.h>
+#include <linux/completion.h>
+#include <scsi/scsi.h>
+#include <scsi/scsi_cmnd.h>
+#include <scsi/scsi_device.h>
+#include <scsi/scsi_host.h>
+#include <scsi/scsi_eh.h>
+#include <scsi/scsi_tcq.h>
+#include <scsi/fc/fc_fcp.h>
+
+#include "csio_defs.h"
+#include "csio_wr.h"
+
+extern struct scsi_host_template csio_fcoe_shost_template;
+extern struct scsi_host_template csio_fcoe_shost_vport_template;
+
+extern int csio_scsi_eqsize;
+extern int csio_scsi_iqlen;
+extern int csio_scsi_ioreqs;
+extern uint32_t csio_max_scan_tmo;
+extern uint32_t csio_delta_scan_tmo;
+extern int csio_lun_qdepth;
+
+/*
+ **************************** NOTE *******************************
+ * How do we calculate MAX FCoE SCSI SGEs? Here is the math:
+ * Max Egress WR size = 512 bytes
+ * One SCSI egress WR has the following fixed no of bytes:
+ * 48 (sizeof(struct fw_scsi_write[read]_wr)) - FW WR
+ * + 32 (sizeof(struct fc_fcp_cmnd)) - Immediate FCP_CMD
+ * ------
+ * 80
+ * ------
+ * That leaves us with 512 - 96 = 432 bytes for data SGE. Using
+ * struct ulptx_sgl header for the SGE consumes:
+ * - 4 bytes for cmnd_sge.
+ * - 12 bytes for the first SGL.
+ * That leaves us with 416 bytes for the remaining SGE pairs. Which is
+ * is 416 / 24 (size(struct ulptx_sge_pair)) = 17 SGE pairs,
+ * or 34 SGEs. Adding the first SGE fetches us 35 SGEs.
+ */
+#define CSIO_SCSI_MAX_SGE 35
+#define CSIO_SCSI_ABRT_TMO_MS 60000
+#define CSIO_SCSI_LUNRST_TMO_MS 60000
+#define CSIO_SCSI_TM_POLL_MS 2000 /* should be less than
+ * all TM timeouts.
+ */
+#define CSIO_SCSI_IQ_WRSZ 128
+#define CSIO_SCSI_IQSIZE (csio_scsi_iqlen * CSIO_SCSI_IQ_WRSZ)
+
+#define CSIO_MAX_SNS_LEN 128
+#define CSIO_SCSI_RSP_LEN (FCP_RESP_WITH_EXT + 4 + CSIO_MAX_SNS_LEN)
+
+/* Reference to scsi_cmnd */
+#define csio_scsi_cmnd(req) ((req)->scratch1)
+
+struct csio_scsi_stats {
+ uint64_t n_tot_success; /* Total number of good I/Os */
+ uint32_t n_rn_nr_error; /* No. of remote-node-not-
+ * ready errors
+ */
+ uint32_t n_hw_nr_error; /* No. of hw-module-not-
+ * ready errors
+ */
+ uint32_t n_dmamap_error; /* No. of DMA map erros */
+ uint32_t n_unsupp_sge_error; /* No. of too-many-SGes
+ * errors.
+ */
+ uint32_t n_no_req_error; /* No. of Out-of-ioreqs error */
+ uint32_t n_busy_error; /* No. of -EBUSY errors */
+ uint32_t n_hosterror; /* No. of FW_HOSTERROR I/O */
+ uint32_t n_rsperror; /* No. of response errors */
+ uint32_t n_autosense; /* No. of auto sense replies */
+ uint32_t n_ovflerror; /* No. of overflow errors */
+ uint32_t n_unflerror; /* No. of underflow errors */
+ uint32_t n_rdev_nr_error;/* No. of rdev not
+ * ready errors
+ */
+ uint32_t n_rdev_lost_error;/* No. of rdev lost errors */
+ uint32_t n_rdev_logo_error;/* No. of rdev logo errors */
+ uint32_t n_link_down_error;/* No. of link down errors */
+ uint32_t n_no_xchg_error; /* No. no exchange error */
+ uint32_t n_unknown_error;/* No. of unhandled errors */
+ uint32_t n_aborted; /* No. of aborted I/Os */
+ uint32_t n_abrt_timedout; /* No. of abort timedouts */
+ uint32_t n_abrt_fail; /* No. of abort failures */
+ uint32_t n_abrt_dups; /* No. of duplicate aborts */
+ uint32_t n_abrt_race_comp; /* No. of aborts that raced
+ * with completions.
+ */
+ uint32_t n_abrt_busy_error;/* No. of abort failures
+ * due to -EBUSY.
+ */
+ uint32_t n_closed; /* No. of closed I/Os */
+ uint32_t n_cls_busy_error; /* No. of close failures
+ * due to -EBUSY.
+ */
+ uint32_t n_active; /* No. of IOs in active_q */
+ uint32_t n_tm_active; /* No. of TMs in active_q */
+ uint32_t n_wcbfn; /* No. of I/Os in worker
+ * cbfn q
+ */
+ uint32_t n_free_ioreq; /* No. of freelist entries */
+ uint32_t n_free_ddp; /* No. of DDP freelist */
+ uint32_t n_unaligned; /* No. of Unaligned SGls */
+ uint32_t n_inval_cplop; /* No. invalid CPL op's in IQ */
+ uint32_t n_inval_scsiop; /* No. invalid scsi op's in IQ*/
+};
+
+struct csio_scsim {
+ struct csio_hw *hw; /* Pointer to HW moduel */
+ uint8_t max_sge; /* Max SGE */
+ uint8_t proto_cmd_len; /* Proto specific SCSI
+ * cmd length
+ */
+ uint16_t proto_rsp_len; /* Proto specific SCSI
+ * response length
+ */
+ spinlock_t freelist_lock; /* Lock for ioreq freelist */
+ struct list_head active_q; /* Outstanding SCSI I/Os */
+ struct list_head ioreq_freelist; /* Free list of ioreq's */
+ struct list_head ddp_freelist; /* DDP descriptor freelist */
+ struct csio_scsi_stats stats; /* This module's statistics */
+};
+
+/* State machine defines */
+enum csio_scsi_ev {
+ CSIO_SCSIE_START_IO = 1, /* Start a regular SCSI IO */
+ CSIO_SCSIE_START_TM, /* Start a TM IO */
+ CSIO_SCSIE_COMPLETED, /* IO Completed */
+ CSIO_SCSIE_ABORT, /* Abort IO */
+ CSIO_SCSIE_ABORTED, /* IO Aborted */
+ CSIO_SCSIE_CLOSE, /* Close exchange */
+ CSIO_SCSIE_CLOSED, /* Exchange closed */
+ CSIO_SCSIE_DRVCLEANUP, /* Driver wants to manually
+ * cleanup this I/O.
+ */
+};
+
+enum csio_scsi_lev {
+ CSIO_LEV_ALL = 1,
+ CSIO_LEV_LNODE,
+ CSIO_LEV_RNODE,
+ CSIO_LEV_LUN,
+};
+
+struct csio_scsi_level_data {
+ enum csio_scsi_lev level;
+ struct csio_rnode *rnode;
+ struct csio_lnode *lnode;
+ uint64_t oslun;
+};
+
+static inline struct csio_ioreq *
+csio_get_scsi_ioreq(struct csio_scsim *scm)
+{
+ struct csio_sm *req;
+
+ if (likely(!list_empty(&scm->ioreq_freelist))) {
+ req = list_first_entry(&scm->ioreq_freelist,
+ struct csio_sm, sm_list);
+ list_del_init(&req->sm_list);
+ CSIO_DEC_STATS(scm, n_free_ioreq);
+ return (struct csio_ioreq *)req;
+ } else
+ return NULL;
+}
+
+static inline void
+csio_put_scsi_ioreq(struct csio_scsim *scm, struct csio_ioreq *ioreq)
+{
+ list_add_tail(&ioreq->sm.sm_list, &scm->ioreq_freelist);
+ CSIO_INC_STATS(scm, n_free_ioreq);
+}
+
+static inline void
+csio_put_scsi_ioreq_list(struct csio_scsim *scm, struct list_head *reqlist,
+ int n)
+{
+ list_splice_init(reqlist, &scm->ioreq_freelist);
+ scm->stats.n_free_ioreq += n;
+}
+
+static inline struct csio_dma_buf *
+csio_get_scsi_ddp(struct csio_scsim *scm)
+{
+ struct csio_dma_buf *ddp;
+
+ if (likely(!list_empty(&scm->ddp_freelist))) {
+ ddp = list_first_entry(&scm->ddp_freelist,
+ struct csio_dma_buf, list);
+ list_del_init(&ddp->list);
+ CSIO_DEC_STATS(scm, n_free_ddp);
+ return ddp;
+ } else
+ return NULL;
+}
+
+static inline void
+csio_put_scsi_ddp(struct csio_scsim *scm, struct csio_dma_buf *ddp)
+{
+ list_add_tail(&ddp->list, &scm->ddp_freelist);
+ CSIO_INC_STATS(scm, n_free_ddp);
+}
+
+static inline void
+csio_put_scsi_ddp_list(struct csio_scsim *scm, struct list_head *reqlist,
+ int n)
+{
+ list_splice_tail_init(reqlist, &scm->ddp_freelist);
+ scm->stats.n_free_ddp += n;
+}
+
+static inline void
+csio_scsi_completed(struct csio_ioreq *ioreq, struct list_head *cbfn_q)
+{
+ csio_post_event(&ioreq->sm, CSIO_SCSIE_COMPLETED);
+ if (csio_list_deleted(&ioreq->sm.sm_list))
+ list_add_tail(&ioreq->sm.sm_list, cbfn_q);
+}
+
+static inline void
+csio_scsi_aborted(struct csio_ioreq *ioreq, struct list_head *cbfn_q)
+{
+ csio_post_event(&ioreq->sm, CSIO_SCSIE_ABORTED);
+ list_add_tail(&ioreq->sm.sm_list, cbfn_q);
+}
+
+static inline void
+csio_scsi_closed(struct csio_ioreq *ioreq, struct list_head *cbfn_q)
+{
+ csio_post_event(&ioreq->sm, CSIO_SCSIE_CLOSED);
+ list_add_tail(&ioreq->sm.sm_list, cbfn_q);
+}
+
+static inline void
+csio_scsi_drvcleanup(struct csio_ioreq *ioreq)
+{
+ csio_post_event(&ioreq->sm, CSIO_SCSIE_DRVCLEANUP);
+}
+
+/*
+ * csio_scsi_start_io - Kick starts the IO SM.
+ * @req: io request SM.
+ *
+ * needs to be called with lock held.
+ */
+static inline int
+csio_scsi_start_io(struct csio_ioreq *ioreq)
+{
+ csio_post_event(&ioreq->sm, CSIO_SCSIE_START_IO);
+ return ioreq->drv_status;
+}
+
+/*
+ * csio_scsi_start_tm - Kicks off the Task management IO SM.
+ * @req: io request SM.
+ *
+ * needs to be called with lock held.
+ */
+static inline int
+csio_scsi_start_tm(struct csio_ioreq *ioreq)
+{
+ csio_post_event(&ioreq->sm, CSIO_SCSIE_START_TM);
+ return ioreq->drv_status;
+}
+
+/*
+ * csio_scsi_abort - Abort an IO request
+ * @req: io request SM.
+ *
+ * needs to be called with lock held.
+ */
+static inline int
+csio_scsi_abort(struct csio_ioreq *ioreq)
+{
+ csio_post_event(&ioreq->sm, CSIO_SCSIE_ABORT);
+ return ioreq->drv_status;
+}
+
+/*
+ * csio_scsi_close - Close an IO request
+ * @req: io request SM.
+ *
+ * needs to be called with lock held.
+ */
+static inline int
+csio_scsi_close(struct csio_ioreq *ioreq)
+{
+ csio_post_event(&ioreq->sm, CSIO_SCSIE_CLOSE);
+ return ioreq->drv_status;
+}
+
+void csio_scsi_cleanup_io_q(struct csio_scsim *, struct list_head *);
+int csio_scsim_cleanup_io(struct csio_scsim *, bool abort);
+int csio_scsim_cleanup_io_lnode(struct csio_scsim *,
+ struct csio_lnode *);
+struct csio_ioreq *csio_scsi_cmpl_handler(struct csio_hw *, void *, uint32_t,
+ struct csio_fl_dma_buf *,
+ void *, uint8_t **);
+int csio_scsi_qconfig(struct csio_hw *);
+int csio_scsim_init(struct csio_scsim *, struct csio_hw *);
+void csio_scsim_exit(struct csio_scsim *);
+
+#endif /* __CSIO_SCSI_H__ */
diff --git a/drivers/scsi/csiostor/csio_wr.h b/drivers/scsi/csiostor/csio_wr.h
new file mode 100644
index 0000000..8d30e7a
--- /dev/null
+++ b/drivers/scsi/csiostor/csio_wr.h
@@ -0,0 +1,512 @@
+/*
+ * This file is part of the Chelsio FCoE driver for Linux.
+ *
+ * Copyright (c) 2008-2012 Chelsio Communications, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#ifndef __CSIO_WR_H__
+#define __CSIO_WR_H__
+
+#include <linux/cache.h>
+
+#include "csio_defs.h"
+#include "t4fw_api.h"
+#include "t4fw_api_stor.h"
+
+/*
+ * SGE register field values.
+ */
+#define X_INGPCIEBOUNDARY_32B 0
+#define X_INGPCIEBOUNDARY_64B 1
+#define X_INGPCIEBOUNDARY_128B 2
+#define X_INGPCIEBOUNDARY_256B 3
+#define X_INGPCIEBOUNDARY_512B 4
+#define X_INGPCIEBOUNDARY_1024B 5
+#define X_INGPCIEBOUNDARY_2048B 6
+#define X_INGPCIEBOUNDARY_4096B 7
+
+/* GTS register */
+#define X_TIMERREG_COUNTER0 0
+#define X_TIMERREG_COUNTER1 1
+#define X_TIMERREG_COUNTER2 2
+#define X_TIMERREG_COUNTER3 3
+#define X_TIMERREG_COUNTER4 4
+#define X_TIMERREG_COUNTER5 5
+#define X_TIMERREG_RESTART_COUNTER 6
+#define X_TIMERREG_UPDATE_CIDX 7
+
+/*
+ * Egress Context field values
+ */
+#define X_FETCHBURSTMIN_16B 0
+#define X_FETCHBURSTMIN_32B 1
+#define X_FETCHBURSTMIN_64B 2
+#define X_FETCHBURSTMIN_128B 3
+
+#define X_FETCHBURSTMAX_64B 0
+#define X_FETCHBURSTMAX_128B 1
+#define X_FETCHBURSTMAX_256B 2
+#define X_FETCHBURSTMAX_512B 3
+
+#define X_HOSTFCMODE_NONE 0
+#define X_HOSTFCMODE_INGRESS_QUEUE 1
+#define X_HOSTFCMODE_STATUS_PAGE 2
+#define X_HOSTFCMODE_BOTH 3
+
+/*
+ * Ingress Context field values
+ */
+#define X_UPDATESCHEDULING_TIMER 0
+#define X_UPDATESCHEDULING_COUNTER_OPTTIMER 1
+
+#define X_UPDATEDELIVERY_NONE 0
+#define X_UPDATEDELIVERY_INTERRUPT 1
+#define X_UPDATEDELIVERY_STATUS_PAGE 2
+#define X_UPDATEDELIVERY_BOTH 3
+
+#define X_INTERRUPTDESTINATION_PCIE 0
+#define X_INTERRUPTDESTINATION_IQ 1
+
+#define X_RSPD_TYPE_FLBUF 0
+#define X_RSPD_TYPE_CPL 1
+#define X_RSPD_TYPE_INTR 2
+
+/* WR status is at the same position as retval in a CMD header */
+#define csio_wr_status(_wr) \
+ (FW_CMD_RETVAL_GET(ntohl(((struct fw_cmd_hdr *)(_wr))->lo)))
+
+struct csio_hw;
+
+extern int csio_intr_coalesce_cnt;
+extern int csio_intr_coalesce_time;
+
+/* Ingress queue params */
+struct csio_iq_params {
+
+ uint8_t iq_start:1;
+ uint8_t iq_stop:1;
+ uint8_t pfn:3;
+
+ uint8_t vfn;
+
+ uint16_t physiqid;
+ uint16_t iqid;
+
+ uint16_t fl0id;
+ uint16_t fl1id;
+
+ uint8_t viid;
+
+ uint8_t type;
+ uint8_t iqasynch;
+ uint8_t reserved4;
+
+ uint8_t iqandst;
+ uint8_t iqanus;
+ uint8_t iqanud;
+
+ uint16_t iqandstindex;
+
+ uint8_t iqdroprss;
+ uint8_t iqpciech;
+ uint8_t iqdcaen;
+
+ uint8_t iqdcacpu;
+ uint8_t iqintcntthresh;
+ uint8_t iqo;
+
+ uint8_t iqcprio;
+ uint8_t iqesize;
+
+ uint16_t iqsize;
+
+ uint64_t iqaddr;
+
+ uint8_t iqflintiqhsen;
+ uint8_t reserved5;
+ uint8_t iqflintcongen;
+ uint8_t iqflintcngchmap;
+
+ uint32_t reserved6;
+
+ uint8_t fl0hostfcmode;
+ uint8_t fl0cprio;
+ uint8_t fl0paden;
+ uint8_t fl0packen;
+ uint8_t fl0congen;
+ uint8_t fl0dcaen;
+
+ uint8_t fl0dcacpu;
+ uint8_t fl0fbmin;
+
+ uint8_t fl0fbmax;
+ uint8_t fl0cidxfthresho;
+ uint8_t fl0cidxfthresh;
+
+ uint16_t fl0size;
+
+ uint64_t fl0addr;
+
+ uint64_t reserved7;
+
+ uint8_t fl1hostfcmode;
+ uint8_t fl1cprio;
+ uint8_t fl1paden;
+ uint8_t fl1packen;
+ uint8_t fl1congen;
+ uint8_t fl1dcaen;
+
+ uint8_t fl1dcacpu;
+ uint8_t fl1fbmin;
+
+ uint8_t fl1fbmax;
+ uint8_t fl1cidxfthresho;
+ uint8_t fl1cidxfthresh;
+
+ uint16_t fl1size;
+
+ uint64_t fl1addr;
+};
+
+/* Egress queue params */
+struct csio_eq_params {
+
+ uint8_t pfn;
+ uint8_t vfn;
+
+ uint8_t eqstart:1;
+ uint8_t eqstop:1;
+
+ uint16_t physeqid;
+ uint32_t eqid;
+
+ uint8_t hostfcmode:2;
+ uint8_t cprio:1;
+ uint8_t pciechn:3;
+
+ uint16_t iqid;
+
+ uint8_t dcaen:1;
+ uint8_t dcacpu:5;
+
+ uint8_t fbmin:3;
+ uint8_t fbmax:3;
+
+ uint8_t cidxfthresho:1;
+ uint8_t cidxfthresh:3;
+
+ uint16_t eqsize;
+
+ uint64_t eqaddr;
+};
+
+struct csio_dma_buf {
+ struct list_head list;
+ void *vaddr; /* Virtual address */
+ dma_addr_t paddr; /* Physical address */
+ uint32_t len; /* Buffer size */
+};
+
+/* Generic I/O request structure */
+struct csio_ioreq {
+ struct csio_sm sm; /* SM, List
+ * should be the first member
+ */
+ int iq_idx; /* Ingress queue index */
+ int eq_idx; /* Egress queue index */
+ uint32_t nsge; /* Number of SG elements */
+ uint32_t tmo; /* Driver timeout */
+ uint32_t datadir; /* Data direction */
+ struct csio_dma_buf dma_buf; /* Req/resp DMA buffers */
+ uint16_t wr_status; /* WR completion status */
+ int16_t drv_status; /* Driver internal status */
+ struct csio_lnode *lnode; /* Owner lnode */
+ struct csio_rnode *rnode; /* Src/destination rnode */
+ void (*io_cbfn) (struct csio_hw *, struct csio_ioreq *);
+ /* completion callback */
+ void *scratch1; /* Scratch area 1.
+ */
+ void *scratch2; /* Scratch area 2. */
+ struct list_head gen_list; /* Any list associated with
+ * this ioreq.
+ */
+ uint64_t fw_handle; /* Unique handle passed
+ * to FW
+ */
+ uint8_t dcopy; /* Data copy required */
+ uint8_t reserved1;
+ uint16_t reserved2;
+ struct completion cmplobj; /* ioreq completion object */
+} ____cacheline_aligned_in_smp;
+
+/*
+ * Egress status page for egress cidx updates
+ */
+struct csio_qstatus_page {
+ __be32 qid;
+ __be16 cidx;
+ __be16 pidx;
+};
+
+
+enum {
+ CSIO_MAX_FLBUF_PER_IQWR = 4,
+ CSIO_QCREDIT_SZ = 64, /* pidx/cidx increments
+ * in bytes
+ */
+ CSIO_MAX_QID = 0xFFFF,
+ CSIO_MAX_IQ = 128,
+
+ CSIO_SGE_NTIMERS = 6,
+ CSIO_SGE_NCOUNTERS = 4,
+ CSIO_SGE_FL_SIZE_REGS = 16,
+};
+
+/* Defines for type */
+enum {
+ CSIO_EGRESS = 1,
+ CSIO_INGRESS = 2,
+ CSIO_FREELIST = 3,
+};
+
+/*
+ * Structure for footer (last 2 flits) of Ingress Queue Entry.
+ */
+struct csio_iqwr_footer {
+ __be32 hdrbuflen_pidx;
+ __be32 pldbuflen_qid;
+ union {
+ u8 type_gen;
+ __be64 last_flit;
+ } u;
+};
+
+#define IQWRF_NEWBUF (1 << 31)
+#define IQWRF_LEN_GET(x) (((x) >> 0) & 0x7fffffffU)
+#define IQWRF_GEN_SHIFT 7
+#define IQWRF_TYPE_GET(x) (((x) >> 4) & 0x3U)
+
+
+/*
+ * WR pair:
+ * ========
+ * A WR can start towards the end of a queue, and then continue at the
+ * beginning, since the queue is considered to be circular. This will
+ * require a pair of address/len to be passed back to the caller -
+ * hence the Work request pair structure.
+ */
+struct csio_wr_pair {
+ void *addr1;
+ uint32_t size1;
+ void *addr2;
+ uint32_t size2;
+};
+
+/*
+ * The following structure is used by ingress processing to return the
+ * free list buffers to consumers.
+ */
+struct csio_fl_dma_buf {
+ struct csio_dma_buf flbufs[CSIO_MAX_FLBUF_PER_IQWR];
+ /* Freelist DMA buffers */
+ int offset; /* Offset within the
+ * first FL buf.
+ */
+ uint32_t totlen; /* Total length */
+ uint8_t defer_free; /* Free of buffer can
+ * deferred
+ */
+};
+
+/* Data-types */
+typedef void (*iq_handler_t)(struct csio_hw *, void *, uint32_t,
+ struct csio_fl_dma_buf *, void *);
+
+struct csio_iq {
+ uint16_t iqid; /* Queue ID */
+ uint16_t physiqid; /* Physical Queue ID */
+ uint16_t genbit; /* Generation bit,
+ * initially set to 1
+ */
+ int flq_idx; /* Freelist queue index */
+ iq_handler_t iq_intx_handler; /* IQ INTx handler routine */
+};
+
+struct csio_eq {
+ uint16_t eqid; /* Qid */
+ uint16_t physeqid; /* Physical Queue ID */
+ uint8_t wrap[512]; /* Temp area for q-wrap around*/
+};
+
+struct csio_fl {
+ uint16_t flid; /* Qid */
+ uint16_t packen; /* Packing enabled? */
+ int offset; /* Offset within FL buf */
+ int sreg; /* Size register */
+ struct csio_dma_buf *bufs; /* Free list buffer ptr array
+ * indexed using flq->cidx/pidx
+ */
+};
+
+struct csio_qstats {
+ uint32_t n_tot_reqs; /* Total no. of Requests */
+ uint32_t n_tot_rsps; /* Total no. of responses */
+ uint32_t n_qwrap; /* Queue wraps */
+ uint32_t n_eq_wr_split; /* Number of split EQ WRs */
+ uint32_t n_qentry; /* Queue entry */
+ uint32_t n_qempty; /* Queue empty */
+ uint32_t n_qfull; /* Queue fulls */
+ uint32_t n_rsp_unknown; /* Unknown response type */
+ uint32_t n_stray_comp; /* Stray completion intr */
+ uint32_t n_flq_refill; /* Number of FL refills */
+};
+
+/* Queue metadata */
+struct csio_q {
+ uint16_t type; /* Type: Ingress/Egress/FL */
+ uint16_t pidx; /* producer index */
+ uint16_t cidx; /* consumer index */
+ uint16_t inc_idx; /* Incremental index */
+ uint32_t wr_sz; /* Size of all WRs in this q
+ * if fixed
+ */
+ void *vstart; /* Base virtual address
+ * of queue
+ */
+ void *vwrap; /* Virtual end address to
+ * wrap around at
+ */
+ uint32_t credits; /* Size of queue in credits */
+ void *owner; /* Owner */
+ union { /* Queue contexts */
+ struct csio_iq iq;
+ struct csio_eq eq;
+ struct csio_fl fl;
+ } un;
+
+ dma_addr_t pstart; /* Base physical address of
+ * queue
+ */
+ uint32_t portid; /* PCIE Channel */
+ uint32_t size; /* Size of queue in bytes */
+ struct csio_qstats stats; /* Statistics */
+} ____cacheline_aligned_in_smp;
+
+struct csio_sge {
+ uint32_t csio_fl_align; /* Calculated and cached
+ * for fast path
+ */
+ uint32_t sge_control; /* padding, boundaries,
+ * lengths, etc.
+ */
+ uint32_t sge_host_page_size; /* Host page size */
+ uint32_t sge_fl_buf_size[CSIO_SGE_FL_SIZE_REGS];
+ /* free list buffer sizes */
+ uint16_t timer_val[CSIO_SGE_NTIMERS];
+ uint8_t counter_val[CSIO_SGE_NCOUNTERS];
+};
+
+/* Work request module */
+struct csio_wrm {
+ int num_q; /* Number of queues */
+ struct csio_q **q_arr; /* Array of queue pointers
+ * allocated dynamically
+ * based on configured values
+ */
+ uint32_t fw_iq_start; /* Start ID of IQ for this fn*/
+ uint32_t fw_eq_start; /* Start ID of EQ for this fn*/
+ struct csio_q *intr_map[CSIO_MAX_IQ];
+ /* IQ-id to IQ map table. */
+ int free_qidx; /* queue idx of free queue */
+ struct csio_sge sge; /* SGE params */
+};
+
+#define csio_get_q(__hw, __idx) ((__hw)->wrm.q_arr[__idx])
+#define csio_q_type(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->type)
+#define csio_q_pidx(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->pidx)
+#define csio_q_cidx(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->cidx)
+#define csio_q_inc_idx(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->inc_idx)
+#define csio_q_vstart(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->vstart)
+#define csio_q_pstart(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->pstart)
+#define csio_q_size(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->size)
+#define csio_q_credits(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->credits)
+#define csio_q_portid(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->portid)
+#define csio_q_wr_sz(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->wr_sz)
+#define csio_q_iqid(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->un.iq.iqid)
+#define csio_q_physiqid(__hw, __idx) \
+ ((__hw)->wrm.q_arr[(__idx)]->un.iq.physiqid)
+#define csio_q_iq_flq_idx(__hw, __idx) \
+ ((__hw)->wrm.q_arr[(__idx)]->un.iq.flq_idx)
+#define csio_q_eqid(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->un.eq.eqid)
+#define csio_q_flid(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->un.fl.flid)
+
+#define csio_q_physeqid(__hw, __idx) \
+ ((__hw)->wrm.q_arr[(__idx)]->un.eq.physeqid)
+#define csio_iq_has_fl(__iq) ((__iq)->un.iq.flq_idx != -1)
+
+#define csio_q_iq_to_flid(__hw, __iq_idx) \
+ csio_q_flid((__hw), (__hw)->wrm.q_arr[(__iq_qidx)]->un.iq.flq_idx)
+#define csio_q_set_intr_map(__hw, __iq_idx, __rel_iq_id) \
+ (__hw)->wrm.intr_map[__rel_iq_id] = csio_get_q(__hw, __iq_idx)
+#define csio_q_eq_wrap(__hw, __idx) ((__hw)->wrm.q_arr[(__idx)]->un.eq.wrap)
+
+struct csio_mb;
+
+int csio_wr_alloc_q(struct csio_hw *, uint32_t, uint32_t,
+ uint16_t, void *, uint32_t, int, iq_handler_t);
+int csio_wr_iq_create(struct csio_hw *, void *, int,
+ uint32_t, uint8_t, bool,
+ void (*)(struct csio_hw *, struct csio_mb *));
+int csio_wr_eq_create(struct csio_hw *, void *, int, int, uint8_t,
+ void (*)(struct csio_hw *, struct csio_mb *));
+int csio_wr_destroy_queues(struct csio_hw *, bool cmd);
+
+
+int csio_wr_get(struct csio_hw *, int, uint32_t,
+ struct csio_wr_pair *);
+void csio_wr_copy_to_wrp(void *, struct csio_wr_pair *, uint32_t, uint32_t);
+int csio_wr_issue(struct csio_hw *, int, bool);
+int csio_wr_process_iq(struct csio_hw *, struct csio_q *,
+ void (*)(struct csio_hw *, void *,
+ uint32_t, struct csio_fl_dma_buf *,
+ void *),
+ void *);
+int csio_wr_process_iq_idx(struct csio_hw *, int,
+ void (*)(struct csio_hw *, void *,
+ uint32_t, struct csio_fl_dma_buf *,
+ void *),
+ void *);
+
+void csio_wr_sge_init(struct csio_hw *);
+int csio_wrm_init(struct csio_wrm *, struct csio_hw *);
+void csio_wrm_exit(struct csio_wrm *, struct csio_hw *);
+
+#endif /* ifndef __CSIO_WR_H__ */
--
1.7.1
^ permalink raw reply related
* [V2 PATCH 4/9] csiostor: Chelsio FCoE offload driver submission (sources part 4).
From: Naresh Kumar Inna @ 2012-09-05 12:33 UTC (permalink / raw)
To: JBottomley, linux-scsi, dm, leedom; +Cc: netdev, naresh, chethan
In-Reply-To: <1346848442-4573-1-git-send-email-naresh@chelsio.com>
This patch contains code to implement the local and remote node port
functionality. It includes tracking the firmware events for changes to
the states of these ports.
Signed-off-by: Naresh Kumar Inna <naresh@chelsio.com>
---
drivers/scsi/csiostor/csio_lnode.c | 2148 ++++++++++++++++++++++++++++++++++++
drivers/scsi/csiostor/csio_rnode.c | 889 +++++++++++++++
2 files changed, 3037 insertions(+), 0 deletions(-)
create mode 100644 drivers/scsi/csiostor/csio_lnode.c
create mode 100644 drivers/scsi/csiostor/csio_rnode.c
diff --git a/drivers/scsi/csiostor/csio_lnode.c b/drivers/scsi/csiostor/csio_lnode.c
new file mode 100644
index 0000000..8f49053
--- /dev/null
+++ b/drivers/scsi/csiostor/csio_lnode.c
@@ -0,0 +1,2148 @@
+/*
+ * This file is part of the Chelsio FCoE driver for Linux.
+ *
+ * Copyright (c) 2008-2012 Chelsio Communications, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/kernel.h>
+#include <linux/delay.h>
+#include <linux/slab.h>
+#include <linux/utsname.h>
+#include <scsi/scsi_transport_fc.h>
+#include <asm/unaligned.h>
+#include <scsi/fc/fc_els.h>
+#include <scsi/fc/fc_fs.h>
+#include <scsi/fc/fc_gs.h>
+#include <scsi/fc/fc_ms.h>
+
+#include "csio_hw.h"
+#include "csio_mb.h"
+#include "csio_lnode.h"
+#include "csio_rnode.h"
+
+int csio_fcoe_rnodes = 512;
+int csio_fdmi_enable = 1;
+
+#define PORT_ID_PTR(_x) ((uint8_t *)(&_x) + 1)
+
+/* Lnode SM declarations */
+static void csio_lns_uninit(struct csio_lnode *, enum csio_ln_ev);
+static void csio_lns_online(struct csio_lnode *, enum csio_ln_ev);
+static void csio_lns_ready(struct csio_lnode *, enum csio_ln_ev);
+static void csio_lns_offline(struct csio_lnode *, enum csio_ln_ev);
+
+static int csio_ln_mgmt_submit_req(struct csio_ioreq *,
+ void (*io_cbfn) (struct csio_hw *, struct csio_ioreq *),
+ enum fcoe_cmn_type, struct csio_dma_buf *, uint32_t);
+
+/* LN event mapping */
+static enum csio_ln_ev fwevt_to_lnevt[] = {
+ CSIO_LNE_NONE, /* None */
+ CSIO_LNE_NONE, /* PLOGI_ACC_RCVD */
+ CSIO_LNE_NONE, /* PLOGI_RJT_RCVD */
+ CSIO_LNE_NONE, /* PLOGI_RCVD */
+ CSIO_LNE_NONE, /* PLOGO_RCVD */
+ CSIO_LNE_NONE, /* PRLI_ACC_RCVD */
+ CSIO_LNE_NONE, /* PRLI_RJT_RCVD */
+ CSIO_LNE_NONE, /* PRLI_RCVD */
+ CSIO_LNE_NONE, /* PRLO_RCVD */
+ CSIO_LNE_NONE, /* NPORT_ID_CHGD */
+ CSIO_LNE_LOGO, /* FLOGO_RCVD */
+ CSIO_LNE_LOGO, /* CLR_VIRT_LNK_RCVD */
+ CSIO_LNE_FAB_INIT_DONE,/* FLOGI_ACC_RCVD */
+ CSIO_LNE_NONE, /* FLOGI_RJT_RCVD */
+ CSIO_LNE_FAB_INIT_DONE,/* FDISC_ACC_RCVD */
+ CSIO_LNE_NONE, /* FDISC_RJT_RCVD */
+ CSIO_LNE_NONE, /* FLOGI_TMO_MAX_RETRY */
+ CSIO_LNE_NONE, /* IMPL_LOGO_ADISC_ACC */
+ CSIO_LNE_NONE, /* IMPL_LOGO_ADISC_RJT */
+ CSIO_LNE_NONE, /* IMPL_LOGO_ADISC_CNFLT */
+ CSIO_LNE_NONE, /* PRLI_TMO */
+ CSIO_LNE_NONE, /* ADISC_TMO */
+ CSIO_LNE_NONE, /* RSCN_DEV_LOST */
+ CSIO_LNE_NONE, /* SCR_ACC_RCVD */
+ CSIO_LNE_NONE, /* ADISC_RJT_RCVD */
+ CSIO_LNE_NONE, /* LOGO_SNT */
+ CSIO_LNE_NONE, /* PROTO_ERR_IMPL_LOGO */
+};
+
+#define CSIO_FWE_TO_LNE(_evt) ((_evt > PROTO_ERR_IMPL_LOGO) ? \
+ CSIO_LNE_NONE : \
+ fwevt_to_lnevt[_evt])
+
+#define csio_ct_rsp(cp) (((struct fc_ct_hdr *)cp)->ct_cmd)
+#define csio_ct_reason(cp) (((struct fc_ct_hdr *)cp)->ct_reason)
+#define csio_ct_expl(cp) (((struct fc_ct_hdr *)cp)->ct_explan)
+#define csio_ct_get_pld(cp) ((void *)(((uint8_t *)cp) + FC_CT_HDR_LEN))
+
+/*
+ * csio_ln_match_by_portid - lookup lnode using given portid.
+ * @hw: HW module
+ * @portid: port-id.
+ *
+ * If found, returns lnode matching given portid otherwise returns NULL.
+ */
+static struct csio_lnode *
+csio_ln_lookup_by_portid(struct csio_hw *hw, uint8_t portid)
+{
+ struct csio_lnode *ln = hw->rln;
+ struct list_head *tmp;
+
+ /* Match siblings lnode with portid */
+ list_for_each(tmp, &hw->sln_head) {
+ ln = (struct csio_lnode *) tmp;
+ if (ln->portid == portid)
+ return ln;
+ }
+
+ return NULL;
+}
+
+/*
+ * csio_ln_lookup_by_vnpi - Lookup lnode using given vnp id.
+ * @hw - HW module
+ * @vnpi - vnp index.
+ * Returns - If found, returns lnode matching given vnp id
+ * otherwise returns NULL.
+ */
+static struct csio_lnode *
+csio_ln_lookup_by_vnpi(struct csio_hw *hw, uint32_t vnp_id)
+{
+ struct list_head *tmp1, *tmp2;
+ struct csio_lnode *sln = NULL, *cln = NULL;
+
+ if (list_empty(&hw->sln_head)) {
+ CSIO_INC_STATS(hw, n_lnlkup_miss);
+ return NULL;
+ }
+ /* Traverse sibling lnodes */
+ list_for_each(tmp1, &hw->sln_head) {
+ sln = (struct csio_lnode *) tmp1;
+
+ /* Match sibling lnode */
+ if (sln->vnp_flowid == vnp_id)
+ return sln;
+
+ if (list_empty(&sln->cln_head))
+ continue;
+
+ /* Traverse children lnodes */
+ list_for_each(tmp2, &sln->cln_head) {
+ cln = (struct csio_lnode *) tmp2;
+
+ if (cln->vnp_flowid == vnp_id)
+ return cln;
+ }
+ }
+ CSIO_INC_STATS(hw, n_lnlkup_miss);
+ return NULL;
+}
+
+/**
+ * csio_lnode_lookup_by_wwpn - Lookup lnode using given wwpn.
+ * @hw: HW module.
+ * @wwpn: WWPN.
+ *
+ * If found, returns lnode matching given wwpn, returns NULL otherwise.
+ */
+struct csio_lnode *
+csio_lnode_lookup_by_wwpn(struct csio_hw *hw, uint8_t *wwpn)
+{
+ struct list_head *tmp1, *tmp2;
+ struct csio_lnode *sln = NULL, *cln = NULL;
+
+ if (list_empty(&hw->sln_head)) {
+ CSIO_INC_STATS(hw, n_lnlkup_miss);
+ return NULL;
+ }
+ /* Traverse sibling lnodes */
+ list_for_each(tmp1, &hw->sln_head) {
+ sln = (struct csio_lnode *) tmp1;
+
+ /* Match sibling lnode */
+ if (!memcmp(csio_ln_wwpn(sln), wwpn, 8))
+ return sln;
+
+ if (list_empty(&sln->cln_head))
+ continue;
+
+ /* Traverse children lnodes */
+ list_for_each(tmp2, &sln->cln_head) {
+ cln = (struct csio_lnode *) tmp2;
+
+ if (!memcmp(csio_ln_wwpn(cln), wwpn, 8))
+ return cln;
+ }
+ }
+ return NULL;
+}
+
+/* FDMI */
+static void
+csio_fill_ct_iu(void *buf, uint8_t type, uint8_t sub_type, uint16_t op)
+{
+ struct fc_ct_hdr *cmd = (struct fc_ct_hdr *)buf;
+ cmd->ct_rev = FC_CT_REV;
+ cmd->ct_fs_type = type;
+ cmd->ct_fs_subtype = sub_type;
+ cmd->ct_cmd = op;
+}
+
+static int
+csio_hostname(uint8_t *buf, size_t buf_len)
+{
+ if (sprintf(buf, "%s", init_utsname()->nodename))
+ return 0;
+ return -1;
+}
+
+static int
+csio_osname(uint8_t *buf, size_t buf_len)
+{
+ uint8_t *ptr = buf;
+
+ strcpy(ptr, init_utsname()->sysname);
+ ptr += strlen(init_utsname()->sysname);
+ *ptr = ' '; /* SPACE */
+ strcpy(ptr, init_utsname()->release);
+ ptr += strlen(init_utsname()->release);
+ *ptr = ' '; /* SPACE */
+ strcpy(ptr, init_utsname()->version);
+ ptr += strlen(init_utsname()->version);
+ *ptr = '\0';
+ return 0;
+}
+
+static inline void
+csio_append_attrib(uint8_t **ptr, uint16_t type, uint8_t *val, uint16_t len)
+{
+ struct fc_fdmi_attr_entry *ae = (struct fc_fdmi_attr_entry *)*ptr;
+ ae->type = htons(type);
+ len += 4; /* includes attribute type and length */
+ len = (len + 3) & ~3; /* should be multiple of 4 bytes */
+ ae->len = htons(len);
+ memset(ae->value, 0, len - 4);
+ memcpy(ae->value, val, len);
+ *ptr += len;
+}
+
+/*
+ * csio_ln_fdmi_done - FDMI registeration completion
+ * @hw: HW context
+ * @fdmi_req: fdmi request
+ */
+static void
+csio_ln_fdmi_done(struct csio_hw *hw, struct csio_ioreq *fdmi_req)
+{
+ void *cmd;
+ struct csio_lnode *ln = fdmi_req->lnode;
+
+ if (fdmi_req->wr_status != FW_SUCCESS) {
+ csio_ln_err(ln, "WR error:%x in processing fdmi rpa cmd\n",
+ fdmi_req->wr_status);
+ CSIO_INC_STATS(ln, n_fdmi_err);
+ }
+
+ cmd = fdmi_req->dma_buf.vaddr;
+ if (ntohs(csio_ct_rsp(cmd)) != FC_FS_ACC) {
+ csio_ln_dbg(ln, "fdmi rpa cmd rejected reason %x expl %x\n",
+ csio_ct_reason(cmd), csio_ct_expl(cmd));
+ }
+}
+
+/*
+ * csio_ln_fdmi_rhba_cbfn - RHBA completion
+ * @hw: HW context
+ * @fdmi_req: fdmi request
+ */
+static void
+csio_ln_fdmi_rhba_cbfn(struct csio_hw *hw, struct csio_ioreq *fdmi_req)
+{
+ void *cmd;
+ uint8_t *pld;
+ uint32_t len = 0;
+ struct csio_lnode *ln = fdmi_req->lnode;
+ struct fs_fdmi_attrs *attrib_blk;
+ struct fc_fdmi_port_name *port_name;
+ uint8_t buf[64];
+ uint32_t val;
+ uint8_t *fc4_type;
+
+ if (fdmi_req->wr_status != FW_SUCCESS) {
+ csio_ln_err(ln, "WR error:%x in processing fdmi rhba cmd\n",
+ fdmi_req->wr_status);
+ CSIO_INC_STATS(ln, n_fdmi_err);
+ }
+
+ cmd = fdmi_req->dma_buf.vaddr;
+ if (ntohs(csio_ct_rsp(cmd)) != FC_FS_ACC) {
+ csio_ln_dbg(ln, "fdmi rhba cmd rejected reason %x expl %x\n",
+ csio_ct_reason(cmd), csio_ct_expl(cmd));
+ }
+
+ if (!csio_is_rnode_ready(fdmi_req->rnode)) {
+ CSIO_INC_STATS(ln, n_fdmi_err);
+ return;
+ }
+
+ /* Prepare CT hdr for RPA cmd */
+ memset(cmd, 0, FC_CT_HDR_LEN);
+ csio_fill_ct_iu(cmd, FC_FST_MGMT, FC_FDMI_SUBTYPE, htons(FC_FDMI_RPA));
+
+ /* Prepare RPA payload */
+ pld = (uint8_t *)csio_ct_get_pld(cmd);
+ port_name = (struct fc_fdmi_port_name *)pld;
+ memcpy(&port_name->portname, csio_ln_wwpn(ln), 8);
+ pld += sizeof(*port_name);
+
+ /* Start appending Port attributes */
+ attrib_blk = (struct fs_fdmi_attrs *)pld;
+ attrib_blk->numattrs = 0;
+ len += sizeof(attrib_blk->numattrs);
+ pld += sizeof(attrib_blk->numattrs);
+
+ fc4_type = &buf[0];
+ memset(fc4_type, 0, FC_FDMI_PORT_ATTR_FC4TYPES_LEN);
+ fc4_type[2] = 1;
+ fc4_type[7] = 1;
+ csio_append_attrib(&pld, FC_FDMI_PORT_ATTR_FC4TYPES,
+ fc4_type, FC_FDMI_PORT_ATTR_FC4TYPES_LEN);
+ attrib_blk->numattrs++;
+ val = htonl(FC_PORTSPEED_1GBIT | FC_PORTSPEED_10GBIT);
+ csio_append_attrib(&pld, FC_FDMI_PORT_ATTR_SUPPORTEDSPEED,
+ (uint8_t *)&val,
+ FC_FDMI_PORT_ATTR_SUPPORTEDSPEED_LEN);
+ attrib_blk->numattrs++;
+
+ if (hw->pport[ln->portid].link_speed == FW_PORT_CAP_SPEED_1G)
+ val = htonl(FC_PORTSPEED_1GBIT);
+ else if (hw->pport[ln->portid].link_speed == FW_PORT_CAP_SPEED_10G)
+ val = htonl(FC_PORTSPEED_10GBIT);
+ else
+ val = htonl(CSIO_HBA_PORTSPEED_UNKNOWN);
+ csio_append_attrib(&pld, FC_FDMI_PORT_ATTR_CURRENTPORTSPEED,
+ (uint8_t *)&val,
+ FC_FDMI_PORT_ATTR_CURRENTPORTSPEED_LEN);
+ attrib_blk->numattrs++;
+
+ val = htonl(ln->ln_sparm.csp.sp_bb_data);
+ csio_append_attrib(&pld, FC_FDMI_PORT_ATTR_MAXFRAMESIZE,
+ (uint8_t *)&val, FC_FDMI_PORT_ATTR_MAXFRAMESIZE_LEN);
+ attrib_blk->numattrs++;
+
+ strcpy(buf, "csiostor");
+ csio_append_attrib(&pld, FC_FDMI_PORT_ATTR_OSDEVICENAME, buf,
+ (uint16_t)strlen(buf));
+ attrib_blk->numattrs++;
+
+ if (!csio_hostname(buf, sizeof(buf))) {
+ csio_append_attrib(&pld, FC_FDMI_PORT_ATTR_HOSTNAME,
+ buf, (uint16_t)strlen(buf));
+ attrib_blk->numattrs++;
+ }
+ attrib_blk->numattrs = ntohl(attrib_blk->numattrs);
+ len = (uint32_t)(pld - (uint8_t *)cmd);
+
+ /* Submit FDMI RPA request */
+ spin_lock_irq(&hw->lock);
+ if (csio_ln_mgmt_submit_req(fdmi_req, csio_ln_fdmi_done,
+ FCOE_CT, &fdmi_req->dma_buf, len)) {
+ CSIO_INC_STATS(ln, n_fdmi_err);
+ csio_ln_err(ln, "Failed to issue fdmi rpa req\n");
+ }
+ spin_unlock_irq(&hw->lock);
+}
+
+/*
+ * csio_ln_fdmi_dprt_cbfn - DPRT completion
+ * @hw: HW context
+ * @fdmi_req: fdmi request
+ */
+static void
+csio_ln_fdmi_dprt_cbfn(struct csio_hw *hw, struct csio_ioreq *fdmi_req)
+{
+ void *cmd;
+ uint8_t *pld;
+ uint32_t len = 0;
+ uint32_t maxpayload = htonl(65536);
+ struct fc_fdmi_hba_identifier *hbaid;
+ struct csio_lnode *ln = fdmi_req->lnode;
+ struct fc_fdmi_rpl *reg_pl;
+ struct fs_fdmi_attrs *attrib_blk;
+ uint8_t buf[64];
+
+ if (fdmi_req->wr_status != FW_SUCCESS) {
+ csio_ln_err(ln, "WR error:%x in processing fdmi dprt cmd\n",
+ fdmi_req->wr_status);
+ CSIO_INC_STATS(ln, n_fdmi_err);
+ }
+
+ if (!csio_is_rnode_ready(fdmi_req->rnode)) {
+ CSIO_INC_STATS(ln, n_fdmi_err);
+ return;
+ }
+ cmd = fdmi_req->dma_buf.vaddr;
+ if (ntohs(csio_ct_rsp(cmd)) != FC_FS_ACC) {
+ csio_ln_dbg(ln, "fdmi dprt cmd rejected reason %x expl %x\n",
+ csio_ct_reason(cmd), csio_ct_expl(cmd));
+ }
+
+ /* Prepare CT hdr for RHBA cmd */
+ memset(cmd, 0, FC_CT_HDR_LEN);
+ csio_fill_ct_iu(cmd, FC_FST_MGMT, FC_FDMI_SUBTYPE, htons(FC_FDMI_RHBA));
+ len = FC_CT_HDR_LEN;
+
+ /* Prepare RHBA payload */
+ pld = (uint8_t *)csio_ct_get_pld(cmd);
+ hbaid = (struct fc_fdmi_hba_identifier *)pld;
+ memcpy(&hbaid->id, csio_ln_wwpn(ln), 8); /* HBA identifer */
+ pld += sizeof(*hbaid);
+
+ /* Register one port per hba */
+ reg_pl = (struct fc_fdmi_rpl *)pld;
+ reg_pl->numport = ntohl(1);
+ memcpy(®_pl->port[0].portname, csio_ln_wwpn(ln), 8);
+ pld += sizeof(*reg_pl);
+
+ /* Start appending HBA attributes hba */
+ attrib_blk = (struct fs_fdmi_attrs *)pld;
+ attrib_blk->numattrs = 0;
+ len += sizeof(attrib_blk->numattrs);
+ pld += sizeof(attrib_blk->numattrs);
+
+ csio_append_attrib(&pld, FC_FDMI_HBA_ATTR_NODENAME, csio_ln_wwnn(ln),
+ FC_FDMI_HBA_ATTR_NODENAME_LEN);
+ attrib_blk->numattrs++;
+
+ memset(buf, 0, sizeof(buf));
+
+ strcpy(buf, "Chelsio Communications");
+ csio_append_attrib(&pld, FC_FDMI_HBA_ATTR_MANUFACTURER, buf,
+ (uint16_t)strlen(buf));
+ attrib_blk->numattrs++;
+ csio_append_attrib(&pld, FC_FDMI_HBA_ATTR_SERIALNUMBER,
+ hw->vpd.sn, (uint16_t)sizeof(hw->vpd.sn));
+ attrib_blk->numattrs++;
+ csio_append_attrib(&pld, FC_FDMI_HBA_ATTR_MODEL, hw->vpd.id,
+ (uint16_t)sizeof(hw->vpd.id));
+ attrib_blk->numattrs++;
+ csio_append_attrib(&pld, FC_FDMI_HBA_ATTR_MODELDESCRIPTION,
+ hw->model_desc, (uint16_t)strlen(hw->model_desc));
+ attrib_blk->numattrs++;
+ csio_append_attrib(&pld, FC_FDMI_HBA_ATTR_HARDWAREVERSION,
+ hw->hw_ver, (uint16_t)sizeof(hw->hw_ver));
+ attrib_blk->numattrs++;
+ csio_append_attrib(&pld, FC_FDMI_HBA_ATTR_FIRMWAREVERSION,
+ hw->fwrev_str, (uint16_t)strlen(hw->fwrev_str));
+ attrib_blk->numattrs++;
+
+ if (!csio_osname(buf, sizeof(buf))) {
+ csio_append_attrib(&pld, FC_FDMI_HBA_ATTR_OSNAMEVERSION,
+ buf, (uint16_t)strlen(buf));
+ attrib_blk->numattrs++;
+ }
+
+ csio_append_attrib(&pld, FC_FDMI_HBA_ATTR_MAXCTPAYLOAD,
+ (uint8_t *)&maxpayload,
+ FC_FDMI_HBA_ATTR_MAXCTPAYLOAD_LEN);
+ len = (uint32_t)(pld - (uint8_t *)cmd);
+ attrib_blk->numattrs++;
+ attrib_blk->numattrs = ntohl(attrib_blk->numattrs);
+
+ /* Submit FDMI RHBA request */
+ spin_lock_irq(&hw->lock);
+ if (csio_ln_mgmt_submit_req(fdmi_req, csio_ln_fdmi_rhba_cbfn,
+ FCOE_CT, &fdmi_req->dma_buf, len)) {
+ CSIO_INC_STATS(ln, n_fdmi_err);
+ csio_ln_err(ln, "Failed to issue fdmi rhba req\n");
+ }
+ spin_unlock_irq(&hw->lock);
+}
+
+/*
+ * csio_ln_fdmi_dhba_cbfn - DHBA completion
+ * @hw: HW context
+ * @fdmi_req: fdmi request
+ */
+static void
+csio_ln_fdmi_dhba_cbfn(struct csio_hw *hw, struct csio_ioreq *fdmi_req)
+{
+ struct csio_lnode *ln = fdmi_req->lnode;
+ void *cmd;
+ struct fc_fdmi_port_name *port_name;
+ uint32_t len;
+
+ if (fdmi_req->wr_status != FW_SUCCESS) {
+ csio_ln_err(ln, "WR error:%x in processing fdmi dhba cmd\n",
+ fdmi_req->wr_status);
+ CSIO_INC_STATS(ln, n_fdmi_err);
+ }
+
+ if (!csio_is_rnode_ready(fdmi_req->rnode)) {
+ CSIO_INC_STATS(ln, n_fdmi_err);
+ return;
+ }
+ cmd = fdmi_req->dma_buf.vaddr;
+ if (ntohs(csio_ct_rsp(cmd)) != FC_FS_ACC) {
+ csio_ln_dbg(ln, "fdmi dhba cmd rejected reason %x expl %x\n",
+ csio_ct_reason(cmd), csio_ct_expl(cmd));
+ }
+
+ /* Send FDMI cmd to de-register any Port attributes if registered
+ * before
+ */
+
+ /* Prepare FDMI DPRT cmd */
+ memset(cmd, 0, FC_CT_HDR_LEN);
+ csio_fill_ct_iu(cmd, FC_FST_MGMT, FC_FDMI_SUBTYPE, htons(FC_FDMI_DPRT));
+ len = FC_CT_HDR_LEN;
+ port_name = (struct fc_fdmi_port_name *)csio_ct_get_pld(cmd);
+ memcpy(&port_name->portname, csio_ln_wwpn(ln), 8);
+ len += sizeof(*port_name);
+
+ /* Submit FDMI request */
+ spin_lock_irq(&hw->lock);
+ if (csio_ln_mgmt_submit_req(fdmi_req, csio_ln_fdmi_dprt_cbfn,
+ FCOE_CT, &fdmi_req->dma_buf, len)) {
+ CSIO_INC_STATS(ln, n_fdmi_err);
+ csio_ln_err(ln, "Failed to issue fdmi dprt req\n");
+ }
+ spin_unlock_irq(&hw->lock);
+}
+
+/**
+ * csio_ln_fdmi_start - Start an FDMI request.
+ * @ln: lnode
+ * @context: session context
+ *
+ * Issued with lock held.
+ */
+int
+csio_ln_fdmi_start(struct csio_lnode *ln, void *context)
+{
+ struct csio_ioreq *fdmi_req;
+ struct csio_rnode *fdmi_rn = (struct csio_rnode *)context;
+ void *cmd;
+ struct fc_fdmi_hba_identifier *hbaid;
+ uint32_t len;
+
+ if (!(ln->flags & CSIO_LNF_FDMI_ENABLE))
+ return -EPROTONOSUPPORT;
+
+ if (!csio_is_rnode_ready(fdmi_rn))
+ CSIO_INC_STATS(ln, n_fdmi_err);
+
+ /* Send FDMI cmd to de-register any HBA attributes if registered
+ * before
+ */
+
+ fdmi_req = ln->mgmt_req;
+ fdmi_req->lnode = ln;
+ fdmi_req->rnode = fdmi_rn;
+
+ /* Prepare FDMI DHBA cmd */
+ cmd = fdmi_req->dma_buf.vaddr;
+ memset(cmd, 0, FC_CT_HDR_LEN);
+ csio_fill_ct_iu(cmd, FC_FST_MGMT, FC_FDMI_SUBTYPE, htons(FC_FDMI_DHBA));
+ len = FC_CT_HDR_LEN;
+
+ hbaid = (struct fc_fdmi_hba_identifier *)csio_ct_get_pld(cmd);
+ memcpy(&hbaid->id, csio_ln_wwpn(ln), 8);
+ len += sizeof(*hbaid);
+
+ /* Submit FDMI request */
+ if (csio_ln_mgmt_submit_req(fdmi_req, csio_ln_fdmi_dhba_cbfn,
+ FCOE_CT, &fdmi_req->dma_buf, len)) {
+ CSIO_INC_STATS(ln, n_fdmi_err);
+ csio_ln_err(ln, "Failed to issue fdmi dhba req\n");
+ }
+
+ return 0;
+}
+
+/*
+ * csio_ln_vnp_read_cbfn - vnp read completion handler.
+ * @hw: HW lnode
+ * @cbfn: Completion handler.
+ *
+ * Reads vnp response and updates ln parameters.
+ */
+static void
+csio_ln_vnp_read_cbfn(struct csio_hw *hw, struct csio_mb *mbp)
+{
+ struct csio_lnode *ln = ((struct csio_lnode *)mbp->priv);
+ struct fw_fcoe_vnp_cmd *rsp = (struct fw_fcoe_vnp_cmd *)(mbp->mb);
+ struct fc_els_csp *csp;
+ struct fc_els_cssp *clsp;
+ enum fw_retval retval;
+
+ spin_lock_irq(&hw->lock);
+
+ retval = FW_CMD_RETVAL_GET(ntohl(rsp->alloc_to_len16));
+ if (retval != FW_SUCCESS) {
+ csio_err(hw, "FCOE VNP read cmd returned error:0x%x\n", retval);
+ spin_unlock_irq(&hw->lock);
+ mempool_free(mbp, hw->mb_mempool);
+ return;
+ }
+
+ memcpy(ln->mac, rsp->vnport_mac, sizeof(ln->mac));
+ memcpy(&ln->nport_id, &rsp->vnport_mac[3],
+ sizeof(uint8_t)*3);
+ ln->nport_id = ntohl(ln->nport_id);
+ ln->nport_id = ln->nport_id>>8;
+
+ /* Update WWNs */
+ /*
+ * This may look like a duplication of what csio_fcoe_enable_link()
+ * does, but is absolutely necessary if the vnpi changes between
+ * a FCOE LINK UP and FCOE LINK DOWN.
+ */
+ memcpy(csio_ln_wwnn(ln), rsp->vnport_wwnn, 8);
+ memcpy(csio_ln_wwpn(ln), rsp->vnport_wwpn, 8);
+
+ /* Copy common sparam */
+ csp = (struct fc_els_csp *)rsp->cmn_srv_parms;
+ ln->ln_sparm.csp.sp_hi_ver = csp->sp_hi_ver;
+ ln->ln_sparm.csp.sp_lo_ver = csp->sp_lo_ver;
+ ln->ln_sparm.csp.sp_bb_cred = ntohs(csp->sp_bb_cred);
+ ln->ln_sparm.csp.sp_features = ntohs(csp->sp_features);
+ ln->ln_sparm.csp.sp_bb_data = ntohs(csp->sp_bb_data);
+ ln->ln_sparm.csp.sp_r_a_tov = ntohl(csp->sp_r_a_tov);
+ ln->ln_sparm.csp.sp_e_d_tov = ntohl(csp->sp_e_d_tov);
+
+ /* Copy word 0 & word 1 of class sparam */
+ clsp = (struct fc_els_cssp *)rsp->clsp_word_0_1;
+ ln->ln_sparm.clsp[2].cp_class = ntohs(clsp->cp_class);
+ ln->ln_sparm.clsp[2].cp_init = ntohs(clsp->cp_init);
+ ln->ln_sparm.clsp[2].cp_recip = ntohs(clsp->cp_recip);
+ ln->ln_sparm.clsp[2].cp_rdfs = ntohs(clsp->cp_rdfs);
+
+ spin_unlock_irq(&hw->lock);
+
+ mempool_free(mbp, hw->mb_mempool);
+
+ /* Send an event to update local attribs */
+ csio_lnode_async_event(ln, CSIO_LN_FC_ATTRIB_UPDATE);
+}
+
+/*
+ * csio_ln_vnp_read - Read vnp params.
+ * @ln: lnode
+ * @cbfn: Completion handler.
+ *
+ * Issued with lock held.
+ */
+static int
+csio_ln_vnp_read(struct csio_lnode *ln,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct csio_hw *hw = ln->hwp;
+ struct csio_mb *mbp;
+
+ /* Allocate Mbox request */
+ mbp = mempool_alloc(hw->mb_mempool, GFP_ATOMIC);
+ if (!mbp) {
+ CSIO_INC_STATS(hw, n_err_nomem);
+ return -ENOMEM;
+ }
+
+ /* Prepare VNP Command */
+ csio_fcoe_vnp_read_init_mb(ln, mbp,
+ CSIO_MB_DEFAULT_TMO,
+ ln->fcf_flowid,
+ ln->vnp_flowid,
+ cbfn);
+
+ /* Issue MBOX cmd */
+ if (csio_mb_issue(hw, mbp)) {
+ csio_err(hw, "Failed to issue mbox FCoE VNP command\n");
+ mempool_free(mbp, hw->mb_mempool);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+/*
+ * csio_fcoe_enable_link - Enable fcoe link.
+ * @ln: lnode
+ * @enable: enable/disable
+ * Issued with lock held.
+ * Issues mbox cmd to bring up FCOE link on port associated with given ln.
+ */
+static int
+csio_fcoe_enable_link(struct csio_lnode *ln, bool enable)
+{
+ struct csio_hw *hw = ln->hwp;
+ struct csio_mb *mbp;
+ enum fw_retval retval;
+ uint8_t portid;
+ uint8_t sub_op;
+ struct fw_fcoe_link_cmd *lcmd;
+ int i;
+
+ mbp = mempool_alloc(hw->mb_mempool, GFP_ATOMIC);
+ if (!mbp) {
+ CSIO_INC_STATS(hw, n_err_nomem);
+ return -ENOMEM;
+ }
+
+ portid = ln->portid;
+ sub_op = enable ? FCOE_LINK_UP : FCOE_LINK_DOWN;
+
+ csio_dbg(hw, "bringing FCOE LINK %s on Port:%d\n",
+ sub_op ? "UP" : "DOWN", portid);
+
+ csio_write_fcoe_link_cond_init_mb(ln, mbp, CSIO_MB_DEFAULT_TMO,
+ portid, sub_op, 0, 0, 0, NULL);
+
+ if (csio_mb_issue(hw, mbp)) {
+ csio_err(hw, "failed to issue FCOE LINK cmd on port[%d]\n",
+ portid);
+ mempool_free(mbp, hw->mb_mempool);
+ return -EINVAL;
+ }
+
+ retval = csio_mb_fw_retval(mbp);
+ if (retval != FW_SUCCESS) {
+ csio_err(hw,
+ "FCOE LINK %s cmd on port[%d] failed with "
+ "ret:x%x\n", sub_op ? "UP" : "DOWN", portid, retval);
+ mempool_free(mbp, hw->mb_mempool);
+ return -EINVAL;
+ }
+
+ if (!enable)
+ goto out;
+
+ lcmd = (struct fw_fcoe_link_cmd *)mbp->mb;
+
+ memcpy(csio_ln_wwnn(ln), lcmd->vnport_wwnn, 8);
+ memcpy(csio_ln_wwpn(ln), lcmd->vnport_wwpn, 8);
+
+ for (i = 0; i < CSIO_MAX_PPORTS; i++)
+ if (hw->pport[i].portid == portid)
+ memcpy(hw->pport[i].mac, lcmd->phy_mac, 6);
+
+out:
+ mempool_free(mbp, hw->mb_mempool);
+ return 0;
+}
+
+/*
+ * csio_ln_read_fcf_cbfn - Read fcf parameters
+ * @ln: lnode
+ *
+ * read fcf response and Update ln fcf information.
+ */
+static void
+csio_ln_read_fcf_cbfn(struct csio_hw *hw, struct csio_mb *mbp)
+{
+ struct csio_lnode *ln = (struct csio_lnode *)mbp->priv;
+ struct csio_fcf_info *fcf_info;
+ struct fw_fcoe_fcf_cmd *rsp =
+ (struct fw_fcoe_fcf_cmd *)(mbp->mb);
+ enum fw_retval retval;
+
+ spin_lock_irq(&hw->lock);
+
+ retval = FW_CMD_RETVAL_GET(ntohl(rsp->retval_len16));
+ if (retval != FW_SUCCESS) {
+ csio_ln_err(ln, "FCOE FCF cmd failed with ret x%x\n",
+ retval);
+ mempool_free(mbp, hw->mb_mempool);
+ spin_unlock_irq(&hw->lock);
+ return;
+ }
+
+ fcf_info = ln->fcfinfo;
+ fcf_info->priority = FW_FCOE_FCF_CMD_PRIORITY_GET(
+ ntohs(rsp->priority_pkd));
+ fcf_info->vf_id = ntohs(rsp->vf_id);
+ fcf_info->vlan_id = rsp->vlan_id;
+ fcf_info->max_fcoe_size = ntohs(rsp->max_fcoe_size);
+ fcf_info->fka_adv = be32_to_cpu(rsp->fka_adv);
+ fcf_info->fcfi = FW_FCOE_FCF_CMD_FCFI_GET(ntohl(rsp->op_to_fcfi));
+ fcf_info->fpma = FW_FCOE_FCF_CMD_FPMA_GET(rsp->fpma_to_portid);
+ fcf_info->spma = FW_FCOE_FCF_CMD_SPMA_GET(rsp->fpma_to_portid);
+ fcf_info->login = FW_FCOE_FCF_CMD_LOGIN_GET(rsp->fpma_to_portid);
+ fcf_info->portid = FW_FCOE_FCF_CMD_PORTID_GET(rsp->fpma_to_portid);
+ memcpy(fcf_info->fc_map, rsp->fc_map, sizeof(fcf_info->fc_map));
+ memcpy(fcf_info->mac, rsp->mac, sizeof(fcf_info->mac));
+ memcpy(fcf_info->name_id, rsp->name_id, sizeof(fcf_info->name_id));
+ memcpy(fcf_info->fabric, rsp->fabric, sizeof(fcf_info->fabric));
+ memcpy(fcf_info->spma_mac, rsp->spma_mac, sizeof(fcf_info->spma_mac));
+
+ spin_unlock_irq(&hw->lock);
+ mempool_free(mbp, hw->mb_mempool);
+
+}
+
+/*
+ * csio_ln_read_fcf_entry - Read fcf entry.
+ * @ln: lnode
+ * @cbfn: Completion handler.
+ *
+ * Issued with lock held.
+ */
+static int
+csio_ln_read_fcf_entry(struct csio_lnode *ln,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct csio_hw *hw = ln->hwp;
+ struct csio_mb *mbp;
+
+ mbp = mempool_alloc(hw->mb_mempool, GFP_ATOMIC);
+ if (!mbp) {
+ CSIO_INC_STATS(hw, n_err_nomem);
+ return -ENOMEM;
+ }
+
+ /* Get FCoE FCF information */
+ csio_fcoe_read_fcf_init_mb(ln, mbp, CSIO_MB_DEFAULT_TMO,
+ ln->portid, ln->fcf_flowid, cbfn);
+
+ if (csio_mb_issue(hw, mbp)) {
+ csio_err(hw, "failed to issue FCOE FCF cmd\n");
+ mempool_free(mbp, hw->mb_mempool);
+ return -EINVAL;
+ }
+
+ if (cbfn == NULL) {
+ spin_unlock_irq(&hw->lock);
+ csio_ln_read_fcf_cbfn(hw, mbp);
+ spin_lock_irq(&hw->lock);
+ }
+
+ return 0;
+}
+
+/*
+ * csio_handle_link_up - Logical Linkup event.
+ * @hw - HW module.
+ * @portid - Physical port number
+ * @fcfi - FCF index.
+ * @vnpi - VNP index.
+ * Returns - none.
+ *
+ * This event is received from FW, when virtual link is established between
+ * Physical port[ENode] and FCF. If its new vnpi, then local node object is
+ * created on this FCF and set to [ONLINE] state.
+ * Lnode waits for FW_RDEV_CMD event to be received indicating that
+ * Fabric login is completed and lnode moves to [READY] state.
+ *
+ * This called with hw lock held
+ */
+static void
+csio_handle_link_up(struct csio_hw *hw, uint8_t portid, uint32_t fcfi,
+ uint32_t vnpi)
+{
+ struct csio_lnode *ln = NULL;
+
+ /* Lookup lnode based on vnpi */
+ ln = csio_ln_lookup_by_vnpi(hw, vnpi);
+ if (!ln) {
+ /* Pick lnode based on portid */
+ ln = csio_ln_lookup_by_portid(hw, portid);
+ if (!ln) {
+ csio_err(hw, "failed to lookup fcoe lnode on port:%d\n",
+ portid);
+ CSIO_DB_ASSERT(0);
+ return;
+ }
+
+ /* Check if lnode has valid vnp flowid */
+ if (ln->vnp_flowid != CSIO_INVALID_IDX) {
+ /* New VN-Port */
+ spin_unlock_irq(&hw->lock);
+ csio_lnode_alloc(hw);
+ spin_lock_irq(&hw->lock);
+ if (!ln) {
+ csio_err(hw,
+ "failed to allocate fcoe lnode"
+ "for port:%d vnpi:x%x\n",
+ portid, vnpi);
+ CSIO_DB_ASSERT(0);
+ return;
+ }
+ ln->portid = portid;
+ }
+ ln->vnp_flowid = vnpi;
+ ln->dev_num &= ~0xFFFF;
+ ln->dev_num |= vnpi;
+ }
+
+ /*Initialize fcfi */
+ ln->fcf_flowid = fcfi;
+
+ csio_info(hw, "Port:%d - FCOE LINK UP\n", portid);
+
+ CSIO_INC_STATS(ln, n_link_up);
+
+ /* Send LINKUP event to SM */
+ csio_post_event(&ln->sm, CSIO_LNE_LINKUP);
+}
+
+/*
+ * csio_post_event_rns
+ * @ln - FCOE lnode
+ * @evt - Given rnode event
+ * Returns - none
+ *
+ * Posts given rnode event to all FCOE rnodes connected with given Lnode.
+ * This routine is invoked when lnode receives LINK_DOWN/DOWN_LINK/CLOSE
+ * event.
+ *
+ * This called with hw lock held
+ */
+static void
+csio_post_event_rns(struct csio_lnode *ln, enum csio_rn_ev evt)
+{
+ struct csio_rnode *rnhead = (struct csio_rnode *) &ln->rnhead;
+ struct list_head *tmp, *next;
+ struct csio_rnode *rn;
+
+ list_for_each_safe(tmp, next, &rnhead->sm.sm_list) {
+ rn = (struct csio_rnode *) tmp;
+ csio_post_event(&rn->sm, evt);
+ }
+}
+
+/*
+ * csio_cleanup_rns
+ * @ln - FCOE lnode
+ * Returns - none
+ *
+ * Frees all FCOE rnodes connected with given Lnode.
+ *
+ * This called with hw lock held
+ */
+static void
+csio_cleanup_rns(struct csio_lnode *ln)
+{
+ struct csio_rnode *rnhead = (struct csio_rnode *) &ln->rnhead;
+ struct list_head *tmp, *next_rn;
+ struct csio_rnode *rn;
+
+ list_for_each_safe(tmp, next_rn, &rnhead->sm.sm_list) {
+ rn = (struct csio_rnode *) tmp;
+ csio_put_rnode(ln, rn);
+ }
+
+}
+
+/*
+ * csio_post_event_lns
+ * @ln - FCOE lnode
+ * @evt - Given lnode event
+ * Returns - none
+ *
+ * Posts given lnode event to all FCOE lnodes connected with given Lnode.
+ * This routine is invoked when lnode receives LINK_DOWN/DOWN_LINK/CLOSE
+ * event.
+ *
+ * This called with hw lock held
+ */
+static void
+csio_post_event_lns(struct csio_lnode *ln, enum csio_ln_ev evt)
+{
+ struct list_head *tmp;
+ struct csio_lnode *cln, *sln;
+
+ /* If NPIV lnode, send evt only to that and return */
+ if (csio_is_npiv_ln(ln)) {
+ csio_post_event(&ln->sm, evt);
+ return;
+ }
+
+ sln = ln;
+ /* Traverse children lnodes list and send evt */
+ list_for_each(tmp, &sln->cln_head) {
+ cln = (struct csio_lnode *) tmp;
+ csio_post_event(&cln->sm, evt);
+ }
+
+ /* Send evt to parent lnode */
+ csio_post_event(&ln->sm, evt);
+}
+
+/*
+ * csio_ln_down - Lcoal nport is down
+ * @ln - FCOE Lnode
+ * Returns - none
+ *
+ * Sends LINK_DOWN events to Lnode and its associated NPIVs lnodes.
+ *
+ * This called with hw lock held
+ */
+static void
+csio_ln_down(struct csio_lnode *ln)
+{
+ csio_post_event_lns(ln, CSIO_LNE_LINK_DOWN);
+}
+
+/*
+ * csio_handle_link_down - Logical Linkdown event.
+ * @hw - HW module.
+ * @portid - Physical port number
+ * @fcfi - FCF index.
+ * @vnpi - VNP index.
+ * Returns - none
+ *
+ * This event is received from FW, when virtual link goes down between
+ * Physical port[ENode] and FCF. Lnode and its associated NPIVs lnode hosted on
+ * this vnpi[VN-Port] will be de-instantiated.
+ *
+ * This called with hw lock held
+ */
+static void
+csio_handle_link_down(struct csio_hw *hw, uint8_t portid, uint32_t fcfi,
+ uint32_t vnpi)
+{
+ struct csio_fcf_info *fp;
+ struct csio_lnode *ln;
+
+ /* Lookup lnode based on vnpi */
+ ln = csio_ln_lookup_by_vnpi(hw, vnpi);
+ if (ln) {
+ fp = ln->fcfinfo;
+ CSIO_INC_STATS(ln, n_link_down);
+
+ /*Warn if linkdown received if lnode is not in ready state */
+ if (!csio_is_lnode_ready(ln)) {
+ csio_ln_warn(ln,
+ "warn: FCOE link is already in offline "
+ "Ignoring Fcoe linkdown event on portid %d\n",
+ portid);
+ CSIO_INC_STATS(ln, n_evt_drop);
+ return;
+ }
+
+ /* Verify portid */
+ if (fp->portid != portid) {
+ csio_ln_warn(ln,
+ "warn: FCOE linkdown recv with "
+ "invalid port %d\n", portid);
+ CSIO_INC_STATS(ln, n_evt_drop);
+ return;
+ }
+
+ /* verify fcfi */
+ if (ln->fcf_flowid != fcfi) {
+ csio_ln_warn(ln,
+ "warn: FCOE linkdown recv with "
+ "invalid fcfi x%x\n", fcfi);
+ CSIO_INC_STATS(ln, n_evt_drop);
+ return;
+ }
+
+ csio_info(hw, "Port:%d - FCOE LINK DOWN\n", portid);
+
+ /* Send LINK_DOWN event to lnode s/m */
+ csio_ln_down(ln);
+
+ return;
+ } else {
+ csio_warn(hw,
+ "warn: FCOE linkdown recv with invalid vnpi x%x\n",
+ vnpi);
+ CSIO_INC_STATS(hw, n_evt_drop);
+ }
+}
+
+/*
+ * csio_is_lnode_ready - Checks FCOE lnode is in ready state.
+ * @ln: Lnode module
+ *
+ * Returns True if FCOE lnode is in ready state.
+ */
+int
+csio_is_lnode_ready(struct csio_lnode *ln)
+{
+ return (csio_get_state(ln) == ((csio_sm_state_t)csio_lns_ready));
+}
+
+/*****************************************************************************/
+/* START: Lnode SM */
+/*****************************************************************************/
+/*
+ * csio_lns_uninit - The request in uninit state.
+ * @ln - FCOE lnode.
+ * @evt - Event to be processed.
+ *
+ * Process the given lnode event which is currently in "uninit" state.
+ * Invoked with HW lock held.
+ * Return - none.
+ */
+static void
+csio_lns_uninit(struct csio_lnode *ln, enum csio_ln_ev evt)
+{
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+ struct csio_lnode *rln = hw->rln;
+ int rv;
+
+ CSIO_INC_STATS(ln, n_evt_sm[evt]);
+ switch (evt) {
+ case CSIO_LNE_LINKUP:
+ csio_set_state(&ln->sm, csio_lns_online);
+ /* Read FCF only for physical lnode */
+ if (csio_is_phys_ln(ln)) {
+ rv = csio_ln_read_fcf_entry(ln,
+ csio_ln_read_fcf_cbfn);
+ if (rv != 0) {
+ /* TODO: Send HW RESET event */
+ CSIO_INC_STATS(ln, n_err);
+ break;
+ }
+
+ /* Add FCF record */
+ list_add_tail(&ln->fcfinfo->list, &rln->fcf_lsthead);
+ }
+
+ rv = csio_ln_vnp_read(ln, csio_ln_vnp_read_cbfn);
+ if (rv != 0) {
+ /* TODO: Send HW RESET event */
+ CSIO_INC_STATS(ln, n_err);
+ }
+ break;
+
+ case CSIO_LNE_DOWN_LINK:
+ break;
+
+ default:
+ csio_ln_dbg(ln,
+ "unexp ln event %d recv from did:x%x in "
+ "ln state[uninit].\n", evt, ln->nport_id);
+ CSIO_INC_STATS(ln, n_evt_unexp);
+ break;
+ } /* switch event */
+}
+
+/*
+ * csio_lns_online - The request in online state.
+ * @ln - FCOE lnode.
+ * @evt - Event to be processed.
+ *
+ * Process the given lnode event which is currently in "online" state.
+ * Invoked with HW lock held.
+ * Return - none.
+ */
+static void
+csio_lns_online(struct csio_lnode *ln, enum csio_ln_ev evt)
+{
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+
+ CSIO_INC_STATS(ln, n_evt_sm[evt]);
+ switch (evt) {
+ case CSIO_LNE_LINKUP:
+ csio_ln_warn(ln,
+ "warn: FCOE link is up already "
+ "Ignoring linkup on port:%d\n", ln->portid);
+ CSIO_INC_STATS(ln, n_evt_drop);
+ break;
+
+ case CSIO_LNE_FAB_INIT_DONE:
+ csio_set_state(&ln->sm, csio_lns_ready);
+
+ spin_unlock_irq(&hw->lock);
+ csio_lnode_async_event(ln, CSIO_LN_FC_LINKUP);
+ spin_lock_irq(&hw->lock);
+
+ break;
+
+ case CSIO_LNE_LINK_DOWN:
+ /* Fall through */
+ case CSIO_LNE_DOWN_LINK:
+ csio_set_state(&ln->sm, csio_lns_uninit);
+ if (csio_is_phys_ln(ln)) {
+ /* Remove FCF entry */
+ list_del_init(&ln->fcfinfo->list);
+ }
+ break;
+
+ default:
+ csio_ln_dbg(ln,
+ "unexp ln event %d recv from did:x%x in "
+ "ln state[uninit].\n", evt, ln->nport_id);
+ CSIO_INC_STATS(ln, n_evt_unexp);
+
+ break;
+ } /* switch event */
+}
+
+/*
+ * csio_lns_ready - The request in ready state.
+ * @ln - FCOE lnode.
+ * @evt - Event to be processed.
+ *
+ * Process the given lnode event which is currently in "ready" state.
+ * Invoked with HW lock held.
+ * Return - none.
+ */
+static void
+csio_lns_ready(struct csio_lnode *ln, enum csio_ln_ev evt)
+{
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+
+ CSIO_INC_STATS(ln, n_evt_sm[evt]);
+ switch (evt) {
+ case CSIO_LNE_FAB_INIT_DONE:
+ csio_ln_err(ln,
+ "ignoring event %d recv from did x%x"
+ "in ln state[ready].\n", evt, ln->nport_id);
+ CSIO_INC_STATS(ln, n_evt_drop);
+ break;
+
+ case CSIO_LNE_LINK_DOWN:
+ csio_set_state(&ln->sm, csio_lns_offline);
+ csio_post_event_rns(ln, CSIO_RNFE_DOWN);
+
+ spin_unlock_irq(&hw->lock);
+ csio_lnode_async_event(ln, CSIO_LN_FC_LINKDOWN);
+ spin_lock_irq(&hw->lock);
+
+ if (csio_is_phys_ln(ln)) {
+ /* Remove FCF entry */
+ list_del_init(&ln->fcfinfo->list);
+ }
+ break;
+
+ case CSIO_LNE_DOWN_LINK:
+ csio_set_state(&ln->sm, csio_lns_offline);
+ csio_post_event_rns(ln, CSIO_RNFE_DOWN);
+
+ /* Host need to issue aborts in case if FW has not returned
+ * WRs with status "ABORTED"
+ */
+ spin_unlock_irq(&hw->lock);
+ csio_lnode_async_event(ln, CSIO_LN_FC_LINKDOWN);
+ spin_lock_irq(&hw->lock);
+
+ if (csio_is_phys_ln(ln)) {
+ /* Remove FCF entry */
+ list_del_init(&ln->fcfinfo->list);
+ }
+ break;
+
+ case CSIO_LNE_CLOSE:
+ csio_set_state(&ln->sm, csio_lns_uninit);
+ csio_post_event_rns(ln, CSIO_RNFE_CLOSE);
+ break;
+
+ case CSIO_LNE_LOGO:
+ csio_set_state(&ln->sm, csio_lns_offline);
+ csio_post_event_rns(ln, CSIO_RNFE_DOWN);
+ break;
+
+ default:
+ csio_ln_dbg(ln,
+ "unexp ln event %d recv from did:x%x in "
+ "ln state[uninit].\n", evt, ln->nport_id);
+ CSIO_INC_STATS(ln, n_evt_unexp);
+ CSIO_DB_ASSERT(0);
+ break;
+ } /* switch event */
+}
+
+/*
+ * csio_lns_offline - The request in offline state.
+ * @ln - FCOE lnode.
+ * @evt - Event to be processed.
+ *
+ * Process the given lnode event which is currently in "offline" state.
+ * Invoked with HW lock held.
+ * Return - none.
+ */
+static void
+csio_lns_offline(struct csio_lnode *ln, enum csio_ln_ev evt)
+{
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+ struct csio_lnode *rln = hw->rln;
+ int rv;
+
+ CSIO_INC_STATS(ln, n_evt_sm[evt]);
+ switch (evt) {
+ case CSIO_LNE_LINKUP:
+ csio_set_state(&ln->sm, csio_lns_online);
+ /* Read FCF only for physical lnode */
+ if (csio_is_phys_ln(ln)) {
+ rv = csio_ln_read_fcf_entry(ln,
+ csio_ln_read_fcf_cbfn);
+ if (rv != 0) {
+ /* TODO: Send HW RESET event */
+ CSIO_INC_STATS(ln, n_err);
+ break;
+ }
+
+ /* Add FCF record */
+ list_add_tail(&ln->fcfinfo->list, &rln->fcf_lsthead);
+ }
+
+ rv = csio_ln_vnp_read(ln, csio_ln_vnp_read_cbfn);
+ if (rv != 0) {
+ /* TODO: Send HW RESET event */
+ CSIO_INC_STATS(ln, n_err);
+ }
+ break;
+
+ case CSIO_LNE_LINK_DOWN:
+ case CSIO_LNE_DOWN_LINK:
+ case CSIO_LNE_LOGO:
+ csio_ln_err(ln,
+ "ignoring event %d recv from did x%x"
+ "in ln state[offline].\n", evt, ln->nport_id);
+ CSIO_INC_STATS(ln, n_evt_drop);
+ break;
+
+ case CSIO_LNE_CLOSE:
+ csio_set_state(&ln->sm, csio_lns_uninit);
+ csio_post_event_rns(ln, CSIO_RNFE_CLOSE);
+ break;
+
+ default:
+ csio_ln_dbg(ln,
+ "unexp ln event %d recv from did:x%x in "
+ "ln state[offline]\n", evt, ln->nport_id);
+ CSIO_INC_STATS(ln, n_evt_unexp);
+ CSIO_DB_ASSERT(0);
+ break;
+ } /* switch event */
+}
+
+/*****************************************************************************/
+/* END: Lnode SM */
+/*****************************************************************************/
+
+static void
+csio_free_fcfinfo(struct kref *kref)
+{
+ struct csio_fcf_info *fcfinfo = container_of(kref,
+ struct csio_fcf_info, kref);
+ kfree(fcfinfo);
+}
+
+/* Helper routines for attributes */
+/*
+ * csio_lnode_state_to_str - Get current state of FCOE lnode.
+ * @ln - lnode
+ * @str - state of lnode.
+ *
+ */
+void
+csio_lnode_state_to_str(struct csio_lnode *ln, int8_t *str)
+{
+ if (csio_get_state(ln) == ((csio_sm_state_t)csio_lns_uninit)) {
+ strcpy(str, "UNINIT");
+ return;
+ }
+ if (csio_get_state(ln) == ((csio_sm_state_t)csio_lns_ready)) {
+ strcpy(str, "READY");
+ return;
+ }
+ if (csio_get_state(ln) == ((csio_sm_state_t)csio_lns_offline)) {
+ strcpy(str, "OFFLINE");
+ return;
+ }
+ strcpy(str, "UNKNOWN");
+} /* csio_lnode_state_to_str */
+
+
+int
+csio_get_phy_port_stats(struct csio_hw *hw, uint8_t portid,
+ struct fw_fcoe_port_stats *port_stats)
+{
+ struct csio_mb *mbp;
+ struct fw_fcoe_port_cmd_params portparams;
+ enum fw_retval retval;
+ int idx;
+
+ mbp = mempool_alloc(hw->mb_mempool, GFP_ATOMIC);
+ if (!mbp) {
+ csio_err(hw, "FCoE FCF PARAMS command out of memory!\n");
+ return -EINVAL;
+ }
+ portparams.portid = portid;
+
+ for (idx = 1; idx <= 3; idx++) {
+ portparams.idx = (idx-1)*6 + 1;
+ portparams.nstats = 6;
+ if (idx == 3)
+ portparams.nstats = 4;
+ csio_fcoe_read_portparams_init_mb(hw, mbp, CSIO_MB_DEFAULT_TMO,
+ &portparams, NULL);
+ if (csio_mb_issue(hw, mbp)) {
+ csio_err(hw, "Issue of FCoE port params failed!\n");
+ mempool_free(mbp, hw->mb_mempool);
+ return -EINVAL;
+ }
+ csio_mb_process_portparams_rsp(hw, mbp, &retval,
+ &portparams, port_stats);
+ }
+
+ mempool_free(mbp, hw->mb_mempool);
+ return 0;
+}
+
+/*
+ * csio_ln_mgmt_wr_handler -Mgmt Work Request handler.
+ * @wr - WR.
+ * @len - WR len.
+ * This handler is invoked when an outstanding mgmt WR is completed.
+ * Its invoked in the context of FW event worker thread for every
+ * mgmt event received.
+ * Return - none.
+ */
+
+static void
+csio_ln_mgmt_wr_handler(struct csio_hw *hw, void *wr, uint32_t len)
+{
+ struct csio_mgmtm *mgmtm = csio_hw_to_mgmtm(hw);
+ struct csio_ioreq *io_req = NULL;
+ struct fw_fcoe_els_ct_wr *wr_cmd;
+
+
+ wr_cmd = (struct fw_fcoe_els_ct_wr *) wr;
+
+ if (len < sizeof(struct fw_fcoe_els_ct_wr)) {
+ csio_err(mgmtm->hw,
+ "Invalid ELS CT WR length recvd, len:%x\n", len);
+ mgmtm->stats.n_err++;
+ return;
+ }
+
+ io_req = (struct csio_ioreq *) ((uintptr_t) wr_cmd->cookie);
+ io_req->wr_status = csio_wr_status(wr_cmd);
+
+ /* lookup ioreq exists in our active Q */
+ spin_lock_irq(&hw->lock);
+ if (csio_mgmt_req_lookup(mgmtm, io_req) != 0) {
+ csio_err(mgmtm->hw,
+ "Error- Invalid IO handle recv in WR. handle: %p\n",
+ io_req);
+ mgmtm->stats.n_err++;
+ spin_unlock_irq(&hw->lock);
+ return;
+ }
+
+ mgmtm = csio_hw_to_mgmtm(hw);
+
+ /* Dequeue from active queue */
+ list_del_init(&io_req->sm.sm_list);
+ mgmtm->stats.n_active--;
+ spin_unlock_irq(&hw->lock);
+
+ /* io_req will be freed by completion handler */
+ if (io_req->io_cbfn)
+ io_req->io_cbfn(hw, io_req);
+}
+
+/**
+ * csio_fcoe_fwevt_handler - Event handler for Firmware FCoE events.
+ * @hw: HW module
+ * @cpl_op: CPL opcode
+ * @cmd: FW cmd/WR.
+ *
+ * Process received FCoE cmd/WR event from FW.
+ */
+void
+csio_fcoe_fwevt_handler(struct csio_hw *hw, __u8 cpl_op, __be64 *cmd)
+{
+ struct csio_lnode *ln;
+ struct csio_rnode *rn;
+ uint8_t portid, opcode = *(uint8_t *)cmd;
+ struct fw_fcoe_link_cmd *lcmd;
+ struct fw_wr_hdr *wr;
+ struct fw_rdev_wr *rdev_wr;
+ enum fw_fcoe_link_status lstatus;
+ uint32_t fcfi, rdev_flowid, vnpi;
+ enum csio_ln_ev evt;
+
+ if (cpl_op == CPL_FW6_MSG && opcode == FW_FCOE_LINK_CMD) {
+
+ lcmd = (struct fw_fcoe_link_cmd *)cmd;
+ lstatus = lcmd->lstatus;
+ portid = FW_FCOE_LINK_CMD_PORTID_GET(
+ ntohl(lcmd->op_to_portid));
+ fcfi = FW_FCOE_LINK_CMD_FCFI_GET(ntohl(lcmd->sub_opcode_fcfi));
+ vnpi = FW_FCOE_LINK_CMD_VNPI_GET(ntohl(lcmd->vnpi_pkd));
+
+ if (lstatus == FCOE_LINKUP) {
+
+ /* HW lock here */
+ spin_lock_irq(&hw->lock);
+ csio_handle_link_up(hw, portid, fcfi, vnpi);
+ spin_unlock_irq(&hw->lock);
+ /* HW un lock here */
+
+ } else if (lstatus == FCOE_LINKDOWN) {
+
+ /* HW lock here */
+ spin_lock_irq(&hw->lock);
+ csio_handle_link_down(hw, portid, fcfi, vnpi);
+ spin_unlock_irq(&hw->lock);
+ /* HW un lock here */
+ } else {
+ csio_warn(hw, "Unexpected FCOE LINK status:0x%x\n",
+ ntohl(lcmd->lstatus));
+ CSIO_INC_STATS(hw, n_cpl_unexp);
+ }
+ } else if (cpl_op == CPL_FW6_PLD) {
+ wr = (struct fw_wr_hdr *) (cmd + 4);
+ if (FW_WR_OP_GET(be32_to_cpu(wr->hi))
+ == FW_RDEV_WR) {
+
+ rdev_wr = (struct fw_rdev_wr *) (cmd + 4);
+
+ rdev_flowid = FW_RDEV_WR_FLOWID_GET(
+ ntohl(rdev_wr->alloc_to_len16));
+ vnpi = FW_RDEV_WR_ASSOC_FLOWID_GET(
+ ntohl(rdev_wr->flags_to_assoc_flowid));
+
+ csio_dbg(hw,
+ "FW_RDEV_WR: flowid:x%x ev_cause:x%x "
+ "vnpi:0x%x\n", rdev_flowid,
+ rdev_wr->event_cause, vnpi);
+
+ if (rdev_wr->protocol != PROT_FCOE) {
+ csio_err(hw,
+ "FW_RDEV_WR: invalid proto:x%x "
+ "received with flowid:x%x\n",
+ rdev_wr->protocol,
+ rdev_flowid);
+ CSIO_INC_STATS(hw, n_evt_drop);
+ return;
+ }
+
+ /* HW lock here */
+ spin_lock_irq(&hw->lock);
+ ln = csio_ln_lookup_by_vnpi(hw, vnpi);
+ if (!ln) {
+ csio_err(hw,
+ "FW_DEV_WR: invalid vnpi:x%x received "
+ "with flowid:x%x\n", vnpi, rdev_flowid);
+ CSIO_INC_STATS(hw, n_evt_drop);
+ spin_unlock_irq(&hw->lock);
+ return;
+ }
+
+ rn = csio_confirm_rnode(ln, rdev_flowid,
+ &rdev_wr->u.fcoe_rdev);
+ if (!rn) {
+ csio_ln_dbg(ln,
+ "Failed to confirm rnode "
+ "for flowid:x%x\n", rdev_flowid);
+ CSIO_INC_STATS(hw, n_evt_drop);
+ spin_unlock_irq(&hw->lock);
+ return;
+ }
+
+ /* save previous event for debugging */
+ ln->prev_evt = ln->cur_evt;
+ ln->cur_evt = rdev_wr->event_cause;
+ CSIO_INC_STATS(ln, n_evt_fw[rdev_wr->event_cause]);
+
+ /* Translate all the fabric events to lnode SM events */
+ evt = CSIO_FWE_TO_LNE(rdev_wr->event_cause);
+ if (evt) {
+ csio_ln_dbg(ln,
+ "Posting event to lnode event:%d "
+ "cause:%d flowid:x%x\n", evt,
+ rdev_wr->event_cause, rdev_flowid);
+ csio_post_event(&ln->sm, evt);
+ }
+
+ /* Handover event to rn SM here. */
+ csio_rnode_fwevt_handler(rn, rdev_wr->event_cause);
+
+ spin_unlock_irq(&hw->lock);
+ } else {
+ csio_warn(hw, "unexpected WR op(0x%x) recv\n",
+ FW_WR_OP_GET(be32_to_cpu((wr->hi))));
+ CSIO_INC_STATS(hw, n_cpl_unexp);
+ }
+ } else if (cpl_op == CPL_FW6_MSG) {
+ wr = (struct fw_wr_hdr *) (cmd);
+ if (FW_WR_OP_GET(be32_to_cpu(wr->hi)) == FW_FCOE_ELS_CT_WR) {
+ csio_ln_mgmt_wr_handler(hw, wr,
+ sizeof(struct fw_fcoe_els_ct_wr));
+ } else {
+ csio_warn(hw, "unexpected WR op(0x%x) recv\n",
+ FW_WR_OP_GET(be32_to_cpu((wr->hi))));
+ CSIO_INC_STATS(hw, n_cpl_unexp);
+ }
+ } else {
+ csio_warn(hw, "unexpected CPL op(0x%x) recv\n", opcode);
+ CSIO_INC_STATS(hw, n_cpl_unexp);
+ }
+}
+
+/**
+ * csio_lnode_start - Kickstart lnode discovery.
+ * @ln: lnode
+ *
+ * This routine kickstarts the discovery by issuing an FCOE_LINK (up) command.
+ */
+int
+csio_lnode_start(struct csio_lnode *ln)
+{
+ int rv = 0;
+ if (csio_is_phys_ln(ln) && !(ln->flags & CSIO_LNF_LINK_ENABLE)) {
+ rv = csio_fcoe_enable_link(ln, 1);
+ ln->flags |= CSIO_LNF_LINK_ENABLE;
+ }
+
+ return rv;
+}
+
+/**
+ * csio_lnode_stop - Stop the lnode.
+ * @ln: lnode
+ *
+ * This routine is invoked by HW module to stop lnode and its associated NPIV
+ * lnodes.
+ */
+void
+csio_lnode_stop(struct csio_lnode *ln)
+{
+ csio_post_event_lns(ln, CSIO_LNE_DOWN_LINK);
+ if (csio_is_phys_ln(ln) && (ln->flags & CSIO_LNF_LINK_ENABLE)) {
+ csio_fcoe_enable_link(ln, 0);
+ ln->flags &= ~CSIO_LNF_LINK_ENABLE;
+ }
+ csio_ln_dbg(ln, "stopping ln :%p\n", ln);
+}
+
+/**
+ * csio_lnode_close - Close an lnode.
+ * @ln: lnode
+ *
+ * This routine is invoked by HW module to close an lnode and its
+ * associated NPIV lnodes. Lnode and its associated NPIV lnodes are
+ * set to uninitialized state.
+ */
+void
+csio_lnode_close(struct csio_lnode *ln)
+{
+ csio_post_event_lns(ln, CSIO_LNE_CLOSE);
+ if (csio_is_phys_ln(ln))
+ ln->vnp_flowid = CSIO_INVALID_IDX;
+
+ csio_ln_dbg(ln, "closed ln :%p\n", ln);
+}
+
+/*
+ * csio_ln_prep_ecwr - Prepare ELS/CT WR.
+ * @io_req - IO request.
+ * @wr_len - WR len
+ * @immd_len - WR immediate data
+ * @sub_op - Sub opcode
+ * @sid - source portid.
+ * @did - destination portid
+ * @flow_id - flowid
+ * @fw_wr - ELS/CT WR to be prepared.
+ * Returns: 0 - on success
+ */
+static int
+csio_ln_prep_ecwr(struct csio_ioreq *io_req, uint32_t wr_len,
+ uint32_t immd_len, uint8_t sub_op, uint32_t sid,
+ uint32_t did, uint32_t flow_id, uint8_t *fw_wr)
+{
+ struct fw_fcoe_els_ct_wr *wr;
+ uint32_t port_id;
+
+ wr = (struct fw_fcoe_els_ct_wr *)fw_wr;
+ wr->op_immdlen = cpu_to_be32(FW_WR_OP(FW_FCOE_ELS_CT_WR) |
+ FW_FCOE_ELS_CT_WR_IMMDLEN(immd_len));
+
+ wr_len = CSIO_ROUNDUP(wr_len, 16);
+ wr->flowid_len16 = cpu_to_be32(FW_WR_FLOWID(flow_id) |
+ FW_WR_LEN16(wr_len));
+ wr->els_ct_type = sub_op;
+ wr->ctl_pri = 0;
+ wr->cp_en_class = 0;
+ wr->cookie = io_req->fw_handle;
+ wr->iqid = (uint16_t)cpu_to_be16(csio_q_physiqid(
+ io_req->lnode->hwp, io_req->iq_idx));
+ wr->fl_to_sp = FW_FCOE_ELS_CT_WR_SP(1);
+ wr->tmo_val = (uint8_t) io_req->tmo;
+ port_id = htonl(sid);
+ memcpy(wr->l_id, PORT_ID_PTR(port_id), 3);
+ port_id = htonl(did);
+ memcpy(wr->r_id, PORT_ID_PTR(port_id), 3);
+
+ /* Prepare RSP SGL */
+ wr->rsp_dmalen = cpu_to_be32(io_req->dma_buf.len);
+ wr->rsp_dmaaddr = cpu_to_be64(io_req->dma_buf.paddr);
+ return 0;
+}
+
+/*
+ * csio_ln_mgmt_submit_wr - Post elsct work request.
+ * @mgmtm - mgmtm
+ * @io_req - io request.
+ * @sub_op - ELS or CT request type
+ * @pld - Dma Payload buffer
+ * @pld_len - Payload len
+ * Prepares ELSCT Work request and sents it to FW.
+ * Returns: 0 - on success
+ */
+static int
+csio_ln_mgmt_submit_wr(struct csio_mgmtm *mgmtm, struct csio_ioreq *io_req,
+ uint8_t sub_op, struct csio_dma_buf *pld,
+ uint32_t pld_len)
+{
+ struct csio_wr_pair wrp;
+ struct csio_lnode *ln = io_req->lnode;
+ struct csio_rnode *rn = io_req->rnode;
+ struct csio_hw *hw = mgmtm->hw;
+ uint8_t fw_wr[64];
+ struct ulptx_sgl dsgl;
+ uint32_t wr_size = 0;
+ uint8_t im_len = 0;
+ uint32_t wr_off = 0;
+
+ int ret = 0;
+
+ /* Calculate WR Size for this ELS REQ */
+ wr_size = sizeof(struct fw_fcoe_els_ct_wr);
+
+ /* Send as immediate data if pld < 256 */
+ if (pld_len < 256) {
+ wr_size += ALIGN(pld_len, 8);
+ im_len = (uint8_t)pld_len;
+ } else
+ wr_size += sizeof(struct ulptx_sgl);
+
+ /* Roundup WR size in units of 16 bytes */
+ wr_size = ALIGN(wr_size, 16);
+
+ /* Get WR to send ELS REQ */
+ ret = csio_wr_get(hw, mgmtm->eq_idx, wr_size, &wrp);
+ if (ret != 0) {
+ csio_err(hw, "Failed to get WR for ec_req %p ret:%d\n",
+ io_req, ret);
+ return ret;
+ }
+
+ /* Prepare Generic WR used by all ELS/CT cmd */
+ csio_ln_prep_ecwr(io_req, wr_size, im_len, sub_op,
+ ln->nport_id, rn->nport_id,
+ csio_rn_flowid(rn),
+ &fw_wr[0]);
+
+ /* Copy ELS/CT WR CMD */
+ csio_wr_copy_to_wrp(&fw_wr[0], &wrp, wr_off,
+ sizeof(struct fw_fcoe_els_ct_wr));
+ wr_off += sizeof(struct fw_fcoe_els_ct_wr);
+
+ /* Copy payload to Immediate section of WR */
+ if (im_len)
+ csio_wr_copy_to_wrp(pld->vaddr, &wrp, wr_off, im_len);
+ else {
+ /* Program DSGL to dma payload */
+ dsgl.cmd_nsge = htonl(ULPTX_CMD(ULP_TX_SC_DSGL) |
+ ULPTX_MORE | ULPTX_NSGE(1));
+ dsgl.len0 = cpu_to_be32(pld_len);
+ dsgl.addr0 = cpu_to_be64(pld->paddr);
+ csio_wr_copy_to_wrp(&dsgl, &wrp, ALIGN(wr_off, 8),
+ sizeof(struct ulptx_sgl));
+ }
+
+ /* Issue work request to xmit ELS/CT req to FW */
+ csio_wr_issue(mgmtm->hw, mgmtm->eq_idx, false);
+ return ret;
+}
+
+/*
+ * csio_ln_mgmt_submit_req - Submit FCOE Mgmt request.
+ * @io_req - IO Request
+ * @io_cbfn - Completion handler.
+ * @req_type - ELS or CT request type
+ * @pld - Dma Payload buffer
+ * @pld_len - Payload len
+ *
+ *
+ * This API used submit managment ELS/CT request.
+ * This called with hw lock held
+ * Returns: 0 - on success
+ * -ENOMEM - on error.
+ */
+static int
+csio_ln_mgmt_submit_req(struct csio_ioreq *io_req,
+ void (*io_cbfn) (struct csio_hw *, struct csio_ioreq *),
+ enum fcoe_cmn_type req_type, struct csio_dma_buf *pld,
+ uint32_t pld_len)
+{
+ struct csio_hw *hw = csio_lnode_to_hw(io_req->lnode);
+ struct csio_mgmtm *mgmtm = csio_hw_to_mgmtm(hw);
+ int rv;
+
+ io_req->io_cbfn = io_cbfn; /* Upper layer callback handler */
+ io_req->fw_handle = (uintptr_t) (io_req);
+ io_req->eq_idx = mgmtm->eq_idx;
+ io_req->iq_idx = mgmtm->iq_idx;
+
+ rv = csio_ln_mgmt_submit_wr(mgmtm, io_req, req_type, pld, pld_len);
+ if (rv == 0) {
+ list_add_tail(&io_req->sm.sm_list, &mgmtm->active_q);
+ mgmtm->stats.n_active++;
+ }
+ return rv;
+}
+
+/*
+ * csio_ln_fdmi_init - FDMI Init entry point.
+ * @ln: lnode
+ */
+static int
+csio_ln_fdmi_init(struct csio_lnode *ln)
+{
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+ struct csio_dma_buf *dma_buf;
+
+ /* Allocate MGMT request required for FDMI */
+ ln->mgmt_req = kzalloc(sizeof(struct csio_ioreq), GFP_KERNEL);
+ if (!ln->mgmt_req) {
+ csio_ln_err(ln, "Failed to alloc ioreq for FDMI\n");
+ CSIO_INC_STATS(hw, n_err_nomem);
+ return -ENOMEM;
+ }
+
+ /* Allocate Dma buffers for FDMI response Payload */
+ dma_buf = &ln->mgmt_req->dma_buf;
+ dma_buf->len = 2048;
+ dma_buf->vaddr = pci_alloc_consistent(hw->pdev, dma_buf->len,
+ &dma_buf->paddr);
+ if (!dma_buf->vaddr) {
+ csio_err(hw, "Failed to alloc DMA buffer for FDMI!\n");
+ kfree(ln->mgmt_req);
+ ln->mgmt_req = NULL;
+ return -ENOMEM;
+ }
+
+ ln->flags |= CSIO_LNF_FDMI_ENABLE;
+ return 0;
+}
+
+/*
+ * csio_ln_fdmi_exit - FDMI exit entry point.
+ * @ln: lnode
+ */
+static int
+csio_ln_fdmi_exit(struct csio_lnode *ln)
+{
+ struct csio_dma_buf *dma_buf;
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+
+ if (!ln->mgmt_req)
+ return 0;
+
+ dma_buf = &ln->mgmt_req->dma_buf;
+ if (dma_buf->vaddr)
+ pci_free_consistent(hw->pdev, dma_buf->len, dma_buf->vaddr,
+ dma_buf->paddr);
+
+ kfree(ln->mgmt_req);
+ return 0;
+}
+
+int
+csio_scan_done(struct csio_lnode *ln, unsigned long ticks,
+ unsigned long time, unsigned long max_scan_ticks,
+ unsigned long delta_scan_ticks)
+{
+ int rv = 0;
+
+ if (time >= max_scan_ticks)
+ return 1;
+
+ if (!ln->tgt_scan_tick)
+ ln->tgt_scan_tick = ticks;
+
+ if (((ticks - ln->tgt_scan_tick) >= delta_scan_ticks)) {
+ if (!ln->last_scan_ntgts)
+ ln->last_scan_ntgts = ln->n_scsi_tgts;
+ else {
+ if (ln->last_scan_ntgts == ln->n_scsi_tgts)
+ return 1;
+
+ ln->last_scan_ntgts = ln->n_scsi_tgts;
+ }
+ ln->tgt_scan_tick = ticks;
+ }
+ return rv;
+}
+
+/*
+ * csio_notify_lnodes:
+ * @hw: HW module
+ * @note: Notification
+ *
+ * Called from the HW SM to fan out notifications to the
+ * Lnode SM. Since the HW SM is entered with lock held,
+ * there is no need to hold locks here.
+ *
+ */
+void
+csio_notify_lnodes(struct csio_hw *hw, enum csio_ln_notify note)
+{
+ struct list_head *tmp;
+ struct csio_lnode *ln;
+
+ csio_dbg(hw, "Notifying all nodes of event %d\n", note);
+
+ /* Traverse children lnodes list and send evt */
+ list_for_each(tmp, &hw->sln_head) {
+ ln = (struct csio_lnode *) tmp;
+
+ switch (note) {
+ case CSIO_LN_NOTIFY_HWREADY:
+ csio_lnode_start(ln);
+ break;
+
+ case CSIO_LN_NOTIFY_HWRESET:
+ case CSIO_LN_NOTIFY_HWREMOVE:
+ csio_lnode_close(ln);
+ break;
+
+ case CSIO_LN_NOTIFY_HWSTOP:
+ csio_lnode_stop(ln);
+ break;
+
+ default:
+ break;
+
+ }
+ }
+}
+
+/*
+ * csio_disable_lnodes:
+ * @hw: HW module
+ * @portid:port id
+ * @disable: disable/enable flag.
+ * If disable=1, disables all lnode hosted on given physical port.
+ * otherwise enables all the lnodes on given phsysical port.
+ * This routine need to called with hw lock held.
+ */
+void
+csio_disable_lnodes(struct csio_hw *hw, uint8_t portid, bool disable)
+{
+ struct list_head *tmp;
+ struct csio_lnode *ln;
+
+ csio_dbg(hw, "Notifying event to all nodes of port:%d\n", portid);
+
+ /* Traverse sibling lnodes list and send evt */
+ list_for_each(tmp, &hw->sln_head) {
+ ln = (struct csio_lnode *) tmp;
+ if (ln->portid != portid)
+ continue;
+
+ if (disable)
+ csio_lnode_stop(ln);
+ else
+ csio_lnode_start(ln);
+ }
+}
+
+/*
+ * csio_ln_init - Initialize an lnode.
+ * @ln: lnode
+ *
+ */
+static int
+csio_ln_init(struct csio_lnode *ln)
+{
+ int rv = -EINVAL;
+ struct csio_lnode *rln, *pln;
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+
+ csio_init_state(&ln->sm, csio_lns_uninit);
+ ln->vnp_flowid = CSIO_INVALID_IDX;
+ ln->fcf_flowid = CSIO_INVALID_IDX;
+
+ if (csio_is_root_ln(ln)) {
+
+ /* This is the lnode used during initialization */
+
+ ln->fcfinfo = kzalloc(sizeof(struct csio_fcf_info), GFP_KERNEL);
+ if (!ln->fcfinfo) {
+ csio_ln_err(ln, "Failed to alloc FCF record\n");
+ CSIO_INC_STATS(hw, n_err_nomem);
+ goto err;
+ }
+
+ INIT_LIST_HEAD(&ln->fcf_lsthead);
+ kref_init(&ln->fcfinfo->kref);
+
+ if (csio_fdmi_enable && csio_ln_fdmi_init(ln))
+ goto err;
+
+ } else { /* Either a non-root physical or a virtual lnode */
+
+ /*
+ * THe rest is common for non-root physical and NPIV lnodes.
+ * Just get references to all other modules
+ */
+ rln = csio_root_lnode(ln);
+
+ if (csio_is_npiv_ln(ln)) {
+ /* NPIV */
+ pln = csio_parent_lnode(ln);
+ kref_get(&pln->fcfinfo->kref);
+ ln->fcfinfo = pln->fcfinfo;
+ } else {
+ /* Another non-root physical lnode (FCF) */
+ ln->fcfinfo = kzalloc(sizeof(struct csio_fcf_info),
+ GFP_KERNEL);
+ if (!ln->fcfinfo) {
+ csio_ln_err(ln,
+ "Failed to alloc FCF info\n");
+ CSIO_INC_STATS(hw, n_err_nomem);
+ goto err;
+ }
+
+ kref_init(&ln->fcfinfo->kref);
+
+ if (csio_fdmi_enable && csio_ln_fdmi_init(ln))
+ goto err;
+ }
+
+ } /* if (!csio_is_root_ln(ln)) */
+
+ return 0;
+err:
+ return rv;
+}
+
+static void
+csio_ln_exit(struct csio_lnode *ln)
+{
+ struct csio_lnode *pln;
+
+ csio_cleanup_rns(ln);
+ if (csio_is_npiv_ln(ln)) {
+ pln = csio_parent_lnode(ln);
+ kref_put(&pln->fcfinfo->kref, csio_free_fcfinfo);
+ } else {
+ kref_put(&ln->fcfinfo->kref, csio_free_fcfinfo);
+ if (csio_fdmi_enable)
+ csio_ln_fdmi_exit(ln);
+ }
+ ln->fcfinfo = NULL;
+}
+
+/**
+ * csio_lnode_init - Initialize the members of an lnode.
+ * @ln: lnode
+ *
+ */
+int
+csio_lnode_init(struct csio_lnode *ln, struct csio_hw *hw,
+ struct csio_lnode *pln)
+{
+ int rv = -EINVAL;
+
+ /* Link this lnode to hw */
+ csio_lnode_to_hw(ln) = hw;
+
+ /* Link child to parent if child lnode */
+ if (pln)
+ ln->pln = pln;
+ else
+ ln->pln = NULL;
+
+ /* Initialize scsi_tgt and timers to zero */
+ ln->n_scsi_tgts = 0;
+ ln->last_scan_ntgts = 0;
+ ln->tgt_scan_tick = 0;
+
+ /* Initialize rnode list */
+ INIT_LIST_HEAD(&ln->rnhead);
+ INIT_LIST_HEAD(&ln->cln_head);
+
+ /* Initialize log level for debug */
+ ln->params.log_level = hw->params.log_level;
+
+ if (csio_ln_init(ln))
+ goto err;
+
+ /* Add lnode to list of sibling or children lnodes */
+ spin_lock_irq(&hw->lock);
+ list_add_tail(&ln->sm.sm_list, pln ? &pln->cln_head : &hw->sln_head);
+ if (pln)
+ pln->num_vports++;
+ spin_unlock_irq(&hw->lock);
+
+ hw->num_lns++;
+
+ return 0;
+err:
+ csio_lnode_to_hw(ln) = NULL;
+ return rv;
+}
+
+/**
+ * csio_lnode_exit - De-instantiate an lnode.
+ * @ln: lnode
+ *
+ */
+void
+csio_lnode_exit(struct csio_lnode *ln)
+{
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+
+ csio_ln_exit(ln);
+
+ /* Remove this lnode from hw->sln_head */
+ spin_lock_irq(&hw->lock);
+
+ list_del_init(&ln->sm.sm_list);
+
+ /* If it is children lnode, decrement the
+ * counter in its parent lnode
+ */
+ if (ln->pln)
+ ln->pln->num_vports--;
+
+ /* Update root lnode pointer */
+ if (list_empty(&hw->sln_head))
+ hw->rln = NULL;
+ else
+ hw->rln = (struct csio_lnode *)csio_list_next(&hw->sln_head);
+
+ spin_unlock_irq(&hw->lock);
+
+ csio_lnode_to_hw(ln) = NULL;
+ hw->num_lns--;
+}
diff --git a/drivers/scsi/csiostor/csio_rnode.c b/drivers/scsi/csiostor/csio_rnode.c
new file mode 100644
index 0000000..5e224a0
--- /dev/null
+++ b/drivers/scsi/csiostor/csio_rnode.c
@@ -0,0 +1,889 @@
+/*
+ * This file is part of the Chelsio FCoE driver for Linux.
+ *
+ * Copyright (c) 2008-2012 Chelsio Communications, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/string.h>
+#include <scsi/scsi_transport_fc.h>
+#include <scsi/fc/fc_els.h>
+#include <scsi/fc/fc_fs.h>
+
+#include "csio_hw.h"
+#include "csio_lnode.h"
+#include "csio_rnode.h"
+
+static int csio_rnode_init(struct csio_rnode *, struct csio_lnode *);
+static void csio_rnode_exit(struct csio_rnode *);
+
+/* Static machine forward declarations */
+static void csio_rns_uninit(struct csio_rnode *, enum csio_rn_ev);
+static void csio_rns_ready(struct csio_rnode *, enum csio_rn_ev);
+static void csio_rns_offline(struct csio_rnode *, enum csio_rn_ev);
+static void csio_rns_disappeared(struct csio_rnode *, enum csio_rn_ev);
+
+/* RNF event mapping */
+static enum csio_rn_ev fwevt_to_rnevt[] = {
+ CSIO_RNFE_NONE, /* None */
+ CSIO_RNFE_LOGGED_IN, /* PLOGI_ACC_RCVD */
+ CSIO_RNFE_NONE, /* PLOGI_RJT_RCVD */
+ CSIO_RNFE_PLOGI_RECV, /* PLOGI_RCVD */
+ CSIO_RNFE_LOGO_RECV, /* PLOGO_RCVD */
+ CSIO_RNFE_PRLI_DONE, /* PRLI_ACC_RCVD */
+ CSIO_RNFE_NONE, /* PRLI_RJT_RCVD */
+ CSIO_RNFE_PRLI_RECV, /* PRLI_RCVD */
+ CSIO_RNFE_PRLO_RECV, /* PRLO_RCVD */
+ CSIO_RNFE_NONE, /* NPORT_ID_CHGD */
+ CSIO_RNFE_LOGO_RECV, /* FLOGO_RCVD */
+ CSIO_RNFE_NONE, /* CLR_VIRT_LNK_RCVD */
+ CSIO_RNFE_LOGGED_IN, /* FLOGI_ACC_RCVD */
+ CSIO_RNFE_NONE, /* FLOGI_RJT_RCVD */
+ CSIO_RNFE_LOGGED_IN, /* FDISC_ACC_RCVD */
+ CSIO_RNFE_NONE, /* FDISC_RJT_RCVD */
+ CSIO_RNFE_NONE, /* FLOGI_TMO_MAX_RETRY */
+ CSIO_RNFE_NONE, /* IMPL_LOGO_ADISC_ACC */
+ CSIO_RNFE_NONE, /* IMPL_LOGO_ADISC_RJT */
+ CSIO_RNFE_NONE, /* IMPL_LOGO_ADISC_CNFLT */
+ CSIO_RNFE_NONE, /* PRLI_TMO */
+ CSIO_RNFE_NONE, /* ADISC_TMO */
+ CSIO_RNFE_NAME_MISSING, /* RSCN_DEV_LOST */
+ CSIO_RNFE_NONE, /* SCR_ACC_RCVD */
+ CSIO_RNFE_NONE, /* ADISC_RJT_RCVD */
+ CSIO_RNFE_NONE, /* LOGO_SNT */
+ CSIO_RNFE_LOGO_RECV, /* PROTO_ERR_IMPL_LOGO */
+};
+
+#define CSIO_FWE_TO_RNFE(_evt) ((_evt > PROTO_ERR_IMPL_LOGO) ? \
+ CSIO_RNFE_NONE : \
+ fwevt_to_rnevt[_evt])
+int
+csio_is_rnode_ready(struct csio_rnode *rn)
+{
+ return csio_match_state(rn, csio_rns_ready);
+}
+
+static int
+csio_is_rnode_uninit(struct csio_rnode *rn)
+{
+ return csio_match_state(rn, csio_rns_uninit);
+}
+
+/*
+ * csio_rn_lookup - Finds the rnode with the given flowid
+ * @ln - lnode
+ * @flowid - flowid.
+ *
+ * Does the rnode lookup on the given lnode and flowid.If no matching entry
+ * found, NULL is returned.
+ */
+static struct csio_rnode *
+csio_rn_lookup(struct csio_lnode *ln, uint32_t flowid)
+{
+ struct csio_rnode *rnhead = (struct csio_rnode *) &ln->rnhead;
+ struct list_head *tmp;
+ struct csio_rnode *rn;
+
+ list_for_each(tmp, &rnhead->sm.sm_list) {
+ rn = (struct csio_rnode *) tmp;
+ if (rn->flowid == flowid)
+ return rn;
+ }
+
+ return NULL;
+}
+
+/*
+ * csio_rn_lookup_wwpn - Finds the rnode with the given wwpn
+ * @ln: lnode
+ * @wwpn: wwpn
+ *
+ * Does the rnode lookup on the given lnode and wwpn. If no matching entry
+ * found, NULL is returned.
+ */
+static struct csio_rnode *
+csio_rn_lookup_wwpn(struct csio_lnode *ln, uint8_t *wwpn)
+{
+ struct csio_rnode *rnhead = (struct csio_rnode *) &ln->rnhead;
+ struct list_head *tmp;
+ struct csio_rnode *rn;
+
+ list_for_each(tmp, &rnhead->sm.sm_list) {
+ rn = (struct csio_rnode *) tmp;
+ if (!memcmp(csio_rn_wwpn(rn), wwpn, 8))
+ return rn;
+ }
+
+ return NULL;
+}
+
+/**
+ * csio_rnode_lookup_portid - Finds the rnode with the given portid
+ * @ln: lnode
+ * @portid: port id
+ *
+ * Lookup the rnode list for a given portid. If no matching entry
+ * found, NULL is returned.
+ */
+struct csio_rnode *
+csio_rnode_lookup_portid(struct csio_lnode *ln, uint32_t portid)
+{
+ struct csio_rnode *rnhead = (struct csio_rnode *) &ln->rnhead;
+ struct list_head *tmp;
+ struct csio_rnode *rn;
+
+ list_for_each(tmp, &rnhead->sm.sm_list) {
+ rn = (struct csio_rnode *) tmp;
+ if (rn->nport_id == portid)
+ return rn;
+ }
+
+ return NULL;
+}
+
+static int
+csio_rn_dup_flowid(struct csio_lnode *ln, uint32_t rdev_flowid,
+ uint32_t *vnp_flowid)
+{
+ struct csio_rnode *rnhead;
+ struct list_head *tmp, *tmp1;
+ struct csio_rnode *rn;
+ struct csio_lnode *ln_tmp;
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+
+ list_for_each(tmp1, &hw->sln_head) {
+ ln_tmp = (struct csio_lnode *) tmp1;
+ if (ln_tmp == ln)
+ continue;
+
+ rnhead = (struct csio_rnode *)&ln_tmp->rnhead;
+ list_for_each(tmp, &rnhead->sm.sm_list) {
+
+ rn = (struct csio_rnode *) tmp;
+ if (csio_is_rnode_ready(rn)) {
+ if (rn->flowid == rdev_flowid) {
+ *vnp_flowid = csio_ln_flowid(ln_tmp);
+ return 1;
+ }
+ }
+ }
+ }
+
+ return 0;
+}
+
+static struct csio_rnode *
+csio_alloc_rnode(struct csio_lnode *ln)
+{
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+
+ struct csio_rnode *rn = mempool_alloc(hw->rnode_mempool, GFP_ATOMIC);
+ if (!rn)
+ goto err;
+
+ memset(rn, 0, sizeof(struct csio_rnode));
+ if (csio_rnode_init(rn, ln))
+ goto err_free;
+
+ CSIO_INC_STATS(ln, n_rnode_alloc);
+
+ return rn;
+
+err_free:
+ mempool_free(rn, hw->rnode_mempool);
+err:
+ CSIO_INC_STATS(ln, n_rnode_nomem);
+ return NULL;
+}
+
+static void
+csio_free_rnode(struct csio_rnode *rn)
+{
+ struct csio_hw *hw = csio_lnode_to_hw(csio_rnode_to_lnode(rn));
+
+ csio_rnode_exit(rn);
+ CSIO_INC_STATS(rn->lnp, n_rnode_free);
+ mempool_free(rn, hw->rnode_mempool);
+}
+
+/*
+ * csio_get_rnode - Gets rnode with the given flowid
+ * @ln - lnode
+ * @flowid - flow id.
+ *
+ * Does the rnode lookup on the given lnode and flowid. If no matching
+ * rnode found, then new rnode with given npid is allocated and returned.
+ */
+static struct csio_rnode *
+csio_get_rnode(struct csio_lnode *ln, uint32_t flowid)
+{
+ struct csio_rnode *rn;
+
+ rn = csio_rn_lookup(ln, flowid);
+ if (!rn) {
+ rn = csio_alloc_rnode(ln);
+ if (!rn)
+ return NULL;
+
+ rn->flowid = flowid;
+ }
+
+ return rn;
+}
+
+/*
+ * csio_put_rnode - Frees the given rnode
+ * @ln - lnode
+ * @flowid - flow id.
+ *
+ * Does the rnode lookup on the given lnode and flowid. If no matching
+ * rnode found, then new rnode with given npid is allocated and returned.
+ */
+void
+csio_put_rnode(struct csio_lnode *ln, struct csio_rnode *rn)
+{
+ CSIO_DB_ASSERT(csio_is_rnode_uninit(rn) != 0);
+ csio_free_rnode(rn);
+}
+
+/*
+ * csio_confirm_rnode - confirms rnode based on wwpn.
+ * @ln: lnode
+ * @rdev_flowid: remote device flowid
+ * @rdevp: remote device params
+ * This routines searches other rnode in list having same wwpn of new rnode.
+ * If there is a match, then matched rnode is returned and otherwise new rnode
+ * is returned.
+ * returns rnode.
+ */
+struct csio_rnode *
+csio_confirm_rnode(struct csio_lnode *ln, uint32_t rdev_flowid,
+ struct fcoe_rdev_entry *rdevp)
+{
+ uint8_t rport_type;
+ struct csio_rnode *rn, *match_rn;
+ uint32_t vnp_flowid;
+
+ rport_type =
+ FW_RDEV_WR_RPORT_TYPE_GET(rdevp->rd_xfer_rdy_to_rport_type);
+ /* Drop rdev event for cntrl port */
+ if (rport_type == FAB_CTLR_VNPORT) {
+ csio_ln_dbg(ln,
+ "Unhandled rport_type:%d recv in rdev evt "
+ "ssni:x%x\n", rport_type, rdev_flowid);
+ return NULL;
+ }
+
+ /* Lookup on flowid */
+ rn = csio_rn_lookup(ln, rdev_flowid);
+ if (!rn) {
+
+ /* Drop events with duplicate flowid */
+ if (csio_rn_dup_flowid(ln, rdev_flowid, &vnp_flowid)) {
+ csio_ln_warn(ln,
+ "ssni:%x already active on vnpi:%x",
+ rdev_flowid, vnp_flowid);
+ return NULL;
+ }
+
+ /* skip wwpn lookup for fabric ports, cntrl port */
+ if (rport_type == FLOGI_VFPORT || rport_type == FDISC_VFPORT
+ || rport_type == FAB_CTLR_VNPORT) {
+ goto alloc_rnode;
+ }
+
+ /* Lookup on wwpn for NPORTs */
+ rn = csio_rn_lookup_wwpn(ln, rdevp->wwpn);
+ if (!rn)
+ goto alloc_rnode;
+
+ /* found rn */
+ goto found_rnode;
+ } else {
+ /* verify rnode found for fabric ports, cntrl port */
+ if (rport_type == FLOGI_VFPORT || rport_type == FDISC_VFPORT
+ || rport_type == FAB_CTLR_VNPORT) {
+
+ /* Rnode role mismatch. Allocate new rnode */
+ if (rn->role == CSIO_RNFR_NS ||
+ rn->role == CSIO_RNFR_NPORT) {
+ csio_ln_dbg(ln,
+ "rnode role mismatch found ssni:x%x "
+ "role:%d new_type:%d\n",
+ rdev_flowid, rn->role, rport_type);
+ if (csio_is_rnode_ready(rn)) {
+ csio_ln_warn(ln,
+ "rnode is already"
+ "active ssni:x%x\n",
+ rdev_flowid);
+ CSIO_DB_ASSERT(0);
+ }
+ csio_rn_flowid(rn) = CSIO_INVALID_IDX;
+ goto alloc_rnode;
+ } else
+ goto found_rnode;
+ }
+
+ /* wwpn match */
+ if (!memcmp(csio_rn_wwpn(rn), rdevp->wwpn, 8)) {
+ /* Update rn */
+ goto found_rnode;
+ }
+
+ /* Search for rnode that have same wwpn */
+ match_rn = csio_rn_lookup_wwpn(ln, rdevp->wwpn);
+ if (match_rn != NULL) {
+ csio_ln_dbg(ln,
+ "ssni:x%x changed for rport name(wwpn):%llx "
+ "did:x%x\n", rdev_flowid,
+ wwn_to_u64(rdevp->wwpn),
+ match_rn->nport_id);
+ csio_rn_flowid(rn) = CSIO_INVALID_IDX;
+ rn = match_rn;
+ CSIO_INC_STATS(ln, n_rnode_match);
+ } else {
+ csio_ln_dbg(ln,
+ "rnode wwpn mismatch found ssni:x%x "
+ "name(wwpn):%llx\n",
+ rdev_flowid,
+ wwn_to_u64(csio_rn_wwpn(rn)));
+ if (csio_is_rnode_ready(rn)) {
+ csio_ln_warn(ln,
+ "rnode is already active "
+ "wwpn:%llx ssni:x%x\n",
+ wwn_to_u64(csio_rn_wwpn(rn)),
+ rdev_flowid);
+ CSIO_DB_ASSERT(0);
+ }
+ csio_rn_flowid(rn) = CSIO_INVALID_IDX;
+ goto alloc_rnode;
+ }
+ }
+
+found_rnode:
+ csio_ln_dbg(ln, "found rnode:%p ssni:x%x name(wwpn):%llx\n",
+ rn, rdev_flowid, wwn_to_u64(rdevp->wwpn));
+
+ /* Update flowid */
+ csio_rn_flowid(rn) = rdev_flowid;
+
+ /* update rdev entry */
+ rn->rdev_entry = rdevp;
+ return rn;
+
+alloc_rnode:
+ rn = csio_get_rnode(ln, rdev_flowid);
+ if (!rn)
+ return NULL;
+
+ csio_ln_dbg(ln, "alloc rnode:%p ssni:x%x name(wwpn):%llx\n",
+ rn, rdev_flowid, wwn_to_u64(rdevp->wwpn));
+
+ /* update rdev entry */
+ rn->rdev_entry = rdevp;
+ return rn;
+}
+
+/*
+ * csio_rn_verify_rparams - verify rparams.
+ * @ln: lnode
+ * @rn: rnode
+ * @rdevp: remote device params
+ * returns success if rparams are verified.
+ */
+static int
+csio_rn_verify_rparams(struct csio_lnode *ln, struct csio_rnode *rn,
+ struct fcoe_rdev_entry *rdevp)
+{
+ uint8_t null[8];
+ uint8_t rport_type;
+ uint8_t fc_class;
+ uint32_t *did;
+
+ did = (uint32_t *) &rdevp->r_id[0];
+ rport_type =
+ FW_RDEV_WR_RPORT_TYPE_GET(rdevp->rd_xfer_rdy_to_rport_type);
+ switch (rport_type) {
+ case FLOGI_VFPORT:
+ rn->role = CSIO_RNFR_FABRIC;
+ if (((ntohl(*did) >> 8) & CSIO_DID_MASK) != FC_FID_FLOGI) {
+ csio_ln_err(ln, "ssni:x%x invalid fabric portid\n",
+ csio_rn_flowid(rn));
+ return -EINVAL;
+ }
+ /* NPIV support */
+ if (FW_RDEV_WR_NPIV_GET(rdevp->vft_to_qos))
+ ln->flags |= CSIO_LNF_NPIVSUPP;
+
+ break;
+
+ case NS_VNPORT:
+ rn->role = CSIO_RNFR_NS;
+ if (((ntohl(*did) >> 8) & CSIO_DID_MASK) != FC_FID_DIR_SERV) {
+ csio_ln_err(ln, "ssni:x%x invalid fabric portid\n",
+ csio_rn_flowid(rn));
+ return -EINVAL;
+ }
+ break;
+
+ case REG_FC4_VNPORT:
+ case REG_VNPORT:
+ rn->role = CSIO_RNFR_NPORT;
+ if (rdevp->event_cause == PRLI_ACC_RCVD ||
+ rdevp->event_cause == PRLI_RCVD) {
+ if (FW_RDEV_WR_TASK_RETRY_ID_GET(
+ rdevp->enh_disc_to_tgt))
+ rn->fcp_flags |= FCP_SPPF_OVLY_ALLOW;
+
+ if (FW_RDEV_WR_RETRY_GET(rdevp->enh_disc_to_tgt))
+ rn->fcp_flags |= FCP_SPPF_RETRY;
+
+ if (FW_RDEV_WR_CONF_CMPL_GET(rdevp->enh_disc_to_tgt))
+ rn->fcp_flags |= FCP_SPPF_CONF_COMPL;
+
+ if (FW_RDEV_WR_TGT_GET(rdevp->enh_disc_to_tgt))
+ rn->role |= CSIO_RNFR_TARGET;
+
+ if (FW_RDEV_WR_INI_GET(rdevp->enh_disc_to_tgt))
+ rn->role |= CSIO_RNFR_INITIATOR;
+ }
+
+ break;
+
+ case FDMI_VNPORT:
+ case FAB_CTLR_VNPORT:
+ rn->role = 0;
+ break;
+
+ default:
+ csio_ln_err(ln, "ssni:x%x invalid rport type recv x%x\n",
+ csio_rn_flowid(rn), rport_type);
+ return -EINVAL;
+ }
+
+ /* validate wwpn/wwnn for Name server/remote port */
+ if (rport_type == REG_VNPORT || rport_type == NS_VNPORT) {
+ memset(null, 0, 8);
+ if (!memcmp(rdevp->wwnn, null, 8)) {
+ csio_ln_err(ln,
+ "ssni:x%x invalid wwnn received from"
+ " rport did:x%x\n",
+ csio_rn_flowid(rn),
+ (ntohl(*did) & CSIO_DID_MASK));
+ return -EINVAL;
+ }
+
+ if (!memcmp(rdevp->wwpn, null, 8)) {
+ csio_ln_err(ln,
+ "ssni:x%x invalid wwpn received from"
+ " rport did:x%x\n",
+ csio_rn_flowid(rn),
+ (ntohl(*did) & CSIO_DID_MASK));
+ return -EINVAL;
+ }
+
+ }
+
+ /* Copy wwnn, wwpn and nport id */
+ rn->nport_id = (ntohl(*did) >> 8) & CSIO_DID_MASK;
+ memcpy(csio_rn_wwnn(rn), rdevp->wwnn, 8);
+ memcpy(csio_rn_wwpn(rn), rdevp->wwpn, 8);
+ rn->rn_sparm.csp.sp_bb_data = ntohs(rdevp->rcv_fr_sz);
+ fc_class = FW_RDEV_WR_CLASS_GET(rdevp->vft_to_qos);
+ rn->rn_sparm.clsp[fc_class - 1].cp_class = htons(FC_CPC_VALID);
+ return 0;
+}
+
+static void
+__csio_reg_rnode(struct csio_rnode *rn)
+{
+ struct csio_hw *hw = csio_lnode_to_hw(csio_rnode_to_lnode(rn));
+ struct csio_lnode *ln = csio_rnode_to_lnode(rn);
+
+ spin_unlock_irq(&hw->lock);
+ csio_reg_rnode(rn);
+ spin_lock_irq(&hw->lock);
+
+ if (rn->nport_id == FC_FID_MGMT_SERV)
+ csio_ln_fdmi_start(ln, (void *) rn);
+}
+
+static void
+__csio_unreg_rnode(struct csio_rnode *rn)
+{
+ struct csio_hw *hw = csio_lnode_to_hw(csio_rnode_to_lnode(rn));
+ LIST_HEAD(tmp_q);
+ int cmpl = 0;
+
+ if (!list_empty(&rn->host_cmpl_q)) {
+ csio_dbg(hw, "Returning completion queue I/Os\n");
+ list_splice_tail_init(&rn->host_cmpl_q, &tmp_q);
+ cmpl = 1;
+ }
+
+ spin_unlock_irq(&hw->lock);
+ csio_unreg_rnode(rn);
+ spin_lock_irq(&hw->lock);
+
+ /* Cleanup I/Os that were waiting for rnode to unregister */
+ if (cmpl)
+ csio_scsi_cleanup_io_q(csio_hw_to_scsim(hw), &tmp_q);
+
+}
+
+/*****************************************************************************/
+/* START: Rnode SM */
+/*****************************************************************************/
+
+/*
+ * csio_rns_uninit -
+ * @rn - rnode
+ * @evt - SM event.
+ *
+ */
+static void
+csio_rns_uninit(struct csio_rnode *rn, enum csio_rn_ev evt)
+{
+ struct csio_lnode *ln = csio_rnode_to_lnode(rn);
+ int ret = 0;
+
+ CSIO_INC_STATS(rn, n_evt_sm[evt]);
+
+ switch (evt) {
+ case CSIO_RNFE_LOGGED_IN:
+ case CSIO_RNFE_PLOGI_RECV:
+ ret = csio_rn_verify_rparams(ln, rn, rn->rdev_entry);
+ if (!ret) {
+ csio_set_state(&rn->sm, csio_rns_ready);
+ __csio_reg_rnode(rn);
+ } else {
+ CSIO_INC_STATS(rn, n_err_inval);
+ }
+ break;
+ case CSIO_RNFE_LOGO_RECV:
+ csio_ln_dbg(ln,
+ "ssni:x%x Ignoring event %d recv "
+ "in rn state[uninit]\n", csio_rn_flowid(rn), evt);
+ CSIO_INC_STATS(rn, n_evt_drop);
+ break;
+ default:
+ csio_ln_dbg(ln,
+ "ssni:x%x unexp event %d recv "
+ "in rn state[uninit]\n", csio_rn_flowid(rn), evt);
+ CSIO_INC_STATS(rn, n_evt_unexp);
+ break;
+ }
+}
+
+/*
+ * csio_rns_ready -
+ * @rn - rnode
+ * @evt - SM event.
+ *
+ */
+static void
+csio_rns_ready(struct csio_rnode *rn, enum csio_rn_ev evt)
+{
+ struct csio_lnode *ln = csio_rnode_to_lnode(rn);
+ int ret = 0;
+
+ CSIO_INC_STATS(rn, n_evt_sm[evt]);
+
+ switch (evt) {
+ case CSIO_RNFE_LOGGED_IN:
+ case CSIO_RNFE_PLOGI_RECV:
+ csio_ln_dbg(ln,
+ "ssni:x%x Ignoring event %d recv from did:x%x "
+ "in rn state[ready]\n", csio_rn_flowid(rn), evt,
+ rn->nport_id);
+ CSIO_INC_STATS(rn, n_evt_drop);
+ break;
+
+ case CSIO_RNFE_PRLI_DONE:
+ case CSIO_RNFE_PRLI_RECV:
+ ret = csio_rn_verify_rparams(ln, rn, rn->rdev_entry);
+ if (!ret)
+ __csio_reg_rnode(rn);
+ else
+ CSIO_INC_STATS(rn, n_err_inval);
+
+ break;
+ case CSIO_RNFE_DOWN:
+ csio_set_state(&rn->sm, csio_rns_offline);
+ __csio_unreg_rnode(rn);
+
+ /* FW expected to internally aborted outstanding SCSI WRs
+ * and return all SCSI WRs to host with status "ABORTED".
+ */
+ break;
+
+ case CSIO_RNFE_LOGO_RECV:
+ csio_set_state(&rn->sm, csio_rns_offline);
+
+ __csio_unreg_rnode(rn);
+
+ /* FW expected to internally aborted outstanding SCSI WRs
+ * and return all SCSI WRs to host with status "ABORTED".
+ */
+ break;
+
+ case CSIO_RNFE_CLOSE:
+ /*
+ * Each rnode receives CLOSE event when driver is removed or
+ * device is reset
+ * Note: All outstanding IOs on remote port need to returned
+ * to uppper layer with appropriate error before sending
+ * CLOSE event
+ */
+ csio_set_state(&rn->sm, csio_rns_uninit);
+ __csio_unreg_rnode(rn);
+ break;
+
+ case CSIO_RNFE_NAME_MISSING:
+ csio_set_state(&rn->sm, csio_rns_disappeared);
+ __csio_unreg_rnode(rn);
+
+ /*
+ * FW expected to internally aborted outstanding SCSI WRs
+ * and return all SCSI WRs to host with status "ABORTED".
+ */
+
+ break;
+
+ default:
+ csio_ln_dbg(ln,
+ "ssni:x%x unexp event %d recv from did:x%x "
+ "in rn state[uninit]\n", csio_rn_flowid(rn), evt,
+ rn->nport_id);
+ CSIO_INC_STATS(rn, n_evt_unexp);
+ break;
+ }
+}
+
+/*
+ * csio_rns_offline -
+ * @rn - rnode
+ * @evt - SM event.
+ *
+ */
+static void
+csio_rns_offline(struct csio_rnode *rn, enum csio_rn_ev evt)
+{
+ struct csio_lnode *ln = csio_rnode_to_lnode(rn);
+ int ret = 0;
+
+ CSIO_INC_STATS(rn, n_evt_sm[evt]);
+
+ switch (evt) {
+ case CSIO_RNFE_LOGGED_IN:
+ case CSIO_RNFE_PLOGI_RECV:
+ ret = csio_rn_verify_rparams(ln, rn, rn->rdev_entry);
+ if (!ret) {
+ csio_set_state(&rn->sm, csio_rns_ready);
+ __csio_reg_rnode(rn);
+ } else {
+ CSIO_INC_STATS(rn, n_err_inval);
+ csio_post_event(&rn->sm, CSIO_RNFE_CLOSE);
+ }
+ break;
+
+ case CSIO_RNFE_DOWN:
+ csio_ln_dbg(ln,
+ "ssni:x%x Ignoring event %d recv from did:x%x "
+ "in rn state[offline]\n", csio_rn_flowid(rn), evt,
+ rn->nport_id);
+ CSIO_INC_STATS(rn, n_evt_drop);
+ break;
+
+ case CSIO_RNFE_CLOSE:
+ /* Each rnode receives CLOSE event when driver is removed or
+ * device is reset
+ * Note: All outstanding IOs on remote port need to returned
+ * to uppper layer with appropriate error before sending
+ * CLOSE event
+ */
+ csio_set_state(&rn->sm, csio_rns_uninit);
+ break;
+
+ case CSIO_RNFE_NAME_MISSING:
+ csio_set_state(&rn->sm, csio_rns_disappeared);
+ break;
+
+ default:
+ csio_ln_dbg(ln,
+ "ssni:x%x unexp event %d recv from did:x%x "
+ "in rn state[offline]\n", csio_rn_flowid(rn), evt,
+ rn->nport_id);
+ CSIO_INC_STATS(rn, n_evt_unexp);
+ break;
+ }
+}
+
+/*
+ * csio_rns_disappeared -
+ * @rn - rnode
+ * @evt - SM event.
+ *
+ */
+static void
+csio_rns_disappeared(struct csio_rnode *rn, enum csio_rn_ev evt)
+{
+ struct csio_lnode *ln = csio_rnode_to_lnode(rn);
+ int ret = 0;
+
+ CSIO_INC_STATS(rn, n_evt_sm[evt]);
+
+ switch (evt) {
+ case CSIO_RNFE_LOGGED_IN:
+ case CSIO_RNFE_PLOGI_RECV:
+ ret = csio_rn_verify_rparams(ln, rn, rn->rdev_entry);
+ if (!ret) {
+ csio_set_state(&rn->sm, csio_rns_ready);
+ __csio_reg_rnode(rn);
+ } else {
+ CSIO_INC_STATS(rn, n_err_inval);
+ csio_post_event(&rn->sm, CSIO_RNFE_CLOSE);
+ }
+ break;
+
+ case CSIO_RNFE_CLOSE:
+ /* Each rnode receives CLOSE event when driver is removed or
+ * device is reset.
+ * Note: All outstanding IOs on remote port need to returned
+ * to uppper layer with appropriate error before sending
+ * CLOSE event
+ */
+ csio_set_state(&rn->sm, csio_rns_uninit);
+ break;
+
+ case CSIO_RNFE_DOWN:
+ case CSIO_RNFE_NAME_MISSING:
+ csio_ln_dbg(ln,
+ "ssni:x%x Ignoring event %d recv from did x%x"
+ "in rn state[disappeared]\n", csio_rn_flowid(rn),
+ evt, rn->nport_id);
+ break;
+
+ default:
+ csio_ln_dbg(ln,
+ "ssni:x%x unexp event %d recv from did x%x"
+ "in rn state[disappeared]\n", csio_rn_flowid(rn),
+ evt, rn->nport_id);
+ CSIO_INC_STATS(rn, n_evt_unexp);
+ break;
+ }
+}
+
+/*****************************************************************************/
+/* END: Rnode SM */
+/*****************************************************************************/
+
+/*
+ * csio_rnode_devloss_handler - Device loss event handler
+ * @rn: rnode
+ *
+ * Post event to close rnode SM and free rnode.
+ */
+void
+csio_rnode_devloss_handler(struct csio_rnode *rn)
+{
+ struct csio_lnode *ln = csio_rnode_to_lnode(rn);
+
+ /* ignore if same rnode came back as online */
+ if (csio_is_rnode_ready(rn))
+ return;
+
+ csio_post_event(&rn->sm, CSIO_RNFE_CLOSE);
+
+ /* Free rn if in uninit state */
+ if (csio_is_rnode_uninit(rn))
+ csio_put_rnode(ln, rn);
+}
+
+/**
+ * csio_rnode_fwevt_handler - Event handler for firmware rnode events.
+ * @rn: rnode
+ *
+ */
+void
+csio_rnode_fwevt_handler(struct csio_rnode *rn, uint8_t fwevt)
+{
+ struct csio_lnode *ln = csio_rnode_to_lnode(rn);
+ enum csio_rn_ev evt;
+
+ evt = CSIO_FWE_TO_RNFE(fwevt);
+ if (!evt) {
+ csio_ln_err(ln, "ssni:x%x Unhandled FW Rdev event: %d\n",
+ csio_rn_flowid(rn), fwevt);
+ CSIO_INC_STATS(rn, n_evt_unexp);
+ return;
+ }
+ CSIO_INC_STATS(rn, n_evt_fw[fwevt]);
+
+ /* Track previous & current events for debugging */
+ rn->prev_evt = rn->cur_evt;
+ rn->cur_evt = fwevt;
+
+ /* Post event to rnode SM */
+ csio_post_event(&rn->sm, evt);
+
+ /* Free rn if in uninit state */
+ if (csio_is_rnode_uninit(rn))
+ csio_put_rnode(ln, rn);
+}
+
+/*
+ * csio_rnode_init - Initialize rnode.
+ * @rn: RNode
+ * @ln: Associated lnode
+ *
+ * Caller is responsible for holding the lock. The lock is required
+ * to be held for inserting the rnode in ln->rnhead list.
+ */
+static int
+csio_rnode_init(struct csio_rnode *rn, struct csio_lnode *ln)
+{
+ csio_rnode_to_lnode(rn) = ln;
+ csio_init_state(&rn->sm, csio_rns_uninit);
+ INIT_LIST_HEAD(&rn->host_cmpl_q);
+ csio_rn_flowid(rn) = CSIO_INVALID_IDX;
+
+ /* Add rnode to list of lnodes->rnhead */
+ list_add_tail(&rn->sm.sm_list, &ln->rnhead);
+
+ return 0;
+}
+
+static void
+csio_rnode_exit(struct csio_rnode *rn)
+{
+ list_del_init(&rn->sm.sm_list);
+ CSIO_DB_ASSERT(list_empty(&rn->host_cmpl_q));
+}
--
1.7.1
^ permalink raw reply related
* [V2 PATCH 3/9] csiostor: Chelsio FCoE offload driver submission (sources part 3).
From: Naresh Kumar Inna @ 2012-09-05 12:33 UTC (permalink / raw)
To: JBottomley, linux-scsi, dm, leedom; +Cc: netdev, naresh, chethan
In-Reply-To: <1346848442-4573-1-git-send-email-naresh@chelsio.com>
This patch contains code for the FC transport template callbacks and the
Mailbox module functionality. The FC transport callbacks include Virtual
Node ports creation and deletion, FC session registration, unregistration and
teardown. The Mailbox module provides services to issue/track/cancel
mailbox commands and wrappers for them.
Signed-off-by: Naresh Kumar Inna <naresh@chelsio.com>
---
drivers/scsi/csiostor/csio_attr.c | 809 +++++++++++++++++
drivers/scsi/csiostor/csio_mb.c | 1769 +++++++++++++++++++++++++++++++++++++
2 files changed, 2578 insertions(+), 0 deletions(-)
create mode 100644 drivers/scsi/csiostor/csio_attr.c
create mode 100644 drivers/scsi/csiostor/csio_mb.c
diff --git a/drivers/scsi/csiostor/csio_attr.c b/drivers/scsi/csiostor/csio_attr.c
new file mode 100644
index 0000000..ad29fd9
--- /dev/null
+++ b/drivers/scsi/csiostor/csio_attr.c
@@ -0,0 +1,809 @@
+/*
+ * This file is part of the Chelsio FCoE driver for Linux.
+ *
+ * Copyright (c) 2008-2012 Chelsio Communications, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/delay.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/pci.h>
+#include <linux/mm.h>
+#include <linux/jiffies.h>
+#include <scsi/fc/fc_fs.h>
+
+#include "csio_init.h"
+
+static void
+csio_vport_set_state(struct csio_lnode *ln);
+
+/*
+ * csio_reg_rnode - Register a remote port with FC transport.
+ * @rn: Rnode representing remote port.
+ *
+ * Call fc_remote_port_add() to register this remote port with FC transport.
+ * If remote port is Initiator OR Target OR both, change the role appropriately.
+ *
+ */
+void
+csio_reg_rnode(struct csio_rnode *rn)
+{
+ struct csio_lnode *ln = csio_rnode_to_lnode(rn);
+ struct Scsi_Host *shost = csio_ln_to_shost(ln);
+ struct fc_rport_identifiers ids;
+ struct fc_rport *rport;
+ struct csio_service_parms *sp;
+
+ ids.node_name = wwn_to_u64(csio_rn_wwnn(rn));
+ ids.port_name = wwn_to_u64(csio_rn_wwpn(rn));
+ ids.port_id = rn->nport_id;
+ ids.roles = FC_RPORT_ROLE_UNKNOWN;
+
+ if (rn->role & CSIO_RNFR_INITIATOR || rn->role & CSIO_RNFR_TARGET) {
+ rport = rn->rport;
+ CSIO_ASSERT(rport != NULL);
+ goto update_role;
+ }
+
+ rn->rport = fc_remote_port_add(shost, 0, &ids);
+ if (!rn->rport) {
+ csio_ln_err(ln, "Failed to register rport = 0x%x.\n",
+ rn->nport_id);
+ return;
+ }
+
+ ln->num_reg_rnodes++;
+ rport = rn->rport;
+ spin_lock_irq(shost->host_lock);
+ *((struct csio_rnode **)rport->dd_data) = rn;
+ spin_unlock_irq(shost->host_lock);
+
+ sp = &rn->rn_sparm;
+ rport->maxframe_size = sp->csp.sp_bb_data;
+ if (ntohs(sp->clsp[2].cp_class) & FC_CPC_VALID)
+ rport->supported_classes = FC_COS_CLASS3;
+ else
+ rport->supported_classes = FC_COS_UNSPECIFIED;
+update_role:
+ if (rn->role & CSIO_RNFR_INITIATOR)
+ ids.roles |= FC_RPORT_ROLE_FCP_INITIATOR;
+ if (rn->role & CSIO_RNFR_TARGET) {
+ ids.roles |= FC_RPORT_ROLE_FCP_TARGET;
+ ln->n_scsi_tgts++;
+ }
+
+ if (ids.roles != FC_RPORT_ROLE_UNKNOWN)
+ fc_remote_port_rolechg(rport, ids.roles);
+
+ rn->scsi_id = rport->scsi_target_id;
+
+ csio_ln_dbg(ln, "Remote port x%x role 0x%x registered\n",
+ rn->nport_id, ids.roles);
+}
+
+/*
+ * csio_unreg_rnode - Unregister a remote port with FC transport.
+ * @rn: Rnode representing remote port.
+ *
+ * Call fc_remote_port_delete() to unregister this remote port with FC
+ * transport.
+ *
+ */
+void
+csio_unreg_rnode(struct csio_rnode *rn)
+{
+ struct csio_lnode *ln = csio_rnode_to_lnode(rn);
+ struct fc_rport *rport = rn->rport;
+
+ rn->role &= ~(CSIO_RNFR_INITIATOR | CSIO_RNFR_TARGET);
+ fc_remote_port_delete(rport);
+ ln->num_reg_rnodes--;
+
+ if (ln->n_scsi_tgts)
+ ln->n_scsi_tgts--;
+
+ if (ln->last_scan_ntgts)
+ ln->last_scan_ntgts--;
+
+ csio_ln_dbg(ln, "Remote port x%x un-registered\n", rn->nport_id);
+}
+
+/*
+ * csio_lnode_async_event - Async events from local port.
+ * @ln: lnode representing local port.
+ *
+ * Async events from local node that FC transport/SCSI ML
+ * should be made aware of (Eg: RSCN).
+ */
+void
+csio_lnode_async_event(struct csio_lnode *ln, enum csio_ln_fc_evt fc_evt)
+{
+ switch (fc_evt) {
+ case CSIO_LN_FC_RSCN:
+ /* Get payload of rscn from ln */
+ /* For each RSCN entry */
+ /*
+ * fc_host_post_event(shost,
+ * fc_get_event_number(),
+ * FCH_EVT_RSCN,
+ * rscn_entry);
+ */
+ break;
+ case CSIO_LN_FC_LINKUP:
+ /* send fc_host_post_event */
+ /* set vport state */
+ if (csio_is_npiv_ln(ln))
+ csio_vport_set_state(ln);
+
+ break;
+ case CSIO_LN_FC_LINKDOWN:
+ /* send fc_host_post_event */
+ /* set vport state */
+ if (csio_is_npiv_ln(ln))
+ csio_vport_set_state(ln);
+
+ break;
+ case CSIO_LN_FC_ATTRIB_UPDATE:
+ csio_fchost_attr_init(ln);
+ break;
+ default:
+ break;
+ }
+}
+
+/*
+ * csio_fchost_attr_init - Initialize FC transport attributes
+ * @ln: Lnode.
+ *
+ */
+void
+csio_fchost_attr_init(struct csio_lnode *ln)
+{
+ struct Scsi_Host *shost = csio_ln_to_shost(ln);
+
+ fc_host_node_name(shost) = wwn_to_u64(csio_ln_wwnn(ln));
+ fc_host_port_name(shost) = wwn_to_u64(csio_ln_wwpn(ln));
+
+ fc_host_supported_classes(shost) = FC_COS_CLASS3;
+ fc_host_max_npiv_vports(shost) =
+ (csio_lnode_to_hw(ln))->fres_info.max_vnps;
+ fc_host_supported_speeds(shost) = FC_PORTSPEED_10GBIT |
+ FC_PORTSPEED_1GBIT;
+
+ fc_host_maxframe_size(shost) = ln->ln_sparm.csp.sp_bb_data;
+ memset(fc_host_supported_fc4s(shost), 0,
+ sizeof(fc_host_supported_fc4s(shost)));
+ fc_host_supported_fc4s(shost)[7] = 1;
+
+ memset(fc_host_active_fc4s(shost), 0,
+ sizeof(fc_host_active_fc4s(shost)));
+ fc_host_active_fc4s(shost)[7] = 1;
+}
+
+/*
+ * csio_get_host_port_id - sysfs entries for nport_id is
+ * populated/cached from this function
+ */
+static void
+csio_get_host_port_id(struct Scsi_Host *shost)
+{
+ struct csio_lnode *ln = shost_priv(shost);
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+
+ spin_lock_irq(&hw->lock);
+ fc_host_port_id(shost) = ln->nport_id;
+ spin_unlock_irq(&hw->lock);
+}
+
+/*
+ * csio_get_port_type - Return FC local port type.
+ * @shost: scsi host.
+ *
+ */
+static void
+csio_get_host_port_type(struct Scsi_Host *shost)
+{
+ struct csio_lnode *ln = shost_priv(shost);
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+
+ spin_lock_irq(&hw->lock);
+ if (csio_is_npiv_ln(ln))
+ fc_host_port_type(shost) = FC_PORTTYPE_NPIV;
+ else
+ fc_host_port_type(shost) = FC_PORTTYPE_NPORT;
+ spin_unlock_irq(&hw->lock);
+}
+
+/*
+ * csio_get_port_state - Return FC local port state.
+ * @shost: scsi host.
+ *
+ */
+static void
+csio_get_host_port_state(struct Scsi_Host *shost)
+{
+ struct csio_lnode *ln = shost_priv(shost);
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+ char state[16];
+
+ spin_lock_irq(&hw->lock);
+
+ csio_lnode_state_to_str(ln, state);
+ if (!strcmp(state, "READY"))
+ fc_host_port_state(shost) = FC_PORTSTATE_ONLINE;
+ else if (!strcmp(state, "OFFLINE"))
+ fc_host_port_state(shost) = FC_PORTSTATE_LINKDOWN;
+ else
+ fc_host_port_state(shost) = FC_PORTSTATE_UNKNOWN;
+
+ spin_unlock_irq(&hw->lock);
+}
+
+/*
+ * csio_get_host_speed - Return link speed to FC transport.
+ * @shost: scsi host.
+ *
+ */
+static void
+csio_get_host_speed(struct Scsi_Host *shost)
+{
+ struct csio_lnode *ln = shost_priv(shost);
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+
+ spin_lock_irq(&hw->lock);
+ switch (hw->pport[ln->portid].link_speed) {
+ case FW_PORT_CAP_SPEED_1G:
+ fc_host_speed(shost) = FC_PORTSPEED_1GBIT;
+ break;
+ case FW_PORT_CAP_SPEED_10G:
+ fc_host_speed(shost) = FC_PORTSPEED_10GBIT;
+ break;
+ default:
+ fc_host_speed(shost) = FC_PORTSPEED_UNKNOWN;
+ break;
+ }
+ spin_unlock_irq(&hw->lock);
+}
+
+/*
+ * csio_get_host_fabric_name - Return fabric name
+ * @shost: scsi host.
+ *
+ */
+static void
+csio_get_host_fabric_name(struct Scsi_Host *shost)
+{
+ struct csio_lnode *ln = shost_priv(shost);
+ struct csio_rnode *rn = NULL;
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+
+ spin_lock_irq(&hw->lock);
+ rn = csio_rnode_lookup_portid(ln, FC_FID_FLOGI);
+ if (rn)
+ fc_host_fabric_name(shost) = wwn_to_u64(csio_rn_wwnn(rn));
+ else
+ fc_host_fabric_name(shost) = 0;
+ spin_unlock_irq(&hw->lock);
+}
+
+/*
+ * csio_get_host_speed - Return FC transport statistics.
+ * @ln: Lnode.
+ *
+ */
+static struct fc_host_statistics *
+csio_get_stats(struct Scsi_Host *shost)
+{
+ struct csio_lnode *ln = shost_priv(shost);
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+ struct fc_host_statistics *fhs = &ln->fch_stats;
+ struct fw_fcoe_port_stats fcoe_port_stats;
+ uint64_t seconds;
+
+ memset(&fcoe_port_stats, 0, sizeof(struct fw_fcoe_port_stats));
+ csio_get_phy_port_stats(hw, ln->portid, &fcoe_port_stats);
+
+ fhs->tx_frames += (fcoe_port_stats.tx_bcast_frames +
+ fcoe_port_stats.tx_mcast_frames +
+ fcoe_port_stats.tx_ucast_frames +
+ fcoe_port_stats.tx_offload_frames);
+ fhs->tx_words += (fcoe_port_stats.tx_bcast_bytes +
+ fcoe_port_stats.tx_mcast_bytes +
+ fcoe_port_stats.tx_ucast_bytes +
+ fcoe_port_stats.tx_offload_bytes) /
+ CSIO_WORD_TO_BYTE;
+ fhs->rx_frames += (fcoe_port_stats.rx_bcast_frames +
+ fcoe_port_stats.rx_mcast_frames +
+ fcoe_port_stats.rx_ucast_frames);
+ fhs->rx_words += (fcoe_port_stats.rx_bcast_bytes +
+ fcoe_port_stats.rx_mcast_bytes +
+ fcoe_port_stats.rx_ucast_bytes) /
+ CSIO_WORD_TO_BYTE;
+ fhs->error_frames += fcoe_port_stats.rx_err_frames;
+ fhs->fcp_input_requests += ln->stats.n_input_requests;
+ fhs->fcp_output_requests += ln->stats.n_output_requests;
+ fhs->fcp_control_requests += ln->stats.n_control_requests;
+ fhs->fcp_input_megabytes += ln->stats.n_input_bytes >> 20;
+ fhs->fcp_output_megabytes += ln->stats.n_output_bytes >> 20;
+ fhs->link_failure_count = ln->stats.n_link_down;
+ /* Reset stats for the device */
+ seconds = jiffies_to_msecs(jiffies);
+ fhs->seconds_since_last_reset =
+ (seconds - hw->stats.n_reset_start) / 1000;
+ return fhs;
+}
+
+/*
+ * csio_set_rport_loss_tmo - Set the rport dev loss timeout
+ * @rport: fc rport.
+ * @timeout: new value for dev loss tmo.
+ *
+ * If timeout is non zero set the dev_loss_tmo to timeout, else set
+ * dev_loss_tmo to one.
+ */
+static void
+csio_set_rport_loss_tmo(struct fc_rport *rport, uint32_t timeout)
+{
+ if (timeout)
+ rport->dev_loss_tmo = timeout;
+ else
+ rport->dev_loss_tmo = 1;
+}
+
+static void
+csio_vport_set_state(struct csio_lnode *ln)
+{
+ struct fc_vport *fc_vport = ln->fc_vport;
+ struct csio_lnode *pln = ln->pln;
+ char state[16];
+
+ /* Set fc vport state based on phyiscal lnode */
+ csio_lnode_state_to_str(pln, state);
+ if (strcmp(state, "READY")) {
+ fc_vport_set_state(fc_vport, FC_VPORT_LINKDOWN);
+ return;
+ }
+
+ if (!(pln->flags & CSIO_LNF_NPIVSUPP)) {
+ fc_vport_set_state(fc_vport, FC_VPORT_NO_FABRIC_SUPP);
+ return;
+ }
+
+ /* Set fc vport state based on virtual lnode */
+ csio_lnode_state_to_str(ln, state);
+ if (strcmp(state, "READY")) {
+ fc_vport_set_state(fc_vport, FC_VPORT_LINKDOWN);
+ return;
+ }
+ fc_vport_set_state(fc_vport, FC_VPORT_ACTIVE);
+}
+
+static int
+csio_fcoe_alloc_vnp(struct csio_hw *hw, struct csio_lnode *ln)
+{
+ struct csio_lnode *pln;
+ struct csio_mb *mbp;
+ struct fw_fcoe_vnp_cmd *rsp;
+ int ret;
+ int retry = 0;
+
+ /* Issue VNP cmd to alloc vport */
+ /* Allocate Mbox request */
+ spin_lock_irq(&hw->lock);
+ mbp = mempool_alloc(hw->mb_mempool, GFP_ATOMIC);
+ if (!mbp) {
+ CSIO_INC_STATS(hw, n_err_nomem);
+ spin_unlock_irq(&hw->lock);
+ return -ENOMEM;
+ }
+
+ pln = ln->pln;
+ ln->fcf_flowid = pln->fcf_flowid;
+ ln->portid = pln->portid;
+
+ csio_fcoe_vnp_alloc_init_mb(ln, mbp, CSIO_MB_DEFAULT_TMO,
+ pln->fcf_flowid, pln->vnp_flowid, 0,
+ csio_ln_wwnn(ln), csio_ln_wwpn(ln),
+ NULL);
+
+ for (retry = 0; retry < 3; retry++) {
+ /* FW is expected to complete vnp cmd in immediate mode
+ * without much delay.
+ * Otherwise, there will be increase in IO latency since HW
+ * lock is held till completion of vnp mbox cmd.
+ */
+ ret = csio_mb_issue(hw, mbp);
+ if (ret != -EBUSY)
+ break;
+
+ /* Retry if mbox returns busy */
+ spin_unlock_irq(&hw->lock);
+ msleep(2000);
+ spin_lock_irq(&hw->lock);
+ }
+
+ if (ret != 0) {
+ csio_ln_err(ln, "Failed to issue mbox FCoE VNP command\n");
+ mempool_free(mbp, hw->mb_mempool);
+ spin_unlock_irq(&hw->lock);
+ return ret;
+ }
+
+ /* Process Mbox response of VNP command */
+ rsp = (struct fw_fcoe_vnp_cmd *)(mbp->mb);
+ if (FW_CMD_RETVAL_GET(ntohl(rsp->alloc_to_len16)) != FW_SUCCESS) {
+ csio_ln_err(ln, "FCOE VNP ALLOC cmd returned 0x%x!\n",
+ FW_CMD_RETVAL_GET(ntohl(rsp->alloc_to_len16)));
+ mempool_free(mbp, hw->mb_mempool);
+ spin_unlock_irq(&hw->lock);
+ return -ENOMEM;
+ }
+
+ ln->vnp_flowid = FW_FCOE_VNP_CMD_VNPI_GET(
+ ntohl(rsp->gen_wwn_to_vnpi));
+ memcpy(csio_ln_wwnn(ln), rsp->vnport_wwnn, 8);
+ memcpy(csio_ln_wwpn(ln), rsp->vnport_wwpn, 8);
+
+ csio_ln_dbg(ln, "FCOE VNPI: 0x%x\n", ln->vnp_flowid);
+ csio_ln_dbg(ln, "\tWWNN: %x%x%x%x%x%x%x%x\n",
+ ln->ln_sparm.wwnn[0], ln->ln_sparm.wwnn[1],
+ ln->ln_sparm.wwnn[2], ln->ln_sparm.wwnn[3],
+ ln->ln_sparm.wwnn[4], ln->ln_sparm.wwnn[5],
+ ln->ln_sparm.wwnn[6], ln->ln_sparm.wwnn[7]);
+ csio_ln_dbg(ln, "\tWWPN: %x%x%x%x%x%x%x%x\n",
+ ln->ln_sparm.wwpn[0], ln->ln_sparm.wwpn[1],
+ ln->ln_sparm.wwpn[2], ln->ln_sparm.wwpn[3],
+ ln->ln_sparm.wwpn[4], ln->ln_sparm.wwpn[5],
+ ln->ln_sparm.wwpn[6], ln->ln_sparm.wwpn[7]);
+
+ mempool_free(mbp, hw->mb_mempool);
+ spin_unlock_irq(&hw->lock);
+
+ return 0;
+}
+
+static int
+csio_fcoe_free_vnp(struct csio_hw *hw, struct csio_lnode *ln)
+{
+ struct csio_lnode *pln;
+ struct csio_mb *mbp;
+ struct fw_fcoe_vnp_cmd *rsp;
+ int ret;
+ int retry = 0;
+
+ /* Issue VNP cmd to free vport */
+ /* Allocate Mbox request */
+
+ spin_lock_irq(&hw->lock);
+ mbp = mempool_alloc(hw->mb_mempool, GFP_ATOMIC);
+ if (!mbp) {
+ CSIO_INC_STATS(hw, n_err_nomem);
+ spin_unlock_irq(&hw->lock);
+ return -ENOMEM;
+ }
+
+ pln = ln->pln;
+
+ csio_fcoe_vnp_free_init_mb(ln, mbp, CSIO_MB_DEFAULT_TMO,
+ ln->fcf_flowid, ln->vnp_flowid,
+ NULL);
+
+ for (retry = 0; retry < 3; retry++) {
+ ret = csio_mb_issue(hw, mbp);
+ if (ret != -EBUSY)
+ break;
+
+ /* Retry if mbox returns busy */
+ spin_unlock_irq(&hw->lock);
+ msleep(2000);
+ spin_lock_irq(&hw->lock);
+ }
+
+ if (ret) {
+ csio_ln_err(ln, "Failed to issue mbox FCoE VNP command\n");
+ mempool_free(mbp, hw->mb_mempool);
+ spin_unlock_irq(&hw->lock);
+ return -EINVAL;
+ }
+
+ /* Process Mbox response of VNP command */
+ rsp = (struct fw_fcoe_vnp_cmd *)(mbp->mb);
+ if (FW_CMD_RETVAL_GET(ntohl(rsp->alloc_to_len16)) != FW_SUCCESS) {
+ csio_ln_err(ln, "FCOE VNP FREE cmd returned 0x%x!\n",
+ FW_CMD_RETVAL_GET(ntohl(rsp->alloc_to_len16)));
+ mempool_free(mbp, hw->mb_mempool);
+ spin_unlock_irq(&hw->lock);
+ return -ENOMEM;
+ }
+
+ mempool_free(mbp, hw->mb_mempool);
+ spin_unlock_irq(&hw->lock);
+
+ return 0;
+}
+
+static int
+csio_vport_create(struct fc_vport *fc_vport, bool disable)
+{
+ struct Scsi_Host *shost = fc_vport->shost;
+ struct csio_lnode *pln = shost_priv(shost);
+ struct csio_lnode *ln = NULL;
+ struct csio_hw *hw = csio_lnode_to_hw(pln);
+ uint8_t wwn[8];
+ int ret = -1;
+
+ ln = csio_shost_init(hw, &fc_vport->dev, false, pln);
+ if (!ln)
+ goto error;
+
+ if (fc_vport->node_name != 0) {
+ u64_to_wwn(fc_vport->node_name, wwn);
+
+ if (!CSIO_VALID_WWN(wwn)) {
+ csio_ln_err(ln,
+ "vport create failed. Invalid wwnn\n");
+ goto error;
+ }
+ memcpy(csio_ln_wwnn(ln), wwn, 8);
+ }
+
+ if (fc_vport->port_name != 0) {
+ u64_to_wwn(fc_vport->port_name, wwn);
+
+ if (!CSIO_VALID_WWN(wwn)) {
+ csio_ln_err(ln,
+ "vport create failed. Invalid wwpn\n");
+ goto error;
+ }
+
+ if (csio_lnode_lookup_by_wwpn(hw, wwn)) {
+ csio_ln_err(ln,
+ "vport create failed. wwpn already exists\n");
+ goto error;
+ }
+ memcpy(csio_ln_wwpn(ln), wwn, 8);
+ }
+
+ fc_vport_set_state(fc_vport, FC_VPORT_INITIALIZING);
+
+ if (csio_fcoe_alloc_vnp(hw, ln))
+ goto error;
+
+ *(struct csio_lnode **)fc_vport->dd_data = ln;
+ ln->fc_vport = fc_vport;
+ if (!fc_vport->node_name)
+ fc_vport->node_name = wwn_to_u64(csio_ln_wwnn(ln));
+ if (!fc_vport->port_name)
+ fc_vport->port_name = wwn_to_u64(csio_ln_wwpn(ln));
+ csio_fchost_attr_init(ln);
+ return 0;
+error:
+ if (ln)
+ csio_shost_exit(ln);
+
+ return ret;
+}
+
+static int
+csio_vport_delete(struct fc_vport *fc_vport)
+{
+ struct csio_lnode *ln = *(struct csio_lnode **)fc_vport->dd_data;
+ struct Scsi_Host *shost = csio_ln_to_shost(ln);
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+
+ spin_lock_irq(&hw->lock);
+ if (csio_is_hw_removing(hw)) {
+ spin_unlock_irq(&hw->lock);
+ csio_shost_exit(ln);
+ return 0;
+ }
+ spin_unlock_irq(&hw->lock);
+
+ /* Quiesce ios and send remove event to lnode */
+ scsi_block_requests(shost);
+ spin_lock_irq(&hw->lock);
+ csio_scsim_cleanup_io_lnode(csio_hw_to_scsim(hw), ln);
+ csio_lnode_close(ln);
+ spin_unlock_irq(&hw->lock);
+ scsi_unblock_requests(shost);
+
+ /* Free vnp */
+ if (fc_vport->vport_state != FC_VPORT_DISABLED)
+ csio_fcoe_free_vnp(hw, ln);
+ csio_ln_err(ln, "vport deleted\n");
+ csio_shost_exit(ln);
+ return 0;
+}
+
+static int
+csio_vport_disable(struct fc_vport *fc_vport, bool disable)
+{
+ struct csio_lnode *ln = *(struct csio_lnode **)fc_vport->dd_data;
+ struct Scsi_Host *shost = csio_ln_to_shost(ln);
+ struct csio_hw *hw = csio_lnode_to_hw(ln);
+
+ /* disable vport */
+ if (disable) {
+ /* Quiesce ios and send stop event to lnode */
+ scsi_block_requests(shost);
+ spin_lock_irq(&hw->lock);
+ csio_scsim_cleanup_io_lnode(csio_hw_to_scsim(hw), ln);
+ csio_lnode_stop(ln);
+ spin_unlock_irq(&hw->lock);
+ scsi_unblock_requests(shost);
+
+ /* Free vnp */
+ csio_fcoe_free_vnp(hw, ln);
+ fc_vport_set_state(fc_vport, FC_VPORT_DISABLED);
+ csio_ln_err(ln, "vport disabled\n");
+ return 0;
+ } else {
+ /* enable vport */
+ fc_vport_set_state(fc_vport, FC_VPORT_INITIALIZING);
+ if (csio_fcoe_alloc_vnp(hw, ln)) {
+ csio_ln_err(ln, "vport enabled failed.\n");
+ return -1;
+ }
+ csio_ln_err(ln, "vport enabled\n");
+ return 0;
+ }
+}
+
+static void
+csio_dev_loss_tmo_callbk(struct fc_rport *rport)
+{
+ struct csio_rnode *rn;
+ struct csio_hw *hw;
+ struct csio_lnode *ln;
+
+ rn = *((struct csio_rnode **)rport->dd_data);
+ ln = csio_rnode_to_lnode(rn);
+ hw = csio_lnode_to_hw(ln);
+
+ spin_lock_irq(&hw->lock);
+
+ /* return if driver is being removed or same rnode comes back online */
+ if (csio_is_hw_removing(hw) || csio_is_rnode_ready(rn)) {
+ spin_unlock_irq(&hw->lock);
+ return;
+ }
+
+ csio_ln_dbg(ln, "devloss timeout on rnode:%p portid:x%x flowid:x%x\n",
+ rn, rn->nport_id, csio_rn_flowid(rn));
+
+ CSIO_INC_STATS(ln, n_dev_loss_tmo);
+
+ /*
+ * enqueue devloss event to event worker thread to serialize all
+ * rnode events.
+ */
+ if (csio_enqueue_evt(hw, CSIO_EVT_DEV_LOSS, &rn, sizeof(rn))) {
+ CSIO_INC_STATS(hw, n_evt_drop);
+ spin_unlock_irq(&hw->lock);
+ return;
+ }
+
+ if (!(hw->flags & CSIO_HWF_FWEVT_PENDING)) {
+ hw->flags |= CSIO_HWF_FWEVT_PENDING;
+ spin_unlock_irq(&hw->lock);
+ schedule_work(&hw->evtq_work);
+ return;
+ }
+
+ spin_unlock_irq(&hw->lock);
+}
+
+/* FC transport functions template - Physical port */
+struct fc_function_template csio_fc_transport_funcs = {
+ .show_host_node_name = 1,
+ .show_host_port_name = 1,
+ .show_host_supported_classes = 1,
+ .show_host_supported_fc4s = 1,
+ .show_host_maxframe_size = 1,
+
+ .get_host_port_id = csio_get_host_port_id,
+ .show_host_port_id = 1,
+
+ .get_host_port_type = csio_get_host_port_type,
+ .show_host_port_type = 1,
+
+ .get_host_port_state = csio_get_host_port_state,
+ .show_host_port_state = 1,
+
+ .show_host_active_fc4s = 1,
+ .get_host_speed = csio_get_host_speed,
+ .show_host_speed = 1,
+ .get_host_fabric_name = csio_get_host_fabric_name,
+ .show_host_fabric_name = 1,
+
+ .get_fc_host_stats = csio_get_stats,
+
+ .dd_fcrport_size = sizeof(struct csio_rnode *),
+ .show_rport_maxframe_size = 1,
+ .show_rport_supported_classes = 1,
+
+ .set_rport_dev_loss_tmo = csio_set_rport_loss_tmo,
+ .show_rport_dev_loss_tmo = 1,
+
+ .show_starget_port_id = 1,
+ .show_starget_node_name = 1,
+ .show_starget_port_name = 1,
+
+ .dev_loss_tmo_callbk = csio_dev_loss_tmo_callbk,
+ .dd_fcvport_size = sizeof(struct csio_lnode *),
+
+ .vport_create = csio_vport_create,
+ .vport_disable = csio_vport_disable,
+ .vport_delete = csio_vport_delete,
+};
+
+/* FC transport functions template - Virtual port */
+struct fc_function_template csio_fc_transport_vport_funcs = {
+ .show_host_node_name = 1,
+ .show_host_port_name = 1,
+ .show_host_supported_classes = 1,
+ .show_host_supported_fc4s = 1,
+ .show_host_maxframe_size = 1,
+
+ .get_host_port_id = csio_get_host_port_id,
+ .show_host_port_id = 1,
+
+ .get_host_port_type = csio_get_host_port_type,
+ .show_host_port_type = 1,
+
+ .get_host_port_state = csio_get_host_port_state,
+ .show_host_port_state = 1,
+ .show_host_active_fc4s = 1,
+
+ .get_host_speed = csio_get_host_speed,
+ .show_host_speed = 1,
+
+ .get_host_fabric_name = csio_get_host_fabric_name,
+ .show_host_fabric_name = 1,
+
+ .get_fc_host_stats = csio_get_stats,
+
+ .dd_fcrport_size = sizeof(struct csio_rnode *),
+ .show_rport_maxframe_size = 1,
+ .show_rport_supported_classes = 1,
+
+ .set_rport_dev_loss_tmo = csio_set_rport_loss_tmo,
+ .show_rport_dev_loss_tmo = 1,
+
+ .show_starget_port_id = 1,
+ .show_starget_node_name = 1,
+ .show_starget_port_name = 1,
+
+ .dev_loss_tmo_callbk = csio_dev_loss_tmo_callbk,
+
+};
diff --git a/drivers/scsi/csiostor/csio_mb.c b/drivers/scsi/csiostor/csio_mb.c
new file mode 100644
index 0000000..100afd1
--- /dev/null
+++ b/drivers/scsi/csiostor/csio_mb.c
@@ -0,0 +1,1769 @@
+/*
+ * This file is part of the Chelsio FCoE driver for Linux.
+ *
+ * Copyright (c) 2008-2012 Chelsio Communications, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include <linux/delay.h>
+#include <linux/jiffies.h>
+#include <linux/string.h>
+#include <scsi/scsi_transport_fc.h>
+
+#include "csio_hw.h"
+#include "csio_lnode.h"
+#include "csio_rnode.h"
+#include "csio_mb.h"
+#include "csio_wr.h"
+
+#define csio_mb_is_host_owner(__owner) ((__owner) == CSIO_MBOWNER_PL)
+
+/* MB Command/Response Helpers */
+/*
+ * csio_mb_fw_retval - FW return value from a mailbox response.
+ * @mbp: Mailbox structure
+ *
+ */
+enum fw_retval
+csio_mb_fw_retval(struct csio_mb *mbp)
+{
+ struct fw_cmd_hdr *hdr;
+
+ hdr = (struct fw_cmd_hdr *)(mbp->mb);
+
+ return FW_CMD_RETVAL_GET(ntohl(hdr->lo));
+}
+
+/*
+ * csio_mb_hello - FW HELLO command helper
+ * @hw: The HW structure
+ * @mbp: Mailbox structure
+ * @m_mbox: Master mailbox number, if any.
+ * @a_mbox: Mailbox number for asycn notifications.
+ * @master: Device mastership.
+ * @cbfn: Callback, if any.
+ *
+ */
+void
+csio_mb_hello(struct csio_hw *hw, struct csio_mb *mbp, uint32_t tmo,
+ uint32_t m_mbox, uint32_t a_mbox, enum csio_dev_master master,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_hello_cmd *cmdp = (struct fw_hello_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, tmo, hw, cbfn, 1);
+
+ cmdp->op_to_write = htonl(FW_CMD_OP(FW_HELLO_CMD) |
+ FW_CMD_REQUEST | FW_CMD_WRITE);
+ cmdp->retval_len16 = htonl(FW_CMD_LEN16(sizeof(*cmdp) / 16));
+ cmdp->err_to_clearinit = htonl(
+ FW_HELLO_CMD_MASTERDIS(master == CSIO_MASTER_CANT) |
+ FW_HELLO_CMD_MASTERFORCE(master == CSIO_MASTER_MUST) |
+ FW_HELLO_CMD_MBMASTER(master == CSIO_MASTER_MUST ?
+ m_mbox : FW_HELLO_CMD_MBMASTER_MASK) |
+ FW_HELLO_CMD_MBASYNCNOT(a_mbox) |
+ FW_HELLO_CMD_STAGE(FW_HELLO_CMD_STAGE_OS) |
+ FW_HELLO_CMD_CLEARINIT);
+
+}
+
+/*
+ * csio_mb_process_hello_rsp - FW HELLO response processing helper
+ * @hw: The HW structure
+ * @mbp: Mailbox structure
+ * @retval: Mailbox return value from Firmware
+ * @state: State that the function is in.
+ * @mpfn: Master pfn
+ *
+ */
+void
+csio_mb_process_hello_rsp(struct csio_hw *hw, struct csio_mb *mbp,
+ enum fw_retval *retval, enum csio_dev_state *state,
+ uint8_t *mpfn)
+{
+ struct fw_hello_cmd *rsp = (struct fw_hello_cmd *)(mbp->mb);
+ uint32_t value;
+
+ *retval = FW_CMD_RETVAL_GET(ntohl(rsp->retval_len16));
+
+ if (*retval == FW_SUCCESS) {
+ hw->fwrev = ntohl(rsp->fwrev);
+
+ value = ntohl(rsp->err_to_clearinit);
+ *mpfn = FW_HELLO_CMD_MBMASTER_GET(value);
+
+ if (value & FW_HELLO_CMD_INIT)
+ *state = CSIO_DEV_STATE_INIT;
+ else if (value & FW_HELLO_CMD_ERR)
+ *state = CSIO_DEV_STATE_ERR;
+ else
+ *state = CSIO_DEV_STATE_UNINIT;
+ }
+}
+
+/*
+ * csio_mb_bye - FW BYE command helper
+ * @hw: The HW structure
+ * @mbp: Mailbox structure
+ * @cbfn: Callback, if any.
+ *
+ */
+void
+csio_mb_bye(struct csio_hw *hw, struct csio_mb *mbp, uint32_t tmo,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_bye_cmd *cmdp = (struct fw_bye_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, tmo, hw, cbfn, 1);
+
+ cmdp->op_to_write = htonl(FW_CMD_OP(FW_BYE_CMD) |
+ FW_CMD_REQUEST | FW_CMD_WRITE);
+ cmdp->retval_len16 = htonl(FW_CMD_LEN16(sizeof(*cmdp) / 16));
+
+}
+
+/*
+ * csio_mb_reset - FW RESET command helper
+ * @hw: The HW structure
+ * @mbp: Mailbox structure
+ * @reset: Type of reset.
+ * @cbfn: Callback, if any.
+ *
+ */
+void
+csio_mb_reset(struct csio_hw *hw, struct csio_mb *mbp, uint32_t tmo,
+ int reset, int halt,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_reset_cmd *cmdp = (struct fw_reset_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, tmo, hw, cbfn, 1);
+
+ cmdp->op_to_write = htonl(FW_CMD_OP(FW_RESET_CMD) |
+ FW_CMD_REQUEST | FW_CMD_WRITE);
+ cmdp->retval_len16 = htonl(FW_CMD_LEN16(sizeof(*cmdp) / 16));
+ cmdp->val = htonl(reset);
+ cmdp->halt_pkd = htonl(halt);
+
+}
+
+/*
+ * csio_mb_params - FW PARAMS command helper
+ * @hw: The HW structure
+ * @mbp: Mailbox structure
+ * @tmo: Command timeout.
+ * @pf: PF number.
+ * @vf: VF number.
+ * @nparams: Number of paramters
+ * @params: Parameter mnemonic array.
+ * @val: Parameter value array.
+ * @wr: Write/Read PARAMS.
+ * @cbfn: Callback, if any.
+ *
+ */
+void
+csio_mb_params(struct csio_hw *hw, struct csio_mb *mbp, uint32_t tmo,
+ unsigned int pf, unsigned int vf, unsigned int nparams,
+ const u32 *params, u32 *val, bool wr,
+ void (*cbfn)(struct csio_hw *, struct csio_mb *))
+{
+ uint32_t i;
+ uint32_t temp_params = 0, temp_val = 0;
+ struct fw_params_cmd *cmdp = (struct fw_params_cmd *)(mbp->mb);
+ __be32 *p = &cmdp->param[0].mnem;
+
+ CSIO_INIT_MBP(mbp, cmdp, tmo, hw, cbfn, 1);
+
+ cmdp->op_to_vfn = htonl(FW_CMD_OP(FW_PARAMS_CMD) |
+ FW_CMD_REQUEST |
+ (wr ? FW_CMD_WRITE : FW_CMD_READ) |
+ FW_PARAMS_CMD_PFN(pf) |
+ FW_PARAMS_CMD_VFN(vf));
+ cmdp->retval_len16 = htonl(FW_CMD_LEN16(sizeof(*cmdp) / 16));
+
+ /* Write Params */
+ if (wr) {
+ while (nparams--) {
+ temp_params = *params++;
+ temp_val = *val++;
+
+ *p++ = htonl(temp_params);
+ *p++ = htonl(temp_val);
+ }
+ } else {
+ for (i = 0; i < nparams; i++, p += 2) {
+ temp_params = *params++;
+ *p = htonl(temp_params);
+ }
+ }
+
+}
+
+/*
+ * csio_mb_process_read_params_rsp - FW PARAMS response processing helper
+ * @hw: The HW structure
+ * @mbp: Mailbox structure
+ * @retval: Mailbox return value from Firmware
+ * @nparams: Number of parameters
+ * @val: Parameter value array.
+ *
+ */
+void
+csio_mb_process_read_params_rsp(struct csio_hw *hw, struct csio_mb *mbp,
+ enum fw_retval *retval, unsigned int nparams,
+ u32 *val)
+{
+ struct fw_params_cmd *rsp = (struct fw_params_cmd *)(mbp->mb);
+ uint32_t i;
+ __be32 *p = &rsp->param[0].val;
+
+ *retval = FW_CMD_RETVAL_GET(ntohl(rsp->retval_len16));
+
+ if (*retval == FW_SUCCESS)
+ for (i = 0; i < nparams; i++, p += 2)
+ *val++ = ntohl(*p);
+}
+
+/*
+ * csio_mb_ldst - FW LDST command
+ * @hw: The HW structure
+ * @mbp: Mailbox structure
+ * @tmo: timeout
+ * @reg: register
+ *
+ */
+void
+csio_mb_ldst(struct csio_hw *hw, struct csio_mb *mbp, uint32_t tmo, int reg)
+{
+ struct fw_ldst_cmd *ldst_cmd = (struct fw_ldst_cmd *)(mbp->mb);
+ CSIO_INIT_MBP(mbp, ldst_cmd, tmo, hw, NULL, 1);
+
+ /*
+ * Construct and send the Firmware LDST Command to retrieve the
+ * specified PCI-E Configuration Space register.
+ */
+ ldst_cmd->op_to_addrspace =
+ htonl(FW_CMD_OP(FW_LDST_CMD) |
+ FW_CMD_REQUEST |
+ FW_CMD_READ |
+ FW_LDST_CMD_ADDRSPACE(FW_LDST_ADDRSPC_FUNC_PCIE));
+ ldst_cmd->cycles_to_len16 = htonl(FW_LEN16(struct fw_ldst_cmd));
+ ldst_cmd->u.pcie.select_naccess = FW_LDST_CMD_NACCESS(1);
+ ldst_cmd->u.pcie.ctrl_to_fn =
+ (FW_LDST_CMD_LC | FW_LDST_CMD_FN(hw->pfn));
+ ldst_cmd->u.pcie.r = (uint8_t)reg;
+}
+
+/*
+ *
+ * csio_mb_caps_config - FW Read/Write Capabilities command helper
+ * @hw: The HW structure
+ * @mbp: Mailbox structure
+ * @wr: Write if 1, Read if 0
+ * @init: Turn on initiator mode.
+ * @tgt: Turn on target mode.
+ * @cofld: If 1, Control Offload for FCoE
+ * @cbfn: Callback, if any.
+ *
+ * This helper assumes that cmdp has MB payload from a previous CAPS
+ * read command.
+ */
+void
+csio_mb_caps_config(struct csio_hw *hw, struct csio_mb *mbp, uint32_t tmo,
+ bool wr, bool init, bool tgt, bool cofld,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_caps_config_cmd *cmdp =
+ (struct fw_caps_config_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, tmo, hw, cbfn, wr ? 0 : 1);
+
+ cmdp->op_to_write = htonl(FW_CMD_OP(FW_CAPS_CONFIG_CMD) |
+ FW_CMD_REQUEST |
+ (wr ? FW_CMD_WRITE : FW_CMD_READ));
+ cmdp->cfvalid_to_len16 = htonl(FW_CMD_LEN16(sizeof(*cmdp) / 16));
+
+ /* Read config */
+ if (!wr)
+ return;
+
+ /* Write config */
+ cmdp->fcoecaps = 0;
+
+ if (cofld)
+ cmdp->fcoecaps |= htons(FW_CAPS_CONFIG_FCOE_CTRL_OFLD);
+ if (init)
+ cmdp->fcoecaps |= htons(FW_CAPS_CONFIG_FCOE_INITIATOR);
+ if (tgt)
+ cmdp->fcoecaps |= htons(FW_CAPS_CONFIG_FCOE_TARGET);
+}
+
+void
+csio_rss_glb_config(struct csio_hw *hw, struct csio_mb *mbp,
+ uint32_t tmo, uint8_t mode, unsigned int flags,
+ void (*cbfn)(struct csio_hw *, struct csio_mb *))
+{
+ struct fw_rss_glb_config_cmd *cmdp =
+ (struct fw_rss_glb_config_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, tmo, hw, cbfn, 1);
+
+ cmdp->op_to_write = htonl(FW_CMD_OP(FW_RSS_GLB_CONFIG_CMD) |
+ FW_CMD_REQUEST | FW_CMD_WRITE);
+ cmdp->retval_len16 = htonl(FW_CMD_LEN16(sizeof(*cmdp) / 16));
+
+ if (mode == FW_RSS_GLB_CONFIG_CMD_MODE_MANUAL) {
+ cmdp->u.manual.mode_pkd =
+ htonl(FW_RSS_GLB_CONFIG_CMD_MODE(mode));
+ } else if (mode == FW_RSS_GLB_CONFIG_CMD_MODE_BASICVIRTUAL) {
+ cmdp->u.basicvirtual.mode_pkd =
+ htonl(FW_RSS_GLB_CONFIG_CMD_MODE(mode));
+ cmdp->u.basicvirtual.synmapen_to_hashtoeplitz = htonl(flags);
+ }
+}
+
+
+/*
+ * csio_mb_pfvf - FW Write PF/VF capabilities command helper.
+ * @hw: The HW structure
+ * @mbp: Mailbox structure
+ * @pf:
+ * @vf:
+ * @txq:
+ * @txq_eht_ctrl:
+ * @rxqi:
+ * @rxq:
+ * @tc:
+ * @vi:
+ * @pmask:
+ * @rcaps:
+ * @wxcaps:
+ * @cbfn: Callback, if any.
+ *
+ */
+void
+csio_mb_pfvf(struct csio_hw *hw, struct csio_mb *mbp, uint32_t tmo,
+ unsigned int pf, unsigned int vf, unsigned int txq,
+ unsigned int txq_eth_ctrl, unsigned int rxqi,
+ unsigned int rxq, unsigned int tc, unsigned int vi,
+ unsigned int cmask, unsigned int pmask, unsigned int nexactf,
+ unsigned int rcaps, unsigned int wxcaps,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_pfvf_cmd *cmdp = (struct fw_pfvf_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, tmo, hw, cbfn, 1);
+
+ cmdp->op_to_vfn = htonl(FW_CMD_OP(FW_PFVF_CMD) |
+ FW_CMD_REQUEST |
+ FW_CMD_WRITE |
+ FW_PFVF_CMD_PFN(pf) |
+ FW_PFVF_CMD_VFN(vf));
+ cmdp->retval_len16 = htonl(FW_CMD_LEN16(sizeof(*cmdp) / 16));
+ cmdp->niqflint_niq = htonl(FW_PFVF_CMD_NIQFLINT(rxqi) |
+ FW_PFVF_CMD_NIQ(rxq));
+
+ cmdp->type_to_neq = htonl(FW_PFVF_CMD_TYPE |
+ FW_PFVF_CMD_CMASK(cmask) |
+ FW_PFVF_CMD_PMASK(pmask) |
+ FW_PFVF_CMD_NEQ(txq));
+ cmdp->tc_to_nexactf = htonl(FW_PFVF_CMD_TC(tc) |
+ FW_PFVF_CMD_NVI(vi) |
+ FW_PFVF_CMD_NEXACTF(nexactf));
+ cmdp->r_caps_to_nethctrl = htonl(FW_PFVF_CMD_R_CAPS(rcaps) |
+ FW_PFVF_CMD_WX_CAPS(wxcaps) |
+ FW_PFVF_CMD_NETHCTRL(txq_eth_ctrl));
+}
+
+#define CSIO_ADVERT_MASK (FW_PORT_CAP_SPEED_100M | FW_PORT_CAP_SPEED_1G |\
+ FW_PORT_CAP_SPEED_10G | FW_PORT_CAP_ANEG)
+
+/*
+ * csio_mb_port- FW PORT command helper
+ * @hw: The HW structure
+ * @mbp: Mailbox structure
+ * @tmo: COmmand timeout
+ * @portid: Port ID to get/set info
+ * @wr: Write/Read PORT information.
+ * @fc: Flow control
+ * @caps: Port capabilites to set.
+ * @cbfn: Callback, if any.
+ *
+ */
+void
+csio_mb_port(struct csio_hw *hw, struct csio_mb *mbp, uint32_t tmo,
+ uint8_t portid, bool wr, uint32_t fc, uint16_t caps,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_port_cmd *cmdp = (struct fw_port_cmd *)(mbp->mb);
+ unsigned int lfc = 0, mdi = FW_PORT_MDI(FW_PORT_MDI_AUTO);
+
+ CSIO_INIT_MBP(mbp, cmdp, tmo, hw, cbfn, 1);
+
+ cmdp->op_to_portid = htonl(FW_CMD_OP(FW_PORT_CMD) |
+ FW_CMD_REQUEST |
+ (wr ? FW_CMD_EXEC : FW_CMD_READ) |
+ FW_PORT_CMD_PORTID(portid));
+ if (!wr) {
+ cmdp->action_to_len16 = htonl(
+ FW_PORT_CMD_ACTION(FW_PORT_ACTION_GET_PORT_INFO) |
+ FW_CMD_LEN16(sizeof(*cmdp) / 16));
+ return;
+ }
+
+ /* Set port */
+ cmdp->action_to_len16 = htonl(
+ FW_PORT_CMD_ACTION(FW_PORT_ACTION_L1_CFG) |
+ FW_CMD_LEN16(sizeof(*cmdp) / 16));
+
+ if (fc & PAUSE_RX)
+ lfc |= FW_PORT_CAP_FC_RX;
+ if (fc & PAUSE_TX)
+ lfc |= FW_PORT_CAP_FC_TX;
+
+ if (!(caps & FW_PORT_CAP_ANEG))
+ cmdp->u.l1cfg.rcap = htonl((caps & CSIO_ADVERT_MASK) | lfc);
+ else
+ cmdp->u.l1cfg.rcap = htonl((caps & CSIO_ADVERT_MASK) |
+ lfc | mdi);
+}
+
+/*
+ * csio_mb_process_read_port_rsp - FW PORT command response processing helper
+ * @hw: The HW structure
+ * @mbp: Mailbox structure
+ * @retval: Mailbox return value from Firmware
+ * @caps: port capabilities
+ *
+ */
+void
+csio_mb_process_read_port_rsp(struct csio_hw *hw, struct csio_mb *mbp,
+ enum fw_retval *retval, uint16_t *caps)
+{
+ struct fw_port_cmd *rsp = (struct fw_port_cmd *)(mbp->mb);
+
+ *retval = FW_CMD_RETVAL_GET(ntohl(rsp->action_to_len16));
+
+ if (*retval == FW_SUCCESS)
+ *caps = ntohs(rsp->u.info.pcap);
+}
+
+/*
+ * csio_mb_initialize - FW INITIALIZE command helper
+ * @hw: The HW structure
+ * @mbp: Mailbox structure
+ * @tmo: COmmand timeout
+ * @cbfn: Callback, if any.
+ *
+ */
+void
+csio_mb_initialize(struct csio_hw *hw, struct csio_mb *mbp, uint32_t tmo,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_initialize_cmd *cmdp = (struct fw_initialize_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, tmo, hw, cbfn, 1);
+
+ cmdp->op_to_write = htonl(FW_CMD_OP(FW_INITIALIZE_CMD) |
+ FW_CMD_REQUEST | FW_CMD_WRITE);
+ cmdp->retval_len16 = htonl(FW_CMD_LEN16(sizeof(*cmdp) / 16));
+
+}
+
+/*
+ * csio_mb_iq_alloc - Initializes the mailbox to allocate an
+ * Ingress DMA queue in the firmware.
+ *
+ * @hw: The hw structure
+ * @mbp: Mailbox structure to initialize
+ * @priv: Private object
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @iq_params: Ingress queue params needed for allocation.
+ * @cbfn: The call-back function
+ *
+ *
+ */
+static void
+csio_mb_iq_alloc(struct csio_hw *hw, struct csio_mb *mbp, void *priv,
+ uint32_t mb_tmo, struct csio_iq_params *iq_params,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_iq_cmd *cmdp = (struct fw_iq_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, mb_tmo, priv, cbfn, 1);
+
+ cmdp->op_to_vfn = htonl(FW_CMD_OP(FW_IQ_CMD) |
+ FW_CMD_REQUEST | FW_CMD_EXEC |
+ FW_IQ_CMD_PFN(iq_params->pfn) |
+ FW_IQ_CMD_VFN(iq_params->vfn));
+
+ cmdp->alloc_to_len16 = htonl(FW_IQ_CMD_ALLOC |
+ FW_CMD_LEN16(sizeof(*cmdp) / 16));
+
+ cmdp->type_to_iqandstindex = htonl(
+ FW_IQ_CMD_VIID(iq_params->viid) |
+ FW_IQ_CMD_TYPE(iq_params->type) |
+ FW_IQ_CMD_IQASYNCH(iq_params->iqasynch));
+
+ cmdp->fl0size = htons(iq_params->fl0size);
+ cmdp->fl0size = htons(iq_params->fl1size);
+
+} /* csio_mb_iq_alloc */
+
+/*
+ * csio_mb_iq_write - Initializes the mailbox for writing into an
+ * Ingress DMA Queue.
+ *
+ * @hw: The HW structure
+ * @mbp: Mailbox structure to initialize
+ * @priv: Private object
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @cascaded_req: TRUE - if this request is cascased with iq-alloc request.
+ * @iq_params: Ingress queue params needed for writing.
+ * @cbfn: The call-back function
+ *
+ * NOTE: We OR relevant bits with cmdp->XXX, instead of just equating,
+ * because this IQ write request can be cascaded with a previous
+ * IQ alloc request, and we dont want to over-write the bits set by
+ * that request. This logic will work even in a non-cascaded case, since the
+ * cmdp structure is zeroed out by CSIO_INIT_MBP.
+ */
+static void
+csio_mb_iq_write(struct csio_hw *hw, struct csio_mb *mbp, void *priv,
+ uint32_t mb_tmo, bool cascaded_req,
+ struct csio_iq_params *iq_params,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_iq_cmd *cmdp = (struct fw_iq_cmd *)(mbp->mb);
+
+ uint32_t iq_start_stop = (iq_params->iq_start) ?
+ FW_IQ_CMD_IQSTART(1) :
+ FW_IQ_CMD_IQSTOP(1);
+
+ /*
+ * If this IQ write is cascaded with IQ alloc request, do not
+ * re-initialize with 0's.
+ *
+ */
+ if (!cascaded_req)
+ CSIO_INIT_MBP(mbp, cmdp, mb_tmo, priv, cbfn, 1);
+
+ cmdp->op_to_vfn |= htonl(FW_CMD_OP(FW_IQ_CMD) |
+ FW_CMD_REQUEST | FW_CMD_WRITE |
+ FW_IQ_CMD_PFN(iq_params->pfn) |
+ FW_IQ_CMD_VFN(iq_params->vfn));
+ cmdp->alloc_to_len16 |= htonl(iq_start_stop |
+ FW_CMD_LEN16(sizeof(*cmdp) / 16));
+ cmdp->iqid |= htons(iq_params->iqid);
+ cmdp->fl0id |= htons(iq_params->fl0id);
+ cmdp->fl1id |= htons(iq_params->fl1id);
+ cmdp->type_to_iqandstindex |= htonl(
+ FW_IQ_CMD_IQANDST(iq_params->iqandst) |
+ FW_IQ_CMD_IQANUS(iq_params->iqanus) |
+ FW_IQ_CMD_IQANUD(iq_params->iqanud) |
+ FW_IQ_CMD_IQANDSTINDEX(iq_params->iqandstindex));
+ cmdp->iqdroprss_to_iqesize |= htons(
+ FW_IQ_CMD_IQPCIECH(iq_params->iqpciech) |
+ FW_IQ_CMD_IQDCAEN(iq_params->iqdcaen) |
+ FW_IQ_CMD_IQDCACPU(iq_params->iqdcacpu) |
+ FW_IQ_CMD_IQINTCNTTHRESH(iq_params->iqintcntthresh) |
+ FW_IQ_CMD_IQCPRIO(iq_params->iqcprio) |
+ FW_IQ_CMD_IQESIZE(iq_params->iqesize));
+
+ cmdp->iqsize |= htons(iq_params->iqsize);
+ cmdp->iqaddr |= cpu_to_be64(iq_params->iqaddr);
+
+ if (iq_params->type == 0) {
+ cmdp->iqns_to_fl0congen |= htonl(
+ FW_IQ_CMD_IQFLINTIQHSEN(iq_params->iqflintiqhsen)|
+ FW_IQ_CMD_IQFLINTCONGEN(iq_params->iqflintcongen));
+ }
+
+ if (iq_params->fl0size && iq_params->fl0addr &&
+ (iq_params->fl0id != 0xFFFF)) {
+
+ cmdp->iqns_to_fl0congen |= htonl(
+ FW_IQ_CMD_FL0HOSTFCMODE(iq_params->fl0hostfcmode)|
+ FW_IQ_CMD_FL0CPRIO(iq_params->fl0cprio) |
+ FW_IQ_CMD_FL0PADEN(iq_params->fl0paden) |
+ FW_IQ_CMD_FL0PACKEN(iq_params->fl0packen));
+ cmdp->fl0dcaen_to_fl0cidxfthresh |= htons(
+ FW_IQ_CMD_FL0DCAEN(iq_params->fl0dcaen) |
+ FW_IQ_CMD_FL0DCACPU(iq_params->fl0dcacpu) |
+ FW_IQ_CMD_FL0FBMIN(iq_params->fl0fbmin) |
+ FW_IQ_CMD_FL0FBMAX(iq_params->fl0fbmax) |
+ FW_IQ_CMD_FL0CIDXFTHRESH(iq_params->fl0cidxfthresh));
+ cmdp->fl0size |= htons(iq_params->fl0size);
+ cmdp->fl0addr |= cpu_to_be64(iq_params->fl0addr);
+ }
+} /* csio_mb_iq_write */
+
+/*
+ * csio_mb_iq_alloc_write - Initializes the mailbox for allocating an
+ * Ingress DMA Queue.
+ *
+ * @hw: The HW structure
+ * @mbp: Mailbox structure to initialize
+ * @priv: Private data.
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @iq_params: Ingress queue params needed for allocation & writing.
+ * @cbfn: The call-back function
+ *
+ *
+ */
+void
+csio_mb_iq_alloc_write(struct csio_hw *hw, struct csio_mb *mbp, void *priv,
+ uint32_t mb_tmo, struct csio_iq_params *iq_params,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ csio_mb_iq_alloc(hw, mbp, priv, mb_tmo, iq_params, cbfn);
+ csio_mb_iq_write(hw, mbp, priv, mb_tmo, true, iq_params, cbfn);
+} /* csio_mb_iq_alloc_write */
+
+/*
+ * csio_mb_iq_alloc_write_rsp - Process the allocation & writing
+ * of ingress DMA queue mailbox's response.
+ *
+ * @hw: The HW structure.
+ * @mbp: Mailbox structure to initialize.
+ * @retval: Firmware return value.
+ * @iq_params: Ingress queue parameters, after allocation and write.
+ *
+ */
+void
+csio_mb_iq_alloc_write_rsp(struct csio_hw *hw, struct csio_mb *mbp,
+ enum fw_retval *ret_val,
+ struct csio_iq_params *iq_params)
+{
+ struct fw_iq_cmd *rsp = (struct fw_iq_cmd *)(mbp->mb);
+
+ *ret_val = FW_CMD_RETVAL_GET(ntohl(rsp->alloc_to_len16));
+ if (*ret_val == FW_SUCCESS) {
+ iq_params->physiqid = ntohs(rsp->physiqid);
+ iq_params->iqid = ntohs(rsp->iqid);
+ iq_params->fl0id = ntohs(rsp->fl0id);
+ iq_params->fl1id = ntohs(rsp->fl1id);
+ } else {
+ iq_params->physiqid = iq_params->iqid =
+ iq_params->fl0id = iq_params->fl1id = 0;
+ }
+} /* csio_mb_iq_alloc_write_rsp */
+
+/*
+ * csio_mb_iq_free - Initializes the mailbox for freeing a
+ * specified Ingress DMA Queue.
+ *
+ * @hw: The HW structure
+ * @mbp: Mailbox structure to initialize
+ * @priv: Private data
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @iq_params: Parameters of ingress queue, that is to be freed.
+ * @cbfn: The call-back function
+ *
+ *
+ */
+void
+csio_mb_iq_free(struct csio_hw *hw, struct csio_mb *mbp, void *priv,
+ uint32_t mb_tmo, struct csio_iq_params *iq_params,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_iq_cmd *cmdp = (struct fw_iq_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, mb_tmo, priv, cbfn, 1);
+
+ cmdp->op_to_vfn = htonl(FW_CMD_OP(FW_IQ_CMD) |
+ FW_CMD_REQUEST | FW_CMD_EXEC |
+ FW_IQ_CMD_PFN(iq_params->pfn) |
+ FW_IQ_CMD_VFN(iq_params->vfn));
+ cmdp->alloc_to_len16 = htonl(FW_IQ_CMD_FREE |
+ FW_CMD_LEN16(sizeof(*cmdp) / 16));
+ cmdp->type_to_iqandstindex = htonl(FW_IQ_CMD_TYPE(iq_params->type));
+
+ cmdp->iqid = htons(iq_params->iqid);
+ cmdp->fl0id = htons(iq_params->fl0id);
+ cmdp->fl1id = htons(iq_params->fl1id);
+
+} /* csio_mb_iq_free */
+
+/*
+ * csio_mb_eq_ofld_alloc - Initializes the mailbox for allocating
+ * an offload-egress queue.
+ *
+ * @hw: The HW structure
+ * @mbp: Mailbox structure to initialize
+ * @priv: Private data
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @eq_ofld_params: (Offload) Egress queue paramters.
+ * @cbfn: The call-back function
+ *
+ *
+ */
+static void
+csio_mb_eq_ofld_alloc(struct csio_hw *hw, struct csio_mb *mbp, void *priv,
+ uint32_t mb_tmo, struct csio_eq_params *eq_ofld_params,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_eq_ofld_cmd *cmdp = (struct fw_eq_ofld_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, mb_tmo, priv, cbfn, 1);
+ cmdp->op_to_vfn = htonl(FW_CMD_OP(FW_EQ_OFLD_CMD) |
+ FW_CMD_REQUEST | FW_CMD_EXEC |
+ FW_EQ_OFLD_CMD_PFN(eq_ofld_params->pfn) |
+ FW_EQ_OFLD_CMD_VFN(eq_ofld_params->vfn));
+ cmdp->alloc_to_len16 = htonl(FW_EQ_OFLD_CMD_ALLOC |
+ FW_CMD_LEN16(sizeof(*cmdp) / 16));
+
+} /* csio_mb_eq_ofld_alloc */
+
+/*
+ * csio_mb_eq_ofld_write - Initializes the mailbox for writing
+ * an alloacted offload-egress queue.
+ *
+ * @hw: The HW structure
+ * @mbp: Mailbox structure to initialize
+ * @priv: Private data
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @cascaded_req: TRUE - if this request is cascased with Eq-alloc request.
+ * @eq_ofld_params: (Offload) Egress queue paramters.
+ * @cbfn: The call-back function
+ *
+ *
+ * NOTE: We OR relevant bits with cmdp->XXX, instead of just equating,
+ * because this EQ write request can be cascaded with a previous
+ * EQ alloc request, and we dont want to over-write the bits set by
+ * that request. This logic will work even in a non-cascaded case, since the
+ * cmdp structure is zeroed out by CSIO_INIT_MBP.
+ */
+static void
+csio_mb_eq_ofld_write(struct csio_hw *hw, struct csio_mb *mbp, void *priv,
+ uint32_t mb_tmo, bool cascaded_req,
+ struct csio_eq_params *eq_ofld_params,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_eq_ofld_cmd *cmdp = (struct fw_eq_ofld_cmd *)(mbp->mb);
+
+ uint32_t eq_start_stop = (eq_ofld_params->eqstart) ?
+ FW_EQ_OFLD_CMD_EQSTART : FW_EQ_OFLD_CMD_EQSTOP;
+
+ /*
+ * If this EQ write is cascaded with EQ alloc request, do not
+ * re-initialize with 0's.
+ *
+ */
+ if (!cascaded_req)
+ CSIO_INIT_MBP(mbp, cmdp, mb_tmo, priv, cbfn, 1);
+
+ cmdp->op_to_vfn |= htonl(FW_CMD_OP(FW_EQ_OFLD_CMD) |
+ FW_CMD_REQUEST | FW_CMD_WRITE |
+ FW_EQ_OFLD_CMD_PFN(eq_ofld_params->pfn) |
+ FW_EQ_OFLD_CMD_VFN(eq_ofld_params->vfn));
+ cmdp->alloc_to_len16 |= htonl(eq_start_stop |
+ FW_CMD_LEN16(sizeof(*cmdp) / 16));
+
+ cmdp->eqid_pkd |= htonl(FW_EQ_OFLD_CMD_EQID(eq_ofld_params->eqid));
+
+ cmdp->fetchszm_to_iqid |= htonl(
+ FW_EQ_OFLD_CMD_HOSTFCMODE(eq_ofld_params->hostfcmode) |
+ FW_EQ_OFLD_CMD_CPRIO(eq_ofld_params->cprio) |
+ FW_EQ_OFLD_CMD_PCIECHN(eq_ofld_params->pciechn) |
+ FW_EQ_OFLD_CMD_IQID(eq_ofld_params->iqid));
+
+ cmdp->dcaen_to_eqsize |= htonl(
+ FW_EQ_OFLD_CMD_DCAEN(eq_ofld_params->dcaen) |
+ FW_EQ_OFLD_CMD_DCACPU(eq_ofld_params->dcacpu) |
+ FW_EQ_OFLD_CMD_FBMIN(eq_ofld_params->fbmin) |
+ FW_EQ_OFLD_CMD_FBMAX(eq_ofld_params->fbmax) |
+ FW_EQ_OFLD_CMD_CIDXFTHRESHO(eq_ofld_params->cidxfthresho) |
+ FW_EQ_OFLD_CMD_CIDXFTHRESH(eq_ofld_params->cidxfthresh) |
+ FW_EQ_OFLD_CMD_EQSIZE(eq_ofld_params->eqsize));
+
+ cmdp->eqaddr |= cpu_to_be64(eq_ofld_params->eqaddr);
+
+} /* csio_mb_eq_ofld_write */
+
+/*
+ * csio_mb_eq_ofld_alloc_write - Initializes the mailbox for allocation
+ * writing into an Engress DMA Queue.
+ *
+ * @hw: The HW structure
+ * @mbp: Mailbox structure to initialize
+ * @priv: Private data.
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @eq_ofld_params: (Offload) Egress queue paramters.
+ * @cbfn: The call-back function
+ *
+ *
+ */
+void
+csio_mb_eq_ofld_alloc_write(struct csio_hw *hw, struct csio_mb *mbp,
+ void *priv, uint32_t mb_tmo,
+ struct csio_eq_params *eq_ofld_params,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ csio_mb_eq_ofld_alloc(hw, mbp, priv, mb_tmo, eq_ofld_params, cbfn);
+ csio_mb_eq_ofld_write(hw, mbp, priv, mb_tmo, true,
+ eq_ofld_params, cbfn);
+} /* csio_mb_eq_ofld_alloc_write */
+
+/*
+ * csio_mb_eq_ofld_alloc_write_rsp - Process the allocation
+ * & write egress DMA queue mailbox's response.
+ *
+ * @hw: The HW structure.
+ * @mbp: Mailbox structure to initialize.
+ * @retval: Firmware return value.
+ * @eq_ofld_params: (Offload) Egress queue paramters.
+ *
+ */
+void
+csio_mb_eq_ofld_alloc_write_rsp(struct csio_hw *hw,
+ struct csio_mb *mbp, enum fw_retval *ret_val,
+ struct csio_eq_params *eq_ofld_params)
+{
+ struct fw_eq_ofld_cmd *rsp = (struct fw_eq_ofld_cmd *)(mbp->mb);
+
+ *ret_val = FW_CMD_RETVAL_GET(ntohl(rsp->alloc_to_len16));
+
+ if (*ret_val == FW_SUCCESS) {
+ eq_ofld_params->eqid = FW_EQ_OFLD_CMD_EQID_GET(
+ ntohl(rsp->eqid_pkd));
+ eq_ofld_params->physeqid = FW_EQ_OFLD_CMD_PHYSEQID_GET(
+ ntohl(rsp->physeqid_pkd));
+ } else
+ eq_ofld_params->eqid = 0;
+
+} /* csio_mb_eq_ofld_alloc_write_rsp */
+
+/*
+ * csio_mb_eq_ofld_free - Initializes the mailbox for freeing a
+ * specified Engress DMA Queue.
+ *
+ * @hw: The HW structure
+ * @mbp: Mailbox structure to initialize
+ * @priv: Private data area.
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @eq_ofld_params: (Offload) Egress queue paramters, that is to be freed.
+ * @cbfn: The call-back function
+ *
+ *
+ */
+void
+csio_mb_eq_ofld_free(struct csio_hw *hw, struct csio_mb *mbp, void *priv,
+ uint32_t mb_tmo, struct csio_eq_params *eq_ofld_params,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_eq_ofld_cmd *cmdp = (struct fw_eq_ofld_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, mb_tmo, priv, cbfn, 1);
+
+ cmdp->op_to_vfn = htonl(FW_CMD_OP(FW_EQ_OFLD_CMD) |
+ FW_CMD_REQUEST | FW_CMD_EXEC |
+ FW_EQ_OFLD_CMD_PFN(eq_ofld_params->pfn) |
+ FW_EQ_OFLD_CMD_VFN(eq_ofld_params->vfn));
+ cmdp->alloc_to_len16 = htonl(FW_EQ_OFLD_CMD_FREE |
+ FW_CMD_LEN16(sizeof(*cmdp) / 16));
+ cmdp->eqid_pkd = htonl(FW_EQ_OFLD_CMD_EQID(eq_ofld_params->eqid));
+
+} /* csio_mb_eq_ofld_free */
+
+/*
+ * csio_write_fcoe_link_cond_init_mb - Initialize Mailbox to write FCoE link
+ * condition.
+ *
+ * @ln: The Lnode structure
+ * @mbp: Mailbox structure to initialize
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @cbfn: The call back function.
+ *
+ *
+ */
+void
+csio_write_fcoe_link_cond_init_mb(struct csio_lnode *ln, struct csio_mb *mbp,
+ uint32_t mb_tmo, uint8_t port_id, uint32_t sub_opcode,
+ uint8_t cos, bool link_status, uint32_t fcfi,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_fcoe_link_cmd *cmdp =
+ (struct fw_fcoe_link_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, mb_tmo, ln, cbfn, 1);
+
+ cmdp->op_to_portid = htonl((
+ FW_CMD_OP(FW_FCOE_LINK_CMD) |
+ FW_CMD_REQUEST |
+ FW_CMD_WRITE |
+ FW_FCOE_LINK_CMD_PORTID(port_id)));
+ cmdp->sub_opcode_fcfi = htonl(
+ FW_FCOE_LINK_CMD_SUB_OPCODE(sub_opcode) |
+ FW_FCOE_LINK_CMD_FCFI(fcfi));
+ cmdp->lstatus = link_status;
+ cmdp->retval_len16 = htonl(FW_CMD_LEN16(sizeof(*cmdp) / 16));
+
+} /* csio_write_fcoe_link_cond_init_mb */
+
+/*
+ * csio_fcoe_read_res_info_init_mb - Initializes the mailbox for reading FCoE
+ * resource information(FW_GET_RES_INFO_CMD).
+ *
+ * @hw: The HW structure
+ * @mbp: Mailbox structure to initialize
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @cbfn: The call-back function
+ *
+ *
+ */
+void
+csio_fcoe_read_res_info_init_mb(struct csio_hw *hw, struct csio_mb *mbp,
+ uint32_t mb_tmo,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_fcoe_res_info_cmd *cmdp =
+ (struct fw_fcoe_res_info_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, mb_tmo, hw, cbfn, 1);
+
+ cmdp->op_to_read = htonl((FW_CMD_OP(FW_FCOE_RES_INFO_CMD) |
+ FW_CMD_REQUEST |
+ FW_CMD_READ));
+
+ cmdp->retval_len16 = htonl(FW_CMD_LEN16(sizeof(*cmdp) / 16));
+
+} /* csio_fcoe_read_res_info_init_mb */
+
+/*
+ * csio_fcoe_vnp_alloc_init_mb - Initializes the mailbox for allocating VNP
+ * in the firmware (FW_FCOE_VNP_CMD).
+ *
+ * @ln: The Lnode structure.
+ * @mbp: Mailbox structure to initialize.
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @fcfi: FCF Index.
+ * @vnpi: vnpi
+ * @iqid: iqid
+ * @vnport_wwnn: vnport WWNN
+ * @vnport_wwpn: vnport WWPN
+ * @cbfn: The call-back function.
+ *
+ *
+ */
+void
+csio_fcoe_vnp_alloc_init_mb(struct csio_lnode *ln, struct csio_mb *mbp,
+ uint32_t mb_tmo, uint32_t fcfi, uint32_t vnpi, uint16_t iqid,
+ uint8_t vnport_wwnn[8], uint8_t vnport_wwpn[8],
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_fcoe_vnp_cmd *cmdp =
+ (struct fw_fcoe_vnp_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, mb_tmo, ln, cbfn, 1);
+
+ cmdp->op_to_fcfi = htonl((FW_CMD_OP(FW_FCOE_VNP_CMD) |
+ FW_CMD_REQUEST |
+ FW_CMD_EXEC |
+ FW_FCOE_VNP_CMD_FCFI(fcfi)));
+
+ cmdp->alloc_to_len16 = htonl(FW_FCOE_VNP_CMD_ALLOC |
+ FW_CMD_LEN16(sizeof(*cmdp) / 16));
+
+ cmdp->gen_wwn_to_vnpi = htonl(FW_FCOE_VNP_CMD_VNPI(vnpi));
+
+ cmdp->iqid = htons(iqid);
+
+ if (!wwn_to_u64(vnport_wwnn) && !wwn_to_u64(vnport_wwpn))
+ cmdp->gen_wwn_to_vnpi |= htonl(FW_FCOE_VNP_CMD_GEN_WWN);
+
+ if (vnport_wwnn)
+ memcpy(cmdp->vnport_wwnn, vnport_wwnn, 8);
+ if (vnport_wwpn)
+ memcpy(cmdp->vnport_wwpn, vnport_wwpn, 8);
+
+} /* csio_fcoe_vnp_alloc_init_mb */
+
+/*
+ * csio_fcoe_vnp_read_init_mb - Prepares VNP read cmd.
+ * @ln: The Lnode structure.
+ * @mbp: Mailbox structure to initialize.
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @fcfi: FCF Index.
+ * @vnpi: vnpi
+ * @cbfn: The call-back handler.
+ */
+void
+csio_fcoe_vnp_read_init_mb(struct csio_lnode *ln, struct csio_mb *mbp,
+ uint32_t mb_tmo, uint32_t fcfi, uint32_t vnpi,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_fcoe_vnp_cmd *cmdp =
+ (struct fw_fcoe_vnp_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, mb_tmo, ln, cbfn, 1);
+ cmdp->op_to_fcfi = htonl(FW_CMD_OP(FW_FCOE_VNP_CMD) |
+ FW_CMD_REQUEST |
+ FW_CMD_READ |
+ FW_FCOE_VNP_CMD_FCFI(fcfi));
+ cmdp->alloc_to_len16 = htonl(FW_CMD_LEN16(sizeof(*cmdp) / 16));
+ cmdp->gen_wwn_to_vnpi = htonl(FW_FCOE_VNP_CMD_VNPI(vnpi));
+}
+
+/*
+ * csio_fcoe_vnp_free_init_mb - Initializes the mailbox for freeing an
+ * alloacted VNP in the firmware (FW_FCOE_VNP_CMD).
+ *
+ * @ln: The Lnode structure.
+ * @mbp: Mailbox structure to initialize.
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @fcfi: FCF flow id
+ * @vnpi: VNP flow id
+ * @cbfn: The call-back function.
+ * Return: None
+ */
+void
+csio_fcoe_vnp_free_init_mb(struct csio_lnode *ln, struct csio_mb *mbp,
+ uint32_t mb_tmo, uint32_t fcfi, uint32_t vnpi,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_fcoe_vnp_cmd *cmdp =
+ (struct fw_fcoe_vnp_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, mb_tmo, ln, cbfn, 1);
+
+ cmdp->op_to_fcfi = htonl(FW_CMD_OP(FW_FCOE_VNP_CMD) |
+ FW_CMD_REQUEST |
+ FW_CMD_EXEC |
+ FW_FCOE_VNP_CMD_FCFI(fcfi));
+ cmdp->alloc_to_len16 = htonl(FW_FCOE_VNP_CMD_FREE |
+ FW_CMD_LEN16(sizeof(*cmdp) / 16));
+ cmdp->gen_wwn_to_vnpi = htonl(FW_FCOE_VNP_CMD_VNPI(vnpi));
+}
+
+/*
+ * csio_fcoe_read_fcf_init_mb - Initializes the mailbox to read the
+ * FCF records.
+ *
+ * @ln: The Lnode structure
+ * @mbp: Mailbox structure to initialize
+ * @mb_tmo: Mailbox time-out period (in ms).
+ * @fcf_params: FC-Forwarder parameters.
+ * @cbfn: The call-back function
+ *
+ *
+ */
+void
+csio_fcoe_read_fcf_init_mb(struct csio_lnode *ln, struct csio_mb *mbp,
+ uint32_t mb_tmo, uint32_t portid, uint32_t fcfi,
+ void (*cbfn) (struct csio_hw *, struct csio_mb *))
+{
+ struct fw_fcoe_fcf_cmd *cmdp =
+ (struct fw_fcoe_fcf_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, mb_tmo, ln, cbfn, 1);
+
+ cmdp->op_to_fcfi = htonl(FW_CMD_OP(FW_FCOE_FCF_CMD) |
+ FW_CMD_REQUEST |
+ FW_CMD_READ |
+ FW_FCOE_FCF_CMD_FCFI(fcfi));
+ cmdp->retval_len16 = htonl(FW_CMD_LEN16(sizeof(*cmdp) / 16));
+
+} /* csio_fcoe_read_fcf_init_mb */
+
+void
+csio_fcoe_read_portparams_init_mb(struct csio_hw *hw, struct csio_mb *mbp,
+ uint32_t mb_tmo,
+ struct fw_fcoe_port_cmd_params *portparams,
+ void (*cbfn)(struct csio_hw *,
+ struct csio_mb *))
+{
+ struct fw_fcoe_stats_cmd *cmdp = (struct fw_fcoe_stats_cmd *)(mbp->mb);
+
+ CSIO_INIT_MBP(mbp, cmdp, mb_tmo, hw, cbfn, 1);
+ mbp->mb_size = 64;
+
+ cmdp->op_to_flowid = htonl(FW_CMD_OP(FW_FCOE_STATS_CMD) |
+ FW_CMD_REQUEST | FW_CMD_READ);
+ cmdp->free_to_len16 = htonl(FW_CMD_LEN16(CSIO_MAX_MB_SIZE/16));
+
+ cmdp->u.ctl.nstats_port = FW_FCOE_STATS_CMD_NSTATS(portparams->nstats) |
+ FW_FCOE_STATS_CMD_PORT(portparams->portid);
+
+ cmdp->u.ctl.port_valid_ix = FW_FCOE_STATS_CMD_IX(portparams->idx) |
+ FW_FCOE_STATS_CMD_PORT_VALID;
+
+} /* csio_fcoe_read_portparams_init_mb */
+
+void
+csio_mb_process_portparams_rsp(
+ struct csio_hw *hw,
+ struct csio_mb *mbp,
+ enum fw_retval *retval,
+ struct fw_fcoe_port_cmd_params *portparams,
+ struct fw_fcoe_port_stats *portstats
+ )
+{
+ struct fw_fcoe_stats_cmd *rsp = (struct fw_fcoe_stats_cmd *)(mbp->mb);
+ struct fw_fcoe_port_stats stats;
+ uint8_t *src;
+ uint8_t *dst;
+
+ *retval = FW_CMD_RETVAL_GET(ntohl(rsp->free_to_len16));
+
+ memset(&stats, 0, sizeof(struct fw_fcoe_port_stats));
+
+ if (*retval == FW_SUCCESS) {
+ dst = (uint8_t *)(&stats) + ((portparams->idx - 1) * 8);
+ src = (uint8_t *)rsp + (CSIO_STATS_OFFSET * 8);
+ memcpy(dst, src, (portparams->nstats * 8));
+ if (portparams->idx == 1) {
+ /* Get the first 6 flits from the Mailbox */
+ portstats->tx_bcast_bytes =
+ be64_to_cpu(stats.tx_bcast_bytes);
+ portstats->tx_bcast_frames =
+ be64_to_cpu(stats.tx_bcast_frames);
+ portstats->tx_mcast_bytes =
+ be64_to_cpu(stats.tx_mcast_bytes);
+ portstats->tx_mcast_frames =
+ be64_to_cpu(stats.tx_mcast_frames);
+ portstats->tx_ucast_bytes =
+ be64_to_cpu(stats.tx_ucast_bytes);
+ portstats->tx_ucast_frames =
+ be64_to_cpu(stats.tx_ucast_frames);
+ }
+ if (portparams->idx == 7) {
+ /* Get the second 6 flits from the Mailbox */
+ portstats->tx_drop_frames =
+ be64_to_cpu(stats.tx_drop_frames);
+ portstats->tx_offload_bytes =
+ be64_to_cpu(stats.tx_offload_bytes);
+ portstats->tx_offload_frames =
+ be64_to_cpu(stats.tx_offload_frames);
+#if 0
+ portstats->rx_pf_bytes =
+ be64_to_cpu(stats.rx_pf_bytes);
+ portstats->rx_pf_frames =
+ be64_to_cpu(stats.rx_pf_frames);
+#endif
+ portstats->rx_bcast_bytes =
+ be64_to_cpu(stats.rx_bcast_bytes);
+ portstats->rx_bcast_frames =
+ be64_to_cpu(stats.rx_bcast_frames);
+ portstats->rx_mcast_bytes =
+ be64_to_cpu(stats.rx_mcast_bytes);
+ }
+ if (portparams->idx == 13) {
+ /* Get the last 4 flits from the Mailbox */
+ portstats->rx_mcast_frames =
+ be64_to_cpu(stats.rx_mcast_frames);
+ portstats->rx_ucast_bytes =
+ be64_to_cpu(stats.rx_ucast_bytes);
+ portstats->rx_ucast_frames =
+ be64_to_cpu(stats.rx_ucast_frames);
+ portstats->rx_err_frames =
+ be64_to_cpu(stats.rx_err_frames);
+ }
+ }
+}
+
+/* Entry points/APIs for MB module */
+/*
+ * csio_mb_intr_enable - Enable Interrupts from mailboxes.
+ * @hw: The HW structure
+ *
+ * Enables CIM interrupt bit in appropriate INT_ENABLE registers.
+ */
+void
+csio_mb_intr_enable(struct csio_hw *hw)
+{
+ csio_wr_reg32(hw, MBMSGRDYINTEN(1), MYPF_REG(CIM_PF_HOST_INT_ENABLE));
+ csio_rd_reg32(hw, MYPF_REG(CIM_PF_HOST_INT_ENABLE));
+}
+
+/*
+ * csio_mb_intr_disable - Disable Interrupts from mailboxes.
+ * @hw: The HW structure
+ *
+ * Disable bit in HostInterruptEnable CIM register.
+ */
+void
+csio_mb_intr_disable(struct csio_hw *hw)
+{
+ csio_wr_reg32(hw, MBMSGRDYINTEN(0), MYPF_REG(CIM_PF_HOST_INT_ENABLE));
+ csio_rd_reg32(hw, MYPF_REG(CIM_PF_HOST_INT_ENABLE));
+}
+
+static void
+csio_mb_dump_fw_dbg(struct csio_hw *hw, __be64 *cmd)
+{
+ struct fw_debug_cmd *dbg = (struct fw_debug_cmd *)cmd;
+
+ if ((FW_DEBUG_CMD_TYPE_GET(ntohl(dbg->op_type))) == 1) {
+ csio_info(hw, "FW print message:\n");
+ csio_info(hw, "\tdebug->dprtstridx = %d\n",
+ ntohs(dbg->u.prt.dprtstridx));
+ csio_info(hw, "\tdebug->dprtstrparam0 = 0x%x\n",
+ ntohl(dbg->u.prt.dprtstrparam0));
+ csio_info(hw, "\tdebug->dprtstrparam1 = 0x%x\n",
+ ntohl(dbg->u.prt.dprtstrparam1));
+ csio_info(hw, "\tdebug->dprtstrparam2 = 0x%x\n",
+ ntohl(dbg->u.prt.dprtstrparam2));
+ csio_info(hw, "\tdebug->dprtstrparam3 = 0x%x\n",
+ ntohl(dbg->u.prt.dprtstrparam3));
+ } else {
+ /* This is a FW assertion */
+ csio_fatal(hw, "FW assertion at %.16s:%u, val0 %#x, val1 %#x\n",
+ dbg->u.assert.filename_0_7,
+ ntohl(dbg->u.assert.line),
+ ntohl(dbg->u.assert.x),
+ ntohl(dbg->u.assert.y));
+ }
+}
+
+static void
+csio_mb_debug_cmd_handler(struct csio_hw *hw)
+{
+ int i;
+ __be64 cmd[CSIO_MB_MAX_REGS];
+ uint32_t ctl_reg = PF_REG(hw->pfn, CIM_PF_MAILBOX_CTRL);
+ uint32_t data_reg = PF_REG(hw->pfn, CIM_PF_MAILBOX_DATA);
+ int size = sizeof(struct fw_debug_cmd);
+
+ /* Copy mailbox data */
+ for (i = 0; i < size; i += 8)
+ cmd[i / 8] = cpu_to_be64(csio_rd_reg64(hw, data_reg + i));
+
+ csio_mb_dump_fw_dbg(hw, cmd);
+
+ /* Notify FW of mailbox by setting owner as UP */
+ csio_wr_reg32(hw, MBMSGVALID | MBINTREQ | MBOWNER(CSIO_MBOWNER_FW),
+ ctl_reg);
+
+ csio_rd_reg32(hw, ctl_reg);
+ wmb();
+}
+
+/*
+ * csio_mb_issue - generic routine for issuing Mailbox commands.
+ * @hw: The HW structure
+ * @mbp: Mailbox command to issue
+ *
+ * Caller should hold hw lock across this call.
+ */
+int
+csio_mb_issue(struct csio_hw *hw, struct csio_mb *mbp)
+{
+ uint32_t owner, ctl;
+ int i;
+ uint32_t ii;
+ __be64 *cmd = mbp->mb;
+ __be64 hdr;
+ struct csio_mbm *mbm = &hw->mbm;
+ uint32_t ctl_reg = PF_REG(hw->pfn, CIM_PF_MAILBOX_CTRL);
+ uint32_t data_reg = PF_REG(hw->pfn, CIM_PF_MAILBOX_DATA);
+ int size = mbp->mb_size;
+ int rv = -EINVAL;
+ struct fw_cmd_hdr *fw_hdr;
+
+ /* Determine mode */
+ if (mbp->mb_cbfn == NULL) {
+ /* Need to issue/get results in the same context */
+ if (mbp->tmo < CSIO_MB_POLL_FREQ) {
+ csio_err(hw, "Invalid tmo: 0x%x\n", mbp->tmo);
+ goto error_out;
+ }
+ } else if (!csio_is_host_intr_enabled(hw) ||
+ !csio_is_hw_intr_enabled(hw)) {
+ csio_err(hw, "Cannot issue mailbox in interrupt mode 0x%x\n",
+ *((uint8_t *)mbp->mb));
+ goto error_out;
+ }
+
+ if (mbm->mcurrent != NULL) {
+ /* Queue mbox cmd, if another mbox cmd is active */
+ if (mbp->mb_cbfn == NULL) {
+ rv = -EBUSY;
+ csio_dbg(hw, "Couldnt own Mailbox %x op:0x%x\n",
+ hw->pfn, *((uint8_t *)mbp->mb));
+
+ goto error_out;
+ } else {
+ list_add_tail(&mbp->list, &mbm->req_q);
+ CSIO_INC_STATS(mbm, n_activeq);
+
+ return 0;
+ }
+ }
+
+ /* Now get ownership of mailbox */
+ owner = MBOWNER_GET(csio_rd_reg32(hw, ctl_reg));
+
+ if (!csio_mb_is_host_owner(owner)) {
+
+ for (i = 0; (owner == CSIO_MBOWNER_NONE) && (i < 3); i++)
+ owner = MBOWNER_GET(csio_rd_reg32(hw, ctl_reg));
+ /*
+ * Mailbox unavailable. In immediate mode, fail the command.
+ * In other modes, enqueue the request.
+ */
+ if (!csio_mb_is_host_owner(owner)) {
+ if (mbp->mb_cbfn == NULL) {
+ rv = owner ? -EBUSY : -ETIMEDOUT;
+
+ csio_dbg(hw,
+ "Couldnt own Mailbox %x op:0x%x "
+ "owner:%x\n",
+ hw->pfn, *((uint8_t *)mbp->mb), owner);
+ goto error_out;
+ } else {
+ if (mbm->mcurrent == NULL) {
+ csio_err(hw,
+ "Couldnt own Mailbox %x "
+ "op:0x%x owner:%x\n",
+ hw->pfn, *((uint8_t *)mbp->mb),
+ owner);
+ csio_err(hw,
+ "No outstanding driver"
+ " mailbox as well\n");
+ goto error_out;
+ }
+ }
+ }
+ }
+
+ /* Mailbox is available, copy mailbox data into it */
+ for (i = 0; i < size; i += 8) {
+ csio_wr_reg64(hw, be64_to_cpu(*cmd), data_reg + i);
+ cmd++;
+ }
+
+ CSIO_DUMP_MB(hw, hw->pfn, data_reg);
+
+ /* Start completion timers in non-immediate modes and notify FW */
+ if (mbp->mb_cbfn != NULL) {
+ mbm->mcurrent = mbp;
+ mod_timer(&mbm->timer, jiffies + msecs_to_jiffies(mbp->tmo));
+ csio_wr_reg32(hw, MBMSGVALID | MBINTREQ |
+ MBOWNER(CSIO_MBOWNER_FW), ctl_reg);
+ } else
+ csio_wr_reg32(hw, MBMSGVALID | MBOWNER(CSIO_MBOWNER_FW),
+ ctl_reg);
+
+ /* Flush posted writes */
+ csio_rd_reg32(hw, ctl_reg);
+ wmb();
+
+ CSIO_INC_STATS(mbm, n_req);
+
+ if (mbp->mb_cbfn)
+ return 0;
+
+ /* Poll for completion in immediate mode */
+ cmd = mbp->mb;
+
+ for (ii = 0; ii < mbp->tmo; ii += CSIO_MB_POLL_FREQ) {
+ mdelay(CSIO_MB_POLL_FREQ);
+
+ /* Check for response */
+ ctl = csio_rd_reg32(hw, ctl_reg);
+ if (csio_mb_is_host_owner(MBOWNER_GET(ctl))) {
+
+ if (!(ctl & MBMSGVALID)) {
+ csio_wr_reg32(hw, 0, ctl_reg);
+ continue;
+ }
+
+ CSIO_DUMP_MB(hw, hw->pfn, data_reg);
+
+ hdr = cpu_to_be64(csio_rd_reg64(hw, data_reg));
+ fw_hdr = (struct fw_cmd_hdr *)&hdr;
+
+ switch (FW_CMD_OP_GET(ntohl(fw_hdr->hi))) {
+ case FW_DEBUG_CMD:
+ csio_mb_debug_cmd_handler(hw);
+ continue;
+ }
+
+ /* Copy response */
+ for (i = 0; i < size; i += 8)
+ *cmd++ = cpu_to_be64(csio_rd_reg64
+ (hw, data_reg + i));
+ csio_wr_reg32(hw, 0, ctl_reg);
+
+ if (FW_CMD_RETVAL_GET(*(mbp->mb)))
+ CSIO_INC_STATS(mbm, n_err);
+
+ CSIO_INC_STATS(mbm, n_rsp);
+ return 0;
+ }
+ }
+
+ CSIO_INC_STATS(mbm, n_tmo);
+
+ csio_err(hw, "Mailbox %x op:0x%x timed out!\n",
+ hw->pfn, *((uint8_t *)cmd));
+
+ return -ETIMEDOUT;
+
+error_out:
+ CSIO_INC_STATS(mbm, n_err);
+ return rv;
+}
+
+/*
+ * csio_mb_completions - Completion handler for Mailbox commands
+ * @hw: The HW structure
+ * @cbfn_q: Completion queue.
+ *
+ */
+void
+csio_mb_completions(struct csio_hw *hw, struct list_head *cbfn_q)
+{
+ struct csio_mb *mbp;
+ struct csio_mbm *mbm = &hw->mbm;
+ enum fw_retval rv;
+
+ while (!list_empty(cbfn_q)) {
+ mbp = list_first_entry(cbfn_q, struct csio_mb, list);
+ list_del_init(&mbp->list);
+
+ rv = csio_mb_fw_retval(mbp);
+ if ((rv != FW_SUCCESS) && (rv != FW_HOSTERROR))
+ CSIO_INC_STATS(mbm, n_err);
+ else if (rv != FW_HOSTERROR)
+ CSIO_INC_STATS(mbm, n_rsp);
+
+ if (mbp->mb_cbfn)
+ mbp->mb_cbfn(hw, mbp);
+ }
+}
+
+static void
+csio_mb_portmod_changed(struct csio_hw *hw, uint8_t port_id)
+{
+ static char *mod_str[] = {
+ NULL, "LR", "SR", "ER", "TWINAX", "active TWINAX", "LRM"
+ };
+
+ struct csio_pport *port = &hw->pport[port_id];
+
+ if (port->mod_type == FW_PORT_MOD_TYPE_NONE)
+ csio_info(hw, "Port:%d - port module unplugged\n", port_id);
+ else if (port->mod_type < ARRAY_SIZE(mod_str))
+ csio_info(hw, "Port:%d - %s port module inserted\n", port_id,
+ mod_str[port->mod_type]);
+ else if (port->mod_type == FW_PORT_MOD_TYPE_NOTSUPPORTED)
+ csio_info(hw,
+ "Port:%d - unsupported optical port module "
+ "inserted\n", port_id);
+ else if (port->mod_type == FW_PORT_MOD_TYPE_UNKNOWN)
+ csio_info(hw,
+ "Port:%d - unknown port module inserted, forcing "
+ "TWINAX\n", port_id);
+ else if (port->mod_type == FW_PORT_MOD_TYPE_ERROR)
+ csio_info(hw, "Port:%d - transceiver module error\n", port_id);
+ else
+ csio_info(hw, "Port:%d - unknown module type %d inserted\n",
+ port_id, port->mod_type);
+}
+
+int
+csio_mb_fwevt_handler(struct csio_hw *hw, __be64 *cmd)
+{
+ uint8_t opcode = *(uint8_t *)cmd;
+ struct fw_port_cmd *pcmd;
+ uint8_t port_id;
+ uint32_t link_status;
+ uint16_t action;
+ uint8_t mod_type;
+
+ if (opcode == FW_PORT_CMD) {
+ pcmd = (struct fw_port_cmd *)cmd;
+ port_id = FW_PORT_CMD_PORTID_GET(
+ ntohl(pcmd->op_to_portid));
+ action = FW_PORT_CMD_ACTION_GET(
+ ntohl(pcmd->action_to_len16));
+ if (action != FW_PORT_ACTION_GET_PORT_INFO) {
+ csio_err(hw, "Unhandled FW_PORT_CMD action: %u\n",
+ action);
+ return -EINVAL;
+ }
+
+ link_status = ntohl(pcmd->u.info.lstatus_to_modtype);
+ mod_type = FW_PORT_CMD_MODTYPE_GET(link_status);
+
+ hw->pport[port_id].link_status =
+ FW_PORT_CMD_LSTATUS_GET(link_status);
+ hw->pport[port_id].link_speed =
+ FW_PORT_CMD_LSPEED_GET(link_status);
+
+ csio_info(hw, "Port:%x - LINK %s\n", port_id,
+ FW_PORT_CMD_LSTATUS_GET(link_status) ? "UP" : "DOWN");
+
+ if (mod_type != hw->pport[port_id].mod_type) {
+ hw->pport[port_id].mod_type = mod_type;
+ csio_mb_portmod_changed(hw, port_id);
+ }
+ } else if (opcode == FW_DEBUG_CMD) {
+ csio_mb_dump_fw_dbg(hw, cmd);
+ } else {
+ csio_dbg(hw, "Gen MB can't handle op:0x%x on evtq.\n", opcode);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+/*
+ * csio_mb_isr_handler - Handle mailboxes related interrupts.
+ * @hw: The HW structure
+ *
+ * Called from the ISR to handle Mailbox related interrupts.
+ * HW Lock should be held across this call.
+ */
+int
+csio_mb_isr_handler(struct csio_hw *hw)
+{
+ struct csio_mbm *mbm = &hw->mbm;
+ struct csio_mb *mbp = mbm->mcurrent;
+ __be64 *cmd;
+ uint32_t ctl, cim_cause, pl_cause;
+ int i;
+ uint32_t ctl_reg = PF_REG(hw->pfn, CIM_PF_MAILBOX_CTRL);
+ uint32_t data_reg = PF_REG(hw->pfn, CIM_PF_MAILBOX_DATA);
+ int size;
+ __be64 hdr;
+ struct fw_cmd_hdr *fw_hdr;
+
+ pl_cause = csio_rd_reg32(hw, MYPF_REG(PL_PF_INT_CAUSE));
+ cim_cause = csio_rd_reg32(hw, MYPF_REG(CIM_PF_HOST_INT_CAUSE));
+
+ if (!(pl_cause & PFCIM) || !(cim_cause & MBMSGRDYINT)) {
+ CSIO_INC_STATS(hw, n_mbint_unexp);
+ return -EINVAL;
+ }
+
+ /*
+ * The cause registers below HAVE to be cleared in the SAME
+ * order as below: The low level cause register followed by
+ * the upper level cause register. In other words, CIM-cause
+ * first followed by PL-Cause next.
+ */
+ csio_wr_reg32(hw, MBMSGRDYINT, MYPF_REG(CIM_PF_HOST_INT_CAUSE));
+ csio_wr_reg32(hw, PFCIM, MYPF_REG(PL_PF_INT_CAUSE));
+
+ ctl = csio_rd_reg32(hw, ctl_reg);
+
+ if (csio_mb_is_host_owner(MBOWNER_GET(ctl))) {
+
+ CSIO_DUMP_MB(hw, hw->pfn, data_reg);
+
+ if (!(ctl & MBMSGVALID)) {
+ csio_warn(hw,
+ "Stray mailbox interrupt recvd,"
+ " mailbox data not valid\n");
+ csio_wr_reg32(hw, 0, ctl_reg);
+ /* Flush */
+ csio_rd_reg32(hw, ctl_reg);
+ return -EINVAL;
+ }
+
+ hdr = cpu_to_be64(csio_rd_reg64(hw, data_reg));
+ fw_hdr = (struct fw_cmd_hdr *)&hdr;
+
+ switch (FW_CMD_OP_GET(ntohl(fw_hdr->hi))) {
+ case FW_DEBUG_CMD:
+ csio_mb_debug_cmd_handler(hw);
+ return -EINVAL;
+#if 0
+ case FW_ERROR_CMD:
+ case FW_INITIALIZE_CMD: /* When we are not master */
+#endif
+ }
+
+ CSIO_ASSERT(mbp != NULL);
+
+ cmd = mbp->mb;
+ size = mbp->mb_size;
+ /* Get response */
+ for (i = 0; i < size; i += 8)
+ *cmd++ = cpu_to_be64(csio_rd_reg64
+ (hw, data_reg + i));
+
+ csio_wr_reg32(hw, 0, ctl_reg);
+ /* Flush */
+ csio_rd_reg32(hw, ctl_reg);
+
+ mbm->mcurrent = NULL;
+
+ /* Add completion to tail of cbfn queue */
+ list_add_tail(&mbp->list, &mbm->cbfn_q);
+ CSIO_INC_STATS(mbm, n_cbfnq);
+
+ /*
+ * Enqueue event to EventQ. Events processing happens
+ * in Event worker thread context
+ */
+ if (csio_enqueue_evt(hw, CSIO_EVT_MBX, mbp, sizeof(mbp)))
+ CSIO_INC_STATS(hw, n_evt_drop);
+
+ return 0;
+
+ } else {
+ /*
+ * We can get here if mailbox MSIX vector is shared,
+ * or in INTx case. Or a stray interrupt.
+ */
+ csio_dbg(hw, "Host not owner, no mailbox interrupt\n");
+ CSIO_INC_STATS(hw, n_int_stray);
+ return -EINVAL;
+ }
+}
+
+/*
+ * csio_mb_tmo_handler - Timeout handler
+ * @hw: The HW structure
+ *
+ */
+struct csio_mb *
+csio_mb_tmo_handler(struct csio_hw *hw)
+{
+ struct csio_mbm *mbm = &hw->mbm;
+ struct csio_mb *mbp = mbm->mcurrent;
+ struct fw_cmd_hdr *fw_hdr;
+
+ /*
+ * Could be a race b/w the completion handler and the timer
+ * and the completion handler won that race.
+ */
+ if (mbp == NULL) {
+ CSIO_DB_ASSERT(0);
+ return NULL;
+ }
+
+ fw_hdr = (struct fw_cmd_hdr *)(mbp->mb);
+
+ csio_dbg(hw, "Mailbox num:%x op:0x%x timed out\n", hw->pfn,
+ FW_CMD_OP_GET(ntohl(fw_hdr->hi)));
+
+ mbm->mcurrent = NULL;
+ CSIO_INC_STATS(mbm, n_tmo);
+ fw_hdr->lo = htonl(FW_CMD_RETVAL(FW_ETIMEDOUT));
+
+ return mbp;
+}
+
+/*
+ * csio_mb_cancel_all - Cancel all waiting commands.
+ * @hw: The HW structure
+ * @cbfn_q: The callback queue.
+ *
+ * Caller should hold hw lock across this call.
+ */
+void
+csio_mb_cancel_all(struct csio_hw *hw, struct list_head *cbfn_q)
+{
+ struct csio_mb *mbp;
+ struct csio_mbm *mbm = &hw->mbm;
+ struct fw_cmd_hdr *hdr;
+ struct list_head *tmp;
+
+ if (mbm->mcurrent) {
+ mbp = mbm->mcurrent;
+
+ /* Stop mailbox completion timer */
+ del_timer_sync(&mbm->timer);
+
+ /* Add completion to tail of cbfn queue */
+ list_add_tail(&mbp->list, cbfn_q);
+ mbm->mcurrent = NULL;
+ }
+
+ if (!list_empty(&mbm->req_q)) {
+ list_splice_tail_init(&mbm->req_q, cbfn_q);
+ mbm->stats.n_activeq = 0;
+ }
+
+ if (!list_empty(&mbm->cbfn_q)) {
+ list_splice_tail_init(&mbm->cbfn_q, cbfn_q);
+ mbm->stats.n_cbfnq = 0;
+ }
+
+ if (list_empty(cbfn_q))
+ return;
+
+ list_for_each(tmp, cbfn_q) {
+ mbp = (struct csio_mb *)tmp;
+ hdr = (struct fw_cmd_hdr *)(mbp->mb);
+
+ csio_dbg(hw, "Cancelling pending mailbox num %x op:%x\n",
+ hw->pfn, FW_CMD_OP_GET(ntohl(hdr->hi)));
+
+ CSIO_INC_STATS(mbm, n_cancel);
+ hdr->lo = htonl(FW_CMD_RETVAL(FW_HOSTERROR));
+ }
+}
+
+/*
+ * csio_mbm_init - Initialize Mailbox module
+ * @mbm: Mailbox module
+ * @hw: The HW structure
+ * @timer: Timing function for interrupting mailboxes
+ *
+ * Initialize timer and the request/response queues.
+ */
+int
+csio_mbm_init(struct csio_mbm *mbm, struct csio_hw *hw,
+ void (*timer_fn)(uintptr_t))
+{
+ struct timer_list *timer = &mbm->timer;
+
+ init_timer(timer);
+ timer->function = timer_fn;
+ timer->data = (unsigned long)hw;
+
+ INIT_LIST_HEAD(&mbm->req_q);
+ INIT_LIST_HEAD(&mbm->cbfn_q);
+ csio_set_mb_intr_idx(mbm, -1);
+
+ return 0;
+}
+
+/*
+ * csio_mbm_exit - Uninitialize mailbox module
+ * @mbm: Mailbox module
+ *
+ * Stop timer.
+ */
+void
+csio_mbm_exit(struct csio_mbm *mbm)
+{
+ del_timer_sync(&mbm->timer);
+
+ CSIO_DB_ASSERT(mbm->mcurrent == NULL);
+ CSIO_DB_ASSERT(list_empty(&mbm->req_q));
+ CSIO_DB_ASSERT(list_empty(&mbm->cbfn_q));
+}
--
1.7.1
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox