* [PATCH net-next-2.6 3/7] be2net: Add error recovery during load for Lancer
From: Padmanabh Ratnakar @ 2011-03-07 13:08 UTC (permalink / raw)
To: netdev, davem; +Cc: Padmanabh Ratnakar, Sathya Perla, Subramanian Seetharaman
Add error recovery during load for Lancer
Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: Subramanian Seetharaman <subbu.seetharaman@emulex.com>
---
drivers/net/benet/be_hw.h | 12 +++++++++
drivers/net/benet/be_main.c | 56 +++++++++++++++++++++++++++++++++++++++++++
2 files changed, 68 insertions(+), 0 deletions(-)
diff --git a/drivers/net/benet/be_hw.h b/drivers/net/benet/be_hw.h
index 3f459f7..dbe67f3 100644
--- a/drivers/net/benet/be_hw.h
+++ b/drivers/net/benet/be_hw.h
@@ -44,6 +44,18 @@
#define POST_STAGE_BE_RESET 0x3 /* Host wants to reset chip */
#define POST_STAGE_ARMFW_RDY 0xc000 /* FW is done with POST */
+
+/* Lancer SLIPORT_CONTROL SLIPORT_STATUS registers */
+#define SLIPORT_STATUS_OFFSET 0x404
+#define SLIPORT_CONTROL_OFFSET 0x408
+
+#define SLIPORT_STATUS_ERR_MASK 0x80000000
+#define SLIPORT_STATUS_RN_MASK 0x01000000
+#define SLIPORT_STATUS_RDY_MASK 0x00800000
+
+
+#define SLI_PORT_CONTROL_IP_MASK 0x08000000
+
/********* Memory BAR register ************/
#define PCICFG_MEMBAR_CTRL_INT_CTRL_OFFSET 0xfc
/* Host Interrupt Enable, if set interrupts are enabled although "PCI Interrupt
diff --git a/drivers/net/benet/be_main.c b/drivers/net/benet/be_main.c
index ac7ae21..665a9b7 100644
--- a/drivers/net/benet/be_main.c
+++ b/drivers/net/benet/be_main.c
@@ -2901,6 +2901,54 @@ static int be_dev_family_check(struct be_adapter *adapter)
return 0;
}
+static int lancer_wait_ready(struct be_adapter *adapter)
+{
+#define SLIPORT_READY_TIMEOUT 500
+ u32 sliport_status;
+ int status = 0, i;
+
+ for (i = 0; i < SLIPORT_READY_TIMEOUT; i++) {
+ sliport_status = ioread32(adapter->db + SLIPORT_STATUS_OFFSET);
+ if (sliport_status & SLIPORT_STATUS_RDY_MASK)
+ break;
+
+ msleep(20);
+ }
+
+ if (i == SLIPORT_READY_TIMEOUT)
+ status = -1;
+
+ return status;
+}
+
+static int lancer_test_and_set_rdy_state(struct be_adapter *adapter)
+{
+ int status;
+ u32 sliport_status, err, reset_needed;
+ status = lancer_wait_ready(adapter);
+ if (!status) {
+ sliport_status = ioread32(adapter->db + SLIPORT_STATUS_OFFSET);
+ err = sliport_status & SLIPORT_STATUS_ERR_MASK;
+ reset_needed = sliport_status & SLIPORT_STATUS_RN_MASK;
+ if (err && reset_needed) {
+ iowrite32(SLI_PORT_CONTROL_IP_MASK,
+ adapter->db + SLIPORT_CONTROL_OFFSET);
+
+ /* check adapter has corrected the error */
+ status = lancer_wait_ready(adapter);
+ sliport_status = ioread32(adapter->db +
+ SLIPORT_STATUS_OFFSET);
+ sliport_status &= (SLIPORT_STATUS_ERR_MASK |
+ SLIPORT_STATUS_RN_MASK);
+ if (status || sliport_status)
+ status = -1;
+ } else if (err || reset_needed) {
+ status = -1;
+ }
+ }
+ return status;
+}
+
static int __devinit be_probe(struct pci_dev *pdev,
const struct pci_device_id *pdev_id)
{
@@ -2950,6 +2998,14 @@ static int __devinit be_probe(struct pci_dev *pdev,
if (status)
goto free_netdev;
+ if (lancer_chip(adapter)) {
+ status = lancer_test_and_set_rdy_state(adapter);
+ if (status) {
+ dev_err(&pdev->dev, "Adapter in non recoverable error\n");
+ goto free_netdev;
+ }
+ }
+
/* sync up with fw's ready state */
if (be_physfn(adapter)) {
status = be_cmd_POST(adapter);
--
1.6.0.2
^ permalink raw reply related
* [PATCH net-next-2.6 4/7] be2net: Change f/w command versions for Lancer
From: Padmanabh Ratnakar @ 2011-03-07 13:08 UTC (permalink / raw)
To: netdev, davem; +Cc: Padmanabh Ratnakar, Sathya Perla, Subramanian Seetharaman
Change f/w command versions for Lancer
Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: Subramanian Seetharaman <subbu.seetharaman@emulex.com>
---
drivers/net/benet/be_cmds.c | 8 +++++++-
drivers/net/benet/be_cmds.h | 2 +-
2 files changed, 8 insertions(+), 2 deletions(-)
diff --git a/drivers/net/benet/be_cmds.c b/drivers/net/benet/be_cmds.c
index 1822ecd..cc3a235 100644
--- a/drivers/net/benet/be_cmds.c
+++ b/drivers/net/benet/be_cmds.c
@@ -726,7 +726,7 @@ int be_cmd_cq_create(struct be_adapter *adapter,
req->num_pages = cpu_to_le16(PAGES_4K_SPANNED(q_mem->va, q_mem->size));
if (lancer_chip(adapter)) {
- req->hdr.version = 1;
+ req->hdr.version = 2;
req->page_size = 1; /* 1 for 4K */
AMAP_SET_BITS(struct amap_cq_context_lancer, coalescwm, ctxt,
coalesce_wm);
@@ -862,6 +862,12 @@ int be_cmd_txq_create(struct be_adapter *adapter,
be_cmd_hdr_prepare(&req->hdr, CMD_SUBSYSTEM_ETH, OPCODE_ETH_TX_CREATE,
sizeof(*req));
+ if (lancer_chip(adapter)) {
+ req->hdr.version = 1;
+ AMAP_SET_BITS(struct amap_tx_context, if_id, ctxt,
+ adapter->if_handle);
+ }
+
req->num_pages = PAGES_4K_SPANNED(q_mem->va, q_mem->size);
req->ulp_num = BE_ULP1_NUM;
req->type = BE_ETH_TX_RING_TYPE_STANDARD;
diff --git a/drivers/net/benet/be_cmds.h b/drivers/net/benet/be_cmds.h
index 93e5768..5bdaa22 100644
--- a/drivers/net/benet/be_cmds.h
+++ b/drivers/net/benet/be_cmds.h
@@ -430,7 +430,7 @@ struct be_cmd_resp_mcc_create {
/* Pseudo amap definition in which each bit of the actual structure is defined
* as a byte: used to calculate offset/shift/mask of each field */
struct amap_tx_context {
- u8 rsvd0[16]; /* dword 0 */
+ u8 if_id[16]; /* dword 0 */
u8 tx_ring_size[4]; /* dword 0 */
u8 rsvd1[26]; /* dword 0 */
u8 pci_func_id[8]; /* dword 1 */
--
1.6.0.2
^ permalink raw reply related
* [PATCH net-next-2.6 2/7] be2net: Checksum field valid only for TCP/UDP
From: Padmanabh Ratnakar @ 2011-03-07 13:08 UTC (permalink / raw)
To: netdev, davem; +Cc: Padmanabh Ratnakar, Sathya Perla, Subramanian Seetharaman
L4 checksum field is valid only for TCP/UDP packets in Lancer
Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: Subramanian Seetharaman <subbu.seetharaman@emulex.com>
---
drivers/net/benet/be_main.c | 9 ++++++---
1 files changed, 6 insertions(+), 3 deletions(-)
diff --git a/drivers/net/benet/be_main.c b/drivers/net/benet/be_main.c
index bf34434..ac7ae21 100644
--- a/drivers/net/benet/be_main.c
+++ b/drivers/net/benet/be_main.c
@@ -865,14 +865,17 @@ static void be_rx_stats_update(struct be_rx_obj *rxo,
static inline bool csum_passed(struct be_eth_rx_compl *rxcp)
{
- u8 l4_cksm, ipv6, ipcksm;
+ u8 l4_cksm, ipv6, ipcksm, tcpf, udpf;
l4_cksm = AMAP_GET_BITS(struct amap_eth_rx_compl, l4_cksm, rxcp);
ipcksm = AMAP_GET_BITS(struct amap_eth_rx_compl, ipcksm, rxcp);
ipv6 = AMAP_GET_BITS(struct amap_eth_rx_compl, ip_version, rxcp);
+ tcpf = AMAP_GET_BITS(struct amap_eth_rx_compl, tcpf, rxcp);
+ udpf = AMAP_GET_BITS(struct amap_eth_rx_compl, udpf, rxcp);
- /* Ignore ipcksm for ipv6 pkts */
- return l4_cksm && (ipcksm || ipv6);
+ /* L4 checksum is not reliable for non TCP/UDP packets.
+ * Also ignore ipcksm for ipv6 pkts */
+ return (tcpf || udpf) && l4_cksm && (ipcksm || ipv6);
}
static struct be_rx_page_info *
--
1.6.0.2
^ permalink raw reply related
* [PATCH net-next-2.6 1/7] be2net: Remove ERR compl workaround for Lancer
From: Padmanabh Ratnakar @ 2011-03-07 13:07 UTC (permalink / raw)
To: netdev, davem; +Cc: Padmanabh Ratnakar, Sathya Perla, Subramanian Seetharaman
Workaround added for Lancer in handling RX ERR completion received
when no RX buffers are posted is not needed.
Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: Subramanian Seetharaman <subbu.seetharaman@emulex.com>
---
drivers/net/benet/be.h | 4 +---
drivers/net/benet/be_main.c | 31 ++++++++-----------------------
2 files changed, 9 insertions(+), 26 deletions(-)
diff --git a/drivers/net/benet/be.h b/drivers/net/benet/be.h
index ed709a5..4ac0d72 100644
--- a/drivers/net/benet/be.h
+++ b/drivers/net/benet/be.h
@@ -220,9 +220,7 @@ struct be_rx_obj {
struct be_rx_stats stats;
u8 rss_id;
bool rx_post_starved; /* Zero rx frags have been posted to BE */
- u16 last_frag_index;
- u16 rsvd;
- u32 cache_line_barrier[15];
+ u32 cache_line_barrier[16];
};
struct be_drv_stats {
diff --git a/drivers/net/benet/be_main.c b/drivers/net/benet/be_main.c
index ef66dc6..bf34434 100644
--- a/drivers/net/benet/be_main.c
+++ b/drivers/net/benet/be_main.c
@@ -909,17 +909,11 @@ static void be_rx_compl_discard(struct be_adapter *adapter,
rxq_idx = AMAP_GET_BITS(struct amap_eth_rx_compl, fragndx, rxcp);
num_rcvd = AMAP_GET_BITS(struct amap_eth_rx_compl, numfrags, rxcp);
- /* Skip out-of-buffer compl(lancer) or flush compl(BE) */
- if (likely(rxq_idx != rxo->last_frag_index && num_rcvd != 0)) {
-
- rxo->last_frag_index = rxq_idx;
-
- for (i = 0; i < num_rcvd; i++) {
- page_info = get_rx_page_info(adapter, rxo, rxq_idx);
- put_page(page_info->page);
- memset(page_info, 0, sizeof(*page_info));
- index_inc(&rxq_idx, rxq->len);
- }
+ for (i = 0; i < num_rcvd; i++) {
+ page_info = get_rx_page_info(adapter, rxo, rxq_idx);
+ put_page(page_info->page);
+ memset(page_info, 0, sizeof(*page_info));
+ index_inc(&rxq_idx, rxq->len);
}
}
@@ -1579,9 +1573,6 @@ static int be_rx_queues_create(struct be_adapter *adapter)
adapter->big_page_size = (1 << get_order(rx_frag_size)) * PAGE_SIZE;
for_all_rx_queues(adapter, rxo, i) {
rxo->adapter = adapter;
- /* Init last_frag_index so that the frag index in the first
- * completion will never match */
- rxo->last_frag_index = 0xffff;
rxo->rx_eq.max_eqd = BE_MAX_EQD;
rxo->rx_eq.enable_aic = true;
@@ -1722,7 +1713,7 @@ static int be_poll_rx(struct napi_struct *napi, int budget)
struct be_queue_info *rx_cq = &rxo->cq;
struct be_eth_rx_compl *rxcp;
u32 work_done;
- u16 frag_index, num_rcvd;
+ u16 num_rcvd;
u8 err;
rxo->stats.rx_polls++;
@@ -1732,16 +1723,10 @@ static int be_poll_rx(struct napi_struct *napi, int budget)
break;
err = AMAP_GET_BITS(struct amap_eth_rx_compl, err, rxcp);
- frag_index = AMAP_GET_BITS(struct amap_eth_rx_compl, fragndx,
- rxcp);
num_rcvd = AMAP_GET_BITS(struct amap_eth_rx_compl, numfrags,
rxcp);
-
- /* Skip out-of-buffer compl(lancer) or flush compl(BE) */
- if (likely(frag_index != rxo->last_frag_index &&
- num_rcvd != 0)) {
- rxo->last_frag_index = frag_index;
-
+ /* Ignore flush completions */
+ if (num_rcvd) {
if (do_gro(rxo, rxcp, err))
be_rx_compl_process_gro(adapter, rxo, rxcp);
else
--
1.6.0.2
^ permalink raw reply related
* [PATCH net-next-2.6 0/7] be2net: Patch series for supporting Lancer
From: Padmanabh Ratnakar @ 2011-03-07 13:05 UTC (permalink / raw)
To: netdev, davem; +Cc: Padmanabh Ratnakar, Sathya Perla, Subramanian Seetharaman
Hi David,
Following are patches to support Lancer family of CNA.
Please apply.
Thanks,
Padmanabh
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: Subramanian Seetharaman <subbu.seetharaman@emulex.com>
Padmanabh Ratnakar (7):
be2net: Remove ERR compl workaround for Lancer
be2net: Checksum field valid only for TCP/UDP
be2net: Add error recovery during load for Lancer
be2net: Change f/w command versions for Lancer
be2net: Remove TX Queue stop in close
be2net: Disarm CQ and EQ to disable interrupt in Lancer
be2net: Add multicast filter capability for Lancer
drivers/net/benet/be.h | 4 +-
drivers/net/benet/be_cmds.c | 8 +++-
drivers/net/benet/be_cmds.h | 5 +-
drivers/net/benet/be_hw.h | 12 ++++
drivers/net/benet/be_main.c | 118 +++++++++++++++++++++++++++++++------------
5 files changed, 108 insertions(+), 39 deletions(-)
^ permalink raw reply
* Re: tun routing is broken
From: Jiri Kosina @ 2011-03-07 12:54 UTC (permalink / raw)
To: Jiri Slaby, Andrew Morton
Cc: David Miller, Jiri Slaby, linux-kernel, mm-commits, netdev, maxk
In-Reply-To: <4D71F194.9010402@suse.cz>
On Sat, 5 Mar 2011, Jiri Slaby wrote:
> > Date: Sat, 05 Mar 2011 09:04:22 +0100
> >
> >> Ok, so I booted the new kernel, and tun is broken there completely. If I
> >> try to ping a vpn peer:
> >
> > -mm tree is missing this fix which went in yesterday. Please if
> > you are going to be testing networking a lot, test against net-next-2.6
> > instead of Andrew's tree which invariable lags behind:
>
> I'm not testing networking, I'm just using my desktop :). (And
> next/master is mostly unusable for these needs. In comparison to mmotm
> which is some kind of a "release".)
I have been hit by similar issues several times already as well.
I guess that question of including -mm into linux-next should be raised
again here ... Andrew, if I remember correctly, on last kernel summit you
were stating that you'd stop eating untill you have -mm included in -next.
Seems like you must be quite hungry these days already :)
Thanks,
--
Jiri Kosina
SUSE Labs, Novell Inc.
^ permalink raw reply
* [patch net-next-2.6] net: reinject arps into bonding slave instead of master
From: Jiri Pirko @ 2011-03-07 12:51 UTC (permalink / raw)
To: Nicolas de Pesloüan
Cc: netdev, davem, shemminger, kaber, fubar, eric.dumazet, andy
In-Reply-To: <20110306133413.GB2795@psychotron.redhat.com>
Recent patch "bonding: move processing of recv handlers into
handle_frame()" caused a regression on following net scheme:
eth0 - bond0 - bond0.5
where arp monitoring is happening over vlan. This patch fixes it by
reinjecting the arp packet into bonding slave device so the bonding
rx_handler can pickup and process it.
Signed-off-by: Jiri Pirko <jpirko@redhat.com>
---
net/core/dev.c | 8 ++++----
1 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/net/core/dev.c b/net/core/dev.c
index c71bd18..3d88458 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -3094,12 +3094,12 @@ void netdev_rx_handler_unregister(struct net_device *dev)
}
EXPORT_SYMBOL_GPL(netdev_rx_handler_unregister);
-static void vlan_on_bond_hook(struct sk_buff *skb)
+static void vlan_on_bond_hook(struct sk_buff *skb, struct net_device *orig_dev)
{
/*
* Make sure ARP frames received on VLAN interfaces stacked on
* bonding interfaces still make their way to any base bonding
- * device that may have registered for a specific ptype.
+ * device by reinjecting the frame into bonding slave (orig_dev)
*/
if (skb->dev->priv_flags & IFF_802_1Q_VLAN &&
vlan_dev_real_dev(skb->dev)->priv_flags & IFF_BONDING &&
@@ -3108,7 +3108,7 @@ static void vlan_on_bond_hook(struct sk_buff *skb)
if (!skb2)
return;
- skb2->dev = vlan_dev_real_dev(skb->dev);
+ skb2->dev = orig_dev;
netif_rx(skb2);
}
}
@@ -3202,7 +3202,7 @@ ncls:
goto out;
}
- vlan_on_bond_hook(skb);
+ vlan_on_bond_hook(skb, orig_dev);
/* deliver only exact match when indicated */
null_or_dev = deliver_exact ? skb->dev : NULL;
--
1.7.4
^ permalink raw reply related
* Re: [PATCH 06/17] mlx4_en: using new mlx4 interrupt scheme
From: Ben Hutchings @ 2011-03-07 12:41 UTC (permalink / raw)
To: Yevgeny Petrilin; +Cc: davem, netdev, markuze
In-Reply-To: <4D74B0AE.5070403@mellanox.co.il>
On Mon, 2011-03-07 at 12:17 +0200, Yevgeny Petrilin wrote:
> Each RX ring will have its own interrupt vector, and TX rings will share one
> (we mostly use polling for TX completions).
> The vectors are assigned first time device is opened, and its name includes
> the interface name and ring number.
[...]
> --- a/drivers/net/mlx4/en_main.c
> +++ b/drivers/net/mlx4/en_main.c
[...]
> @@ -601,10 +602,19 @@ int mlx4_en_start_port(struct net_device *dev)
> goto cq_err;
> }
>
> + if (mdev->dev->caps.comp_pool && !priv->tx_vector) {
> + sprintf(name , "%s_tx", priv->dev->name);
> + if (mlx4_assign_eq(mdev->dev , name, &priv->tx_vector)) {
> + mlx4_warn(mdev, "Failed Assigning an EQ to "
> + "%s_tx ,Falling back to legacy "
> + "EQ's\n", priv->dev->name);
> + }
> + }
[...]
If this string is being copied to the IRQ name, it should be
"<ifname>-tx" not "<ifname>_tx" for consistency with other drivers.
Ben.
--
Ben Hutchings, Senior Software Engineer, Solarflare Communications
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
* [PATCH 8/8] Phonet: kill the ST-Ericsson pipe controller Kconfig
From: Rémi Denis-Courmont @ 2011-03-07 10:46 UTC (permalink / raw)
To: netdev
In-Reply-To: <201103071245.47722.remi.denis-courmont@nokia.com>
This is now a run-time choice so that a single kernel can support both
old and new generation ISI modems.
Signed-off-by: Rémi Denis-Courmont <remi.denis-courmont@nokia.com>
---
Documentation/networking/phonet.txt | 13 -------------
include/linux/phonet.h | 2 --
net/phonet/Kconfig | 12 ------------
net/phonet/pep.c | 25 -------------------------
4 files changed, 0 insertions(+), 52 deletions(-)
diff --git a/Documentation/networking/phonet.txt b/Documentation/networking/phonet.txt
index 3d12779..8100358 100644
--- a/Documentation/networking/phonet.txt
+++ b/Documentation/networking/phonet.txt
@@ -205,19 +205,6 @@ The pipe protocol provides two socket options at the SOL_PNPIPE level:
socket descriptors that are already connected or being connected.
-Phonet Pipe-controller Implementation
--------------------------------------
-
-Phonet Pipe-controller is enabled by selecting the CONFIG_PHONET_PIPECTRLR
-Kconfig option.
-
-The implementation adds socket options at SOL_PNPIPE level:
-
- PNPIPE_ENABLE accepts one integer value (int). If set to zero, the pipe
- is disabled. If the value is non-zero, the pipe is enabled. If the pipe
- is not (yet) connected, ENOTCONN is error is returned.
-
-
Authors
-------
diff --git a/include/linux/phonet.h b/include/linux/phonet.h
index 32a0965..6fb1384 100644
--- a/include/linux/phonet.h
+++ b/include/linux/phonet.h
@@ -37,8 +37,6 @@
#define PNPIPE_ENCAP 1
#define PNPIPE_IFINDEX 2
#define PNPIPE_HANDLE 3
-#define PNPIPE_ENABLE 4
-/* unused slot */
#define PNADDR_ANY 0
#define PNADDR_BROADCAST 0xFC
diff --git a/net/phonet/Kconfig b/net/phonet/Kconfig
index 0d9b8a2..6ec7d55 100644
--- a/net/phonet/Kconfig
+++ b/net/phonet/Kconfig
@@ -14,15 +14,3 @@ config PHONET
To compile this driver as a module, choose M here: the module
will be called phonet. If unsure, say N.
-
-config PHONET_PIPECTRLR
- bool "Phonet Pipe Controller (EXPERIMENTAL)"
- depends on PHONET && EXPERIMENTAL
- default N
- help
- The Pipe Controller implementation in Phonet stack to support Pipe
- data with Nokia Slim modems like WG2.5 used on ST-Ericsson U8500
- platform.
-
- This option is incompatible with older Nokia modems.
- Say N here unless you really know what you are doing.
diff --git a/net/phonet/pep.c b/net/phonet/pep.c
index 671effb..68e635f 100644
--- a/net/phonet/pep.c
+++ b/net/phonet/pep.c
@@ -167,15 +167,6 @@ static int pipe_handler_send_created_ind(struct sock *sk)
data, 4, GFP_ATOMIC);
}
-#ifdef CONFIG_PHONET_PIPECTRLR
-static int pipe_handler_enable_pipe(struct sock *sk, int enable)
-{
- u8 id = enable ? PNS_PEP_ENABLE_REQ : PNS_PEP_DISABLE_REQ;
-
- return pipe_handler_request(sk, id, PAD, NULL, 0);
-}
-#endif
-
static int pep_accept_conn(struct sock *sk, struct sk_buff *skb)
{
static const u8 data[20] = {
@@ -968,16 +959,6 @@ static int pep_setsockopt(struct sock *sk, int level, int optname,
}
goto out_norel;
-#ifdef CONFIG_PHONET_PIPECTRLR
- case PNPIPE_ENABLE:
- if ((1 << sk->sk_state) & ~(TCPF_SYN_RECV|TCPF_ESTABLISHED)) {
- err = -ENOTCONN;
- break;
- }
- err = pipe_handler_enable_pipe(sk, val);
- break;
-#endif
-
default:
err = -ENOPROTOOPT;
}
@@ -1013,12 +994,6 @@ static int pep_getsockopt(struct sock *sk, int level, int optname,
return -EINVAL;
break;
-#ifdef CONFIG_PHONET_PIPECTRLR
- case PNPIPE_ENABLE:
- val = sk->sk_state == TCP_ESTABLISHED;
- break;
-#endif
-
default:
return -ENOPROTOOPT;
}
--
1.7.1
^ permalink raw reply related
* [PATCH 6/8] Phonet: provide pipe socket option to retrieve the pipe identifier
From: Rémi Denis-Courmont @ 2011-03-07 10:46 UTC (permalink / raw)
To: netdev
In-Reply-To: <201103071245.47722.remi.denis-courmont@nokia.com>
User-space sometimes needs this information.
Signed-off-by: Rémi Denis-Courmont <remi.denis-courmont@nokia.com>
---
Documentation/networking/phonet.txt | 7 ++++---
include/linux/phonet.h | 2 +-
net/phonet/pep.c | 15 +++++++--------
3 files changed, 12 insertions(+), 12 deletions(-)
diff --git a/Documentation/networking/phonet.txt b/Documentation/networking/phonet.txt
index 24ad2ad..cacac96 100644
--- a/Documentation/networking/phonet.txt
+++ b/Documentation/networking/phonet.txt
@@ -181,6 +181,10 @@ The pipe protocol provides two socket options at the SOL_PNPIPE level:
interface index of the network interface created by PNPIPE_ENCAP,
or zero if encapsulation is off.
+ PNPIPE_HANDLE is a read-only integer value. It contains the underlying
+ identifier ("pipe handle") of the pipe. This is only defined for
+ socket descriptors that are already connected or being connected.
+
Phonet Pipe-controller Implementation
-------------------------------------
@@ -199,9 +203,6 @@ between itself and a remote pipe-end point (e.g. modem).
The implementation adds socket options at SOL_PNPIPE level:
- PNPIPE_PIPE_HANDLE
- It accepts an integer argument for setting value of pipe handle.
-
PNPIPE_ENABLE accepts one integer value (int). If set to zero, the pipe
is disabled. If the value is non-zero, the pipe is enabled. If the pipe
is not (yet) connected, ENOTCONN is error is returned.
diff --git a/include/linux/phonet.h b/include/linux/phonet.h
index 26c8df7..32a0965 100644
--- a/include/linux/phonet.h
+++ b/include/linux/phonet.h
@@ -36,7 +36,7 @@
/* Socket options for SOL_PNPIPE level */
#define PNPIPE_ENCAP 1
#define PNPIPE_IFINDEX 2
-#define PNPIPE_PIPE_HANDLE 3
+#define PNPIPE_HANDLE 3
#define PNPIPE_ENABLE 4
/* unused slot */
diff --git a/net/phonet/pep.c b/net/phonet/pep.c
index c0fab4c..abfb795 100644
--- a/net/phonet/pep.c
+++ b/net/phonet/pep.c
@@ -853,6 +853,7 @@ static int pep_sock_connect(struct sock *sk, struct sockaddr *addr, int len)
pn->pn_sk.dobject = pn_sockaddr_get_object(spn);
pn->pn_sk.resource = pn_sockaddr_get_resource(spn);
+ pn->pipe_handle = 1; /* anything but INVALID_HANDLE */
return pipe_handler_request(sk, PNS_PEP_CONNECT_REQ,
PN_PIPE_DISABLE, data, 4);
}
@@ -909,14 +910,6 @@ static int pep_setsockopt(struct sock *sk, int level, int optname,
lock_sock(sk);
switch (optname) {
-#ifdef CONFIG_PHONET_PIPECTRLR
- case PNPIPE_PIPE_HANDLE:
- if (val) {
- pn->pipe_handle = val;
- break;
- }
-#endif
-
case PNPIPE_ENCAP:
if (val && val != PNPIPE_ENCAP_IP) {
err = -EINVAL;
@@ -982,6 +975,12 @@ static int pep_getsockopt(struct sock *sk, int level, int optname,
val = pn->ifindex;
break;
+ case PNPIPE_HANDLE:
+ val = pn->pipe_handle;
+ if (val == PN_PIPE_INVALID_HANDLE)
+ return -EINVAL;
+ break;
+
#ifdef CONFIG_PHONET_PIPECTRLR
case PNPIPE_ENABLE:
val = sk->sk_state == TCP_ESTABLISHED;
--
1.7.1
^ permalink raw reply related
* [PATCH 7/8] Phonet: support active connection without pipe controller on modem
From: Rémi Denis-Courmont @ 2011-03-07 10:46 UTC (permalink / raw)
To: netdev
In-Reply-To: <201103071245.47722.remi.denis-courmont@nokia.com>
This provides support for newer ISI modems with no need for the
earlier experimental compile-time alternative choice. This also fixes
a number of remaining bugs in the old ST-E code.
Signed-off-by: Rémi Denis-Courmont <remi.denis-courmont@nokia.com>
---
Documentation/networking/phonet.txt | 53 +++++------
net/phonet/pep.c | 172 ++++++++++++++++++++--------------
net/phonet/socket.c | 102 ++++++++-------------
3 files changed, 165 insertions(+), 162 deletions(-)
diff --git a/Documentation/networking/phonet.txt b/Documentation/networking/phonet.txt
index cacac96..3d12779 100644
--- a/Documentation/networking/phonet.txt
+++ b/Documentation/networking/phonet.txt
@@ -154,9 +154,28 @@ connections, one per accept()'d socket.
write(cfd, msg, msglen);
}
-Connections are established between two endpoints by a "third party"
-application. This means that both endpoints are passive; so connect()
-is not possible.
+Connections are traditionally established between two endpoints by a
+"third party" application. This means that both endpoints are passive.
+
+
+As of Linux kernel version 2.6.39, it is also possible to connect
+two endpoints directly, using connect() on the active side. This is
+intended to support the newer Nokia Wireless Modem API, as found in
+e.g. the Nokia Slim Modem in the ST-Ericsson U8500 platform:
+
+ struct sockaddr_spn spn;
+ int fd;
+
+ fd = socket(PF_PHONET, SOCK_SEQPACKET, PN_PROTO_PIPE);
+ memset(&spn, 0, sizeof(spn));
+ spn.spn_family = AF_PHONET;
+ spn.spn_obj = ...;
+ spn.spn_dev = ...;
+ spn.spn_resource = 0xD9;
+ connect(fd, (struct sockaddr *)&spn, sizeof(spn));
+ /* normal I/O here ... */
+ close(fd);
+
WARNING:
When polling a connected pipe socket for writability, there is an
@@ -189,17 +208,8 @@ The pipe protocol provides two socket options at the SOL_PNPIPE level:
Phonet Pipe-controller Implementation
-------------------------------------
-Phonet Pipe-controller is enabled by selecting the CONFIG_PHONET_PIPECTRLR Kconfig
-option. It is useful when communicating with those Nokia Modems which do not
-implement Pipe controller in them e.g. Nokia Slim Modem used in ST-Ericsson
-U8500 platform.
-
-The implementation is based on the Data Connection Establishment Sequence
-depicted in 'Nokia Wireless Modem API - Wireless_modem_user_guide.pdf'
-document.
-
-It allows a phonet sequenced socket (host-pep) to initiate a Pipe connection
-between itself and a remote pipe-end point (e.g. modem).
+Phonet Pipe-controller is enabled by selecting the CONFIG_PHONET_PIPECTRLR
+Kconfig option.
The implementation adds socket options at SOL_PNPIPE level:
@@ -207,21 +217,6 @@ The implementation adds socket options at SOL_PNPIPE level:
is disabled. If the value is non-zero, the pipe is enabled. If the pipe
is not (yet) connected, ENOTCONN is error is returned.
-The implementation also adds socket 'connect'. On calling the 'connect', pipe
-will be created between the source socket and the destination, and the pipe
-state will be set to PIPE_DISABLED.
-
-After a pipe has been created and enabled successfully, the Pipe data can be
-exchanged between the host-pep and remote-pep (modem).
-
-User-space would typically follow below sequence with Pipe controller:-
--socket
--bind
--setsockopt for PNPIPE_PIPE_HANDLE
--connect
--setsockopt for PNPIPE_ENCAP_IP
--setsockopt for PNPIPE_ENABLE
-
Authors
-------
diff --git a/net/phonet/pep.c b/net/phonet/pep.c
index abfb795..671effb 100644
--- a/net/phonet/pep.c
+++ b/net/phonet/pep.c
@@ -136,7 +136,6 @@ static int pep_indicate(struct sock *sk, u8 id, u8 code,
#define PAD 0x00
-#ifdef CONFIG_PHONET_PIPECTRLR
static int pipe_handler_request(struct sock *sk, u8 id, u8 code,
const void *data, int len)
{
@@ -168,11 +167,7 @@ static int pipe_handler_send_created_ind(struct sock *sk)
data, 4, GFP_ATOMIC);
}
-static int pipe_handler_send_ind(struct sock *sk, u8 id)
-{
- return pep_indicate(sk, id, PAD, NULL, 0, GFP_ATOMIC);
-}
-
+#ifdef CONFIG_PHONET_PIPECTRLR
static int pipe_handler_enable_pipe(struct sock *sk, int enable)
{
u8 id = enable ? PNS_PEP_ENABLE_REQ : PNS_PEP_DISABLE_REQ;
@@ -376,32 +371,11 @@ static int pipe_do_rcv(struct sock *sk, struct sk_buff *skb)
sk->sk_state_change(sk);
break;
-#ifdef CONFIG_PHONET_PIPECTRLR
- case PNS_PEP_DISCONNECT_RESP:
- sk->sk_state = TCP_CLOSE;
- break;
-#endif
-
case PNS_PEP_ENABLE_REQ:
/* Wait for PNS_PIPE_(ENABLED|REDIRECTED)_IND */
pep_reply(sk, skb, PN_PIPE_NO_ERROR, NULL, 0, GFP_ATOMIC);
break;
-#ifdef CONFIG_PHONET_PIPECTRLR
- case PNS_PEP_ENABLE_RESP:
- pipe_handler_send_ind(sk, PNS_PIPE_ENABLED_IND);
-
- if (!pn_flow_safe(pn->tx_fc)) {
- atomic_set(&pn->tx_credits, 1);
- sk->sk_write_space(sk);
- }
- if (sk->sk_state == TCP_ESTABLISHED)
- break; /* Nothing to do */
- sk->sk_state = TCP_ESTABLISHED;
- pipe_grant_credits(sk, GFP_ATOMIC);
- break;
-#endif
-
case PNS_PEP_RESET_REQ:
switch (hdr->state_after_reset) {
case PN_PIPE_DISABLE:
@@ -420,15 +394,6 @@ static int pipe_do_rcv(struct sock *sk, struct sk_buff *skb)
pep_reply(sk, skb, PN_PIPE_NO_ERROR, NULL, 0, GFP_ATOMIC);
break;
-#ifdef CONFIG_PHONET_PIPECTRLR
- case PNS_PEP_DISABLE_RESP:
- atomic_set(&pn->tx_credits, 0);
- pipe_handler_send_ind(sk, PNS_PIPE_DISABLED_IND);
- sk->sk_state = TCP_SYN_RECV;
- pn->rx_credits = 0;
- break;
-#endif
-
case PNS_PEP_CTRL_REQ:
if (skb_queue_len(&pn->ctrlreq_queue) >= PNPIPE_CTRLREQ_MAX) {
atomic_inc(&sk->sk_drops);
@@ -521,7 +486,6 @@ static void pipe_destruct(struct sock *sk)
skb_queue_purge(&pn->ctrlreq_queue);
}
-#ifdef CONFIG_PHONET_PIPECTRLR
static u8 pipe_negotiate_fc(const u8 *fcs, unsigned n)
{
unsigned i;
@@ -546,6 +510,8 @@ static int pep_connresp_rcv(struct sock *sk, struct sk_buff *skb)
return -EINVAL;
hdr = pnp_hdr(skb);
+ if (hdr->error_code != PN_PIPE_NO_ERROR)
+ return -ECONNREFUSED;
/* Parse sub-blocks */
n_sb = hdr->data[4];
@@ -573,14 +539,74 @@ static int pep_connresp_rcv(struct sock *sk, struct sk_buff *skb)
n_sb--;
}
- sk->sk_state = TCP_SYN_RECV;
- sk->sk_backlog_rcv = pipe_do_rcv;
- pn->rx_credits = 0;
- sk->sk_state_change(sk);
-
return pipe_handler_send_created_ind(sk);
}
-#endif
+
+/* Queue an skb to an actively connected sock.
+ * Socket lock must be held. */
+static int pipe_handler_do_rcv(struct sock *sk, struct sk_buff *skb)
+{
+ struct pep_sock *pn = pep_sk(sk);
+ struct pnpipehdr *hdr = pnp_hdr(skb);
+ int err = NET_RX_SUCCESS;
+
+ switch (hdr->message_id) {
+ case PNS_PIPE_ALIGNED_DATA:
+ __skb_pull(skb, 1);
+ /* fall through */
+ case PNS_PIPE_DATA:
+ __skb_pull(skb, 3); /* Pipe data header */
+ if (!pn_flow_safe(pn->rx_fc)) {
+ err = sock_queue_rcv_skb(sk, skb);
+ if (!err)
+ return NET_RX_SUCCESS;
+ err = NET_RX_DROP;
+ break;
+ }
+
+ if (pn->rx_credits == 0) {
+ atomic_inc(&sk->sk_drops);
+ err = NET_RX_DROP;
+ break;
+ }
+ pn->rx_credits--;
+ skb->dev = NULL;
+ skb_set_owner_r(skb, sk);
+ err = skb->len;
+ skb_queue_tail(&sk->sk_receive_queue, skb);
+ if (!sock_flag(sk, SOCK_DEAD))
+ sk->sk_data_ready(sk, err);
+ return NET_RX_SUCCESS;
+
+ case PNS_PEP_CONNECT_RESP:
+ if (sk->sk_state != TCP_SYN_SENT)
+ break;
+ if (!sock_flag(sk, SOCK_DEAD))
+ sk->sk_state_change(sk);
+ if (pep_connresp_rcv(sk, skb)) {
+ sk->sk_state = TCP_CLOSE_WAIT;
+ break;
+ }
+
+ sk->sk_state = TCP_ESTABLISHED;
+ if (!pn_flow_safe(pn->tx_fc)) {
+ atomic_set(&pn->tx_credits, 1);
+ sk->sk_write_space(sk);
+ }
+ pipe_grant_credits(sk, GFP_ATOMIC);
+ break;
+
+ case PNS_PEP_DISCONNECT_RESP:
+ /* sock should already be dead, nothing to do */
+ break;
+
+ case PNS_PEP_STATUS_IND:
+ pipe_rcv_status(sk, skb);
+ break;
+ }
+ kfree_skb(skb);
+ return err;
+}
/* Listening sock must be locked */
static struct sock *pep_find_pipe(const struct hlist_head *hlist,
@@ -649,12 +675,6 @@ static int pep_do_rcv(struct sock *sk, struct sk_buff *skb)
sk->sk_data_ready(sk, 0);
return NET_RX_SUCCESS;
-#ifdef CONFIG_PHONET_PIPECTRLR
- case PNS_PEP_CONNECT_RESP:
- pep_connresp_rcv(sk, skb);
- break;
-#endif
-
case PNS_PEP_DISCONNECT_REQ:
pep_reply(sk, skb, PN_PIPE_NO_ERROR, NULL, 0, GFP_ATOMIC);
break;
@@ -667,15 +687,19 @@ static int pep_do_rcv(struct sock *sk, struct sk_buff *skb)
case PNS_PEP_ENABLE_REQ:
case PNS_PEP_DISABLE_REQ:
/* invalid handle is not even allowed here! */
- default:
break;
+
+ default:
+ if ((1 << sk->sk_state)
+ & ~(TCPF_CLOSE|TCPF_LISTEN|TCPF_CLOSE_WAIT))
+ /* actively connected socket */
+ return pipe_handler_do_rcv(sk, skb);
}
drop:
kfree_skb(skb);
return NET_RX_SUCCESS;
}
-#ifndef CONFIG_PHONET_PIPECTRLR
static int pipe_do_remove(struct sock *sk)
{
struct pep_sock *pn = pep_sk(sk);
@@ -693,7 +717,6 @@ static int pipe_do_remove(struct sock *sk)
ph->data[0] = PAD;
return pn_skb_send(sk, skb, NULL);
}
-#endif
/* associated socket ceases to exist */
static void pep_sock_close(struct sock *sk, long timeout)
@@ -706,13 +729,12 @@ static void pep_sock_close(struct sock *sk, long timeout)
lock_sock(sk);
if ((1 << sk->sk_state) & (TCPF_SYN_RECV|TCPF_ESTABLISHED)) {
-#ifndef CONFIG_PHONET_PIPECTRLR
- /* Forcefully remove dangling Phonet pipe */
- pipe_do_remove(sk);
-#else
- /* send pep disconnect request */
- pipe_handler_request(sk, PNS_PEP_DISCONNECT_REQ, PAD, NULL, 0);
-#endif
+ if (sk->sk_backlog_rcv == pipe_do_rcv)
+ /* Forcefully remove dangling Phonet pipe */
+ pipe_do_remove(sk);
+ else
+ pipe_handler_request(sk, PNS_PEP_DISCONNECT_REQ, PAD,
+ NULL, 0);
}
sk->sk_state = TCP_CLOSE;
@@ -844,20 +866,22 @@ drop:
return newsk;
}
-#ifdef CONFIG_PHONET_PIPECTRLR
static int pep_sock_connect(struct sock *sk, struct sockaddr *addr, int len)
{
struct pep_sock *pn = pep_sk(sk);
- const struct sockaddr_pn *spn = (struct sockaddr_pn *)addr;
+ int err;
u8 data[4] = { 0 /* sub-blocks */, PAD, PAD, PAD };
- pn->pn_sk.dobject = pn_sockaddr_get_object(spn);
- pn->pn_sk.resource = pn_sockaddr_get_resource(spn);
pn->pipe_handle = 1; /* anything but INVALID_HANDLE */
- return pipe_handler_request(sk, PNS_PEP_CONNECT_REQ,
- PN_PIPE_DISABLE, data, 4);
+ err = pipe_handler_request(sk, PNS_PEP_CONNECT_REQ,
+ PN_PIPE_ENABLE, data, 4);
+ if (err) {
+ pn->pipe_handle = PN_PIPE_INVALID_HANDLE;
+ return err;
+ }
+ sk->sk_state = TCP_SYN_SENT;
+ return 0;
}
-#endif
static int pep_ioctl(struct sock *sk, int cmd, unsigned long arg)
{
@@ -890,8 +914,16 @@ static int pep_init(struct sock *sk)
sk->sk_destruct = pipe_destruct;
INIT_HLIST_HEAD(&pn->hlist);
+ pn->listener = NULL;
skb_queue_head_init(&pn->ctrlreq_queue);
+ atomic_set(&pn->tx_credits, 0);
+ pn->ifindex = 0;
+ pn->peer_type = 0;
pn->pipe_handle = PN_PIPE_INVALID_HANDLE;
+ pn->rx_credits = 0;
+ pn->rx_fc = pn->tx_fc = PN_LEGACY_FLOW_CONTROL;
+ pn->init_enable = 1;
+ pn->aligned = 0;
return 0;
}
@@ -1219,9 +1251,9 @@ static void pep_sock_unhash(struct sock *sk)
lock_sock(sk);
-#ifndef CONFIG_PHONET_PIPECTRLR
- if ((1 << sk->sk_state) & ~(TCPF_CLOSE|TCPF_LISTEN)) {
+ if (pn->listener != NULL) {
skparent = pn->listener;
+ pn->listener = NULL;
release_sock(sk);
pn = pep_sk(skparent);
@@ -1229,7 +1261,7 @@ static void pep_sock_unhash(struct sock *sk)
sk_del_node_init(sk);
sk = skparent;
}
-#endif
+
/* Unhash a listening sock only when it is closed
* and all of its active connected pipes are closed. */
if (hlist_empty(&pn->hlist))
@@ -1243,9 +1275,7 @@ static void pep_sock_unhash(struct sock *sk)
static struct proto pep_proto = {
.close = pep_sock_close,
.accept = pep_sock_accept,
-#ifdef CONFIG_PHONET_PIPECTRLR
.connect = pep_sock_connect,
-#endif
.ioctl = pep_ioctl,
.init = pep_init,
.setsockopt = pep_setsockopt,
diff --git a/net/phonet/socket.c b/net/phonet/socket.c
index 1eccfc3..b1adafa 100644
--- a/net/phonet/socket.c
+++ b/net/phonet/socket.c
@@ -225,15 +225,18 @@ static int pn_socket_autobind(struct socket *sock)
return 0; /* socket was already bound */
}
-#ifdef CONFIG_PHONET_PIPECTRLR
static int pn_socket_connect(struct socket *sock, struct sockaddr *addr,
int len, int flags)
{
struct sock *sk = sock->sk;
+ struct pn_sock *pn = pn_sk(sk);
struct sockaddr_pn *spn = (struct sockaddr_pn *)addr;
- long timeo;
+ struct task_struct *tsk = current;
+ long timeo = sock_rcvtimeo(sk, flags & O_NONBLOCK);
int err;
+ if (pn_socket_autobind(sock))
+ return -ENOBUFS;
if (len < sizeof(struct sockaddr_pn))
return -EINVAL;
if (spn->spn_family != AF_PHONET)
@@ -243,82 +246,61 @@ static int pn_socket_connect(struct socket *sock, struct sockaddr *addr,
switch (sock->state) {
case SS_UNCONNECTED:
- sk->sk_state = TCP_CLOSE;
- break;
- case SS_CONNECTING:
- switch (sk->sk_state) {
- case TCP_SYN_RECV:
- sock->state = SS_CONNECTED;
- err = -EISCONN;
- goto out;
- case TCP_CLOSE:
- err = -EALREADY;
- if (flags & O_NONBLOCK)
- goto out;
- goto wait_connect;
- }
- break;
- case SS_CONNECTED:
- switch (sk->sk_state) {
- case TCP_SYN_RECV:
+ if (sk->sk_state != TCP_CLOSE) {
err = -EISCONN;
goto out;
- case TCP_CLOSE:
- sock->state = SS_UNCONNECTED;
- break;
}
break;
- case SS_DISCONNECTING:
- case SS_FREE:
- break;
+ case SS_CONNECTING:
+ err = -EALREADY;
+ goto out;
+ default:
+ err = -EISCONN;
+ goto out;
}
- sk->sk_state = TCP_CLOSE;
- sk_stream_kill_queues(sk);
+ pn->dobject = pn_sockaddr_get_object(spn);
+ pn->resource = pn_sockaddr_get_resource(spn);
sock->state = SS_CONNECTING;
+
err = sk->sk_prot->connect(sk, addr, len);
- if (err < 0) {
+ if (err) {
sock->state = SS_UNCONNECTED;
- sk->sk_state = TCP_CLOSE;
+ pn->dobject = 0;
goto out;
}
- err = -EINPROGRESS;
-wait_connect:
- if (sk->sk_state != TCP_SYN_RECV && (flags & O_NONBLOCK))
- goto out;
-
- timeo = sock_sndtimeo(sk, flags & O_NONBLOCK);
- release_sock(sk);
+ while (sk->sk_state == TCP_SYN_SENT) {
+ DEFINE_WAIT(wait);
- err = -ERESTARTSYS;
- timeo = wait_event_interruptible_timeout(*sk_sleep(sk),
- sk->sk_state != TCP_CLOSE,
- timeo);
-
- lock_sock(sk);
- if (timeo < 0)
- goto out; /* -ERESTARTSYS */
-
- err = -ETIMEDOUT;
- if (timeo == 0 && sk->sk_state != TCP_SYN_RECV)
- goto out;
+ if (!timeo) {
+ err = -EINPROGRESS;
+ goto out;
+ }
+ if (signal_pending(tsk)) {
+ err = sock_intr_errno(timeo);
+ goto out;
+ }
- if (sk->sk_state != TCP_SYN_RECV) {
- sock->state = SS_UNCONNECTED;
- err = sock_error(sk);
- if (!err)
- err = -ECONNREFUSED;
- goto out;
+ prepare_to_wait_exclusive(sk_sleep(sk), &wait,
+ TASK_INTERRUPTIBLE);
+ release_sock(sk);
+ timeo = schedule_timeout(timeo);
+ lock_sock(sk);
+ finish_wait(sk_sleep(sk), &wait);
}
- sock->state = SS_CONNECTED;
- err = 0;
+ if ((1 << sk->sk_state) & (TCPF_SYN_RECV|TCPF_ESTABLISHED))
+ err = 0;
+ else if (sk->sk_state == TCP_CLOSE_WAIT)
+ err = -ECONNRESET;
+ else
+ err = -ECONNREFUSED;
+ sock->state = err ? SS_UNCONNECTED : SS_CONNECTED;
out:
release_sock(sk);
return err;
}
-#endif
static int pn_socket_accept(struct socket *sock, struct socket *newsock,
int flags)
@@ -486,11 +468,7 @@ const struct proto_ops phonet_stream_ops = {
.owner = THIS_MODULE,
.release = pn_socket_release,
.bind = pn_socket_bind,
-#ifdef CONFIG_PHONET_PIPECTRLR
.connect = pn_socket_connect,
-#else
- .connect = sock_no_connect,
-#endif
.socketpair = sock_no_socketpair,
.accept = pn_socket_accept,
.getname = pn_socket_getname,
--
1.7.1
^ permalink raw reply related
* [PATCH 5/8] Phonet: allocate sock from accept syscall rather than soft IRQ
From: Rémi Denis-Courmont @ 2011-03-07 10:46 UTC (permalink / raw)
To: netdev
In-Reply-To: <201103071245.47722.remi.denis-courmont@nokia.com>
This moves most of the accept logic to process context, and simplifies
the code a bit at the same time.
Signed-off-by: Rémi Denis-Courmont <remi.denis-courmont@nokia.com>
---
include/net/phonet/pep.h | 1 -
net/phonet/pep.c | 284 +++++++++++++++++++---------------------------
net/phonet/socket.c | 10 +-
3 files changed, 121 insertions(+), 174 deletions(-)
diff --git a/include/net/phonet/pep.h b/include/net/phonet/pep.h
index 38eed1b..b669fe6 100644
--- a/include/net/phonet/pep.h
+++ b/include/net/phonet/pep.h
@@ -28,7 +28,6 @@ struct pep_sock {
/* XXX: union-ify listening vs connected stuff ? */
/* Listening socket stuff: */
- struct hlist_head ackq;
struct hlist_head hlist;
/* Connected socket stuff: */
diff --git a/net/phonet/pep.c b/net/phonet/pep.c
index 610794a..c0fab4c 100644
--- a/net/phonet/pep.c
+++ b/net/phonet/pep.c
@@ -42,7 +42,7 @@
* TCP_ESTABLISHED connected pipe in enabled state
*
* pep_sock locking:
- * - sk_state, ackq, hlist: sock lock needed
+ * - sk_state, hlist: sock lock needed
* - listener: read only
* - pipe_handle: read only
*/
@@ -202,11 +202,12 @@ static int pep_accept_conn(struct sock *sk, struct sk_buff *skb)
GFP_KERNEL);
}
-static int pep_reject_conn(struct sock *sk, struct sk_buff *skb, u8 code)
+static int pep_reject_conn(struct sock *sk, struct sk_buff *skb, u8 code,
+ gfp_t priority)
{
static const u8 data[4] = { PAD, PAD, PAD, 0 /* sub-blocks */ };
WARN_ON(code == PN_PIPE_NO_ERROR);
- return pep_reply(sk, skb, code, data, sizeof(data), GFP_ATOMIC);
+ return pep_reply(sk, skb, code, data, sizeof(data), priority);
}
/* Control requests are not sent by the pipe service and have a specific
@@ -365,7 +366,7 @@ static int pipe_do_rcv(struct sock *sk, struct sk_buff *skb)
switch (hdr->message_id) {
case PNS_PEP_CONNECT_REQ:
- pep_reject_conn(sk, skb, PN_PIPE_ERR_PEP_IN_USE);
+ pep_reject_conn(sk, skb, PN_PIPE_ERR_PEP_IN_USE, GFP_ATOMIC);
break;
case PNS_PEP_DISCONNECT_REQ:
@@ -574,7 +575,6 @@ static int pep_connresp_rcv(struct sock *sk, struct sk_buff *skb)
sk->sk_state = TCP_SYN_RECV;
sk->sk_backlog_rcv = pipe_do_rcv;
- sk->sk_destruct = pipe_destruct;
pn->rx_credits = 0;
sk->sk_state_change(sk);
@@ -582,96 +582,6 @@ static int pep_connresp_rcv(struct sock *sk, struct sk_buff *skb)
}
#endif
-static int pep_connreq_rcv(struct sock *sk, struct sk_buff *skb)
-{
- struct sock *newsk;
- struct pep_sock *newpn, *pn = pep_sk(sk);
- struct pnpipehdr *hdr;
- struct sockaddr_pn dst, src;
- u16 peer_type;
- u8 pipe_handle, enabled, n_sb;
- u8 aligned = 0;
-
- if (!pskb_pull(skb, sizeof(*hdr) + 4))
- return -EINVAL;
-
- hdr = pnp_hdr(skb);
- pipe_handle = hdr->pipe_handle;
- switch (hdr->state_after_connect) {
- case PN_PIPE_DISABLE:
- enabled = 0;
- break;
- case PN_PIPE_ENABLE:
- enabled = 1;
- break;
- default:
- pep_reject_conn(sk, skb, PN_PIPE_ERR_INVALID_PARAM);
- return -EINVAL;
- }
- peer_type = hdr->other_pep_type << 8;
-
- /* Parse sub-blocks (options) */
- n_sb = hdr->data[4];
- while (n_sb > 0) {
- u8 type, buf[1], len = sizeof(buf);
- const u8 *data = pep_get_sb(skb, &type, &len, buf);
-
- if (data == NULL)
- return -EINVAL;
- switch (type) {
- case PN_PIPE_SB_CONNECT_REQ_PEP_SUB_TYPE:
- if (len < 1)
- return -EINVAL;
- peer_type = (peer_type & 0xff00) | data[0];
- break;
- case PN_PIPE_SB_ALIGNED_DATA:
- aligned = data[0] != 0;
- break;
- }
- n_sb--;
- }
-
- skb = skb_clone(skb, GFP_ATOMIC);
- if (!skb)
- return -ENOMEM;
-
- /* Create a new to-be-accepted sock */
- newsk = sk_alloc(sock_net(sk), PF_PHONET, GFP_ATOMIC, sk->sk_prot);
- if (!newsk) {
- kfree_skb(skb);
- return -ENOMEM;
- }
- sock_init_data(NULL, newsk);
- newsk->sk_state = TCP_SYN_RECV;
- newsk->sk_backlog_rcv = pipe_do_rcv;
- newsk->sk_protocol = sk->sk_protocol;
- newsk->sk_destruct = pipe_destruct;
-
- newpn = pep_sk(newsk);
- pn_skb_get_dst_sockaddr(skb, &dst);
- pn_skb_get_src_sockaddr(skb, &src);
- newpn->pn_sk.sobject = pn_sockaddr_get_object(&dst);
- newpn->pn_sk.dobject = pn_sockaddr_get_object(&src);
- newpn->pn_sk.resource = pn_sockaddr_get_resource(&dst);
- skb_queue_head_init(&newpn->ctrlreq_queue);
- newpn->pipe_handle = pipe_handle;
- atomic_set(&newpn->tx_credits, 0);
- newpn->peer_type = peer_type;
- newpn->rx_credits = 0;
- newpn->rx_fc = newpn->tx_fc = PN_LEGACY_FLOW_CONTROL;
- newpn->init_enable = enabled;
- newpn->aligned = aligned;
-
- BUG_ON(!skb_queue_empty(&newsk->sk_receive_queue));
- skb_queue_head(&newsk->sk_receive_queue, skb);
- if (!sock_flag(sk, SOCK_DEAD))
- sk->sk_data_ready(sk, 0);
-
- sk_acceptq_added(sk);
- sk_add_node(newsk, &pn->ackq);
- return 0;
-}
-
/* Listening sock must be locked */
static struct sock *pep_find_pipe(const struct hlist_head *hlist,
const struct sockaddr_pn *dst,
@@ -726,22 +636,18 @@ static int pep_do_rcv(struct sock *sk, struct sk_buff *skb)
if (sknode)
return sk_receive_skb(sknode, skb, 1);
- /* Look for a pipe handle pending accept */
- sknode = pep_find_pipe(&pn->ackq, &dst, pipe_handle);
- if (sknode) {
- sock_put(sknode);
- if (net_ratelimit())
- printk(KERN_WARNING"Phonet unconnected PEP ignored");
- goto drop;
- }
-
switch (hdr->message_id) {
case PNS_PEP_CONNECT_REQ:
- if (sk->sk_state == TCP_LISTEN && !sk_acceptq_is_full(sk))
- pep_connreq_rcv(sk, skb);
- else
- pep_reject_conn(sk, skb, PN_PIPE_ERR_PEP_IN_USE);
- break;
+ if (sk->sk_state != TCP_LISTEN || sk_acceptq_is_full(sk)) {
+ pep_reject_conn(sk, skb, PN_PIPE_ERR_PEP_IN_USE,
+ GFP_ATOMIC);
+ break;
+ }
+ skb_queue_head(&sk->sk_receive_queue, skb);
+ sk_acceptq_added(sk);
+ if (!sock_flag(sk, SOCK_DEAD))
+ sk->sk_data_ready(sk, 0);
+ return NET_RX_SUCCESS;
#ifdef CONFIG_PHONET_PIPECTRLR
case PNS_PEP_CONNECT_RESP:
@@ -799,24 +705,16 @@ static void pep_sock_close(struct sock *sk, long timeout)
sk_common_release(sk);
lock_sock(sk);
- if (sk->sk_state == TCP_LISTEN) {
- /* Destroy the listen queue */
- struct sock *sknode;
- struct hlist_node *p, *n;
-
- sk_for_each_safe(sknode, p, n, &pn->ackq)
- sk_del_node_init(sknode);
- sk->sk_state = TCP_CLOSE;
- } else if ((1 << sk->sk_state) & (TCPF_SYN_RECV|TCPF_ESTABLISHED)) {
+ if ((1 << sk->sk_state) & (TCPF_SYN_RECV|TCPF_ESTABLISHED)) {
#ifndef CONFIG_PHONET_PIPECTRLR
/* Forcefully remove dangling Phonet pipe */
pipe_do_remove(sk);
#else
/* send pep disconnect request */
pipe_handler_request(sk, PNS_PEP_DISCONNECT_REQ, PAD, NULL, 0);
- sk->sk_state = TCP_CLOSE;
#endif
}
+ sk->sk_state = TCP_CLOSE;
ifindex = pn->ifindex;
pn->ifindex = 0;
@@ -827,69 +725,121 @@ static void pep_sock_close(struct sock *sk, long timeout)
sock_put(sk);
}
-static int pep_wait_connreq(struct sock *sk, int noblock)
+static struct sock *pep_sock_accept(struct sock *sk, int flags, int *errp)
{
- struct task_struct *tsk = current;
- struct pep_sock *pn = pep_sk(sk);
- long timeo = sock_rcvtimeo(sk, noblock);
-
- for (;;) {
- DEFINE_WAIT(wait);
+ struct pep_sock *pn = pep_sk(sk), *newpn;
+ struct sock *newsk = NULL;
+ struct sk_buff *skb;
+ struct pnpipehdr *hdr;
+ struct sockaddr_pn dst, src;
+ int err;
+ u16 peer_type;
+ u8 pipe_handle, enabled, n_sb;
+ u8 aligned = 0;
- if (sk->sk_state != TCP_LISTEN)
- return -EINVAL;
- if (!hlist_empty(&pn->ackq))
- break;
- if (!timeo)
- return -EWOULDBLOCK;
- if (signal_pending(tsk))
- return sock_intr_errno(timeo);
+ skb = skb_recv_datagram(sk, 0, flags & O_NONBLOCK, errp);
+ if (!skb)
+ return NULL;
- prepare_to_wait_exclusive(sk_sleep(sk), &wait,
- TASK_INTERRUPTIBLE);
- release_sock(sk);
- timeo = schedule_timeout(timeo);
- lock_sock(sk);
- finish_wait(sk_sleep(sk), &wait);
+ lock_sock(sk);
+ if (sk->sk_state != TCP_LISTEN) {
+ err = -EINVAL;
+ goto drop;
}
+ sk_acceptq_removed(sk);
- return 0;
-}
+ err = -EPROTO;
+ if (!pskb_may_pull(skb, sizeof(*hdr) + 4))
+ goto drop;
-static struct sock *pep_sock_accept(struct sock *sk, int flags, int *errp)
-{
- struct pep_sock *pn = pep_sk(sk);
- struct sock *newsk = NULL;
- struct sk_buff *oskb;
- int err;
+ hdr = pnp_hdr(skb);
+ pipe_handle = hdr->pipe_handle;
+ switch (hdr->state_after_connect) {
+ case PN_PIPE_DISABLE:
+ enabled = 0;
+ break;
+ case PN_PIPE_ENABLE:
+ enabled = 1;
+ break;
+ default:
+ pep_reject_conn(sk, skb, PN_PIPE_ERR_INVALID_PARAM,
+ GFP_KERNEL);
+ goto drop;
+ }
+ peer_type = hdr->other_pep_type << 8;
- lock_sock(sk);
- err = pep_wait_connreq(sk, flags & O_NONBLOCK);
- if (err)
- goto out;
+ /* Parse sub-blocks (options) */
+ n_sb = hdr->data[4];
+ while (n_sb > 0) {
+ u8 type, buf[1], len = sizeof(buf);
+ const u8 *data = pep_get_sb(skb, &type, &len, buf);
- newsk = __sk_head(&pn->ackq);
+ if (data == NULL)
+ goto drop;
+ switch (type) {
+ case PN_PIPE_SB_CONNECT_REQ_PEP_SUB_TYPE:
+ if (len < 1)
+ goto drop;
+ peer_type = (peer_type & 0xff00) | data[0];
+ break;
+ case PN_PIPE_SB_ALIGNED_DATA:
+ aligned = data[0] != 0;
+ break;
+ }
+ n_sb--;
+ }
- oskb = skb_dequeue(&newsk->sk_receive_queue);
- err = pep_accept_conn(newsk, oskb);
- if (err) {
- skb_queue_head(&newsk->sk_receive_queue, oskb);
+ /* Check for duplicate pipe handle */
+ newsk = pep_find_pipe(&pn->hlist, &dst, pipe_handle);
+ if (unlikely(newsk)) {
+ __sock_put(newsk);
newsk = NULL;
- goto out;
+ pep_reject_conn(sk, skb, PN_PIPE_ERR_PEP_IN_USE, GFP_KERNEL);
+ goto drop;
+ }
+
+ /* Create a new to-be-accepted sock */
+ newsk = sk_alloc(sock_net(sk), PF_PHONET, GFP_KERNEL, sk->sk_prot);
+ if (!newsk) {
+ pep_reject_conn(sk, skb, PN_PIPE_ERR_OVERLOAD, GFP_KERNEL);
+ err = -ENOBUFS;
+ goto drop;
}
- kfree_skb(oskb);
+ sock_init_data(NULL, newsk);
+ newsk->sk_state = TCP_SYN_RECV;
+ newsk->sk_backlog_rcv = pipe_do_rcv;
+ newsk->sk_protocol = sk->sk_protocol;
+ newsk->sk_destruct = pipe_destruct;
+
+ newpn = pep_sk(newsk);
+ pn_skb_get_dst_sockaddr(skb, &dst);
+ pn_skb_get_src_sockaddr(skb, &src);
+ newpn->pn_sk.sobject = pn_sockaddr_get_object(&dst);
+ newpn->pn_sk.dobject = pn_sockaddr_get_object(&src);
+ newpn->pn_sk.resource = pn_sockaddr_get_resource(&dst);
sock_hold(sk);
- pep_sk(newsk)->listener = sk;
+ newpn->listener = sk;
+ skb_queue_head_init(&newpn->ctrlreq_queue);
+ newpn->pipe_handle = pipe_handle;
+ atomic_set(&newpn->tx_credits, 0);
+ newpn->ifindex = 0;
+ newpn->peer_type = peer_type;
+ newpn->rx_credits = 0;
+ newpn->rx_fc = newpn->tx_fc = PN_LEGACY_FLOW_CONTROL;
+ newpn->init_enable = enabled;
+ newpn->aligned = aligned;
- sock_hold(newsk);
- sk_del_node_init(newsk);
- sk_acceptq_removed(sk);
+ err = pep_accept_conn(newsk, skb);
+ if (err) {
+ sock_put(newsk);
+ newsk = NULL;
+ goto drop;
+ }
sk_add_node(newsk, &pn->hlist);
- __sock_put(newsk);
-
-out:
+drop:
release_sock(sk);
+ kfree_skb(skb);
*errp = err;
return newsk;
}
@@ -937,7 +887,7 @@ static int pep_init(struct sock *sk)
{
struct pep_sock *pn = pep_sk(sk);
- INIT_HLIST_HEAD(&pn->ackq);
+ sk->sk_destruct = pipe_destruct;
INIT_HLIST_HEAD(&pn->hlist);
skb_queue_head_init(&pn->ctrlreq_queue);
pn->pipe_handle = PN_PIPE_INVALID_HANDLE;
diff --git a/net/phonet/socket.c b/net/phonet/socket.c
index 65a0333..1eccfc3 100644
--- a/net/phonet/socket.c
+++ b/net/phonet/socket.c
@@ -327,6 +327,9 @@ static int pn_socket_accept(struct socket *sock, struct socket *newsock,
struct sock *newsk;
int err;
+ if (unlikely(sk->sk_state != TCP_LISTEN))
+ return -EINVAL;
+
newsk = sk->sk_prot->accept(sk, flags, &err);
if (!newsk)
return err;
@@ -363,13 +366,8 @@ static unsigned int pn_socket_poll(struct file *file, struct socket *sock,
poll_wait(file, sk_sleep(sk), wait);
- switch (sk->sk_state) {
- case TCP_LISTEN:
- return hlist_empty(&pn->ackq) ? 0 : POLLIN;
- case TCP_CLOSE:
+ if (sk->sk_state == TCP_CLOSE)
return POLLERR;
- }
-
if (!skb_queue_empty(&sk->sk_receive_queue))
mask |= POLLIN | POLLRDNORM;
if (!skb_queue_empty(&pn->ctrlreq_queue))
--
1.7.1
^ permalink raw reply related
* [PATCH 4/8] Phonet: factor common code to send control messages
From: Rémi Denis-Courmont @ 2011-03-07 10:46 UTC (permalink / raw)
To: netdev
In-Reply-To: <201103071245.47722.remi.denis-courmont@nokia.com>
Signed-off-by: Rémi Denis-Courmont <remi.denis-courmont@nokia.com>
---
net/phonet/pep.c | 225 ++++++++++++++++++------------------------------------
1 files changed, 73 insertions(+), 152 deletions(-)
diff --git a/net/phonet/pep.c b/net/phonet/pep.c
index 40952c7..610794a 100644
--- a/net/phonet/pep.c
+++ b/net/phonet/pep.c
@@ -77,24 +77,34 @@ static unsigned char *pep_get_sb(struct sk_buff *skb, u8 *ptype, u8 *plen,
return data;
}
-static int pep_reply(struct sock *sk, struct sk_buff *oskb,
- u8 code, const void *data, int len, gfp_t priority)
+static struct sk_buff *pep_alloc_skb(struct sock *sk, const void *payload,
+ int len, gfp_t priority)
+{
+ struct sk_buff *skb = alloc_skb(MAX_PNPIPE_HEADER + len, priority);
+ if (!skb)
+ return NULL;
+ skb_set_owner_w(skb, sk);
+
+ skb_reserve(skb, MAX_PNPIPE_HEADER);
+ __skb_put(skb, len);
+ skb_copy_to_linear_data(skb, payload, len);
+ __skb_push(skb, sizeof(struct pnpipehdr));
+ skb_reset_transport_header(skb);
+ return skb;
+}
+
+static int pep_reply(struct sock *sk, struct sk_buff *oskb, u8 code,
+ const void *data, int len, gfp_t priority)
{
const struct pnpipehdr *oph = pnp_hdr(oskb);
struct pnpipehdr *ph;
struct sk_buff *skb;
struct sockaddr_pn peer;
- skb = alloc_skb(MAX_PNPIPE_HEADER + len, priority);
+ skb = pep_alloc_skb(sk, data, len, priority);
if (!skb)
return -ENOMEM;
- skb_set_owner_w(skb, sk);
- skb_reserve(skb, MAX_PNPIPE_HEADER);
- __skb_put(skb, len);
- skb_copy_to_linear_data(skb, data, len);
- __skb_push(skb, sizeof(*ph));
- skb_reset_transport_header(skb);
ph = pnp_hdr(skb);
ph->utid = oph->utid;
ph->message_id = oph->message_id + 1; /* REQ -> RESP */
@@ -105,135 +115,69 @@ static int pep_reply(struct sock *sk, struct sk_buff *oskb,
return pn_skb_send(sk, skb, &peer);
}
-#define PAD 0x00
-
-#ifdef CONFIG_PHONET_PIPECTRLR
-static int pipe_handler_send_req(struct sock *sk, u8 msg_id, gfp_t priority)
+static int pep_indicate(struct sock *sk, u8 id, u8 code,
+ const void *data, int len, gfp_t priority)
{
- int len;
+ struct pep_sock *pn = pep_sk(sk);
struct pnpipehdr *ph;
struct sk_buff *skb;
- struct pep_sock *pn = pep_sk(sk);
-
- static const u8 data[4] = {
- PAD, PAD, PAD, PAD,
- };
- switch (msg_id) {
- case PNS_PEP_CONNECT_REQ:
- len = sizeof(data);
- break;
-
- case PNS_PEP_DISCONNECT_REQ:
- case PNS_PEP_ENABLE_REQ:
- case PNS_PEP_DISABLE_REQ:
- len = 0;
- break;
-
- default:
- return -EINVAL;
- }
-
- skb = alloc_skb(MAX_PNPIPE_HEADER + len, priority);
+ skb = pep_alloc_skb(sk, data, len, priority);
if (!skb)
return -ENOMEM;
- skb_set_owner_w(skb, sk);
- skb_reserve(skb, MAX_PNPIPE_HEADER);
- if (len) {
- __skb_put(skb, len);
- skb_copy_to_linear_data(skb, data, len);
- }
- __skb_push(skb, sizeof(*ph));
- skb_reset_transport_header(skb);
ph = pnp_hdr(skb);
- ph->utid = msg_id; /* whatever */
- ph->message_id = msg_id;
+ ph->utid = 0;
+ ph->message_id = id;
ph->pipe_handle = pn->pipe_handle;
- ph->error_code = PN_PIPE_NO_ERROR;
-
+ ph->data[0] = code;
return pn_skb_send(sk, skb, NULL);
}
-static int pipe_handler_send_created_ind(struct sock *sk, u8 msg_id)
+#define PAD 0x00
+
+#ifdef CONFIG_PHONET_PIPECTRLR
+static int pipe_handler_request(struct sock *sk, u8 id, u8 code,
+ const void *data, int len)
{
- int err_code;
+ struct pep_sock *pn = pep_sk(sk);
struct pnpipehdr *ph;
struct sk_buff *skb;
- struct pep_sock *pn = pep_sk(sk);
- static u8 data[4] = {
- 0x03, 0x04,
- };
- data[2] = pn->tx_fc;
- data[3] = pn->rx_fc;
-
- /*
- * actually, below is number of sub-blocks and not error code.
- * Pipe_created_ind message format does not have any
- * error code field. However, the Phonet stack will always send
- * an error code as part of pnpipehdr. So, use that err_code to
- * specify the number of sub-blocks.
- */
- err_code = 0x01;
-
- skb = alloc_skb(MAX_PNPIPE_HEADER + sizeof(data), GFP_ATOMIC);
+ skb = pep_alloc_skb(sk, data, len, GFP_KERNEL);
if (!skb)
return -ENOMEM;
- skb_set_owner_w(skb, sk);
- skb_reserve(skb, MAX_PNPIPE_HEADER);
- __skb_put(skb, sizeof(data));
- skb_copy_to_linear_data(skb, data, sizeof(data));
- __skb_push(skb, sizeof(*ph));
- skb_reset_transport_header(skb);
ph = pnp_hdr(skb);
- ph->utid = 0;
- ph->message_id = msg_id;
+ ph->utid = id; /* whatever */
+ ph->message_id = id;
ph->pipe_handle = pn->pipe_handle;
- ph->error_code = err_code;
-
+ ph->data[0] = code;
return pn_skb_send(sk, skb, NULL);
}
-static int pipe_handler_send_ind(struct sock *sk, u8 msg_id)
+static int pipe_handler_send_created_ind(struct sock *sk)
{
- int err_code;
- struct pnpipehdr *ph;
- struct sk_buff *skb;
struct pep_sock *pn = pep_sk(sk);
+ u8 data[4] = {
+ PN_PIPE_SB_NEGOTIATED_FC, pep_sb_size(2),
+ pn->tx_fc, pn->rx_fc,
+ };
- /*
- * actually, below is a filler.
- * Pipe_enabled/disabled_ind message format does not have any
- * error code field. However, the Phonet stack will always send
- * an error code as part of pnpipehdr. So, use that err_code to
- * specify the filler value.
- */
- err_code = 0x0;
-
- skb = alloc_skb(MAX_PNPIPE_HEADER, GFP_ATOMIC);
- if (!skb)
- return -ENOMEM;
- skb_set_owner_w(skb, sk);
-
- skb_reserve(skb, MAX_PNPIPE_HEADER);
- __skb_push(skb, sizeof(*ph));
- skb_reset_transport_header(skb);
- ph = pnp_hdr(skb);
- ph->utid = 0;
- ph->message_id = msg_id;
- ph->pipe_handle = pn->pipe_handle;
- ph->error_code = err_code;
+ return pep_indicate(sk, PNS_PIPE_CREATED_IND, 1 /* sub-blocks */,
+ data, 4, GFP_ATOMIC);
+}
- return pn_skb_send(sk, skb, NULL);
+static int pipe_handler_send_ind(struct sock *sk, u8 id)
+{
+ return pep_indicate(sk, id, PAD, NULL, 0, GFP_ATOMIC);
}
static int pipe_handler_enable_pipe(struct sock *sk, int enable)
{
u8 id = enable ? PNS_PEP_ENABLE_REQ : PNS_PEP_DISABLE_REQ;
- return pipe_handler_send_req(sk, id, GFP_KERNEL);
+ return pipe_handler_request(sk, id, PAD, NULL, 0);
}
#endif
@@ -274,23 +218,21 @@ static int pep_ctrlreq_error(struct sock *sk, struct sk_buff *oskb, u8 code,
struct sk_buff *skb;
struct pnpipehdr *ph;
struct sockaddr_pn dst;
+ u8 data[4] = {
+ oph->data[0], /* PEP type */
+ code, /* error code, at an unusual offset */
+ PAD, PAD,
+ };
- skb = alloc_skb(MAX_PNPIPE_HEADER + 4, priority);
+ skb = pep_alloc_skb(sk, data, 4, priority);
if (!skb)
return -ENOMEM;
- skb_set_owner_w(skb, sk);
-
- skb_reserve(skb, MAX_PHONET_HEADER);
- ph = (struct pnpipehdr *)skb_put(skb, sizeof(*ph) + 4);
+ ph = pnp_hdr(skb);
ph->utid = oph->utid;
ph->message_id = PNS_PEP_CTRL_RESP;
ph->pipe_handle = oph->pipe_handle;
ph->data[0] = oph->data[1]; /* CTRL id */
- ph->data[1] = oph->data[0]; /* PEP type */
- ph->data[2] = code; /* error code, at an usual offset */
- ph->data[3] = PAD;
- ph->data[4] = PAD;
pn_skb_get_src_sockaddr(oskb, &dst);
return pn_skb_send(sk, skb, &dst);
@@ -298,34 +240,15 @@ static int pep_ctrlreq_error(struct sock *sk, struct sk_buff *oskb, u8 code,
static int pipe_snd_status(struct sock *sk, u8 type, u8 status, gfp_t priority)
{
- struct pep_sock *pn = pep_sk(sk);
- struct pnpipehdr *ph;
- struct sk_buff *skb;
+ u8 data[4] = { type, PAD, PAD, status };
- skb = alloc_skb(MAX_PNPIPE_HEADER + 4, priority);
- if (!skb)
- return -ENOMEM;
- skb_set_owner_w(skb, sk);
-
- skb_reserve(skb, MAX_PNPIPE_HEADER + 4);
- __skb_push(skb, sizeof(*ph) + 4);
- skb_reset_transport_header(skb);
- ph = pnp_hdr(skb);
- ph->utid = 0;
- ph->message_id = PNS_PEP_STATUS_IND;
- ph->pipe_handle = pn->pipe_handle;
- ph->pep_type = PN_PEP_TYPE_COMMON;
- ph->data[1] = type;
- ph->data[2] = PAD;
- ph->data[3] = PAD;
- ph->data[4] = status;
-
- return pn_skb_send(sk, skb, NULL);
+ return pep_indicate(sk, PNS_PEP_STATUS_IND, PN_PEP_TYPE_COMMON,
+ data, 4, priority);
}
/* Send our RX flow control information to the sender.
* Socket must be locked. */
-static void pipe_grant_credits(struct sock *sk)
+static void pipe_grant_credits(struct sock *sk, gfp_t priority)
{
struct pep_sock *pn = pep_sk(sk);
@@ -335,16 +258,16 @@ static void pipe_grant_credits(struct sock *sk)
case PN_LEGACY_FLOW_CONTROL: /* TODO */
break;
case PN_ONE_CREDIT_FLOW_CONTROL:
- pipe_snd_status(sk, PN_PEP_IND_FLOW_CONTROL,
- PEP_IND_READY, GFP_ATOMIC);
- pn->rx_credits = 1;
+ if (pipe_snd_status(sk, PN_PEP_IND_FLOW_CONTROL,
+ PEP_IND_READY, priority) == 0)
+ pn->rx_credits = 1;
break;
case PN_MULTI_CREDIT_FLOW_CONTROL:
if ((pn->rx_credits + CREDITS_THR) > CREDITS_MAX)
break;
if (pipe_snd_status(sk, PN_PEP_IND_ID_MCFC_GRANT_CREDITS,
CREDITS_MAX - pn->rx_credits,
- GFP_ATOMIC) == 0)
+ priority) == 0)
pn->rx_credits = CREDITS_MAX;
break;
}
@@ -474,7 +397,7 @@ static int pipe_do_rcv(struct sock *sk, struct sk_buff *skb)
if (sk->sk_state == TCP_ESTABLISHED)
break; /* Nothing to do */
sk->sk_state = TCP_ESTABLISHED;
- pipe_grant_credits(sk);
+ pipe_grant_credits(sk, GFP_ATOMIC);
break;
#endif
@@ -561,7 +484,7 @@ static int pipe_do_rcv(struct sock *sk, struct sk_buff *skb)
if (sk->sk_state == TCP_ESTABLISHED)
break; /* Nothing to do */
sk->sk_state = TCP_ESTABLISHED;
- pipe_grant_credits(sk);
+ pipe_grant_credits(sk, GFP_ATOMIC);
break;
case PNS_PIPE_DISABLED_IND:
@@ -655,7 +578,7 @@ static int pep_connresp_rcv(struct sock *sk, struct sk_buff *skb)
pn->rx_credits = 0;
sk->sk_state_change(sk);
- return pipe_handler_send_created_ind(sk, PNS_PIPE_CREATED_IND);
+ return pipe_handler_send_created_ind(sk);
}
#endif
@@ -853,19 +776,15 @@ static int pipe_do_remove(struct sock *sk)
struct pnpipehdr *ph;
struct sk_buff *skb;
- skb = alloc_skb(MAX_PNPIPE_HEADER, GFP_KERNEL);
+ skb = pep_alloc_skb(sk, NULL, 0, GFP_KERNEL);
if (!skb)
return -ENOMEM;
- skb_reserve(skb, MAX_PNPIPE_HEADER);
- __skb_push(skb, sizeof(*ph));
- skb_reset_transport_header(skb);
ph = pnp_hdr(skb);
ph->utid = 0;
ph->message_id = PNS_PIPE_REMOVE_REQ;
ph->pipe_handle = pn->pipe_handle;
ph->data[0] = PAD;
-
return pn_skb_send(sk, skb, NULL);
}
#endif
@@ -894,7 +813,7 @@ static void pep_sock_close(struct sock *sk, long timeout)
pipe_do_remove(sk);
#else
/* send pep disconnect request */
- pipe_handler_send_req(sk, PNS_PEP_DISCONNECT_REQ, GFP_KERNEL);
+ pipe_handler_request(sk, PNS_PEP_DISCONNECT_REQ, PAD, NULL, 0);
sk->sk_state = TCP_CLOSE;
#endif
}
@@ -980,10 +899,12 @@ static int pep_sock_connect(struct sock *sk, struct sockaddr *addr, int len)
{
struct pep_sock *pn = pep_sk(sk);
const struct sockaddr_pn *spn = (struct sockaddr_pn *)addr;
+ u8 data[4] = { 0 /* sub-blocks */, PAD, PAD, PAD };
pn->pn_sk.dobject = pn_sockaddr_get_object(spn);
pn->pn_sk.resource = pn_sockaddr_get_resource(spn);
- return pipe_handler_send_req(sk, PNS_PEP_CONNECT_REQ, GFP_KERNEL);
+ return pipe_handler_request(sk, PNS_PEP_CONNECT_REQ,
+ PN_PIPE_DISABLE, data, 4);
}
#endif
@@ -1280,7 +1201,7 @@ struct sk_buff *pep_read(struct sock *sk)
struct sk_buff *skb = skb_dequeue(&sk->sk_receive_queue);
if (sk->sk_state == TCP_ESTABLISHED)
- pipe_grant_credits(sk);
+ pipe_grant_credits(sk, GFP_ATOMIC);
return skb;
}
@@ -1325,7 +1246,7 @@ static int pep_recvmsg(struct kiocb *iocb, struct sock *sk,
}
if (sk->sk_state == TCP_ESTABLISHED)
- pipe_grant_credits(sk);
+ pipe_grant_credits(sk, GFP_KERNEL);
release_sock(sk);
copy:
msg->msg_flags |= MSG_EOR;
--
1.7.1
^ permalink raw reply related
* [PATCH 3/8] Phonet: fix backlog callback return value
From: Rémi Denis-Courmont @ 2011-03-07 10:46 UTC (permalink / raw)
To: netdev
In-Reply-To: <201103071245.47722.remi.denis-courmont@nokia.com>
Signed-off-by: Rémi Denis-Courmont <remi.denis-courmont@nokia.com>
---
net/phonet/pep.c | 25 +++++++++++--------------
1 files changed, 11 insertions(+), 14 deletions(-)
diff --git a/net/phonet/pep.c b/net/phonet/pep.c
index 875e86c..40952c7 100644
--- a/net/phonet/pep.c
+++ b/net/phonet/pep.c
@@ -522,7 +522,8 @@ static int pipe_do_rcv(struct sock *sk, struct sk_buff *skb)
if (!pn_flow_safe(pn->rx_fc)) {
err = sock_queue_rcv_skb(sk, skb);
if (!err)
- return 0;
+ return NET_RX_SUCCESS;
+ err = -ENOBUFS;
break;
}
@@ -575,7 +576,7 @@ static int pipe_do_rcv(struct sock *sk, struct sk_buff *skb)
}
out:
kfree_skb(skb);
- return err;
+ return (err == -ENOBUFS) ? NET_RX_DROP : NET_RX_SUCCESS;
queue:
skb->dev = NULL;
@@ -584,7 +585,7 @@ queue:
skb_queue_tail(queue, skb);
if (!sock_flag(sk, SOCK_DEAD))
sk->sk_data_ready(sk, err);
- return 0;
+ return NET_RX_SUCCESS;
}
/* Destroy connected sock. */
@@ -686,11 +687,6 @@ static int pep_connreq_rcv(struct sock *sk, struct sk_buff *skb)
}
peer_type = hdr->other_pep_type << 8;
- if (unlikely(sk->sk_state != TCP_LISTEN) || sk_acceptq_is_full(sk)) {
- pep_reject_conn(sk, skb, PN_PIPE_ERR_PEP_IN_USE);
- return -ENOBUFS;
- }
-
/* Parse sub-blocks (options) */
n_sb = hdr->data[4];
while (n_sb > 0) {
@@ -790,7 +786,6 @@ static int pep_do_rcv(struct sock *sk, struct sk_buff *skb)
struct sock *sknode;
struct pnpipehdr *hdr;
struct sockaddr_pn dst;
- int err = NET_RX_SUCCESS;
u8 pipe_handle;
if (!pskb_may_pull(skb, sizeof(*hdr)))
@@ -814,18 +809,20 @@ static int pep_do_rcv(struct sock *sk, struct sk_buff *skb)
sock_put(sknode);
if (net_ratelimit())
printk(KERN_WARNING"Phonet unconnected PEP ignored");
- err = NET_RX_DROP;
goto drop;
}
switch (hdr->message_id) {
case PNS_PEP_CONNECT_REQ:
- err = pep_connreq_rcv(sk, skb);
+ if (sk->sk_state == TCP_LISTEN && !sk_acceptq_is_full(sk))
+ pep_connreq_rcv(sk, skb);
+ else
+ pep_reject_conn(sk, skb, PN_PIPE_ERR_PEP_IN_USE);
break;
#ifdef CONFIG_PHONET_PIPECTRLR
case PNS_PEP_CONNECT_RESP:
- err = pep_connresp_rcv(sk, skb);
+ pep_connresp_rcv(sk, skb);
break;
#endif
@@ -842,11 +839,11 @@ static int pep_do_rcv(struct sock *sk, struct sk_buff *skb)
case PNS_PEP_DISABLE_REQ:
/* invalid handle is not even allowed here! */
default:
- err = NET_RX_DROP;
+ break;
}
drop:
kfree_skb(skb);
- return err;
+ return NET_RX_SUCCESS;
}
#ifndef CONFIG_PHONET_PIPECTRLR
--
1.7.1
^ permalink raw reply related
* [PATCH 2/8] Phonet: return an error when TX fails
From: Rémi Denis-Courmont @ 2011-03-07 10:46 UTC (permalink / raw)
To: netdev
In-Reply-To: <201103071245.47722.remi.denis-courmont@nokia.com>
Signed-off-by: Rémi Denis-Courmont <remi.denis-courmont@nokia.com>
---
net/phonet/af_phonet.c | 8 +++-----
1 files changed, 3 insertions(+), 5 deletions(-)
diff --git a/net/phonet/af_phonet.c b/net/phonet/af_phonet.c
index 4706b77..c6fffd9 100644
--- a/net/phonet/af_phonet.c
+++ b/net/phonet/af_phonet.c
@@ -195,11 +195,7 @@ static int pn_send(struct sk_buff *skb, struct net_device *dev,
if (skb->pkt_type == PACKET_LOOPBACK) {
skb_reset_mac_header(skb);
skb_orphan(skb);
- if (irq)
- netif_rx(skb);
- else
- netif_rx_ni(skb);
- err = 0;
+ err = (irq ? netif_rx(skb) : netif_rx_ni(skb)) ? -ENOBUFS : 0;
} else {
err = dev_hard_header(skb, dev, ntohs(skb->protocol),
NULL, NULL, skb->len);
@@ -208,6 +204,8 @@ static int pn_send(struct sk_buff *skb, struct net_device *dev,
goto drop;
}
err = dev_queue_xmit(skb);
+ if (unlikely(err > 0))
+ err = net_xmit_errno(err);
}
return err;
--
1.7.1
^ permalink raw reply related
* [PATCH 1/8] Phonet: fix NULL-deref in previous patch series
From: Rémi Denis-Courmont @ 2011-03-07 10:46 UTC (permalink / raw)
To: netdev
In-Reply-To: <201103071245.47722.remi.denis-courmont@nokia.com>
Signed-off-by: Rémi Denis-Courmont <remi.denis-courmont@nokia.com>
---
net/phonet/af_phonet.c | 5 ++---
1 files changed, 2 insertions(+), 3 deletions(-)
diff --git a/net/phonet/af_phonet.c b/net/phonet/af_phonet.c
index 30cc676..4706b77 100644
--- a/net/phonet/af_phonet.c
+++ b/net/phonet/af_phonet.c
@@ -262,10 +262,9 @@ int pn_skb_send(struct sock *sk, struct sk_buff *skb,
else if (phonet_address_lookup(net, daddr) == 0) {
dev = phonet_device_get(net);
skb->pkt_type = PACKET_LOOPBACK;
- } else if (pn_sockaddr_get_object(target) == 0) {
+ } else if (dst == 0) {
/* Resource routing (small race until phonet_rcv()) */
- struct sock *sk = pn_find_sock_by_res(net,
- target->spn_resource);
+ struct sock *sk = pn_find_sock_by_res(net, res);
if (sk) {
sock_put(sk);
dev = phonet_device_get(net);
--
1.7.1
^ permalink raw reply related
* [RFC] [PATCH 0/8] net-next: Phonet fixes and cleanup
From: Rémi Denis-Courmont @ 2011-03-07 10:45 UTC (permalink / raw)
To: netdev
As "promised", here is the cleaned up and unified Phonet pipe code.
The same kernel can now support both old (e.g. Nokia N900) and new
(e.g. ST-Ericsson U8500 SoC) modems.
Changes are available in the git repository at:
git://git.remlab.net/linux-phonet.git master
Rémi Denis-Courmont (8):
Phonet: fix NULL-deref in previous patch series
Phonet: return an error when TX fails
Phonet: fix backlog callback return value
Phonet: factor common code to send control messages
Phonet: allocate sock from accept syscall rather than soft IRQ
Phonet: provide pipe socket option to retrieve the pipe identifier
Phonet: support active connection without pipe controller on modem
Phonet: kill the ST-Ericsson pipe controller Kconfig
Documentation/networking/phonet.txt | 67 ++---
include/linux/phonet.h | 4 +-
include/net/phonet/pep.h | 1 -
net/phonet/Kconfig | 12 -
net/phonet/af_phonet.c | 13 +-
net/phonet/pep.c | 682 ++++++++++++++---------------------
net/phonet/socket.c | 112 +++----
7 files changed, 352 insertions(+), 539 deletions(-)
--
Rémi Denis-Courmont
http://www.remlab.net/
^ permalink raw reply
* [PATCH 16/17] mlx4_en: Using blue flame support
From: Yevgeny Petrilin @ 2011-03-07 10:18 UTC (permalink / raw)
To: davem; +Cc: netdev, yevgenyp
Doorbell is used according to usage of BlueFlame.
For Blue Flame to work in Ethernet mode QP number should have 0
at bits 6,7.
Allocating range of QPs accordingly.
Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
---
drivers/net/mlx4/en_netdev.c | 10 +++++-
drivers/net/mlx4/en_tx.c | 72 ++++++++++++++++++++++++++++++-----------
drivers/net/mlx4/mlx4_en.h | 4 ++-
3 files changed, 64 insertions(+), 22 deletions(-)
diff --git a/drivers/net/mlx4/en_netdev.c b/drivers/net/mlx4/en_netdev.c
index 4a30047..78717aa 100644
--- a/drivers/net/mlx4/en_netdev.c
+++ b/drivers/net/mlx4/en_netdev.c
@@ -931,6 +931,13 @@ int mlx4_en_alloc_resources(struct mlx4_en_priv *priv)
{
struct mlx4_en_port_profile *prof = priv->prof;
int i;
+ int base_tx_qpn, err;
+
+ err = mlx4_qp_reserve_range(priv->mdev->dev, priv->tx_ring_num, 256, &base_tx_qpn);
+ if (err) {
+ en_err(priv, "failed reserving range for TX rings\n");
+ return err;
+ }
/* Create tx Rings */
for (i = 0; i < priv->tx_ring_num; i++) {
@@ -938,7 +945,7 @@ int mlx4_en_alloc_resources(struct mlx4_en_priv *priv)
prof->tx_ring_size, i, TX))
goto err;
- if (mlx4_en_create_tx_ring(priv, &priv->tx_ring[i],
+ if (mlx4_en_create_tx_ring(priv, &priv->tx_ring[i], base_tx_qpn + i,
prof->tx_ring_size, TXBB_SIZE))
goto err;
}
@@ -958,6 +965,7 @@ int mlx4_en_alloc_resources(struct mlx4_en_priv *priv)
err:
en_err(priv, "Failed to allocate NIC resources\n");
+ mlx4_qp_release_range(priv->mdev->dev, base_tx_qpn, priv->tx_ring_num);
return -ENOMEM;
}
diff --git a/drivers/net/mlx4/en_tx.c b/drivers/net/mlx4/en_tx.c
index a680cd4..01feb8f 100644
--- a/drivers/net/mlx4/en_tx.c
+++ b/drivers/net/mlx4/en_tx.c
@@ -44,6 +44,7 @@
enum {
MAX_INLINE = 104, /* 128 - 16 - 4 - 4 */
+ MAX_BF = 256,
};
static int inline_thold __read_mostly = MAX_INLINE;
@@ -52,7 +53,7 @@ module_param_named(inline_thold, inline_thold, int, 0444);
MODULE_PARM_DESC(inline_thold, "threshold for using inline data");
int mlx4_en_create_tx_ring(struct mlx4_en_priv *priv,
- struct mlx4_en_tx_ring *ring, u32 size,
+ struct mlx4_en_tx_ring *ring, int qpn, u32 size,
u16 stride)
{
struct mlx4_en_dev *mdev = priv->mdev;
@@ -103,23 +104,25 @@ int mlx4_en_create_tx_ring(struct mlx4_en_priv *priv,
"buf_size:%d dma:%llx\n", ring, ring->buf, ring->size,
ring->buf_size, (unsigned long long) ring->wqres.buf.direct.map);
- err = mlx4_qp_reserve_range(mdev->dev, 1, 1, &ring->qpn);
- if (err) {
- en_err(priv, "Failed reserving qp for tx ring.\n");
- goto err_map;
- }
-
+ ring->qpn = qpn;
err = mlx4_qp_alloc(mdev->dev, ring->qpn, &ring->qp);
if (err) {
en_err(priv, "Failed allocating qp %d\n", ring->qpn);
- goto err_reserve;
+ goto err_map;
}
ring->qp.event = mlx4_en_sqp_event;
+ err = mlx4_bf_alloc(mdev->dev, &ring->bf);
+ if (err) {
+ en_dbg(DRV, priv, "working without blueflame (%d)", err);
+ ring->bf.uar = &mdev->priv_uar;
+ ring->bf.uar->map = mdev->uar_map;
+ ring->bf_enabled = false;
+ } else
+ ring->bf_enabled = true;
+
return 0;
-err_reserve:
- mlx4_qp_release_range(mdev->dev, ring->qpn, 1);
err_map:
mlx4_en_unmap_buffer(&ring->wqres.buf);
err_hwq_res:
@@ -139,6 +142,8 @@ void mlx4_en_destroy_tx_ring(struct mlx4_en_priv *priv,
struct mlx4_en_dev *mdev = priv->mdev;
en_dbg(DRV, priv, "Destroying tx ring, qpn: %d\n", ring->qpn);
+ if (ring->bf_enabled)
+ mlx4_bf_free(mdev->dev, &ring->bf);
mlx4_qp_remove(mdev->dev, &ring->qp);
mlx4_qp_free(mdev->dev, &ring->qp);
mlx4_qp_release_range(mdev->dev, ring->qpn, 1);
@@ -171,6 +176,8 @@ int mlx4_en_activate_tx_ring(struct mlx4_en_priv *priv,
mlx4_en_fill_qp_context(priv, ring->size, ring->stride, 1, 0, ring->qpn,
ring->cqn, &ring->context);
+ if (ring->bf_enabled)
+ ring->context.usr_page = cpu_to_be32(ring->bf.uar->index);
err = mlx4_qp_to_ready(mdev->dev, &ring->wqres.mtt, &ring->context,
&ring->qp, &ring->qp_state);
@@ -591,6 +598,11 @@ u16 mlx4_en_select_queue(struct net_device *dev, struct sk_buff *skb)
return skb_tx_hash(dev, skb);
}
+static void mlx4_bf_copy(unsigned long *dst, unsigned long *src, unsigned bytecnt)
+{
+ __iowrite64_copy(dst, src, bytecnt / 8);
+}
+
netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
{
struct mlx4_en_priv *priv = netdev_priv(dev);
@@ -609,12 +621,13 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
int desc_size;
int real_size;
dma_addr_t dma;
- u32 index;
+ u32 index, bf_index;
__be32 op_own;
u16 vlan_tag = 0;
int i;
int lso_header_size;
void *fragptr;
+ bool bounce = false;
if (!priv->port_up)
goto tx_drop;
@@ -657,13 +670,16 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
/* Packet is good - grab an index and transmit it */
index = ring->prod & ring->size_mask;
+ bf_index = ring->prod;
/* See if we have enough space for whole descriptor TXBB for setting
* SW ownership on next descriptor; if not, use a bounce buffer. */
if (likely(index + nr_txbb <= ring->size))
tx_desc = ring->buf + index * TXBB_SIZE;
- else
+ else {
tx_desc = (struct mlx4_en_tx_desc *) ring->bounce_buf;
+ bounce = true;
+ }
/* Save skb in tx_info ring */
tx_info = &ring->tx_info[index];
@@ -768,21 +784,37 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
ring->prod += nr_txbb;
/* If we used a bounce buffer then copy descriptor back into place */
- if (tx_desc == (struct mlx4_en_tx_desc *) ring->bounce_buf)
+ if (bounce)
tx_desc = mlx4_en_bounce_to_desc(priv, ring, index, desc_size);
/* Run destructor before passing skb to HW */
if (likely(!skb_shared(skb)))
skb_orphan(skb);
- /* Ensure new descirptor hits memory
- * before setting ownership of this descriptor to HW */
- wmb();
- tx_desc->ctrl.owner_opcode = op_own;
+ if (ring->bf_enabled && desc_size <= MAX_BF && !bounce && !vlan_tag) {
+ *(u32 *) (&tx_desc->ctrl.vlan_tag) |= ring->doorbell_qpn;
+ op_own |= htonl((bf_index & 0xffff) << 8);
+ /* Ensure new descirptor hits memory
+ * before setting ownership of this descriptor to HW */
+ wmb();
+ tx_desc->ctrl.owner_opcode = op_own;
- /* Ring doorbell! */
- wmb();
- writel(ring->doorbell_qpn, mdev->uar_map + MLX4_SEND_DOORBELL);
+ wmb();
+
+ mlx4_bf_copy(ring->bf.reg + ring->bf.offset, (unsigned long *) &tx_desc->ctrl,
+ desc_size);
+
+ wmb();
+
+ ring->bf.offset ^= ring->bf.buf_size;
+ } else {
+ /* Ensure new descirptor hits memory
+ * before setting ownership of this descriptor to HW */
+ wmb();
+ tx_desc->ctrl.owner_opcode = op_own;
+ wmb();
+ writel(ring->doorbell_qpn, ring->bf.uar->map + MLX4_SEND_DOORBELL);
+ }
/* Poll CQ here */
mlx4_en_xmit_poll(priv, tx_ind);
diff --git a/drivers/net/mlx4/mlx4_en.h b/drivers/net/mlx4/mlx4_en.h
index 5a2c560..edcb535 100644
--- a/drivers/net/mlx4/mlx4_en.h
+++ b/drivers/net/mlx4/mlx4_en.h
@@ -248,6 +248,8 @@ struct mlx4_en_tx_ring {
unsigned long bytes;
unsigned long packets;
spinlock_t comp_lock;
+ struct mlx4_bf bf;
+ bool bf_enabled;
};
struct mlx4_en_rx_desc {
@@ -518,7 +520,7 @@ u16 mlx4_en_select_queue(struct net_device *dev, struct sk_buff *skb);
netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev);
int mlx4_en_create_tx_ring(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring *ring,
- u32 size, u16 stride);
+ int qpn, u32 size, u16 stride);
void mlx4_en_destroy_tx_ring(struct mlx4_en_priv *priv, struct mlx4_en_tx_ring *ring);
int mlx4_en_activate_tx_ring(struct mlx4_en_priv *priv,
struct mlx4_en_tx_ring *ring,
--
1.6.0.2
^ permalink raw reply related
* [PATCH 13/17] mlx4: Add blue flame support for kernel consumers
From: Yevgeny Petrilin @ 2011-03-07 10:17 UTC (permalink / raw)
To: davem; +Cc: netdev, yevgenyp, eli
From: Eli Cohen <eli@mellanox.co.il>
Using blue flame can improve latency by allowing the HW to more efficiently
access the WQE. This patch presents two functions that are used to allocate or
release HW resources for using blue flame; the caller need to supply a struct
mlx4_bf object when allocating resources. Consumers that make use of this API
should post doorbells to the UAR object pointed by the initialized struct
mlx4_bf;
Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
---
drivers/net/mlx4/main.c | 31 ++++++++++++++
drivers/net/mlx4/mlx4.h | 3 +
drivers/net/mlx4/pd.c | 94 +++++++++++++++++++++++++++++++++++++++++++
include/linux/mlx4/device.h | 13 ++++++
include/linux/mlx4/qp.h | 1 +
5 files changed, 142 insertions(+), 0 deletions(-)
diff --git a/drivers/net/mlx4/main.c b/drivers/net/mlx4/main.c
index c8e2761..34581d2 100644
--- a/drivers/net/mlx4/main.c
+++ b/drivers/net/mlx4/main.c
@@ -39,6 +39,7 @@
#include <linux/pci.h>
#include <linux/dma-mapping.h>
#include <linux/slab.h>
+#include <linux/io-mapping.h>
#include <linux/mlx4/device.h>
#include <linux/mlx4/doorbell.h>
@@ -721,8 +722,31 @@ static void mlx4_free_icms(struct mlx4_dev *dev)
mlx4_free_icm(dev, priv->fw.aux_icm, 0);
}
+static int map_bf_area(struct mlx4_dev *dev)
+{
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ resource_size_t bf_start;
+ resource_size_t bf_len;
+ int err = 0;
+
+ bf_start = pci_resource_start(dev->pdev, 2) + (dev->caps.num_uars << PAGE_SHIFT);
+ bf_len = pci_resource_len(dev->pdev, 2) - (dev->caps.num_uars << PAGE_SHIFT);
+ priv->bf_mapping = io_mapping_create_wc(bf_start, bf_len);
+ if (!priv->bf_mapping)
+ err = -ENOMEM;
+
+ return err;
+}
+
+static void unmap_bf_area(struct mlx4_dev *dev)
+{
+ if (mlx4_priv(dev)->bf_mapping)
+ io_mapping_free(mlx4_priv(dev)->bf_mapping);
+}
+
static void mlx4_close_hca(struct mlx4_dev *dev)
{
+ unmap_bf_area(dev);
mlx4_CLOSE_HCA(dev, 0);
mlx4_free_icms(dev);
mlx4_UNMAP_FA(dev);
@@ -775,6 +799,9 @@ static int mlx4_init_hca(struct mlx4_dev *dev)
goto err_stop_fw;
}
+ if (map_bf_area(dev))
+ mlx4_dbg(dev, "Kernel support for blue flame is not available for kernels < 2.6.28\n");
+
init_hca.log_uar_sz = ilog2(dev->caps.num_uars);
err = mlx4_init_icm(dev, &dev_cap, &init_hca, icm_size);
@@ -805,6 +832,7 @@ err_free_icm:
mlx4_free_icms(dev);
err_stop_fw:
+ unmap_bf_area(dev);
mlx4_UNMAP_FA(dev);
mlx4_free_icm(dev, priv->fw.fw_icm, 0);
@@ -1196,6 +1224,9 @@ static int __mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
pci_read_config_byte(pdev, PCI_REVISION_ID, &dev->rev_id);
+ INIT_LIST_HEAD(&priv->bf_list);
+ mutex_init(&priv->bf_mutex);
+
/*
* Now reset the HCA before we touch the PCI capabilities or
* attempt a firmware command, since a boot ROM may have left
diff --git a/drivers/net/mlx4/mlx4.h b/drivers/net/mlx4/mlx4.h
index a50923a..008bf95 100644
--- a/drivers/net/mlx4/mlx4.h
+++ b/drivers/net/mlx4/mlx4.h
@@ -353,6 +353,9 @@ struct mlx4_priv {
struct mutex port_mutex;
struct mlx4_msix_ctl msix_ctl;
struct mlx4_steer *steer;
+ struct list_head bf_list;
+ struct mutex bf_mutex;
+ struct io_mapping *bf_mapping;
};
static inline struct mlx4_priv *mlx4_priv(struct mlx4_dev *dev)
diff --git a/drivers/net/mlx4/pd.c b/drivers/net/mlx4/pd.c
index c4988d6..5210a0f 100644
--- a/drivers/net/mlx4/pd.c
+++ b/drivers/net/mlx4/pd.c
@@ -32,6 +32,7 @@
*/
#include <linux/errno.h>
+#include <linux/io-mapping.h>
#include <asm/page.h>
@@ -77,6 +78,7 @@ int mlx4_uar_alloc(struct mlx4_dev *dev, struct mlx4_uar *uar)
return -ENOMEM;
uar->pfn = (pci_resource_start(dev->pdev, 2) >> PAGE_SHIFT) + uar->index;
+ uar->map = NULL;
return 0;
}
@@ -88,6 +90,98 @@ void mlx4_uar_free(struct mlx4_dev *dev, struct mlx4_uar *uar)
}
EXPORT_SYMBOL_GPL(mlx4_uar_free);
+int mlx4_bf_alloc(struct mlx4_dev *dev, struct mlx4_bf *bf)
+{
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ struct mlx4_uar *uar;
+ int err = 0;
+ int idx;
+
+ if (!priv->bf_mapping)
+ return -ENOMEM;
+
+ mutex_lock(&priv->bf_mutex);
+ if (!list_empty(&priv->bf_list))
+ uar = list_entry(priv->bf_list.next, struct mlx4_uar, bf_list);
+ else {
+ uar = kmalloc(sizeof *uar, GFP_KERNEL);
+ if (!uar) {
+ err = -ENOMEM;
+ goto out;
+ }
+ err = mlx4_uar_alloc(dev, uar);
+ if (err)
+ goto free_kmalloc;
+
+ uar->map = ioremap(uar->pfn << PAGE_SHIFT, PAGE_SIZE);
+ if (!uar->map) {
+ err = -ENOMEM;
+ goto free_uar;
+ }
+
+ uar->bf_map = io_mapping_map_wc(priv->bf_mapping, uar->index << PAGE_SHIFT);
+ if (!uar->bf_map) {
+ err = -ENOMEM;
+ goto unamp_uar;
+ }
+ uar->free_bf_bmap = 0;
+ list_add(&uar->bf_list, &priv->bf_list);
+ }
+
+ bf->uar = uar;
+ idx = ffz(uar->free_bf_bmap);
+ uar->free_bf_bmap |= 1 << idx;
+ bf->uar = uar;
+ bf->offset = 0;
+ bf->buf_size = dev->caps.bf_reg_size / 2;
+ bf->reg = uar->bf_map + idx * dev->caps.bf_reg_size;
+ if (uar->free_bf_bmap == (1 << dev->caps.bf_regs_per_page) - 1)
+ list_del_init(&uar->bf_list);
+
+ goto out;
+
+unamp_uar:
+ bf->uar = NULL;
+ iounmap(uar->map);
+
+free_uar:
+ mlx4_uar_free(dev, uar);
+
+free_kmalloc:
+ kfree(uar);
+
+out:
+ mutex_unlock(&priv->bf_mutex);
+ return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_bf_alloc);
+
+void mlx4_bf_free(struct mlx4_dev *dev, struct mlx4_bf *bf)
+{
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ int idx;
+
+ if (!bf->uar || !bf->uar->bf_map)
+ return;
+
+ mutex_lock(&priv->bf_mutex);
+ idx = (bf->reg - bf->uar->bf_map) / dev->caps.bf_reg_size;
+ bf->uar->free_bf_bmap &= ~(1 << idx);
+ if (!bf->uar->free_bf_bmap) {
+ if (!list_empty(&bf->uar->bf_list))
+ list_del(&bf->uar->bf_list);
+
+ io_mapping_unmap(bf->uar->bf_map);
+ iounmap(bf->uar->map);
+ mlx4_uar_free(dev, bf->uar);
+ kfree(bf->uar);
+ } else if (list_empty(&bf->uar->bf_list))
+ list_add(&bf->uar->bf_list, &priv->bf_list);
+
+ mutex_unlock(&priv->bf_mutex);
+}
+EXPORT_SYMBOL_GPL(mlx4_bf_free);
+
int mlx4_init_uar_table(struct mlx4_dev *dev)
{
if (dev->caps.num_uars <= 128) {
diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index f1cb31b..229cd65 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -351,6 +351,17 @@ struct mlx4_fmr {
struct mlx4_uar {
unsigned long pfn;
int index;
+ struct list_head bf_list;
+ unsigned free_bf_bmap;
+ void __iomem *map;
+ void __iomem *bf_map;
+};
+
+struct mlx4_bf {
+ unsigned long offset;
+ int buf_size;
+ struct mlx4_uar *uar;
+ void __iomem *reg;
};
struct mlx4_cq {
@@ -478,6 +489,8 @@ void mlx4_pd_free(struct mlx4_dev *dev, u32 pdn);
int mlx4_uar_alloc(struct mlx4_dev *dev, struct mlx4_uar *uar);
void mlx4_uar_free(struct mlx4_dev *dev, struct mlx4_uar *uar);
+int mlx4_bf_alloc(struct mlx4_dev *dev, struct mlx4_bf *bf);
+void mlx4_bf_free(struct mlx4_dev *dev, struct mlx4_bf *bf);
int mlx4_mtt_init(struct mlx4_dev *dev, int npages, int page_shift,
struct mlx4_mtt *mtt);
diff --git a/include/linux/mlx4/qp.h b/include/linux/mlx4/qp.h
index 0eeb2a1..9e9eb21 100644
--- a/include/linux/mlx4/qp.h
+++ b/include/linux/mlx4/qp.h
@@ -303,6 +303,7 @@ struct mlx4_wqe_data_seg {
enum {
MLX4_INLINE_ALIGN = 64,
+ MLX4_INLINE_SEG = 1 << 31,
};
struct mlx4_wqe_inline_seg {
--
1.6.0.2
^ permalink raw reply related
* [PATCH 08/17] mlx4_en: Reporting HW revision in ethtool -i
From: Yevgeny Petrilin @ 2011-03-07 10:17 UTC (permalink / raw)
To: davem; +Cc: netdev, yevgenyp, eugenia
HW revision is derived from device ID and rev id.
Signed-off-by: Eugenia Emantayev <eugenia@mellanox.co.il>
Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
---
drivers/infiniband/hw/mlx4/main.c | 1 -
drivers/net/mlx4/en_ethtool.c | 15 ++++++++++++++-
drivers/net/mlx4/main.c | 2 ++
drivers/net/mlx4/mlx4_en.h | 3 +++
include/linux/mlx4/device.h | 2 +-
5 files changed, 20 insertions(+), 3 deletions(-)
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index c7a6213..66e3eec 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -721,7 +721,6 @@ static int init_node_data(struct mlx4_ib_dev *dev)
if (err)
goto out;
- dev->dev->rev_id = be32_to_cpup((__be32 *) (out_mad->data + 32));
memcpy(&dev->ib_dev.node_guid, out_mad->data + 12, 8);
out:
diff --git a/drivers/net/mlx4/en_ethtool.c b/drivers/net/mlx4/en_ethtool.c
index c1f351f..62ace6c 100644
--- a/drivers/net/mlx4/en_ethtool.c
+++ b/drivers/net/mlx4/en_ethtool.c
@@ -45,7 +45,20 @@ mlx4_en_get_drvinfo(struct net_device *dev, struct ethtool_drvinfo *drvinfo)
struct mlx4_en_priv *priv = netdev_priv(dev);
struct mlx4_en_dev *mdev = priv->mdev;
- sprintf(drvinfo->driver, DRV_NAME " (%s)", mdev->dev->board_id);
+ switch (mdev->dev->rev_id) {
+ case 0xa0:
+ if (dev->dev_id >= MLX4_EN_CX3_LOW_ID && dev->dev_id <= MLX4_EN_CX3_HIGH_ID)
+ sprintf(drvinfo->driver, DRV_NAME " (%s_CX-3)", mdev->dev->board_id);
+ else
+ sprintf(drvinfo->driver, DRV_NAME " (%s_CX)", mdev->dev->board_id);
+ break;
+ case 0xb0:
+ sprintf(drvinfo->driver, DRV_NAME " (%s_CX-2)", mdev->dev->board_id);
+ break;
+ default:
+ sprintf(drvinfo->driver, DRV_NAME " (%s)", mdev->dev->board_id);
+ break;
+ }
strncpy(drvinfo->version, DRV_VERSION " (" DRV_RELDATE ")", 32);
sprintf(drvinfo->fw_version, "%d.%d.%d",
(u16) (mdev->dev->caps.fw_ver >> 32),
diff --git a/drivers/net/mlx4/main.c b/drivers/net/mlx4/main.c
index 42d4fb4..5bebb88 100644
--- a/drivers/net/mlx4/main.c
+++ b/drivers/net/mlx4/main.c
@@ -1139,6 +1139,8 @@ static int __mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
INIT_LIST_HEAD(&priv->pgdir_list);
mutex_init(&priv->pgdir_mutex);
+ pci_read_config_byte(pdev, PCI_REVISION_ID, &dev->rev_id);
+
/*
* Now reset the HCA before we touch the PCI capabilities or
* attempt a firmware command, since a boot ROM may have left
diff --git a/drivers/net/mlx4/mlx4_en.h b/drivers/net/mlx4/mlx4_en.h
index 07aea8d..ad4df66 100644
--- a/drivers/net/mlx4/mlx4_en.h
+++ b/drivers/net/mlx4/mlx4_en.h
@@ -216,6 +216,9 @@ struct mlx4_en_tx_desc {
#define MLX4_EN_USE_SRQ 0x01000000
+#define MLX4_EN_CX3_LOW_ID 0x1000
+#define MLX4_EN_CX3_HIGH_ID 0x1005
+
struct mlx4_en_rx_alloc {
struct page *page;
u16 offset;
diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index 2460356..fe2a3a3 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -422,7 +422,7 @@ struct mlx4_dev {
unsigned long flags;
struct mlx4_caps caps;
struct radix_tree_root qp_table_tree;
- u32 rev_id;
+ u8 rev_id;
char board_id[MLX4_BOARD_ID_LEN];
};
--
1.6.0.2
^ permalink raw reply related
* [PATCH 07/17] mlx4: Wake on LAN support
From: Yevgeny Petrilin @ 2011-03-07 10:17 UTC (permalink / raw)
To: davem; +Cc: netdev, yevgenyp, igory
The driver queries the FW for WOL support.
Ethtool get/set_wol is implemented accordingly.
Only magic packets are supported at the time.
Signed-off-by: Igor Yarovinsky <igory@mellanox.co.il>
Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
---
drivers/net/mlx4/en_ethtool.c | 62 +++++++++++++++++++++++++++++++++++++++-
drivers/net/mlx4/fw.c | 20 +++++++++++++
drivers/net/mlx4/fw.h | 1 +
drivers/net/mlx4/main.c | 1 +
drivers/net/mlx4/mlx4_en.h | 7 ++++
include/linux/mlx4/device.h | 4 ++
6 files changed, 93 insertions(+), 2 deletions(-)
diff --git a/drivers/net/mlx4/en_ethtool.c b/drivers/net/mlx4/en_ethtool.c
index 8cfe858..c1f351f 100644
--- a/drivers/net/mlx4/en_ethtool.c
+++ b/drivers/net/mlx4/en_ethtool.c
@@ -131,8 +131,65 @@ static void mlx4_en_set_msglevel(struct net_device *dev, u32 val)
static void mlx4_en_get_wol(struct net_device *netdev,
struct ethtool_wolinfo *wol)
{
- wol->supported = 0;
- wol->wolopts = 0;
+ struct mlx4_en_priv *priv = netdev_priv(netdev);
+ int err = 0;
+ u64 config = 0;
+
+ if (!priv->mdev->dev->caps.wol) {
+ wol->supported = 0;
+ wol->wolopts = 0;
+ return;
+ }
+
+ err = mlx4_wol_read(priv->mdev->dev, &config, priv->port);
+ if (err) {
+ en_err(priv, "Failed to get WoL information\n");
+ return;
+ }
+
+ if (config & MLX4_EN_WOL_MAGIC)
+ wol->supported = WAKE_MAGIC;
+ else
+ wol->supported = 0;
+
+ if (config & MLX4_EN_WOL_ENABLED)
+ wol->wolopts = WAKE_MAGIC;
+ else
+ wol->wolopts = 0;
+}
+
+static int mlx4_en_set_wol(struct net_device *netdev,
+ struct ethtool_wolinfo *wol)
+{
+ struct mlx4_en_priv *priv = netdev_priv(netdev);
+ u64 config = 0;
+ int err = 0;
+
+ if (!priv->mdev->dev->caps.wol)
+ return -EOPNOTSUPP;
+
+ if (wol->supported & ~WAKE_MAGIC)
+ return -EINVAL;
+
+ err = mlx4_wol_read(priv->mdev->dev, &config, priv->port);
+ if (err) {
+ en_err(priv, "Failed to get WoL info, unable to modify\n");
+ return err;
+ }
+
+ if (wol->wolopts & WAKE_MAGIC) {
+ config |= MLX4_EN_WOL_DO_MODIFY | MLX4_EN_WOL_ENABLED |
+ MLX4_EN_WOL_MAGIC;
+ } else {
+ config &= ~(MLX4_EN_WOL_ENABLED | MLX4_EN_WOL_MAGIC);
+ config |= MLX4_EN_WOL_DO_MODIFY;
+ }
+
+ err = mlx4_wol_write(priv->mdev->dev, config, priv->port);
+ if (err)
+ en_err(priv, "Failed to set WoL information\n");
+
+ return err;
}
static int mlx4_en_get_sset_count(struct net_device *dev, int sset)
@@ -442,6 +499,7 @@ const struct ethtool_ops mlx4_en_ethtool_ops = {
.get_ethtool_stats = mlx4_en_get_ethtool_stats,
.self_test = mlx4_en_self_test,
.get_wol = mlx4_en_get_wol,
+ .set_wol = mlx4_en_set_wol,
.get_msglevel = mlx4_en_get_msglevel,
.set_msglevel = mlx4_en_set_msglevel,
.get_coalesce = mlx4_en_get_coalesce,
diff --git a/drivers/net/mlx4/fw.c b/drivers/net/mlx4/fw.c
index 5de1db8..fd1c51b 100644
--- a/drivers/net/mlx4/fw.c
+++ b/drivers/net/mlx4/fw.c
@@ -276,6 +276,7 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
dev_cap->udp_rss = field & 0x1;
MLX4_GET(field, outbox, QUERY_DEV_CAP_ETH_UC_LOOPBACK_OFFSET);
dev_cap->loopback_support = field & 0x1;
+ dev_cap->wol = field & 0x40;
MLX4_GET(dev_cap->flags, outbox, QUERY_DEV_CAP_FLAGS_OFFSET);
MLX4_GET(field, outbox, QUERY_DEV_CAP_RSVD_UAR_OFFSET);
dev_cap->reserved_uars = field >> 4;
@@ -908,3 +909,22 @@ int mlx4_NOP(struct mlx4_dev *dev)
/* Input modifier of 0x1f means "finish as soon as possible." */
return mlx4_cmd(dev, 0, 0x1f, 0, MLX4_CMD_NOP, 100);
}
+
+#define MLX4_WOL_SETUP_MODE (5 << 28)
+int mlx4_wol_read(struct mlx4_dev *dev, u64 *config, int port)
+{
+ u32 in_mod = MLX4_WOL_SETUP_MODE | port << 8;
+
+ return mlx4_cmd_imm(dev, 0, config, in_mod, 0x3,
+ MLX4_CMD_MOD_STAT_CFG, MLX4_CMD_TIME_CLASS_A);
+}
+EXPORT_SYMBOL_GPL(mlx4_wol_read);
+
+int mlx4_wol_write(struct mlx4_dev *dev, u64 config, int port)
+{
+ u32 in_mod = MLX4_WOL_SETUP_MODE | port << 8;
+
+ return mlx4_cmd(dev, config, in_mod, 0x1, MLX4_CMD_MOD_STAT_CFG,
+ MLX4_CMD_TIME_CLASS_A);
+}
+EXPORT_SYMBOL_GPL(mlx4_wol_write);
diff --git a/drivers/net/mlx4/fw.h b/drivers/net/mlx4/fw.h
index 65cc72e..f7b9cc2 100644
--- a/drivers/net/mlx4/fw.h
+++ b/drivers/net/mlx4/fw.h
@@ -80,6 +80,7 @@ struct mlx4_dev_cap {
u16 stat_rate_support;
int udp_rss;
int loopback_support;
+ int wol;
u32 flags;
int reserved_uars;
int uar_size;
diff --git a/drivers/net/mlx4/main.c b/drivers/net/mlx4/main.c
index 517ca34..42d4fb4 100644
--- a/drivers/net/mlx4/main.c
+++ b/drivers/net/mlx4/main.c
@@ -227,6 +227,7 @@ static int mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
dev->caps.stat_rate_support = dev_cap->stat_rate_support;
dev->caps.udp_rss = dev_cap->udp_rss;
dev->caps.loopback_support = dev_cap->loopback_support;
+ dev->caps.wol = dev_cap->wol;
dev->caps.max_gso_sz = dev_cap->max_gso_sz;
dev->caps.log_num_macs = log_num_mac;
diff --git a/drivers/net/mlx4/mlx4_en.h b/drivers/net/mlx4/mlx4_en.h
index 2db245f..07aea8d 100644
--- a/drivers/net/mlx4/mlx4_en.h
+++ b/drivers/net/mlx4/mlx4_en.h
@@ -479,6 +479,13 @@ struct mlx4_en_priv {
int mc_addrs_cnt;
struct mlx4_en_stat_out_mbox hw_stats;
int vids[128];
+ bool wol;
+};
+
+enum mlx4_en_wol {
+ MLX4_EN_WOL_MAGIC = (1ULL << 61),
+ MLX4_EN_WOL_ENABLED = (1ULL << 62),
+ MLX4_EN_WOL_DO_MODIFY = (1ULL << 63),
};
diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index 7838082..2460356 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -251,6 +251,7 @@ struct mlx4_caps {
u16 stat_rate_support;
int udp_rss;
int loopback_support;
+ int wol;
u8 port_width_cap[MLX4_MAX_PORTS + 1];
int max_gso_sz;
int reserved_qps_cnt[MLX4_NUM_QP_REGION];
@@ -535,4 +536,7 @@ int mlx4_test_interrupts(struct mlx4_dev *dev);
int mlx4_assign_eq(struct mlx4_dev *dev, char* name , int* vector);
void mlx4_release_eq(struct mlx4_dev *dev, int vec);
+int mlx4_wol_read(struct mlx4_dev *dev, u64 *config, int port);
+int mlx4_wol_write(struct mlx4_dev *dev, u64 config, int port);
+
#endif /* MLX4_DEVICE_H */
--
1.6.0.2
^ permalink raw reply related
* [PATCH 06/17] mlx4_en: using new mlx4 interrupt scheme
From: Yevgeny Petrilin @ 2011-03-07 10:17 UTC (permalink / raw)
To: davem; +Cc: netdev, yevgenyp, markuze
Each RX ring will have its own interrupt vector, and TX rings will share one
(we mostly use polling for TX completions).
The vectors are assigned first time device is opened, and its name includes
the interface name and ring number.
Signed-off-by: Markuze Alex <markuze@mellanox.co.il>
Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
---
drivers/net/mlx4/en_cq.c | 38 +++++++++++++++++++++++++++++++-------
drivers/net/mlx4/en_ethtool.c | 2 +-
drivers/net/mlx4/en_main.c | 20 +++++++++++---------
drivers/net/mlx4/en_netdev.c | 18 ++++++++++++++----
drivers/net/mlx4/mlx4_en.h | 7 +++++--
5 files changed, 62 insertions(+), 23 deletions(-)
diff --git a/drivers/net/mlx4/en_cq.c b/drivers/net/mlx4/en_cq.c
index 21786ad..ec4b6d0 100644
--- a/drivers/net/mlx4/en_cq.c
+++ b/drivers/net/mlx4/en_cq.c
@@ -51,13 +51,10 @@ int mlx4_en_create_cq(struct mlx4_en_priv *priv,
int err;
cq->size = entries;
- if (mode == RX) {
+ if (mode == RX)
cq->buf_size = cq->size * sizeof(struct mlx4_cqe);
- cq->vector = ring % mdev->dev->caps.num_comp_vectors;
- } else {
+ else
cq->buf_size = sizeof(struct mlx4_cqe);
- cq->vector = 0;
- }
cq->ring = ring;
cq->is_tx = mode;
@@ -80,7 +77,8 @@ int mlx4_en_create_cq(struct mlx4_en_priv *priv,
int mlx4_en_activate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq)
{
struct mlx4_en_dev *mdev = priv->mdev;
- int err;
+ int err = 0;
+ char name[25];
cq->dev = mdev->pndev[priv->port];
cq->mcq.set_ci_db = cq->wqres.db.db;
@@ -89,6 +87,29 @@ int mlx4_en_activate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq)
*cq->mcq.arm_db = 0;
memset(cq->buf, 0, cq->buf_size);
+ if (cq->is_tx == RX) {
+ if (mdev->dev->caps.comp_pool) {
+ if (!cq->vector) {
+ sprintf(name , "%s-rx-%d", priv->dev->name, cq->ring);
+ if (mlx4_assign_eq(mdev->dev, name, &cq->vector)) {
+ cq->vector = (cq->ring + 1 + priv->port) %
+ mdev->dev->caps.num_comp_vectors;
+ mlx4_warn(mdev, "Failed Assigning an EQ to "
+ "%s_rx-%d ,Falling back to legacy EQ's\n",
+ priv->dev->name, cq->ring);
+ }
+ }
+ } else {
+ cq->vector = (cq->ring + 1 + priv->port) %
+ mdev->dev->caps.num_comp_vectors;
+ }
+ } else {
+ if (!cq->vector || !mdev->dev->caps.comp_pool) {
+ /*Fallback to legacy pool in case of error*/
+ cq->vector = 0;
+ }
+ }
+
if (!cq->is_tx)
cq->size = priv->rx_ring[cq->ring].actual_size;
@@ -112,12 +133,15 @@ int mlx4_en_activate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq)
return 0;
}
-void mlx4_en_destroy_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq)
+void mlx4_en_destroy_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq,
+ bool reserve_vectors)
{
struct mlx4_en_dev *mdev = priv->mdev;
mlx4_en_unmap_buffer(&cq->wqres.buf);
mlx4_free_hwq_res(mdev->dev, &cq->wqres, cq->buf_size);
+ if (priv->mdev->dev->caps.comp_pool && cq->vector && !reserve_vectors)
+ mlx4_release_eq(priv->mdev->dev, cq->vector);
cq->buf_size = 0;
cq->buf = NULL;
}
diff --git a/drivers/net/mlx4/en_ethtool.c b/drivers/net/mlx4/en_ethtool.c
index 056152b..8cfe858 100644
--- a/drivers/net/mlx4/en_ethtool.c
+++ b/drivers/net/mlx4/en_ethtool.c
@@ -388,7 +388,7 @@ static int mlx4_en_set_ringparam(struct net_device *dev,
mlx4_en_stop_port(dev);
}
- mlx4_en_free_resources(priv);
+ mlx4_en_free_resources(priv, true);
priv->prof->tx_ring_size = tx_size;
priv->prof->rx_ring_size = rx_size;
diff --git a/drivers/net/mlx4/en_main.c b/drivers/net/mlx4/en_main.c
index 1ff6ca6..29aaa43 100644
--- a/drivers/net/mlx4/en_main.c
+++ b/drivers/net/mlx4/en_main.c
@@ -241,16 +241,18 @@ static void *mlx4_en_add(struct mlx4_dev *dev)
mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH)
mdev->port_cnt++;
- /* If we did not receive an explicit number of Rx rings, default to
- * the number of completion vectors populated by the mlx4_core */
mlx4_foreach_port(i, dev, MLX4_PORT_TYPE_ETH) {
- mlx4_info(mdev, "Using %d tx rings for port:%d\n",
- mdev->profile.prof[i].tx_ring_num, i);
- mdev->profile.prof[i].rx_ring_num = min_t(int,
- roundup_pow_of_two(dev->caps.num_comp_vectors),
- MAX_RX_RINGS);
- mlx4_info(mdev, "Defaulting to %d rx rings for port:%d\n",
- mdev->profile.prof[i].rx_ring_num, i);
+ if (!dev->caps.comp_pool) {
+ mdev->profile.prof[i].rx_ring_num =
+ rounddown_pow_of_two(max_t(int, MIN_RX_RINGS,
+ min_t(int,
+ dev->caps.num_comp_vectors,
+ MAX_RX_RINGS)));
+ } else {
+ mdev->profile.prof[i].rx_ring_num = rounddown_pow_of_two(
+ min_t(int, dev->caps.comp_pool/
+ dev->caps.num_ports - 1 , MAX_MSIX_P_PORT - 1));
+ }
}
/* Create our own workqueue for reset/multicast tasks
diff --git a/drivers/net/mlx4/en_netdev.c b/drivers/net/mlx4/en_netdev.c
index 5727bf5..f0d06a0 100644
--- a/drivers/net/mlx4/en_netdev.c
+++ b/drivers/net/mlx4/en_netdev.c
@@ -557,6 +557,7 @@ int mlx4_en_start_port(struct net_device *dev)
int err = 0;
int i;
int j;
+ char name[32];
if (priv->port_up) {
en_dbg(DRV, priv, "start port called while port already up\n");
@@ -601,10 +602,19 @@ int mlx4_en_start_port(struct net_device *dev)
goto cq_err;
}
+ if (mdev->dev->caps.comp_pool && !priv->tx_vector) {
+ sprintf(name , "%s_tx", priv->dev->name);
+ if (mlx4_assign_eq(mdev->dev , name, &priv->tx_vector)) {
+ mlx4_warn(mdev, "Failed Assigning an EQ to "
+ "%s_tx ,Falling back to legacy "
+ "EQ's\n", priv->dev->name);
+ }
+ }
/* Configure tx cq's and rings */
for (i = 0; i < priv->tx_ring_num; i++) {
/* Configure cq */
cq = &priv->tx_cq[i];
+ cq->vector = priv->tx_vector;
err = mlx4_en_activate_cq(priv, cq);
if (err) {
en_err(priv, "Failed allocating Tx CQ\n");
@@ -819,7 +829,7 @@ static int mlx4_en_close(struct net_device *dev)
return 0;
}
-void mlx4_en_free_resources(struct mlx4_en_priv *priv)
+void mlx4_en_free_resources(struct mlx4_en_priv *priv, bool reserve_vectors)
{
int i;
@@ -827,14 +837,14 @@ void mlx4_en_free_resources(struct mlx4_en_priv *priv)
if (priv->tx_ring[i].tx_info)
mlx4_en_destroy_tx_ring(priv, &priv->tx_ring[i]);
if (priv->tx_cq[i].buf)
- mlx4_en_destroy_cq(priv, &priv->tx_cq[i]);
+ mlx4_en_destroy_cq(priv, &priv->tx_cq[i], reserve_vectors);
}
for (i = 0; i < priv->rx_ring_num; i++) {
if (priv->rx_ring[i].rx_info)
mlx4_en_destroy_rx_ring(priv, &priv->rx_ring[i]);
if (priv->rx_cq[i].buf)
- mlx4_en_destroy_cq(priv, &priv->rx_cq[i]);
+ mlx4_en_destroy_cq(priv, &priv->rx_cq[i], reserve_vectors);
}
}
@@ -896,7 +906,7 @@ void mlx4_en_destroy_netdev(struct net_device *dev)
mdev->pndev[priv->port] = NULL;
mutex_unlock(&mdev->state_lock);
- mlx4_en_free_resources(priv);
+ mlx4_en_free_resources(priv, false);
free_netdev(dev);
}
diff --git a/drivers/net/mlx4/mlx4_en.h b/drivers/net/mlx4/mlx4_en.h
index 5387c4e..2db245f 100644
--- a/drivers/net/mlx4/mlx4_en.h
+++ b/drivers/net/mlx4/mlx4_en.h
@@ -62,6 +62,7 @@
#define MLX4_EN_PAGE_SHIFT 12
#define MLX4_EN_PAGE_SIZE (1 << MLX4_EN_PAGE_SHIFT)
#define MAX_RX_RINGS 16
+#define MIN_RX_RINGS 4
#define TXBB_SIZE 64
#define HEADROOM (2048 / TXBB_SIZE + 1)
#define STAMP_STRIDE 64
@@ -462,6 +463,7 @@ struct mlx4_en_priv {
u16 log_rx_info;
struct mlx4_en_tx_ring tx_ring[MAX_TX_RINGS];
+ int tx_vector;
struct mlx4_en_rx_ring rx_ring[MAX_RX_RINGS];
struct mlx4_en_cq tx_cq[MAX_TX_RINGS];
struct mlx4_en_cq rx_cq[MAX_RX_RINGS];
@@ -487,12 +489,13 @@ int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port,
int mlx4_en_start_port(struct net_device *dev);
void mlx4_en_stop_port(struct net_device *dev);
-void mlx4_en_free_resources(struct mlx4_en_priv *priv);
+void mlx4_en_free_resources(struct mlx4_en_priv *priv, bool reserve_vectors);
int mlx4_en_alloc_resources(struct mlx4_en_priv *priv);
int mlx4_en_create_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq,
int entries, int ring, enum cq_type mode);
-void mlx4_en_destroy_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq);
+void mlx4_en_destroy_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq,
+ bool reserve_vectors);
int mlx4_en_activate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq);
void mlx4_en_deactivate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq);
int mlx4_en_set_cq_moder(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq);
--
1.6.0.2
^ permalink raw reply related
* [PATCH 04/17] mlx4_en: bringing link up when registering net device
From: Yevgeny Petrilin @ 2011-03-07 10:17 UTC (permalink / raw)
To: davem; +Cc: netdev, yevgenyp, eugenia
Signed-off-by: Eugenia Emantayev <eugenia@mellanox.co.il>
Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
---
drivers/net/mlx4/en_netdev.c | 17 +++++++++++++++++
1 files changed, 17 insertions(+), 0 deletions(-)
diff --git a/drivers/net/mlx4/en_netdev.c b/drivers/net/mlx4/en_netdev.c
index b1180dd..5727bf5 100644
--- a/drivers/net/mlx4/en_netdev.c
+++ b/drivers/net/mlx4/en_netdev.c
@@ -1069,6 +1069,23 @@ int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port,
en_warn(priv, "Using %d TX rings\n", prof->tx_ring_num);
en_warn(priv, "Using %d RX rings\n", prof->rx_ring_num);
+ /* Configure port */
+ err = mlx4_SET_PORT_general(mdev->dev, priv->port,
+ MLX4_EN_MIN_MTU,
+ 0, 0, 0, 0);
+ if (err) {
+ en_err(priv, "Failed setting port general configurations "
+ "for port %d, with error %d\n", priv->port, err);
+ goto out;
+ }
+
+ /* Init port */
+ en_warn(priv, "Initializing port\n");
+ err = mlx4_INIT_PORT(mdev->dev, priv->port);
+ if (err) {
+ en_err(priv, "Failed Initializing port\n");
+ goto out;
+ }
priv->registered = 1;
mlx4_en_set_default_moderation(priv);
queue_delayed_work(mdev->workqueue, &priv->stats_task, STATS_DELAY);
--
1.6.0.2
^ permalink raw reply related
* [PATCH 02/17] mlx4_en: moderation parameters are not reseted.
From: Yevgeny Petrilin @ 2011-03-07 10:17 UTC (permalink / raw)
To: davem; +Cc: netdev, yevgenyp
Instead of reseting the module parameters each ifup or mtu change,
they are being set once at device initialization
Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
---
drivers/net/mlx4/en_netdev.c | 3 +--
1 files changed, 1 insertions(+), 2 deletions(-)
diff --git a/drivers/net/mlx4/en_netdev.c b/drivers/net/mlx4/en_netdev.c
index 897f576..56627c2 100644
--- a/drivers/net/mlx4/en_netdev.c
+++ b/drivers/net/mlx4/en_netdev.c
@@ -801,7 +801,6 @@ static int mlx4_en_open(struct net_device *dev)
priv->rx_ring[i].packets = 0;
}
- mlx4_en_set_default_moderation(priv);
err = mlx4_en_start_port(dev);
if (err)
en_err(priv, "Failed starting port:%d\n", priv->port);
@@ -932,7 +931,6 @@ static int mlx4_en_change_mtu(struct net_device *dev, int new_mtu)
en_dbg(DRV, priv, "Change MTU called with card down!?\n");
} else {
mlx4_en_stop_port(dev);
- mlx4_en_set_default_moderation(priv);
err = mlx4_en_start_port(dev);
if (err) {
en_err(priv, "Failed restarting port:%d\n",
@@ -1080,6 +1078,7 @@ int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port,
en_warn(priv, "Using %d RX rings\n", prof->rx_ring_num);
priv->registered = 1;
+ mlx4_en_set_default_moderation(priv);
queue_delayed_work(mdev->workqueue, &priv->stats_task, STATS_DELAY);
return 0;
--
1.6.0.2
^ permalink raw reply related
* [PATCH 00/17] mlx4_en: driver updates
From: Yevgeny Petrilin @ 2011-03-07 10:16 UTC (permalink / raw)
To: davem; +Cc: netdev, yevgenyp
Hello David,
This patch set presents a set of fixes, performance improvements and new features for the mlx4_core and mlx4_en modules.
main changes are:
- MSI-X per ring
- Changing RX packet steering model for supported devices
- BlueFlame support for better TX latency.
- Adaptive moderation improvements.
Patches were generated against net-next-2.6.git
Thanks,
Yevgeny
^ permalink raw reply
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