* Re: [PATCH net-next v2] block/drbd: use nla_put_u64_64bit()
From: Eric Dumazet @ 2016-05-04 17:13 UTC (permalink / raw)
To: David Miller
Cc: nicolas.dichtel, netdev, philipp.reisner, drbd-dev, linux-kernel
In-Reply-To: <20160504.125057.1606520320212390037.davem@davemloft.net>
On Wed, 2016-05-04 at 12:50 -0400, David Miller wrote:
> From: Eric Dumazet <eric.dumazet@gmail.com>
> Date: Wed, 04 May 2016 07:27:06 -0700
>
> > kernel was fine, and most user land apps were fine as well.
>
> Userland should really not have to deal with garbage like this.
>
> And because it quietly works just fine on x86-64, nothing makes
> sure that applications will universally get this right anyways.
>
> Better to align things properly and magically all of these issues
> simply disappear.
Sure, but in practice we end up consuming 16 bytes (instead of 12) per
u64 attribute, only on some arches. 33 % space overhead.
So maybe some dumps will abort on those arches, while on x86 the size of
skb might be below some magic limit.
I guess this is fine, we do not break ABI in any way.
^ permalink raw reply
* [PATCH net-next 2/2] sfc: allocate rx pages on the same node as the interrupt
From: Edward Cree @ 2016-05-04 17:02 UTC (permalink / raw)
To: linux-net-drivers, netdev, David Miller; +Cc: Daniel Pieczko
In-Reply-To: <572A29AF.1040607@solarflare.com>
From: Daniel Pieczko <dpieczko@solarflare.com>
When the interrupt servicing a channel is on a NUMA node that is
not local to the device, performance is improved by allocating
rx pages on the node local to the interrupt (remote to the device)
The performance-optimal case, where interrupts and applications
are pinned to CPUs on the same node as the device, is not altered
by this change.
This change gave a 1% improvement in transaction rate using Nginx
with all interrupts and Nginx threads on the node remote to the
device. It also gave a small reduction in round-trip latency,
again with the interrupt and application on a different node to
the device.
Allocating rx pages based on the channel->irq_node value is only
valid for the initial driver-load interrupt affinities; if an
interrupt is moved later, the wrong node may be used for the
allocation.
Signed-off-by: Bert Kenward <bkenward@solarflare.com>
Signed-off-by: Edward Cree <ecree@solarflare.com>
---
drivers/net/ethernet/sfc/efx.c | 38 +++++++++++++++++++++++++++++++++++
drivers/net/ethernet/sfc/net_driver.h | 3 +++
drivers/net/ethernet/sfc/rx.c | 18 ++++++++++++++---
3 files changed, 56 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/sfc/efx.c b/drivers/net/ethernet/sfc/efx.c
index e6fdf35..f7e6ce5 100644
--- a/drivers/net/ethernet/sfc/efx.c
+++ b/drivers/net/ethernet/sfc/efx.c
@@ -449,6 +449,7 @@ efx_alloc_channel(struct efx_nic *efx, int i, struct efx_channel *old_channel)
channel->efx = efx;
channel->channel = i;
channel->type = &efx_default_channel_type;
+ channel->irq_mem_node = NUMA_NO_NODE;
for (j = 0; j < EFX_TXQ_TYPES; j++) {
tx_queue = &channel->tx_queue[j];
@@ -1569,6 +1570,38 @@ static int efx_probe_interrupts(struct efx_nic *efx)
return 0;
}
+#ifndef CONFIG_SMP
+static void efx_set_interrupt_affinity(struct efx_nic *efx __always_unused)
+{
+}
+
+static void efx_clear_interrupt_affinity(struct efx_nic *efx __always_unused)
+{
+}
+#else
+static void efx_set_interrupt_affinity(struct efx_nic *efx)
+{
+ struct efx_channel *channel;
+ unsigned int cpu;
+
+ efx_for_each_channel(channel, efx) {
+ cpu = cpumask_local_spread(channel->channel,
+ pcibus_to_node(efx->pci_dev->bus));
+
+ irq_set_affinity_hint(channel->irq, cpumask_of(cpu));
+ channel->irq_mem_node = cpu_to_mem(cpu);
+ }
+}
+
+static void efx_clear_interrupt_affinity(struct efx_nic *efx)
+{
+ struct efx_channel *channel;
+
+ efx_for_each_channel(channel, efx)
+ irq_set_affinity_hint(channel->irq, NULL);
+}
+#endif /* CONFIG_SMP */
+
static int efx_soft_enable_interrupts(struct efx_nic *efx)
{
struct efx_channel *channel, *end_channel;
@@ -3017,6 +3050,7 @@ static void efx_pci_remove_main(struct efx_nic *efx)
cancel_work_sync(&efx->reset_work);
efx_disable_interrupts(efx);
+ efx_clear_interrupt_affinity(efx);
efx_nic_fini_interrupt(efx);
efx_fini_port(efx);
efx->type->fini(efx);
@@ -3166,6 +3200,9 @@ static int efx_pci_probe_main(struct efx_nic *efx)
rc = efx_nic_init_interrupt(efx);
if (rc)
goto fail5;
+
+ efx_set_interrupt_affinity(efx);
+
rc = efx_enable_interrupts(efx);
if (rc)
goto fail6;
@@ -3173,6 +3210,7 @@ static int efx_pci_probe_main(struct efx_nic *efx)
return 0;
fail6:
+ efx_clear_interrupt_affinity(efx);
efx_nic_fini_interrupt(efx);
fail5:
efx_fini_port(efx);
diff --git a/drivers/net/ethernet/sfc/net_driver.h b/drivers/net/ethernet/sfc/net_driver.h
index 38c4223..28c3673 100644
--- a/drivers/net/ethernet/sfc/net_driver.h
+++ b/drivers/net/ethernet/sfc/net_driver.h
@@ -423,6 +423,7 @@ enum efx_sync_events_state {
* @sync_events_state: Current state of sync events on this channel
* @sync_timestamp_major: Major part of the last ptp sync event
* @sync_timestamp_minor: Minor part of the last ptp sync event
+ * @irq_mem_node: Memory NUMA node of interrupt
*/
struct efx_channel {
struct efx_nic *efx;
@@ -468,6 +469,8 @@ struct efx_channel {
enum efx_sync_events_state sync_events_state;
u32 sync_timestamp_major;
u32 sync_timestamp_minor;
+
+ int irq_mem_node;
};
#ifdef CONFIG_NET_RX_BUSY_POLL
diff --git a/drivers/net/ethernet/sfc/rx.c b/drivers/net/ethernet/sfc/rx.c
index 8956995..eed0c3b 100644
--- a/drivers/net/ethernet/sfc/rx.c
+++ b/drivers/net/ethernet/sfc/rx.c
@@ -163,9 +163,21 @@ static int efx_init_rx_buffers(struct efx_rx_queue *rx_queue, bool atomic)
do {
page = efx_reuse_page(rx_queue);
if (page == NULL) {
- page = alloc_pages(__GFP_COLD | __GFP_COMP |
- (atomic ? GFP_ATOMIC : GFP_KERNEL),
- efx->rx_buffer_order);
+ /* GFP_ATOMIC may fail because of various reasons,
+ * and we re-schedule rx_fill from non-atomic
+ * context in such a case. So, use __GFP_NO_WARN
+ * in case of atomic.
+ */
+ struct efx_channel *channel;
+
+ channel = efx_rx_queue_channel(rx_queue);
+ page = alloc_pages_node(channel->irq_mem_node,
+ __GFP_COMP |
+ (atomic ?
+ (GFP_ATOMIC | __GFP_NOWARN)
+ : GFP_KERNEL),
+ efx->rx_buffer_order);
+
if (unlikely(page == NULL))
return -ENOMEM;
dma_addr =
^ permalink raw reply related
* [PATCH net-next 1/2] sfc: Support setting rss_cpus to 'cores', 'packages' or 'hyperthreads'
From: Edward Cree @ 2016-05-04 17:01 UTC (permalink / raw)
To: linux-net-drivers, netdev, David Miller
In-Reply-To: <572A29AF.1040607@solarflare.com>
These settings autoconfigure the number of RSS channels to match the number of
CPUs present.
Signed-off-by: Edward Cree <ecree@solarflare.com>
---
drivers/net/ethernet/sfc/efx.c | 143 ++++++++++++++++++++++++++++++++---------
1 file changed, 113 insertions(+), 30 deletions(-)
diff --git a/drivers/net/ethernet/sfc/efx.c b/drivers/net/ethernet/sfc/efx.c
index 0705ec86..e6fdf35 100644
--- a/drivers/net/ethernet/sfc/efx.c
+++ b/drivers/net/ethernet/sfc/efx.c
@@ -22,6 +22,7 @@
#include <linux/gfp.h>
#include <linux/aer.h>
#include <linux/interrupt.h>
+#include <xen/xen.h>
#include "net_driver.h"
#include "efx.h"
#include "nic.h"
@@ -161,16 +162,19 @@ static unsigned int tx_irq_mod_usec = 150;
*/
static unsigned int interrupt_mode;
-/* This is the requested number of CPUs to use for Receive-Side Scaling (RSS),
- * i.e. the number of CPUs among which we may distribute simultaneous
- * interrupt handling.
+/* This is the requested number of CPUs to use for Receive-Side Scaling
+ * (RSS), i.e. the number of CPUs among which we may distribute
+ * simultaneous interrupt handling. Or alternatively it may be set to
+ * "packages", "cores" or "hyperthreads" to get one receive channel per
+ * package, core or hyperthread. The default is "cores".
*
- * Cards without MSI-X will only target one CPU via legacy or MSI interrupt.
- * The default (0) means to assign an interrupt to each core.
+ * Systems without MSI-X will only target one CPU via legacy or MSI
+ * interrupt.
*/
-static unsigned int rss_cpus;
-module_param(rss_cpus, uint, 0444);
-MODULE_PARM_DESC(rss_cpus, "Number of CPUs to use for Receive-Side Scaling");
+static char *rss_cpus;
+module_param(rss_cpus, charp, 0444);
+MODULE_PARM_DESC(rss_cpus, "Number of CPUs to use for Receive-Side Scaling, "
+ "or 'packages', 'cores' or 'hyperthreads'");
static bool phy_flash_cfg;
module_param(phy_flash_cfg, bool, 0644);
@@ -1324,51 +1328,130 @@ void efx_set_default_rx_indir_table(struct efx_nic *efx)
ethtool_rxfh_indir_default(i, efx->rss_spread);
}
-static unsigned int efx_wanted_parallelism(struct efx_nic *efx)
+/* Count the number of unique packages in the given cpumask */
+static unsigned int efx_num_packages(const cpumask_t *in)
{
- cpumask_var_t thread_mask;
+ cpumask_var_t core_mask;
+ unsigned int count;
+ int cpu, cpu2;
+
+ if (unlikely(!zalloc_cpumask_var(&core_mask, GFP_KERNEL))) {
+ pr_warn("sfc: RSS disabled due to allocation failure\n");
+ return 1;
+ }
+
+ count = 0;
+ for_each_cpu(cpu, in) {
+ if (!cpumask_test_cpu(cpu, core_mask)) {
+ ++count;
+
+ /* Treat each numa node as a seperate package */
+ for_each_cpu(cpu2, topology_core_cpumask(cpu)) {
+ if (cpu_to_node(cpu) == cpu_to_node(cpu2))
+ cpumask_set_cpu(cpu2, core_mask);
+ }
+ }
+ }
+
+ free_cpumask_var(core_mask);
+
+ return count;
+}
+
+/* Count the number of unique cores in the given cpumask */
+static unsigned int efx_num_cores(const cpumask_t *in)
+{
+ cpumask_var_t core_mask;
unsigned int count;
int cpu;
- if (rss_cpus) {
- count = rss_cpus;
- } else {
- if (unlikely(!zalloc_cpumask_var(&thread_mask, GFP_KERNEL))) {
- netif_warn(efx, probe, efx->net_dev,
- "RSS disabled due to allocation failure\n");
- return 1;
+ if (unlikely(!zalloc_cpumask_var(&core_mask, GFP_KERNEL))) {
+ pr_warn("sfc: RSS disabled due to allocation failure\n");
+ return 1;
+ }
+
+ count = 0;
+ for_each_cpu(cpu, in) {
+ if (!cpumask_test_cpu(cpu, core_mask)) {
+ ++count;
+ cpumask_or(core_mask, core_mask,
+ topology_sibling_cpumask(cpu));
}
+ }
- count = 0;
- for_each_online_cpu(cpu) {
- if (!cpumask_test_cpu(cpu, thread_mask)) {
- ++count;
- cpumask_or(thread_mask, thread_mask,
- topology_sibling_cpumask(cpu));
- }
+ free_cpumask_var(core_mask);
+ return count;
+}
+
+static unsigned int efx_wanted_parallelism(struct efx_nic *efx)
+{
+ struct net_device *net_dev = efx->net_dev;
+ unsigned int n_rxq;
+
+ if (rss_cpus == NULL) {
+ /* default to 'cores' */
+ goto cores;
+ } else if (strcmp(rss_cpus, "packages") == 0) {
+ if (xen_domain()) {
+ netif_warn(efx, drv, net_dev,
+ "Unable to determine CPU topology on Xen reliably. Using 4 rss channels.\n");
+ n_rxq = 4;
+ } else {
+ netif_dbg(efx, drv, net_dev,
+ "using efx_num_packages()\n");
+ n_rxq = efx_num_packages(cpu_online_mask);
+ /* Create two RSS queues even with a single package */
+ if (n_rxq == 1)
+ n_rxq = 2;
}
+ } else if (strcmp(rss_cpus, "cores") == 0) {
+cores:
+ if (xen_domain()) {
+ netif_warn(efx, drv, net_dev,
+ "Unable to determine CPU topology on Xen reliably. Assuming hyperthreading enabled.\n");
+ n_rxq = max_t(int, 1, num_online_cpus() / 2);
+ } else {
+ netif_dbg(efx, drv, net_dev,
+ "using efx_num_cores()\n");
+ n_rxq = efx_num_cores(cpu_online_mask);
+ }
+ } else if (strcmp(rss_cpus, "hyperthreads") == 0) {
+ n_rxq = num_online_cpus();
+ } else if (sscanf(rss_cpus, "%u", &n_rxq) == 1 && n_rxq > 0) {
+ /* nothing to do */
+ } else {
+ netif_err(efx, drv, net_dev,
+ "Bad value for module parameter rss_cpus='%s'\n",
+ rss_cpus);
+ /* default to 'cores' */
+ goto cores;
+ }
- free_cpumask_var(thread_mask);
+ if (n_rxq > EFX_MAX_RX_QUEUES) {
+ netif_warn(efx, drv, net_dev,
+ "Reducing number of rss channels from %u to %u.\n",
+ n_rxq, EFX_MAX_RX_QUEUES);
+ n_rxq = EFX_MAX_RX_QUEUES;
}
+#ifdef CONFIG_SFC_SRIOV
/* If RSS is requested for the PF *and* VFs then we can't write RSS
* table entries that are inaccessible to VFs
*/
-#ifdef CONFIG_SFC_SRIOV
if (efx->type->sriov_wanted) {
if (efx->type->sriov_wanted(efx) && efx_vf_size(efx) > 1 &&
- count > efx_vf_size(efx)) {
+ n_rxq > efx_vf_size(efx)) {
netif_warn(efx, probe, efx->net_dev,
"Reducing number of RSS channels from %u to %u for "
"VF support. Increase vf-msix-limit to use more "
"channels on the PF.\n",
- count, efx_vf_size(efx));
- count = efx_vf_size(efx);
+ n_rxq, efx_vf_size(efx));
+ n_rxq = efx_vf_size(efx);
}
}
#endif
- return count;
+ return n_rxq;
}
/* Probe the number and type of interrupts we are able to obtain, and
^ permalink raw reply related
* [PATCH v9 net-next 2/2] net: add the AF_HYPERV entries to family name tables
From: Dexuan Cui @ 2016-05-04 16:57 UTC (permalink / raw)
To: gregkh, davem, netdev, linux-kernel, devel, olaf, apw, jasowang,
cavery, kys, haiyangz
Cc: joe
This is for the hv_sock driver, which introduces AF_HYPERV(42).
Signed-off-by: Dexuan Cui <decui@microsoft.com>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Cathy Avery <cavery@redhat.com>
---
net/core/sock.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/net/core/sock.c b/net/core/sock.c
index e16a5db..c0884c7 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -222,7 +222,7 @@ static const char *const af_family_key_strings[AF_MAX+1] = {
"sk_lock-AF_RXRPC" , "sk_lock-AF_ISDN" , "sk_lock-AF_PHONET" ,
"sk_lock-AF_IEEE802154", "sk_lock-AF_CAIF" , "sk_lock-AF_ALG" ,
"sk_lock-AF_NFC" , "sk_lock-AF_VSOCK" , "sk_lock-AF_KCM" ,
- "sk_lock-AF_MAX"
+ "sk_lock-AF_HYPERV", "sk_lock-AF_MAX"
};
static const char *const af_family_slock_key_strings[AF_MAX+1] = {
"slock-AF_UNSPEC", "slock-AF_UNIX" , "slock-AF_INET" ,
@@ -239,7 +239,7 @@ static const char *const af_family_slock_key_strings[AF_MAX+1] = {
"slock-AF_RXRPC" , "slock-AF_ISDN" , "slock-AF_PHONET" ,
"slock-AF_IEEE802154", "slock-AF_CAIF" , "slock-AF_ALG" ,
"slock-AF_NFC" , "slock-AF_VSOCK" ,"slock-AF_KCM" ,
- "slock-AF_MAX"
+ "slock-AF_HYPERV", "slock-AF_MAX"
};
static const char *const af_family_clock_key_strings[AF_MAX+1] = {
"clock-AF_UNSPEC", "clock-AF_UNIX" , "clock-AF_INET" ,
@@ -256,7 +256,7 @@ static const char *const af_family_clock_key_strings[AF_MAX+1] = {
"clock-AF_RXRPC" , "clock-AF_ISDN" , "clock-AF_PHONET" ,
"clock-AF_IEEE802154", "clock-AF_CAIF" , "clock-AF_ALG" ,
"clock-AF_NFC" , "clock-AF_VSOCK" , "clock-AF_KCM" ,
- "clock-AF_MAX"
+ "clock-AF_HYPERV", "clock-AF_MAX"
};
/*
--
2.1.0
^ permalink raw reply related
* [PATCH v9 net-next 1/2] hv_sock: introduce Hyper-V Sockets
From: Dexuan Cui @ 2016-05-04 16:56 UTC (permalink / raw)
To: gregkh, davem, netdev, linux-kernel, devel, olaf, apw, jasowang,
cavery, kys, haiyangz
Cc: joe, vkuznets
Hyper-V Sockets (hv_sock) supplies a byte-stream based communication
mechanism between the host and the guest. It's somewhat like TCP over
VMBus, but the transportation layer (VMBus) is much simpler than IP.
With Hyper-V Sockets, applications between the host and the guest can talk
to each other directly by the traditional BSD-style socket APIs.
Hyper-V Sockets is only available on new Windows hosts, like Windows Server
2016. More info is in this article "Make your own integration services":
https://msdn.microsoft.com/en-us/virtualization/hyperv_on_windows/develop/make_mgmt_service
The patch implements the necessary support in the guest side by introducing
a new socket address family AF_HYPERV.
Signed-off-by: Dexuan Cui <decui@microsoft.com>
Cc: "K. Y. Srinivasan" <kys@microsoft.com>
Cc: Haiyang Zhang <haiyangz@microsoft.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Cathy Avery <cavery@redhat.com>
---
MAINTAINERS | 2 +
include/linux/hyperv.h | 16 +
include/linux/socket.h | 5 +-
include/net/af_hvsock.h | 55 ++
include/uapi/linux/hyperv.h | 25 +
net/Kconfig | 1 +
net/Makefile | 1 +
net/hv_sock/Kconfig | 10 +
net/hv_sock/Makefile | 3 +
net/hv_sock/af_hvsock.c | 1434 +++++++++++++++++++++++++++++++++++++++++++
10 files changed, 1550 insertions(+), 2 deletions(-)
diff --git a/MAINTAINERS b/MAINTAINERS
index fa02825..b32716f 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -5268,7 +5268,9 @@ F: drivers/pci/host/pci-hyperv.c
F: drivers/net/hyperv/
F: drivers/scsi/storvsc_drv.c
F: drivers/video/fbdev/hyperv_fb.c
+F: net/hv_sock/
F: include/linux/hyperv.h
+F: include/net/af_hvsock.h
F: tools/hv/
F: Documentation/ABI/stable/sysfs-bus-vmbus
diff --git a/include/linux/hyperv.h b/include/linux/hyperv.h
index aa0fadc..e756719 100644
--- a/include/linux/hyperv.h
+++ b/include/linux/hyperv.h
@@ -1338,4 +1338,20 @@ extern __u32 vmbus_proto_version;
int vmbus_send_tl_connect_request(const uuid_le *shv_guest_servie_id,
const uuid_le *shv_host_servie_id);
+struct vmpipe_proto_header {
+ u32 pkt_type;
+ u32 data_size;
+};
+
+#define HVSOCK_HEADER_LEN (sizeof(struct vmpacket_descriptor) + \
+ sizeof(struct vmpipe_proto_header))
+
+/* See 'prev_indices' in hv_ringbuffer_read(), hv_ringbuffer_write() */
+#define PREV_INDICES_LEN (sizeof(u64))
+
+#define HVSOCK_PKT_LEN(payload_len) (HVSOCK_HEADER_LEN + \
+ ALIGN((payload_len), 8) + \
+ PREV_INDICES_LEN)
+#define HVSOCK_MIN_PKT_LEN HVSOCK_PKT_LEN(1)
+
#endif /* _HYPERV_H */
diff --git a/include/linux/socket.h b/include/linux/socket.h
index 73bf6c6..88b1ccd 100644
--- a/include/linux/socket.h
+++ b/include/linux/socket.h
@@ -201,8 +201,8 @@ struct ucred {
#define AF_NFC 39 /* NFC sockets */
#define AF_VSOCK 40 /* vSockets */
#define AF_KCM 41 /* Kernel Connection Multiplexor*/
-
-#define AF_MAX 42 /* For now.. */
+#define AF_HYPERV 42 /* Hyper-V Sockets */
+#define AF_MAX 43 /* For now.. */
/* Protocol families, same as address families. */
#define PF_UNSPEC AF_UNSPEC
@@ -249,6 +249,7 @@ struct ucred {
#define PF_NFC AF_NFC
#define PF_VSOCK AF_VSOCK
#define PF_KCM AF_KCM
+#define PF_HYPERV AF_HYPERV
#define PF_MAX AF_MAX
/* Maximum queue length specifiable by listen. */
diff --git a/include/net/af_hvsock.h b/include/net/af_hvsock.h
new file mode 100644
index 0000000..04bc40c
--- /dev/null
+++ b/include/net/af_hvsock.h
@@ -0,0 +1,55 @@
+#ifndef __AF_HVSOCK_H__
+#define __AF_HVSOCK_H__
+
+#include <linux/kernel.h>
+#include <linux/hyperv.h>
+#include <net/sock.h>
+
+#define VMBUS_RINGBUFFER_SIZE_HVSOCK_RECV (5 * PAGE_SIZE)
+#define VMBUS_RINGBUFFER_SIZE_HVSOCK_SEND (5 * PAGE_SIZE)
+
+#define HVSOCK_RCV_BUF_SZ VMBUS_RINGBUFFER_SIZE_HVSOCK_RECV
+#define HVSOCK_SND_BUF_SZ PAGE_SIZE
+
+#define sk_to_hvsock(__sk) ((struct hvsock_sock *)(__sk))
+#define hvsock_to_sk(__hvsk) ((struct sock *)(__hvsk))
+
+struct hvsock_sock {
+ /* sk must be the first member. */
+ struct sock sk;
+
+ struct sockaddr_hv local_addr;
+ struct sockaddr_hv remote_addr;
+
+ /* protected by the global hvsock_mutex */
+ struct list_head bound_list;
+ struct list_head connected_list;
+
+ struct list_head accept_queue;
+ /* used by enqueue and dequeue */
+ struct mutex accept_queue_mutex;
+
+ struct delayed_work dwork;
+
+ u32 peer_shutdown;
+
+ struct vmbus_channel *channel;
+
+ /* The 'hdr' and 'buf' in the below 'send' and 'recv' definitions must
+ * be consecutive: see hvsock_send_data() and hvsock_recv_data().
+ */
+ struct {
+ struct vmpipe_proto_header hdr;
+ u8 buf[HVSOCK_SND_BUF_SZ];
+ } send;
+
+ struct {
+ struct vmpipe_proto_header hdr;
+ u8 buf[HVSOCK_RCV_BUF_SZ];
+
+ unsigned int data_len;
+ unsigned int data_offset;
+ } recv;
+};
+
+#endif /* __AF_HVSOCK_H__ */
diff --git a/include/uapi/linux/hyperv.h b/include/uapi/linux/hyperv.h
index e347b24..408b832 100644
--- a/include/uapi/linux/hyperv.h
+++ b/include/uapi/linux/hyperv.h
@@ -26,6 +26,7 @@
#define _UAPI_HYPERV_H
#include <linux/uuid.h>
+#include <linux/socket.h>
/*
* Framework version for util services.
@@ -396,4 +397,28 @@ struct hv_kvp_ip_msg {
struct hv_kvp_ipaddr_value kvp_ip_val;
} __attribute__((packed));
+/*
+ * This is the address fromat of Hyper-V Sockets.
+ * Note: here we just borrow the kernel's built-in type uuid_le. When
+ * an application calls bind() or connect(), the 2 members of struct
+ * sockaddr_hv must be of GUID.
+ * The GUID format differs from the UUID format only in the byte order of
+ * the first 3 fields. Refer to:
+ * https://en.wikipedia.org/wiki/Globally_unique_identifier
+ */
+#define guid_t uuid_le
+struct sockaddr_hv {
+ __kernel_sa_family_t shv_family; /* Address family */
+ __le16 reserved; /* Must be Zero */
+ guid_t shv_vm_id; /* VM ID */
+ guid_t shv_service_id; /* Service ID */
+};
+
+#define SHV_VMID_GUEST NULL_UUID_LE
+#define SHV_VMID_HOST NULL_UUID_LE
+
+#define SHV_SERVICE_ID_ANY NULL_UUID_LE
+
+#define SHV_PROTO_RAW 1
+
#endif /* _UAPI_HYPERV_H */
diff --git a/net/Kconfig b/net/Kconfig
index a8934d8..68d13d7 100644
--- a/net/Kconfig
+++ b/net/Kconfig
@@ -231,6 +231,7 @@ source "net/dns_resolver/Kconfig"
source "net/batman-adv/Kconfig"
source "net/openvswitch/Kconfig"
source "net/vmw_vsock/Kconfig"
+source "net/hv_sock/Kconfig"
source "net/netlink/Kconfig"
source "net/mpls/Kconfig"
source "net/hsr/Kconfig"
diff --git a/net/Makefile b/net/Makefile
index 81d1411..d115c31 100644
--- a/net/Makefile
+++ b/net/Makefile
@@ -70,6 +70,7 @@ obj-$(CONFIG_BATMAN_ADV) += batman-adv/
obj-$(CONFIG_NFC) += nfc/
obj-$(CONFIG_OPENVSWITCH) += openvswitch/
obj-$(CONFIG_VSOCKETS) += vmw_vsock/
+obj-$(CONFIG_HYPERV_SOCK) += hv_sock/
obj-$(CONFIG_MPLS) += mpls/
obj-$(CONFIG_HSR) += hsr/
ifneq ($(CONFIG_NET_SWITCHDEV),)
diff --git a/net/hv_sock/Kconfig b/net/hv_sock/Kconfig
new file mode 100644
index 0000000..1f41848
--- /dev/null
+++ b/net/hv_sock/Kconfig
@@ -0,0 +1,10 @@
+config HYPERV_SOCK
+ tristate "Hyper-V Sockets"
+ depends on HYPERV
+ default m if HYPERV
+ help
+ Hyper-V Sockets is somewhat like TCP over VMBus, allowing
+ communication between Linux guest and Hyper-V host without TCP/IP.
+
+ To compile this driver as a module, choose M here: the module
+ will be called hv_sock.
diff --git a/net/hv_sock/Makefile b/net/hv_sock/Makefile
new file mode 100644
index 0000000..716c012
--- /dev/null
+++ b/net/hv_sock/Makefile
@@ -0,0 +1,3 @@
+obj-$(CONFIG_HYPERV_SOCK) += hv_sock.o
+
+hv_sock-y += af_hvsock.o
diff --git a/net/hv_sock/af_hvsock.c b/net/hv_sock/af_hvsock.c
new file mode 100644
index 0000000..114a9e4
--- /dev/null
+++ b/net/hv_sock/af_hvsock.c
@@ -0,0 +1,1434 @@
+/*
+ * Hyper-V Sockets -- a socket-based communication channel between the
+ * Hyper-V host and the virtual machines running on it.
+ *
+ * Copyright(c) 2016, Microsoft Corporation. All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ * 3. The name of the author may not be used to endorse or promote
+ * products derived from this software without specific prior written
+ * permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
+ * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT,
+ * INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+ * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
+ * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/init.h>
+#include <linux/module.h>
+#include <net/af_hvsock.h>
+
+static struct proto hvsock_proto = {
+ .name = "HV_SOCK",
+ .owner = THIS_MODULE,
+ .obj_size = sizeof(struct hvsock_sock),
+};
+
+#define SS_LISTEN 255
+
+static LIST_HEAD(hvsock_bound_list);
+static LIST_HEAD(hvsock_connected_list);
+static DEFINE_MUTEX(hvsock_mutex);
+
+static bool uuid_equals(uuid_le u1, uuid_le u2)
+{
+ return !uuid_le_cmp(u1, u2);
+}
+
+static struct sock *hvsock_find_bound_socket(const struct sockaddr_hv *addr)
+{
+ struct hvsock_sock *hvsk;
+
+ list_for_each_entry(hvsk, &hvsock_bound_list, bound_list) {
+ if (uuid_equals(addr->shv_service_id,
+ hvsk->local_addr.shv_service_id))
+ return hvsock_to_sk(hvsk);
+ }
+ return NULL;
+}
+
+static struct sock *hvsock_find_connected_socket_by_channel(
+ const struct vmbus_channel *channel)
+{
+ struct hvsock_sock *hvsk;
+
+ list_for_each_entry(hvsk, &hvsock_connected_list, connected_list) {
+ if (hvsk->channel == channel)
+ return hvsock_to_sk(hvsk);
+ }
+ return NULL;
+}
+
+static
+void hvsock_enqueue_accept(struct sock *listener, struct sock *connected)
+{
+ struct hvsock_sock *hvlistener;
+ struct hvsock_sock *hvconnected;
+
+ hvlistener = sk_to_hvsock(listener);
+ hvconnected = sk_to_hvsock(connected);
+
+ sock_hold(connected);
+ sock_hold(listener);
+
+ mutex_lock(&hvlistener->accept_queue_mutex);
+ list_add_tail(&hvconnected->accept_queue, &hvlistener->accept_queue);
+ listener->sk_ack_backlog++;
+ mutex_unlock(&hvlistener->accept_queue_mutex);
+}
+
+static struct sock *hvsock_dequeue_accept(struct sock *listener)
+{
+ struct hvsock_sock *hvlistener;
+ struct hvsock_sock *hvconnected;
+
+ hvlistener = sk_to_hvsock(listener);
+
+ mutex_lock(&hvlistener->accept_queue_mutex);
+
+ if (list_empty(&hvlistener->accept_queue)) {
+ mutex_unlock(&hvlistener->accept_queue_mutex);
+ return NULL;
+ }
+
+ hvconnected = list_entry(hvlistener->accept_queue.next,
+ struct hvsock_sock, accept_queue);
+
+ list_del_init(&hvconnected->accept_queue);
+ listener->sk_ack_backlog--;
+
+ mutex_unlock(&hvlistener->accept_queue_mutex);
+
+ sock_put(listener);
+ /* The caller will need a reference on the connected socket so we let
+ * it call sock_put().
+ */
+
+ return hvsock_to_sk(hvconnected);
+}
+
+static bool hvsock_is_accept_queue_empty(struct sock *sk)
+{
+ struct hvsock_sock *hvsk = sk_to_hvsock(sk);
+ int ret;
+
+ mutex_lock(&hvsk->accept_queue_mutex);
+ ret = list_empty(&hvsk->accept_queue);
+ mutex_unlock(&hvsk->accept_queue_mutex);
+
+ return ret;
+}
+
+static void hvsock_addr_init(struct sockaddr_hv *addr, uuid_le service_id)
+{
+ memset(addr, 0, sizeof(*addr));
+ addr->shv_family = AF_HYPERV;
+ addr->shv_service_id = service_id;
+}
+
+static int hvsock_addr_validate(const struct sockaddr_hv *addr)
+{
+ if (!addr)
+ return -EFAULT;
+
+ if (addr->shv_family != AF_HYPERV)
+ return -EAFNOSUPPORT;
+
+ if (addr->reserved != 0)
+ return -EINVAL;
+
+ return 0;
+}
+
+static bool hvsock_addr_bound(const struct sockaddr_hv *addr)
+{
+ return !uuid_equals(addr->shv_service_id, SHV_SERVICE_ID_ANY);
+}
+
+static int hvsock_addr_cast(const struct sockaddr *addr, size_t len,
+ struct sockaddr_hv **out_addr)
+{
+ if (len < sizeof(**out_addr))
+ return -EFAULT;
+
+ *out_addr = (struct sockaddr_hv *)addr;
+ return hvsock_addr_validate(*out_addr);
+}
+
+static int __hvsock_do_bind(struct hvsock_sock *hvsk,
+ struct sockaddr_hv *addr)
+{
+ struct sockaddr_hv hv_addr;
+ int ret = 0;
+
+ hvsock_addr_init(&hv_addr, addr->shv_service_id);
+
+ mutex_lock(&hvsock_mutex);
+
+ if (uuid_equals(addr->shv_service_id, SHV_SERVICE_ID_ANY)) {
+ do {
+ uuid_le_gen(&hv_addr.shv_service_id);
+ } while (hvsock_find_bound_socket(&hv_addr));
+ } else {
+ if (hvsock_find_bound_socket(&hv_addr)) {
+ ret = -EADDRINUSE;
+ goto out;
+ }
+ }
+
+ hvsock_addr_init(&hvsk->local_addr, hv_addr.shv_service_id);
+
+ sock_hold(&hvsk->sk);
+ list_add(&hvsk->bound_list, &hvsock_bound_list);
+out:
+ mutex_unlock(&hvsock_mutex);
+
+ return ret;
+}
+
+static int __hvsock_bind(struct sock *sk, struct sockaddr_hv *addr)
+{
+ struct hvsock_sock *hvsk = sk_to_hvsock(sk);
+ int ret;
+
+ if (hvsock_addr_bound(&hvsk->local_addr))
+ return -EINVAL;
+
+ switch (sk->sk_socket->type) {
+ case SOCK_STREAM:
+ ret = __hvsock_do_bind(hvsk, addr);
+ break;
+
+ default:
+ ret = -EINVAL;
+ break;
+ }
+
+ return ret;
+}
+
+/* Autobind this socket to the local address if necessary. */
+static int hvsock_auto_bind(struct hvsock_sock *hvsk)
+{
+ struct sock *sk = hvsock_to_sk(hvsk);
+ struct sockaddr_hv local_addr;
+
+ if (hvsock_addr_bound(&hvsk->local_addr))
+ return 0;
+ hvsock_addr_init(&local_addr, SHV_SERVICE_ID_ANY);
+ return __hvsock_bind(sk, &local_addr);
+}
+
+static void hvsock_sk_destruct(struct sock *sk)
+{
+ struct hvsock_sock *hvsk = sk_to_hvsock(sk);
+ struct vmbus_channel *channel = hvsk->channel;
+
+ if (!channel)
+ return;
+
+ vmbus_hvsock_device_unregister(channel);
+}
+
+static void __hvsock_release(struct sock *sk)
+{
+ struct hvsock_sock *hvsk;
+ struct sock *pending;
+
+ hvsk = sk_to_hvsock(sk);
+
+ mutex_lock(&hvsock_mutex);
+
+ if (!list_empty(&hvsk->bound_list)) {
+ list_del_init(&hvsk->bound_list);
+ sock_put(&hvsk->sk);
+ }
+
+ if (!list_empty(&hvsk->connected_list)) {
+ list_del_init(&hvsk->connected_list);
+ sock_put(&hvsk->sk);
+ }
+
+ mutex_unlock(&hvsock_mutex);
+
+ lock_sock(sk);
+ sock_orphan(sk);
+ sk->sk_shutdown = SHUTDOWN_MASK;
+
+ /* Clean up any sockets that never were accepted. */
+ while ((pending = hvsock_dequeue_accept(sk)) != NULL) {
+ __hvsock_release(pending);
+ sock_put(pending);
+ }
+
+ release_sock(sk);
+ sock_put(sk);
+}
+
+static int hvsock_release(struct socket *sock)
+{
+ /* If accept() is interrupted by a signal, the temporary socket
+ * struct's sock->sk is NULL.
+ */
+ if (sock->sk) {
+ __hvsock_release(sock->sk);
+ sock->sk = NULL;
+ }
+
+ sock->state = SS_FREE;
+ return 0;
+}
+
+static struct sock *__hvsock_create(struct net *net, struct socket *sock,
+ gfp_t priority, unsigned short type)
+{
+ struct hvsock_sock *hvsk;
+ struct sock *sk;
+
+ sk = sk_alloc(net, AF_HYPERV, priority, &hvsock_proto, 0);
+ if (!sk)
+ return NULL;
+
+ sock_init_data(sock, sk);
+
+ /* sk->sk_type is normally set in sock_init_data, but only if sock is
+ * non-NULL. We make sure that our sockets always have a type by
+ * setting it here if needed.
+ */
+ if (!sock)
+ sk->sk_type = type;
+
+ hvsk = sk_to_hvsock(sk);
+ hvsock_addr_init(&hvsk->local_addr, SHV_SERVICE_ID_ANY);
+ hvsock_addr_init(&hvsk->remote_addr, SHV_SERVICE_ID_ANY);
+
+ sk->sk_destruct = hvsock_sk_destruct;
+
+ /* Looks stream-based socket doesn't need this. */
+ sk->sk_backlog_rcv = NULL;
+
+ sk->sk_state = 0;
+ sock_reset_flag(sk, SOCK_DONE);
+
+ INIT_LIST_HEAD(&hvsk->bound_list);
+ INIT_LIST_HEAD(&hvsk->connected_list);
+
+ INIT_LIST_HEAD(&hvsk->accept_queue);
+ mutex_init(&hvsk->accept_queue_mutex);
+
+ hvsk->peer_shutdown = 0;
+
+ hvsk->recv.data_len = 0;
+ hvsk->recv.data_offset = 0;
+
+ return sk;
+}
+
+static int hvsock_bind(struct socket *sock, struct sockaddr *addr,
+ int addr_len)
+{
+ struct sockaddr_hv *hv_addr;
+ struct sock *sk;
+ int ret;
+
+ sk = sock->sk;
+
+ if (hvsock_addr_cast(addr, addr_len, &hv_addr) != 0)
+ return -EINVAL;
+
+ if (!uuid_equals(hv_addr->shv_vm_id, NULL_UUID_LE))
+ return -EINVAL;
+
+ lock_sock(sk);
+ ret = __hvsock_bind(sk, hv_addr);
+ release_sock(sk);
+
+ return ret;
+}
+
+static int hvsock_getname(struct socket *sock,
+ struct sockaddr *addr, int *addr_len, int peer)
+{
+ struct sockaddr_hv *hv_addr;
+ struct hvsock_sock *hvsk;
+ struct sock *sk;
+ int ret;
+
+ sk = sock->sk;
+ hvsk = sk_to_hvsock(sk);
+ ret = 0;
+
+ lock_sock(sk);
+
+ if (peer) {
+ if (sock->state != SS_CONNECTED) {
+ ret = -ENOTCONN;
+ goto out;
+ }
+ hv_addr = &hvsk->remote_addr;
+ } else {
+ hv_addr = &hvsk->local_addr;
+ }
+
+ __sockaddr_check_size(sizeof(*hv_addr));
+
+ memcpy(addr, hv_addr, sizeof(*hv_addr));
+ *addr_len = sizeof(*hv_addr);
+
+out:
+ release_sock(sk);
+ return ret;
+}
+
+static int hvsock_shutdown(struct socket *sock, int mode)
+{
+ struct sock *sk;
+
+ if (mode < SHUT_RD || mode > SHUT_RDWR)
+ return -EINVAL;
+ /* This maps:
+ * SHUT_RD (0) -> RCV_SHUTDOWN (1)
+ * SHUT_WR (1) -> SEND_SHUTDOWN (2)
+ * SHUT_RDWR (2) -> SHUTDOWN_MASK (3)
+ */
+ ++mode;
+
+ if (sock->state == SS_UNCONNECTED)
+ return -ENOTCONN;
+
+ sock->state = SS_DISCONNECTING;
+
+ sk = sock->sk;
+
+ lock_sock(sk);
+
+ sk->sk_shutdown |= mode;
+ sk->sk_state_change(sk);
+
+ /* TODO: how to send a FIN if we haven't done that? */
+ if (mode & SEND_SHUTDOWN)
+ ;
+
+ release_sock(sk);
+
+ return 0;
+}
+
+static void get_ringbuffer_rw_status(struct vmbus_channel *channel,
+ bool *can_read, bool *can_write)
+{
+ u32 avl_read_bytes, avl_write_bytes, dummy;
+
+ if (can_read) {
+ hv_get_ringbuffer_availbytes(&channel->inbound,
+ &avl_read_bytes,
+ &dummy);
+ *can_read = avl_read_bytes >= HVSOCK_MIN_PKT_LEN;
+ }
+
+ /* We write into the ringbuffer only when we're able to write a
+ * a payload of 4096 bytes (the actual written payload's length may be
+ * less than 4096).
+ */
+ if (can_write) {
+ hv_get_ringbuffer_availbytes(&channel->outbound,
+ &dummy,
+ &avl_write_bytes);
+ *can_write = avl_write_bytes > HVSOCK_PKT_LEN(PAGE_SIZE);
+ }
+}
+
+static unsigned int hvsock_poll(struct file *file, struct socket *sock,
+ poll_table *wait)
+{
+ struct vmbus_channel *channel;
+ bool can_read, can_write;
+ struct hvsock_sock *hvsk;
+ struct sock *sk;
+ unsigned int mask;
+
+ sk = sock->sk;
+ hvsk = sk_to_hvsock(sk);
+
+ poll_wait(file, sk_sleep(sk), wait);
+ mask = 0;
+
+ if (sk->sk_err)
+ /* Signify that there has been an error on this socket. */
+ mask |= POLLERR;
+
+ /* INET sockets treat local write shutdown and peer write shutdown as a
+ * case of POLLHUP set.
+ */
+ if ((sk->sk_shutdown == SHUTDOWN_MASK) ||
+ ((sk->sk_shutdown & SEND_SHUTDOWN) &&
+ (hvsk->peer_shutdown & SEND_SHUTDOWN))) {
+ mask |= POLLHUP;
+ }
+
+ if (sk->sk_shutdown & RCV_SHUTDOWN ||
+ hvsk->peer_shutdown & SEND_SHUTDOWN) {
+ mask |= POLLRDHUP;
+ }
+
+ lock_sock(sk);
+
+ /* Listening sockets that have connections in their accept
+ * queue can be read.
+ */
+ if (sk->sk_state == SS_LISTEN && !hvsock_is_accept_queue_empty(sk))
+ mask |= POLLIN | POLLRDNORM;
+
+ /* The mutex is to against hvsock_open_connection() */
+ mutex_lock(&hvsock_mutex);
+
+ channel = hvsk->channel;
+ if (channel) {
+ /* If there is something in the queue then we can read */
+ get_ringbuffer_rw_status(channel, &can_read, &can_write);
+
+ if (!can_read && hvsk->recv.data_len > 0)
+ can_read = true;
+
+ if (!(sk->sk_shutdown & RCV_SHUTDOWN) && can_read)
+ mask |= POLLIN | POLLRDNORM;
+ } else {
+ can_read = false;
+ can_write = false;
+ }
+
+ mutex_unlock(&hvsock_mutex);
+
+ /* Sockets whose connections have been closed terminated should
+ * also be considered read, and we check the shutdown flag for that.
+ */
+ if (sk->sk_shutdown & RCV_SHUTDOWN ||
+ hvsk->peer_shutdown & SEND_SHUTDOWN) {
+ mask |= POLLIN | POLLRDNORM;
+ }
+
+ /* Connected sockets that can produce data can be written. */
+ if (sk->sk_state == SS_CONNECTED && can_write &&
+ !(sk->sk_shutdown & SEND_SHUTDOWN)) {
+ /* Remove POLLWRBAND since INET sockets are not setting it.
+ */
+ mask |= POLLOUT | POLLWRNORM;
+ }
+
+ /* Simulate INET socket poll behaviors, which sets
+ * POLLOUT|POLLWRNORM when peer is closed and nothing to read,
+ * but local send is not shutdown.
+ */
+ if (sk->sk_state == SS_UNCONNECTED &&
+ !(sk->sk_shutdown & SEND_SHUTDOWN))
+ mask |= POLLOUT | POLLWRNORM;
+
+ release_sock(sk);
+
+ return mask;
+}
+
+/* This function runs in the tasklet context of process_chn_event() */
+static void hvsock_on_channel_cb(void *ctx)
+{
+ struct sock *sk = (struct sock *)ctx;
+ struct hvsock_sock *hvsk = sk_to_hvsock(sk);
+ struct vmbus_channel *channel = hvsk->channel;
+ bool can_read, can_write;
+
+ if (!channel) {
+ WARN_ONCE(1, "NULL channel! There is a programming bug.\n");
+ return;
+ }
+
+ get_ringbuffer_rw_status(channel, &can_read, &can_write);
+
+ if (can_read)
+ sk->sk_data_ready(sk);
+
+ if (can_write)
+ sk->sk_write_space(sk);
+}
+
+static void hvsock_close_connection(struct vmbus_channel *channel)
+{
+ struct hvsock_sock *hvsk;
+ struct sock *sk;
+
+ mutex_lock(&hvsock_mutex);
+
+ sk = hvsock_find_connected_socket_by_channel(channel);
+
+ /* The guest has already closed the connection? */
+ if (!sk)
+ goto out;
+
+ sk->sk_socket->state = SS_UNCONNECTED;
+ sk->sk_state = SS_UNCONNECTED;
+ sock_set_flag(sk, SOCK_DONE);
+
+ hvsk = sk_to_hvsock(sk);
+ hvsk->peer_shutdown |= SEND_SHUTDOWN | RCV_SHUTDOWN;
+
+ sk->sk_state_change(sk);
+out:
+ mutex_unlock(&hvsock_mutex);
+}
+
+static int hvsock_open_connection(struct vmbus_channel *channel)
+{
+ struct hvsock_sock *hvsk, *new_hvsk;
+ struct sockaddr_hv hv_addr;
+ struct sock *sk, *new_sk;
+ unsigned char conn_from_host;
+
+ uuid_le *instance, *service_id;
+ int ret;
+
+ instance = &channel->offermsg.offer.if_instance;
+ service_id = &channel->offermsg.offer.if_type;
+
+ /* The first byte != 0 means the host initiated the connection. */
+ conn_from_host = channel->offermsg.offer.u.pipe.user_def[0];
+
+ mutex_lock(&hvsock_mutex);
+
+ hvsock_addr_init(&hv_addr, conn_from_host ? *service_id : *instance);
+ sk = hvsock_find_bound_socket(&hv_addr);
+
+ if (!sk || (conn_from_host && sk->sk_state != SS_LISTEN) ||
+ (!conn_from_host && sk->sk_state != SS_CONNECTING)) {
+ ret = -ENXIO;
+ goto out;
+ }
+
+ if (conn_from_host) {
+ if (sk->sk_ack_backlog >= sk->sk_max_ack_backlog) {
+ ret = -EMFILE;
+ goto out;
+ }
+
+ new_sk = __hvsock_create(sock_net(sk), NULL, GFP_KERNEL,
+ sk->sk_type);
+ if (!new_sk) {
+ ret = -ENOMEM;
+ goto out;
+ }
+ new_sk->sk_state = SS_CONNECTING;
+ new_hvsk = sk_to_hvsock(new_sk);
+ new_hvsk->channel = channel;
+ hvsock_addr_init(&new_hvsk->local_addr, *service_id);
+ hvsock_addr_init(&new_hvsk->remote_addr, *instance);
+ } else {
+ hvsk = sk_to_hvsock(sk);
+ hvsk->channel = channel;
+ }
+
+ set_channel_read_state(channel, false);
+ vmbus_set_chn_rescind_callback(channel, hvsock_close_connection);
+ ret = vmbus_open(channel, VMBUS_RINGBUFFER_SIZE_HVSOCK_SEND,
+ VMBUS_RINGBUFFER_SIZE_HVSOCK_RECV, NULL, 0,
+ hvsock_on_channel_cb, conn_from_host ? new_sk : sk);
+ if (ret != 0) {
+ if (conn_from_host) {
+ new_hvsk->channel = NULL;
+ sock_put(new_sk);
+ } else {
+ hvsk->channel = NULL;
+ }
+ goto out;
+ }
+
+ set_channel_pending_send_size(channel, HVSOCK_PKT_LEN(PAGE_SIZE));
+
+ if (conn_from_host) {
+ new_sk->sk_state = SS_CONNECTED;
+
+ sock_hold(&new_hvsk->sk);
+ list_add(&new_hvsk->connected_list, &hvsock_connected_list);
+
+ hvsock_enqueue_accept(sk, new_sk);
+ } else {
+ sk->sk_state = SS_CONNECTED;
+ sk->sk_socket->state = SS_CONNECTED;
+
+ sock_hold(&hvsk->sk);
+ list_add(&hvsk->connected_list, &hvsock_connected_list);
+ }
+
+ sk->sk_state_change(sk);
+out:
+ mutex_unlock(&hvsock_mutex);
+ return ret;
+}
+
+static void hvsock_connect_timeout(struct work_struct *work)
+{
+ struct hvsock_sock *hvsk;
+ struct sock *sk;
+
+ hvsk = container_of(work, struct hvsock_sock, dwork.work);
+ sk = hvsock_to_sk(hvsk);
+
+ lock_sock(sk);
+ if ((sk->sk_state == SS_CONNECTING) &&
+ (sk->sk_shutdown != SHUTDOWN_MASK)) {
+ sk->sk_state = SS_UNCONNECTED;
+ sk->sk_err = ETIMEDOUT;
+ sk->sk_error_report(sk);
+ }
+ release_sock(sk);
+
+ sock_put(sk);
+}
+
+static int hvsock_connect_wait(struct socket *sock,
+ int flags, int current_ret)
+{
+ struct sock *sk = sock->sk;
+ struct hvsock_sock *hvsk = sk_to_hvsock(sk);
+
+ int ret = current_ret;
+
+ long timeout = 30 * HZ;
+ DEFINE_WAIT(wait);
+
+ prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
+
+ while (sk->sk_state != SS_CONNECTED && sk->sk_err == 0) {
+ if (flags & O_NONBLOCK) {
+ /* If we're not going to block, we schedule a timeout
+ * function to generate a timeout on the connection
+ * attempt, in case the peer doesn't respond in a
+ * timely manner. We hold on to the socket until the
+ * timeout fires.
+ */
+ sock_hold(sk);
+ INIT_DELAYED_WORK(&hvsk->dwork,
+ hvsock_connect_timeout);
+ schedule_delayed_work(&hvsk->dwork, timeout);
+
+ /* Skip ahead to preserve error code set above. */
+ goto out_wait;
+ }
+
+ release_sock(sk);
+ timeout = schedule_timeout(timeout);
+ lock_sock(sk);
+
+ if (signal_pending(current)) {
+ ret = sock_intr_errno(timeout);
+ goto out_wait_error;
+ } else if (timeout == 0) {
+ ret = -ETIMEDOUT;
+ goto out_wait_error;
+ }
+
+ prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
+ }
+
+ ret = sk->sk_err ? -sk->sk_err : 0;
+
+out_wait_error:
+ if (ret < 0) {
+ sk->sk_state = SS_UNCONNECTED;
+ sock->state = SS_UNCONNECTED;
+ }
+out_wait:
+ finish_wait(sk_sleep(sk), &wait);
+ return ret;
+}
+
+static int hvsock_connect(struct socket *sock, struct sockaddr *addr,
+ int addr_len, int flags)
+{
+ struct sockaddr_hv *remote_addr;
+ struct hvsock_sock *hvsk;
+ struct sock *sk;
+ int ret = 0;
+
+ sk = sock->sk;
+ hvsk = sk_to_hvsock(sk);
+
+ lock_sock(sk);
+
+ switch (sock->state) {
+ case SS_CONNECTED:
+ ret = -EISCONN;
+ goto out;
+ case SS_DISCONNECTING:
+ ret = -EINVAL;
+ goto out;
+ case SS_CONNECTING:
+ /* This continues on so we can move sock into the SS_CONNECTED
+ * state once the connection has completed (at which point err
+ * will be set to zero also). Otherwise, we will either wait
+ * for the connection or return -EALREADY should this be a
+ * non-blocking call.
+ */
+ ret = -EALREADY;
+ break;
+ default:
+ if ((sk->sk_state == SS_LISTEN) ||
+ hvsock_addr_cast(addr, addr_len, &remote_addr) != 0) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ /* Set the remote address that we are connecting to. */
+ memcpy(&hvsk->remote_addr, remote_addr,
+ sizeof(hvsk->remote_addr));
+
+ ret = hvsock_auto_bind(hvsk);
+ if (ret)
+ goto out;
+
+ sk->sk_state = SS_CONNECTING;
+
+ ret = vmbus_send_tl_connect_request(
+ &hvsk->local_addr.shv_service_id,
+ &hvsk->remote_addr.shv_service_id);
+ if (ret < 0)
+ goto out;
+
+ /* Mark sock as connecting and set the error code to in
+ * progress in case this is a non-blocking connect.
+ */
+ sock->state = SS_CONNECTING;
+ ret = -EINPROGRESS;
+ }
+
+ ret = hvsock_connect_wait(sock, flags, ret);
+out:
+ release_sock(sk);
+ return ret;
+}
+
+static int hvsock_accept_wait(struct sock *listener,
+ struct socket *newsock, int flags)
+{
+ struct hvsock_sock *hvconnected;
+ struct sock *connected;
+
+ DEFINE_WAIT(wait);
+ long timeout;
+
+ int ret = 0;
+
+ /* Wait for children sockets to appear; these are the new sockets
+ * created upon connection establishment.
+ */
+ timeout = sock_sndtimeo(listener, flags & O_NONBLOCK);
+ prepare_to_wait(sk_sleep(listener), &wait, TASK_INTERRUPTIBLE);
+
+ while ((connected = hvsock_dequeue_accept(listener)) == NULL &&
+ listener->sk_err == 0) {
+ release_sock(listener);
+ timeout = schedule_timeout(timeout);
+ lock_sock(listener);
+
+ if (signal_pending(current)) {
+ ret = sock_intr_errno(timeout);
+ goto out_wait;
+ } else if (timeout == 0) {
+ ret = -EAGAIN;
+ goto out_wait;
+ }
+
+ prepare_to_wait(sk_sleep(listener), &wait, TASK_INTERRUPTIBLE);
+ }
+
+ if (listener->sk_err)
+ ret = -listener->sk_err;
+
+ if (connected) {
+ lock_sock(connected);
+ hvconnected = sk_to_hvsock(connected);
+
+ if (ret) {
+ release_sock(connected);
+ sock_put(connected);
+ } else {
+ newsock->state = SS_CONNECTED;
+ sock_graft(connected, newsock);
+ release_sock(connected);
+ sock_put(connected);
+ }
+ }
+
+out_wait:
+ finish_wait(sk_sleep(listener), &wait);
+ return ret;
+}
+
+static
+int hvsock_accept(struct socket *sock, struct socket *newsock, int flags)
+{
+ struct sock *listener;
+ int ret;
+
+ listener = sock->sk;
+
+ lock_sock(listener);
+
+ if (sock->type != SOCK_STREAM) {
+ ret = -EOPNOTSUPP;
+ goto out;
+ }
+
+ if (listener->sk_state != SS_LISTEN) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ ret = hvsock_accept_wait(listener, newsock, flags);
+out:
+ release_sock(listener);
+ return ret;
+}
+
+static int hvsock_listen(struct socket *sock, int backlog)
+{
+ struct hvsock_sock *hvsk;
+ struct sock *sk;
+ int ret = 0;
+
+ sk = sock->sk;
+ lock_sock(sk);
+
+ if (sock->type != SOCK_STREAM) {
+ ret = -EOPNOTSUPP;
+ goto out;
+ }
+
+ if (sock->state != SS_UNCONNECTED) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ if (backlog <= 0) {
+ ret = -EINVAL;
+ goto out;
+ }
+ /* This is an artificial limit */
+ if (backlog > 128)
+ backlog = 128;
+
+ hvsk = sk_to_hvsock(sk);
+ if (!hvsock_addr_bound(&hvsk->local_addr)) {
+ ret = -EINVAL;
+ goto out;
+ }
+
+ sk->sk_ack_backlog = 0;
+ sk->sk_max_ack_backlog = backlog;
+ sk->sk_state = SS_LISTEN;
+out:
+ release_sock(sk);
+ return ret;
+}
+
+static int hvsock_send_data(struct vmbus_channel *channel,
+ struct hvsock_sock *hvsk,
+ size_t to_write)
+{
+ hvsk->send.hdr.pkt_type = 1;
+ hvsk->send.hdr.data_size = to_write;
+ return vmbus_sendpacket(channel, &hvsk->send.hdr,
+ sizeof(hvsk->send.hdr) + to_write,
+ 0, VM_PKT_DATA_INBAND, 0);
+}
+
+static
+int hvsock_sendmsg_wait(struct sock *sk, struct msghdr *msg, size_t len)
+{
+ struct hvsock_sock *hvsk = sk_to_hvsock(sk);
+ struct vmbus_channel *channel = hvsk->channel;
+
+ size_t total_to_write = len;
+ size_t total_written = 0;
+ bool can_write;
+
+ int ret = 0;
+
+ DEFINE_WAIT(wait);
+ long timeout;
+
+ timeout = sock_sndtimeo(sk, msg->msg_flags & MSG_DONTWAIT);
+ prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
+
+ while (total_to_write > 0) {
+ size_t to_write;
+
+ while (1) {
+ get_ringbuffer_rw_status(channel, NULL, &can_write);
+
+ if (can_write || sk->sk_err != 0 ||
+ (sk->sk_shutdown & SEND_SHUTDOWN) ||
+ (hvsk->peer_shutdown & RCV_SHUTDOWN))
+ break;
+
+ /* Don't wait for non-blocking sockets. */
+ if (timeout == 0) {
+ ret = -EAGAIN;
+ goto out_wait;
+ }
+
+ release_sock(sk);
+
+ timeout = schedule_timeout(timeout);
+
+ lock_sock(sk);
+ if (signal_pending(current)) {
+ ret = sock_intr_errno(timeout);
+ goto out_wait;
+ } else if (timeout == 0) {
+ ret = -EAGAIN;
+ goto out_wait;
+ }
+
+ prepare_to_wait(sk_sleep(sk), &wait,
+ TASK_INTERRUPTIBLE);
+ }
+
+ /* These checks occur both as part of and after the loop
+ * conditional since we need to check before and after
+ * sleeping.
+ */
+ if (sk->sk_err) {
+ ret = -sk->sk_err;
+ goto out_wait;
+ } else if ((sk->sk_shutdown & SEND_SHUTDOWN) ||
+ (hvsk->peer_shutdown & RCV_SHUTDOWN)) {
+ ret = -EPIPE;
+ goto out_wait;
+ }
+
+ /* Note: that write will only write as many bytes as possible
+ * in the ringbuffer. It is the caller's responsibility to
+ * check how many bytes we actually wrote.
+ */
+ do {
+ to_write = min_t(size_t, HVSOCK_SND_BUF_SZ,
+ total_to_write);
+ ret = memcpy_from_msg(hvsk->send.buf, msg, to_write);
+ if (ret != 0)
+ goto out_wait;
+
+ ret = hvsock_send_data(channel, hvsk, to_write);
+ if (ret != 0)
+ goto out_wait;
+
+ total_written += to_write;
+ total_to_write -= to_write;
+ } while (total_to_write > 0);
+ }
+
+out_wait:
+ if (total_written > 0)
+ ret = total_written;
+
+ finish_wait(sk_sleep(sk), &wait);
+ return ret;
+}
+
+static int hvsock_sendmsg(struct socket *sock, struct msghdr *msg, size_t len)
+{
+ struct hvsock_sock *hvsk;
+ struct sock *sk;
+ int ret;
+
+ if (len == 0)
+ return -EINVAL;
+
+ if (msg->msg_flags & ~MSG_DONTWAIT) {
+ pr_err("%s: unsupported flags=0x%x\n", __func__,
+ msg->msg_flags);
+ return -EOPNOTSUPP;
+ }
+
+ sk = sock->sk;
+ hvsk = sk_to_hvsock(sk);
+
+ lock_sock(sk);
+
+ /* Callers should not provide a destination with stream sockets. */
+ if (msg->msg_namelen) {
+ ret = -EOPNOTSUPP;
+ goto out;
+ }
+
+ /* Send data only if both sides are not shutdown in the direction. */
+ if (sk->sk_shutdown & SEND_SHUTDOWN ||
+ hvsk->peer_shutdown & RCV_SHUTDOWN) {
+ ret = -EPIPE;
+ goto out;
+ }
+
+ if (sk->sk_state != SS_CONNECTED ||
+ !hvsock_addr_bound(&hvsk->local_addr)) {
+ ret = -ENOTCONN;
+ goto out;
+ }
+
+ if (!hvsock_addr_bound(&hvsk->remote_addr)) {
+ ret = -EDESTADDRREQ;
+ goto out;
+ }
+
+ ret = hvsock_sendmsg_wait(sk, msg, len);
+out:
+ release_sock(sk);
+
+ /* ret is a bigger-than-0 total_written or a negative err code. */
+ if (ret == 0) {
+ WARN(1, "unexpected return value of 0\n");
+ ret = -EIO;
+ }
+
+ return ret;
+}
+
+static int hvsock_recv_data(struct vmbus_channel *channel,
+ struct hvsock_sock *hvsk,
+ size_t *payload_len)
+{
+ u32 buffer_actual_len;
+ u64 dummy_req_id;
+ int ret;
+
+ ret = vmbus_recvpacket(channel, &hvsk->recv.hdr,
+ sizeof(hvsk->recv.hdr) + sizeof(hvsk->recv.buf),
+ &buffer_actual_len, &dummy_req_id);
+ if (ret != 0 || buffer_actual_len <= sizeof(hvsk->recv.hdr))
+ *payload_len = 0;
+ else
+ *payload_len = hvsk->recv.hdr.data_size;
+
+ return ret;
+}
+
+static int hvsock_recvmsg_wait(struct sock *sk, struct msghdr *msg,
+ size_t len, int flags)
+{
+ struct hvsock_sock *hvsk = sk_to_hvsock(sk);
+ struct vmbus_channel *channel = hvsk->channel;
+
+ size_t to_read, total_to_read = len;
+ size_t copied = 0;
+ bool can_read;
+
+ int ret = 0;
+
+ DEFINE_WAIT(wait);
+ long timeout;
+
+ timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
+ prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
+
+ while (1) {
+ bool need_refill = hvsk->recv.data_len == 0;
+
+ if (need_refill)
+ get_ringbuffer_rw_status(channel, &can_read, NULL);
+ else
+ can_read = true;
+
+ if (can_read) {
+ size_t payload_len;
+
+ if (need_refill) {
+ ret = hvsock_recv_data(channel, hvsk,
+ &payload_len);
+ if (ret != 0 || payload_len == 0 ||
+ payload_len > HVSOCK_RCV_BUF_SZ) {
+ ret = -EIO;
+ goto out_wait;
+ }
+
+ hvsk->recv.data_len = payload_len;
+ hvsk->recv.data_offset = 0;
+ }
+
+ to_read = min_t(size_t, total_to_read,
+ hvsk->recv.data_len);
+
+ ret = memcpy_to_msg(msg, hvsk->recv.buf +
+ hvsk->recv.data_offset,
+ to_read);
+ if (ret != 0)
+ break;
+
+ copied += to_read;
+ total_to_read -= to_read;
+
+ hvsk->recv.data_len -= to_read;
+
+ if (hvsk->recv.data_len == 0)
+ hvsk->recv.data_offset = 0;
+ else
+ hvsk->recv.data_offset += to_read;
+
+ if (total_to_read == 0)
+ break;
+ } else {
+ if (sk->sk_err || (sk->sk_shutdown & RCV_SHUTDOWN) ||
+ (hvsk->peer_shutdown & SEND_SHUTDOWN))
+ break;
+
+ /* Don't wait for non-blocking sockets. */
+ if (timeout == 0) {
+ ret = -EAGAIN;
+ break;
+ }
+
+ if (copied > 0)
+ break;
+
+ release_sock(sk);
+ timeout = schedule_timeout(timeout);
+ lock_sock(sk);
+
+ if (signal_pending(current)) {
+ ret = sock_intr_errno(timeout);
+ break;
+ } else if (timeout == 0) {
+ ret = -EAGAIN;
+ break;
+ }
+
+ prepare_to_wait(sk_sleep(sk), &wait,
+ TASK_INTERRUPTIBLE);
+ }
+ }
+
+ if (sk->sk_err)
+ ret = -sk->sk_err;
+ else if (sk->sk_shutdown & RCV_SHUTDOWN)
+ ret = 0;
+
+ if (copied > 0) {
+ ret = copied;
+
+ /* If the other side has shutdown for sending and there
+ * is nothing more to read, then we modify the socket
+ * state.
+ */
+ if ((hvsk->peer_shutdown & SEND_SHUTDOWN) &&
+ hvsk->recv.data_len == 0) {
+ get_ringbuffer_rw_status(channel, &can_read, NULL);
+ if (!can_read) {
+ sk->sk_state = SS_UNCONNECTED;
+ sock_set_flag(sk, SOCK_DONE);
+ sk->sk_state_change(sk);
+ }
+ }
+ }
+out_wait:
+ finish_wait(sk_sleep(sk), &wait);
+ return ret;
+}
+
+static int hvsock_recvmsg(struct socket *sock, struct msghdr *msg,
+ size_t len, int flags)
+{
+ struct sock *sk = sock->sk;
+ int ret;
+
+ lock_sock(sk);
+
+ if (sk->sk_state != SS_CONNECTED) {
+ /* Recvmsg is supposed to return 0 if a peer performs an
+ * orderly shutdown. Differentiate between that case and when a
+ * peer has not connected or a local shutdown occurred with the
+ * SOCK_DONE flag.
+ */
+ if (sock_flag(sk, SOCK_DONE))
+ ret = 0;
+ else
+ ret = -ENOTCONN;
+
+ goto out;
+ }
+
+ /* We ignore msg->addr_name/len. */
+ if (flags & ~MSG_DONTWAIT) {
+ pr_err("%s: unsupported flags=0x%x\n", __func__, flags);
+ ret = -EOPNOTSUPP;
+ goto out;
+ }
+
+ /* We don't check peer_shutdown flag here since peer may actually shut
+ * down, but there can be data in the queue that a local socket can
+ * receive.
+ */
+ if (sk->sk_shutdown & RCV_SHUTDOWN) {
+ ret = 0;
+ goto out;
+ }
+
+ /* It is valid on Linux to pass in a zero-length receive buffer. This
+ * is not an error. We may as well bail out now.
+ */
+ if (!len) {
+ ret = 0;
+ goto out;
+ }
+
+ ret = hvsock_recvmsg_wait(sk, msg, len, flags);
+out:
+ release_sock(sk);
+ return ret;
+}
+
+static const struct proto_ops hvsock_ops = {
+ .family = PF_HYPERV,
+ .owner = THIS_MODULE,
+ .release = hvsock_release,
+ .bind = hvsock_bind,
+ .connect = hvsock_connect,
+ .socketpair = sock_no_socketpair,
+ .accept = hvsock_accept,
+ .getname = hvsock_getname,
+ .poll = hvsock_poll,
+ .ioctl = sock_no_ioctl,
+ .listen = hvsock_listen,
+ .shutdown = hvsock_shutdown,
+ .setsockopt = sock_no_setsockopt,
+ .getsockopt = sock_no_getsockopt,
+ .sendmsg = hvsock_sendmsg,
+ .recvmsg = hvsock_recvmsg,
+ .mmap = sock_no_mmap,
+ .sendpage = sock_no_sendpage,
+};
+
+static int hvsock_create(struct net *net, struct socket *sock,
+ int protocol, int kern)
+{
+ if (!capable(CAP_SYS_ADMIN) && !capable(CAP_NET_ADMIN))
+ return -EPERM;
+
+ if (protocol != 0 && protocol != SHV_PROTO_RAW)
+ return -EPROTONOSUPPORT;
+
+ switch (sock->type) {
+ case SOCK_STREAM:
+ sock->ops = &hvsock_ops;
+ break;
+ default:
+ return -ESOCKTNOSUPPORT;
+ }
+
+ sock->state = SS_UNCONNECTED;
+
+ return __hvsock_create(net, sock, GFP_KERNEL, 0) ? 0 : -ENOMEM;
+}
+
+static const struct net_proto_family hvsock_family_ops = {
+ .family = AF_HYPERV,
+ .create = hvsock_create,
+ .owner = THIS_MODULE,
+};
+
+static int hvsock_probe(struct hv_device *hdev,
+ const struct hv_vmbus_device_id *dev_id)
+{
+ struct vmbus_channel *channel = hdev->channel;
+
+ /* We ignore the error return code to suppress the unnecessary
+ * error message in vmbus_probe(): on error the host will rescind
+ * the offer in 30 seconds and we can do cleanup at that time.
+ */
+ (void)hvsock_open_connection(channel);
+
+ return 0;
+}
+
+static int hvsock_remove(struct hv_device *hdev)
+{
+ struct vmbus_channel *channel = hdev->channel;
+
+ vmbus_close(channel);
+
+ return 0;
+}
+
+/* It's not really used. See vmbus_match() and vmbus_probe(). */
+static const struct hv_vmbus_device_id id_table[] = {
+ {},
+};
+
+static struct hv_driver hvsock_drv = {
+ .name = "hv_sock",
+ .hvsock = true,
+ .id_table = id_table,
+ .probe = hvsock_probe,
+ .remove = hvsock_remove,
+};
+
+static int __init hvsock_init(void)
+{
+ int ret;
+
+ /* Hyper-V Sockets requires at least VMBus 4.0 */
+ if ((vmbus_proto_version >> 16) < 4) {
+ pr_err("failed to load: VMBus 4 or later is required\n");
+ return -ENODEV;
+ }
+
+ ret = vmbus_driver_register(&hvsock_drv);
+ if (ret) {
+ pr_err("failed to register hv_sock driver\n");
+ return ret;
+ }
+
+ ret = proto_register(&hvsock_proto, 0);
+ if (ret) {
+ pr_err("failed to register protocol\n");
+ goto unreg_hvsock_drv;
+ }
+
+ ret = sock_register(&hvsock_family_ops);
+ if (ret) {
+ pr_err("failed to register address family\n");
+ goto unreg_proto;
+ }
+
+ return 0;
+
+unreg_proto:
+ proto_unregister(&hvsock_proto);
+unreg_hvsock_drv:
+ vmbus_driver_unregister(&hvsock_drv);
+ return ret;
+}
+
+static void __exit hvsock_exit(void)
+{
+ sock_unregister(AF_HYPERV);
+ proto_unregister(&hvsock_proto);
+ vmbus_driver_unregister(&hvsock_drv);
+}
+
+module_init(hvsock_init);
+module_exit(hvsock_exit);
+
+MODULE_DESCRIPTION("Hyper-V Sockets");
+MODULE_LICENSE("Dual BSD/GPL");
--
2.1.0
^ permalink raw reply related
* [PATCH net-next 0/2] sfc: RSS enhancements
From: Edward Cree @ 2016-05-04 16:56 UTC (permalink / raw)
To: linux-net-drivers, netdev, David Miller; +Cc: Daniel Pieczko
Daniel Pieczko (1):
sfc: allocate rx pages on the same node as the interrupt
Edward Cree (1):
sfc: Support setting rss_cpus to 'cores', 'packages' or 'hyperthreads'
drivers/net/ethernet/sfc/efx.c | 181 ++++++++++++++++++++++++++++------
drivers/net/ethernet/sfc/net_driver.h | 3 +
drivers/net/ethernet/sfc/rx.c | 18 +++-
3 files changed, 169 insertions(+), 33 deletions(-)
^ permalink raw reply
* [PATCH v9 net-next 0/2] introduce Hyper-V VM Sockets(hv_sock)
From: Dexuan Cui @ 2016-05-04 16:56 UTC (permalink / raw)
To: gregkh, davem, netdev, linux-kernel, devel, olaf, apw, jasowang,
cavery, kys, haiyangz
Cc: joe
Hyper-V Sockets (hv_sock) supplies a byte-stream based communication
mechanism between the host and the guest. It's somewhat like TCP over
VMBus, but the transportation layer (VMBus) is much simpler than IP.
With Hyper-V Sockets, applications between the host and the guest can talk
to each other directly by the traditional BSD-style socket APIs.
Hyper-V Sockets is only available on new Windows hosts, like Windows Server
2016. More info is in this article "Make your own integration services":
https://msdn.microsoft.com/en-us/virtualization/hyperv_on_windows/develop/make_mgmt_service
The patch implements the necessary support in the guest side by
introducing a new socket address family AF_HYPERV.
You can also get the patch by:
https://github.com/dcui/linux/commits/decui/hv_sock/net-next/20160502_v09
Note: the VMBus driver side's supporting patches have been in the mainline
tree.
I know the kernel has already had a VM Sockets driver (AF_VSOCK) based
on VMware VMCI (net/vmw_vsock/, drivers/misc/vmw_vmci), and KVM is
proposing AF_VSOCK of virtio version:
http://marc.info/?l=linux-netdev&m=145952064004765&w=2
However, though Hyper-V Sockets may seem conceptually similar to
AF_VOSCK, there are differences in the transportation layer, and IMO these
make the direct code reusing impractical:
1. In AF_VSOCK, the endpoint type is: <u32 ContextID, u32 Port>, but in
AF_HYPERV, the endpoint type is: <GUID VM_ID, GUID ServiceID>. Here GUID
is 128-bit.
2. AF_VSOCK supports SOCK_DGRAM, while AF_HYPERV doesn't.
3. AF_VSOCK supports some special sock opts, like SO_VM_SOCKETS_BUFFER_SIZE,
SO_VM_SOCKETS_BUFFER_MIN/MAX_SIZE and SO_VM_SOCKETS_CONNECT_TIMEOUT.
These are meaningless to AF_HYPERV.
4. Some AF_VSOCK's VMCI transportation ops are meanless to AF_HYPERV/VMBus,
like .notify_recv_init
.notify_recv_pre_block
.notify_recv_pre_dequeue
.notify_recv_post_dequeue
.notify_send_init
.notify_send_pre_block
.notify_send_pre_enqueue
.notify_send_post_enqueue
etc.
So I think we'd better introduce a new address family: AF_HYPERV.
Please review the patch.
Looking forward to your comments, especially comments from David. :-)
Changes since v1:
- updated "[PATCH 6/7] hvsock: introduce Hyper-V VM Sockets feature"
- added __init and __exit for the module init/exit functions
- net/hv_sock/Kconfig: "default m" -> "default m if HYPERV"
- MODULE_LICENSE: "Dual MIT/GPL" -> "Dual BSD/GPL"
Changes since v2:
- fixed various coding issue pointed out by David Miller
- fixed indentation issues
- removed pr_debug in net/hv_sock/af_hvsock.c
- used reverse-Chrismas-tree style for local variables.
- EXPORT_SYMBOL -> EXPORT_SYMBOL_GPL
Changes since v3:
- fixed a few coding issue pointed by Vitaly Kuznetsov and Dan Carpenter
- fixed the ret value in vmbus_recvpacket_hvsock on error
- fixed the style of multi-line comment: vmbus_get_hvsock_rw_status()
Changes since v4 (https://lkml.org/lkml/2015/7/28/404):
- addressed all the comments about V4.
- treat the hvsock offers/channels as special VMBus devices
- add a mechanism to pass hvsock events to the hvsock driver
- fixed some corner cases with proper locking when a connection is closed
- rebased to the latest Greg's tree
Changes since v5 (https://lkml.org/lkml/2015/12/24/103):
- addressed the coding style issues (Vitaly Kuznetsov & David Miller, thanks!)
- used a better coding for the per-channel rescind callback (Thank Vitaly!)
- avoided the introduction of new VMBUS driver APIs vmbus_sendpacket_hvsock()
and vmbus_recvpacket_hvsock() and used vmbus_sendpacket()/vmbus_recvpacket()
in the higher level (i.e., the vmsock driver). Thank Vitaly!
Changes since v6 (http://lkml.iu.edu/hypermail/linux/kernel/1601.3/01813.html)
- only a few minor changes of coding style and comments
Changes since v7
- a few minor changes of coding style: thanks, Joe Perches!
- added some lines of comments about GUID/UUID before the struct sockaddr_hv.
Changes since v8
- removed the unnecessary __packed for some definitions: thanks, David!
- hvsock_open_connection: use offer.u.pipe.user_def[0] to know the connection
and reorganized the function
direction
- reorganized the code according to suggestions from Cathy Avery: split big
functions into small ones, set .setsockopt and getsockopt to
sock_no_setsockopt/sock_no_getsockopt
- inline'd some small list helper functions
Dexuan Cui (2):
hv_sock: introduce Hyper-V Sockets
net: add the AF_HYPERV entries to family name tables
MAINTAINERS | 2 +
include/linux/hyperv.h | 16 +
include/linux/socket.h | 5 +-
include/net/af_hvsock.h | 55 ++
include/uapi/linux/hyperv.h | 25 +
net/Kconfig | 1 +
net/Makefile | 1 +
net/core/sock.c | 6 +-
net/hv_sock/Kconfig | 10 +
net/hv_sock/Makefile | 3 +
net/hv_sock/af_hvsock.c | 1434 +++++++++++++++++++++++++++++++++++++++++++
11 files changed, 1553 insertions(+), 5 deletions(-)
create mode 100644 include/net/af_hvsock.h
create mode 100644 net/hv_sock/Kconfig
create mode 100644 net/hv_sock/Makefile
create mode 100644 net/hv_sock/af_hvsock.c
--
2.1.0
^ permalink raw reply
* Re: [PATCH net-next v2] block/drbd: use nla_put_u64_64bit()
From: David Miller @ 2016-05-04 16:50 UTC (permalink / raw)
To: eric.dumazet
Cc: nicolas.dichtel, netdev, philipp.reisner, drbd-dev, linux-kernel
In-Reply-To: <1462372026.5535.344.camel@edumazet-glaptop3.roam.corp.google.com>
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Wed, 04 May 2016 07:27:06 -0700
> kernel was fine, and most user land apps were fine as well.
Userland should really not have to deal with garbage like this.
And because it quietly works just fine on x86-64, nothing makes
sure that applications will universally get this right anyways.
Better to align things properly and magically all of these issues
simply disappear.
^ permalink raw reply
* Re: [PATCH net] igmp: fix byte order in /proc/net/igmp output
From: David Miller @ 2016-05-04 16:48 UTC (permalink / raw)
To: eric.dumazet; +Cc: Eugene.Crosser, crosser, netdev
In-Reply-To: <1462367616.5535.340.camel@edumazet-glaptop3.roam.corp.google.com>
From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Wed, 04 May 2016 06:13:36 -0700
> On Wed, 2016-05-04 at 14:11 +0200, Eugene Crosser wrote:
>> /proc/net/igmp is a readonly attribute that shows multicast groups
>> to which different network interfaces are subscribed. Among other
>> things, it displays `multiaddr` which is a 32 bit network-byte-order
>> field, in hexadecimal format. Prior to this patch, the field was
>> displayed as an integer, resulting in reverse byte order on little
>> endian architectures. This patch converts it with ntohl() for display
>> the same way as this is done for the /proc/net/mcfilter attribute.
>>
>> The patch changes (corrects) user-visible behaviour.
>>
>> Signed-off-by: Eugene Crosser <Eugene.Crosser@ru.ibm.com>
>> ---
>> net/ipv4/igmp.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/net/ipv4/igmp.c b/net/ipv4/igmp.c
>> index b3086cf..f9d2139 100644
>> --- a/net/ipv4/igmp.c
>> +++ b/net/ipv4/igmp.c
>> @@ -2722,7 +2722,7 @@ static int igmp_mc_seq_show(struct seq_file *seq, void *v)
>> delta = im->timer.expires - jiffies;
>> seq_printf(seq,
>> "\t\t\t\t%08X %5d %d:%08lX\t\t%d\n",
>> - im->multiaddr, im->users,
>> + ntohl(im->multiaddr), im->users,
>> im->tm_running,
>> im->tm_running ? jiffies_delta_to_clock_t(delta) : 0,
>> im->reporter);
>
>
> I do not believe we can change this. This is unfortunately too late.
>
> /proc/net/tcp has the same 'issue' or if you prefer, well known
> behavior.
>
> Such change would break basically all /proc/net/igmp users.
Agreed.
^ permalink raw reply
* Re: [PATCH net-next v2] block/drbd: use nla_put_u64_64bit()
From: David Miller @ 2016-05-04 16:47 UTC (permalink / raw)
To: nicolas.dichtel; +Cc: netdev, philipp.reisner, drbd-dev, linux-kernel
In-Reply-To: <5729EFBC.7040002@6wind.com>
From: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Date: Wed, 4 May 2016 14:49:00 +0200
> Le 04/05/2016 11:05, Lars Ellenberg a écrit :
> [snip]
>> We don't have an "alignment problem" there, btw.
>> Last time I checked, we did work fine without this alignment magic,
>> we already take care of that, yes, even on affected architectures.
> The code adds several consecutive u64 attributes. The nl attribute header is 4
> bytes, thus the full attribute length is 12 bytes. If the first u64 is aligned
> on 8 (nla_data()), the next one is not aligned on 8: it starts 12 bytes (8 (u64)
> + 4 (nl attr hdr)) after the previous u64.
Right.
^ permalink raw reply
* Re: [PATCH net-next] gre: change gre_parse_header to return the header length
From: David Miller @ 2016-05-04 16:45 UTC (permalink / raw)
To: jbenc; +Cc: netdev, tom
In-Reply-To: <3627136025fbd767d70e2ce563aa5e37373c880a.1462280197.git.jbenc@redhat.com>
From: Jiri Benc <jbenc@redhat.com>
Date: Tue, 3 May 2016 15:00:21 +0200
> It's easier for gre_parse_header to return the header length instead of
> filing it into a parameter. That way, the callers that don't care about the
> header length can just check whether the returned value is lower than zero.
>
> In gre_err, the tunnel header must not be pulled. See commit b7f8fe251e46
> ("gre: do not pull header in ICMP error processing") for details.
>
> This patch reduces the conflict between the mentioned commit and commit
> 95f5c64c3c13 ("gre: Move utility functions to common headers").
>
> Signed-off-by: Jiri Benc <jbenc@redhat.com>
Applied.
^ permalink raw reply
* Re: [PATCH next-next 2/7] gre: Move utility functions to common headers
From: David Miller @ 2016-05-04 16:45 UTC (permalink / raw)
To: jbenc; +Cc: tom, netdev, kernel-team
In-Reply-To: <20160504103549.228c9e8a@griffin>
From: Jiri Benc <jbenc@redhat.com>
Date: Wed, 4 May 2016 10:35:49 +0200
> On Wed, 04 May 2016 00:53:16 -0400 (EDT), David Miller wrote:
>> From: Jiri Benc <jbenc@redhat.com>
>> Date: Tue, 3 May 2016 13:29:44 +0200
>>
>> > How do we resolve the conflict between net and net-next? I'd prefer
>> > gre_parse_header to return the header length. I can submit a patch for
>> > net-next that does this; that would substantially ease the merge.
>>
>> Jiri, I just did the net --> net-next merge. Please send me something on
>> top of that which does what you like.
>
> The patch http://patchwork.ozlabs.org/patch/617940/ still applies and
> does the right thing even post merge.
>
> Note that the merge reintroduced the bug fixed by b7f8fe251e46 (the
> mentioned patch fixes this, too).
Awesome, I'll use that, thanks Jiri.
^ permalink raw reply
* [ethtool 2/3] ethtool/ixgbe: Correct offsets and support x550, x550em_x, x550em_a
From: Jeff Kirsher @ 2016-05-04 16:44 UTC (permalink / raw)
To: davem; +Cc: Preethi Banala, netdev, nhorman, sassmann, jogreene, Jeff Kirsher
In-Reply-To: <1462380293-4132-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Preethi Banala <preethi.banala@intel.com>
Correct hard-coded wrong offset values and add several conditions to
display registers for x550, x550em_x and x550em_a based on datasheet/
EAS document.
Signed-off-by: Preethi Banala <preethi.banala@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
ixgbe.c | 74 ++++++++++++++++++++++++++++++++++++-----------------------------
1 file changed, 41 insertions(+), 33 deletions(-)
diff --git a/ixgbe.c b/ixgbe.c
index a375782..4f5af77 100644
--- a/ixgbe.c
+++ b/ixgbe.c
@@ -173,6 +173,7 @@ ixgbe_dump_regs(struct ethtool_drvinfo *info, struct ethtool_regs *regs)
u32 *regs_buff = (u32 *)regs->data;
u32 regs_buff_len = regs->len / sizeof(*regs_buff);
u32 reg;
+ u32 offset;
u16 hw_device_id = (u16) regs->version;
u8 version = (u8)(regs->version >> 24);
u8 i;
@@ -273,7 +274,7 @@ ixgbe_dump_regs(struct ethtool_drvinfo *info, struct ethtool_regs *regs)
reg = regs_buff[1047];
fprintf(stdout,
- "0x04250: HLREG0 (Highlander Control 0 register) 0x%08X\n"
+ "0x04240: HLREG0 (Highlander Control 0 register) 0x%08X\n"
" Transmit CRC: %s\n"
" Receive CRC Strip: %s\n"
" Jumbo Frames: %s\n"
@@ -320,17 +321,19 @@ ixgbe_dump_regs(struct ethtool_drvinfo *info, struct ethtool_regs *regs)
regs_buff[7]);
/* NVM Register */
+ offset = mac_type == ixgbe_mac_x550em_a ? 0x15FF8 : 0x10010;
fprintf(stdout,
- "0x10010: EEC (EEPROM/Flash Control) 0x%08X\n",
- regs_buff[8]);
+ "0x%05X: EEC (EEPROM/Flash Control) 0x%08X\n",
+ offset, regs_buff[8]);
fprintf(stdout,
"0x10014: EERD (EEPROM Read) 0x%08X\n",
regs_buff[9]);
+ offset = mac_type == ixgbe_mac_x550em_a ? 0x15F6C : 0x1001C;
fprintf(stdout,
- "0x1001C: FLA (Flash Access) 0x%08X\n",
- regs_buff[10]);
+ "0x%05X: FLA (Flash Access) 0x%08X\n",
+ offset, regs_buff[10]);
fprintf(stdout,
"0x10110: EEMNGCTL (Manageability EEPROM Control) 0x%08X\n",
@@ -341,7 +344,7 @@ ixgbe_dump_regs(struct ethtool_drvinfo *info, struct ethtool_regs *regs)
regs_buff[12]);
fprintf(stdout,
- "0x10110: FLMNGCTL (Manageability Flash Control) 0x%08X\n",
+ "0x10118: FLMNGCTL (Manageability Flash Control) 0x%08X\n",
regs_buff[13]);
fprintf(stdout,
@@ -356,9 +359,10 @@ ixgbe_dump_regs(struct ethtool_drvinfo *info, struct ethtool_regs *regs)
"0x1013C: FLOP (Flash Opcode) 0x%08X\n",
regs_buff[16]);
+ offset = mac_type == ixgbe_mac_x550em_a ? 0x15F64 : 0x10200;
fprintf(stdout,
- "0x10200: GRC (General Receive Control) 0x%08X\n",
- regs_buff[17]);
+ "0x%05X: GRC (General Receive Control) 0x%08X\n",
+ offset, regs_buff[17]);
/* Interrupt */
fprintf(stdout,
@@ -690,7 +694,7 @@ ixgbe_dump_regs(struct ethtool_drvinfo *info, struct ethtool_regs *regs)
fprintf(stdout,
"0x%05X: TDPT2TCSR%d (Tx Data Plane T2 TC Status %d) 0x%08X\n",
0x0CD40 + (4 * i), i, i, regs_buff[873 + i]);
- } else if (mac_type >= ixgbe_mac_82599EB) {
+ } else if (mac_type >= ixgbe_mac_82599EB && mac_type <= ixgbe_mac_x550) {
fprintf(stdout,
"0x04900: RTTDCS (Tx Descr Plane Ctrl&Status) 0x%08X\n",
regs_buff[830]);
@@ -718,60 +722,64 @@ ixgbe_dump_regs(struct ethtool_drvinfo *info, struct ethtool_regs *regs)
"0x%05X: RTTDT2C%d (Tx Descr Plane T2 Config %d) 0x%08X\n",
0x04910 + (4 * i), i, i, regs_buff[849 + i]);
- for (i = 0; i < 8; i++)
- fprintf(stdout,
- "0x%05X: RTTDT2S%d (Tx Descr Plane T2 Status %d) 0x%08X\n",
- 0x04930 + (4 * i), i, i, regs_buff[857 + i]);
+ if (mac_type < ixgbe_mac_x550)
+ for (i = 0; i < 8; i++)
+ fprintf(stdout,
+ "0x%05X: RTTDT2S%d (Tx Descr Plane T2 Status %d) 0x%08X\n",
+ 0x04930 + (4 * i), i, i, regs_buff[857 + i]);
for (i = 0; i < 8; i++)
fprintf(stdout,
"0x%05X: RTTPT2C%d (Tx Packet Plane T2 Config %d) 0x%08X\n",
0x0CD20 + (4 * i), i, i, regs_buff[865]);
- for (i = 0; i < 8; i++)
- fprintf(stdout,
- "0x%05X: RTTPT2S%d (Tx Packet Plane T2 Status %d) 0x%08X\n",
- 0x0CD40 + (4 * i), i, i, regs_buff[873 + i]);
+ if (mac_type < ixgbe_mac_x550)
+ for (i = 0; i < 8; i++)
+ fprintf(stdout,
+ "0x%05X: RTTPT2S%d (Tx Packet Plane T2 Status %d) 0x%08X\n",
+ 0x0CD40 + (4 * i), i, i, regs_buff[873 + i]);
+ }
- if (regs_buff_len > 1129) {
- fprintf(stdout,
+ if (regs_buff_len > 1129 && mac_type != ixgbe_mac_82598EB) {
+ fprintf(stdout,
"0x03020: RTRUP2TC (Rx User Prio to Traffic Classes)0x%08X\n",
regs_buff[1129]);
- fprintf(stdout,
+ fprintf(stdout,
"0x0C800: RTTUP2TC (Tx User Prio to Traffic Classes)0x%08X\n",
regs_buff[1130]);
+ if (mac_type <= ixgbe_mac_x550)
for (i = 0; i < 4; i++)
fprintf(stdout,
- "0x%05X: TXLLQ%d (Strict Low Lat Tx Queues %d) 0x%08X\n",
- 0x082E0 + (4 * i), i, i, regs_buff[1131 + i]);
+ "0x%05X: TXLLQ%d (Strict Low Lat Tx Queues %d) 0x%08X\n",
+ 0x082E0 + (4 * i), i, i, regs_buff[1131 + i]);
- if (mac_type == ixgbe_mac_82599EB) {
- fprintf(stdout,
+ if (mac_type == ixgbe_mac_82599EB) {
+ fprintf(stdout,
"0x04980: RTTBCNRM (DCB TX Rate Sched MMW) 0x%08X\n",
regs_buff[1135]);
- fprintf(stdout,
+ fprintf(stdout,
"0x0498C: RTTBCNRD (DCB TX Rate-Scheduler Drift) 0x%08X\n",
regs_buff[1136]);
- } else if (mac_type == ixgbe_mac_X540) {
- fprintf(stdout,
+ } else if (mac_type <= ixgbe_mac_x550) {
+ fprintf(stdout,
"0x04980: RTTQCNRM (DCB TX QCN Rate Sched MMW) 0x%08X\n",
regs_buff[1135]);
- fprintf(stdout,
+ fprintf(stdout,
"0x0498C: RTTQCNRR (DCB TX QCN Rate Reset) 0x%08X\n",
regs_buff[1136]);
+ if (mac_type < ixgbe_mac_x550)
fprintf(stdout,
- "0x08B00: RTTQCNCR (DCB TX QCN Control) 0x%08X\n",
- regs_buff[1137]);
+ "0x08B00: RTTQCNCR (DCB TX QCN Control) 0x%08X\n",
+ regs_buff[1137]);
- fprintf(stdout,
+ fprintf(stdout,
"0x04A90: RTTQCNTG (DCB TX QCN Tagging) 0x%08X\n",
regs_buff[1138]);
- }
}
}
@@ -1127,7 +1135,7 @@ ixgbe_dump_regs(struct ethtool_drvinfo *info, struct ethtool_regs *regs)
regs_buff[1068]);
fprintf(stdout,
- "0x042B0: ANLP2 (Auto Neg Lnk Part. Ctrl Word 2) 0x%08X\n",
+ "0x042B4: ANLP2 (Auto Neg Lnk Part. Ctrl Word 2) 0x%08X\n",
regs_buff[1069]);
}
--
2.5.5
^ permalink raw reply related
* [ethtool 3/3] ethtool/ixgbe: Reduce format string storage
From: Jeff Kirsher @ 2016-05-04 16:44 UTC (permalink / raw)
To: davem; +Cc: Preethi Banala, netdev, nhorman, sassmann, jogreene, Jeff Kirsher
In-Reply-To: <1462380293-4132-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Preethi Banala <preethi.banala@intel.com>
Reduce format string storage using for loop.
Signed-off-by: Preethi Banala <preethi.banala@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
ixgbe.c | 57 ++++++++++++---------------------------------------------
1 file changed, 12 insertions(+), 45 deletions(-)
diff --git a/ixgbe.c b/ixgbe.c
index 4f5af77..ff0e769 100644
--- a/ixgbe.c
+++ b/ixgbe.c
@@ -1193,21 +1193,10 @@ ixgbe_dump_regs(struct ethtool_drvinfo *info, struct ethtool_regs *regs)
"0x07F08: TDHMPN (Tx Desc Handler Mem Page Num) 0x%08X\n",
regs_buff[1095]);
- fprintf(stdout,
- "0x07F10: TIC_DW0 (Tx Desc Hand. Mem Read Data 0) 0x%08X\n",
- regs_buff[1096]);
-
- fprintf(stdout,
- "0x07F14: TIC_DW1 (Tx Desc Hand. Mem Read Data 1) 0x%08X\n",
- regs_buff[1097]);
-
- fprintf(stdout,
- "0x07F18: TIC_DW2 (Tx Desc Hand. Mem Read Data 2) 0x%08X\n",
- regs_buff[1098]);
-
- fprintf(stdout,
- "0x07F1C: TIC_DW3 (Tx Desc Hand. Mem Read Data 3) 0x%08X\n",
- regs_buff[1099]);
+ for (i = 0; i < 4; i++)
+ fprintf(stdout,
+ "0x%05X: TIC_DW%d (Tx Desc Hand. Mem Read Data %d) 0x%08X\n",
+ 0x07F10 + (4 * i), i, i, regs_buff[1096 + i]);
fprintf(stdout,
"0x07F20: TDPROBE (Tx Probe Mode Status) 0x%08X\n",
@@ -1217,41 +1206,19 @@ ixgbe_dump_regs(struct ethtool_drvinfo *info, struct ethtool_regs *regs)
"0x0C600: TXBUFCTRL (TX Buffer Access Control) 0x%08X\n",
regs_buff[1101]);
- fprintf(stdout,
- "0x0C610: TXBUFDATA0 (TX Buffer DATA 0) 0x%08X\n",
- regs_buff[1102]);
-
- fprintf(stdout,
- "0x0C614: TXBUFDATA1 (TX Buffer DATA 1) 0x%08X\n",
- regs_buff[1103]);
-
- fprintf(stdout,
- "0x0C618: TXBUFDATA2 (TX Buffer DATA 2) 0x%08X\n",
- regs_buff[1104]);
-
- fprintf(stdout,
- "0x0C61C: TXBUFDATA3 (TX Buffer DATA 3) 0x%08X\n",
- regs_buff[1105]);
+ for (i = 0; i < 4; i++)
+ fprintf(stdout,
+ "0x%05X: TXBUFDATA%d (TX Buffer DATA %d) 0x%08X\n",
+ 0x0C610 + (4 * i), i, i, regs_buff[1102 + i]);
fprintf(stdout,
"0x03600: RXBUFCTRL (RX Buffer Access Control) 0x%08X\n",
regs_buff[1106]);
- fprintf(stdout,
- "0x03610: RXBUFDATA0 (RX Buffer DATA 0) 0x%08X\n",
- regs_buff[1107]);
-
- fprintf(stdout,
- "0x03614: RXBUFDATA1 (RX Buffer DATA 1) 0x%08X\n",
- regs_buff[1108]);
-
- fprintf(stdout,
- "0x03618: RXBUFDATA2 (RX Buffer DATA 2) 0x%08X\n",
- regs_buff[1109]);
-
- fprintf(stdout,
- "0x0361C: RXBUFDATA3 (RX Buffer DATA 3) 0x%08X\n",
- regs_buff[1110]);
+ for (i = 0; i < 4; i++)
+ fprintf(stdout,
+ "0x%05X: RXBUFDATA%d (RX Buffer DATA %d) 0x%08X\n",
+ 0x03610 + (4 * i), i, i, regs_buff[1107 + i]);
for (i = 0; i < 8; i++)
fprintf(stdout,
--
2.5.5
^ permalink raw reply related
* [ethtool 0/3][pull request] Intel Wired LAN Driver Updates 2016-05-03
From: Jeff Kirsher @ 2016-05-04 16:44 UTC (permalink / raw)
To: davem; +Cc: Jeff Kirsher, netdev, nhorman, sassmann, jogreene
This series contains updates to ixgbe in ethtool.
Preethi adds missing device IDs and mac_type definitions, also updated
the display registers for x550, x550em_x/a. Cleaned up the format string
storage by taking advantage of "for" loops.
The following are changes since commit deb1c6613ec14fd828d321e38c7bea45fe559bd5:
Release version 4.5.
and are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/ethtool master
Preethi Banala (3):
ethtool/ixgbe: Add device ID and mac_type definitions
ethtool/ixgbe: Correct offsets and support x550, x550em_x, x550em_a
ethtool/ixgbe: Reduce format string storage
ixgbe.c | 173 +++++++++++++++++++++++++++++++++++-----------------------------
1 file changed, 95 insertions(+), 78 deletions(-)
--
2.5.5
^ permalink raw reply
* [ethtool 1/3] ethtool/ixgbe: Add device ID and mac_type definitions
From: Jeff Kirsher @ 2016-05-04 16:44 UTC (permalink / raw)
To: davem; +Cc: Preethi Banala, netdev, nhorman, sassmann, jogreene, Jeff Kirsher
In-Reply-To: <1462380293-4132-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Preethi Banala <preethi.banala@intel.com>
Add device id macro definitions, mac_type enum definition and update
get_mac_type() accordingly.
Signed-off-by: Preethi Banala <preethi.banala@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
ixgbe.c | 42 ++++++++++++++++++++++++++++++++++++++++++
1 file changed, 42 insertions(+)
diff --git a/ixgbe.c b/ixgbe.c
index 3928d77..a375782 100644
--- a/ixgbe.c
+++ b/ixgbe.c
@@ -63,6 +63,23 @@
#define IXGBE_SUBDEV_ID_82599_KX4_KR_MEZZ 0x000C
#define IXGBE_DEV_ID_82599_LS 0x154F
#define IXGBE_DEV_ID_X540T 0x1528
+#define IXGBE_DEV_ID_82599_SFP_SF_QP 0x154A
+#define IXGBE_DEV_ID_82599_QSFP_SF_QP 0x1558
+#define IXGBE_DEV_ID_X540T1 0x1560
+
+#define IXGBE_DEV_ID_X550T 0x1563
+#define IXGBE_DEV_ID_X550T1 0x15D1
+#define IXGBE_DEV_ID_X550EM_X_KX4 0x15AA
+#define IXGBE_DEV_ID_X550EM_X_KR 0x15AB
+#define IXGBE_DEV_ID_X550EM_X_SFP 0x15AC
+#define IXGBE_DEV_ID_X550EM_X_10G_T 0x15AD
+#define IXGBE_DEV_ID_X550EM_X_1G_T 0x15AE
+#define IXGBE_DEV_ID_X550EM_A_KR 0x15C2
+#define IXGBE_DEV_ID_X550EM_A_KR_L 0x15C3
+#define IXGBE_DEV_ID_X550EM_A_SFP_N 0x15C4
+#define IXGBE_DEV_ID_X550EM_A_SGMII 0x15C6
+#define IXGBE_DEV_ID_X550EM_A_SGMII_L 0x15C7
+#define IXGBE_DEV_ID_X550EM_A_SFP 0x15CE
/*
* Enumerated types specific to the ixgbe hardware
@@ -73,6 +90,9 @@ enum ixgbe_mac_type {
ixgbe_mac_82598EB,
ixgbe_mac_82599EB,
ixgbe_mac_X540,
+ ixgbe_mac_x550,
+ ixgbe_mac_x550em_x,
+ ixgbe_mac_x550em_a,
ixgbe_num_macs
};
@@ -112,11 +132,33 @@ ixgbe_get_mac_type(u16 device_id)
case IXGBE_DEV_ID_82599_COMBO_BACKPLANE:
case IXGBE_SUBDEV_ID_82599_KX4_KR_MEZZ:
case IXGBE_DEV_ID_82599_LS:
+ case IXGBE_DEV_ID_82599_SFP_SF_QP:
+ case IXGBE_DEV_ID_82599_QSFP_SF_QP:
mac_type = ixgbe_mac_82599EB;
break;
case IXGBE_DEV_ID_X540T:
+ case IXGBE_DEV_ID_X540T1:
mac_type = ixgbe_mac_X540;
break;
+ case IXGBE_DEV_ID_X550T:
+ case IXGBE_DEV_ID_X550T1:
+ mac_type = ixgbe_mac_x550;
+ break;
+ case IXGBE_DEV_ID_X550EM_X_KX4:
+ case IXGBE_DEV_ID_X550EM_X_KR:
+ case IXGBE_DEV_ID_X550EM_X_SFP:
+ case IXGBE_DEV_ID_X550EM_X_10G_T:
+ case IXGBE_DEV_ID_X550EM_X_1G_T:
+ mac_type = ixgbe_mac_x550em_x;
+ break;
+ case IXGBE_DEV_ID_X550EM_A_KR:
+ case IXGBE_DEV_ID_X550EM_A_KR_L:
+ case IXGBE_DEV_ID_X550EM_A_SFP_N:
+ case IXGBE_DEV_ID_X550EM_A_SGMII:
+ case IXGBE_DEV_ID_X550EM_A_SGMII_L:
+ case IXGBE_DEV_ID_X550EM_A_SFP:
+ mac_type = ixgbe_mac_x550em_a;
+ break;
default:
mac_type = ixgbe_mac_82598EB;
break;
--
2.5.5
^ permalink raw reply related
* [net-next 01/14] net_sched: act_mirred: add helper inlines to access tcf_mirred info
From: Jeff Kirsher @ 2016-05-04 16:26 UTC (permalink / raw)
To: davem; +Cc: Sridhar Samudrala, netdev, nhorman, sassmann, jogreene,
Jeff Kirsher
In-Reply-To: <1462379208-3997-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Sridhar Samudrala <sridhar.samudrala@intel.com>
Signed-off-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
include/net/tc_act/tc_mirred.h | 15 +++++++++++++++
1 file changed, 15 insertions(+)
diff --git a/include/net/tc_act/tc_mirred.h b/include/net/tc_act/tc_mirred.h
index dae96ba..e891835 100644
--- a/include/net/tc_act/tc_mirred.h
+++ b/include/net/tc_act/tc_mirred.h
@@ -2,6 +2,7 @@
#define __NET_TC_MIR_H
#include <net/act_api.h>
+#include <linux/tc_act/tc_mirred.h>
struct tcf_mirred {
struct tcf_common common;
@@ -14,4 +15,18 @@ struct tcf_mirred {
#define to_mirred(a) \
container_of(a->priv, struct tcf_mirred, common)
+static inline bool is_tcf_mirred_redirect(const struct tc_action *a)
+{
+#ifdef CONFIG_NET_CLS_ACT
+ if (a->ops && a->ops->type == TCA_ACT_MIRRED)
+ return to_mirred(a)->tcfm_eaction == TCA_EGRESS_REDIR;
+#endif
+ return false;
+}
+
+static inline int tcf_mirred_ifindex(const struct tc_action *a)
+{
+ return to_mirred(a)->tcfm_ifindex;
+}
+
#endif /* __NET_TC_MIR_H */
--
2.5.5
^ permalink raw reply related
* [net-next 02/14] ixgbe: Add support for redirect action to cls_u32 offloads
From: Jeff Kirsher @ 2016-05-04 16:26 UTC (permalink / raw)
To: davem; +Cc: Sridhar Samudrala, netdev, nhorman, sassmann, jogreene,
Jeff Kirsher
In-Reply-To: <1462379208-3997-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Sridhar Samudrala <sridhar.samudrala@intel.com>
This patch enables 'redirect' to a SRIOV VF or a offloaded macvlan
device queue via tc 'mirred' action.
Verified with the following script that creates SRIOV VFs, offloaded
macvlan and adds tc u32 filters with redirect action to the associated
netdevs.
# add ingress qdisc.
tc qdisc add dev p4p1 ingress
# enable hw tc offload.
ethtool -K p4p1 hw-tc-offload on
# create 4 sriov VFs and bring up the first one.
echo 4 > /sys/class/net/p4p1/device/sriov_numvfs
sleep 1
ip link set p4p1 up
ip link set p4p1_0 up
# create a offloaded macvlan device and bring it up.
ethtool -K p4p1 l2-fwd-offload on
ip link add link p4p1 name mvlan_1 type macvlan
ip link set mvlan_1 up
# add u32 filter with action to redirect to VF netdev
tc filter add dev p4p1 parent ffff: protocol ip prio 99 \
handle 800:0:1 u32 ht 800: \
match ip src 192.168.1.3/32 \
action mirred egress redirect dev p4p1_0
# add u32 filter with action to redirect to macvlan netdev
tc filter add dev p4p1 parent ffff: protocol ip prio 99 \
handle 800:0:2 u32 ht 800: \
match ip src 192.168.2.3/32 \
action mirred egress redirect dev mvlan_1
Signed-off-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 96 +++++++++++++++++++++++----
1 file changed, 83 insertions(+), 13 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index 0ef4a15..5aa22ac 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -53,6 +53,7 @@
#include <net/vxlan.h>
#include <net/pkt_cls.h>
#include <net/tc_act/tc_gact.h>
+#include <net/tc_act/tc_mirred.h>
#include "ixgbe.h"
#include "ixgbe_common.h"
@@ -8319,6 +8320,85 @@ static int ixgbe_configure_clsu32_del_hnode(struct ixgbe_adapter *adapter,
return 0;
}
+#ifdef CONFIG_NET_CLS_ACT
+static int handle_redirect_action(struct ixgbe_adapter *adapter, int ifindex,
+ u8 *queue, u64 *action)
+{
+ unsigned int num_vfs = adapter->num_vfs, vf;
+ struct net_device *upper;
+ struct list_head *iter;
+
+ /* redirect to a SRIOV VF */
+ for (vf = 0; vf < num_vfs; ++vf) {
+ upper = pci_get_drvdata(adapter->vfinfo[vf].vfdev);
+ if (upper->ifindex == ifindex) {
+ if (adapter->num_rx_pools > 1)
+ *queue = vf * 2;
+ else
+ *queue = vf * adapter->num_rx_queues_per_pool;
+
+ *action = vf + 1;
+ *action <<= ETHTOOL_RX_FLOW_SPEC_RING_VF_OFF;
+ return 0;
+ }
+ }
+
+ /* redirect to a offloaded macvlan netdev */
+ netdev_for_each_all_upper_dev_rcu(adapter->netdev, upper, iter) {
+ if (netif_is_macvlan(upper)) {
+ struct macvlan_dev *dfwd = netdev_priv(upper);
+ struct ixgbe_fwd_adapter *vadapter = dfwd->fwd_priv;
+
+ if (vadapter && vadapter->netdev->ifindex == ifindex) {
+ *queue = adapter->rx_ring[vadapter->rx_base_queue]->reg_idx;
+ *action = *queue;
+ return 0;
+ }
+ }
+ }
+
+ return -EINVAL;
+}
+
+static int parse_tc_actions(struct ixgbe_adapter *adapter,
+ struct tcf_exts *exts, u64 *action, u8 *queue)
+{
+ const struct tc_action *a;
+ int err;
+
+ if (tc_no_actions(exts))
+ return -EINVAL;
+
+ tc_for_each_action(a, exts) {
+
+ /* Drop action */
+ if (is_tcf_gact_shot(a)) {
+ *action = IXGBE_FDIR_DROP_QUEUE;
+ *queue = IXGBE_FDIR_DROP_QUEUE;
+ return 0;
+ }
+
+ /* Redirect to a VF or a offloaded macvlan */
+ if (is_tcf_mirred_redirect(a)) {
+ int ifindex = tcf_mirred_ifindex(a);
+
+ err = handle_redirect_action(adapter, ifindex, queue,
+ action);
+ if (err == 0)
+ return err;
+ }
+ }
+
+ return -EINVAL;
+}
+#else
+static int parse_tc_actions(struct ixgbe_adapter *adapter,
+ struct tcf_exts *exts, u64 *action, u8 *queue)
+{
+ return -EINVAL;
+}
+#endif /* CONFIG_NET_CLS_ACT */
+
static int ixgbe_configure_clsu32(struct ixgbe_adapter *adapter,
__be16 protocol,
struct tc_cls_u32_offload *cls)
@@ -8328,9 +8408,6 @@ static int ixgbe_configure_clsu32(struct ixgbe_adapter *adapter,
struct ixgbe_mat_field *field_ptr;
struct ixgbe_fdir_filter *input;
union ixgbe_atr_input mask;
-#ifdef CONFIG_NET_CLS_ACT
- const struct tc_action *a;
-#endif
int i, err = 0;
u8 queue;
u32 uhtid, link_uhtid;
@@ -8432,18 +8509,11 @@ static int ixgbe_configure_clsu32(struct ixgbe_adapter *adapter,
if (input->filter.formatted.flow_type == IXGBE_ATR_FLOW_TYPE_IPV4)
mask.formatted.flow_type &= IXGBE_ATR_L4TYPE_IPV6_MASK;
-#ifdef CONFIG_NET_CLS_ACT
- if (list_empty(&cls->knode.exts->actions))
+ err = parse_tc_actions(adapter, cls->knode.exts, &input->action,
+ &queue);
+ if (err < 0)
goto err_out;
- list_for_each_entry(a, &cls->knode.exts->actions, list) {
- if (!is_tcf_gact_shot(a))
- goto err_out;
- }
-#endif
-
- input->action = IXGBE_FDIR_DROP_QUEUE;
- queue = IXGBE_FDIR_DROP_QUEUE;
input->sw_idx = loc;
spin_lock(&adapter->fdir_perfect_lock);
--
2.5.5
^ permalink raw reply related
* [net-next 00/14][pull request] 10GbE Intel Wired LAN Driver Updates 2016-05-04
From: Jeff Kirsher @ 2016-05-04 16:26 UTC (permalink / raw)
To: davem; +Cc: Jeff Kirsher, netdev, nhorman, sassmann, jogreene, john.ronciak
This series contains updates to ixgbe, ixgbevf and traffic class helpers.
Sridhar adds helper functions to the tc_mirred header to access tcf_mirred
information and then implements them for ixgbe to enable redirection to
a SRIOV VF or an offloaded MACVLAN device queue via tc 'mirred' action.
Amritha adds support to set filters with multiple header fields (L3,L4)
to match on.
KY Srinivasan from Microsoft add Hyper-V support into ixgbevf.
Emil adds 82599 sub-device IDs that were missing from the list of parts
that support WoL. Then simplified the logic we use to determine WoL
support by reading the EEPROM bits for MACs X540 and newer.
Preethi cleaned up duplicate and unused device IDs. Fixed our ethtool
stat reporting where we were ignoring higher 32 bits of stats registers,
so fill out 64 bit stat values into two 32 bit words.
Babu Moger from Oracle improves VF performance issues on SPARC.
Alex Duyck cleans up some of the Hyper-V implementation from KY so that
we can just use function pointers instead of having to identify if a
given VF is running on a Linux or Windows PF.
Usha makes sure that DCB and FCoE is disabled for X550EM_x/a MACs and
cleans up the DCB initialization in the process.
Tony cleans up the API for ixgbevf_update_xcast_mode() so we do not
have to pass in the netdev parameter, since it was never used in the
function.
The following are changes since commit 26879da58711aa604a1b866cbeedd7e0f78f90ad:
ipv6: add new struct ipcm6_cookie
and are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next-queue 10GbE
Alexander Duyck (1):
ixgbevf: Use mac_ops instead of trying to identify NIC type
Amritha Nambiar (1):
ixgbe: Match on multiple headers for cls_u32 offloads
Babu Moger (1):
ixgbevf: Change the relaxed order settings in VF driver for sparc
Emil Tantilov (2):
ixgbe: add WoL support for some 82599 subdevice IDs
ixgbe: check EEPROM for WOL support for X540 and above
KY Srinivasan (2):
ixgbevf: Add the device ID's presented while running on Hyper-V
ixgbevf: Support Windows hosts (Hyper-V)
Preethi Banala (3):
ixgbe: Remove duplicate and unused device ID definitions
ixgbe: Return 64 bit stats values
ixgbe: Revise populating few registers and macro definitions
Sridhar Samudrala (2):
net_sched: act_mirred: add helper inlines to access tcf_mirred info
ixgbe: Add support for redirect action to cls_u32 offloads
Tony Nguyen (1):
ixgbevf: Remove unused parameter
Usha Ketineni (1):
ixgbe: Disable DCB and FCoE for X550EM_x and x550em_a
drivers/net/ethernet/intel/ixgbe/ixgbe.h | 7 +-
drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c | 27 +-
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 480 +++++++++++++++-------
drivers/net/ethernet/intel/ixgbe/ixgbe_model.h | 6 +
drivers/net/ethernet/intel/ixgbe/ixgbe_type.h | 25 +-
drivers/net/ethernet/intel/ixgbevf/defines.h | 5 +
drivers/net/ethernet/intel/ixgbevf/ixgbevf.h | 10 +
drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c | 28 +-
drivers/net/ethernet/intel/ixgbevf/mbx.c | 11 +
drivers/net/ethernet/intel/ixgbevf/vf.c | 219 +++++++++-
drivers/net/ethernet/intel/ixgbevf/vf.h | 4 +-
include/net/tc_act/tc_mirred.h | 15 +
12 files changed, 640 insertions(+), 197 deletions(-)
--
2.5.5
^ permalink raw reply
* [net-next 14/14] ixgbevf: Remove unused parameter
From: Jeff Kirsher @ 2016-05-04 16:26 UTC (permalink / raw)
To: davem; +Cc: Tony Nguyen, netdev, nhorman, sassmann, jogreene, Jeff Kirsher
In-Reply-To: <1462379208-3997-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Tony Nguyen <anthony.l.nguyen@intel.com>
ixgbevf_update_xcast_mode() is not using the netdev parameter;
removing it since it's unnecessary.
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c | 2 +-
drivers/net/ethernet/intel/ixgbevf/vf.c | 8 ++------
drivers/net/ethernet/intel/ixgbevf/vf.h | 2 +-
3 files changed, 4 insertions(+), 8 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c b/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
index ba17a61..5e348b1 100644
--- a/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
+++ b/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
@@ -1922,7 +1922,7 @@ static void ixgbevf_set_rx_mode(struct net_device *netdev)
spin_lock_bh(&adapter->mbx_lock);
- hw->mac.ops.update_xcast_mode(hw, netdev, xcast_mode);
+ hw->mac.ops.update_xcast_mode(hw, xcast_mode);
/* reprogram multicast list */
hw->mac.ops.update_mc_addr_list(hw, netdev);
diff --git a/drivers/net/ethernet/intel/ixgbevf/vf.c b/drivers/net/ethernet/intel/ixgbevf/vf.c
index f66055aa..e670d3b 100644
--- a/drivers/net/ethernet/intel/ixgbevf/vf.c
+++ b/drivers/net/ethernet/intel/ixgbevf/vf.c
@@ -536,13 +536,11 @@ static s32 ixgbevf_hv_update_mc_addr_list_vf(struct ixgbe_hw *hw,
/**
* ixgbevf_update_xcast_mode - Update Multicast mode
* @hw: pointer to the HW structure
- * @netdev: pointer to net device structure
* @xcast_mode: new multicast mode
*
* Updates the Multicast Mode of VF.
**/
-static s32 ixgbevf_update_xcast_mode(struct ixgbe_hw *hw,
- struct net_device *netdev, int xcast_mode)
+static s32 ixgbevf_update_xcast_mode(struct ixgbe_hw *hw, int xcast_mode)
{
struct ixgbe_mbx_info *mbx = &hw->mbx;
u32 msgbuf[2];
@@ -576,9 +574,7 @@ static s32 ixgbevf_update_xcast_mode(struct ixgbe_hw *hw,
/**
* Hyper-V variant - just a stub.
*/
-static s32 ixgbevf_hv_update_xcast_mode(struct ixgbe_hw *hw,
- struct net_device *netdev,
- int xcast_mode)
+static s32 ixgbevf_hv_update_xcast_mode(struct ixgbe_hw *hw, int xcast_mode)
{
return -EOPNOTSUPP;
}
diff --git a/drivers/net/ethernet/intel/ixgbevf/vf.h b/drivers/net/ethernet/intel/ixgbevf/vf.h
index 6770f10..2cac610 100644
--- a/drivers/net/ethernet/intel/ixgbevf/vf.h
+++ b/drivers/net/ethernet/intel/ixgbevf/vf.h
@@ -64,7 +64,7 @@ struct ixgbe_mac_operations {
s32 (*set_uc_addr)(struct ixgbe_hw *, u32, u8 *);
s32 (*init_rx_addrs)(struct ixgbe_hw *);
s32 (*update_mc_addr_list)(struct ixgbe_hw *, struct net_device *);
- s32 (*update_xcast_mode)(struct ixgbe_hw *, struct net_device *, int);
+ s32 (*update_xcast_mode)(struct ixgbe_hw *, int);
s32 (*enable_mc)(struct ixgbe_hw *);
s32 (*disable_mc)(struct ixgbe_hw *);
s32 (*clear_vfta)(struct ixgbe_hw *);
--
2.5.5
^ permalink raw reply related
* [net-next 11/14] ixgbevf: Change the relaxed order settings in VF driver for sparc
From: Jeff Kirsher @ 2016-05-04 16:26 UTC (permalink / raw)
To: davem; +Cc: Babu Moger, netdev, nhorman, sassmann, jogreene, Jeff Kirsher
In-Reply-To: <1462379208-3997-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Babu Moger <babu.moger@oracle.com>
We noticed performance issues with VF interface on sparc compared
to PF. Setting the RX to IXGBE_DCA_RXCTRL_DATA_WRO_EN brings it
on far with PF. Also this matches to the default sparc setting in
PF driver.
Signed-off-by: Babu Moger <babu.moger@oracle.com>
Acked-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c b/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
index f39c8cb..14d4729 100644
--- a/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
+++ b/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
@@ -1760,9 +1760,15 @@ static void ixgbevf_configure_rx_ring(struct ixgbevf_adapter *adapter,
IXGBE_WRITE_REG(hw, IXGBE_VFRDLEN(reg_idx),
ring->count * sizeof(union ixgbe_adv_rx_desc));
+#ifndef CONFIG_SPARC
/* enable relaxed ordering */
IXGBE_WRITE_REG(hw, IXGBE_VFDCA_RXCTRL(reg_idx),
IXGBE_DCA_RXCTRL_DESC_RRO_EN);
+#else
+ IXGBE_WRITE_REG(hw, IXGBE_VFDCA_RXCTRL(reg_idx),
+ IXGBE_DCA_RXCTRL_DESC_RRO_EN |
+ IXGBE_DCA_RXCTRL_DATA_WRO_EN);
+#endif
/* reset head and tail pointers */
IXGBE_WRITE_REG(hw, IXGBE_VFRDH(reg_idx), 0);
--
2.5.5
^ permalink raw reply related
* [net-next 13/14] ixgbe: Disable DCB and FCoE for X550EM_x and x550em_a
From: Jeff Kirsher @ 2016-05-04 16:26 UTC (permalink / raw)
To: davem; +Cc: Usha Ketineni, netdev, nhorman, sassmann, jogreene, Jeff Kirsher
In-Reply-To: <1462379208-3997-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Usha Ketineni <usha.k.ketineni@intel.com>
This patch adds IXGBE_FLAG_DCB_CAPABLE flag that is set
for all MACs other than X550EM_x and x550em_a. DCB and
FCoE is disabled for these MACS. DCB initialization
code is moved to a separate function.
Signed-off-by: Usha Ketineni <usha.k.ketineni@intel.com>
Tested-by: Ronald Bynoe <ronald.j.bynoe@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe.h | 1 +
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 111 ++++++++++++++++----------
drivers/net/ethernet/intel/ixgbe/ixgbe_type.h | 1 +
3 files changed, 71 insertions(+), 42 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe.h b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
index e216a3d..9f2db18 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe.h
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe.h
@@ -644,6 +644,7 @@ struct ixgbe_adapter {
#define IXGBE_FLAG_VXLAN_OFFLOAD_CAPABLE BIT(24)
#define IXGBE_FLAG_RX_HWTSTAMP_ENABLED BIT(25)
#define IXGBE_FLAG_RX_HWTSTAMP_IN_REGISTER BIT(26)
+#define IXGBE_FLAG_DCB_CAPABLE BIT(27)
u32 flags2;
#define IXGBE_FLAG2_RSC_CAPABLE BIT(0)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
index 403b3cc..2f69842 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
@@ -5559,6 +5559,58 @@ static void ixgbe_tx_timeout(struct net_device *netdev)
ixgbe_tx_timeout_reset(adapter);
}
+#ifdef CONFIG_IXGBE_DCB
+static void ixgbe_init_dcb(struct ixgbe_adapter *adapter)
+{
+ struct ixgbe_hw *hw = &adapter->hw;
+ struct tc_configuration *tc;
+ int j;
+
+ switch (hw->mac.type) {
+ case ixgbe_mac_82598EB:
+ case ixgbe_mac_82599EB:
+ adapter->dcb_cfg.num_tcs.pg_tcs = MAX_TRAFFIC_CLASS;
+ adapter->dcb_cfg.num_tcs.pfc_tcs = MAX_TRAFFIC_CLASS;
+ break;
+ case ixgbe_mac_X540:
+ case ixgbe_mac_X550:
+ adapter->dcb_cfg.num_tcs.pg_tcs = X540_TRAFFIC_CLASS;
+ adapter->dcb_cfg.num_tcs.pfc_tcs = X540_TRAFFIC_CLASS;
+ break;
+ case ixgbe_mac_X550EM_x:
+ case ixgbe_mac_x550em_a:
+ default:
+ adapter->dcb_cfg.num_tcs.pg_tcs = DEF_TRAFFIC_CLASS;
+ adapter->dcb_cfg.num_tcs.pfc_tcs = DEF_TRAFFIC_CLASS;
+ break;
+ }
+
+ /* Configure DCB traffic classes */
+ for (j = 0; j < MAX_TRAFFIC_CLASS; j++) {
+ tc = &adapter->dcb_cfg.tc_config[j];
+ tc->path[DCB_TX_CONFIG].bwg_id = 0;
+ tc->path[DCB_TX_CONFIG].bwg_percent = 12 + (j & 1);
+ tc->path[DCB_RX_CONFIG].bwg_id = 0;
+ tc->path[DCB_RX_CONFIG].bwg_percent = 12 + (j & 1);
+ tc->dcb_pfc = pfc_disabled;
+ }
+
+ /* Initialize default user to priority mapping, UPx->TC0 */
+ tc = &adapter->dcb_cfg.tc_config[0];
+ tc->path[DCB_TX_CONFIG].up_to_tc_bitmap = 0xFF;
+ tc->path[DCB_RX_CONFIG].up_to_tc_bitmap = 0xFF;
+
+ adapter->dcb_cfg.bw_percentage[DCB_TX_CONFIG][0] = 100;
+ adapter->dcb_cfg.bw_percentage[DCB_RX_CONFIG][0] = 100;
+ adapter->dcb_cfg.pfc_mode_enable = false;
+ adapter->dcb_set_bitmap = 0x00;
+ if (adapter->flags & IXGBE_FLAG_DCB_CAPABLE)
+ adapter->dcbx_cap = DCB_CAP_DCBX_HOST | DCB_CAP_DCBX_VER_CEE;
+ memcpy(&adapter->temp_dcb_cfg, &adapter->dcb_cfg,
+ sizeof(adapter->temp_dcb_cfg));
+}
+#endif
+
/**
* ixgbe_sw_init - Initialize general software structures (struct ixgbe_adapter)
* @adapter: board private structure to initialize
@@ -5575,10 +5627,6 @@ static int ixgbe_sw_init(struct ixgbe_adapter *adapter)
u32 fwsm;
u16 device_caps;
int i;
-#ifdef CONFIG_IXGBE_DCB
- int j;
- struct tc_configuration *tc;
-#endif
/* PCI config space info */
@@ -5600,6 +5648,10 @@ static int ixgbe_sw_init(struct ixgbe_adapter *adapter)
#ifdef CONFIG_IXGBE_DCA
adapter->flags |= IXGBE_FLAG_DCA_CAPABLE;
#endif
+#ifdef CONFIG_IXGBE_DCB
+ adapter->flags |= IXGBE_FLAG_DCB_CAPABLE;
+ adapter->flags &= ~IXGBE_FLAG_DCB_ENABLED;
+#endif
#ifdef IXGBE_FCOE
adapter->flags |= IXGBE_FLAG_FCOE_CAPABLE;
adapter->flags &= ~IXGBE_FLAG_FCOE_ENABLED;
@@ -5656,6 +5708,16 @@ static int ixgbe_sw_init(struct ixgbe_adapter *adapter)
break;
case ixgbe_mac_X550EM_x:
case ixgbe_mac_x550em_a:
+#ifdef CONFIG_IXGBE_DCB
+ adapter->flags &= ~IXGBE_FLAG_DCB_CAPABLE;
+#endif
+#ifdef IXGBE_FCOE
+ adapter->flags &= ~IXGBE_FLAG_FCOE_CAPABLE;
+#ifdef CONFIG_IXGBE_DCB
+ adapter->fcoe.up = 0;
+#endif /* IXGBE_DCB */
+#endif /* IXGBE_FCOE */
+ /* Fall Through */
case ixgbe_mac_X550:
#ifdef CONFIG_IXGBE_DCA
adapter->flags &= ~IXGBE_FLAG_DCA_CAPABLE;
@@ -5677,43 +5739,7 @@ static int ixgbe_sw_init(struct ixgbe_adapter *adapter)
spin_lock_init(&adapter->fdir_perfect_lock);
#ifdef CONFIG_IXGBE_DCB
- switch (hw->mac.type) {
- case ixgbe_mac_X540:
- case ixgbe_mac_X550:
- case ixgbe_mac_X550EM_x:
- case ixgbe_mac_x550em_a:
- adapter->dcb_cfg.num_tcs.pg_tcs = X540_TRAFFIC_CLASS;
- adapter->dcb_cfg.num_tcs.pfc_tcs = X540_TRAFFIC_CLASS;
- break;
- default:
- adapter->dcb_cfg.num_tcs.pg_tcs = MAX_TRAFFIC_CLASS;
- adapter->dcb_cfg.num_tcs.pfc_tcs = MAX_TRAFFIC_CLASS;
- break;
- }
-
- /* Configure DCB traffic classes */
- for (j = 0; j < MAX_TRAFFIC_CLASS; j++) {
- tc = &adapter->dcb_cfg.tc_config[j];
- tc->path[DCB_TX_CONFIG].bwg_id = 0;
- tc->path[DCB_TX_CONFIG].bwg_percent = 12 + (j & 1);
- tc->path[DCB_RX_CONFIG].bwg_id = 0;
- tc->path[DCB_RX_CONFIG].bwg_percent = 12 + (j & 1);
- tc->dcb_pfc = pfc_disabled;
- }
-
- /* Initialize default user to priority mapping, UPx->TC0 */
- tc = &adapter->dcb_cfg.tc_config[0];
- tc->path[DCB_TX_CONFIG].up_to_tc_bitmap = 0xFF;
- tc->path[DCB_RX_CONFIG].up_to_tc_bitmap = 0xFF;
-
- adapter->dcb_cfg.bw_percentage[DCB_TX_CONFIG][0] = 100;
- adapter->dcb_cfg.bw_percentage[DCB_RX_CONFIG][0] = 100;
- adapter->dcb_cfg.pfc_mode_enable = false;
- adapter->dcb_set_bitmap = 0x00;
- adapter->dcbx_cap = DCB_CAP_DCBX_HOST | DCB_CAP_DCBX_VER_CEE;
- memcpy(&adapter->temp_dcb_cfg, &adapter->dcb_cfg,
- sizeof(adapter->temp_dcb_cfg));
-
+ ixgbe_init_dcb(adapter);
#endif
/* default flow control settings */
@@ -9495,7 +9521,8 @@ skip_sriov:
netdev->priv_flags |= IFF_SUPP_NOFCS;
#ifdef CONFIG_IXGBE_DCB
- netdev->dcbnl_ops = &dcbnl_ops;
+ if (adapter->flags & IXGBE_FLAG_DCB_CAPABLE)
+ netdev->dcbnl_ops = &dcbnl_ops;
#endif
#ifdef IXGBE_FCOE
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h b/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h
index cd6c1dd..da3d835 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h
@@ -548,6 +548,7 @@ struct ixgbe_thermal_sensor_data {
/* DCB registers */
#define MAX_TRAFFIC_CLASS 8
#define X540_TRAFFIC_CLASS 4
+#define DEF_TRAFFIC_CLASS 1
#define IXGBE_RMCS 0x03D00
#define IXGBE_DPMCS 0x07F40
#define IXGBE_PDPMCS 0x0CD00
--
2.5.5
^ permalink raw reply related
* [net-next 12/14] ixgbevf: Use mac_ops instead of trying to identify NIC type
From: Jeff Kirsher @ 2016-05-04 16:26 UTC (permalink / raw)
To: davem; +Cc: Alexander Duyck, netdev, nhorman, sassmann, jogreene,
Jeff Kirsher
In-Reply-To: <1462379208-3997-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Alexander Duyck <aduyck@mirantis.com>
This change makes it so that we can just use function pointers instead of
having to identify if a given VF is running on a Linux or Windows PF. By
doing this we can avoid having to pull too much information out of the
lower layers and can instead just make use of the mac_ops pointers since
they should differ between the two types of VFs anyway.
Signed-off-by: Alexander Duyck <aduyck@mirantis.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbevf/ixgbevf.h | 1 -
drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c | 15 +++------------
drivers/net/ethernet/intel/ixgbevf/vf.c | 20 +++++++++-----------
drivers/net/ethernet/intel/ixgbevf/vf.h | 4 +---
4 files changed, 13 insertions(+), 27 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbevf/ixgbevf.h b/drivers/net/ethernet/intel/ixgbevf/ixgbevf.h
index f2bafa6..d5944c3 100644
--- a/drivers/net/ethernet/intel/ixgbevf/ixgbevf.h
+++ b/drivers/net/ethernet/intel/ixgbevf/ixgbevf.h
@@ -494,7 +494,6 @@ void ixgbevf_free_rx_resources(struct ixgbevf_ring *);
void ixgbevf_free_tx_resources(struct ixgbevf_ring *);
void ixgbevf_update_stats(struct ixgbevf_adapter *adapter);
int ethtool_ioctl(struct ifreq *ifr);
-bool ixgbevf_on_hyperv(struct ixgbe_hw *hw);
extern void ixgbevf_write_eitr(struct ixgbevf_q_vector *q_vector);
diff --git a/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c b/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
index 14d4729..ba17a61 100644
--- a/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
+++ b/drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c
@@ -1809,10 +1809,7 @@ static void ixgbevf_configure_rx(struct ixgbevf_adapter *adapter)
ixgbevf_setup_vfmrqc(adapter);
/* notify the PF of our intent to use this size of frame */
- if (!ixgbevf_on_hyperv(hw))
- ixgbevf_rlpml_set_vf(hw, netdev->mtu + ETH_HLEN + ETH_FCS_LEN);
- else
- ixgbevf_hv_rlpml_set_vf(hw, netdev->mtu + ETH_HLEN + ETH_FCS_LEN);
+ hw->mac.ops.set_rlpml(hw, netdev->mtu + ETH_HLEN + ETH_FCS_LEN);
/* Setup the HW Rx Head and Tail Descriptor Pointers and
* the Base and Length of the Rx Descriptor Ring
@@ -2073,10 +2070,7 @@ static void ixgbevf_negotiate_api(struct ixgbevf_adapter *adapter)
spin_lock_bh(&adapter->mbx_lock);
while (api[idx] != ixgbe_mbox_api_unknown) {
- if (!ixgbevf_on_hyperv(hw))
- err = hw->mac.ops.negotiate_api_version(hw, api[idx]);
- else
- err = ixgbevf_hv_negotiate_api_version(hw, api[idx]);
+ err = hw->mac.ops.negotiate_api_version(hw, api[idx]);
if (!err)
break;
idx++;
@@ -3760,10 +3754,7 @@ static int ixgbevf_change_mtu(struct net_device *netdev, int new_mtu)
netdev->mtu = new_mtu;
/* notify the PF of our intent to use this size of frame */
- if (!ixgbevf_on_hyperv(hw))
- ixgbevf_rlpml_set_vf(hw, max_frame);
- else
- ixgbevf_hv_rlpml_set_vf(hw, max_frame);
+ hw->mac.ops.set_rlpml(hw, max_frame);
return 0;
}
diff --git a/drivers/net/ethernet/intel/ixgbevf/vf.c b/drivers/net/ethernet/intel/ixgbevf/vf.c
index 8a4eb08..f66055aa 100644
--- a/drivers/net/ethernet/intel/ixgbevf/vf.c
+++ b/drivers/net/ethernet/intel/ixgbevf/vf.c
@@ -797,11 +797,11 @@ out:
}
/**
- * ixgbevf_rlpml_set_vf - Set the maximum receive packet length
+ * ixgbevf_set_rlpml_vf - Set the maximum receive packet length
* @hw: pointer to the HW structure
* @max_size: value to assign to max frame size
**/
-void ixgbevf_rlpml_set_vf(struct ixgbe_hw *hw, u16 max_size)
+static void ixgbevf_set_rlpml_vf(struct ixgbe_hw *hw, u16 max_size)
{
u32 msgbuf[2];
@@ -811,12 +811,12 @@ void ixgbevf_rlpml_set_vf(struct ixgbe_hw *hw, u16 max_size)
}
/**
- * ixgbevf_hv_rlpml_set_vf - Set the maximum receive packet length
+ * ixgbevf_hv_set_rlpml_vf - Set the maximum receive packet length
* @hw: pointer to the HW structure
* @max_size: value to assign to max frame size
* Hyper-V variant.
**/
-void ixgbevf_hv_rlpml_set_vf(struct ixgbe_hw *hw, u16 max_size)
+static void ixgbevf_hv_set_rlpml_vf(struct ixgbe_hw *hw, u16 max_size)
{
u32 reg;
@@ -864,12 +864,12 @@ static int ixgbevf_negotiate_api_version_vf(struct ixgbe_hw *hw, int api)
}
/**
- * ixgbevf_hv_negotiate_api_version - Negotiate supported API version
+ * ixgbevf_hv_negotiate_api_version_vf - Negotiate supported API version
* @hw: pointer to the HW structure
* @api: integer containing requested API version
* Hyper-V version - only ixgbe_mbox_api_10 supported.
**/
-int ixgbevf_hv_negotiate_api_version(struct ixgbe_hw *hw, int api)
+static int ixgbevf_hv_negotiate_api_version_vf(struct ixgbe_hw *hw, int api)
{
/* Hyper-V only supports api version ixgbe_mbox_api_10 */
if (api != ixgbe_mbox_api_10)
@@ -950,6 +950,7 @@ static const struct ixgbe_mac_operations ixgbevf_mac_ops = {
.update_xcast_mode = ixgbevf_update_xcast_mode,
.set_uc_addr = ixgbevf_set_uc_addr_vf,
.set_vfta = ixgbevf_set_vfta_vf,
+ .set_rlpml = ixgbevf_set_rlpml_vf,
};
static const struct ixgbe_mac_operations ixgbevf_hv_mac_ops = {
@@ -960,11 +961,13 @@ static const struct ixgbe_mac_operations ixgbevf_hv_mac_ops = {
.stop_adapter = ixgbevf_stop_hw_vf,
.setup_link = ixgbevf_setup_mac_link_vf,
.check_link = ixgbevf_hv_check_mac_link_vf,
+ .negotiate_api_version = ixgbevf_hv_negotiate_api_version_vf,
.set_rar = ixgbevf_hv_set_rar_vf,
.update_mc_addr_list = ixgbevf_hv_update_mc_addr_list_vf,
.update_xcast_mode = ixgbevf_hv_update_xcast_mode,
.set_uc_addr = ixgbevf_hv_set_uc_addr_vf,
.set_vfta = ixgbevf_hv_set_vfta_vf,
+ .set_rlpml = ixgbevf_hv_set_rlpml_vf,
};
const struct ixgbevf_info ixgbevf_82599_vf_info = {
@@ -1006,8 +1009,3 @@ const struct ixgbevf_info ixgbevf_X550EM_x_vf_hv_info = {
.mac = ixgbe_mac_X550EM_x_vf,
.mac_ops = &ixgbevf_hv_mac_ops,
};
-
-bool ixgbevf_on_hyperv(struct ixgbe_hw *hw)
-{
- return hw->mbx.ops.check_for_msg == NULL;
-}
diff --git a/drivers/net/ethernet/intel/ixgbevf/vf.h b/drivers/net/ethernet/intel/ixgbevf/vf.h
index f7c0170..6770f10 100644
--- a/drivers/net/ethernet/intel/ixgbevf/vf.h
+++ b/drivers/net/ethernet/intel/ixgbevf/vf.h
@@ -69,6 +69,7 @@ struct ixgbe_mac_operations {
s32 (*disable_mc)(struct ixgbe_hw *);
s32 (*clear_vfta)(struct ixgbe_hw *);
s32 (*set_vfta)(struct ixgbe_hw *, u32, u32, bool);
+ void (*set_rlpml)(struct ixgbe_hw *, u16);
};
enum ixgbe_mac_type {
@@ -208,9 +209,6 @@ static inline u32 ixgbe_read_reg_array(struct ixgbe_hw *hw, u32 reg,
#define IXGBE_READ_REG_ARRAY(h, r, o) ixgbe_read_reg_array(h, r, o)
-void ixgbevf_rlpml_set_vf(struct ixgbe_hw *hw, u16 max_size);
-void ixgbevf_hv_rlpml_set_vf(struct ixgbe_hw *hw, u16 max_size);
-int ixgbevf_hv_negotiate_api_version(struct ixgbe_hw *hw, int api);
int ixgbevf_get_queues(struct ixgbe_hw *hw, unsigned int *num_tcs,
unsigned int *default_tc);
int ixgbevf_get_reta_locked(struct ixgbe_hw *hw, u32 *reta, int num_rx_queues);
--
2.5.5
^ permalink raw reply related
* [net-next 09/14] ixgbe: Return 64 bit stats values
From: Jeff Kirsher @ 2016-05-04 16:26 UTC (permalink / raw)
To: davem; +Cc: Preethi Banala, netdev, nhorman, sassmann, jogreene, Jeff Kirsher
In-Reply-To: <1462379208-3997-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Preethi Banala <preethi.banala@intel.com>
The code was ignoring higher 32 bits of stats registers. This patch
correctly fills out 64 bit value in two 32 bit words.
Signed-off-by: Preethi Banala <preethi.banala@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c | 9 ++++++---
1 file changed, 6 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
index d3efcb4..a696192 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
@@ -720,8 +720,10 @@ static void ixgbe_get_regs(struct net_device *netdev,
regs_buff[939] = IXGBE_GET_STAT(adapter, bprc);
regs_buff[940] = IXGBE_GET_STAT(adapter, mprc);
regs_buff[941] = IXGBE_GET_STAT(adapter, gptc);
- regs_buff[942] = IXGBE_GET_STAT(adapter, gorc);
- regs_buff[944] = IXGBE_GET_STAT(adapter, gotc);
+ regs_buff[942] = (u32)IXGBE_GET_STAT(adapter, gorc);
+ regs_buff[943] = (u32)(IXGBE_GET_STAT(adapter, gorc) >> 32);
+ regs_buff[944] = (u32)IXGBE_GET_STAT(adapter, gotc);
+ regs_buff[945] = (u32)(IXGBE_GET_STAT(adapter, gotc) >> 32);
for (i = 0; i < 8; i++)
regs_buff[946 + i] = IXGBE_GET_STAT(adapter, rnbc[i]);
regs_buff[954] = IXGBE_GET_STAT(adapter, ruc);
@@ -731,7 +733,8 @@ static void ixgbe_get_regs(struct net_device *netdev,
regs_buff[958] = IXGBE_GET_STAT(adapter, mngprc);
regs_buff[959] = IXGBE_GET_STAT(adapter, mngpdc);
regs_buff[960] = IXGBE_GET_STAT(adapter, mngptc);
- regs_buff[961] = IXGBE_GET_STAT(adapter, tor);
+ regs_buff[961] = (u32)IXGBE_GET_STAT(adapter, tor);
+ regs_buff[962] = (u32)(IXGBE_GET_STAT(adapter, tor) >> 32);
regs_buff[963] = IXGBE_GET_STAT(adapter, tpr);
regs_buff[964] = IXGBE_GET_STAT(adapter, tpt);
regs_buff[965] = IXGBE_GET_STAT(adapter, ptc64);
--
2.5.5
^ permalink raw reply related
* [net-next 10/14] ixgbe: Revise populating few registers and macro definitions
From: Jeff Kirsher @ 2016-05-04 16:26 UTC (permalink / raw)
To: davem; +Cc: Preethi Banala, netdev, nhorman, sassmann, jogreene, Jeff Kirsher
In-Reply-To: <1462379208-3997-1-git-send-email-jeffrey.t.kirsher@intel.com>
From: Preethi Banala <preethi.banala@intel.com>
Revise populating few registers in ixgbe_get_regs() and macro
definitions.
Before applying patch:
$ du -k objs/drivers/net/ethernet/intel/ixgbe/ixgbe.ko
8572 objs/drivers/net/ethernet/intel/ixgbe/ixgbe.ko
After applying patch:
$ du -k objs/drivers/net/ethernet/intel/ixgbe/ixgbe.ko
8568 objs/drivers/net/ethernet/intel/ixgbe/ixgbe.ko
Signed-off-by: Preethi Banala <preethi.banala@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c | 18 ++++++------------
drivers/net/ethernet/intel/ixgbe/ixgbe_type.h | 10 ++--------
2 files changed, 8 insertions(+), 20 deletions(-)
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
index a696192..59b771b 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_ethtool.c
@@ -533,10 +533,8 @@ static void ixgbe_get_regs(struct net_device *netdev,
/* Flow Control */
regs_buff[30] = IXGBE_READ_REG(hw, IXGBE_PFCTOP);
- regs_buff[31] = IXGBE_READ_REG(hw, IXGBE_FCTTV(0));
- regs_buff[32] = IXGBE_READ_REG(hw, IXGBE_FCTTV(1));
- regs_buff[33] = IXGBE_READ_REG(hw, IXGBE_FCTTV(2));
- regs_buff[34] = IXGBE_READ_REG(hw, IXGBE_FCTTV(3));
+ for (i = 0; i < 4; i++)
+ regs_buff[31 + i] = IXGBE_READ_REG(hw, IXGBE_FCTTV(i));
for (i = 0; i < 8; i++) {
switch (hw->mac.type) {
case ixgbe_mac_82598EB:
@@ -806,15 +804,11 @@ static void ixgbe_get_regs(struct net_device *netdev,
regs_buff[1096 + i] = IXGBE_READ_REG(hw, IXGBE_TIC_DW(i));
regs_buff[1100] = IXGBE_READ_REG(hw, IXGBE_TDPROBE);
regs_buff[1101] = IXGBE_READ_REG(hw, IXGBE_TXBUFCTRL);
- regs_buff[1102] = IXGBE_READ_REG(hw, IXGBE_TXBUFDATA0);
- regs_buff[1103] = IXGBE_READ_REG(hw, IXGBE_TXBUFDATA1);
- regs_buff[1104] = IXGBE_READ_REG(hw, IXGBE_TXBUFDATA2);
- regs_buff[1105] = IXGBE_READ_REG(hw, IXGBE_TXBUFDATA3);
+ for (i = 0; i < 4; i++)
+ regs_buff[1102 + i] = IXGBE_READ_REG(hw, IXGBE_TXBUFDATA(i));
regs_buff[1106] = IXGBE_READ_REG(hw, IXGBE_RXBUFCTRL);
- regs_buff[1107] = IXGBE_READ_REG(hw, IXGBE_RXBUFDATA0);
- regs_buff[1108] = IXGBE_READ_REG(hw, IXGBE_RXBUFDATA1);
- regs_buff[1109] = IXGBE_READ_REG(hw, IXGBE_RXBUFDATA2);
- regs_buff[1110] = IXGBE_READ_REG(hw, IXGBE_RXBUFDATA3);
+ for (i = 0; i < 4; i++)
+ regs_buff[1107 + i] = IXGBE_READ_REG(hw, IXGBE_RXBUFDATA(i));
for (i = 0; i < 8; i++)
regs_buff[1111 + i] = IXGBE_READ_REG(hw, IXGBE_PCIE_DIAG(i));
regs_buff[1119] = IXGBE_READ_REG(hw, IXGBE_RFVAL);
diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h b/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h
index 1912f7e..cd6c1dd 100644
--- a/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h
+++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_type.h
@@ -1060,15 +1060,9 @@ struct ixgbe_thermal_sensor_data {
#define IXGBE_TIC_DW2(_i) (0x082B0 + ((_i) * 4))
#define IXGBE_TDPROBE 0x07F20
#define IXGBE_TXBUFCTRL 0x0C600
-#define IXGBE_TXBUFDATA0 0x0C610
-#define IXGBE_TXBUFDATA1 0x0C614
-#define IXGBE_TXBUFDATA2 0x0C618
-#define IXGBE_TXBUFDATA3 0x0C61C
+#define IXGBE_TXBUFDATA(_i) (0x0C610 + ((_i) * 4)) /* 4 of these (0-3) */
#define IXGBE_RXBUFCTRL 0x03600
-#define IXGBE_RXBUFDATA0 0x03610
-#define IXGBE_RXBUFDATA1 0x03614
-#define IXGBE_RXBUFDATA2 0x03618
-#define IXGBE_RXBUFDATA3 0x0361C
+#define IXGBE_RXBUFDATA(_i) (0x03610 + ((_i) * 4)) /* 4 of these (0-3) */
#define IXGBE_PCIE_DIAG(_i) (0x11090 + ((_i) * 4)) /* 8 of these */
#define IXGBE_RFVAL 0x050A4
#define IXGBE_MDFTC1 0x042B8
--
2.5.5
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox