* [PATCH v13 10/16] Add a hook to intercept external buffers from NIC driver.
From: xiaohui.xin @ 2010-10-15 9:12 UTC (permalink / raw)
To: netdev, kvm, linux-kernel, mst, mingo, davem, herbert, jdike; +Cc: Xin Xiaohui
In-Reply-To: <fc6e95d63a2c62aaf77f8ded22fc43ccefcdbbff.1287132437.git.xiaohui.xin@intel.com>
From: Xin Xiaohui <xiaohui.xin@intel.com>
The hook is called in __netif_receive_skb().
Signed-off-by: Xin Xiaohui <xiaohui.xin@intel.com>
Signed-off-by: Zhao Yu <yzhao81new@gmail.com>
Reviewed-by: Jeff Dike <jdike@linux.intel.com>
---
net/core/dev.c | 37 +++++++++++++++++++++++++++++++++++++
1 files changed, 37 insertions(+), 0 deletions(-)
diff --git a/net/core/dev.c b/net/core/dev.c
index e48639d..235eaab 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -2814,6 +2814,37 @@ int __skb_bond_should_drop(struct sk_buff *skb, struct net_device *master)
}
EXPORT_SYMBOL(__skb_bond_should_drop);
+#if defined(CONFIG_MEDIATE_PASSTHRU) || defined(CONFIG_MEDIATE_PASSTHRU_MODULE)
+/* Add a hook to intercept mediate passthru(zero-copy) packets,
+ * and insert it to the socket queue owned by mp_port specially.
+ */
+static inline struct sk_buff *handle_mpassthru(struct sk_buff *skb,
+ struct packet_type **pt_prev,
+ int *ret,
+ struct net_device *orig_dev)
+{
+ struct mp_port *mp_port = NULL;
+ struct sock *sk = NULL;
+
+ if (!dev_is_mpassthru(skb->dev))
+ return skb;
+ mp_port = skb->dev->mp_port;
+
+ if (*pt_prev) {
+ *ret = deliver_skb(skb, *pt_prev, orig_dev);
+ *pt_prev = NULL;
+ }
+
+ sk = mp_port->sock->sk;
+ skb_queue_tail(&sk->sk_receive_queue, skb);
+ sk->sk_state_change(sk);
+
+ return NULL;
+}
+#else
+#define handle_mpassthru(skb, pt_prev, ret, orig_dev) (skb)
+#endif
+
static int __netif_receive_skb(struct sk_buff *skb)
{
struct packet_type *ptype, *pt_prev;
@@ -2891,6 +2922,11 @@ static int __netif_receive_skb(struct sk_buff *skb)
ncls:
#endif
+ /* To intercept mediate passthru(zero-copy) packets here */
+ skb = handle_mpassthru(skb, &pt_prev, &ret, orig_dev);
+ if (!skb)
+ goto out;
+
/* Handle special case of bridge or macvlan */
rx_handler = rcu_dereference(skb->dev->rx_handler);
if (rx_handler) {
@@ -2991,6 +3027,7 @@ err:
EXPORT_SYMBOL(netdev_mp_port_prep);
#endif
+
/**
* netif_receive_skb - process receive buffer from network
* @skb: buffer to process
--
1.7.3
^ permalink raw reply related
* [PATCH v13 04/16] Add a function make external buffer owner to query capability.
From: xiaohui.xin @ 2010-10-15 9:12 UTC (permalink / raw)
To: netdev, kvm, linux-kernel, mst, mingo, davem, herbert, jdike; +Cc: Xin Xiaohui
In-Reply-To: <fc6e95d63a2c62aaf77f8ded22fc43ccefcdbbff.1287132437.git.xiaohui.xin@intel.com>
From: Xin Xiaohui <xiaohui.xin@intel.com>
The external buffer owner can use the functions to get
the capability of the underlying NIC driver.
Signed-off-by: Xin Xiaohui <xiaohui.xin@intel.com>
Signed-off-by: Zhao Yu <yzhaonew@gmail.com>
Reviewed-by: Jeff Dike <jdike@linux.intel.com>
---
include/linux/netdevice.h | 2 +
net/core/dev.c | 49 +++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 51 insertions(+), 0 deletions(-)
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 575777f..8dcf6de 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -1736,6 +1736,8 @@ extern gro_result_t napi_frags_finish(struct napi_struct *napi,
gro_result_t ret);
extern struct sk_buff * napi_frags_skb(struct napi_struct *napi);
extern gro_result_t napi_gro_frags(struct napi_struct *napi);
+extern int netdev_mp_port_prep(struct net_device *dev,
+ struct mp_port *port);
static inline void napi_free_frags(struct napi_struct *napi)
{
diff --git a/net/core/dev.c b/net/core/dev.c
index 660dd41..e48639d 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -2942,6 +2942,55 @@ out:
return ret;
}
+/* To support meidate passthru(zero-copy) with NIC driver,
+ * we'd better query NIC driver for the capability it can
+ * provide, especially for packet split mode, now we only
+ * query for the header size, and the payload a descriptor
+ * may carry. If a driver does not use the API to export,
+ * then we may try to use a default value, currently,
+ * we use the default value from an IGB driver. Now,
+ * it's only called by mpassthru device.
+ */
+#if defined(CONFIG_MEDIATE_PASSTHRU) || defined(CONFIG_MEDIATE_PASSTHRU_MODULE)
+int netdev_mp_port_prep(struct net_device *dev,
+ struct mp_port *port)
+{
+ int rc;
+ int npages, data_len;
+ const struct net_device_ops *ops = dev->netdev_ops;
+
+ if (ops->ndo_mp_port_prep) {
+ rc = ops->ndo_mp_port_prep(dev, port);
+ if (rc)
+ return rc;
+ } else {
+ /* If the NIC driver did not report this,
+ * then we try to use default value.
+ */
+ port->hdr_len = 128;
+ port->data_len = 2048;
+ port->npages = 1;
+ }
+
+ if (port->hdr_len <= 0)
+ goto err;
+
+ npages = port->npages;
+ data_len = port->data_len;
+ if (npages <= 0 || npages > MAX_SKB_FRAGS ||
+ (data_len < PAGE_SIZE * (npages - 1) ||
+ data_len > PAGE_SIZE * npages))
+ goto err;
+
+ return 0;
+err:
+ dev_warn(&dev->dev, "invalid page constructor parameters\n");
+
+ return -EINVAL;
+}
+EXPORT_SYMBOL(netdev_mp_port_prep);
+#endif
+
/**
* netif_receive_skb - process receive buffer from network
* @skb: buffer to process
--
1.7.3
^ permalink raw reply related
* [PATCH v13 00/16] Provide a zero-copy method on KVM virtio-net.
From: xiaohui.xin @ 2010-10-15 9:12 UTC (permalink / raw)
To: netdev, kvm, linux-kernel, mst, mingo, davem, herbert, jdike
We provide an zero-copy method which driver side may get external
buffers to DMA. Here external means driver don't use kernel space
to allocate skb buffers. Currently the external buffer can be from
guest virtio-net driver.
The idea is simple, just to pin the guest VM user space and then
let host NIC driver has the chance to directly DMA to it.
The patches are based on vhost-net backend driver. We add a device
which provides proto_ops as sendmsg/recvmsg to vhost-net to
send/recv directly to/from the NIC driver. KVM guest who use the
vhost-net backend may bind any ethX interface in the host side to
get copyless data transfer thru guest virtio-net frontend.
patch 01-10: net core and kernel changes.
patch 11-13: new device as interface to mantpulate external buffers.
patch 14: for vhost-net.
patch 15: An example on modifying NIC driver to using napi_gro_frags().
patch 16: An example how to get guest buffers based on driver
who using napi_gro_frags().
The guest virtio-net driver submits multiple requests thru vhost-net
backend driver to the kernel. And the requests are queued and then
completed after corresponding actions in h/w are done.
For read, user space buffers are dispensed to NIC driver for rx when
a page constructor API is invoked. Means NICs can allocate user buffers
from a page constructor. We add a hook in netif_receive_skb() function
to intercept the incoming packets, and notify the zero-copy device.
For write, the zero-copy deivce may allocates a new host skb and puts
payload on the skb_shinfo(skb)->frags, and copied the header to skb->data.
The request remains pending until the skb is transmitted by h/w.
We provide multiple submits and asynchronous notifiicaton to
vhost-net too.
Our goal is to improve the bandwidth and reduce the CPU usage.
Exact performance data will be provided later.
What we have not done yet:
Performance tuning
what we have done in v1:
polish the RCU usage
deal with write logging in asynchroush mode in vhost
add notifier block for mp device
rename page_ctor to mp_port in netdevice.h to make it looks generic
add mp_dev_change_flags() for mp device to change NIC state
add CONIFG_VHOST_MPASSTHRU to limit the usage when module is not load
a small fix for missing dev_put when fail
using dynamic minor instead of static minor number
a __KERNEL__ protect to mp_get_sock()
what we have done in v2:
remove most of the RCU usage, since the ctor pointer is only
changed by BIND/UNBIND ioctl, and during that time, NIC will be
stopped to get good cleanup(all outstanding requests are finished),
so the ctor pointer cannot be raced into wrong situation.
Remove the struct vhost_notifier with struct kiocb.
Let vhost-net backend to alloc/free the kiocb and transfer them
via sendmsg/recvmsg.
use get_user_pages_fast() and set_page_dirty_lock() when read.
Add some comments for netdev_mp_port_prep() and handle_mpassthru().
what we have done in v3:
the async write logging is rewritten
a drafted synchronous write function for qemu live migration
a limit for locked pages from get_user_pages_fast() to prevent Dos
by using RLIMIT_MEMLOCK
what we have done in v4:
add iocb completion callback from vhost-net to queue iocb in mp device
replace vq->receiver by mp_sock_data_ready()
remove stuff in mp device which access structures from vhost-net
modify skb_reserve() to ignore host NIC driver reserved space
rebase to the latest vhost tree
split large patches into small pieces, especially for net core part.
what we have done in v5:
address Arnd Bergmann's comments
-remove IFF_MPASSTHRU_EXCL flag in mp device
-Add CONFIG_COMPAT macro
-remove mp_release ops
move dev_is_mpassthru() as inline func
fix a bug in memory relinquish
Apply to current git (2.6.34-rc6) tree.
what we have done in v6:
move create_iocb() out of page_dtor which may happen in interrupt context
-This remove the potential issues which lock called in interrupt context
make the cache used by mp, vhost as static, and created/destoryed during
modules init/exit functions.
-This makes multiple mp guest created at the same time.
what we have done in v7:
some cleanup prepared to suppprt PS mode
what we have done in v8:
discarding the modifications to point skb->data to guest buffer directly.
Add code to modify driver to support napi_gro_frags() with Herbert's comments.
To support PS mode.
Add mergeable buffer support in mp device.
Add GSO/GRO support in mp deice.
Address comments from Eric Dumazet about cache line and rcu usage.
what we have done in v9:
v8 patch is based on a fix in dev_gro_receive().
But Herbert did not agree with the fix we have sent out.
And he suggest another fix. v9 is modified to base on that fix.
what we have done in v10:
Fix a partial csum error.
Cleanup some unused fields with struct page_info{} in mp device.
Modify kmem_cache_zalloc() to kmem_cache_alloc() based on Michael S. Thirkin.
what we have done in v11:
Address comments from Michael S. Thirkin to add two new ioctls in mp device.
But still need to revise.
what we have done in v12:
Address most comments from Ben Hutchings, except the compat ioctls.
As the comments are sparse, so do not make a split patch.
Change struct mpassthru_port to struct mp_port, and struct page_ctor
to struct page_pool.
what we have done in v13:
Export functions to other drivers like macvtap, in case it want to reuse it to
get zero-copy.
Rebase on 2.6.36-rc7.
Performance:
We have seen the performance data request from mailling-list.
And we are now looking into this.
^ permalink raw reply
* Re: tbf/htb qdisc limitations
From: Jarek Poplawski @ 2010-10-15 8:18 UTC (permalink / raw)
To: Bill Fink; +Cc: Rick Jones, Steven Brudenell, netdev
In-Reply-To: <20101015023749.f085006b.billfink@mindspring.com>
On Fri, Oct 15, 2010 at 02:37:49AM -0400, Bill Fink wrote:
> On Thu, 14 Oct 2010, Jarek Poplawski wrote:
>
> > On Thu, Oct 14, 2010 at 08:09:39AM +0000, Jarek Poplawski wrote:
> > > On Thu, Oct 14, 2010 at 03:13:54AM -0400, Bill Fink wrote:
> > > > TSO/GSO was disabled and was using 9000-byte jumbo frames
> > > > (and specified mtu 9000 to tc command).
> > > >
> > > > Here was one attempt I made using tbf:
> > > >
> > > > tc qdisc add dev eth2 root handle 1: prio
> > > > tc qdisc add dev eth2 parent 1:1 handle 10: tbf rate 8900mbit buffer 1112500 limit 10000 mtu 9000
> > > > tc filter add dev eth2 protocol ip parent 1: prio 1 u32 match ip dst 192.168.1.23 flowid 10:1
> > > >
> > > > I tried many variations of the above, all without success.
> > >
> > > The main problem are smaller packets. If you had (almost) only 9000b
> > > frames this probably could work. [...]
> >
> > On the other hand, e.g. the limit above seems too low wrt mtu & rate.
>
> Actually, I discovered my commands above work just fine on
> a 2.6.35 box:
>
> i7test7% nuttcp -T10 -i1 192.168.1.17
> 1045.3125 MB / 1.00 sec = 8768.3573 Mbps 0 retrans
> 1045.6875 MB / 1.00 sec = 8772.0292 Mbps 0 retrans
> 1049.5625 MB / 1.00 sec = 8804.2627 Mbps 0 retrans
> 1043.1875 MB / 1.00 sec = 8750.9960 Mbps 0 retrans
> 1048.6875 MB / 1.00 sec = 8796.3246 Mbps 0 retrans
> 1033.4375 MB / 1.00 sec = 8669.3188 Mbps 0 retrans
> 1040.7500 MB / 1.00 sec = 8730.7057 Mbps 0 retrans
> 1047.0000 MB / 1.00 sec = 8783.2063 Mbps 0 retrans
> 1040.0000 MB / 1.00 sec = 8724.0564 Mbps 0 retrans
> 1037.4375 MB / 1.00 sec = 8702.5434 Mbps 0 retrans
>
> 10431.5608 MB / 10.00 sec = 8749.7542 Mbps 25 %TX 35 %RX 0 retrans 0.11 msRTT
>
> The problems I encountered were on a field system running
> 2.6.30.10. I will investigate upgrading the field system
> to 2.6.35.
This change from 2.6.31 should matter here:
http://git.kernel.org/?p=linux/kernel/git/stable/linux-2.6.35.y.git;a=commit;h=a4a710c4a7490587406462bf1d54504b7783d7d7
Jarek P.
^ permalink raw reply
* Re: -j MARK in raw vs. mangle (was Re: xfrm by MARK: tcp problems when mark for in and out differ)
From: Gerd v. Egidy @ 2010-10-15 8:05 UTC (permalink / raw)
To: Patrick McHardy; +Cc: hadi, netfilter-devel, netdev
In-Reply-To: <4CB7FCEB.5070804@trash.net>
Hi Patrick,
> > So it seems this has nothing to do with xfrm, but that the MARK target
> > has different effects when used in raw than in mangle. I was using raw
> > because I had to set conntrack zones too and it was more conveniant to
> > do both in one place.
> >
> > Can one of the netfilter guys comment on this? Is using MARK in raw not
> > fully supported or has known deficiencies?
>
> No, the problem is most likely that for outgoing packets, the XFRM
> lookup is done with the route lookup before the packet is even sent,
> so once it hits the raw or mangle table, it is too late. mangle however
> performs rerouting when the mark value changes, at which point a new
> XFRM lookup is performed.
ah, this would explain it. Thanks for the explanation. I'll just stick with
mangle for marking.
Kind regards,
Gerd
--
Address (better: trap) for people I really don't want to get mail from:
jonas@cactusamerica.com
^ permalink raw reply
* [PATCH 1/1] ARC vmac ethernet driver.
From: Andreas Fenkart @ 2010-10-15 7:54 UTC (permalink / raw)
To: netdev; +Cc: Andreas Fenkart
In-Reply-To: <1287129254-18078-1-git-send-email-andreas.fenkart@streamunlimited.com>
Signed-off-by: Andreas Fenkart <andreas.fenkart@streamunlimited.com>
---
drivers/net/Kconfig | 9 +
drivers/net/Makefile | 1 +
drivers/net/arcvmac.c | 1411 +++++++++++++++++++++++++++++++++++++++++++++++++
drivers/net/arcvmac.h | 268 ++++++++++
4 files changed, 1689 insertions(+), 0 deletions(-)
diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 5db667c..f534587 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -248,6 +248,15 @@ config AX88796_93CX6
help
Select this if your platform comes with an external 93CX6 eeprom.
+config ARCVMAC
+ tristate "ARC VMAC ethernet support"
+ depends on HAS_DMA
+ select MII
+ select PHYLIB
+ select CRC32
+ help
+ MAC device present on Zoran Quatro43XX
+
config MACE
tristate "MACE (Power Mac ethernet) support"
depends on PPC_PMAC && PPC32
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index 1d05ea5..da41896 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -137,6 +137,7 @@ obj-$(CONFIG_ULTRA32) += smc-ultra32.o 8390.o
obj-$(CONFIG_E2100) += e2100.o 8390.o
obj-$(CONFIG_ES3210) += es3210.o 8390.o
obj-$(CONFIG_LNE390) += lne390.o 8390.o
+obj-$(CONFIG_ARCVMAC) += arcvmac.o
obj-$(CONFIG_NE3210) += ne3210.o 8390.o
obj-$(CONFIG_SB1250_MAC) += sb1250-mac.o
obj-$(CONFIG_B44) += b44.o
diff --git a/drivers/net/arcvmac.c b/drivers/net/arcvmac.c
new file mode 100644
index 0000000..e49e1c1
--- /dev/null
+++ b/drivers/net/arcvmac.c
@@ -0,0 +1,1411 @@
+/*
+ * linux/arch/arc/drivers/arcvmac.c
+ *
+ * Copyright (C) 2003-2006 Codito Technologies, for linux-2.4 port
+ * Copyright (C) 2006-2007 Celunite Inc, for linux-2.6 port
+ * Copyright (C) 2007-2008 Sagem Communications, Fehmi HAFSI
+ * Copyright (C) 2009 Sagem Communications, Andreas Fenkart
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ *
+ * external PHY support based on dnet.c
+ * ring management based on bcm63xx_enet.c
+ *
+ * Authors: amit.bhor@celunite.com, sameer.dhavale@celunite.com
+ */
+
+#undef DEBUG
+
+#include <linux/clk.h>
+#include <linux/crc32.h>
+#include <linux/delay.h>
+#include <linux/dma-mapping.h>
+#include <linux/etherdevice.h>
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/moduleparam.h>
+#include <linux/netdevice.h>
+#include <linux/phy.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+#include <linux/types.h>
+
+#include "arcvmac.h"
+
+/* Register access macros */
+#define vmac_writel(port, value, reg) \
+ writel((value), (port)->regs + reg##_OFFSET)
+#define vmac_readl(port, reg) readl((port)->regs + reg##_OFFSET)
+
+static unsigned char *read_mac_reg(struct net_device *dev,
+ unsigned char hwaddr[ETH_ALEN])
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ unsigned mac_lo, mac_hi;
+
+ WARN_ON(!hwaddr);
+ mac_lo = vmac_readl(ap, ADDRL);
+ mac_hi = vmac_readl(ap, ADDRH);
+
+ hwaddr[0] = (mac_lo >> 0) & 0xff;
+ hwaddr[1] = (mac_lo >> 8) & 0xff;
+ hwaddr[2] = (mac_lo >> 16) & 0xff;
+ hwaddr[3] = (mac_lo >> 24) & 0xff;
+ hwaddr[4] = (mac_hi >> 0) & 0xff;
+ hwaddr[5] = (mac_hi >> 8) & 0xff;
+ return hwaddr;
+}
+
+static void write_mac_reg(struct net_device *dev, unsigned char* hwaddr)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ unsigned mac_lo, mac_hi;
+
+ mac_lo = hwaddr[3] << 24 | hwaddr[2] << 16 | hwaddr[1] << 8 | hwaddr[0];
+ mac_hi = hwaddr[5] << 8 | hwaddr[4];
+
+ vmac_writel(ap, mac_lo, ADDRL);
+ vmac_writel(ap, mac_hi, ADDRH);
+}
+
+static void vmac_mdio_xmit(struct vmac_priv *ap, unsigned val)
+{
+ init_completion(&ap->mdio_complete);
+ vmac_writel(ap, val, MDIO_DATA);
+ wait_for_completion(&ap->mdio_complete);
+}
+
+static int vmac_mdio_read(struct mii_bus *bus, int phy_id, int phy_reg)
+{
+ struct vmac_priv *vmac = bus->priv;
+ unsigned int val;
+ /* only 5 bits allowed for phy-addr and reg_offset */
+ WARN_ON(phy_id & ~0x1f || phy_reg & ~0x1f);
+
+ val = MDIO_BASE | MDIO_OP_READ;
+ val |= phy_id << 23 | phy_reg << 18;
+ vmac_mdio_xmit(vmac, val);
+
+ val = vmac_readl(vmac, MDIO_DATA);
+ return val & MDIO_DATA_MASK;
+}
+
+static int vmac_mdio_write(struct mii_bus *bus, int phy_id, int phy_reg,
+ u16 value)
+{
+ struct vmac_priv *vmac = bus->priv;
+ unsigned int val;
+ /* only 5 bits allowed for phy-addr and reg_offset */
+ WARN_ON(phy_id & ~0x1f || phy_reg & ~0x1f);
+
+ val = MDIO_BASE | MDIO_OP_WRITE;
+ val |= phy_id << 23 | phy_reg << 18;
+ val |= (value & MDIO_DATA_MASK);
+ vmac_mdio_xmit(vmac, val);
+ return 0;
+}
+
+static void vmac_handle_link_change(struct net_device *dev)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ struct phy_device *phydev = ap->phy_dev;
+ unsigned long flags;
+ int report_change = 0;
+
+ spin_lock_irqsave(&ap->lock, flags);
+
+ if (phydev->duplex != ap->duplex) {
+ unsigned tmp;
+
+ tmp = vmac_readl(ap, ENABLE);
+
+ if (phydev->duplex)
+ tmp |= ENFL_MASK;
+ else
+ tmp &= ~ENFL_MASK;
+
+ vmac_writel(ap, tmp, ENABLE);
+
+ ap->duplex = phydev->duplex;
+ report_change = 1;
+ }
+
+ if (phydev->speed != ap->speed) {
+ ap->speed = phydev->speed;
+ report_change = 1;
+ }
+
+ if (phydev->link != ap->link) {
+ ap->link = phydev->link;
+ report_change = 1;
+ }
+
+ spin_unlock_irqrestore(&ap->lock, flags);
+
+ if (report_change)
+ phy_print_status(ap->phy_dev);
+}
+
+static int __devinit vmac_mii_probe(struct net_device *dev)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ struct phy_device *phydev = NULL;
+ struct clk *sys_clk;
+ unsigned long clock_rate;
+ int phy_addr, err;
+
+ /* find the first phy */
+ for (phy_addr = 0; phy_addr < PHY_MAX_ADDR; phy_addr++) {
+ if (ap->mii_bus->phy_map[phy_addr]) {
+ phydev = ap->mii_bus->phy_map[phy_addr];
+ break;
+ }
+ }
+
+ if (!phydev) {
+ dev_err(&dev->dev, "no PHY found\n");
+ return -ENODEV;
+ }
+
+ /* add pin_irq, if avail */
+
+ phydev = phy_connect(dev, dev_name(&phydev->dev),
+ &vmac_handle_link_change, 0,
+ PHY_INTERFACE_MODE_MII);
+
+ if (IS_ERR(phydev)) {
+ err = PTR_ERR(phydev);
+ dev_err(&dev->dev, "could not attach to PHY %d\n", err);
+ goto err_out;
+ }
+
+ phydev->supported &= PHY_BASIC_FEATURES;
+ phydev->supported |= SUPPORTED_Asym_Pause | SUPPORTED_Pause;
+
+ sys_clk = clk_get(&ap->pdev->dev, "arcvmac");
+ if (IS_ERR(sys_clk)) {
+ err = PTR_ERR(sys_clk);
+ goto err_disconnect;
+ }
+
+ clock_rate = clk_get_rate(sys_clk);
+ clk_put(sys_clk);
+
+ dev_dbg(&ap->pdev->dev, "clk_get: dev_name : %s %lu\n",
+ dev_name(&ap->pdev->dev),
+ clock_rate);
+
+ if (clock_rate < 25000000)
+ phydev->supported &= ~(SUPPORTED_100baseT_Half |
+ SUPPORTED_100baseT_Full);
+
+ phydev->advertising = phydev->supported;
+
+ ap->link = 0;
+ ap->speed = 0;
+ ap->duplex = -1;
+ ap->phy_dev = phydev;
+
+ return 0;
+
+err_disconnect:
+ phy_disconnect(phydev);
+err_out:
+ return err;
+}
+
+static int __devinit vmac_mii_init(struct vmac_priv *ap)
+{
+ int err, i;
+
+ ap->mii_bus = mdiobus_alloc();
+ if (ap->mii_bus == NULL)
+ return -ENOMEM;
+
+ ap->mii_bus->name = "vmac_mii_bus";
+ ap->mii_bus->read = &vmac_mdio_read;
+ ap->mii_bus->write = &vmac_mdio_write;
+
+ snprintf(ap->mii_bus->id, MII_BUS_ID_SIZE, "%x", 0);
+
+ ap->mii_bus->priv = ap;
+
+ err = -ENOMEM;
+ ap->mii_bus->irq = kmalloc(sizeof(int) * PHY_MAX_ADDR, GFP_KERNEL);
+ if (!ap->mii_bus->irq)
+ goto err_out;
+
+ for (i = 0; i < PHY_MAX_ADDR; i++)
+ ap->mii_bus->irq[i] = PHY_POLL;
+
+#if 0
+ /* FIXME: what is it used for? */
+ platform_set_drvdata(ap->dev, ap->mii_bus);
+#endif
+
+ err = mdiobus_register(ap->mii_bus);
+ if (err)
+ goto err_out_free_mdio_irq;
+
+ err = vmac_mii_probe(ap->dev);
+ if (err)
+ goto err_out_unregister_bus;
+
+ return 0;
+
+err_out_unregister_bus:
+ mdiobus_unregister(ap->mii_bus);
+err_out_free_mdio_irq:
+ kfree(ap->mii_bus->irq);
+err_out:
+ mdiobus_free(ap->mii_bus);
+ return err;
+}
+
+static void vmac_mii_exit(struct net_device *dev)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+
+ if (ap->phy_dev)
+ phy_disconnect(ap->phy_dev);
+
+ mdiobus_unregister(ap->mii_bus);
+ kfree(ap->mii_bus->irq);
+ mdiobus_free(ap->mii_bus);
+}
+
+static int vmacether_get_settings(struct net_device *dev,
+ struct ethtool_cmd *cmd)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ struct phy_device *phydev = ap->phy_dev;
+
+ if (!phydev)
+ return -ENODEV;
+
+ return phy_ethtool_gset(phydev, cmd);
+}
+
+static int vmacether_set_settings(struct net_device *dev,
+ struct ethtool_cmd *cmd)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ struct phy_device *phydev = ap->phy_dev;
+
+ if (!phydev)
+ return -ENODEV;
+
+ return phy_ethtool_sset(phydev, cmd);
+}
+
+static int vmac_ioctl(struct net_device *dev, struct ifreq *rq, int cmd)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ struct phy_device *phydev = ap->phy_dev;
+
+ if (!netif_running(dev))
+ return -EINVAL;
+
+ if (!phydev)
+ return -ENODEV;
+
+ return phy_mii_ioctl(phydev, rq, cmd);
+}
+
+static void vmacether_get_drvinfo(struct net_device *dev,
+ struct ethtool_drvinfo *info)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+
+ strlcpy(info->driver, DRV_NAME, sizeof(info->driver));
+ strlcpy(info->version, DRV_VERSION, sizeof(info->version));
+ snprintf(info->bus_info, sizeof(info->bus_info),
+ "platform 0x%x", ap->mem_base);
+}
+
+static int update_error_counters(struct net_device *dev, int status)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ dev_dbg(&ap->pdev->dev, "rx error counter overrun. status = 0x%x\n",
+ status);
+
+ /* programming error */
+ WARN_ON(status & TXCH_MASK);
+ WARN_ON(!(status & (MSER_MASK | RXCR_MASK | RXFR_MASK | RXFL_MASK)));
+
+ if (status & MSER_MASK)
+ ap->stats.rx_over_errors += 256; /* ran out of BD */
+ if (status & RXCR_MASK)
+ ap->stats.rx_crc_errors += 256;
+ if (status & RXFR_MASK)
+ ap->stats.rx_frame_errors += 256;
+ if (status & RXFL_MASK)
+ ap->stats.rx_fifo_errors += 256;
+
+ return 0;
+}
+
+static void update_tx_errors(struct net_device *dev, int status)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+
+ if (status & UFLO)
+ ap->stats.tx_fifo_errors++;
+
+ if (ap->duplex)
+ return;
+
+ /* half duplex flags */
+ if (status & LTCL)
+ ap->stats.tx_window_errors++;
+ if (status & RETRY_CT)
+ ap->stats.collisions += (status & RETRY_CT) >> 24;
+ if (status & DROP) /* too many retries */
+ ap->stats.tx_aborted_errors++;
+ if (status & DEFER)
+ dev_vdbg(&ap->pdev->dev, "\"defer to traffic\"\n");
+ if (status & CARLOSS)
+ ap->stats.tx_carrier_errors++;
+}
+
+static int vmac_rx_reclaim_force(struct net_device *dev)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ int ct;
+
+ ct = 0;
+
+ dev_dbg(&ap->pdev->dev, "%s need to release %d rx sk_buff\n",
+ __func__, fifo_used(&ap->rx_ring));
+
+ while (!fifo_empty(&ap->rx_ring) && ct++ < ap->rx_ring.size) {
+ struct vmac_buffer_desc *desc;
+ struct sk_buff *skb;
+ int desc_idx;
+
+ desc_idx = ap->rx_ring.tail;
+ desc = &ap->rxbd[desc_idx];
+ fifo_inc_tail(&ap->rx_ring);
+
+ if (!ap->rx_skbuff[desc_idx]) {
+ dev_err(&ap->pdev->dev, "non-populated rx_skbuff found %d\n",
+ desc_idx);
+ continue;
+ }
+
+ skb = ap->rx_skbuff[desc_idx];
+ ap->rx_skbuff[desc_idx] = NULL;
+
+ dma_unmap_single(&ap->pdev->dev, desc->data, skb->len,
+ DMA_TO_DEVICE);
+
+ dev_kfree_skb(skb);
+ }
+
+ if (!fifo_empty(&ap->rx_ring)) {
+ dev_err(&ap->pdev->dev, "failed to reclaim %d rx sk_buff\n",
+ fifo_used(&ap->rx_ring));
+ }
+
+ return 0;
+}
+
+static int vmac_rx_refill(struct net_device *dev)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+
+ WARN_ON(fifo_full(&ap->rx_ring));
+
+ while (!fifo_full(&ap->rx_ring)) {
+ struct vmac_buffer_desc *desc;
+ struct sk_buff *skb;
+ dma_addr_t p;
+ int desc_idx;
+
+ desc_idx = ap->rx_ring.head;
+ desc = &ap->rxbd[desc_idx];
+
+ /* make sure we read the actual descriptor status */
+ rmb();
+
+ if (ap->rx_skbuff[desc_idx]) {
+ /* dropped packet / buffer chaining */
+ fifo_inc_head(&ap->rx_ring);
+
+ /* return to DMA */
+ wmb();
+ desc->info = OWN_MASK | ap->rx_skb_size;
+ continue;
+ }
+
+ skb = netdev_alloc_skb(dev, ap->rx_skb_size + 2);
+ if (!skb) {
+ dev_info(&ap->pdev->dev, "failed to allocate rx_skb, skb's left %d\n",
+ fifo_used(&ap->rx_ring));
+ break;
+ }
+
+ /* IP header Alignment (14 byte Ethernet header) */
+ skb_reserve(skb, 2);
+ WARN_ON(skb->len != 0); /* nothing received yet */
+
+ ap->rx_skbuff[desc_idx] = skb;
+
+ p = dma_map_single(&ap->pdev->dev, skb->data, ap->rx_skb_size,
+ DMA_FROM_DEVICE);
+
+ desc->data = p;
+
+ wmb();
+ desc->info = OWN_MASK | ap->rx_skb_size;
+
+ fifo_inc_head(&ap->rx_ring);
+ }
+
+ /* If rx ring is still empty, set a timer to try allocating
+ * again at a later time. */
+ if (fifo_empty(&ap->rx_ring) && netif_running(dev)) {
+ dev_warn(&ap->pdev->dev, "unable to refill rx ring\n");
+ ap->rx_timeout.expires = jiffies + HZ;
+ add_timer(&ap->rx_timeout);
+ }
+
+ return 0;
+}
+
+/*
+ * timer callback to defer refill rx queue in case we're OOM
+ */
+static void vmac_refill_rx_timer(unsigned long data)
+{
+ struct net_device *dev;
+ struct vmac_priv *ap;
+
+ dev = (struct net_device *)data;
+ ap = netdev_priv(dev);
+
+ spin_lock(&ap->rx_lock);
+ vmac_rx_refill(dev);
+ spin_unlock(&ap->rx_lock);
+}
+
+/* merge buffer chaining */
+struct sk_buff *vmac_merge_rx_buffers(struct net_device *dev,
+ struct vmac_buffer_desc *after,
+ int pkt_len) /* data */
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ struct sk_buff *merge_skb, *cur_skb;
+ struct dma_fifo *rx_ring;
+ struct vmac_buffer_desc *desc;
+
+ rx_ring = &ap->rx_ring;
+ desc = &ap->rxbd[rx_ring->tail];
+
+ WARN_ON(desc == after);
+
+ /* strip FCS */
+ pkt_len -= 4;
+
+ /* IP header Alignment (14 byte Ethernet header) */
+ merge_skb = netdev_alloc_skb(dev, pkt_len + 2);
+ if (!merge_skb) {
+ dev_err(&ap->pdev->dev, "failed to allocate merged rx_skb, rx skb's left %d\n",
+ fifo_used(rx_ring));
+
+ return NULL;
+ }
+
+ skb_reserve(merge_skb, 2);
+
+ while (desc != after && pkt_len) {
+ struct vmac_buffer_desc *desc;
+ int buf_len, valid;
+
+ /* desc needs wrapping */
+ desc = &ap->rxbd[rx_ring->tail];
+ cur_skb = ap->rx_skbuff[rx_ring->tail];
+ WARN_ON(!cur_skb);
+
+ dma_unmap_single(&ap->pdev->dev, desc->data, ap->rx_skb_size,
+ DMA_FROM_DEVICE);
+
+ /* do not copy FCS */
+ buf_len = desc->info & LEN_MASK;
+ valid = min(pkt_len, buf_len);
+ pkt_len -= valid;
+
+ memcpy(skb_put(merge_skb, valid), cur_skb->data, valid);
+
+ fifo_inc_tail(rx_ring);
+ }
+
+ /* merging_pressure++ */
+
+ if (unlikely(pkt_len != 0))
+ dev_err(&ap->pdev->dev, "buffer chaining bytes missing %d\n",
+ pkt_len);
+
+ WARN_ON(desc != after);
+
+ return merge_skb;
+}
+
+int vmac_rx_receive(struct net_device *dev, int budget)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ struct vmac_buffer_desc *first;
+ int processed, pkt_len, pkt_err;
+ struct dma_fifo lookahead;
+
+ processed = 0;
+
+ first = NULL;
+ pkt_err = pkt_len = 0;
+
+ /* look ahead, till packet complete */
+ lookahead = ap->rx_ring;
+
+ do {
+ struct vmac_buffer_desc *desc; /* cur_ */
+ int desc_idx; /* cur_ */
+ struct sk_buff *skb; /* pkt_ */
+
+ desc_idx = lookahead.tail;
+ desc = &ap->rxbd[desc_idx];
+
+ /* make sure we read the actual descriptor status */
+ rmb();
+
+ /* break if dma ownership belongs to hw */
+ if (desc->info & OWN_MASK) {
+ ap->mac_rxring_head = vmac_readl(ap, MAC_RXRING_HEAD);
+ break;
+ }
+
+ if (desc->info & FRST_MASK) {
+ pkt_len = 0;
+ pkt_err = 0;
+
+ /* don't free current */
+ ap->rx_ring.tail = lookahead.tail;
+ first = desc;
+ }
+
+ fifo_inc_tail(&lookahead);
+
+ /* check bd */
+
+ pkt_len += desc->info & LEN_MASK;
+ pkt_err |= (desc->info & BUFF);
+
+ if (!(desc->info & LAST_MASK))
+ continue;
+
+ /* received complete packet */
+
+ if (unlikely(pkt_err || !first)) {
+ /* recycle buffers */
+ ap->rx_ring.tail = lookahead.tail;
+ continue;
+ }
+
+ WARN_ON(!(first->info & FRST_MASK) ||
+ !(desc->info & LAST_MASK));
+ WARN_ON(pkt_err);
+
+ /* -- valid packet -- */
+
+ if (first != desc) {
+ skb = vmac_merge_rx_buffers(dev, desc, pkt_len);
+
+ if (!skb) {
+ /* kill packet */
+ ap->rx_ring.tail = lookahead.tail;
+ ap->rx_merge_error++;
+ continue;
+ }
+ } else {
+ dma_unmap_single(&ap->pdev->dev, desc->data,
+ ap->rx_skb_size, DMA_FROM_DEVICE);
+
+ skb = ap->rx_skbuff[desc_idx];
+ ap->rx_skbuff[desc_idx] = NULL;
+ /* desc->data != skb->data => desc->data DMA mapped */
+
+ /* strip FCS */
+ skb_put(skb, pkt_len - 4);
+ }
+
+ /* free buffers */
+ ap->rx_ring.tail = lookahead.tail;
+
+ WARN_ON(skb->len != pkt_len - 4);
+ processed++;
+ skb->dev = dev;
+ skb->protocol = eth_type_trans(skb, dev);
+ ap->stats.rx_packets++;
+ ap->stats.rx_bytes += skb->len;
+ dev->last_rx = jiffies;
+ netif_rx(skb);
+
+ } while (!fifo_empty(&lookahead) && (processed < budget));
+
+ dev_vdbg(&ap->pdev->dev, "processed pkt %d, remaining rx buff %d\n",
+ processed,
+ fifo_used(&ap->rx_ring));
+
+ if (processed || fifo_empty(&ap->rx_ring))
+ vmac_rx_refill(dev);
+
+ return processed;
+}
+
+static void vmac_toggle_irqmask(struct net_device *dev, int enable, int mask)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ unsigned long tmp;
+
+ tmp = vmac_readl(ap, ENABLE);
+ if (enable)
+ tmp |= mask;
+ else
+ tmp &= ~mask;
+ vmac_writel(ap, tmp, ENABLE);
+}
+
+static void vmac_toggle_txint(struct net_device *dev, int enable)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ unsigned long flags;
+
+ spin_lock_irqsave(&ap->lock, flags);
+ vmac_toggle_irqmask(dev, enable, TXINT_MASK);
+ spin_unlock_irqrestore(&ap->lock, flags);
+}
+
+static void vmac_toggle_rxint(struct net_device *dev, int enable)
+{
+ vmac_toggle_irqmask(dev, enable, RXINT_MASK);
+}
+
+static int vmac_poll(struct napi_struct *napi, int budget)
+{
+ struct vmac_priv *ap;
+ struct net_device *dev;
+ int rx_work_done;
+ unsigned long flags;
+
+ ap = container_of(napi, struct vmac_priv, napi);
+ dev = ap->dev;
+
+ /* ack interrupt */
+ vmac_writel(ap, RXINT_MASK, STAT);
+
+ spin_lock(&ap->rx_lock);
+ rx_work_done = vmac_rx_receive(dev, budget);
+ spin_unlock(&ap->rx_lock);
+
+#ifdef VERBOSE_DEBUG
+ if (printk_ratelimit()) {
+ dev_vdbg(&ap->pdev->dev, "poll budget %d receive rx_work_done %d\n",
+ budget,
+ rx_work_done);
+ }
+#endif
+
+ if (rx_work_done >= budget) {
+ /* rx queue is not yet empty/clean */
+ return rx_work_done;
+ }
+
+ /* no more packet in rx/tx queue, remove device from poll
+ * queue */
+ spin_lock_irqsave(&ap->lock, flags);
+ napi_complete(napi);
+ vmac_toggle_rxint(dev, 1);
+ spin_unlock_irqrestore(&ap->lock, flags);
+
+ return rx_work_done;
+}
+
+static int vmac_tx_reclaim(struct net_device *dev, int force);
+
+static irqreturn_t vmac_intr(int irq, void *dev_instance)
+{
+ struct net_device *dev = dev_instance;
+ struct vmac_priv *ap = netdev_priv(dev);
+ unsigned int status;
+
+ spin_lock(&ap->lock);
+
+ status = vmac_readl(ap, STAT);
+ vmac_writel(ap, status, STAT);
+
+#ifdef DEBUG
+ if (unlikely(ap->shutdown))
+ dev_err(&ap->pdev->dev, "ISR during close\n");
+
+ if (unlikely(!status & (RXINT_MASK|MDIO_MASK|ERR_MASK)))
+ dev_err(&ap->pdev->dev, "No source of IRQ found\n");
+#endif
+
+ if ((status & RXINT_MASK) &&
+ (ap->mac_rxring_head !=
+ vmac_readl(ap, MAC_RXRING_HEAD))) {
+ vmac_toggle_rxint(dev, 0);
+ napi_schedule(&ap->napi);
+ }
+
+ if (unlikely(netif_queue_stopped(dev) && (status & TXINT_MASK)))
+ vmac_tx_reclaim(dev, 0);
+
+ if (status & MDIO_MASK)
+ complete(&ap->mdio_complete);
+
+ if (unlikely(status & ERR_MASK))
+ update_error_counters(dev, status);
+
+ spin_unlock(&ap->lock);
+
+ return IRQ_HANDLED;
+}
+
+static int vmac_tx_reclaim(struct net_device *dev, int force)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ int released = 0;
+
+ /* buffer chaining not used, see vmac_start_xmit */
+
+ while (!fifo_empty(&ap->tx_ring)) {
+ struct vmac_buffer_desc *desc;
+ struct sk_buff *skb;
+ int desc_idx;
+
+ desc_idx = ap->tx_ring.tail;
+ desc = &ap->txbd[desc_idx];
+
+ /* ensure other field of the descriptor were not read
+ * before we checked ownership */
+ rmb();
+
+ if ((desc->info & OWN_MASK) && !force)
+ break;
+
+ if (desc->info & ERR_MSK_TX) {
+ update_tx_errors(dev, desc->info);
+ /* recycle packet, let upper level deal with it */
+ }
+
+ skb = ap->tx_skbuff[desc_idx];
+ ap->tx_skbuff[desc_idx] = NULL;
+ WARN_ON(!skb);
+
+ dma_unmap_single(&ap->pdev->dev, desc->data, skb->len,
+ DMA_TO_DEVICE);
+
+ dev_kfree_skb_any(skb);
+
+ released++;
+ fifo_inc_tail(&ap->tx_ring);
+ }
+
+ if (netif_queue_stopped(dev) && released) {
+ netif_wake_queue(dev);
+ vmac_toggle_txint(dev, 0);
+ }
+
+ if (unlikely(force && !fifo_empty(&ap->tx_ring))) {
+ dev_err(&ap->pdev->dev, "failed to reclaim %d tx sk_buff\n",
+ fifo_used(&ap->tx_ring));
+ }
+
+ return released;
+}
+
+int vmac_start_xmit(struct sk_buff *skb, struct net_device *dev)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ struct vmac_buffer_desc *desc;
+ unsigned int tmp;
+
+ /* running under xmit lock */
+
+ /* no scatter/gatter see features below */
+ WARN_ON(skb_shinfo(skb)->nr_frags != 0);
+ WARN_ON(skb->len > MAX_TX_BUFFER_LEN);
+
+ if (unlikely(fifo_full(&ap->tx_ring))) {
+ netif_stop_queue(dev);
+ vmac_toggle_txint(dev, 1);
+ dev_err(&ap->pdev->dev, "xmit called with no tx desc available\n");
+ return NETDEV_TX_BUSY;
+ }
+
+ if (unlikely(skb->len < ETH_ZLEN)) {
+ struct sk_buff *short_skb;
+ short_skb = netdev_alloc_skb(dev, ETH_ZLEN);
+ if (!short_skb)
+ return NETDEV_TX_LOCKED;
+
+ memset(short_skb->data, 0, ETH_ZLEN);
+ memcpy(skb_put(short_skb, ETH_ZLEN), skb->data, skb->len);
+ dev_kfree_skb(skb);
+ skb = short_skb;
+ }
+
+ /* fill descriptor */
+ ap->tx_skbuff[ap->tx_ring.head] = skb;
+
+ desc = &ap->txbd[ap->tx_ring.head];
+ desc->data = dma_map_single(&ap->pdev->dev, skb->data, skb->len,
+ DMA_TO_DEVICE);
+
+ /* dma might already be polling */
+ wmb();
+ desc->info = OWN_MASK | FRST_MASK | LAST_MASK | skb->len;
+ wmb();
+
+ /* kick tx dma */
+ tmp = vmac_readl(ap, STAT);
+ vmac_writel(ap, tmp | TXPL_MASK, STAT);
+
+ ap->stats.tx_packets++;
+ ap->stats.tx_bytes += skb->len;
+ dev->trans_start = jiffies;
+ fifo_inc_head(&ap->tx_ring);
+
+ /* vmac_tx_reclaim independent of vmac_tx_timeout */
+ if (fifo_used(&ap->tx_ring) > 8)
+ vmac_tx_reclaim(dev, 0);
+
+ /* stop queue if no more desc available */
+ if (fifo_full(&ap->tx_ring)) {
+ netif_stop_queue(dev);
+ vmac_toggle_txint(dev, 1);
+ }
+
+ return NETDEV_TX_OK;
+}
+
+static int alloc_buffers(struct net_device *dev)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ int err = -ENOMEM;
+ int size;
+
+ fifo_init(&ap->rx_ring, RX_BDT_LEN);
+ fifo_init(&ap->tx_ring, TX_BDT_LEN);
+
+ /* initialize skb list */
+ memset(ap->rx_skbuff, 0, sizeof(ap->rx_skbuff));
+ memset(ap->tx_skbuff, 0, sizeof(ap->tx_skbuff));
+
+ /* allocate DMA received descriptors */
+ size = sizeof(*ap->rxbd) * ap->rx_ring.size;
+ ap->rxbd = dma_alloc_coherent(&ap->pdev->dev, size,
+ &ap->rxbd_dma,
+ GFP_KERNEL);
+ if (ap->rxbd == NULL)
+ goto err_out;
+
+ /* allocate DMA transmit descriptors */
+ size = sizeof(*ap->txbd) * ap->tx_ring.size;
+ ap->txbd = dma_alloc_coherent(&ap->pdev->dev, size,
+ &ap->txbd_dma,
+ GFP_KERNEL);
+ if (ap->txbd == NULL)
+ goto err_free_rxbd;
+
+ /* ensure 8-byte aligned */
+ WARN_ON(((int)ap->txbd & 0x7) || ((int)ap->rxbd & 0x7));
+
+ memset(ap->txbd, 0, sizeof(*ap->txbd) * ap->tx_ring.size);
+ memset(ap->rxbd, 0, sizeof(*ap->rxbd) * ap->rx_ring.size);
+
+ /* allocate rx skb */
+ err = vmac_rx_refill(dev);
+ if (err)
+ goto err_free_txbd;
+
+ return 0;
+
+err_free_txbd:
+ dma_free_coherent(&ap->pdev->dev, sizeof(*ap->txbd) * ap->tx_ring.size,
+ ap->txbd, ap->txbd_dma);
+err_free_rxbd:
+ dma_free_coherent(&ap->pdev->dev, sizeof(*ap->rxbd) * ap->rx_ring.size,
+ ap->rxbd, ap->rxbd_dma);
+err_out:
+ return err;
+}
+
+static int free_buffers(struct net_device *dev)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+
+ /* free skbuff */
+ vmac_tx_reclaim(dev, 1);
+ vmac_rx_reclaim_force(dev);
+
+ /* free DMA ring */
+ dma_free_coherent(&ap->pdev->dev, sizeof(ap->txbd) * ap->tx_ring.size,
+ ap->txbd, ap->txbd_dma);
+ dma_free_coherent(&ap->pdev->dev, sizeof(ap->rxbd) * ap->rx_ring.size,
+ ap->rxbd, ap->rxbd_dma);
+
+ return 0;
+}
+
+static int vmac_hw_init(struct net_device *dev)
+{
+ struct vmac_priv *priv = netdev_priv(dev);
+
+ /* clear IRQ mask */
+ vmac_writel(priv, 0, ENABLE);
+
+ /* clear pending IRQ */
+ vmac_writel(priv, 0xffffffff, STAT);
+
+ /* Initialize logical address filter */
+ vmac_writel(priv, 0x0, LAFL);
+ vmac_writel(priv, 0x0, LAFH);
+
+ return 0;
+}
+
+int vmac_open(struct net_device *dev)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ struct phy_device *phydev;
+ unsigned int temp;
+ int err = 0;
+
+ if (ap == NULL)
+ return -ENODEV;
+
+ ap->shutdown = 0;
+
+ vmac_hw_init(dev);
+
+ /* mac address changed? */
+ write_mac_reg(dev, dev->dev_addr);
+
+ err = alloc_buffers(dev);
+ if (err)
+ goto err_out;
+
+ err = request_irq(dev->irq, &vmac_intr, 0, dev->name, dev);
+ if (err) {
+ dev_err(&ap->pdev->dev, "Unable to request IRQ %d (error %d)\n",
+ dev->irq, err);
+ goto err_free_buffers;
+ }
+
+ /* install DMA ring pointers */
+ vmac_writel(ap, ap->rxbd_dma, RXRINGPTR);
+ vmac_writel(ap, ap->txbd_dma, TXRINGPTR);
+
+ /* set poll rate to 1 ms */
+ vmac_writel(ap, POLLRATE_TIME, POLLRATE);
+
+ /* make sure we enable napi before rx interrupt */
+ napi_enable(&ap->napi);
+
+ /* IRQ mask */
+ temp = RXINT_MASK | ERR_MASK | TXCH_MASK | MDIO_MASK;
+ vmac_writel(ap, temp, ENABLE);
+
+ /* Set control */
+ temp = (RX_BDT_LEN << 24) | (TX_BDT_LEN << 16) | TXRN_MASK | RXRN_MASK;
+ vmac_writel(ap, temp, CONTROL);
+
+ /* enable, after all other bits are set */
+ vmac_writel(ap, temp | EN_MASK, CONTROL);
+
+ netif_start_queue(dev);
+ netif_carrier_off(dev);
+
+ /* register the PHY board fixup, if needed */
+ err = vmac_mii_init(ap);
+ if (err)
+ goto err_free_irq;
+
+ /* schedule a link state check */
+ phy_start(ap->phy_dev);
+
+ phydev = ap->phy_dev;
+ dev_info(&ap->pdev->dev, "PHY driver [%s] (mii_bus:phy_addr=%s, irq=%d)\n",
+ phydev->drv->name, dev_name(&phydev->dev), phydev->irq);
+
+ return 0;
+
+err_free_irq:
+ free_irq(dev->irq, dev);
+err_free_buffers:
+ free_buffers(dev);
+err_out:
+ return err;
+}
+
+int vmac_close(struct net_device *dev)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ unsigned int temp;
+
+ netif_stop_queue(dev);
+ napi_disable(&ap->napi);
+
+ /* stop running transfers */
+ temp = vmac_readl(ap, CONTROL);
+ temp &= ~(TXRN_MASK | RXRN_MASK);
+ vmac_writel(ap, temp, CONTROL);
+
+ del_timer_sync(&ap->rx_timeout);
+
+ /* disable phy */
+ phy_stop(ap->phy_dev);
+ vmac_mii_exit(dev);
+ netif_carrier_off(dev);
+
+ /* disable interrupts */
+ vmac_writel(ap, 0, ENABLE);
+ free_irq(dev->irq, dev);
+
+ /* turn off vmac */
+ vmac_writel(ap, 0, CONTROL);
+ /* vmac_reset_hw(vmac) */
+
+ ap->shutdown = 1;
+ wmb();
+
+ free_buffers(dev);
+ return 0;
+}
+
+void vmac_update_stats(struct vmac_priv *ap)
+{
+ struct net_device_stats *_stats = &ap->stats;
+ unsigned long miss, rxerr;
+ unsigned long rxfram, rxcrc, rxoflow;
+
+ /* compare with /proc/net/dev,
+ * see net/core/dev.c:dev_seq_printf_stats */
+
+ /* rx stats */
+ rxerr = vmac_readl(ap, RXERR);
+ miss = vmac_readl(ap, MISS);
+
+ rxcrc = (rxerr & RXERR_CRC);
+ rxfram = (rxerr & RXERR_FRM) >> 8;
+ rxoflow = (rxerr & RXERR_OFLO) >> 16;
+
+ _stats->rx_length_errors = 0;
+ _stats->rx_over_errors += miss;
+ _stats->rx_crc_errors += rxcrc;
+ _stats->rx_frame_errors += rxfram;
+ _stats->rx_fifo_errors += rxoflow;
+ _stats->rx_missed_errors = 0;
+
+ /* TODO check rx_dropped/rx_errors/tx_dropped/tx_errors have not
+ * been updated elsewhere */
+ _stats->rx_dropped = _stats->rx_over_errors +
+ _stats->rx_fifo_errors +
+ ap->rx_merge_error;
+
+ _stats->rx_errors = _stats->rx_length_errors + _stats->rx_crc_errors +
+ _stats->rx_frame_errors +
+ _stats->rx_missed_errors +
+ _stats->rx_dropped;
+
+ /* tx stats */
+ _stats->tx_dropped = 0; /* otherwise queue stopped */
+
+ _stats->tx_errors = _stats->tx_aborted_errors +
+ _stats->tx_carrier_errors +
+ _stats->tx_fifo_errors +
+ _stats->tx_heartbeat_errors +
+ _stats->tx_window_errors +
+ _stats->tx_dropped +
+ ap->tx_timeout_error;
+}
+
+struct net_device_stats *vmac_stats(struct net_device *dev)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ unsigned long flags;
+
+ spin_lock_irqsave(&ap->lock, flags);
+ vmac_update_stats(ap);
+ spin_unlock_irqrestore(&ap->lock, flags);
+
+ return &ap->stats;
+}
+
+void vmac_tx_timeout(struct net_device *dev)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ unsigned int status;
+ unsigned long flags;
+
+ spin_lock_irqsave(&ap->lock, flags);
+
+ /* queue did not progress for timeo jiffies */
+ WARN_ON(!netif_queue_stopped(dev));
+ WARN_ON(!fifo_full(&ap->tx_ring));
+
+ /* TX IRQ lost? */
+ status = vmac_readl(ap, STAT);
+ if (status & TXINT_MASK) {
+ dev_err(&ap->pdev->dev, "lost tx interrupt, IRQ mask %x\n",
+ vmac_readl(ap, ENABLE));
+ vmac_writel(ap, TXINT_MASK, STAT);
+ }
+
+ /* TODO RX/MDIO/ERR as well? */
+
+ vmac_tx_reclaim(dev, 0);
+ if (fifo_full(&ap->tx_ring))
+ dev_err(&ap->pdev->dev, "DMA state machine not active\n");
+
+ /* We can accept TX packets again */
+ ap->tx_timeout_error++;
+ dev->trans_start = jiffies;
+ netif_wake_queue(dev);
+
+ spin_unlock_irqrestore(&ap->lock, flags);
+}
+
+static void create_multicast_filter(struct net_device *dev,
+ unsigned long *bitmask)
+{
+ struct netdev_hw_addr *ha;
+ unsigned long crc;
+ char *addrs;
+
+ WARN_ON(netdev_mc_count(dev) == 0);
+ WARN_ON(dev->flags & IFF_ALLMULTI);
+
+ bitmask[0] = bitmask[1] = 0;
+
+ netdev_for_each_mc_addr(ha, dev) {
+ addrs = ha->addr;
+
+ /* skip non-multicast addresses */
+ if (!(*addrs & 1))
+ continue;
+
+ crc = ether_crc_le(ETH_ALEN, addrs);
+ set_bit(crc >> 26, bitmask);
+ }
+}
+
+static void vmac_set_multicast_list(struct net_device *dev)
+{
+ struct vmac_priv *ap = netdev_priv(dev);
+ unsigned long flags, bitmask[2];
+ int promisc, reg;
+
+ spin_lock_irqsave(&ap->lock, flags);
+
+ promisc = !!(dev->flags & IFF_PROMISC);
+ reg = vmac_readl(ap, ENABLE);
+ if (promisc != !!(reg & PROM_MASK)) {
+ reg ^= PROM_MASK;
+ vmac_writel(ap, reg, ENABLE);
+ }
+
+ if (dev->flags & IFF_ALLMULTI)
+ memset(bitmask, 1, sizeof(bitmask));
+ else if (netdev_mc_count(dev) == 0)
+ memset(bitmask, 0, sizeof(bitmask));
+ else
+ create_multicast_filter(dev, bitmask);
+
+ vmac_writel(ap, bitmask[0], LAFL);
+ vmac_writel(ap, bitmask[1], LAFH);
+
+ spin_unlock_irqrestore(&ap->lock, flags);
+}
+
+static struct ethtool_ops vmac_ethtool_ops = {
+ .get_settings = vmacether_get_settings,
+ .set_settings = vmacether_set_settings,
+ .get_drvinfo = vmacether_get_drvinfo,
+ .get_link = ethtool_op_get_link,
+};
+
+static const struct net_device_ops vmac_netdev_ops = {
+ .ndo_open = vmac_open,
+ .ndo_stop = vmac_close,
+ .ndo_get_stats = vmac_stats,
+ .ndo_start_xmit = vmac_start_xmit,
+ .ndo_do_ioctl = vmac_ioctl,
+ .ndo_set_mac_address = eth_mac_addr,
+ .ndo_tx_timeout = vmac_tx_timeout,
+ .ndo_set_multicast_list = vmac_set_multicast_list,
+ .ndo_validate_addr = eth_validate_addr,
+ .ndo_change_mtu = eth_change_mtu,
+};
+
+static int __devinit vmac_probe(struct platform_device *pdev)
+{
+ struct net_device *dev;
+ struct vmac_priv *ap;
+ struct resource *res;
+ unsigned int mem_base, mem_size, irq;
+ int err;
+
+ dev = alloc_etherdev(sizeof(*ap));
+ if (!dev) {
+ dev_err(&pdev->dev, "etherdev alloc failed, aborting.\n");
+ return -ENOMEM;
+ }
+
+ ap = netdev_priv(dev);
+
+ err = -ENODEV;
+ res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+ if (!res) {
+ dev_err(&pdev->dev, "no mmio resource defined\n");
+ goto err_out;
+ }
+ mem_base = res->start;
+ mem_size = resource_size(res);
+ irq = platform_get_irq(pdev, 0);
+
+ err = -EBUSY;
+ if (!request_mem_region(mem_base, mem_size, DRV_NAME)) {
+ dev_err(&pdev->dev, "no memory region available\n");
+ goto err_out;
+ }
+
+ err = -ENOMEM;
+ ap->regs = ioremap(mem_base, mem_size);
+ if (!ap->regs) {
+ dev_err(&pdev->dev, "failed to map registers, aborting.\n");
+ goto err_out_release_mem;
+ }
+
+ /* no checksum support, hence no scatter/gather */
+ dev->features |= NETIF_F_HIGHDMA;
+
+ spin_lock_init(&ap->lock);
+
+ SET_NETDEV_DEV(dev, &pdev->dev);
+ ap->dev = dev;
+ ap->pdev = pdev;
+
+ /* init rx timeout (used for oom) */
+ init_timer(&ap->rx_timeout);
+ ap->rx_timeout.function = vmac_refill_rx_timer;
+ ap->rx_timeout.data = (unsigned long)dev;
+
+ netif_napi_add(dev, &ap->napi, vmac_poll, 2);
+ dev->netdev_ops = &vmac_netdev_ops;
+ dev->ethtool_ops = &vmac_ethtool_ops;
+ dev->irq = irq;
+
+ dev->flags |= IFF_MULTICAST;
+
+ dev->base_addr = (unsigned long)ap->regs;
+ ap->mem_base = mem_base;
+
+ /* prevent buffer chaining, favor speed over space */
+ ap->rx_skb_size = ETH_FRAME_LEN + VMAC_BUFFER_PAD;
+
+ /* private struct functional */
+
+ /* mac address intialize, set vmac_open */
+ read_mac_reg(dev, dev->dev_addr);
+
+ if (!is_valid_ether_addr(dev->dev_addr))
+ random_ether_addr(dev->dev_addr);
+
+ err = register_netdev(dev);
+ if (err) {
+ dev_err(&pdev->dev, "Cannot register net device, aborting.\n");
+ goto err_out_iounmap;
+ }
+
+ dev_info(&pdev->dev, "ARC VMAC at 0x%08x irq %d %pM\n", mem_base,
+ dev->irq, dev->dev_addr);
+ platform_set_drvdata(pdev, dev);
+
+ return 0;
+
+err_out_iounmap:
+ iounmap(ap->regs);
+err_out_release_mem:
+ release_mem_region(mem_base, mem_size);
+err_out:
+ free_netdev(dev);
+ return err;
+}
+
+static int __devexit vmac_remove(struct platform_device *pdev)
+{
+ struct net_device *dev;
+ struct vmac_priv *ap;
+ struct resource *res;
+
+ dev = platform_get_drvdata(pdev);
+ if (!dev) {
+ dev_err(&pdev->dev, "%s no valid dev found\n", __func__);
+ return 0;
+ }
+
+ ap = netdev_priv(dev);
+
+ /* MAC */
+ unregister_netdev(dev);
+ iounmap(ap->regs);
+
+ res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+ release_mem_region(res->start, resource_size(res));
+
+ platform_set_drvdata(pdev, NULL);
+ free_netdev(dev);
+ return 0;
+}
+
+static struct platform_driver arcvmac_driver = {
+ .probe = vmac_probe,
+ .remove = __devexit_p(vmac_remove),
+ .driver = {
+ .name = "arcvmac",
+ },
+};
+
+static int __init vmac_init(void)
+{
+ return platform_driver_register(&arcvmac_driver);
+}
+
+static void __exit vmac_exit(void)
+{
+ platform_driver_unregister(&arcvmac_driver);
+}
+
+module_init(vmac_init);
+module_exit(vmac_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("ARC VMAC Ethernet driver");
+MODULE_AUTHOR("amit.bhor@celunite.com, sameer.dhavale@celunite.com, andreas.fenkart@streamunlimited.com");
diff --git a/drivers/net/arcvmac.h b/drivers/net/arcvmac.h
new file mode 100644
index 0000000..44c0587
--- /dev/null
+++ b/drivers/net/arcvmac.h
@@ -0,0 +1,268 @@
+/*
+ * linux/arch/arc/drivers/arcvmac.h
+ *
+ * Copyright (C) 2003-2006 Codito Technologies, for linux-2.4 port
+ * Copyright (C) 2006-2007 Celunite Inc, for linux-2.6 port
+ * Copyright (C) 2007-2008 Sagem Communications, Fehmi HAFSI
+ * Copyright (C) 2009 Sagem Communications, Andreas Fenkart
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ *
+ * Authors: amit.bhor@celunite.com, sameer.dhavale@celunite.com
+ */
+
+#ifndef _ARCVMAC_H
+#define _ARCVMAC_H
+
+#define DRV_NAME "arcvmac"
+#define DRV_VERSION "1.0"
+
+/* Buffer descriptors */
+#define TX_BDT_LEN 16 /* Number of receive BD's */
+#define RX_BDT_LEN 256 /* Number of transmit BD's */
+
+/* BD poll rate, in 1024 cycles. @100Mhz: x * 1024 cy * 10ns = 1ms */
+#define POLLRATE_TIME 200
+
+/* next power of two, bigger than ETH_FRAME_LEN + VLAN */
+#define MAX_RX_BUFFER_LEN 0x800 /* 2^11 = 2048 = 0x800 */
+#define MAX_TX_BUFFER_LEN 0x800 /* 2^11 = 2048 = 0x800 */
+
+/* 14 bytes of ethernet header, 4 bytes VLAN, FCS,
+ * plus extra pad to prevent buffer chaining of
+ * maximum sized ethernet packets (1514 bytes) */
+#define VMAC_BUFFER_PAD (ETH_HLEN + 4 + ETH_FCS_LEN + 4)
+
+/* VMAC register definitions, offsets in the ref manual are in bytes */
+#define ID_OFFSET (0x00/0x4)
+#define STAT_OFFSET (0x04/0x4)
+#define ENABLE_OFFSET (0x08/0x4)
+#define CONTROL_OFFSET (0x0c/0x4)
+#define POLLRATE_OFFSET (0x10/0x4)
+#define RXERR_OFFSET (0x14/0x4)
+#define MISS_OFFSET (0x18/0x4)
+#define TXRINGPTR_OFFSET (0x1c/0x4)
+#define RXRINGPTR_OFFSET (0x20/0x4)
+#define ADDRL_OFFSET (0x24/0x4)
+#define ADDRH_OFFSET (0x28/0x4)
+#define LAFL_OFFSET (0x2c/0x4)
+#define LAFH_OFFSET (0x30/0x4)
+#define MDIO_DATA_OFFSET (0x34/0x4)
+#define MAC_TXRING_HEAD_OFFSET (0x38/0x4)
+#define MAC_RXRING_HEAD_OFFSET (0x3C/0x4)
+
+/* STATUS and ENABLE register bit masks */
+#define TXINT_MASK (1<<0) /* Transmit interrupt */
+#define RXINT_MASK (1<<1) /* Receive interrupt */
+#define ERR_MASK (1<<2) /* Error interrupt */
+#define TXCH_MASK (1<<3) /* Transmit chaining error interrupt */
+#define MSER_MASK (1<<4) /* Missed packet counter error */
+#define RXCR_MASK (1<<8) /* RXCRCERR counter rolled over */
+#define RXFR_MASK (1<<9) /* RXFRAMEERR counter rolled over */
+#define RXFL_MASK (1<<10) /* RXOFLOWERR counter rolled over */
+#define MDIO_MASK (1<<12) /* MDIO complete */
+#define TXPL_MASK (1<<31) /* TXPOLL */
+
+/* CONTROL register bitmasks */
+#define EN_MASK (1<<0) /* VMAC enable */
+#define TXRN_MASK (1<<3) /* TX enable */
+#define RXRN_MASK (1<<4) /* RX enable */
+#define DSBC_MASK (1<<8) /* Disable receive broadcast */
+#define ENFL_MASK (1<<10) /* Enable Full Duplex */
+#define PROM_MASK (1<<11) /* Promiscuous mode */
+
+/* RXERR register bitmasks */
+#define RXERR_CRC 0x000000ff
+#define RXERR_FRM 0x0000ff00
+#define RXERR_OFLO 0x00ff0000 /* fifo overflow */
+
+/* MDIO data register bit masks */
+#define MDIO_SFD 0xC0000000
+#define MDIO_OP 0x30000000
+#define MDIO_ID_MASK 0x0F800000
+#define MDIO_REG_MASK 0x007C0000
+#define MDIO_TA 0x00030000
+#define MDIO_DATA_MASK 0x0000FFFF
+
+#define MDIO_BASE 0x40020000
+#define MDIO_OP_READ 0x20000000
+#define MDIO_OP_WRITE 0x10000000
+
+/* Buffer descriptor INFO bit masks */
+#define OWN_MASK (1<<31) /* ownership of buffer, 0 CPU, 1 DMA */
+#define BUFF (1<<30) /* buffer invalid, rx */
+#define UFLO (1<<29) /* underflow, tx */
+#define LTCL (1<<28) /* late collision, tx */
+#define RETRY_CT (0xf<<24) /* tx */
+#define DROP (1<<23) /* drop, more than 16 retries, tx */
+#define DEFER (1<<22) /* traffic on the wire, tx */
+#define CARLOSS (1<<21) /* carrier loss while transmission, tx, rx? */
+/* 20:19 reserved */
+#define ADCR (1<<18) /* add crc, ignored if not disaddcrc */
+#define LAST_MASK (1<<17) /* Last buffer in chain */
+#define FRST_MASK (1<<16) /* First buffer in chain */
+/* 15:11 reserved */
+#define LEN_MASK 0x000007FF
+
+#define ERR_MSK_TX 0x3fe00000 /* UFLO | LTCL | RTRY | DROP | DEFER | CRLS */
+
+
+/* arcvmac private data structures */
+struct vmac_buffer_desc {
+ unsigned int info;
+ dma_addr_t data;
+};
+
+struct dma_fifo {
+ int head; /* head */
+ int tail; /* tail */
+ int size;
+};
+
+struct vmac_priv {
+ struct net_device *dev;
+ struct platform_device *pdev;
+ struct net_device_stats stats;
+
+ spinlock_t lock; /* TODO revisit */
+ struct completion mdio_complete;
+
+ /* base address of register set */
+ int *regs;
+ unsigned int mem_base;
+
+ /* DMA ring buffers */
+ struct vmac_buffer_desc *rxbd;
+ dma_addr_t rxbd_dma;
+
+ struct vmac_buffer_desc *txbd;
+ dma_addr_t txbd_dma;
+
+ /* socket buffers */
+ struct sk_buff *rx_skbuff[RX_BDT_LEN];
+ struct sk_buff *tx_skbuff[TX_BDT_LEN];
+ int rx_skb_size;
+
+ /* skb / dma desc managing */
+ struct dma_fifo rx_ring;
+ struct dma_fifo tx_ring;
+
+ /* descriptor last polled/processed by the VMAC */
+ unsigned long mac_rxring_head;
+ /* used when rx skb allocation failed, so we defer rx queue
+ * refill */
+ struct timer_list rx_timeout;
+
+ /* lock rx_timeout against rx normal operation */
+ spinlock_t rx_lock;
+
+ struct napi_struct napi;
+
+ /* rx buffer chaining */
+ int rx_merge_error;
+ int tx_timeout_error;
+
+ /* PHY stuff */
+ struct mii_bus *mii_bus;
+ struct phy_device *phy_dev;
+
+ int link;
+ int speed;
+ int duplex;
+
+ /* debug */
+ int shutdown;
+};
+
+/* DMA ring management */
+
+/* for a fifo with size n,
+ * - [0..n] fill levels are n + 1 states
+ * - there are only n different deltas (head - tail) values
+ * => not all fill levels can be represented with head, tail
+ * pointers only
+ * we give up the n fill level, aka fifo full */
+
+/* sacrifice one elt as a sentinel */
+static inline int fifo_used(struct dma_fifo *f);
+static inline int fifo_inc_ct(int ct, int size);
+static inline void fifo_dump(struct dma_fifo *fifo);
+
+static inline int fifo_empty(struct dma_fifo *f)
+{
+ return (f->head == f->tail);
+}
+
+static inline int fifo_free(struct dma_fifo *f)
+{
+ int free;
+
+ free = f->tail - f->head;
+ if (free <= 0)
+ free += f->size;
+
+ return free;
+}
+
+static inline int fifo_used(struct dma_fifo *f)
+{
+ int used;
+
+ used = f->head - f->tail;
+ if (used < 0)
+ used += f->size;
+
+ return used;
+}
+
+static inline int fifo_full(struct dma_fifo *f)
+{
+ return (fifo_used(f) + 1) == f->size;
+}
+
+/* manipulate */
+static inline void fifo_init(struct dma_fifo *fifo, int size)
+{
+ fifo->size = size;
+ fifo->head = fifo->tail = 0; /* empty */
+}
+
+static inline void fifo_inc_head(struct dma_fifo *fifo)
+{
+ BUG_ON(fifo_full(fifo));
+ fifo->head = fifo_inc_ct(fifo->head, fifo->size);
+}
+
+static inline void fifo_inc_tail(struct dma_fifo *fifo)
+{
+ BUG_ON(fifo_empty(fifo));
+ fifo->tail = fifo_inc_ct(fifo->tail, fifo->size);
+}
+
+/* internal funcs */
+static inline void fifo_dump(struct dma_fifo *fifo)
+{
+ printk(KERN_INFO "fifo: head %d, tail %d, size %d\n", fifo->head,
+ fifo->tail,
+ fifo->size);
+}
+
+static inline int fifo_inc_ct(int ct, int size)
+{
+ return (++ct == size) ? 0 : ct;
+}
+
+#endif /* _ARCVMAC_H */
--
1.7.1
^ permalink raw reply related
* [PATCH 0/1] ARC vmac ethernet driver.
From: Andreas Fenkart @ 2010-10-15 7:54 UTC (permalink / raw)
To: netdev
This is a driver for the MAC IP block from Arc international. It
is based on an existing driver found in Arc linux distribution,
but essentially, it is a full rewrite. Influenced heavily by
dnet/bcm63xx_enet drivers.
This is a resubmission, changes are:
- Kconfig dependencies, HAS_DMA/CRC32/PHYLIB/MII
- removed mac_addr module parameter
- rebased linux tree: 3b72090a7317a034d1276a8fbe3b68c3cb77bd92
regards
Andreas
^ permalink raw reply
* [PATCH net-next] can-raw: add msg_flags to distinguish local traffic
From: Kurt Van Dijck @ 2010-10-15 7:37 UTC (permalink / raw)
To: netdev, socketcan-core-0fE9KPoRgkgATYTw5x5z8w; +Cc: Oliver Hartkopp
CAN has no addressing scheme. It is currently impossible
for userspace to tell is a received CAN frame comes from
another process on the local host, or from a remote CAN
device.
This patch add support for userspace applications to distinguish
between 'own', 'local' and 'remote' CAN traffic.
Distinction is made by returning some flags in msg->msg_flags
in the call to recvmsg.
MSG_CONFIRM flag means 'own', as in 'transmission confirmation'
MSG_DONTROUTE flag means 'local', not routed.
Obviously, msgs with MSG_CONFIRM will have MSG_DONTROUTE set too.
Please note that on SocketCAN mailing list, different opinions
exist on the exact meaning of MSG_DONTROUTE. Better (=more
intuitive) alternatives are appreciated.
Signed-off-by: Kurt Van Dijck <kurt.van.dijck-AgBVmzD5pcezQB+pC5nmwQ@public.gmane.org>
---
net/can/raw.c | 33 ++++++++++++++++++++++++++++++---
1 files changed, 30 insertions(+), 3 deletions(-)
diff --git a/net/can/raw.c b/net/can/raw.c
index 7d77e67..f98709e 100644
--- a/net/can/raw.c
+++ b/net/can/raw.c
@@ -90,23 +90,39 @@ struct raw_sock {
can_err_mask_t err_mask;
};
+/*
+ * return some space to store extra msg flags in.
+ * We use 1 int beyond the 'struct sockaddr_can' in skb->cb
+ * to store those.
+ * These flags will be use in raw_recvmsg()
+ */
+static inline int *raw_flags(struct sk_buff *skb)
+{
+ BUILD_BUG_ON(sizeof(skb->cb)
+ <= (sizeof(struct sockaddr_can) + sizeof(int)));
+ /* return pointer after struct sockaddr_can */
+ return (int *)(&((struct sockaddr_can *)skb->cb)[1]);
+}
+
static inline struct raw_sock *raw_sk(const struct sock *sk)
{
return (struct raw_sock *)sk;
}
-static void raw_rcv(struct sk_buff *skb, void *data)
+static void raw_rcv(struct sk_buff *oskb, void *data)
{
struct sock *sk = (struct sock *)data;
struct raw_sock *ro = raw_sk(sk);
struct sockaddr_can *addr;
+ struct sk_buff *skb;
+ int *pflags;
/* check the received tx sock reference */
- if (!ro->recv_own_msgs && skb->sk == sk)
+ if (!ro->recv_own_msgs && oskb->sk == sk)
return;
/* clone the given skb to be able to enqueue it into the rcv queue */
- skb = skb_clone(skb, GFP_ATOMIC);
+ skb = skb_clone(oskb, GFP_ATOMIC);
if (!skb)
return;
@@ -123,6 +139,14 @@ static void raw_rcv(struct sk_buff *skb, void *data)
addr->can_family = AF_CAN;
addr->can_ifindex = skb->dev->ifindex;
+ /* prepare the flags for raw_recvmsg() */
+ pflags = raw_flags(skb);
+ *pflags = 0;
+ if (oskb->sk)
+ *pflags |= MSG_DONTROUTE;
+ if (oskb->sk == sk)
+ *pflags |= MSG_CONFIRM;
+
if (sock_queue_rcv_skb(sk, skb) < 0)
kfree_skb(skb);
}
@@ -707,6 +731,9 @@ static int raw_recvmsg(struct kiocb *iocb, struct socket *sock,
memcpy(msg->msg_name, skb->cb, msg->msg_namelen);
}
+ /* assign the flags that have been recorded in in raw_rcv() */
+ msg->msg_flags |= *(raw_flags(skb));
+
skb_free_datagram(sk, skb);
return size;
^ permalink raw reply related
* Re: [GIT PATCH] ioat2: fix performance regression
From: Linus Walleij @ 2010-10-15 7:15 UTC (permalink / raw)
To: Dan Williams
Cc: akpm@linux-foundation.org, torvalds@linux-foundation.org,
Dave Jiang, netdev@vger.kernel.org, Maciej Sosnowski,
Jesse Brandeburg, linux-kernel@vger.kernel.org, stable@kernel.org,
Richard Scobie
In-Reply-To: <20101013230248.32594.35857.stgit@localhost6.localdomain6>
Dan Williams wrote:
> Commit 0793448 "DMAENGINE: generic channel status v2" changed the interface for
> how dma channel progress is retrieved. It inadvertently exported an internal
> helper function ioat_tx_status() instead of ioat_dma_tx_status().
Acked-by: Linus Walleij <linus.walleij@stericsson.com>
Yours,
Linus Walleij
^ permalink raw reply
* Re: -j MARK in raw vs. mangle (was Re: xfrm by MARK: tcp problems when mark for in and out differ)
From: Patrick McHardy @ 2010-10-15 7:04 UTC (permalink / raw)
To: Gerd v. Egidy; +Cc: hadi, netfilter-devel, netdev
In-Reply-To: <201010141616.58795.lists@egidy.de>
On 14.10.2010 16:16, Gerd v. Egidy wrote:
> Hi Jamal,
>
> thanks for your help.
>
>>> So it seems like the fl->mark is never initialized with the packet mark
>>> in the first place. What would be the correct stage in the kernel
>>> network stack to do that?
>>
>> Can you try a simple setup without xfrm/ipsec and see if this reverse
>> path works? Was there a kernel where it worked?
>
> I just tried opening a simple tcp connection without any xfrm or other weird
> stuff. I just had one iptables rule in place:
>
> -t raw -A OUTPUT -d 192.168.5.200 -j MARK --set-mark 99
>
> 192.168.5.200 is the other system I open the tcp connection from. So this
> should mark all response packets to the client.
>
> But the moment __xfrm_lookup is called (this is where my debug printk sits),
> fl->mark is always 0.
>
> By chance I changed the rule over to the mangle table:
>
> -t mangle -A OUTPUT -d 192.168.5.200 -j MARK --set-mark 99
>
> Now it works, the mark in the flow is 99!
>
> So it seems this has nothing to do with xfrm, but that the MARK target has
> different effects when used in raw than in mangle. I was using raw because I
> had to set conntrack zones too and it was more conveniant to do both in one
> place.
>
> Can one of the netfilter guys comment on this? Is using MARK in raw not fully
> supported or has known deficiencies?
No, the problem is most likely that for outgoing packets, the XFRM
lookup is done with the route lookup before the packet is even sent,
so once it hits the raw or mangle table, it is too late. mangle however
performs rerouting when the mark value changes, at which point a new
XFRM lookup is performed.
^ permalink raw reply
* [PATCH net-next 1/3] fib_hash: embed initial hash table in fn_zone
From: Eric Dumazet @ 2010-10-15 6:53 UTC (permalink / raw)
To: David Miller; +Cc: netdev
In-Reply-To: <1287084159.2659.0.camel@edumazet-laptop>
While looking for false sharing problems, I noticed
sizeof(struct fn_zone) was small (28 bytes) and possibly sharing a cache
line with an often written kernel structure.
Most of the time, fn_zone uses its initial hash table of 16 slots.
We can avoid the false sharing problem by embedding this initial hash
table in fn_zone itself, so that sizeof(fn_zone) > L1_CACHE_BYTES
We did a similar optimization in commit a6501e080c (Reduce memory needs
and speedup lookups)
Add a fz_revorder field to speedup fn_hash() a bit.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
---
net/ipv4/fib_hash.c | 52 ++++++++++++++++++------------------------
1 file changed, 23 insertions(+), 29 deletions(-)
diff --git a/net/ipv4/fib_hash.c b/net/ipv4/fib_hash.c
index 83cca68..10001aa 100644
--- a/net/ipv4/fib_hash.c
+++ b/net/ipv4/fib_hash.c
@@ -54,23 +54,23 @@ struct fib_node {
struct fib_alias fn_embedded_alias;
};
+#define EMBEDDED_HASH_SIZE (L1_CACHE_BYTES / sizeof(struct hlist_head))
+
struct fn_zone {
struct fn_zone *fz_next; /* Next not empty zone */
struct hlist_head *fz_hash; /* Hash table pointer */
- int fz_nent; /* Number of entries */
-
- int fz_divisor; /* Hash divisor */
u32 fz_hashmask; /* (fz_divisor - 1) */
-#define FZ_HASHMASK(fz) ((fz)->fz_hashmask)
- int fz_order; /* Zone order */
- __be32 fz_mask;
+ u8 fz_order; /* Zone order (0..32) */
+ u8 fz_revorder; /* 32 - fz_order */
+ __be32 fz_mask; /* inet_make_mask(order) */
#define FZ_MASK(fz) ((fz)->fz_mask)
-};
-/* NOTE. On fast computers evaluation of fz_hashmask and fz_mask
- * can be cheaper than memory lookup, so that FZ_* macros are used.
- */
+ struct hlist_head fz_embedded_hash[EMBEDDED_HASH_SIZE];
+
+ int fz_nent; /* Number of entries */
+ int fz_divisor; /* Hash size (mask+1) */
+};
struct fn_hash {
struct fn_zone *fn_zones[33];
@@ -79,11 +79,11 @@ struct fn_hash {
static inline u32 fn_hash(__be32 key, struct fn_zone *fz)
{
- u32 h = ntohl(key)>>(32 - fz->fz_order);
+ u32 h = ntohl(key) >> fz->fz_revorder;
h ^= (h>>20);
h ^= (h>>10);
h ^= (h>>5);
- h &= FZ_HASHMASK(fz);
+ h &= fz->fz_hashmask;
return h;
}
@@ -147,14 +147,14 @@ static void fn_rehash_zone(struct fn_zone *fz)
int old_divisor, new_divisor;
u32 new_hashmask;
- old_divisor = fz->fz_divisor;
+ new_divisor = old_divisor = fz->fz_divisor;
switch (old_divisor) {
- case 16:
- new_divisor = 256;
+ case EMBEDDED_HASH_SIZE:
+ new_divisor *= EMBEDDED_HASH_SIZE;
break;
- case 256:
- new_divisor = 1024;
+ case EMBEDDED_HASH_SIZE*EMBEDDED_HASH_SIZE:
+ new_divisor *= (EMBEDDED_HASH_SIZE/2);
break;
default:
if ((old_divisor << 1) > FZ_MAX_DIVISOR) {
@@ -184,7 +184,8 @@ static void fn_rehash_zone(struct fn_zone *fz)
fib_hash_genid++;
write_unlock_bh(&fib_hash_lock);
- fz_hash_free(old_ht, old_divisor);
+ if (old_ht != fz->fz_embedded_hash)
+ fz_hash_free(old_ht, old_divisor);
}
}
@@ -210,18 +211,11 @@ fn_new_zone(struct fn_hash *table, int z)
if (!fz)
return NULL;
- if (z) {
- fz->fz_divisor = 16;
- } else {
- fz->fz_divisor = 1;
- }
- fz->fz_hashmask = (fz->fz_divisor - 1);
- fz->fz_hash = fz_hash_alloc(fz->fz_divisor);
- if (!fz->fz_hash) {
- kfree(fz);
- return NULL;
- }
+ fz->fz_divisor = z ? EMBEDDED_HASH_SIZE : 1;
+ fz->fz_hashmask = fz->fz_divisor - 1;
+ fz->fz_hash = fz->fz_embedded_hash;
fz->fz_order = z;
+ fz->fz_revorder = 32 - z;
fz->fz_mask = inet_make_mask(z);
/* Find the first not empty zone with more specific mask */
^ permalink raw reply related
* [PATCH net-next 3/3] fib_hash: RCU conversion phase 2
From: Eric Dumazet @ 2010-10-15 6:56 UTC (permalink / raw)
To: David Miller; +Cc: netdev
Get rid of fib_hash_lock rwlock.
The fn_zone hash table resize is the noticeable part of this patch.
I added a seqlock per fn_zone, so that readers can restart their lookup
in the (very rare) case a writer expanded the hash table.
Add rcu heads in fib_alias and fib_node, use call_rcu() to defer their
freeing, and use appropriate _rcu list manipulations.
Stress test (160.000.000 udp frames sent, IP route cache disabled to
mimic DDOS attack, FIB_HASH)
Before:
real 0m41.191s
user 0m13.137s
sys 8m55.241s
After:
real 0m38.091s
user 0m13.189s
sys 7m53.018s
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
---
net/ipv4/fib_hash.c | 176 ++++++++++++++++++++++------------------
net/ipv4/fib_lookup.h | 2
2 files changed, 101 insertions(+), 77 deletions(-)
diff --git a/net/ipv4/fib_hash.c b/net/ipv4/fib_hash.c
index 04f05a9..4f1aafd 100644
--- a/net/ipv4/fib_hash.c
+++ b/net/ipv4/fib_hash.c
@@ -58,7 +58,8 @@ struct fib_node {
struct fn_zone {
struct fn_zone __rcu *fz_next; /* Next not empty zone */
- struct hlist_head *fz_hash; /* Hash table pointer */
+ struct hlist_head __rcu *fz_hash; /* Hash table pointer */
+ seqlock_t fz_lock;
u32 fz_hashmask; /* (fz_divisor - 1) */
u8 fz_order; /* Zone order (0..32) */
@@ -92,7 +93,6 @@ static inline __be32 fz_key(__be32 dst, struct fn_zone *fz)
return dst & FZ_MASK(fz);
}
-static DEFINE_RWLOCK(fib_hash_lock);
static unsigned int fib_hash_genid;
#define FZ_MAX_DIVISOR ((PAGE_SIZE<<MAX_ORDER) / sizeof(struct hlist_head))
@@ -101,12 +101,11 @@ static struct hlist_head *fz_hash_alloc(int divisor)
{
unsigned long size = divisor * sizeof(struct hlist_head);
- if (size <= PAGE_SIZE) {
+ if (size <= PAGE_SIZE)
return kzalloc(size, GFP_KERNEL);
- } else {
- return (struct hlist_head *)
- __get_free_pages(GFP_KERNEL | __GFP_ZERO, get_order(size));
- }
+
+ return (struct hlist_head *)
+ __get_free_pages(GFP_KERNEL | __GFP_ZERO, get_order(size));
}
/* The fib hash lock must be held when this is called. */
@@ -121,12 +120,12 @@ static inline void fn_rebuild_zone(struct fn_zone *fz,
struct fib_node *f;
hlist_for_each_entry_safe(f, node, n, &old_ht[i], fn_hash) {
- struct hlist_head *new_head;
+ struct hlist_head __rcu *new_head;
- hlist_del(&f->fn_hash);
+ hlist_del_rcu(&f->fn_hash);
new_head = &fz->fz_hash[fn_hash(f->fn_key, fz)];
- hlist_add_head(&f->fn_hash, new_head);
+ hlist_add_head_rcu(&f->fn_hash, new_head);
}
}
}
@@ -175,32 +174,55 @@ static void fn_rehash_zone(struct fn_zone *fz)
ht = fz_hash_alloc(new_divisor);
if (ht) {
- write_lock_bh(&fib_hash_lock);
+ struct fn_zone nfz;
+
+ memcpy(&nfz, fz, sizeof(nfz));
+
+ write_seqlock_bh(&fz->fz_lock);
old_ht = fz->fz_hash;
- fz->fz_hash = ht;
+ nfz.fz_hash = ht;
+ nfz.fz_hashmask = new_hashmask;
+ nfz.fz_divisor = new_divisor;
+ fn_rebuild_zone(&nfz, old_ht, old_divisor);
+ fib_hash_genid++;
+ rcu_assign_pointer(fz->fz_hash, ht);
fz->fz_hashmask = new_hashmask;
fz->fz_divisor = new_divisor;
- fn_rebuild_zone(fz, old_ht, old_divisor);
- fib_hash_genid++;
- write_unlock_bh(&fib_hash_lock);
+ write_sequnlock_bh(&fz->fz_lock);
- if (old_ht != fz->fz_embedded_hash)
+ if (old_ht != fz->fz_embedded_hash) {
+ synchronize_rcu();
fz_hash_free(old_ht, old_divisor);
+ }
}
}
-static inline void fn_free_node(struct fib_node * f)
+static void fn_free_node_rcu(struct rcu_head *head)
{
+ struct fib_node *f = container_of(head, struct fib_node, fn_embedded_alias.rcu);
+
kmem_cache_free(fn_hash_kmem, f);
}
+static inline void fn_free_node(struct fib_node *f)
+{
+ call_rcu(&f->fn_embedded_alias.rcu, fn_free_node_rcu);
+}
+
+static void fn_free_alias_rcu(struct rcu_head *head)
+{
+ struct fib_alias *fa = container_of(head, struct fib_alias, rcu);
+
+ kmem_cache_free(fn_alias_kmem, fa);
+}
+
static inline void fn_free_alias(struct fib_alias *fa, struct fib_node *f)
{
fib_release_info(fa->fa_info);
if (fa == &f->fn_embedded_alias)
fa->fa_info = NULL;
else
- kmem_cache_free(fn_alias_kmem, fa);
+ call_rcu(&fa->rcu, fn_free_alias_rcu);
}
static struct fn_zone *
@@ -211,6 +233,7 @@ fn_new_zone(struct fn_hash *table, int z)
if (!fz)
return NULL;
+ seqlock_init(&fz->fz_lock);
fz->fz_divisor = z ? EMBEDDED_HASH_SIZE : 1;
fz->fz_hashmask = fz->fz_divisor - 1;
fz->fz_hash = fz->fz_embedded_hash;
@@ -246,30 +269,34 @@ int fib_table_lookup(struct fib_table *tb,
struct fn_hash *t = (struct fn_hash *)tb->tb_data;
rcu_read_lock();
- read_lock(&fib_hash_lock);
for (fz = rcu_dereference(t->fn_zone_list);
fz != NULL;
fz = rcu_dereference(fz->fz_next)) {
- struct hlist_head *head;
+ struct hlist_head __rcu *head;
struct hlist_node *node;
struct fib_node *f;
- __be32 k = fz_key(flp->fl4_dst, fz);
+ __be32 k;
+ unsigned int seq;
- head = &fz->fz_hash[fn_hash(k, fz)];
- hlist_for_each_entry(f, node, head, fn_hash) {
- if (f->fn_key != k)
- continue;
+ do {
+ seq = read_seqbegin(&fz->fz_lock);
+ k = fz_key(flp->fl4_dst, fz);
+
+ head = &fz->fz_hash[fn_hash(k, fz)];
+ hlist_for_each_entry_rcu(f, node, head, fn_hash) {
+ if (f->fn_key != k)
+ continue;
- err = fib_semantic_match(&f->fn_alias,
+ err = fib_semantic_match(&f->fn_alias,
flp, res,
fz->fz_order, fib_flags);
- if (err <= 0)
- goto out;
- }
+ if (err <= 0)
+ goto out;
+ }
+ } while (read_seqretry(&fz->fz_lock, seq));
}
err = 1;
out:
- read_unlock(&fib_hash_lock);
rcu_read_unlock();
return err;
}
@@ -292,11 +319,11 @@ void fib_table_select_default(struct fib_table *tb,
last_resort = NULL;
order = -1;
- read_lock(&fib_hash_lock);
- hlist_for_each_entry(f, node, &fz->fz_hash[0], fn_hash) {
+ rcu_read_lock();
+ hlist_for_each_entry_rcu(f, node, &fz->fz_hash[0], fn_hash) {
struct fib_alias *fa;
- list_for_each_entry(fa, &f->fn_alias, fa_list) {
+ list_for_each_entry_rcu(fa, &f->fn_alias, fa_list) {
struct fib_info *next_fi = fa->fa_info;
if (fa->fa_scope != res->scope ||
@@ -340,7 +367,7 @@ void fib_table_select_default(struct fib_table *tb,
fib_result_assign(res, last_resort);
tb->tb_default = last_idx;
out:
- read_unlock(&fib_hash_lock);
+ rcu_read_unlock();
}
/* Insert node F to FZ. */
@@ -348,7 +375,7 @@ static inline void fib_insert_node(struct fn_zone *fz, struct fib_node *f)
{
struct hlist_head *head = &fz->fz_hash[fn_hash(f->fn_key, fz)];
- hlist_add_head(&f->fn_hash, head);
+ hlist_add_head_rcu(&f->fn_hash, head);
}
/* Return the node in FZ matching KEY. */
@@ -358,7 +385,7 @@ static struct fib_node *fib_find_node(struct fn_zone *fz, __be32 key)
struct hlist_node *node;
struct fib_node *f;
- hlist_for_each_entry(f, node, head, fn_hash) {
+ hlist_for_each_entry_rcu(f, node, head, fn_hash) {
if (f->fn_key == key)
return f;
}
@@ -366,6 +393,16 @@ static struct fib_node *fib_find_node(struct fn_zone *fz, __be32 key)
return NULL;
}
+
+static struct fib_alias *fib_fast_alloc(struct fib_node *f)
+{
+ struct fib_alias *fa = &f->fn_embedded_alias;
+
+ if (fa->fa_info != NULL)
+ fa = kmem_cache_alloc(fn_alias_kmem, GFP_KERNEL);
+ return fa;
+}
+
/* Caller must hold RTNL. */
int fib_table_insert(struct fib_table *tb, struct fib_config *cfg)
{
@@ -451,7 +488,6 @@ int fib_table_insert(struct fib_table *tb, struct fib_config *cfg)
}
if (cfg->fc_nlflags & NLM_F_REPLACE) {
- struct fib_info *fi_drop;
u8 state;
fa = fa_first;
@@ -460,21 +496,25 @@ int fib_table_insert(struct fib_table *tb, struct fib_config *cfg)
err = 0;
goto out;
}
- write_lock_bh(&fib_hash_lock);
- fi_drop = fa->fa_info;
- fa->fa_info = fi;
- fa->fa_type = cfg->fc_type;
- fa->fa_scope = cfg->fc_scope;
+ err = -ENOBUFS;
+ new_fa = fib_fast_alloc(f);
+ if (new_fa == NULL)
+ goto out;
+
+ new_fa->fa_tos = fa->fa_tos;
+ new_fa->fa_info = fi;
+ new_fa->fa_type = cfg->fc_type;
+ new_fa->fa_scope = cfg->fc_scope;
state = fa->fa_state;
- fa->fa_state &= ~FA_S_ACCESSED;
+ new_fa->fa_state = state & ~FA_S_ACCESSED;
fib_hash_genid++;
- write_unlock_bh(&fib_hash_lock);
+ list_replace_rcu(&fa->fa_list, &new_fa->fa_list);
- fib_release_info(fi_drop);
+ fn_free_alias(fa, f);
if (state & FA_S_ACCESSED)
rt_cache_flush(cfg->fc_nlinfo.nl_net, -1);
- rtmsg_fib(RTM_NEWROUTE, key, fa, cfg->fc_dst_len, tb->tb_id,
- &cfg->fc_nlinfo, NLM_F_REPLACE);
+ rtmsg_fib(RTM_NEWROUTE, key, new_fa, cfg->fc_dst_len,
+ tb->tb_id, &cfg->fc_nlinfo, NLM_F_REPLACE);
return 0;
}
@@ -506,12 +546,10 @@ int fib_table_insert(struct fib_table *tb, struct fib_config *cfg)
f = new_f;
}
- new_fa = &f->fn_embedded_alias;
- if (new_fa->fa_info != NULL) {
- new_fa = kmem_cache_alloc(fn_alias_kmem, GFP_KERNEL);
- if (new_fa == NULL)
- goto out;
- }
+ new_fa = fib_fast_alloc(f);
+ if (new_fa == NULL)
+ goto out;
+
new_fa->fa_info = fi;
new_fa->fa_tos = tos;
new_fa->fa_type = cfg->fc_type;
@@ -522,13 +560,11 @@ int fib_table_insert(struct fib_table *tb, struct fib_config *cfg)
* Insert new entry to the list.
*/
- write_lock_bh(&fib_hash_lock);
if (new_f)
fib_insert_node(fz, new_f);
- list_add_tail(&new_fa->fa_list,
+ list_add_tail_rcu(&new_fa->fa_list,
(fa ? &fa->fa_list : &f->fn_alias));
fib_hash_genid++;
- write_unlock_bh(&fib_hash_lock);
if (new_f)
fz->fz_nent++;
@@ -603,14 +639,12 @@ int fib_table_delete(struct fib_table *tb, struct fib_config *cfg)
tb->tb_id, &cfg->fc_nlinfo, 0);
kill_fn = 0;
- write_lock_bh(&fib_hash_lock);
- list_del(&fa->fa_list);
+ list_del_rcu(&fa->fa_list);
if (list_empty(&f->fn_alias)) {
- hlist_del(&f->fn_hash);
+ hlist_del_rcu(&f->fn_hash);
kill_fn = 1;
}
fib_hash_genid++;
- write_unlock_bh(&fib_hash_lock);
if (fa->fa_state & FA_S_ACCESSED)
rt_cache_flush(cfg->fc_nlinfo.nl_net, -1);
@@ -641,14 +675,12 @@ static int fn_flush_list(struct fn_zone *fz, int idx)
struct fib_info *fi = fa->fa_info;
if (fi && (fi->fib_flags&RTNH_F_DEAD)) {
- write_lock_bh(&fib_hash_lock);
- list_del(&fa->fa_list);
+ list_del_rcu(&fa->fa_list);
if (list_empty(&f->fn_alias)) {
- hlist_del(&f->fn_hash);
+ hlist_del_rcu(&f->fn_hash);
kill_f = 1;
}
fib_hash_genid++;
- write_unlock_bh(&fib_hash_lock);
fn_free_alias(fa, f);
found++;
@@ -693,10 +725,10 @@ fn_hash_dump_bucket(struct sk_buff *skb, struct netlink_callback *cb,
s_i = cb->args[4];
i = 0;
- hlist_for_each_entry(f, node, head, fn_hash) {
+ hlist_for_each_entry_rcu(f, node, head, fn_hash) {
struct fib_alias *fa;
- list_for_each_entry(fa, &f->fn_alias, fa_list) {
+ list_for_each_entry_rcu(fa, &f->fn_alias, fa_list) {
if (i < s_i)
goto next;
@@ -714,7 +746,7 @@ fn_hash_dump_bucket(struct sk_buff *skb, struct netlink_callback *cb,
cb->args[4] = i;
return -1;
}
- next:
+next:
i++;
}
}
@@ -755,7 +787,6 @@ int fib_table_dump(struct fib_table *tb, struct sk_buff *skb,
s_m = cb->args[2];
rcu_read_lock();
- read_lock(&fib_hash_lock);
for (fz = rcu_dereference(table->fn_zone_list);
fz != NULL;
fz = rcu_dereference(fz->fz_next), m++) {
@@ -763,14 +794,12 @@ int fib_table_dump(struct fib_table *tb, struct sk_buff *skb,
continue;
if (fn_hash_dump_zone(skb, cb, tb, fz) < 0) {
cb->args[2] = m;
- read_unlock(&fib_hash_lock);
rcu_read_unlock();
return -1;
}
memset(&cb->args[3], 0,
sizeof(cb->args) - 3*sizeof(cb->args[0]));
}
- read_unlock(&fib_hash_lock);
rcu_read_unlock();
cb->args[2] = m;
return skb->len;
@@ -960,13 +989,11 @@ static struct fib_alias *fib_get_idx(struct seq_file *seq, loff_t pos)
}
static void *fib_seq_start(struct seq_file *seq, loff_t *pos)
- __acquires(fib_hash_lock)
__acquires(RCU)
{
void *v = NULL;
rcu_read_lock();
- read_lock(&fib_hash_lock);
if (fib_get_table(seq_file_net(seq), RT_TABLE_MAIN))
v = *pos ? fib_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
return v;
@@ -979,17 +1006,16 @@ static void *fib_seq_next(struct seq_file *seq, void *v, loff_t *pos)
}
static void fib_seq_stop(struct seq_file *seq, void *v)
- __releases(fib_hash_lock)
__releases(RCU)
{
- read_unlock(&fib_hash_lock);
rcu_read_unlock();
}
static unsigned fib_flag_trans(int type, __be32 mask, struct fib_info *fi)
{
static const unsigned type2flags[RTN_MAX + 1] = {
- [7] = RTF_REJECT, [8] = RTF_REJECT,
+ [7] = RTF_REJECT,
+ [8] = RTF_REJECT,
};
unsigned flags = type2flags[type];
diff --git a/net/ipv4/fib_lookup.h b/net/ipv4/fib_lookup.h
index b9c9a9f..5072d8e 100644
--- a/net/ipv4/fib_lookup.h
+++ b/net/ipv4/fib_lookup.h
@@ -12,9 +12,7 @@ struct fib_alias {
u8 fa_type;
u8 fa_scope;
u8 fa_state;
-#ifdef CONFIG_IP_FIB_TRIE
struct rcu_head rcu;
-#endif
};
#define FA_S_ACCESSED 0x01
^ permalink raw reply related
* [PATCH net-next 2/3] fib_hash: RCU conversion phase 1
From: Eric Dumazet @ 2010-10-15 6:53 UTC (permalink / raw)
To: David Miller; +Cc: netdev
First step for RCU conversion of fib_hash :
struct fn_zone are created and never deleted.
Very classic conversion, using rcu_assign_pointer(), rcu_dereference()
and rtnl_dereference() verbs.
__rcu markers on fz_next and fn_zone_list
They are created under RTNL, we dont need fib_hash_lock anymore in
fn_new_zone().
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
---
net/ipv4/fib_hash.c | 57 ++++++++++++++++++++++++++++--------------
1 files changed, 38 insertions(+), 19 deletions(-)
diff --git a/net/ipv4/fib_hash.c b/net/ipv4/fib_hash.c
index 10001aa..04f05a9 100644
--- a/net/ipv4/fib_hash.c
+++ b/net/ipv4/fib_hash.c
@@ -57,7 +57,7 @@ struct fib_node {
#define EMBEDDED_HASH_SIZE (L1_CACHE_BYTES / sizeof(struct hlist_head))
struct fn_zone {
- struct fn_zone *fz_next; /* Next not empty zone */
+ struct fn_zone __rcu *fz_next; /* Next not empty zone */
struct hlist_head *fz_hash; /* Hash table pointer */
u32 fz_hashmask; /* (fz_divisor - 1) */
@@ -73,8 +73,8 @@ struct fn_zone {
};
struct fn_hash {
- struct fn_zone *fn_zones[33];
- struct fn_zone *fn_zone_list;
+ struct fn_zone *fn_zones[33];
+ struct fn_zone __rcu *fn_zone_list;
};
static inline u32 fn_hash(__be32 key, struct fn_zone *fz)
@@ -219,21 +219,21 @@ fn_new_zone(struct fn_hash *table, int z)
fz->fz_mask = inet_make_mask(z);
/* Find the first not empty zone with more specific mask */
- for (i=z+1; i<=32; i++)
+ for (i = z + 1; i <= 32; i++)
if (table->fn_zones[i])
break;
- write_lock_bh(&fib_hash_lock);
- if (i>32) {
+ if (i > 32) {
/* No more specific masks, we are the first. */
- fz->fz_next = table->fn_zone_list;
- table->fn_zone_list = fz;
+ rcu_assign_pointer(fz->fz_next,
+ rtnl_dereference(table->fn_zone_list));
+ rcu_assign_pointer(table->fn_zone_list, fz);
} else {
- fz->fz_next = table->fn_zones[i]->fz_next;
- table->fn_zones[i]->fz_next = fz;
+ rcu_assign_pointer(fz->fz_next,
+ rtnl_dereference(table->fn_zones[i]->fz_next));
+ rcu_assign_pointer(table->fn_zones[i]->fz_next, fz);
}
table->fn_zones[z] = fz;
fib_hash_genid++;
- write_unlock_bh(&fib_hash_lock);
return fz;
}
@@ -245,8 +245,11 @@ int fib_table_lookup(struct fib_table *tb,
struct fn_zone *fz;
struct fn_hash *t = (struct fn_hash *)tb->tb_data;
+ rcu_read_lock();
read_lock(&fib_hash_lock);
- for (fz = t->fn_zone_list; fz; fz = fz->fz_next) {
+ for (fz = rcu_dereference(t->fn_zone_list);
+ fz != NULL;
+ fz = rcu_dereference(fz->fz_next)) {
struct hlist_head *head;
struct hlist_node *node;
struct fib_node *f;
@@ -267,6 +270,7 @@ int fib_table_lookup(struct fib_table *tb,
err = 1;
out:
read_unlock(&fib_hash_lock);
+ rcu_read_unlock();
return err;
}
@@ -362,6 +366,7 @@ static struct fib_node *fib_find_node(struct fn_zone *fz, __be32 key)
return NULL;
}
+/* Caller must hold RTNL. */
int fib_table_insert(struct fib_table *tb, struct fib_config *cfg)
{
struct fn_hash *table = (struct fn_hash *) tb->tb_data;
@@ -657,13 +662,16 @@ static int fn_flush_list(struct fn_zone *fz, int idx)
return found;
}
+/* caller must hold RTNL. */
int fib_table_flush(struct fib_table *tb)
{
struct fn_hash *table = (struct fn_hash *) tb->tb_data;
struct fn_zone *fz;
int found = 0;
- for (fz = table->fn_zone_list; fz; fz = fz->fz_next) {
+ for (fz = rtnl_dereference(table->fn_zone_list);
+ fz != NULL;
+ fz = rtnl_dereference(fz->fz_next)) {
int i;
for (i = fz->fz_divisor - 1; i >= 0; i--)
@@ -741,23 +749,29 @@ fn_hash_dump_zone(struct sk_buff *skb, struct netlink_callback *cb,
int fib_table_dump(struct fib_table *tb, struct sk_buff *skb,
struct netlink_callback *cb)
{
- int m, s_m;
+ int m = 0, s_m;
struct fn_zone *fz;
struct fn_hash *table = (struct fn_hash *)tb->tb_data;
s_m = cb->args[2];
+ rcu_read_lock();
read_lock(&fib_hash_lock);
- for (fz = table->fn_zone_list, m=0; fz; fz = fz->fz_next, m++) {
- if (m < s_m) continue;
+ for (fz = rcu_dereference(table->fn_zone_list);
+ fz != NULL;
+ fz = rcu_dereference(fz->fz_next), m++) {
+ if (m < s_m)
+ continue;
if (fn_hash_dump_zone(skb, cb, tb, fz) < 0) {
cb->args[2] = m;
read_unlock(&fib_hash_lock);
+ rcu_read_unlock();
return -1;
}
memset(&cb->args[3], 0,
sizeof(cb->args) - 3*sizeof(cb->args[0]));
}
read_unlock(&fib_hash_lock);
+ rcu_read_unlock();
cb->args[2] = m;
return skb->len;
}
@@ -820,8 +834,9 @@ static struct fib_alias *fib_get_first(struct seq_file *seq)
iter->genid = fib_hash_genid;
iter->valid = 1;
- for (iter->zone = table->fn_zone_list; iter->zone;
- iter->zone = iter->zone->fz_next) {
+ for (iter->zone = rcu_dereference(table->fn_zone_list);
+ iter->zone != NULL;
+ iter->zone = rcu_dereference(iter->zone->fz_next)) {
int maxslot;
if (!iter->zone->fz_nent)
@@ -906,7 +921,7 @@ static struct fib_alias *fib_get_next(struct seq_file *seq)
}
}
- iter->zone = iter->zone->fz_next;
+ iter->zone = rcu_dereference(iter->zone->fz_next);
if (!iter->zone)
goto out;
@@ -946,9 +961,11 @@ static struct fib_alias *fib_get_idx(struct seq_file *seq, loff_t pos)
static void *fib_seq_start(struct seq_file *seq, loff_t *pos)
__acquires(fib_hash_lock)
+ __acquires(RCU)
{
void *v = NULL;
+ rcu_read_lock();
read_lock(&fib_hash_lock);
if (fib_get_table(seq_file_net(seq), RT_TABLE_MAIN))
v = *pos ? fib_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
@@ -963,8 +980,10 @@ static void *fib_seq_next(struct seq_file *seq, void *v, loff_t *pos)
static void fib_seq_stop(struct seq_file *seq, void *v)
__releases(fib_hash_lock)
+ __releases(RCU)
{
read_unlock(&fib_hash_lock);
+ rcu_read_unlock();
}
static unsigned fib_flag_trans(int type, __be32 mask, struct fib_info *fi)
^ permalink raw reply related
* Re: tbf/htb qdisc limitations
From: Eric Dumazet @ 2010-10-15 6:44 UTC (permalink / raw)
To: Bill Fink; +Cc: Jarek Poplawski, Rick Jones, Steven Brudenell, netdev
In-Reply-To: <20101015023749.f085006b.billfink@mindspring.com>
Le vendredi 15 octobre 2010 à 02:37 -0400, Bill Fink a écrit :
> On Thu, 14 Oct 2010, Jarek Poplawski wrote:
>
> > On Thu, Oct 14, 2010 at 08:09:39AM +0000, Jarek Poplawski wrote:
> > > On Thu, Oct 14, 2010 at 03:13:54AM -0400, Bill Fink wrote:
> > > > TSO/GSO was disabled and was using 9000-byte jumbo frames
> > > > (and specified mtu 9000 to tc command).
> > > >
> > > > Here was one attempt I made using tbf:
> > > >
> > > > tc qdisc add dev eth2 root handle 1: prio
> > > > tc qdisc add dev eth2 parent 1:1 handle 10: tbf rate 8900mbit buffer 1112500 limit 10000 mtu 9000
> > > > tc filter add dev eth2 protocol ip parent 1: prio 1 u32 match ip dst 192.168.1.23 flowid 10:1
> > > >
> > > > I tried many variations of the above, all without success.
> > >
> > > The main problem are smaller packets. If you had (almost) only 9000b
> > > frames this probably could work. [...]
> >
> > On the other hand, e.g. the limit above seems too low wrt mtu & rate.
>
> Actually, I discovered my commands above work just fine on
> a 2.6.35 box:
>
> i7test7% nuttcp -T10 -i1 192.168.1.17
> 1045.3125 MB / 1.00 sec = 8768.3573 Mbps 0 retrans
> 1045.6875 MB / 1.00 sec = 8772.0292 Mbps 0 retrans
> 1049.5625 MB / 1.00 sec = 8804.2627 Mbps 0 retrans
> 1043.1875 MB / 1.00 sec = 8750.9960 Mbps 0 retrans
> 1048.6875 MB / 1.00 sec = 8796.3246 Mbps 0 retrans
> 1033.4375 MB / 1.00 sec = 8669.3188 Mbps 0 retrans
> 1040.7500 MB / 1.00 sec = 8730.7057 Mbps 0 retrans
> 1047.0000 MB / 1.00 sec = 8783.2063 Mbps 0 retrans
> 1040.0000 MB / 1.00 sec = 8724.0564 Mbps 0 retrans
> 1037.4375 MB / 1.00 sec = 8702.5434 Mbps 0 retrans
>
> 10431.5608 MB / 10.00 sec = 8749.7542 Mbps 25 %TX 35 %RX 0 retrans 0.11 msRTT
>
> The problems I encountered were on a field system running
> 2.6.30.10. I will investigate upgrading the field system
> to 2.6.35.
>
Yes, I noticed same thing for me on net-next-2.6
Please report :
tc -s -d qdisc
^ permalink raw reply
* Re: tbf/htb qdisc limitations
From: Bill Fink @ 2010-10-15 6:37 UTC (permalink / raw)
To: Jarek Poplawski; +Cc: Rick Jones, Steven Brudenell, netdev
In-Reply-To: <20101014085005.GA8349@ff.dom.local>
On Thu, 14 Oct 2010, Jarek Poplawski wrote:
> On Thu, Oct 14, 2010 at 08:09:39AM +0000, Jarek Poplawski wrote:
> > On Thu, Oct 14, 2010 at 03:13:54AM -0400, Bill Fink wrote:
> > > TSO/GSO was disabled and was using 9000-byte jumbo frames
> > > (and specified mtu 9000 to tc command).
> > >
> > > Here was one attempt I made using tbf:
> > >
> > > tc qdisc add dev eth2 root handle 1: prio
> > > tc qdisc add dev eth2 parent 1:1 handle 10: tbf rate 8900mbit buffer 1112500 limit 10000 mtu 9000
> > > tc filter add dev eth2 protocol ip parent 1: prio 1 u32 match ip dst 192.168.1.23 flowid 10:1
> > >
> > > I tried many variations of the above, all without success.
> >
> > The main problem are smaller packets. If you had (almost) only 9000b
> > frames this probably could work. [...]
>
> On the other hand, e.g. the limit above seems too low wrt mtu & rate.
Actually, I discovered my commands above work just fine on
a 2.6.35 box:
i7test7% nuttcp -T10 -i1 192.168.1.17
1045.3125 MB / 1.00 sec = 8768.3573 Mbps 0 retrans
1045.6875 MB / 1.00 sec = 8772.0292 Mbps 0 retrans
1049.5625 MB / 1.00 sec = 8804.2627 Mbps 0 retrans
1043.1875 MB / 1.00 sec = 8750.9960 Mbps 0 retrans
1048.6875 MB / 1.00 sec = 8796.3246 Mbps 0 retrans
1033.4375 MB / 1.00 sec = 8669.3188 Mbps 0 retrans
1040.7500 MB / 1.00 sec = 8730.7057 Mbps 0 retrans
1047.0000 MB / 1.00 sec = 8783.2063 Mbps 0 retrans
1040.0000 MB / 1.00 sec = 8724.0564 Mbps 0 retrans
1037.4375 MB / 1.00 sec = 8702.5434 Mbps 0 retrans
10431.5608 MB / 10.00 sec = 8749.7542 Mbps 25 %TX 35 %RX 0 retrans 0.11 msRTT
The problems I encountered were on a field system running
2.6.30.10. I will investigate upgrading the field system
to 2.6.35.
-Bill
^ permalink raw reply
* [net-next-2.6 PATCH] ixgbe: DCB: remove DCB check config
From: Jeff Kirsher @ 2010-10-15 5:21 UTC (permalink / raw)
To: davem; +Cc: netdev, gospo, bphilips, John Fastabend, Jeff Kirsher
From: John Fastabend <john.r.fastabend@intel.com>
Remove a DCB check config from DCB configuration we
continue to configure DCB even if it fails so don't
even bother to check. Plus user space (lldpad) checks
this before programming the hw anyways.
Worse case is we program some values into the hw that
don't make total sense resulting in incorrect bandwidth
allocation.
Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
---
drivers/net/ixgbe/ixgbe_main.c | 1 -
1 files changed, 0 insertions(+), 1 deletions(-)
diff --git a/drivers/net/ixgbe/ixgbe_main.c b/drivers/net/ixgbe/ixgbe_main.c
index 95dbf60..790a0da 100644
--- a/drivers/net/ixgbe/ixgbe_main.c
+++ b/drivers/net/ixgbe/ixgbe_main.c
@@ -3374,7 +3374,6 @@ static void ixgbe_configure_dcb(struct ixgbe_adapter *adapter)
if (hw->mac.type == ixgbe_mac_82598EB)
netif_set_gso_max_size(adapter->netdev, 32768);
- ixgbe_dcb_check_config(&adapter->dcb_cfg);
ixgbe_dcb_calculate_tc_credits(&adapter->dcb_cfg, DCB_TX_CONFIG);
ixgbe_dcb_calculate_tc_credits(&adapter->dcb_cfg, DCB_RX_CONFIG);
^ permalink raw reply related
* ixgbe: net-next build breakage
From: Harvey Harrison @ 2010-10-15 4:33 UTC (permalink / raw)
To: emil.s.tantilov
Cc: jeffrey.t.kirsher, stephen.s.ko, shemminger, David Miller, netdev
commit f32f837b75233588cd4f8542214a30915ab7847b
Author: Emil Tantilov <emil.s.tantilov@intel.com>
Date: Tue Oct 12 22:20:34 2010 +0000
ixgbe: remove unused functions
Breaks the build:
drivers/net/ixgbe/ixgbe_main.c: In function ‘ixgbe_configure_dcb’:
drivers/net/ixgbe/ixgbe_main.c:3377: error: implicit declaration of
function ‘ixgbe_dcb_check_config’
make[2]: *** [drivers/net/ixgbe/ixgbe_main.o] Error 1
make[1]: *** [drivers/net/ixgbe] Error 2
make: *** [drivers/net/] Error 2
As ixgbe_dcb_check_config no longer exists in the tree.
Cheers,
Harvey
^ permalink raw reply
* [PATCHv2 net-2.6] r6040: Fix multicast filter some more
From: Ben Hutchings @ 2010-10-15 3:41 UTC (permalink / raw)
To: David Miller; +Cc: Florian Fainelli, netdev
In-Reply-To: <1287112636.20865.13.camel@localhost>
This code has been broken forever, but in several different and
creative ways.
So far as I can work out, the R6040 MAC filter has 4 exact-match
entries, the first of which the driver uses for its assigned unicast
address, plus a 64-entry hash-based filter for multicast addresses
(maybe unicast as well?).
The original version of this code would write the first 4 multicast
addresses as exact-match entries from offset 1 (bug #1: there is no
entry 4 so this could write to some PHY registers). It would fill the
remainder of the exact-match entries with the broadcast address (bug #2:
this would overwrite the last used entry). If more than 4 multicast
addresses were configured, it would set up the hash table, write some
random crap to the MAC control register (bug #3) and finally walk off
the end of the list when filling the exact-match entries (bug #4).
All of this seems to be pointless, since it sets the promiscuous bit
when the interface is made promiscuous or if >4 multicast addresses
are enabled, and never clears it (bug #5, masking bug #2).
The recent(ish) changes to the multicast list fixed bug #4, but
completely removed the limit on iteration over the exact-match entries
(bug #6).
Bug #4 was reported as
<https://bugzilla.kernel.org/show_bug.cgi?id=15355> and more recently
as <http://bugs.debian.org/600155>. Florian Fainelli attempted to fix
these in commit 3bcf8229a8c49769e48d3e0bd1e20d8e003f8106, but that
actually dealt with bugs #1-3, bug #4 having been fixed in mainline at
that point.
That commit fixes the most important current bug #6.
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Cc: stable@kernel.org [2.6.35 only]
---
Commit message was slightly mangled in the previous message.
Ben.
drivers/net/r6040.c | 22 ++++++++++++----------
1 files changed, 12 insertions(+), 10 deletions(-)
diff --git a/drivers/net/r6040.c b/drivers/net/r6040.c
index 142c381..80666f0 100644
--- a/drivers/net/r6040.c
+++ b/drivers/net/r6040.c
@@ -893,16 +893,18 @@ static void r6040_multicast_list(struct net_device *dev)
/* Multicast Address 1~4 case */
i = 0;
netdev_for_each_mc_addr(ha, dev) {
- if (i < MCAST_MAX) {
- adrp = (u16 *) ha->addr;
- iowrite16(adrp[0], ioaddr + MID_1L + 8 * i);
- iowrite16(adrp[1], ioaddr + MID_1M + 8 * i);
- iowrite16(adrp[2], ioaddr + MID_1H + 8 * i);
- } else {
- iowrite16(0xffff, ioaddr + MID_1L + 8 * i);
- iowrite16(0xffff, ioaddr + MID_1M + 8 * i);
- iowrite16(0xffff, ioaddr + MID_1H + 8 * i);
- }
+ if (i >= MCAST_MAX)
+ break;
+ adrp = (u16 *) ha->addr;
+ iowrite16(adrp[0], ioaddr + MID_1L + 8 * i);
+ iowrite16(adrp[1], ioaddr + MID_1M + 8 * i);
+ iowrite16(adrp[2], ioaddr + MID_1H + 8 * i);
+ i++;
+ }
+ while (i < MCAST_MAX) {
+ iowrite16(0xffff, ioaddr + MID_1L + 8 * i);
+ iowrite16(0xffff, ioaddr + MID_1M + 8 * i);
+ iowrite16(0xffff, ioaddr + MID_1H + 8 * i);
i++;
}
}
--
1.7.1
^ permalink raw reply related
* [PATCH net-2.6] r6040: Fix multicast filter some more
From: Ben Hutchings @ 2010-10-15 3:17 UTC (permalink / raw)
To: David Miller, Florian Fainelli; +Cc: netdev
This code has been broken forever, but in several different and
creative ways.
So far as I can work out, the R6040 MAC filter has 4 exact-match
entries, the first of which the driver uses for its assigned unicast
address, plus a 64-entry hash-based filter for multicast addresses
(maybe unicast as well?).
The original version of this code would write the first 4 multicast
addresses as exact-match entries from offset 1 (bug #1: there is no
entry 4 so this could write to some PHY registers). It would fill the
remainder of the exact-match entries with the broadcast address (bug
addresses were configured, it would set up the hash table, write some
random crap to the MAC control register (bug #3) and finally walk off
the end of the list when filling the exact-match entries (bug #4).
All of this seems to be pointless, since it sets the promiscuous bit
when the interface is made promiscuous or if >4 multicast addresses
are enabled, and never clears it (bug #5, masking bug #2).
The recent(ish) changes to the multicast list fixed bug #4, but
completely removed the limit on iteration over the exact-match entries
(bug #6).
Bug #4 was reported as
<https://bugzilla.kernel.org/show_bug.cgi?id=15355> and more recently
as <http://bugs.debian.org/600155>. Florian Fainelli attempted to fix
these in commit 3bcf8229a8c49769e48d3e0bd1e20d8e003f8106, but that
actually dealt with bugs #1-3, bug #4 having been fixed in mainline at
that point.
That commit fixes the most important current bug #6.
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Cc: stable@kernel.org [2.6.35 only]
---
Compile-tested only.
Ben.
drivers/net/r6040.c | 22 ++++++++++++----------
1 files changed, 12 insertions(+), 10 deletions(-)
diff --git a/drivers/net/r6040.c b/drivers/net/r6040.c
index 142c381..80666f0 100644
--- a/drivers/net/r6040.c
+++ b/drivers/net/r6040.c
@@ -893,16 +893,18 @@ static void r6040_multicast_list(struct net_device *dev)
/* Multicast Address 1~4 case */
i = 0;
netdev_for_each_mc_addr(ha, dev) {
- if (i < MCAST_MAX) {
- adrp = (u16 *) ha->addr;
- iowrite16(adrp[0], ioaddr + MID_1L + 8 * i);
- iowrite16(adrp[1], ioaddr + MID_1M + 8 * i);
- iowrite16(adrp[2], ioaddr + MID_1H + 8 * i);
- } else {
- iowrite16(0xffff, ioaddr + MID_1L + 8 * i);
- iowrite16(0xffff, ioaddr + MID_1M + 8 * i);
- iowrite16(0xffff, ioaddr + MID_1H + 8 * i);
- }
+ if (i >= MCAST_MAX)
+ break;
+ adrp = (u16 *) ha->addr;
+ iowrite16(adrp[0], ioaddr + MID_1L + 8 * i);
+ iowrite16(adrp[1], ioaddr + MID_1M + 8 * i);
+ iowrite16(adrp[2], ioaddr + MID_1H + 8 * i);
+ i++;
+ }
+ while (i < MCAST_MAX) {
+ iowrite16(0xffff, ioaddr + MID_1L + 8 * i);
+ iowrite16(0xffff, ioaddr + MID_1M + 8 * i);
+ iowrite16(0xffff, ioaddr + MID_1H + 8 * i);
i++;
}
}
--
1.7.1
^ permalink raw reply related
* Re: [RFC PATCH 3/7] bnx2: Update bnx2 to use new vlan accleration.
From: Jesse Gross @ 2010-10-15 1:36 UTC (permalink / raw)
To: Michael Chan; +Cc: davem@davemloft.net, netdev@vger.kernel.org
In-Reply-To: <1287096960.14523.31.camel@nseg_linux_HP1.broadcom.com>
On Thu, Oct 14, 2010 at 3:56 PM, Michael Chan <mchan@broadcom.com> wrote:
>> if ((status & L2_FHDR_STATUS_L2_VLAN_TAG) &&
>> - !(bp->rx_mode & BNX2_EMAC_RX_MODE_KEEP_VLAN_TAG)) {
>> - vtag = rx_hdr->l2_fhdr_vlan_tag;
>> -#ifdef BCM_VLAN
>> - if (bp->vlgrp)
>> - hw_vlan = 1;
>> - else
>> -#endif
>> - {
>> - struct vlan_ethhdr *ve = (struct vlan_ethhdr *)
>> - __skb_push(skb, 4);
>> -
>> - memmove(ve, skb->data + 4, ETH_ALEN * 2);
>> - ve->h_vlan_proto = htons(ETH_P_8021Q);
>> - ve->h_vlan_TCI = htons(vtag);
>> - len += 4;
>> - }
>> - }
>> + !(bp->rx_mode & BNX2_EMAC_RX_MODE_KEEP_VLAN_TAG))
>
> This check for the rx_mode bit is no longer necessary if we always
> configure the chip to strip out the vlan tag.
Thanks, I've dropped it from my local tree. I may add support for
ethtool to disable vlan stripping in the future, so I might have to
add it back.
^ permalink raw reply
* Re: [PATCH net-next 3/5] tipc: Optimizations to bearer enabling logic
From: Paul Gortmaker @ 2010-10-15 1:11 UTC (permalink / raw)
To: Neil Horman; +Cc: davem, netdev, allan.stephens
In-Reply-To: <20101013145843.GD31379@hmsreliant.think-freely.org>
[Re: [PATCH net-next 3/5] tipc: Optimizations to bearer enabling logic] On 13/10/2010 (Wed 10:58) Neil Horman wrote:
> On Tue, Oct 12, 2010 at 08:25:56PM -0400, Paul Gortmaker wrote:
> > From: Allan Stephens <allan.stephens@windriver.com>
> >
> > Introduces "enabling" state during activation of a new TIPC bearer,
> > which supplements the existing "disabled" and "enabled" states.
> > This change allows the new bearer to be added without having to
> > temporarily block the processing of incoming packets on existing
> > bearers during the binding of the new bearer to its associated
> > interface. It also makes it unnecessary to zero out the entire
> > bearer structure at the start of activation.
> >
[...]
> > + b_ptr->state = BEARER_ENABLING;
> > strcpy(b_ptr->publ.name, name);
> > + b_ptr->priority = priority;
> > +
> > + write_unlock_bh(&tipc_net_lock);
> Why the 3rd state? Doesn't seem needed.
I'm a bit disappointed in myself for also not noticing that it
was set but never tested for. The following should give the
same end result but without the obfuscation of an extra state.
This one also doesn't explicitly depend on any other changes,
so if it is now OK, the option is there for it to be applied
independently of the others that haven't been reworked yet.
Thanks,
Paul.
>From 86d0d5c92439d0a3f5a0f165aa8bd842d377dae9 Mon Sep 17 00:00:00 2001
From: Allan Stephens <allan.stephens@windriver.com>
Date: Thu, 14 Oct 2010 16:09:23 -0400
Subject: [PATCH] tipc: Optimizations to bearer enabling logic
Allow new bearers to be added without having to temporarily block
the processing of incoming packets on existing bearers during the
binding of the new bearer to its associated interface. Eliminates
zeroing out of the new bearer structure at the start of activation,
since it is already in that state.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/bearer.c | 7 ++++---
1 files changed, 4 insertions(+), 3 deletions(-)
diff --git a/net/tipc/bearer.c b/net/tipc/bearer.c
index fd9c06c..2ff8181 100644
--- a/net/tipc/bearer.c
+++ b/net/tipc/bearer.c
@@ -556,14 +556,15 @@ restart:
}
b_ptr = &tipc_bearers[bearer_id];
- memset(b_ptr, 0, sizeof(struct bearer));
-
strcpy(b_ptr->publ.name, name);
+
+ write_unlock_bh(&tipc_net_lock);
res = m_ptr->enable_bearer(&b_ptr->publ);
if (res) {
warn("Bearer <%s> rejected, enable failure (%d)\n", name, -res);
- goto failed;
+ return res;
}
+ write_lock_bh(&tipc_net_lock);
b_ptr->identity = bearer_id;
b_ptr->media = m_ptr;
--
1.7.2.1
^ permalink raw reply related
* linux-next: build failure after merge of the net tree
From: Stephen Rothwell @ 2010-10-15 0:38 UTC (permalink / raw)
To: David Miller, netdev
Cc: linux-next, linux-kernel, Emil Tantilov, Jeff Kirsher
[-- Attachment #1: Type: text/plain, Size: 569 bytes --]
Hi all,
After merging the net tree, today's linux-next build (x86_64 allmodconfig)
failed like this:
drivers/net/ixgbe/ixgbe_main.c: In function 'ixgbe_configure_dcb':
drivers/net/ixgbe/ixgbe_main.c:3377: error: implicit declaration of function 'ixgbe_dcb_check_config'
Caused by commit f32f837b75233588cd4f8542214a30915ab7847b ("ixgbe: remove
unused functions").
Grep is your friend :-)
I have used the net tree from next-20101014 for today.
--
Cheers,
Stephen Rothwell sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/
[-- Attachment #2: Type: application/pgp-signature, Size: 490 bytes --]
^ permalink raw reply
* [PATCH net-next] 3c52x: remove IRQF_SAMPLE_RANDOM from legacy MCA drivers.
From: Paul Gortmaker @ 2010-10-15 0:21 UTC (permalink / raw)
To: davem; +Cc: netdev
If you are genuinely using one of these legacy MCA drivers
then you are tragically on hardware where you really don't
have the extra CPU cycles to be wasting on this.
In addition, it makes two less cases for people to inadvertently
blindly copy flags from without explicitly thinking whether it
makes sense -- see the addition to feature-removal.txt as per
commit 9d9b8fb0e5ebf4b0398e579f6061d4451fea3242.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
drivers/net/3c523.c | 2 +-
drivers/net/3c527.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/3c523.c b/drivers/net/3c523.c
index 1719079..3aa9baa 100644
--- a/drivers/net/3c523.c
+++ b/drivers/net/3c523.c
@@ -287,7 +287,7 @@ static int elmc_open(struct net_device *dev)
elmc_id_attn586(); /* disable interrupts */
- ret = request_irq(dev->irq, elmc_interrupt, IRQF_SHARED | IRQF_SAMPLE_RANDOM,
+ ret = request_irq(dev->irq, elmc_interrupt, IRQF_SHARED,
dev->name, dev);
if (ret) {
pr_err("%s: couldn't get irq %d\n", dev->name, dev->irq);
diff --git a/drivers/net/3c527.c b/drivers/net/3c527.c
index 5c07b14..cf4cfd3 100644
--- a/drivers/net/3c527.c
+++ b/drivers/net/3c527.c
@@ -443,7 +443,7 @@ static int __init mc32_probe1(struct net_device *dev, int slot)
* Grab the IRQ
*/
- err = request_irq(dev->irq, mc32_interrupt, IRQF_SHARED | IRQF_SAMPLE_RANDOM, DRV_NAME, dev);
+ err = request_irq(dev->irq, mc32_interrupt, IRQF_SHARED, DRV_NAME, dev);
if (err) {
release_region(dev->base_addr, MC32_IO_EXTENT);
pr_err("%s: unable to get IRQ %d.\n", DRV_NAME, dev->irq);
--
1.7.2.1
^ permalink raw reply related
* Re: [PATCH net-next 2/5] tipc: Simplify bearer shutdown logic
From: Paul Gortmaker @ 2010-10-14 23:58 UTC (permalink / raw)
To: Neil Horman; +Cc: davem, netdev, allan.stephens
In-Reply-To: <20101013143922.GC31379@hmsreliant.think-freely.org>
[Re: [PATCH net-next 2/5] tipc: Simplify bearer shutdown logic] On 13/10/2010 (Wed 10:39) Neil Horman wrote:
> On Tue, Oct 12, 2010 at 08:25:55PM -0400, Paul Gortmaker wrote:
> > From: Allan Stephens <allan.stephens@windriver.com>
> >
> > Disable all active bearers when TIPC is shut down without having to do
> > a name-based search to locate each bearer object.
> >
> It seems like you're doing a good deal more in this patch than just disabling
> all active bearers without doing a name search. The description is implemented
> in the for loop of tipc_bearer_stop. Whats the rest of it for?
It seems the original needlessly bloated out the patch size by
swapping the order of tipc_bearer_find_interface & bearer_find
in the file (now fixed) - and you are right, the locking change
wasn't properly covered in the commit log. The extra test you'd
suggested tossing out is also now gone.
This change doesn't explicitly depend on any other changes,
so if it is now OK, the option is there for it to be applied
independently of the others that haven't been reworked yet.
Thanks,
Paul.
>From 1771ad642cb076dbeb71e3533a25cb2f07df9cd8 Mon Sep 17 00:00:00 2001
From: Allan Stephens <allan.stephens@windriver.com>
Date: Sat, 4 Sep 2010 09:29:04 -0400
Subject: [PATCH] tipc: Simplify bearer shutdown logic
Optimize processing in TIPC's bearer shutdown code, including:
1. Remove an unnecessary check to see if TIPC bearer's can exist.
2. Don't release spinlocks before calling a media-specific disabling
routine, since the routine can't sleep.
3. Make bearer_disable() operate directly on a struct bearer, instead
of needlessly taking a name and then mapping that to the struct.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
---
net/tipc/bearer.c | 38 +++++++++++---------------------------
1 files changed, 11 insertions(+), 27 deletions(-)
diff --git a/net/tipc/bearer.c b/net/tipc/bearer.c
index 9c10c6b..fd9c06c 100644
--- a/net/tipc/bearer.c
+++ b/net/tipc/bearer.c
@@ -288,9 +288,6 @@ static struct bearer *bearer_find(const char *name)
struct bearer *b_ptr;
u32 i;
- if (tipc_mode != TIPC_NET_MODE)
- return NULL;
-
for (i = 0, b_ptr = tipc_bearers; i < MAX_BEARERS; i++, b_ptr++) {
if (b_ptr->active && (!strcmp(b_ptr->publ.name, name)))
return b_ptr;
@@ -630,30 +627,17 @@ int tipc_block_bearer(const char *name)
* Note: This routine assumes caller holds tipc_net_lock.
*/
-static int bearer_disable(const char *name)
+static int bearer_disable(struct bearer *b_ptr)
{
- struct bearer *b_ptr;
struct link *l_ptr;
struct link *temp_l_ptr;
- b_ptr = bearer_find(name);
- if (!b_ptr) {
- warn("Attempt to disable unknown bearer <%s>\n", name);
- return -EINVAL;
- }
-
- info("Disabling bearer <%s>\n", name);
+ info("Disabling bearer <%s>\n", b_ptr->publ.name);
tipc_disc_stop_link_req(b_ptr->link_req);
spin_lock_bh(&b_ptr->publ.lock);
b_ptr->link_req = NULL;
b_ptr->publ.blocked = 1;
- if (b_ptr->media->disable_bearer) {
- spin_unlock_bh(&b_ptr->publ.lock);
- write_unlock_bh(&tipc_net_lock);
- b_ptr->media->disable_bearer(&b_ptr->publ);
- write_lock_bh(&tipc_net_lock);
- spin_lock_bh(&b_ptr->publ.lock);
- }
+ b_ptr->media->disable_bearer(&b_ptr->publ);
list_for_each_entry_safe(l_ptr, temp_l_ptr, &b_ptr->links, link_list) {
tipc_link_delete(l_ptr);
}
@@ -664,10 +648,16 @@ static int bearer_disable(const char *name)
int tipc_disable_bearer(const char *name)
{
+ struct bearer *b_ptr;
int res;
write_lock_bh(&tipc_net_lock);
- res = bearer_disable(name);
+ b_ptr = bearer_find(name);
+ if (b_ptr == NULL) {
+ warn("Attempt to disable unknown bearer <%s>\n", name);
+ res = -EINVAL;
+ } else
+ res = bearer_disable(b_ptr);
write_unlock_bh(&tipc_net_lock);
return res;
}
@@ -680,13 +670,7 @@ void tipc_bearer_stop(void)
for (i = 0; i < MAX_BEARERS; i++) {
if (tipc_bearers[i].active)
- tipc_bearers[i].publ.blocked = 1;
- }
- for (i = 0; i < MAX_BEARERS; i++) {
- if (tipc_bearers[i].active)
- bearer_disable(tipc_bearers[i].publ.name);
+ bearer_disable(&tipc_bearers[i]);
}
media_count = 0;
}
-
-
--
1.7.2.1
^ permalink raw reply related
* Couple tc filter questions.
From: Jonathan Thibault @ 2010-10-14 23:31 UTC (permalink / raw)
To: netdev
Since the lartc mailing list appears to be dead, I'll ask here and hope not to offend anyone.
1- This page:
http://lartc.org/howto/lartc.qdisc.filters.html
States: "Also, with HTB, you should attach all filters to the root!"
Why? Is it still true? My setup would be a lot easier with cascading filters. If it's just a matter of there not being any efficiency gains from cascading filters, that's fine. If there is a risk of things exploding randomly and without notice, I'd be keen to know. Testing shows that cascading works okay, but I haven't tried under any serious load.
2- Are filter flowid (classify) actions terminating? Meaning if two consecutive filters would match the same packet, only the first match would ever apply and no further filter is evaluated? Are there actions for which this isn't the case? Intuitively and experimentally, I'd answer no but if anyone knowledgeable in the matter would care to expand on that topic I'd be grateful. Especially considering cascading classes/filters.
Another area where termination isn't entirely clear is when using mirred and ifb devices. I might want to send a copy of all my traffic to an ifb device, but then I would still want subsequent filters to match in the current qdisc. In such a case, a filter that matches all traffic with a mirred action should probably not be terminating.
Maybe I'm thinking too much in terms of iptables here :P
Jonathan
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox