* [PATCH] [1/2] Skip empty hash buckets faster in /proc/net/tcp
From: Andi Kleen @ 2008-01-30 8:01 UTC (permalink / raw)
To: meissner, netdev, davem
On most systems most of the TCP established/time-wait hash buckets are empty.
When walking the hash table for /proc/net/tcp their read locks would
always be aquired just to find out they're empty. This patch changes the code
to check first if the buckets have any entries before taking the lock, which
is much cheaper than taking a lock. Since the hash tables are large
this makes a measurable difference on processing /proc/net/tcp,
especially on architectures with slow read_lock (e.g. PPC)
On a 2GB Core2 system here I see a time cat /proc/net/tcp > /dev/null
constently dropping from 0.44s to 0.4-0.8s system time with this change.
This is with mostly empty hash tables.
On systems with slower atomics (like P4 or POWER4) or larger hash tables
(more RAM) the difference is much higher.
This can be noticeable because there are some daemons around who regularly
scan /proc/net/tcp.
Original idea for this patch from Marcus Meissner, but redone by me.
Cc: meissner@suse.de
Signed-off-by: Andi Kleen <ak@suse.de>
---
net/ipv4/tcp_ipv4.c | 30 ++++++++++++++++++------------
1 file changed, 18 insertions(+), 12 deletions(-)
Index: linux/net/ipv4/tcp_ipv4.c
===================================================================
--- linux.orig/net/ipv4/tcp_ipv4.c
+++ linux/net/ipv4/tcp_ipv4.c
@@ -2039,6 +2039,12 @@ static void *listening_get_idx(struct se
return rc;
}
+static inline int empty_bucket(struct tcp_iter_state *st)
+{
+ return hlist_empty(&tcp_hashinfo.ehash[st->bucket].chain) &&
+ hlist_empty(&tcp_hashinfo.ehash[st->bucket].twchain);
+}
+
static void *established_get_first(struct seq_file *seq)
{
struct tcp_iter_state* st = seq->private;
@@ -2050,6 +2056,10 @@ static void *established_get_first(struc
struct inet_timewait_sock *tw;
rwlock_t *lock = inet_ehash_lockp(&tcp_hashinfo, st->bucket);
+ /* Lockless fast path for the common case of empty buckets */
+ if (empty_bucket(st))
+ continue;
+
read_lock_bh(lock);
sk_for_each(sk, node, &tcp_hashinfo.ehash[st->bucket].chain) {
if (sk->sk_family != st->family) {
@@ -2097,13 +2107,15 @@ get_tw:
read_unlock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
st->state = TCP_SEQ_STATE_ESTABLISHED;
- if (++st->bucket < tcp_hashinfo.ehash_size) {
- read_lock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
- sk = sk_head(&tcp_hashinfo.ehash[st->bucket].chain);
- } else {
- cur = NULL;
- goto out;
- }
+ /* Look for next non empty bucket */
+ while (++st->bucket < tcp_hashinfo.ehash_size &&
+ empty_bucket(st))
+ ;
+ if (st->bucket >= tcp_hashinfo.ehash_size)
+ return NULL;
+
+ read_lock_bh(inet_ehash_lockp(&tcp_hashinfo, st->bucket));
+ sk = sk_head(&tcp_hashinfo.ehash[st->bucket].chain);
} else
sk = sk_next(sk);
^ permalink raw reply
* Re: [PATCH] [2/2] Remove some unnecessary gotos in established_get_first()
From: Oliver Neukum @ 2008-01-30 8:25 UTC (permalink / raw)
To: Andi Kleen; +Cc: netdev, davem
In-Reply-To: <20080130080110.35B691B416F@basil.firstfloor.org>
Am Mittwoch, 30. Januar 2008 09:01:10 schrieb Andi Kleen:
>
> gcc does not generate different code for return foo vs bar = foo; goto x;
> x: return bar; So convert it all to direct returns for better readability.
Now suppose somebody needs to change locking. He'll have to convert
it back. IMHO a conditional return is worse than "goto clearly_named_label"
Regards
Oliver
^ permalink raw reply
* Re: sis190 build breakage
From: Jeff Garzik @ 2008-01-30 8:22 UTC (permalink / raw)
To: Sam Ravnborg; +Cc: Francois Romieu, maximilian attems, netdev
In-Reply-To: <20080130032838.GA17881@uranus.ravnborg.org>
Sam Ravnborg wrote:
> On Tue, Jan 29, 2008 at 11:03:10PM +0100, Francois Romieu wrote:
>> maximilian attems <max@stro.at> :
>>> CC [M] drivers/net/sis190.o
>>> drivers/net/sis190.c:329: error: sis190_pci_tbl causes a section type conflict
>>> make[5]: *** [drivers/net/sis190.o] Error 1
>>>
>>> gcc --version
>>> gcc (GCC) 4.1.2 20061115 (prerelease) (Debian 4.1.1-21)
>
> Looks like a bug where __initdata has been used
> for const data.
> Searching:
> static int __devinit sis190_get_mac_addr_from_apc(struct pci_dev *pdev,
> struct net_device *dev)
> {
> static const u16 __devinitdata ids[] = { 0x0965, 0x0966, 0x0968 };
> struct sis190_private *tp = netdev_priv(dev);
> struct pci_dev *isa_bridge;
> u8 reg, tmp8;
>
> Try to change this is __initconst and it should be fixed.
We have __initconst now?
Three cheers, and a beer, to whomever did that...
Jeff
^ permalink raw reply
* Re: [PATCH] [2/2] Remove some unnecessary gotos in established_get_first()
From: Andi Kleen @ 2008-01-30 8:27 UTC (permalink / raw)
To: Oliver Neukum; +Cc: netdev, davem
In-Reply-To: <200801300925.12397.oliver@neukum.org>
Oliver Neukum <oliver@neukum.org> writes:
> Am Mittwoch, 30. Januar 2008 09:01:10 schrieb Andi Kleen:
>>
>> gcc does not generate different code for return foo vs bar = foo; goto x;
>> x: return bar; So convert it all to direct returns for better readability.
>
> Now suppose somebody needs to change locking. He'll have to convert
> it back.
Please take a look at the overall /proc/net/tcp logic. Any locking
change will be a major change to the code flow of the whole family
of funtions.
-Andi
^ permalink raw reply
* Re: [PATCH 1/7] bonding: fix parameter parsing
From: Jeff Garzik @ 2008-01-30 8:38 UTC (permalink / raw)
To: Jay Vosburgh; +Cc: netdev
In-Reply-To: <12016588701127-git-send-email-fubar@us.ibm.com>
Jay Vosburgh wrote:
> My last fix (commit ece95f7fefe3afae19e641e1b3f5e64b00d5b948)
> didn't handle one case correctly. This resolves that, and it will now
> correctly parse parameters with arbitrary white space, and either text
> names or mode values.
>
> Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
> ---
> drivers/net/bonding/bond_main.c | 17 +++++++++++------
> 1 files changed, 11 insertions(+), 6 deletions(-)
applied 1-7
^ permalink raw reply
* Re: [PATCH] cxgb3: Remove incorrect __devinit annotations
From: Jeff Garzik @ 2008-01-30 8:38 UTC (permalink / raw)
To: Roland Dreier; +Cc: Divy Le Ray, netdev
In-Reply-To: <adak5lsdzew.fsf@cisco.com>
Roland Dreier wrote:
> When PCI error recovery was added to cxgb3, a function t3_io_slot_reset()
> was added. This function can call back into t3_prep_adapter() at any
> time, so t3_prep_adapter() can no longer be marked __devinit.
> This patch removes the __devinit annotation from t3_prep_adapter() and
> all the functions that it calls, which fixes
>
> WARNING: drivers/net/cxgb3/built-in.o(.text+0x2427): Section mismatch in reference from the function t3_io_slot_reset() to the function .devinit.text:t3_prep_adapter()
>
> Signed-off-by: Roland Dreier <rolandd@cisco.com>
> ---
> drivers/net/cxgb3/mc5.c | 2 +-
> drivers/net/cxgb3/sge.c | 2 +-
> drivers/net/cxgb3/t3_hw.c | 22 ++++++++++------------
> 3 files changed, 12 insertions(+), 14 deletions(-)
applied
^ permalink raw reply
* Re: [PATCH 1/1 resend][arm/at91_ether.c] logical/bitand typo in function reset_phy() (inactive), drivers/net/arm/at91_ether.c
From: Jeff Garzik @ 2008-01-30 8:39 UTC (permalink / raw)
To: Roel Kluin; +Cc: andrew, netdev, linux-arm-kernel
In-Reply-To: <479F8940.7030301@tiscali.nl>
Roel Kluin wrote:
> include/linux/mii.h:48:#define BMCR_RESET 0x8000
>
> The function reset_phy() is in "#if 0" inactivated code
> --
> Replace logical "&&" by bit "&" before BMCR_RESET
>
> Signed-off-by: Roel Kluin <12o3l@tiscali.nl>
> ---
> diff --git a/drivers/net/arm/at91_ether.c b/drivers/net/arm/at91_ether.c
> index 25b114a..0ae0d83 100644
> --- a/drivers/net/arm/at91_ether.c
> +++ b/drivers/net/arm/at91_ether.c
> @@ -384,7 +384,7 @@ static void reset_phy(struct net_device *dev)
> /* Wait until PHY reset is complete */
> do {
> read_phy(lp->phy_address, MII_BMCR, &bmcr);
> - } while (!(bmcr && BMCR_RESET));
> + } while (!(bmcr & BMCR_RESET));
>
> disable_mdi();
> spin_unlock_irq(&lp->lock);
applied
^ permalink raw reply
* Re: [PATCH 1/5] forcedeth: reset register fix
From: Jeff Garzik @ 2008-01-30 8:39 UTC (permalink / raw)
To: Ayaz Abdulla; +Cc: Manfred Spraul, Andrew Morton, nedev
In-Reply-To: <478A7C72.8060903@nvidia.com>
Ayaz Abdulla wrote:
> This patch fixes the reset register definition from 0x3C to 0x34.
>
> Signed-off-by: Ayaz Abdulla <aabdulla@nvidia.com>
applied 1-5
^ permalink raw reply
* Re: [PATCH] [NET]: Remove PowerPC code from fec.c
From: Jeff Garzik @ 2008-01-30 8:41 UTC (permalink / raw)
To: Jochen Friedrich
Cc: Vitaly Bordug, Scott Wood, Kumar Gala, Geert Uytterhoeven,
Kernel, Linux, netdev@vger.kernel.org, linuxppc-dev list,
linux-m68k
In-Reply-To: <4799F349.9090102@scram.de>
Jochen Friedrich wrote:
> fec.c is only used on M68k Coldfire CPUs. Remove leftover
> PowerPC code from this driver.
>
> Signed-off-by: Jochen Friedrich <jochen@scram.de>
> ---
> drivers/net/fec.c | 136 +---------------------------------------------------
> 1 files changed, 3 insertions(+), 133 deletions(-)
Seems OK to me, but I feel I don't have enough knowledge to ACK or NAK.
Please pass through an arch tree, thanks.
Jeff
^ permalink raw reply
* Re: [PATCH] forcedeth: mac address mcp77/79
From: Jeff Garzik @ 2008-01-30 8:48 UTC (permalink / raw)
To: Ayaz Abdulla; +Cc: Andrew Morton, nedev, stable
In-Reply-To: <479DF3B8.2000204@nvidia.com>
Ayaz Abdulla wrote:
> This patch is a critical fix for MCP77 and MCP79 devices. The feature
> flags were missing the define for correct mac address
> (DEV_HAS_CORRECT_MACADDR).
>
> Signed-off-by: Ayaz Abdulla <aabdulla@nvidia.com>
applied (upstream)
^ permalink raw reply
* Re: [PATCH] PHYLIB: Locking fixes for PHY I/O potentially sleeping
From: Jeff Garzik @ 2008-01-30 8:48 UTC (permalink / raw)
To: Nate Case; +Cc: Andy Fleming, David S. Miller, netdev
In-Reply-To: <1201622709.12444.118.camel@localhost.localdomain>
Nate Case wrote:
> PHY read/write functions can potentially sleep (e.g., a PHY accessed
> via I2C). The following changes were made to account for this:
>
> * Change spin locks to mutex locks
> * Add a BUG_ON() to phy_read() phy_write() to warn against
> calling them from an interrupt context.
> * Use work queue for PHY state machine handling since
> it can potentially sleep
> * Change phydev lock from spinlock to mutex
>
> Signed-off-by: Nate Case <ncase@xes-inc.com>
> Acked-by: Andy Fleming <afleming@freescale.com>
>
> ---
> Note: This is a resend of the patch submitted on January 3rd, 2008
>
> drivers/net/phy/mdio_bus.c | 2 +-
> drivers/net/phy/phy.c | 68 ++++++++++++++++++++++++++++-------------
> drivers/net/phy/phy_device.c | 11 +++----
> include/linux/phy.h | 5 ++-
> 4 files changed, 55 insertions(+), 31 deletions(-)
applied
^ permalink raw reply
* Re: [PATCH] natsemi: Update locking documentation
From: Jeff Garzik @ 2008-01-30 8:48 UTC (permalink / raw)
To: Mark Brown; +Cc: Tim Hockin, netdev, linux-kernel
In-Reply-To: <1201442291-13668-1-git-send-email-broonie@sirena.org.uk>
Mark Brown wrote:
> The documentation regarding synchronisation at the head of the natsemi
> driver was badly bitrotted so replace it with a general statement about
> the techniques used which is less likely to bitrot.
>
> Also remove the note saying these chips are uncommon - it makes little
> difference but they were used in a number of laptops and at least one mass
> market PCI ethernet card.
>
> Signed-off-by: Mark Brown <broonie@sirena.org.uk>
> ---
> drivers/net/natsemi.c | 18 ++----------------
> 1 files changed, 2 insertions(+), 16 deletions(-)
applied
^ permalink raw reply
* Re: [PATCH] phylib: Add Realtek 821x eth PHY support
From: Jeff Garzik @ 2008-01-30 8:49 UTC (permalink / raw)
To: Kim Phillips; +Cc: netdev, Johnson Leung, Kevin Lam, Joe D'Abbraccio
In-Reply-To: <20080124202826.a50fa29c.kim.phillips@freescale.com>
Kim Phillips wrote:
> this PHY present on the MPC8315E and MPC837xE RDB boards.
>
> Signed-off-by: Johnson Leung <r58129@freescale.com>
> Signed-off-by: Kevin Lam <r43770@freescale.com>
> Signed-off-by: Joe D'Abbraccio <ljd015@freescale.com>
> Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
> ---
> drivers/net/phy/Kconfig | 5 +++
> drivers/net/phy/Makefile | 1 +
> drivers/net/phy/realtek.c | 80 +++++++++++++++++++++++++++++++++++++++++++++
> 3 files changed, 86 insertions(+), 0 deletions(-)
> create mode 100644 drivers/net/phy/realtek.c
applied (though admittedly the filename seems a bit too generic)
^ permalink raw reply
* Re: [PATCH] pci-skeleton: Misc fixes to build neatly
From: Jeff Garzik @ 2008-01-30 8:49 UTC (permalink / raw)
To: Jike Song; +Cc: netdev, linux-kernel
In-Reply-To: <1201157496-20665-2-git-send-email-albcamus@gmail.com>
Jike Song wrote:
> The pci-skeleton.c has several problems with compilation, such as missing args
> when calling synchronize_irq(). Fix it.
>
> Signed-off-by: Jike Song <albcamus@gmail.com>
> ---
> drivers/net/pci-skeleton.c | 49 ++++++++++++++++++++++---------------------
> 1 files changed, 25 insertions(+), 24 deletions(-)
applied
^ permalink raw reply
* Re: [PATCH 1/2] sky2: restore multicast addresses after recovery
From: Jeff Garzik @ 2008-01-30 8:51 UTC (permalink / raw)
To: Stephen Hemminger; +Cc: netdev
In-Reply-To: <20080123191151.1423a045@deepthought>
Stephen Hemminger wrote:
> If the sky2 deadman timer forces a recovery, the multicast hash
> list is lost. Move the call to sky2_set_multicast to the end
> of sky2_up() so all paths that bring device up will restore multicast.
>
> Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
>
> ---
> Please apply for 2.6.24
applied 1-2
You'll want to send this to stable@kernel.org, since by the time I read
your mail, 2.6.24 had been released, just around 24 hours thereafter.
^ permalink raw reply
* [PATCH 0/8] [Blackfin] EMAC driver updates
From: Bryan Wu @ 2008-01-30 8:52 UTC (permalink / raw)
To: jeff, netdev; +Cc: linux-kernel
Several bug fixing for this driver.
^ permalink raw reply
* [PATCH 2/8] [Blackfin] EMAC driver: define MDC_CLK=2.5MHz and caculate mdc_div according to SCLK.
From: Bryan Wu @ 2008-01-30 8:52 UTC (permalink / raw)
To: jeff, netdev; +Cc: linux-kernel, Bryan Wu
In-Reply-To: <1201683148-23931-1-git-send-email-bryan.wu@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
---
drivers/net/bfin_mac.c | 16 ++++++++++++----
1 files changed, 12 insertions(+), 4 deletions(-)
diff --git a/drivers/net/bfin_mac.c b/drivers/net/bfin_mac.c
index 4006a5d..ee39819 100644
--- a/drivers/net/bfin_mac.c
+++ b/drivers/net/bfin_mac.c
@@ -412,20 +412,26 @@ static void bf537_adjust_link(struct net_device *dev)
spin_unlock_irqrestore(&lp->lock, flags);
}
+/* MDC = 2.5 MHz */
+#define MDC_CLK 2500000
+
static int mii_probe(struct net_device *dev)
{
struct bf537mac_local *lp = netdev_priv(dev);
struct phy_device *phydev = NULL;
unsigned short sysctl;
int i;
+ u32 sclk, mdc_div;
/* Enable PHY output early */
if (!(bfin_read_VR_CTL() & PHYCLKOE))
bfin_write_VR_CTL(bfin_read_VR_CTL() | PHYCLKOE);
- /* MDC = 2.5 MHz */
+ sclk = get_sclk();
+ mdc_div = ((sclk / MDC_CLK) / 2) - 1;
+
sysctl = bfin_read_EMAC_SYSCTL();
- sysctl |= SET_MDCDIV(24);
+ sysctl |= SET_MDCDIV(mdc_div);
bfin_write_EMAC_SYSCTL(sysctl);
/* search for connect PHY device */
@@ -477,8 +483,10 @@ static int mii_probe(struct net_device *dev)
lp->phydev = phydev;
printk(KERN_INFO "%s: attached PHY driver [%s] "
- "(mii_bus:phy_addr=%s, irq=%d)\n",
- DRV_NAME, phydev->drv->name, phydev->dev.bus_id, phydev->irq);
+ "(mii_bus:phy_addr=%s, irq=%d, mdc_clk=%dHz(mdc_div=%d)"
+ "@sclk=%dMHz)\n",
+ DRV_NAME, phydev->drv->name, phydev->dev.bus_id, phydev->irq,
+ MDC_CLK, mdc_div, sclk/1000000);
return 0;
}
--
1.5.3.4
^ permalink raw reply related
* [PATCH 1/8] [Blackfin] EMAC driver: shorten the mdelay value to solve netperf performance issue
From: Bryan Wu @ 2008-01-30 8:52 UTC (permalink / raw)
To: jeff, netdev; +Cc: linux-kernel, Bryan Wu
In-Reply-To: <1201683148-23931-1-git-send-email-bryan.wu@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
---
drivers/net/bfin_mac.c | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/bfin_mac.c b/drivers/net/bfin_mac.c
index eb97175..4006a5d 100644
--- a/drivers/net/bfin_mac.c
+++ b/drivers/net/bfin_mac.c
@@ -296,7 +296,7 @@ static void mdio_poll(void)
/* poll the STABUSY bit */
while ((bfin_read_EMAC_STAADD()) & STABUSY) {
- mdelay(10);
+ udelay(1);
if (timeout_cnt-- < 0) {
printk(KERN_ERR DRV_NAME
": wait MDC/MDIO transaction to complete timeout\n");
@@ -551,7 +551,7 @@ static void adjust_tx_list(void)
*/
if (current_tx_ptr->next->next == tx_list_head) {
while (tx_list_head->status.status_word == 0) {
- mdelay(10);
+ mdelay(1);
if (tx_list_head->status.status_word != 0
|| !(bfin_read_DMA2_IRQ_STATUS() & 0x08)) {
goto adjust_head;
--
1.5.3.4
^ permalink raw reply related
* [PATCH 3/8] [Blackfin] EMAC driver: bf537 MAC multicast hash filtering patch
From: Bryan Wu @ 2008-01-30 8:52 UTC (permalink / raw)
To: jeff, netdev; +Cc: linux-kernel, Aidan Williams, Bryan Wu
In-Reply-To: <1201683148-23931-1-git-send-email-bryan.wu@analog.com>
From: Aidan Williams <aidan@nicta.com.au>
The bf537 Ethernet MAC driver in the 2007R1.1-RC3 kernel (and the
current kernel) do not implement multicast hash filtering. This
is a performance problem if you have lots of multicast on your network.
This patch plugs the right bits into the multicast hash registers.
Signed-off-by: Aidan Williams <aidan@nicta.com.au>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
---
drivers/net/bfin_mac.c | 42 +++++++++++++++++++++++++++++++++++++++++-
1 files changed, 41 insertions(+), 1 deletions(-)
diff --git a/drivers/net/bfin_mac.c b/drivers/net/bfin_mac.c
index ee39819..c6586cd 100644
--- a/drivers/net/bfin_mac.c
+++ b/drivers/net/bfin_mac.c
@@ -11,6 +11,7 @@
* Description:
*
* Modified:
+ * 2006-12-19 Aidan Williams, multicast hash support
* Copyright 2004-2006 Analog Devices Inc.
*
* Bugs: Enter bugs at http://blackfin.uclinux.org/
@@ -800,6 +801,39 @@ static void bf537mac_timeout(struct net_device *dev)
netif_wake_queue(dev);
}
+static void bf537mac_multicast_hash(struct net_device *dev)
+{
+ u32 emac_hashhi, emac_hashlo;
+ struct dev_mc_list *dmi = dev->mc_list;
+ char *addrs;
+ int i;
+ u32 crc;
+
+ emac_hashhi = emac_hashlo = 0;
+
+ for (i = 0; i < dev->mc_count; i++) {
+ addrs = dmi->dmi_addr;
+ dmi = dmi->next;
+
+ /* skip non-multicast addresses */
+ if (!(*addrs & 1))
+ continue;
+
+ crc = ether_crc(ETH_ALEN, addrs);
+ crc >>= 26;
+
+ if (crc & 0x20)
+ emac_hashhi |= 1 << (crc & 0x1f);
+ else
+ emac_hashlo |= 1 << (crc & 0x1f);
+ }
+
+ bfin_write_EMAC_HASHHI(emac_hashhi);
+ bfin_write_EMAC_HASHLO(emac_hashlo);
+
+ return;
+}
+
/*
* This routine will, depending on the values passed to it,
* either make it accept multicast packets, go into
@@ -815,11 +849,17 @@ static void bf537mac_set_multicast_list(struct net_device *dev)
sysctl = bfin_read_EMAC_OPMODE();
sysctl |= RAF;
bfin_write_EMAC_OPMODE(sysctl);
- } else if (dev->flags & IFF_ALLMULTI || dev->mc_count) {
+ } else if (dev->flags & IFF_ALLMULTI) {
/* accept all multicast */
sysctl = bfin_read_EMAC_OPMODE();
sysctl |= PAM;
bfin_write_EMAC_OPMODE(sysctl);
+ } else if (dev->mc_count) {
+ /* set up multicast hash table */
+ sysctl = bfin_read_EMAC_OPMODE();
+ sysctl |= HM;
+ bfin_write_EMAC_OPMODE(sysctl);
+ bf537mac_multicast_hash(dev);
} else {
/* clear promisc or multicast mode */
sysctl = bfin_read_EMAC_OPMODE();
--
1.5.3.4
^ permalink raw reply related
* [PATCH 6/8] [Blackfin] EMAC driver: add power down mode
From: Bryan Wu @ 2008-01-30 8:52 UTC (permalink / raw)
To: jeff, netdev; +Cc: linux-kernel, Vitja Makarov, Bryan Wu
In-Reply-To: <1201683148-23931-1-git-send-email-bryan.wu@analog.com>
From: Vitja Makarov <vitja.makarov@gmail.com>
This patch puts phy in power-down mode when interface is down.
Also we should think about energy detect power-down mode, that will
decrease power consumption when no link.
Signed-off-by: Vitja Makarov <vitja.makarov@gmail.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
---
drivers/net/bfin_mac.c | 3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/drivers/net/bfin_mac.c b/drivers/net/bfin_mac.c
index e9bd059..f2368b7 100644
--- a/drivers/net/bfin_mac.c
+++ b/drivers/net/bfin_mac.c
@@ -884,10 +884,10 @@ static int bf537mac_open(struct net_device *dev)
return retval;
phy_start(lp->phydev);
+ phy_write(lp->phydev, MII_BMCR, BMCR_RESET);
setup_system_regs(dev);
bf537mac_disable();
bf537mac_enable();
-
pr_debug("hardware init finished\n");
netif_start_queue(dev);
netif_carrier_on(dev);
@@ -910,6 +910,7 @@ static int bf537mac_close(struct net_device *dev)
netif_carrier_off(dev);
phy_stop(lp->phydev);
+ phy_write(lp->phydev, MII_BMCR, BMCR_PDOWN);
/* clear everything */
bf537mac_shutdown(dev);
--
1.5.3.4
^ permalink raw reply related
* [PATCH 8/8] [Blackfin] EMAC driver: Fix bug: The clock divisor is set to all ones at reset.
From: Bryan Wu @ 2008-01-30 8:52 UTC (permalink / raw)
To: jeff, netdev; +Cc: linux-kernel, Bryan Wu, Kalle Pokki
In-Reply-To: <1201683148-23931-1-git-send-email-bryan.wu@analog.com>
Signed-off-by: Kalle Pokki <kalle.pokki@eke.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
---
drivers/net/bfin_mac.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/net/bfin_mac.c b/drivers/net/bfin_mac.c
index 0a17fb4..c993a32 100644
--- a/drivers/net/bfin_mac.c
+++ b/drivers/net/bfin_mac.c
@@ -408,7 +408,7 @@ static int mii_probe(struct net_device *dev)
mdc_div = ((sclk / MDC_CLK) / 2) - 1;
sysctl = bfin_read_EMAC_SYSCTL();
- sysctl |= SET_MDCDIV(mdc_div);
+ sysctl = (sysctl & ~MDCDIV) | SET_MDCDIV(mdc_div);
bfin_write_EMAC_SYSCTL(sysctl);
/* search for connect PHY device */
--
1.5.3.4
^ permalink raw reply related
* [PATCH 7/8] [Blackfin] EMAC driver: fix bug - invalidate data cache of new_skb->data range when cache is WB
From: Bryan Wu @ 2008-01-30 8:52 UTC (permalink / raw)
To: jeff, netdev; +Cc: linux-kernel, Alexey Demin, Bryan Wu
In-Reply-To: <1201683148-23931-1-git-send-email-bryan.wu@analog.com>
From: Alexey Demin <bf53x@ya.ru>
It prevents overwritting new data from DMA.
Signed-off-by: Alexey Demin <bf53x@ya.ru>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
---
drivers/net/bfin_mac.c | 6 ++++++
1 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/drivers/net/bfin_mac.c b/drivers/net/bfin_mac.c
index f2368b7..0a17fb4 100644
--- a/drivers/net/bfin_mac.c
+++ b/drivers/net/bfin_mac.c
@@ -651,6 +651,12 @@ static void bf537mac_rx(struct net_device *dev)
current_rx_ptr->skb = new_skb;
current_rx_ptr->desc_a.start_addr = (unsigned long)new_skb->data - 2;
+ /* Invidate the data cache of skb->data range when it is write back
+ * cache. It will prevent overwritting the new data from DMA
+ */
+ blackfin_dcache_invalidate_range((unsigned long)new_skb->head,
+ (unsigned long)new_skb->end);
+
len = (unsigned short)((current_rx_ptr->status.status_word) & RX_FRLEN);
skb_put(skb, len);
blackfin_dcache_invalidate_range((unsigned long)skb->head,
--
1.5.3.4
^ permalink raw reply related
* [PATCH 5/8] [Blackfin] EMAC driver: ADSP-BF52x arch/mach support
From: Bryan Wu @ 2008-01-30 8:52 UTC (permalink / raw)
To: jeff, netdev; +Cc: linux-kernel, Michael Hennerich, Bryan Wu
In-Reply-To: <1201683148-23931-1-git-send-email-bryan.wu@analog.com>
From: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
---
drivers/net/Kconfig | 9 +++++----
drivers/net/bfin_mac.c | 6 +++---
2 files changed, 8 insertions(+), 7 deletions(-)
diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 5a2d1dd..ca2552b 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -814,8 +814,8 @@ config ULTRA32
will be called smc-ultra32.
config BFIN_MAC
- tristate "Blackfin 536/537 on-chip mac support"
- depends on NET_ETHERNET && (BF537 || BF536) && (!BF537_PORT_H)
+ tristate "Blackfin 527/536/537 on-chip mac support"
+ depends on NET_ETHERNET && (BF527 || BF537 || BF536) && (!BF537_PORT_H)
select CRC32
select MII
select PHYLIB
@@ -828,7 +828,7 @@ config BFIN_MAC
config BFIN_MAC_USE_L1
bool "Use L1 memory for rx/tx packets"
- depends on BFIN_MAC && BF537
+ depends on BFIN_MAC && (BF527 || BF537)
default y
help
To get maximum network performance, you should use L1 memory as rx/tx buffers.
@@ -855,7 +855,8 @@ config BFIN_RX_DESC_NUM
config BFIN_MAC_RMII
bool "RMII PHY Interface (EXPERIMENTAL)"
depends on BFIN_MAC && EXPERIMENTAL
- default n
+ default y if BFIN527_EZKIT
+ default n if BFIN537_STAMP
help
Use Reduced PHY MII Interface
diff --git a/drivers/net/bfin_mac.c b/drivers/net/bfin_mac.c
index ed935e1..e9bd059 100644
--- a/drivers/net/bfin_mac.c
+++ b/drivers/net/bfin_mac.c
@@ -42,7 +42,7 @@
#define DRV_NAME "bfin_mac"
#define DRV_VERSION "1.1"
#define DRV_AUTHOR "Bryan Wu, Luke Yang"
-#define DRV_DESC "Blackfin BF53[67] on-chip Ethernet MAC driver"
+#define DRV_DESC "Blackfin BF53[67] BF527 on-chip Ethernet MAC driver"
MODULE_AUTHOR(DRV_AUTHOR);
MODULE_LICENSE("GPL");
@@ -752,7 +752,7 @@ static void bf537mac_enable(void)
#if defined(CONFIG_BFIN_MAC_RMII)
opmode |= RMII; /* For Now only 100MBit are supported */
-#ifdef CONFIG_BF_REV_0_2
+#if (defined(CONFIG_BF537) || defined(CONFIG_BF536)) && CONFIG_BF_REV_0_2
opmode |= TE;
#endif
#endif
@@ -994,7 +994,7 @@ static int __init bf537mac_probe(struct net_device *dev)
/* register irq handler */
if (request_irq
(IRQ_MAC_RX, bf537mac_interrupt, IRQF_DISABLED | IRQF_SHARED,
- "BFIN537_MAC_RX", dev)) {
+ "EMAC_RX", dev)) {
printk(KERN_WARNING DRV_NAME
": Unable to attach BlackFin MAC RX interrupt\n");
return -EBUSY;
--
1.5.3.4
^ permalink raw reply related
* [PATCH 4/8] [Blackfin] EMAC driver: use simpler comment headers and strip out information that is maintained in the scm's log
From: Bryan Wu @ 2008-01-30 8:52 UTC (permalink / raw)
To: jeff, netdev; +Cc: linux-kernel, Mike Frysinger, Bryan Wu
In-Reply-To: <1201683148-23931-1-git-send-email-bryan.wu@analog.com>
From: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Mike Frysinger <michael.frysinger@analog.com>
Signed-off-by: Bryan Wu <bryan.wu@analog.com>
---
drivers/net/bfin_mac.c | 32 ++++----------------------------
drivers/net/bfin_mac.h | 31 ++++---------------------------
2 files changed, 8 insertions(+), 55 deletions(-)
diff --git a/drivers/net/bfin_mac.c b/drivers/net/bfin_mac.c
index c6586cd..ed935e1 100644
--- a/drivers/net/bfin_mac.c
+++ b/drivers/net/bfin_mac.c
@@ -1,35 +1,11 @@
/*
- * File: drivers/net/bfin_mac.c
- * Based on:
- * Maintainer:
- * Bryan Wu <bryan.wu@analog.com>
+ * Blackfin On-Chip MAC Driver
*
- * Original author:
- * Luke Yang <luke.yang@analog.com>
+ * Copyright 2004-2007 Analog Devices Inc.
*
- * Created:
- * Description:
+ * Enter bugs at http://blackfin.uclinux.org/
*
- * Modified:
- * 2006-12-19 Aidan Williams, multicast hash support
- * Copyright 2004-2006 Analog Devices Inc.
- *
- * Bugs: Enter bugs at http://blackfin.uclinux.org/
- *
- * 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, 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 ; see the file COPYING.
- * If not, write to the Free Software Foundation,
- * 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ * Licensed under the GPL-2 or later.
*/
#include <linux/init.h>
diff --git a/drivers/net/bfin_mac.h b/drivers/net/bfin_mac.h
index 5970ea7..f774d5a 100644
--- a/drivers/net/bfin_mac.h
+++ b/drivers/net/bfin_mac.h
@@ -1,34 +1,11 @@
/*
- * File: drivers/net/bfin_mac.c
- * Based on:
- * Maintainer:
- * Bryan Wu <bryan.wu@analog.com>
+ * Blackfin On-Chip MAC Driver
*
- * Original author:
- * Luke Yang <luke.yang@analog.com>
+ * Copyright 2004-2007 Analog Devices Inc.
*
- * Created:
- * Description:
+ * Enter bugs at http://blackfin.uclinux.org/
*
- * Modified:
- * Copyright 2004-2006 Analog Devices Inc.
- *
- * Bugs: Enter bugs at http://blackfin.uclinux.org/
- *
- * 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, 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 ; see the file COPYING.
- * If not, write to the Free Software Foundation,
- * 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+ * Licensed under the GPL-2 or later.
*/
#define BFIN_MAC_CSUM_OFFLOAD
--
1.5.3.4
^ permalink raw reply related
* [PATCH] [1/1] Deprecate tcp_tw_{reuse,recycle}
From: Andi Kleen @ 2008-01-30 8:38 UTC (permalink / raw)
To: netdev
We've recently had a long discussion about the CVE-2005-0356 time stamp denial-of-service
attack. It turned out that Linux is only vunerable to this problem when tcp_tw_recycle
is enabled (which it is not by default).
In general these two options are not really usable in today's internet because they
make the (often false) assumption that a single IP address has a single TCP time stamp /
PAWS clock. This assumption breaks both NAT/masquerading and also opens Linux to denial
of service attacks (see the CVE description)
Due to these numerous problems I propose to remove this code for 2.6.26
Signed-off-by: Andi Kleen <ak@suse.de>
Index: linux/Documentation/feature-removal-schedule.txt
===================================================================
--- linux.orig/Documentation/feature-removal-schedule.txt
+++ linux/Documentation/feature-removal-schedule.txt
@@ -354,3 +354,15 @@ Why: The support code for the old firmwa
and slightly hurts runtime performance. Bugfixes for the old firmware
are not provided by Broadcom anymore.
Who: Michael Buesch <mb@bu3sch.de>
+
+---------------------------
+
+What: Support for /proc/sys/net/ipv4/tcp_tw_{reuse,recycle} = 1
+When: 2.6.26
+Why: Enabling either of those makes Linux TCP incompatible with masquerading and
+ also opens Linux to the CVE-2005-0356 denial of service attack. And these
+ optimizations are explicitely disallowed by some benchmarks. They also have
+ been disabled by default for more than ten years so they're unlikely to be used
+ much. Due to these fatal flaws it doesn't make sense to keep the code.
+Who: Andi Kleen <andi@firstfloor.org>
+
^ 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