* [PATCH 00/10] net/macb: driver enhancement concerning GEM support, ring logic and cleanup
From: Nicolas Ferre @ 2012-09-05 8:19 UTC (permalink / raw)
To: netdev
Cc: patrice.vilchez, nicolas.ferre, linux-kernel, havard, jamie,
plagnioj, davem, linux-arm-kernel
This is an enhancement work that began several years ago. I try to catchup with
some performance improvement that has been implemented then by Havard.
The ring index logic and the TX error path modification are the biggest changes
but some cleanup/debugging have been added along the way.
The GEM revision will benefit from the Gigabit support.
The series has been tested on several Atmel AT91 SoC with the two MACB/GEM
flavors.
Havard Skinnemoen (5):
net/macb: memory barriers cleanup
net/macb: change debugging messages
net/macb: Fix a race in macb_start_xmit()
net/macb: clean up ring buffer logic
net/macb: Offset first RX buffer by two bytes
Nicolas Ferre (4):
net/macb: better manage tx errors
net/macb: tx status is more than 8 bits now
net/macb: macb_get_drvinfo: add GEM/MACB suffix to differentiate
revision
net/macb: ethtool interface: add register dump feature
Patrice Vilchez (1):
net/macb: Add support for Gigabit Ethernet mode
drivers/net/ethernet/cadence/macb.c | 408 ++++++++++++++++++++++++-----------
drivers/net/ethernet/cadence/macb.h | 29 ++-
2 files changed, 304 insertions(+), 133 deletions(-)
--
1.7.10
^ permalink raw reply
* [PATCH 01/10] net/macb: Add support for Gigabit Ethernet mode
From: Nicolas Ferre @ 2012-09-05 8:19 UTC (permalink / raw)
To: netdev
Cc: linux-arm-kernel, davem, havard, nicolas.ferre, plagnioj, jamie,
linux-kernel, patrice.vilchez
In-Reply-To: <cover.1346775479.git.nicolas.ferre@atmel.com>
From: Patrice Vilchez <patrice.vilchez@atmel.com>
Add Gigabit Ethernet mode to GEM cadence IP and enable RGMII connection.
Signed-off-by: Patrice Vilchez <patrice.vilchez@atmel.com>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
drivers/net/ethernet/cadence/macb.c | 15 ++++++++++++---
drivers/net/ethernet/cadence/macb.h | 4 ++++
2 files changed, 16 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/cadence/macb.c b/drivers/net/ethernet/cadence/macb.c
index 033064b..9a10f69 100644
--- a/drivers/net/ethernet/cadence/macb.c
+++ b/drivers/net/ethernet/cadence/macb.c
@@ -152,13 +152,17 @@ static void macb_handle_link_change(struct net_device *dev)
reg = macb_readl(bp, NCFGR);
reg &= ~(MACB_BIT(SPD) | MACB_BIT(FD));
+ if (macb_is_gem(bp))
+ reg &= ~GEM_BIT(GBE);
if (phydev->duplex)
reg |= MACB_BIT(FD);
if (phydev->speed == SPEED_100)
reg |= MACB_BIT(SPD);
+ if (phydev->speed == SPEED_1000)
+ reg |= GEM_BIT(GBE);
- macb_writel(bp, NCFGR, reg);
+ macb_or_gem_writel(bp, NCFGR, reg);
bp->speed = phydev->speed;
bp->duplex = phydev->duplex;
@@ -216,7 +220,10 @@ static int macb_mii_probe(struct net_device *dev)
}
/* mask with MAC supported features */
- phydev->supported &= PHY_BASIC_FEATURES;
+ if (macb_is_gem(bp))
+ phydev->supported &= PHY_GBIT_FEATURES;
+ else
+ phydev->supported &= PHY_BASIC_FEATURES;
phydev->advertising = phydev->supported;
@@ -1384,7 +1391,9 @@ static int __init macb_probe(struct platform_device *pdev)
bp->phy_interface = err;
}
- if (bp->phy_interface == PHY_INTERFACE_MODE_RMII)
+ if (bp->phy_interface == PHY_INTERFACE_MODE_RGMII)
+ macb_or_gem_writel(bp, USRIO, GEM_BIT(RGMII));
+ else if (bp->phy_interface == PHY_INTERFACE_MODE_RMII)
#if defined(CONFIG_ARCH_AT91)
macb_or_gem_writel(bp, USRIO, (MACB_BIT(RMII) |
MACB_BIT(CLKEN)));
diff --git a/drivers/net/ethernet/cadence/macb.h b/drivers/net/ethernet/cadence/macb.h
index 335e288..f69ceef 100644
--- a/drivers/net/ethernet/cadence/macb.h
+++ b/drivers/net/ethernet/cadence/macb.h
@@ -145,6 +145,8 @@
#define MACB_IRXFCS_SIZE 1
/* GEM specific NCFGR bitfields. */
+#define GEM_GBE_OFFSET 10
+#define GEM_GBE_SIZE 1
#define GEM_CLK_OFFSET 18
#define GEM_CLK_SIZE 3
#define GEM_DBW_OFFSET 21
@@ -246,6 +248,8 @@
/* Bitfields in USRIO (AT91) */
#define MACB_RMII_OFFSET 0
#define MACB_RMII_SIZE 1
+#define GEM_RGMII_OFFSET 0 /* GEM gigabit mode */
+#define GEM_RGMII_SIZE 1
#define MACB_CLKEN_OFFSET 1
#define MACB_CLKEN_SIZE 1
--
1.7.10
^ permalink raw reply related
* [PATCH 02/10] net/macb: memory barriers cleanup
From: Nicolas Ferre @ 2012-09-05 8:19 UTC (permalink / raw)
To: netdev
Cc: patrice.vilchez, nicolas.ferre, linux-kernel, havard, jamie,
plagnioj, davem, linux-arm-kernel
In-Reply-To: <cover.1346775479.git.nicolas.ferre@atmel.com>
From: Havard Skinnemoen <havard@skinnemoen.net>
Remove a couple of unneeded barriers and document the remaining ones.
Signed-off-by: Havard Skinnemoen <havard@skinnemoen.net>
[nicolas.ferre@atmel.com: split patch in topics]
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
drivers/net/ethernet/cadence/macb.c | 18 ++++++++++++++----
1 file changed, 14 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ethernet/cadence/macb.c b/drivers/net/ethernet/cadence/macb.c
index 9a10f69..26ca01e 100644
--- a/drivers/net/ethernet/cadence/macb.c
+++ b/drivers/net/ethernet/cadence/macb.c
@@ -372,7 +372,9 @@ static void macb_tx(struct macb *bp)
BUG_ON(skb == NULL);
+ /* Make hw descriptor updates visible to CPU */
rmb();
+
bufstat = bp->tx_ring[tail].ctrl;
if (!(bufstat & MACB_BIT(TX_USED)))
@@ -415,7 +417,10 @@ static int macb_rx_frame(struct macb *bp, unsigned int first_frag,
if (frag == last_frag)
break;
}
+
+ /* Make descriptor updates visible to hardware */
wmb();
+
return 1;
}
@@ -436,12 +441,14 @@ static int macb_rx_frame(struct macb *bp, unsigned int first_frag,
frag_len);
offset += RX_BUFFER_SIZE;
bp->rx_ring[frag].addr &= ~MACB_BIT(RX_USED);
- wmb();
if (frag == last_frag)
break;
}
+ /* Make descriptor updates visible to hardware */
+ wmb();
+
skb->protocol = eth_type_trans(skb, bp->dev);
bp->stats.rx_packets++;
@@ -461,6 +468,8 @@ static void discard_partial_frame(struct macb *bp, unsigned int begin,
for (frag = begin; frag != end; frag = NEXT_RX(frag))
bp->rx_ring[frag].addr &= ~MACB_BIT(RX_USED);
+
+ /* Make descriptor updates visible to hardware */
wmb();
/*
@@ -479,7 +488,9 @@ static int macb_rx(struct macb *bp, int budget)
for (; budget > 0; tail = NEXT_RX(tail)) {
u32 addr, ctrl;
+ /* Make hw descriptor updates visible to CPU */
rmb();
+
addr = bp->rx_ring[tail].addr;
ctrl = bp->rx_ring[tail].ctrl;
@@ -674,6 +685,8 @@ static int macb_start_xmit(struct sk_buff *skb, struct net_device *dev)
bp->tx_ring[entry].addr = mapping;
bp->tx_ring[entry].ctrl = ctrl;
+
+ /* Make newly initialized descriptor visible to hardware */
wmb();
entry = NEXT_TX(entry);
@@ -782,9 +795,6 @@ static void macb_init_rings(struct macb *bp)
static void macb_reset_hw(struct macb *bp)
{
- /* Make sure we have the write buffer for ourselves */
- wmb();
-
/*
* Disable RX and TX (XXX: Should we halt the transmission
* more gracefully?)
--
1.7.10
^ permalink raw reply related
* [PATCH 03/10] net/macb: change debugging messages
From: Nicolas Ferre @ 2012-09-05 8:19 UTC (permalink / raw)
To: netdev
Cc: linux-arm-kernel, davem, havard, nicolas.ferre, plagnioj, jamie,
linux-kernel, patrice.vilchez
In-Reply-To: <cover.1346775479.git.nicolas.ferre@atmel.com>
From: Havard Skinnemoen <havard@skinnemoen.net>
Convert some noisy netdev_dbg() statements to netdev_vdbg(). Defining
DEBUG will no longer fill up the logs; VERBOSE_DEBUG still does.
Add one more verbose debug for ISR status.
Signed-off-by: Havard Skinnemoen <havard@skinnemoen.net>
[nicolas.ferre@atmel.com: split patch in topics, add ISR status]
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
drivers/net/ethernet/cadence/macb.c | 22 ++++++++++++----------
1 file changed, 12 insertions(+), 10 deletions(-)
diff --git a/drivers/net/ethernet/cadence/macb.c b/drivers/net/ethernet/cadence/macb.c
index 26ca01e..2228dfc 100644
--- a/drivers/net/ethernet/cadence/macb.c
+++ b/drivers/net/ethernet/cadence/macb.c
@@ -313,7 +313,7 @@ static void macb_tx(struct macb *bp)
status = macb_readl(bp, TSR);
macb_writel(bp, TSR, status);
- netdev_dbg(bp->dev, "macb_tx status = %02lx\n", (unsigned long)status);
+ netdev_vdbg(bp->dev, "macb_tx status = %02lx\n", (unsigned long)status);
if (status & (MACB_BIT(UND) | MACB_BIT(TSR_RLE))) {
int i;
@@ -380,7 +380,7 @@ static void macb_tx(struct macb *bp)
if (!(bufstat & MACB_BIT(TX_USED)))
break;
- netdev_dbg(bp->dev, "skb %u (data %p) TX complete\n",
+ netdev_vdbg(bp->dev, "skb %u (data %p) TX complete\n",
tail, skb->data);
dma_unmap_single(&bp->pdev->dev, rp->mapping, skb->len,
DMA_TO_DEVICE);
@@ -406,7 +406,7 @@ static int macb_rx_frame(struct macb *bp, unsigned int first_frag,
len = MACB_BFEXT(RX_FRMLEN, bp->rx_ring[last_frag].ctrl);
- netdev_dbg(bp->dev, "macb_rx_frame frags %u - %u (len %u)\n",
+ netdev_vdbg(bp->dev, "macb_rx_frame frags %u - %u (len %u)\n",
first_frag, last_frag, len);
skb = netdev_alloc_skb(bp->dev, len + RX_OFFSET);
@@ -453,7 +453,7 @@ static int macb_rx_frame(struct macb *bp, unsigned int first_frag,
bp->stats.rx_packets++;
bp->stats.rx_bytes += len;
- netdev_dbg(bp->dev, "received skb of length %u, csum: %08x\n",
+ netdev_vdbg(bp->dev, "received skb of length %u, csum: %08x\n",
skb->len, skb->csum);
netif_receive_skb(skb);
@@ -535,7 +535,7 @@ static int macb_poll(struct napi_struct *napi, int budget)
work_done = 0;
- netdev_dbg(bp->dev, "poll: status = %08lx, budget = %d\n",
+ netdev_vdbg(bp->dev, "poll: status = %08lx, budget = %d\n",
(unsigned long)status, budget);
work_done = macb_rx(bp, budget);
@@ -574,6 +574,8 @@ static irqreturn_t macb_interrupt(int irq, void *dev_id)
break;
}
+ netdev_vdbg(bp->dev, "isr = 0x%08lx\n", (unsigned long)status);
+
if (status & MACB_RX_INT_FLAGS) {
/*
* There's no point taking any more interrupts
@@ -585,7 +587,7 @@ static irqreturn_t macb_interrupt(int irq, void *dev_id)
macb_writel(bp, IDR, MACB_RX_INT_FLAGS);
if (napi_schedule_prep(&bp->napi)) {
- netdev_dbg(bp->dev, "scheduling RX softirq\n");
+ netdev_vdbg(bp->dev, "scheduling RX softirq\n");
__napi_schedule(&bp->napi);
}
}
@@ -647,8 +649,8 @@ static int macb_start_xmit(struct sk_buff *skb, struct net_device *dev)
u32 ctrl;
unsigned long flags;
-#ifdef DEBUG
- netdev_dbg(bp->dev,
+#if defined(DEBUG) && defined(VERBOSE_DEBUG)
+ netdev_vdbg(bp->dev,
"start_xmit: len %u head %p data %p tail %p end %p\n",
skb->len, skb->head, skb->data,
skb_tail_pointer(skb), skb_end_pointer(skb));
@@ -670,12 +672,12 @@ static int macb_start_xmit(struct sk_buff *skb, struct net_device *dev)
}
entry = bp->tx_head;
- netdev_dbg(bp->dev, "Allocated ring entry %u\n", entry);
+ netdev_vdbg(bp->dev, "Allocated ring entry %u\n", entry);
mapping = dma_map_single(&bp->pdev->dev, skb->data,
len, DMA_TO_DEVICE);
bp->tx_skb[entry].skb = skb;
bp->tx_skb[entry].mapping = mapping;
- netdev_dbg(bp->dev, "Mapped skb data %p to DMA addr %08lx\n",
+ netdev_vdbg(bp->dev, "Mapped skb data %p to DMA addr %08lx\n",
skb->data, (unsigned long)mapping);
ctrl = MACB_BF(TX_FRMLEN, len);
--
1.7.10
^ permalink raw reply related
* [PATCH 04/10] net/macb: Fix a race in macb_start_xmit()
From: Nicolas Ferre @ 2012-09-05 8:19 UTC (permalink / raw)
To: netdev
Cc: linux-arm-kernel, davem, havard, nicolas.ferre, plagnioj, jamie,
linux-kernel, patrice.vilchez
In-Reply-To: <cover.1346775479.git.nicolas.ferre@atmel.com>
From: Havard Skinnemoen <havard@skinnemoen.net>
Fix a race in macb_start_xmit() where we unconditionally set the TSTART bit.
If an underrun just happened (we do this with interrupts disabled, so it might
not have been handled yet), the controller starts transmitting from the first
entry in the ring, which is usually wrong.
Restart the controller after error handling.
Signed-off-by: Havard Skinnemoen <havard@skinnemoen.net>
[nicolas.ferre@atmel.com: split patch in topics]
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
drivers/net/ethernet/cadence/macb.c | 20 +++++++++++++++++++-
1 file changed, 19 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/cadence/macb.c b/drivers/net/ethernet/cadence/macb.c
index 2228dfc..f4b8adf 100644
--- a/drivers/net/ethernet/cadence/macb.c
+++ b/drivers/net/ethernet/cadence/macb.c
@@ -390,6 +390,13 @@ static void macb_tx(struct macb *bp)
dev_kfree_skb_irq(skb);
}
+ /*
+ * Someone may have submitted a new frame while this interrupt
+ * was pending, or we may just have handled an error.
+ */
+ if (head != tail && !(status & MACB_BIT(TGO)))
+ macb_writel(bp, NCR, macb_readl(bp, NCR) | MACB_BIT(TSTART));
+
bp->tx_tail = tail;
if (netif_queue_stopped(bp->dev) &&
TX_BUFFS_AVAIL(bp) > MACB_TX_WAKEUP_THRESH)
@@ -696,7 +703,18 @@ static int macb_start_xmit(struct sk_buff *skb, struct net_device *dev)
skb_tx_timestamp(skb);
- macb_writel(bp, NCR, macb_readl(bp, NCR) | MACB_BIT(TSTART));
+ /*
+ * Only start the controller if the queue was empty; otherwise
+ * we may race against the hardware resetting the ring pointer
+ * due to a transmit error.
+ *
+ * If the controller is idle but the queue isn't empty, there
+ * must be a pending interrupt that will trigger as soon as we
+ * re-enable interrupts, and the interrupt handler will make
+ * sure the controler is started.
+ */
+ if (NEXT_TX(bp->tx_tail) == bp->tx_head)
+ macb_writel(bp, NCR, macb_readl(bp, NCR) | MACB_BIT(TSTART));
if (TX_BUFFS_AVAIL(bp) < 1)
netif_stop_queue(dev);
--
1.7.10
^ permalink raw reply related
* Re: Commit "ipconfig wait for carrier" makes boot hang for 2 mins if no carrier
From: Micha Nelissen @ 2012-09-05 8:30 UTC (permalink / raw)
To: Joakim Tjernlund; +Cc: netdev
In-Reply-To: <OFFA3AC1B1.08FC3AAD-ONC1257A70.00266EDC-C1257A70.0026E8BC@transmode.se>
Op 2012-09-05 9:04, Joakim Tjernlund schreef:
>> Because that's where my root filesystem is? The IP autoconfiguration
>> code exists for this purpose.
>
> This is not the only purpose.
Documentation/filesystems/nfs/nfsroot.txt seems to suggest so (this is
where the ip= parameter is documented), but it works independently indeed.
So explain your reasons?
>>> The answer is probably the same, it is much easier to
>>> manage our IP config in one place for our embedded system.
>>
>> You retrieve the kernel via TFTP or so when booting?
>
> Yes, but mostly not. This really doesn't matter
Seems to me that if you boot standalone there is no reason to let the IP
address be configured by the kernel? Retrieve the IP address in user
space from your bootloader environment or whatever. And if you boot from
ethernet (or some other networking interface), then you have a carrier,
and there is no 2 minute delay (maybe less even than before with this
patch!).
>>> The wait should be conditional on NFS root or not so that non NFS roots
>>> can skip this stage altogether.
>
> Your patch broke other use cases so my patch would just revert or change the tmo
> to 2 secs or so.
> Or you could clean up your stuff so it works for all and not just for you.
It didn't break anything, it does work for you also, you just need to
wait somewhat longer. Or make sure there is a carrier. The intent of the
original 1 second delay was to let the link come up!
Micha
^ permalink raw reply
* Re: [Bug 47021] New: kernel panic with l2tpv3 & mtu > 1500
From: Eric Dumazet @ 2012-09-05 8:34 UTC (permalink / raw)
To: a1; +Cc: Stephen Hemminger, netdev
In-Reply-To: <50470324.5040805@atlas.cz>
On Wed, 2012-09-05 at 09:45 +0200, a1 wrote:
> Thanks, it definitely helped, no more panics now...
It seems MTU of device is wrong, and lot of packets are fragmented...
Its currently 1488, but it really should be less than that (accounting
for the IP+UDP header)
I wonder if anybody ever used this code ?
^ permalink raw reply
* Re: [PATCH] decnet: fix shutdown parameter checking
From: Steven Whitehouse @ 2012-09-05 8:37 UTC (permalink / raw)
To: David Miller; +Cc: xi.wang, netdev, linux-kernel
In-Reply-To: <20120831.155719.2228399422573401952.davem@davemloft.net>
Hi,
On Fri, 2012-08-31 at 15:57 -0400, David Miller wrote:
> From: Steven Whitehouse <swhiteho@redhat.com>
> Date: Mon, 27 Aug 2012 10:16:41 +0100
>
> > On Sun, 2012-08-26 at 22:37 -0400, Xi Wang wrote:
> >> The allowed value of "how" is SHUT_RD/SHUT_WR/SHUT_RDWR (0/1/2),
> >> rather than SHUTDOWN_MASK (3).
> >>
> >> Signed-off-by: Xi Wang <xi.wang@gmail.com>
> > Acked-by: Steven Whitehouse <swhiteho@redhat.com>
>
> Applied to net-next.
>
> > Although it could be argued that we should also continue to accept the
> > value 3 just in case there is any userland software out there which
> > sends that value,
>
> True, but this is a rather standard BSD socket interface with a very
> specific small set of legitimate input parameters. Allowing
> deviation, even for compatability for specific protocols, is largely
> unwise.
Yes, I'd agree on the whole, and certainly if this was a recent
addition. However since this code has been around for somewhere close to
16 years now, I'd say that means that either (a) nobody calls shutdown
for DECnet or (b) existing users are buggy too.
We do have a precedent for this kind of compatibility, such as the AX.25
use of SOCK_SEQPACKET.
However, I'm not overly worried and we'll soon know if it will cause any
problems or not,
Steve.
^ permalink raw reply
* Re: Commit "ipconfig wait for carrier" makes boot hang for 2 mins if no carrier
From: Joakim Tjernlund @ 2012-09-05 8:47 UTC (permalink / raw)
To: Micha Nelissen; +Cc: netdev
In-Reply-To: <50470DAC.8030009@neli.hopto.org>
Micha Nelissen <micha@neli.hopto.org> wrote on 2012/09/05 10:30:36:
> From: Micha Nelissen <micha@neli.hopto.org>
> To: Joakim Tjernlund <joakim.tjernlund@transmode.se>,
> Cc: netdev@vger.kernel.org
> Date: 2012/09/05 10:30
> Subject: Re: Commit "ipconfig wait for carrier" makes boot hang for 2 mins if no carrier
>
> Op 2012-09-05 9:04, Joakim Tjernlund schreef:
> >> Because that's where my root filesystem is? The IP autoconfiguration
> >> code exists for this purpose.
> >
> > This is not the only purpose.
>
> Documentation/filesystems/nfs/nfsroot.txt seems to suggest so (this is
> where the ip= parameter is documented), but it works independently indeed.
>
> So explain your reasons?
If you read that doc you find:
ip=<client-ip>:<server-ip>:<gw-ip>:<netmask>:<hostname>:<device>:<autoconf>
This parameter tells the kernel how to configure IP addresses of devices
and also how to set up the IP routing table. It was originally called
`nfsaddrs', but now the boot-time IP configuration works independently of
NFS, so it was renamed to `ip' and the old name remained as an alias for
compatibility reasons.
>
> >>> The answer is probably the same, it is much easier to
> >>> manage our IP config in one place for our embedded system.
> >>
> >> You retrieve the kernel via TFTP or so when booting?
> >
> > Yes, but mostly not. This really doesn't matter
>
> Seems to me that if you boot standalone there is no reason to let the IP
> address be configured by the kernel? Retrieve the IP address in user
> space from your bootloader environment or whatever. And if you boot from
> ethernet (or some other networking interface), then you have a carrier,
> and there is no 2 minute delay (maybe less even than before with this
> patch!).
Everything is possible but we choosed to use already built-in functionality, as
did you.
You could have added an initram FS and done your NFS mount there so this
argument goes nowhere.
>
> >>> The wait should be conditional on NFS root or not so that non NFS roots
> >>> can skip this stage altogether.
> >
> > Your patch broke other use cases so my patch would just revert or change the tmo
> > to 2 secs or so.
> > Or you could clean up your stuff so it works for all and not just for you.
>
> It didn't break anything, it does work for you also, you just need to
> wait somewhat longer. Or make sure there is a carrier. The intent of the
> original 1 second delay was to let the link come up!
Sure it did, our system(and not only ours I bet) can not accept a 2 minute delay in
booting up the system for no reason.
Please adjust the 2 min wait to nfsroot= only and keep the old way for ip=
Jocke
^ permalink raw reply
* [PATCH 05/10] net/macb: clean up ring buffer logic
From: Nicolas Ferre @ 2012-09-05 9:00 UTC (permalink / raw)
To: netdev
Cc: linux-arm-kernel, davem, havard, nicolas.ferre, plagnioj, jamie,
linux-kernel, patrice.vilchez
In-Reply-To: <cover.1346775479.git.nicolas.ferre@atmel.com>
From: Havard Skinnemoen <havard@skinnemoen.net>
Instead of masking head and tail every time we increment them, just let them
wrap through UINT_MAX and mask them when subscripting. Add simple accessor
functions to do the subscripting properly to minimize the chances of messing
this up.
This makes the code slightly smaller, and hopefully faster as well. Also,
doing the ring buffer management this way will simplify things a lot when
making the ring sizes configurable in the future.
Signed-off-by: Havard Skinnemoen <havard@skinnemoen.net>
[nicolas.ferre@atmel.com: split patch in topics, adapt to newer kernel]
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
drivers/net/ethernet/cadence/macb.c | 170 ++++++++++++++++++++++-------------
drivers/net/ethernet/cadence/macb.h | 22 +++--
2 files changed, 123 insertions(+), 69 deletions(-)
diff --git a/drivers/net/ethernet/cadence/macb.c b/drivers/net/ethernet/cadence/macb.c
index f4b8adf..3d3a077 100644
--- a/drivers/net/ethernet/cadence/macb.c
+++ b/drivers/net/ethernet/cadence/macb.c
@@ -31,24 +31,13 @@
#define RX_BUFFER_SIZE 128
#define RX_RING_SIZE 512
-#define RX_RING_BYTES (sizeof(struct dma_desc) * RX_RING_SIZE)
+#define RX_RING_BYTES (sizeof(struct macb_dma_desc) * RX_RING_SIZE)
/* Make the IP header word-aligned (the ethernet header is 14 bytes) */
#define RX_OFFSET 2
#define TX_RING_SIZE 128
-#define DEF_TX_RING_PENDING (TX_RING_SIZE - 1)
-#define TX_RING_BYTES (sizeof(struct dma_desc) * TX_RING_SIZE)
-
-#define TX_RING_GAP(bp) \
- (TX_RING_SIZE - (bp)->tx_pending)
-#define TX_BUFFS_AVAIL(bp) \
- (((bp)->tx_tail <= (bp)->tx_head) ? \
- (bp)->tx_tail + (bp)->tx_pending - (bp)->tx_head : \
- (bp)->tx_tail - (bp)->tx_head - TX_RING_GAP(bp))
-#define NEXT_TX(n) (((n) + 1) & (TX_RING_SIZE - 1))
-
-#define NEXT_RX(n) (((n) + 1) & (RX_RING_SIZE - 1))
+#define TX_RING_BYTES (sizeof(struct macb_dma_desc) * TX_RING_SIZE)
/* minimum number of free TX descriptors before waking up TX process */
#define MACB_TX_WAKEUP_THRESH (TX_RING_SIZE / 4)
@@ -56,6 +45,51 @@
#define MACB_RX_INT_FLAGS (MACB_BIT(RCOMP) | MACB_BIT(RXUBR) \
| MACB_BIT(ISR_ROVR))
+/* Ring buffer accessors */
+static unsigned int macb_tx_ring_wrap(unsigned int index)
+{
+ return index & (TX_RING_SIZE - 1);
+}
+
+static unsigned int macb_tx_ring_avail(struct macb *bp)
+{
+ return TX_RING_SIZE - (bp->tx_head - bp->tx_tail);
+}
+
+static struct macb_dma_desc *macb_tx_desc(struct macb *bp, unsigned int index)
+{
+ return &bp->tx_ring[macb_tx_ring_wrap(index)];
+}
+
+static struct macb_tx_skb *macb_tx_skb(struct macb *bp, unsigned int index)
+{
+ return &bp->tx_skb[macb_tx_ring_wrap(index)];
+}
+
+static dma_addr_t macb_tx_dma(struct macb *bp, unsigned int index)
+{
+ dma_addr_t offset;
+
+ offset = macb_tx_ring_wrap(index) * sizeof(struct macb_dma_desc);
+
+ return bp->tx_ring_dma + offset;
+}
+
+static unsigned int macb_rx_ring_wrap(unsigned int index)
+{
+ return index & (RX_RING_SIZE - 1);
+}
+
+static struct macb_dma_desc *macb_rx_desc(struct macb *bp, unsigned int index)
+{
+ return &bp->rx_ring[macb_rx_ring_wrap(index)];
+}
+
+static void *macb_rx_buffer(struct macb *bp, unsigned int index)
+{
+ return bp->rx_buffers + RX_BUFFER_SIZE * macb_rx_ring_wrap(index);
+}
+
static void __macb_set_hwaddr(struct macb *bp)
{
u32 bottom;
@@ -335,17 +369,18 @@ static void macb_tx(struct macb *bp)
bp->tx_ring[TX_RING_SIZE - 1].ctrl |= MACB_BIT(TX_WRAP);
/* free transmit buffer in upper layer*/
- for (tail = bp->tx_tail; tail != head; tail = NEXT_TX(tail)) {
- struct ring_info *rp = &bp->tx_skb[tail];
- struct sk_buff *skb = rp->skb;
-
- BUG_ON(skb == NULL);
+ for (tail = bp->tx_tail; tail != head; tail++) {
+ struct macb_tx_skb *tx_skb;
+ struct sk_buff *skb;
rmb();
- dma_unmap_single(&bp->pdev->dev, rp->mapping, skb->len,
- DMA_TO_DEVICE);
- rp->skb = NULL;
+ tx_skb = macb_tx_skb(bp, tail);
+ skb = tx_skb->skb;
+
+ dma_unmap_single(&bp->pdev->dev, tx_skb->mapping,
+ skb->len, DMA_TO_DEVICE);
+ tx_skb->skb = NULL;
dev_kfree_skb_irq(skb);
}
@@ -365,28 +400,32 @@ static void macb_tx(struct macb *bp)
return;
head = bp->tx_head;
- for (tail = bp->tx_tail; tail != head; tail = NEXT_TX(tail)) {
- struct ring_info *rp = &bp->tx_skb[tail];
- struct sk_buff *skb = rp->skb;
- u32 bufstat;
+ for (tail = bp->tx_tail; tail != head; tail++) {
+ struct macb_tx_skb *tx_skb;
+ struct sk_buff *skb;
+ struct macb_dma_desc *desc;
+ u32 ctrl;
- BUG_ON(skb == NULL);
+ desc = macb_tx_desc(bp, tail);
/* Make hw descriptor updates visible to CPU */
rmb();
- bufstat = bp->tx_ring[tail].ctrl;
+ ctrl = desc->ctrl;
- if (!(bufstat & MACB_BIT(TX_USED)))
+ if (!(ctrl & MACB_BIT(TX_USED)))
break;
+ tx_skb = macb_tx_skb(bp, tail);
+ skb = tx_skb->skb;
+
netdev_vdbg(bp->dev, "skb %u (data %p) TX complete\n",
- tail, skb->data);
- dma_unmap_single(&bp->pdev->dev, rp->mapping, skb->len,
+ macb_tx_ring_wrap(tail), skb->data);
+ dma_unmap_single(&bp->pdev->dev, tx_skb->mapping, skb->len,
DMA_TO_DEVICE);
bp->stats.tx_packets++;
bp->stats.tx_bytes += skb->len;
- rp->skb = NULL;
+ tx_skb->skb = NULL;
dev_kfree_skb_irq(skb);
}
@@ -398,8 +437,8 @@ static void macb_tx(struct macb *bp)
macb_writel(bp, NCR, macb_readl(bp, NCR) | MACB_BIT(TSTART));
bp->tx_tail = tail;
- if (netif_queue_stopped(bp->dev) &&
- TX_BUFFS_AVAIL(bp) > MACB_TX_WAKEUP_THRESH)
+ if (netif_queue_stopped(bp->dev)
+ && macb_tx_ring_avail(bp) > MACB_TX_WAKEUP_THRESH)
netif_wake_queue(bp->dev);
}
@@ -410,17 +449,21 @@ static int macb_rx_frame(struct macb *bp, unsigned int first_frag,
unsigned int frag;
unsigned int offset = 0;
struct sk_buff *skb;
+ struct macb_dma_desc *desc;
- len = MACB_BFEXT(RX_FRMLEN, bp->rx_ring[last_frag].ctrl);
+ desc = macb_rx_desc(bp, last_frag);
+ len = MACB_BFEXT(RX_FRMLEN, desc->ctrl);
netdev_vdbg(bp->dev, "macb_rx_frame frags %u - %u (len %u)\n",
- first_frag, last_frag, len);
+ macb_rx_ring_wrap(first_frag),
+ macb_rx_ring_wrap(last_frag), len);
skb = netdev_alloc_skb(bp->dev, len + RX_OFFSET);
if (!skb) {
bp->stats.rx_dropped++;
- for (frag = first_frag; ; frag = NEXT_RX(frag)) {
- bp->rx_ring[frag].addr &= ~MACB_BIT(RX_USED);
+ for (frag = first_frag; ; frag++) {
+ desc = macb_rx_desc(bp, frag);
+ desc->addr &= ~MACB_BIT(RX_USED);
if (frag == last_frag)
break;
}
@@ -435,7 +478,7 @@ static int macb_rx_frame(struct macb *bp, unsigned int first_frag,
skb_checksum_none_assert(skb);
skb_put(skb, len);
- for (frag = first_frag; ; frag = NEXT_RX(frag)) {
+ for (frag = first_frag; ; frag++) {
unsigned int frag_len = RX_BUFFER_SIZE;
if (offset + frag_len > len) {
@@ -443,11 +486,10 @@ static int macb_rx_frame(struct macb *bp, unsigned int first_frag,
frag_len = len - offset;
}
skb_copy_to_linear_data_offset(skb, offset,
- (bp->rx_buffers +
- (RX_BUFFER_SIZE * frag)),
- frag_len);
+ macb_rx_buffer(bp, frag), frag_len);
offset += RX_BUFFER_SIZE;
- bp->rx_ring[frag].addr &= ~MACB_BIT(RX_USED);
+ desc = macb_rx_desc(bp, frag);
+ desc->addr &= ~MACB_BIT(RX_USED);
if (frag == last_frag)
break;
@@ -473,8 +515,10 @@ static void discard_partial_frame(struct macb *bp, unsigned int begin,
{
unsigned int frag;
- for (frag = begin; frag != end; frag = NEXT_RX(frag))
- bp->rx_ring[frag].addr &= ~MACB_BIT(RX_USED);
+ for (frag = begin; frag != end; frag++) {
+ struct macb_dma_desc *desc = macb_rx_desc(bp, frag);
+ desc->addr &= ~MACB_BIT(RX_USED);
+ }
/* Make descriptor updates visible to hardware */
wmb();
@@ -489,17 +533,18 @@ static void discard_partial_frame(struct macb *bp, unsigned int begin,
static int macb_rx(struct macb *bp, int budget)
{
int received = 0;
- unsigned int tail = bp->rx_tail;
+ unsigned int tail;
int first_frag = -1;
- for (; budget > 0; tail = NEXT_RX(tail)) {
+ for (tail = bp->rx_tail; budget > 0; tail++) {
+ struct macb_dma_desc *desc = macb_rx_desc(bp, tail);
u32 addr, ctrl;
/* Make hw descriptor updates visible to CPU */
rmb();
- addr = bp->rx_ring[tail].addr;
- ctrl = bp->rx_ring[tail].ctrl;
+ addr = desc->addr;
+ ctrl = desc->ctrl;
if (!(addr & MACB_BIT(RX_USED)))
break;
@@ -653,6 +698,8 @@ static int macb_start_xmit(struct sk_buff *skb, struct net_device *dev)
struct macb *bp = netdev_priv(dev);
dma_addr_t mapping;
unsigned int len, entry;
+ struct macb_dma_desc *desc;
+ struct macb_tx_skb *tx_skb;
u32 ctrl;
unsigned long flags;
@@ -669,7 +716,7 @@ static int macb_start_xmit(struct sk_buff *skb, struct net_device *dev)
spin_lock_irqsave(&bp->lock, flags);
/* This is a hard error, log it. */
- if (TX_BUFFS_AVAIL(bp) < 1) {
+ if (macb_tx_ring_avail(bp) < 1) {
netif_stop_queue(dev);
spin_unlock_irqrestore(&bp->lock, flags);
netdev_err(bp->dev, "BUG! Tx Ring full when queue awake!\n");
@@ -678,12 +725,15 @@ static int macb_start_xmit(struct sk_buff *skb, struct net_device *dev)
return NETDEV_TX_BUSY;
}
- entry = bp->tx_head;
+ entry = macb_tx_ring_wrap(bp->tx_head);
+ bp->tx_head++;
netdev_vdbg(bp->dev, "Allocated ring entry %u\n", entry);
mapping = dma_map_single(&bp->pdev->dev, skb->data,
len, DMA_TO_DEVICE);
- bp->tx_skb[entry].skb = skb;
- bp->tx_skb[entry].mapping = mapping;
+
+ tx_skb = &bp->tx_skb[entry];
+ tx_skb->skb = skb;
+ tx_skb->mapping = mapping;
netdev_vdbg(bp->dev, "Mapped skb data %p to DMA addr %08lx\n",
skb->data, (unsigned long)mapping);
@@ -692,15 +742,13 @@ static int macb_start_xmit(struct sk_buff *skb, struct net_device *dev)
if (entry == (TX_RING_SIZE - 1))
ctrl |= MACB_BIT(TX_WRAP);
- bp->tx_ring[entry].addr = mapping;
- bp->tx_ring[entry].ctrl = ctrl;
+ desc = &bp->tx_ring[entry];
+ desc->addr = mapping;
+ desc->ctrl = ctrl;
/* Make newly initialized descriptor visible to hardware */
wmb();
- entry = NEXT_TX(entry);
- bp->tx_head = entry;
-
skb_tx_timestamp(skb);
/*
@@ -713,10 +761,10 @@ static int macb_start_xmit(struct sk_buff *skb, struct net_device *dev)
* re-enable interrupts, and the interrupt handler will make
* sure the controler is started.
*/
- if (NEXT_TX(bp->tx_tail) == bp->tx_head)
+ if (bp->tx_tail == bp->tx_head - 1)
macb_writel(bp, NCR, macb_readl(bp, NCR) | MACB_BIT(TSTART));
- if (TX_BUFFS_AVAIL(bp) < 1)
+ if (macb_tx_ring_avail(bp) < 1)
netif_stop_queue(dev);
spin_unlock_irqrestore(&bp->lock, flags);
@@ -752,7 +800,7 @@ static int macb_alloc_consistent(struct macb *bp)
{
int size;
- size = TX_RING_SIZE * sizeof(struct ring_info);
+ size = TX_RING_SIZE * sizeof(struct macb_tx_skb);
bp->tx_skb = kmalloc(size, GFP_KERNEL);
if (!bp->tx_skb)
goto out_err;
@@ -1437,8 +1485,6 @@ static int __init macb_probe(struct platform_device *pdev)
macb_or_gem_writel(bp, USRIO, MACB_BIT(MII));
#endif
- bp->tx_pending = DEF_TX_RING_PENDING;
-
err = register_netdev(dev);
if (err) {
dev_err(&pdev->dev, "Cannot register net device, aborting.\n");
diff --git a/drivers/net/ethernet/cadence/macb.h b/drivers/net/ethernet/cadence/macb.h
index f69ceef..8a4ee2f 100644
--- a/drivers/net/ethernet/cadence/macb.h
+++ b/drivers/net/ethernet/cadence/macb.h
@@ -356,7 +356,12 @@
__v; \
})
-struct dma_desc {
+/**
+ * struct macb_dma_desc - Hardware DMA descriptor
+ * @addr: DMA address of data buffer
+ * @ctrl: Control and status bits
+ */
+struct macb_dma_desc {
u32 addr;
u32 ctrl;
};
@@ -421,7 +426,12 @@ struct dma_desc {
#define MACB_TX_USED_OFFSET 31
#define MACB_TX_USED_SIZE 1
-struct ring_info {
+/**
+ * struct macb_tx_skb - data about an skb which is being transmitted
+ * @skb: skb currently being transmitted
+ * @mapping: DMA address of the skb's data buffer
+ */
+struct macb_tx_skb {
struct sk_buff *skb;
dma_addr_t mapping;
};
@@ -506,12 +516,12 @@ struct macb {
void __iomem *regs;
unsigned int rx_tail;
- struct dma_desc *rx_ring;
+ struct macb_dma_desc *rx_ring;
void *rx_buffers;
unsigned int tx_head, tx_tail;
- struct dma_desc *tx_ring;
- struct ring_info *tx_skb;
+ struct macb_dma_desc *tx_ring;
+ struct macb_tx_skb *tx_skb;
spinlock_t lock;
struct platform_device *pdev;
@@ -529,8 +539,6 @@ struct macb {
dma_addr_t tx_ring_dma;
dma_addr_t rx_buffers_dma;
- unsigned int rx_pending, tx_pending;
-
struct mii_bus *mii_bus;
struct phy_device *phy_dev;
unsigned int link;
--
1.7.10
^ permalink raw reply related
* [PATCH 06/10] net/macb: better manage tx errors
From: Nicolas Ferre @ 2012-09-05 9:00 UTC (permalink / raw)
To: netdev
Cc: patrice.vilchez, nicolas.ferre, linux-kernel, havard, jamie,
plagnioj, davem, linux-arm-kernel
In-Reply-To: <cover.1346775479.git.nicolas.ferre@atmel.com>
Handle all TX errors, not only underruns.
Reinitialize the TX ring after skipping all remaining frames, and
restart the controller when everything has been cleaned up properly.
Original idea from a patch by Havard Skinnemoen.
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
drivers/net/ethernet/cadence/macb.c | 124 ++++++++++++++++++++---------------
1 file changed, 71 insertions(+), 53 deletions(-)
diff --git a/drivers/net/ethernet/cadence/macb.c b/drivers/net/ethernet/cadence/macb.c
index 3d3a077..af71151 100644
--- a/drivers/net/ethernet/cadence/macb.c
+++ b/drivers/net/ethernet/cadence/macb.c
@@ -44,6 +44,10 @@
#define MACB_RX_INT_FLAGS (MACB_BIT(RCOMP) | MACB_BIT(RXUBR) \
| MACB_BIT(ISR_ROVR))
+#define MACB_TX_INT_FLAGS (MACB_BIT(ISR_TUND) \
+ | MACB_BIT(ISR_RLE) \
+ | MACB_BIT(TXERR) \
+ | MACB_BIT(TCOMP))
/* Ring buffer accessors */
static unsigned int macb_tx_ring_wrap(unsigned int index)
@@ -338,66 +342,56 @@ static void macb_update_stats(struct macb *bp)
*p += __raw_readl(reg);
}
-static void macb_tx(struct macb *bp)
+static void macb_handle_tx_error(struct macb *bp, unsigned int err_tail, u32 ctrl)
{
- unsigned int tail;
- unsigned int head;
- u32 status;
-
- status = macb_readl(bp, TSR);
- macb_writel(bp, TSR, status);
+ struct macb_tx_skb *tx_skb;
+ struct sk_buff *skb;
+ unsigned int head = bp->tx_head;
- netdev_vdbg(bp->dev, "macb_tx status = %02lx\n", (unsigned long)status);
+ netdev_dbg(bp->dev, "TX error: ctrl 0x%08x, head %u, error tail %u\n",
+ ctrl, head, err_tail);
- if (status & (MACB_BIT(UND) | MACB_BIT(TSR_RLE))) {
- int i;
- netdev_err(bp->dev, "TX %s, resetting buffers\n",
- status & MACB_BIT(UND) ?
- "underrun" : "retry limit exceeded");
-
- /* Transfer ongoing, disable transmitter, to avoid confusion */
- if (status & MACB_BIT(TGO))
- macb_writel(bp, NCR, macb_readl(bp, NCR) & ~MACB_BIT(TE));
-
- head = bp->tx_head;
-
- /*Mark all the buffer as used to avoid sending a lost buffer*/
- for (i = 0; i < TX_RING_SIZE; i++)
- bp->tx_ring[i].ctrl = MACB_BIT(TX_USED);
-
- /* Add wrap bit */
- bp->tx_ring[TX_RING_SIZE - 1].ctrl |= MACB_BIT(TX_WRAP);
+ /*
+ * "Buffers exhausted mid-frame" errors may only happen if the
+ * driver is buggy, so complain loudly about those. Statistics
+ * are updated by hardware.
+ */
+ if (ctrl & MACB_BIT(TX_BUF_EXHAUSTED))
+ netdev_err(bp->dev, "BUG: TX buffers exhausted mid-frame\n");
- /* free transmit buffer in upper layer*/
- for (tail = bp->tx_tail; tail != head; tail++) {
- struct macb_tx_skb *tx_skb;
- struct sk_buff *skb;
+ /*
+ * Drop the frames that caused the error plus all remaining in queue.
+ * Free transmit buffers in upper layer.
+ */
+ for (; err_tail != head; err_tail++) {
+ struct macb_dma_desc *desc;
- rmb();
+ tx_skb = macb_tx_skb(bp, err_tail);
+ skb = tx_skb->skb;
+ dma_unmap_single(&bp->pdev->dev, tx_skb->mapping, skb->len,
+ DMA_TO_DEVICE);
+ dev_kfree_skb_irq(skb);
+ tx_skb->skb = NULL;
- tx_skb = macb_tx_skb(bp, tail);
- skb = tx_skb->skb;
+ desc = macb_tx_desc(bp, err_tail);
+ desc->ctrl |= MACB_BIT(TX_USED);
+ }
- dma_unmap_single(&bp->pdev->dev, tx_skb->mapping,
- skb->len, DMA_TO_DEVICE);
- tx_skb->skb = NULL;
- dev_kfree_skb_irq(skb);
- }
+ /* Make descriptor updates visible to hardware */
+ wmb();
+}
- bp->tx_head = bp->tx_tail = 0;
+static void macb_tx_interrupt(struct macb *bp)
+{
+ unsigned int tail;
+ unsigned int head;
+ u32 status;
- /* Enable the transmitter again */
- if (status & MACB_BIT(TGO))
- macb_writel(bp, NCR, macb_readl(bp, NCR) | MACB_BIT(TE));
- }
+ status = macb_readl(bp, TSR);
+ macb_writel(bp, TSR, status);
- if (!(status & MACB_BIT(COMP)))
- /*
- * This may happen when a buffer becomes complete
- * between reading the ISR and scanning the
- * descriptors. Nothing to worry about.
- */
- return;
+ netdev_vdbg(bp->dev, "macb_tx_interrupt status = %02lx\n",
+ (unsigned long)status);
head = bp->tx_head;
for (tail = bp->tx_tail; tail != head; tail++) {
@@ -413,6 +407,31 @@ static void macb_tx(struct macb *bp)
ctrl = desc->ctrl;
+ if (unlikely(ctrl & (MACB_BIT(TX_ERROR)
+ | MACB_BIT(TX_UNDERRUN)
+ | MACB_BIT(TX_BUF_EXHAUSTED)))) {
+ /*
+ * In case of transfer ongoing, disable transmitter.
+ * Should already be the case due to hardware,
+ * but make sure to avoid confusion.
+ */
+ if (status & MACB_BIT(TGO))
+ macb_writel(bp, NCR, macb_readl(bp, NCR) & ~MACB_BIT(TE));
+
+ /*
+ * An error should always stop the queue from advancing.
+ * reset entries in the ring and exit from the loop.
+ */
+ macb_handle_tx_error(bp, tail, ctrl);
+ bp->tx_head = bp->tx_tail = head = tail = 0;
+
+ /* Enable the transmitter again, start TX will be done elsewhere */
+ if (status & MACB_BIT(TGO))
+ macb_writel(bp, NCR, macb_readl(bp, NCR) | MACB_BIT(TE));
+
+ break;
+ }
+
if (!(ctrl & MACB_BIT(TX_USED)))
break;
@@ -644,9 +663,8 @@ static irqreturn_t macb_interrupt(int irq, void *dev_id)
}
}
- if (status & (MACB_BIT(TCOMP) | MACB_BIT(ISR_TUND) |
- MACB_BIT(ISR_RLE)))
- macb_tx(bp);
+ if (status & MACB_TX_INT_FLAGS)
+ macb_tx_interrupt(bp);
/*
* Link change detection isn't possible with RMII, so we'll
--
1.7.10
^ permalink raw reply related
* [PATCH 07/10] net/macb: tx status is more than 8 bits now
From: Nicolas Ferre @ 2012-09-05 9:00 UTC (permalink / raw)
To: netdev
Cc: linux-arm-kernel, davem, havard, nicolas.ferre, plagnioj, jamie,
linux-kernel, patrice.vilchez
In-Reply-To: <cover.1346775479.git.nicolas.ferre@atmel.com>
On some revision of GEM, TSR status register is has more information.
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
drivers/net/ethernet/cadence/macb.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/cadence/macb.c b/drivers/net/ethernet/cadence/macb.c
index af71151..bd331fd 100644
--- a/drivers/net/ethernet/cadence/macb.c
+++ b/drivers/net/ethernet/cadence/macb.c
@@ -390,7 +390,7 @@ static void macb_tx_interrupt(struct macb *bp)
status = macb_readl(bp, TSR);
macb_writel(bp, TSR, status);
- netdev_vdbg(bp->dev, "macb_tx_interrupt status = %02lx\n",
+ netdev_vdbg(bp->dev, "macb_tx_interrupt status = 0x%03lx\n",
(unsigned long)status);
head = bp->tx_head;
--
1.7.10
^ permalink raw reply related
* [PATCH 08/10] net/macb: macb_get_drvinfo: add GEM/MACB suffix to differentiate revision
From: Nicolas Ferre @ 2012-09-05 9:00 UTC (permalink / raw)
To: netdev
Cc: linux-arm-kernel, davem, havard, nicolas.ferre, plagnioj, jamie,
linux-kernel, patrice.vilchez
In-Reply-To: <cover.1346775479.git.nicolas.ferre@atmel.com>
Add an indication about which revision of the hardware we are running in
info->driver string.
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
drivers/net/ethernet/cadence/macb.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/drivers/net/ethernet/cadence/macb.c b/drivers/net/ethernet/cadence/macb.c
index bd331fd..c7c39f1 100644
--- a/drivers/net/ethernet/cadence/macb.c
+++ b/drivers/net/ethernet/cadence/macb.c
@@ -1313,6 +1313,10 @@ static void macb_get_drvinfo(struct net_device *dev,
struct macb *bp = netdev_priv(dev);
strcpy(info->driver, bp->pdev->dev.driver->name);
+ if (macb_is_gem(bp))
+ strcat(info->driver, " GEM");
+ else
+ strcat(info->driver, " MACB");
strcpy(info->version, "$Revision: 1.14 $");
strcpy(info->bus_info, dev_name(&bp->pdev->dev));
}
--
1.7.10
^ permalink raw reply related
* [PATCH 09/10] net/macb: ethtool interface: add register dump feature
From: Nicolas Ferre @ 2012-09-05 9:00 UTC (permalink / raw)
To: netdev
Cc: linux-arm-kernel, davem, havard, nicolas.ferre, plagnioj, jamie,
linux-kernel, patrice.vilchez
In-Reply-To: <cover.1346775479.git.nicolas.ferre@atmel.com>
Add macb_get_regs() ethtool function and its helper function:
macb_get_regs_len().
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
drivers/net/ethernet/cadence/macb.c | 40 +++++++++++++++++++++++++++++++++++
drivers/net/ethernet/cadence/macb.h | 3 +++
2 files changed, 43 insertions(+)
diff --git a/drivers/net/ethernet/cadence/macb.c b/drivers/net/ethernet/cadence/macb.c
index c7c39f1..f31c0a7 100644
--- a/drivers/net/ethernet/cadence/macb.c
+++ b/drivers/net/ethernet/cadence/macb.c
@@ -1321,10 +1321,50 @@ static void macb_get_drvinfo(struct net_device *dev,
strcpy(info->bus_info, dev_name(&bp->pdev->dev));
}
+static int macb_get_regs_len(struct net_device *netdev)
+{
+ return MACB_GREGS_LEN * sizeof(u32);
+}
+
+static void macb_get_regs(struct net_device *dev, struct ethtool_regs *regs,
+ void *p)
+{
+ struct macb *bp = netdev_priv(dev);
+ unsigned int tail, head;
+ u32 *regs_buff = p;
+
+ memset(p, 0, MACB_GREGS_LEN * sizeof(u32));
+ regs->version = MACB_BFEXT(IDNUM, macb_readl(bp, MID));
+
+ tail = macb_tx_ring_wrap(bp->tx_tail);
+ head = macb_tx_ring_wrap(bp->tx_head);
+
+ regs_buff[0] = macb_readl(bp, NCR);
+ regs_buff[1] = macb_or_gem_readl(bp, NCFGR);
+ regs_buff[2] = macb_readl(bp, NSR);
+ regs_buff[3] = macb_readl(bp, TSR);
+ regs_buff[4] = macb_readl(bp, RBQP);
+ regs_buff[5] = macb_readl(bp, TBQP);
+ regs_buff[6] = macb_readl(bp, RSR);
+ regs_buff[7] = macb_readl(bp, IMR);
+
+ regs_buff[8] = tail;
+ regs_buff[9] = head;
+ regs_buff[10] = macb_tx_dma(bp, tail);
+ regs_buff[11] = macb_tx_dma(bp, head);
+
+ if (macb_is_gem(bp)) {
+ regs_buff[12] = gem_readl(bp, USRIO);
+ regs_buff[13] = gem_readl(bp, DMACFG);
+ }
+}
+
static const struct ethtool_ops macb_ethtool_ops = {
.get_settings = macb_get_settings,
.set_settings = macb_set_settings,
.get_drvinfo = macb_get_drvinfo,
+ .get_regs_len = macb_get_regs_len,
+ .get_regs = macb_get_regs,
.get_link = ethtool_op_get_link,
.get_ts_info = ethtool_op_get_ts_info,
};
diff --git a/drivers/net/ethernet/cadence/macb.h b/drivers/net/ethernet/cadence/macb.h
index 8a4ee2f..d509e88 100644
--- a/drivers/net/ethernet/cadence/macb.h
+++ b/drivers/net/ethernet/cadence/macb.h
@@ -10,6 +10,9 @@
#ifndef _MACB_H
#define _MACB_H
+
+#define MACB_GREGS_LEN 32
+
/* MACB register offsets */
#define MACB_NCR 0x0000
#define MACB_NCFGR 0x0004
--
1.7.10
^ permalink raw reply related
* Aw: Re: Question regarding kernel panic in net/ipv4/tcp_output.c
From: "Sascha Mühlbach" @ 2012-09-05 9:01 UTC (permalink / raw)
To: Eric Dumazet; +Cc: netdev
In-Reply-To: <1346761388.13121.21.camel@edumazet-glaptop>
Hi,
this is the stack trace I could save by making a screenshot of the remote console:
[<ffffffff8101193b>] ? invalid_op+0x1b/0x20
[<ffffffff81289e6a>] ? tcp_retransmit_skb+0x66/0x5aa
[<ffffffff8128a546>] ? tcp_xmit_retransmit_queue+0x198/0x223
[<ffffffff81286113>] ? tcp_ack+0x1744/0x1952
[<ffffffff81286871>] ? tcp_validate_incoming+0x1ba/0x2be
[<ffffffff81286f5e>] ? tcp_rcv_established+0x5e9/0x6d9
[<ffffffff8128e00f>] ? tcp_v4_do_rcv+0x1bb/0x376
[<ffffffff8128e639>] ? tcp_v4_rcv+0x46f/0x6f8
[<ffffffff81273afa>] ? ip_local_deliver_finish+0x0/0x1e9
[<ffffffff81273afa>] ? ip_local_deliver_finish+0x0/0x1e9
[<ffffffff81273c40>] ? ip_local_deliver_finish+0x146/0x1e9
[<ffffffff8127378f>] ? ip_rcv_finish+0x373/0x38d
[<ffffffffa004d95c>] ? bnx2_poll_work+0x954/0xa7e [bnx2]
[<ffffffffa004d95c>] ? bnx2_poll_work+0x954/0xa7e [bnx2]
[<ffffffff8105aeb6>] ? __mod_timer+0x141/0x153
[<ffffffff810964a1>] ? handle_edge_irq+0xdd/0x101
[<ffffffffa004daae>] ? bnx2_poll_msix+0x28/0xa6 [bnx2]
[<ffffffff8125039f>] ? net_rx_action+0xae/0x1c9
[<ffffffff81053d6f>] ? __do_softirq+0xdd/0x1a6
[<ffffffff81011cac>] ? call_softirq+0x1c/0x30
[<ffffffff8101322b>] ? do_softirq+0x3f/0x7c
[<ffffffff81053bdf>] ? irq_exit+0x36/0x76
[<ffffffff81012922>] ? do_IRQ+0xa0/0xb6
[<ffffffff810114d3>] ? ret_from_intr+0x0/0x11
<E0I> [<ffffffffa0135509>] ? acpi_idle_enter_bm+0x27d/0x2af [processor]
[<ffffffffa0135509>] ? acpi_idle_enter_bm+0x27d/0x2af [processor]
[<ffffffffa0135502>] ? acpi_idle_enter_bm+0x276/0x2af [processor]
[<ffffffff8123a2c6>] ? cpuidle_idle_call+0x94/0xee
[<ffffffff8100fe97>] ? cpu_idle+0xa2/0xda
[<ffffffff8151c140>] ? early_idt_handler+0x0/0x71
[<ffffffff8151ccdd>] ? start_kernel+0x3dc/0x3e8
[<ffffffff8151c3b7>] ? x86_64_start_kernel+0xf9/0x106
Kind Regards,
Sascha
>Gesendet: Dienstag, 04. September 2012 um 14:23 Uhr
>Von: "Eric Dumazet" <eric.dumazet@gmail.com>
>An: smka2012@email.de
>Cc: netdev@vger.kernel.org
>Betreff: Re: Question regarding kernel panic in net/ipv4/tcp_output.c
>On Tue, 2012-09-04 at 13:55 +0200, smka2012@email.de wrote:
>> Hi,
>>
>> I recently had a severe issue with multiple servers that stopped
>> working at the same time. By using the stacktrace, I could locate the
>> following line in the tcp_retransmit_skb function of the tcp_output.c
>> kernel source that led to a kernel panic on all servers. I am using
>> Debian 6.0 with the kernel version 2.6.32-45.
>>
>> (gdb) list *(tcp_retransmit_skb+0x66)
>> 0x1a93 is in tcp_retransmit_skb (net/ipv4/tcp_output.c:1905).
>> 1900 if (atomic_read(&sk->sk_wmem_alloc) >
>> 1901 min(sk->sk_wmem_queued + (sk->sk_wmem_queued >> 2),
>> sk->sk_sndbuf))
>> 1902 return -EAGAIN;
>> 1903
>> 1904 if (before(TCP_SKB_CB(skb)->seq, tp->snd_una)) {
>>
>> --->
>> 1905 if (before(TCP_SKB_CB(skb)->end_seq, tp->snd_una))
>> 1906 BUG();
>> <---
>>
>> 1907 if (tcp_trim_head(sk, skb, tp->snd_una - TCP_SKB_CB(skb)->seq))
>> 1908 return -ENOMEM;
>> 1909 }
>>
>> Now I am interested in getting to know what can lead to a situation
>> that the condition in line 1905 evaluates true and why the kernel goes
>> into the BUG() function in that case and does not only return an
>> error. All servers reached this line of code. They all were connected
>> to a switch that broke the same time. However, I cannot say if the
>> switch broke before the servers and eventually affected the servers or
>> if the switch was also itself affected by some external event.
>>
>> Thank you very much for your help.
>>
>> Kind Regards,
>> Sascha
>> --
>
>You can see this BUG() as an early notification of a hard to
>debug/diagnose bug.
>
>We shouldnt take this path at all.
>
>If we do, we have an earlier bug that we should fix anyway, because
>machine is going to crash.
>
>It would be nice you sent the stack trace you had.
^ permalink raw reply
* [PATCH 10/10] net/macb: Offset first RX buffer by two bytes
From: Nicolas Ferre @ 2012-09-05 9:04 UTC (permalink / raw)
To: netdev
Cc: linux-arm-kernel, davem, havard, nicolas.ferre, plagnioj, jamie,
linux-kernel, patrice.vilchez
In-Reply-To: <cover.1346775479.git.nicolas.ferre@atmel.com>
From: Havard Skinnemoen <havard@skinnemoen.net>
Make the ethernet frame payload word-aligned, possibly making the
memcpy into the skb a bit faster. This will be even more important
after we eliminate the copy altogether.
Also eliminate the redundant RX_OFFSET constant -- it has the same
definition and purpose as NET_IP_ALIGN.
Signed-off-by: Havard Skinnemoen <havard@skinnemoen.net>
[nicolas.ferre@atmel.com: adapt to newer kernel]
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
---
drivers/net/ethernet/cadence/macb.c | 23 ++++++++++++++++-------
1 file changed, 16 insertions(+), 7 deletions(-)
diff --git a/drivers/net/ethernet/cadence/macb.c b/drivers/net/ethernet/cadence/macb.c
index f31c0a7..f7716b6 100644
--- a/drivers/net/ethernet/cadence/macb.c
+++ b/drivers/net/ethernet/cadence/macb.c
@@ -33,9 +33,6 @@
#define RX_RING_SIZE 512
#define RX_RING_BYTES (sizeof(struct macb_dma_desc) * RX_RING_SIZE)
-/* Make the IP header word-aligned (the ethernet header is 14 bytes) */
-#define RX_OFFSET 2
-
#define TX_RING_SIZE 128
#define TX_RING_BYTES (sizeof(struct macb_dma_desc) * TX_RING_SIZE)
@@ -466,7 +463,7 @@ static int macb_rx_frame(struct macb *bp, unsigned int first_frag,
{
unsigned int len;
unsigned int frag;
- unsigned int offset = 0;
+ unsigned int offset;
struct sk_buff *skb;
struct macb_dma_desc *desc;
@@ -477,7 +474,16 @@ static int macb_rx_frame(struct macb *bp, unsigned int first_frag,
macb_rx_ring_wrap(first_frag),
macb_rx_ring_wrap(last_frag), len);
- skb = netdev_alloc_skb(bp->dev, len + RX_OFFSET);
+ /*
+ * The ethernet header starts NET_IP_ALIGN bytes into the
+ * first buffer. Since the header is 14 bytes, this makes the
+ * payload word-aligned.
+ *
+ * Instead of calling skb_reserve(NET_IP_ALIGN), we just copy
+ * the two padding bytes into the skb so that we avoid hitting
+ * the slowpath in memcpy(), and pull them off afterwards.
+ */
+ skb = netdev_alloc_skb(bp->dev, len + NET_IP_ALIGN);
if (!skb) {
bp->stats.rx_dropped++;
for (frag = first_frag; ; frag++) {
@@ -493,7 +499,8 @@ static int macb_rx_frame(struct macb *bp, unsigned int first_frag,
return 1;
}
- skb_reserve(skb, RX_OFFSET);
+ offset = 0;
+ len += NET_IP_ALIGN;
skb_checksum_none_assert(skb);
skb_put(skb, len);
@@ -517,10 +524,11 @@ static int macb_rx_frame(struct macb *bp, unsigned int first_frag,
/* Make descriptor updates visible to hardware */
wmb();
+ __skb_pull(skb, NET_IP_ALIGN);
skb->protocol = eth_type_trans(skb, bp->dev);
bp->stats.rx_packets++;
- bp->stats.rx_bytes += len;
+ bp->stats.rx_bytes += skb->len;
netdev_vdbg(bp->dev, "received skb of length %u, csum: %08x\n",
skb->len, skb->csum);
netif_receive_skb(skb);
@@ -985,6 +993,7 @@ static void macb_init_hw(struct macb *bp)
__macb_set_hwaddr(bp);
config = macb_mdc_clk_div(bp);
+ config |= MACB_BF(RBOF, NET_IP_ALIGN); /* Make eth data aligned */
config |= MACB_BIT(PAE); /* PAuse Enable */
config |= MACB_BIT(DRFCS); /* Discard Rx FCS */
config |= MACB_BIT(BIG); /* Receive oversized frames */
--
1.7.10
^ permalink raw reply related
* [RFC PATCH v2] ipv6: fix handling of blackhole and prohibit routes
From: Nicolas Dichtel @ 2012-09-05 11:34 UTC (permalink / raw)
To: davem; +Cc: netdev, Nicolas Dichtel
In-Reply-To: <20120904.155836.51164588359279575.davem@davemloft.net>
When adding a blackhole or a prohibit route, they were handling like classic
routes. Moreover, it was only possible to add this kind of routes by specifying
an interface.
Bug already reported here:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=498498
Before the patch:
$ ip route add blackhole 2001::1/128
RTNETLINK answers: No such device
$ ip route add blackhole 2001::1/128 dev eth0
$ ip -6 route | grep 2001
2001::1 dev eth0 metric 1024
After:
$ ip route add blackhole 2001::1/128
$ ip -6 route | grep 2001
blackhole 2001::1 dev lo metric 1024 error -22
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
---
include/net/ip6_fib.h | 1 +
net/ipv6/route.c | 32 ++++++++++++++++++++++++++++----
2 files changed, 29 insertions(+), 4 deletions(-)
diff --git a/include/net/ip6_fib.h b/include/net/ip6_fib.h
index 0fedbd8..cd64cf3 100644
--- a/include/net/ip6_fib.h
+++ b/include/net/ip6_fib.h
@@ -37,6 +37,7 @@ struct fib6_config {
int fc_ifindex;
u32 fc_flags;
u32 fc_protocol;
+ u32 fc_type; /* only 8 bits are used */
struct in6_addr fc_dst;
struct in6_addr fc_src;
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index 8e80fd2..5642fb5 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -1463,8 +1463,18 @@ int ip6_route_add(struct fib6_config *cfg)
}
rt->dst.output = ip6_pkt_discard_out;
rt->dst.input = ip6_pkt_discard;
- rt->dst.error = -ENETUNREACH;
rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
+ switch (cfg->fc_type) {
+ case RTM_BLACKHOLE:
+ rt->dst.error = -EINVAL;
+ break;
+ case RTM_PROHIBIT:
+ rt->dst.error = -EACCES;
+ break;
+ default:
+ rt->dst.error = -ENETUNREACH;
+ break;
+ }
goto install_route;
}
@@ -2261,8 +2271,11 @@ static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
cfg->fc_src_len = rtm->rtm_src_len;
cfg->fc_flags = RTF_UP;
cfg->fc_protocol = rtm->rtm_protocol;
+ cfg->type = rtm->rtm_type;
- if (rtm->rtm_type == RTN_UNREACHABLE)
+ if (rtm->rtm_type == RTN_UNREACHABLE ||
+ rtm->rtm_type == RTN_BLACKHOLE ||
+ rtm->rtm_type == RTN_PROHIBIT)
cfg->fc_flags |= RTF_REJECT;
if (rtm->rtm_type == RTN_LOCAL)
@@ -2391,8 +2404,19 @@ static int rt6_fill_node(struct net *net,
rtm->rtm_table = table;
if (nla_put_u32(skb, RTA_TABLE, table))
goto nla_put_failure;
- if (rt->rt6i_flags & RTF_REJECT)
- rtm->rtm_type = RTN_UNREACHABLE;
+ if (rt->rt6i_flags & RTF_REJECT) {
+ switch (rt->dst.error) {
+ case -EINVAL:
+ rtm->rtm_type = RTN_BLACKHOLE;
+ break;
+ case -EACCES:
+ rtm->rtm_type = RTN_PROHIBIT;
+ break;
+ default:
+ rtm->rtm_type = RTN_UNREACHABLE;
+ break;
+ }
+ }
else if (rt->rt6i_flags & RTF_LOCAL)
rtm->rtm_type = RTN_LOCAL;
else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
--
1.7.12
^ permalink raw reply related
* Re: [RFC PATCH v2] ipv6: fix handling of blackhole and prohibit routes
From: Nicolas Dichtel @ 2012-09-05 10:03 UTC (permalink / raw)
To: davem; +Cc: netdev
In-Reply-To: <1346844858-3210-1-git-send-email-nicolas.dichtel@6wind.com>
Please, forget this patch, it's a wrong version.
Sorry for that.
Regards,
Nicolas
Le 05/09/2012 13:34, Nicolas Dichtel a écrit :
> When adding a blackhole or a prohibit route, they were handling like classic
> routes. Moreover, it was only possible to add this kind of routes by specifying
> an interface.
>
> Bug already reported here:
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=498498
>
> Before the patch:
> $ ip route add blackhole 2001::1/128
> RTNETLINK answers: No such device
> $ ip route add blackhole 2001::1/128 dev eth0
> $ ip -6 route | grep 2001
> 2001::1 dev eth0 metric 1024
>
> After:
> $ ip route add blackhole 2001::1/128
> $ ip -6 route | grep 2001
> blackhole 2001::1 dev lo metric 1024 error -22
>
> Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
> ---
> include/net/ip6_fib.h | 1 +
> net/ipv6/route.c | 32 ++++++++++++++++++++++++++++----
> 2 files changed, 29 insertions(+), 4 deletions(-)
>
> diff --git a/include/net/ip6_fib.h b/include/net/ip6_fib.h
> index 0fedbd8..cd64cf3 100644
> --- a/include/net/ip6_fib.h
> +++ b/include/net/ip6_fib.h
> @@ -37,6 +37,7 @@ struct fib6_config {
> int fc_ifindex;
> u32 fc_flags;
> u32 fc_protocol;
> + u32 fc_type; /* only 8 bits are used */
>
> struct in6_addr fc_dst;
> struct in6_addr fc_src;
> diff --git a/net/ipv6/route.c b/net/ipv6/route.c
> index 8e80fd2..5642fb5 100644
> --- a/net/ipv6/route.c
> +++ b/net/ipv6/route.c
> @@ -1463,8 +1463,18 @@ int ip6_route_add(struct fib6_config *cfg)
> }
> rt->dst.output = ip6_pkt_discard_out;
> rt->dst.input = ip6_pkt_discard;
> - rt->dst.error = -ENETUNREACH;
> rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
> + switch (cfg->fc_type) {
> + case RTM_BLACKHOLE:
> + rt->dst.error = -EINVAL;
> + break;
> + case RTM_PROHIBIT:
> + rt->dst.error = -EACCES;
> + break;
> + default:
> + rt->dst.error = -ENETUNREACH;
> + break;
> + }
> goto install_route;
> }
>
> @@ -2261,8 +2271,11 @@ static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
> cfg->fc_src_len = rtm->rtm_src_len;
> cfg->fc_flags = RTF_UP;
> cfg->fc_protocol = rtm->rtm_protocol;
> + cfg->type = rtm->rtm_type;
>
> - if (rtm->rtm_type == RTN_UNREACHABLE)
> + if (rtm->rtm_type == RTN_UNREACHABLE ||
> + rtm->rtm_type == RTN_BLACKHOLE ||
> + rtm->rtm_type == RTN_PROHIBIT)
> cfg->fc_flags |= RTF_REJECT;
>
> if (rtm->rtm_type == RTN_LOCAL)
> @@ -2391,8 +2404,19 @@ static int rt6_fill_node(struct net *net,
> rtm->rtm_table = table;
> if (nla_put_u32(skb, RTA_TABLE, table))
> goto nla_put_failure;
> - if (rt->rt6i_flags & RTF_REJECT)
> - rtm->rtm_type = RTN_UNREACHABLE;
> + if (rt->rt6i_flags & RTF_REJECT) {
> + switch (rt->dst.error) {
> + case -EINVAL:
> + rtm->rtm_type = RTN_BLACKHOLE;
> + break;
> + case -EACCES:
> + rtm->rtm_type = RTN_PROHIBIT;
> + break;
> + default:
> + rtm->rtm_type = RTN_UNREACHABLE;
> + break;
> + }
> + }
> else if (rt->rt6i_flags & RTF_LOCAL)
> rtm->rtm_type = RTN_LOCAL;
> else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
>
--
Nicolas DICHTEL
6WIND
R&D Engineer
Tel: +33 1 39 30 92 10
Fax: +33 1 39 30 92 11
nicolas.dichtel@6wind.com
www.6wind.com
Twitter: http://twitter.com/6windsoftware
Join the Multicore Packet Processing Forum: www.multicorepacketprocessing.com
Ce courriel ainsi que toutes les pièces jointes, est uniquement destiné à son ou
ses destinataires. Il contient des informations confidentielles qui sont la
propriété de 6WIND. Toute révélation, distribution ou copie des informations
qu'il contient est strictement interdite. Si vous avez reçu ce message par
erreur, veuillez immédiatement le signaler à l'émetteur et détruire toutes les
données reçues.
This e-mail message, including any attachments, is for the sole use of the
intended recipient(s) and contains information that is confidential and
proprietary to 6WIND. All unauthorized review, use, disclosure or distribution
is prohibited. If you are not the intended recipient, please contact the sender
by reply e-mail and destroy all copies of the original message.
^ permalink raw reply
* [RFC PATCH v3] ipv6: fix handling of blackhole and prohibit routes
From: Nicolas Dichtel @ 2012-09-05 12:12 UTC (permalink / raw)
To: davem; +Cc: netdev, Nicolas Dichtel
In-Reply-To: <50472379.10608@6wind.com>
When adding a blackhole or a prohibit route, they were handling like classic
routes. Moreover, it was only possible to add this kind of routes by specifying
an interface.
Bug already reported here:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=498498
Before the patch:
$ ip route add blackhole 2001::1/128
RTNETLINK answers: No such device
$ ip route add blackhole 2001::1/128 dev eth0
$ ip -6 route | grep 2001
2001::1 dev eth0 metric 1024
After:
$ ip route add blackhole 2001::1/128
$ ip -6 route | grep 2001
blackhole 2001::1 dev lo metric 1024 error -22
v2: wrong patch
v3: add a field fc_type in struct fib6_config to store RTN_* type
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
---
include/net/ip6_fib.h | 1 +
net/ipv6/route.c | 32 ++++++++++++++++++++++++++++----
2 files changed, 29 insertions(+), 4 deletions(-)
diff --git a/include/net/ip6_fib.h b/include/net/ip6_fib.h
index 0fedbd8..cd64cf3 100644
--- a/include/net/ip6_fib.h
+++ b/include/net/ip6_fib.h
@@ -37,6 +37,7 @@ struct fib6_config {
int fc_ifindex;
u32 fc_flags;
u32 fc_protocol;
+ u32 fc_type; /* only 8 bits are used */
struct in6_addr fc_dst;
struct in6_addr fc_src;
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index 8e80fd2..63de1c1 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -1463,8 +1463,18 @@ int ip6_route_add(struct fib6_config *cfg)
}
rt->dst.output = ip6_pkt_discard_out;
rt->dst.input = ip6_pkt_discard;
- rt->dst.error = -ENETUNREACH;
rt->rt6i_flags = RTF_REJECT|RTF_NONEXTHOP;
+ switch (cfg->fc_type) {
+ case RTN_BLACKHOLE:
+ rt->dst.error = -EINVAL;
+ break;
+ case RTN_PROHIBIT:
+ rt->dst.error = -EACCES;
+ break;
+ default:
+ rt->dst.error = -ENETUNREACH;
+ break;
+ }
goto install_route;
}
@@ -2261,8 +2271,11 @@ static int rtm_to_fib6_config(struct sk_buff *skb, struct nlmsghdr *nlh,
cfg->fc_src_len = rtm->rtm_src_len;
cfg->fc_flags = RTF_UP;
cfg->fc_protocol = rtm->rtm_protocol;
+ cfg->fc_type = rtm->rtm_type;
- if (rtm->rtm_type == RTN_UNREACHABLE)
+ if (rtm->rtm_type == RTN_UNREACHABLE ||
+ rtm->rtm_type == RTN_BLACKHOLE ||
+ rtm->rtm_type == RTN_PROHIBIT)
cfg->fc_flags |= RTF_REJECT;
if (rtm->rtm_type == RTN_LOCAL)
@@ -2391,8 +2404,19 @@ static int rt6_fill_node(struct net *net,
rtm->rtm_table = table;
if (nla_put_u32(skb, RTA_TABLE, table))
goto nla_put_failure;
- if (rt->rt6i_flags & RTF_REJECT)
- rtm->rtm_type = RTN_UNREACHABLE;
+ if (rt->rt6i_flags & RTF_REJECT) {
+ switch (rt->dst.error) {
+ case -EINVAL:
+ rtm->rtm_type = RTN_BLACKHOLE;
+ break;
+ case -EACCES:
+ rtm->rtm_type = RTN_PROHIBIT;
+ break;
+ default:
+ rtm->rtm_type = RTN_UNREACHABLE;
+ break;
+ }
+ }
else if (rt->rt6i_flags & RTF_LOCAL)
rtm->rtm_type = RTN_LOCAL;
else if (rt->dst.dev && (rt->dst.dev->flags & IFF_LOOPBACK))
--
1.7.12
^ permalink raw reply related
* [PATCH net-next] net: qdisc busylock needs lockdep annotations
From: Eric Dumazet @ 2012-09-05 11:02 UTC (permalink / raw)
To: David Miller; +Cc: netdev
From: Eric Dumazet <edumazet@google.com>
It seems we need to provide ability for stacked devices
to use specific lock_class_key for sch->busylock
We could instead default l2tpeth tx_queue_len to 0 (no qdisc), but
a user might use a qdisc anyway.
(So same fixes are probably needed on non LLTX stacked drivers)
Noticed while stressing L2TPV3 setup :
======================================================
[ INFO: possible circular locking dependency detected ]
3.6.0-rc3+ #788 Not tainted
-------------------------------------------------------
netperf/4660 is trying to acquire lock:
(l2tpsock){+.-...}, at: [<ffffffffa0208db2>] l2tp_xmit_skb+0x172/0xa50 [l2tp_core]
but task is already holding lock:
(&(&sch->busylock)->rlock){+.-...}, at: [<ffffffff81596595>] dev_queue_xmit+0xd75/0xe00
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (&(&sch->busylock)->rlock){+.-...}:
[<ffffffff810a5df0>] lock_acquire+0x90/0x200
[<ffffffff817499fc>] _raw_spin_lock_irqsave+0x4c/0x60
[<ffffffff81074872>] __wake_up+0x32/0x70
[<ffffffff8136d39e>] tty_wakeup+0x3e/0x80
[<ffffffff81378fb3>] pty_write+0x73/0x80
[<ffffffff8136cb4c>] tty_put_char+0x3c/0x40
[<ffffffff813722b2>] process_echoes+0x142/0x330
[<ffffffff813742ab>] n_tty_receive_buf+0x8fb/0x1230
[<ffffffff813777b2>] flush_to_ldisc+0x142/0x1c0
[<ffffffff81062818>] process_one_work+0x198/0x760
[<ffffffff81063236>] worker_thread+0x186/0x4b0
[<ffffffff810694d3>] kthread+0x93/0xa0
[<ffffffff81753e24>] kernel_thread_helper+0x4/0x10
-> #0 (l2tpsock){+.-...}:
[<ffffffff810a5288>] __lock_acquire+0x1628/0x1b10
[<ffffffff810a5df0>] lock_acquire+0x90/0x200
[<ffffffff817498c1>] _raw_spin_lock+0x41/0x50
[<ffffffffa0208db2>] l2tp_xmit_skb+0x172/0xa50 [l2tp_core]
[<ffffffffa021a802>] l2tp_eth_dev_xmit+0x32/0x60 [l2tp_eth]
[<ffffffff815952b2>] dev_hard_start_xmit+0x502/0xa70
[<ffffffff815b63ce>] sch_direct_xmit+0xfe/0x290
[<ffffffff81595a05>] dev_queue_xmit+0x1e5/0xe00
[<ffffffff815d9d60>] ip_finish_output+0x3d0/0x890
[<ffffffff815db019>] ip_output+0x59/0xf0
[<ffffffff815da36d>] ip_local_out+0x2d/0xa0
[<ffffffff815da5a3>] ip_queue_xmit+0x1c3/0x680
[<ffffffff815f4192>] tcp_transmit_skb+0x402/0xa60
[<ffffffff815f4a94>] tcp_write_xmit+0x1f4/0xa30
[<ffffffff815f5300>] tcp_push_one+0x30/0x40
[<ffffffff815e6672>] tcp_sendmsg+0xe82/0x1040
[<ffffffff81614495>] inet_sendmsg+0x125/0x230
[<ffffffff81576cdc>] sock_sendmsg+0xdc/0xf0
[<ffffffff81579ece>] sys_sendto+0xfe/0x130
[<ffffffff81752c92>] system_call_fastpath+0x16/0x1b
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(&(&sch->busylock)->rlock);
lock(l2tpsock);
lock(&(&sch->busylock)->rlock);
lock(l2tpsock);
*** DEADLOCK ***
5 locks held by netperf/4660:
#0: (sk_lock-AF_INET){+.+.+.}, at: [<ffffffff815e581c>] tcp_sendmsg+0x2c/0x1040
#1: (rcu_read_lock){.+.+..}, at: [<ffffffff815da3e0>] ip_queue_xmit+0x0/0x680
#2: (rcu_read_lock_bh){.+....}, at: [<ffffffff815d9ac5>] ip_finish_output+0x135/0x890
#3: (rcu_read_lock_bh){.+....}, at: [<ffffffff81595820>] dev_queue_xmit+0x0/0xe00
#4: (&(&sch->busylock)->rlock){+.-...}, at: [<ffffffff81596595>] dev_queue_xmit+0xd75/0xe00
stack backtrace:
Pid: 4660, comm: netperf Not tainted 3.6.0-rc3+ #788
Call Trace:
[<ffffffff8173dbf8>] print_circular_bug+0x1fb/0x20c
[<ffffffff810a5288>] __lock_acquire+0x1628/0x1b10
[<ffffffff810a334b>] ? check_usage+0x9b/0x4d0
[<ffffffff810a3f44>] ? __lock_acquire+0x2e4/0x1b10
[<ffffffff810a5df0>] lock_acquire+0x90/0x200
[<ffffffffa0208db2>] ? l2tp_xmit_skb+0x172/0xa50 [l2tp_core]
[<ffffffff817498c1>] _raw_spin_lock+0x41/0x50
[<ffffffffa0208db2>] ? l2tp_xmit_skb+0x172/0xa50 [l2tp_core]
[<ffffffffa0208db2>] l2tp_xmit_skb+0x172/0xa50 [l2tp_core]
[<ffffffffa021a802>] l2tp_eth_dev_xmit+0x32/0x60 [l2tp_eth]
[<ffffffff815952b2>] dev_hard_start_xmit+0x502/0xa70
[<ffffffff81594e0e>] ? dev_hard_start_xmit+0x5e/0xa70
[<ffffffff81595961>] ? dev_queue_xmit+0x141/0xe00
[<ffffffff815b63ce>] sch_direct_xmit+0xfe/0x290
[<ffffffff81595a05>] dev_queue_xmit+0x1e5/0xe00
[<ffffffff81595820>] ? dev_hard_start_xmit+0xa70/0xa70
[<ffffffff815d9d60>] ip_finish_output+0x3d0/0x890
[<ffffffff815d9ac5>] ? ip_finish_output+0x135/0x890
[<ffffffff815db019>] ip_output+0x59/0xf0
[<ffffffff815da36d>] ip_local_out+0x2d/0xa0
[<ffffffff815da5a3>] ip_queue_xmit+0x1c3/0x680
[<ffffffff815da3e0>] ? ip_local_out+0xa0/0xa0
[<ffffffff815f4192>] tcp_transmit_skb+0x402/0xa60
[<ffffffff815fa25e>] ? tcp_md5_do_lookup+0x18e/0x1a0
[<ffffffff815f4a94>] tcp_write_xmit+0x1f4/0xa30
[<ffffffff815f5300>] tcp_push_one+0x30/0x40
[<ffffffff815e6672>] tcp_sendmsg+0xe82/0x1040
[<ffffffff81614495>] inet_sendmsg+0x125/0x230
[<ffffffff81614370>] ? inet_create+0x6b0/0x6b0
[<ffffffff8157e6e2>] ? sock_update_classid+0xc2/0x3b0
[<ffffffff8157e750>] ? sock_update_classid+0x130/0x3b0
[<ffffffff81576cdc>] sock_sendmsg+0xdc/0xf0
[<ffffffff81162579>] ? fget_light+0x3f9/0x4f0
[<ffffffff81579ece>] sys_sendto+0xfe/0x130
[<ffffffff810a69ad>] ? trace_hardirqs_on+0xd/0x10
[<ffffffff8174a0b0>] ? _raw_spin_unlock_irq+0x30/0x50
[<ffffffff810757e3>] ? finish_task_switch+0x83/0xf0
[<ffffffff810757a6>] ? finish_task_switch+0x46/0xf0
[<ffffffff81752cb7>] ? sysret_check+0x1b/0x56
[<ffffffff81752c92>] system_call_fastpath+0x16/0x1b
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
include/linux/netdevice.h | 2 ++
net/l2tp/l2tp_eth.c | 3 ++-
net/sched/sch_generic.c | 9 ++++++++-
3 files changed, 12 insertions(+), 2 deletions(-)
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index ccac82e..ae3153c0 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -1322,6 +1322,8 @@ struct net_device {
/* phy device may attach itself for hardware timestamping */
struct phy_device *phydev;
+ struct lock_class_key *qdisc_tx_busylock;
+
/* group the device belongs to */
int group;
diff --git a/net/l2tp/l2tp_eth.c b/net/l2tp/l2tp_eth.c
index f9ee74d..37b8b8b 100644
--- a/net/l2tp/l2tp_eth.c
+++ b/net/l2tp/l2tp_eth.c
@@ -67,6 +67,7 @@ static inline struct l2tp_eth_net *l2tp_eth_pernet(struct net *net)
return net_generic(net, l2tp_eth_net_id);
}
+static struct lock_class_key l2tp_eth_tx_busylock;
static int l2tp_eth_dev_init(struct net_device *dev)
{
struct l2tp_eth *priv = netdev_priv(dev);
@@ -74,7 +75,7 @@ static int l2tp_eth_dev_init(struct net_device *dev)
priv->dev = dev;
eth_hw_addr_random(dev);
memset(&dev->broadcast[0], 0xff, 6);
-
+ dev->qdisc_tx_busylock = &l2tp_eth_tx_busylock;
return 0;
}
diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
index 6c4d5fe..aefc150 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -527,6 +527,8 @@ struct Qdisc_ops pfifo_fast_ops __read_mostly = {
};
EXPORT_SYMBOL(pfifo_fast_ops);
+static struct lock_class_key qdisc_tx_busylock;
+
struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
struct Qdisc_ops *ops)
{
@@ -534,6 +536,7 @@ struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
struct Qdisc *sch;
unsigned int size = QDISC_ALIGN(sizeof(*sch)) + ops->priv_size;
int err = -ENOBUFS;
+ struct net_device *dev = dev_queue->dev;
p = kzalloc_node(size, GFP_KERNEL,
netdev_queue_numa_node_read(dev_queue));
@@ -553,12 +556,16 @@ struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
}
INIT_LIST_HEAD(&sch->list);
skb_queue_head_init(&sch->q);
+
spin_lock_init(&sch->busylock);
+ lockdep_set_class(&sch->busylock,
+ dev->qdisc_tx_busylock ?: &qdisc_tx_busylock);
+
sch->ops = ops;
sch->enqueue = ops->enqueue;
sch->dequeue = ops->dequeue;
sch->dev_queue = dev_queue;
- dev_hold(qdisc_dev(sch));
+ dev_hold(dev);
atomic_set(&sch->refcnt, 1);
return sch;
^ permalink raw reply related
* [PATCH net-next] ipv6: Export nd_tbl to allow modules to support IPv6
From: Thomas Graf @ 2012-09-05 11:14 UTC (permalink / raw)
To: davem; +Cc: netdev
Several out of the tree modules use the exported symbol arp_tbl
to read the ARP table. These modules now seek to support IPv6
but can't because nd_tbl is not exported.
Having access to the protocol specific neigh_table structure is
required in order to use the functions neigh_lookup() or
neigh_create() which are already exported to modules.
It does not make sense to export these functions if we don't
export the table itself as well.
Signed-off-by: Thomas Graf <tgraf@suug.ch>
---
net/ipv6/ndisc.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/net/ipv6/ndisc.c b/net/ipv6/ndisc.c
index ff36194..0a0eb3f 100644
--- a/net/ipv6/ndisc.c
+++ b/net/ipv6/ndisc.c
@@ -143,6 +143,8 @@ struct neigh_table nd_tbl = {
.gc_thresh3 = 1024,
};
+EXPORT_SYMBOL(nd_tbl);
+
static inline int ndisc_opt_addr_space(struct net_device *dev)
{
return NDISC_OPT_SPACE(dev->addr_len + ndisc_addr_option_pad(dev->type));
--
1.7.11.4
^ permalink raw reply related
* [PATCH V2 0/2] Remove duplicate register definitions in Chelsio cxgb4
From: Vipul Pandya @ 2012-09-05 12:01 UTC (permalink / raw)
To: linux-rdma-u79uwXL29TY76Z2rM5mHXA, netdev-u79uwXL29TY76Z2rM5mHXA
Cc: roland-BHEL68pLQRGGvPXPguhicg, davem-fT/PcQaiUtIeIZ0/mPfg9Q,
divy-ut6Up61K2wZBDgjK7y7TUQ, dm-ut6Up61K2wZBDgjK7y7TUQ,
swise-7bPotxP6k4+P2YhJcF5u+vpXobYPEAuW,
kumaras-ut6Up61K2wZBDgjK7y7TUQ, santosh-ut6Up61K2wZBDgjK7y7TUQ,
sivasu-ut6Up61K2wZBDgjK7y7TUQ, Vipul Pandya
Hi David Miller/Roland,
This patch series has minor changes in cxgb4 driver. It removes the duplicate
definitions of the registers in cxgb4 driver. It also has a minor update in
RDMA/cxgb4 driver due to change in cxgb4 driver.
This patch series is built against Stephen Rothwell's linux-next tree. We
request to merge patch series thorough single tree to avoid build failures.
Thanks,
Vipul Pandya
Vipul Pandya (2):
RDMA/cxgb4: Update RDMA/cxgb4 due to macro definition removal in
cxgb4 driver
cxgb4: Remove duplicate register definitions
drivers/infiniband/hw/cxgb4/qp.c | 2 +-
drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c | 18 ++++----
drivers/net/ethernet/chelsio/cxgb4/sge.c | 4 +-
drivers/net/ethernet/chelsio/cxgb4/t4_hw.c | 12 +++---
drivers/net/ethernet/chelsio/cxgb4/t4_regs.h | 54 +++++-----------------
5 files changed, 31 insertions(+), 59 deletions(-)
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
^ permalink raw reply
* [PATCH V2 1/2] RDMA/cxgb4: Update RDMA/cxgb4 due to macro definition removal in cxgb4 driver
From: Vipul Pandya @ 2012-09-05 12:01 UTC (permalink / raw)
To: linux-rdma, netdev
Cc: roland, davem, divy, dm, swise, kumaras, santosh, sivasu,
Vipul Pandya
In-Reply-To: <1346846515-25997-1-git-send-email-vipul@chelsio.com>
cxgb4 driver has duplicate definitions of registers which will be removed. This
patch updates the RDMA/cxgb4 driver accordingly.
Signed-off-by: Santosh Rastapur <santosh@chelsio.com>
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Reviewed-by: Sivakumar Subramani <sivasu@chelsio.com>
---
V2: Changed the order of the patch in patch series to avoid build failure
between two changes.
drivers/infiniband/hw/cxgb4/qp.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/infiniband/hw/cxgb4/qp.c b/drivers/infiniband/hw/cxgb4/qp.c
index 45aedf1..5213bab 100644
--- a/drivers/infiniband/hw/cxgb4/qp.c
+++ b/drivers/infiniband/hw/cxgb4/qp.c
@@ -1155,7 +1155,7 @@ static int ring_kernel_db(struct c4iw_qp *qhp, u32 qid, u16 inc)
*/
if (cxgb4_dbfifo_count(qhp->rhp->rdev.lldi.ports[0], 1) <
(qhp->rhp->rdev.lldi.dbfifo_int_thresh << 5)) {
- writel(V_QID(qid) | V_PIDX(inc), qhp->wq.db);
+ writel(QID(qid) | PIDX(inc), qhp->wq.db);
break;
}
set_current_state(TASK_UNINTERRUPTIBLE);
--
1.7.1
^ permalink raw reply related
* [PATCH V2 2/2] cxgb4: Remove duplicate register definitions
From: Vipul Pandya @ 2012-09-05 12:01 UTC (permalink / raw)
To: linux-rdma, netdev
Cc: roland, davem, divy, dm, swise, kumaras, santosh, sivasu,
Vipul Pandya
In-Reply-To: <1346846515-25997-1-git-send-email-vipul@chelsio.com>
Removed duplicate definition for SGE_PF_KDOORBELL, SGE_INT_ENABLE3,
PCIE_MEM_ACCESS_OFFSET registers.
Moved the register field definitions around the register definition.
Signed-off-by: Santosh Rastapur <santosh@chelsio.com>
Signed-off-by: Vipul Pandya <vipul@chelsio.com>
Reviewed-by: Sivakumar Subramani <sivasu@chelsio.com>
---
V2: Changed the order of the patch in patch series to avoid build failure
between two changes.
drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c | 18 ++++----
drivers/net/ethernet/chelsio/cxgb4/sge.c | 4 +-
drivers/net/ethernet/chelsio/cxgb4/t4_hw.c | 12 +++---
drivers/net/ethernet/chelsio/cxgb4/t4_regs.h | 54 +++++-----------------
4 files changed, 30 insertions(+), 58 deletions(-)
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
index 4a20821..5497eaa 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
@@ -2470,8 +2470,8 @@ int cxgb4_sync_txq_pidx(struct net_device *dev, u16 qid, u16 pidx,
else
delta = size - hw_pidx + pidx;
wmb();
- t4_write_reg(adap, MYPF_REG(A_SGE_PF_KDOORBELL),
- V_QID(qid) | V_PIDX(delta));
+ t4_write_reg(adap, MYPF_REG(SGE_PF_KDOORBELL),
+ QID(qid) | PIDX(delta));
}
out:
return ret;
@@ -2579,8 +2579,8 @@ static void sync_txq_pidx(struct adapter *adap, struct sge_txq *q)
else
delta = q->size - hw_pidx + q->db_pidx;
wmb();
- t4_write_reg(adap, MYPF_REG(A_SGE_PF_KDOORBELL),
- V_QID(q->cntxt_id) | V_PIDX(delta));
+ t4_write_reg(adap, MYPF_REG(SGE_PF_KDOORBELL),
+ QID(q->cntxt_id) | PIDX(delta));
}
out:
q->db_disabled = 0;
@@ -2617,9 +2617,9 @@ static void process_db_full(struct work_struct *work)
notify_rdma_uld(adap, CXGB4_CONTROL_DB_FULL);
drain_db_fifo(adap, dbfifo_drain_delay);
- t4_set_reg_field(adap, A_SGE_INT_ENABLE3,
- F_DBFIFO_HP_INT | F_DBFIFO_LP_INT,
- F_DBFIFO_HP_INT | F_DBFIFO_LP_INT);
+ t4_set_reg_field(adap, SGE_INT_ENABLE3,
+ DBFIFO_HP_INT | DBFIFO_LP_INT,
+ DBFIFO_HP_INT | DBFIFO_LP_INT);
notify_rdma_uld(adap, CXGB4_CONTROL_DB_EMPTY);
}
@@ -2639,8 +2639,8 @@ static void process_db_drop(struct work_struct *work)
void t4_db_full(struct adapter *adap)
{
- t4_set_reg_field(adap, A_SGE_INT_ENABLE3,
- F_DBFIFO_HP_INT | F_DBFIFO_LP_INT, 0);
+ t4_set_reg_field(adap, SGE_INT_ENABLE3,
+ DBFIFO_HP_INT | DBFIFO_LP_INT, 0);
queue_work(workq, &adap->db_full_task);
}
diff --git a/drivers/net/ethernet/chelsio/cxgb4/sge.c b/drivers/net/ethernet/chelsio/cxgb4/sge.c
index d49933e..1fde57d 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/sge.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/sge.c
@@ -769,8 +769,8 @@ static inline void ring_tx_db(struct adapter *adap, struct sge_txq *q, int n)
wmb(); /* write descriptors before telling HW */
spin_lock(&q->db_lock);
if (!q->db_disabled) {
- t4_write_reg(adap, MYPF_REG(A_SGE_PF_KDOORBELL),
- V_QID(q->cntxt_id) | V_PIDX(n));
+ t4_write_reg(adap, MYPF_REG(SGE_PF_KDOORBELL),
+ QID(q->cntxt_id) | PIDX(n));
}
q->db_pidx = q->pidx;
spin_unlock(&q->db_lock);
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
index af16013..dccecdc 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
@@ -1018,9 +1018,9 @@ static void sge_intr_handler(struct adapter *adapter)
{ ERR_INVALID_CIDX_INC,
"SGE GTS CIDX increment too large", -1, 0 },
{ ERR_CPL_OPCODE_0, "SGE received 0-length CPL", -1, 0 },
- { F_DBFIFO_LP_INT, NULL, -1, 0, t4_db_full },
- { F_DBFIFO_HP_INT, NULL, -1, 0, t4_db_full },
- { F_ERR_DROPPED_DB, NULL, -1, 0, t4_db_dropped },
+ { DBFIFO_LP_INT, NULL, -1, 0, t4_db_full },
+ { DBFIFO_HP_INT, NULL, -1, 0, t4_db_full },
+ { ERR_DROPPED_DB, NULL, -1, 0, t4_db_dropped },
{ ERR_DATA_CPL_ON_HIGH_QID1 | ERR_DATA_CPL_ON_HIGH_QID0,
"SGE IQID > 1023 received CPL for FL", -1, 0 },
{ ERR_BAD_DB_PIDX3, "SGE DBP 3 pidx increment too large", -1,
@@ -1520,7 +1520,7 @@ void t4_intr_enable(struct adapter *adapter)
ERR_BAD_DB_PIDX2 | ERR_BAD_DB_PIDX1 |
ERR_BAD_DB_PIDX0 | ERR_ING_CTXT_PRIO |
ERR_EGR_CTXT_PRIO | INGRESS_SIZE_ERR |
- F_DBFIFO_HP_INT | F_DBFIFO_LP_INT |
+ DBFIFO_HP_INT | DBFIFO_LP_INT |
EGRESS_SIZE_ERR);
t4_write_reg(adapter, MYPF_REG(PL_PF_INT_ENABLE), PF_INTR_MASK);
t4_set_reg_field(adapter, PL_INT_MAP0, 0, 1 << pf);
@@ -2033,8 +2033,8 @@ int t4_mem_win_read_len(struct adapter *adap, u32 addr, __be32 *data, int len)
if ((addr & 3) || (len + off) > MEMWIN0_APERTURE)
return -EINVAL;
- t4_write_reg(adap, A_PCIE_MEM_ACCESS_OFFSET, addr & ~15);
- t4_read_reg(adap, A_PCIE_MEM_ACCESS_OFFSET);
+ t4_write_reg(adap, PCIE_MEM_ACCESS_OFFSET, addr & ~15);
+ t4_read_reg(adap, PCIE_MEM_ACCESS_OFFSET);
for (i = 0; i < len; i += 4)
*data++ = t4_read_reg(adap, (MEMWIN0_BASE + off + i));
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h b/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
index 111fc32..8e814bc 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_regs.h
@@ -190,58 +190,30 @@
#define SGE_DEBUG_DATA_LOW 0x10d4
#define SGE_INGRESS_QUEUES_PER_PAGE_PF 0x10f4
-#define S_LP_INT_THRESH 12
-#define V_LP_INT_THRESH(x) ((x) << S_LP_INT_THRESH)
#define S_HP_INT_THRESH 28
+#define M_HP_INT_THRESH 0xfU
#define V_HP_INT_THRESH(x) ((x) << S_HP_INT_THRESH)
+#define M_HP_COUNT 0x7ffU
+#define S_HP_COUNT 16
+#define G_HP_COUNT(x) (((x) >> S_HP_COUNT) & M_HP_COUNT)
+#define S_LP_INT_THRESH 12
+#define M_LP_INT_THRESH 0xfU
+#define V_LP_INT_THRESH(x) ((x) << S_LP_INT_THRESH)
+#define M_LP_COUNT 0x7ffU
+#define S_LP_COUNT 0
+#define G_LP_COUNT(x) (((x) >> S_LP_COUNT) & M_LP_COUNT)
#define A_SGE_DBFIFO_STATUS 0x10a4
#define S_ENABLE_DROP 13
#define V_ENABLE_DROP(x) ((x) << S_ENABLE_DROP)
#define F_ENABLE_DROP V_ENABLE_DROP(1U)
-#define A_SGE_DOORBELL_CONTROL 0x10a8
-
-#define A_SGE_CTXT_CMD 0x11fc
-#define A_SGE_DBQ_CTXT_BADDR 0x1084
-
-#define A_SGE_PF_KDOORBELL 0x0
-
-#define S_QID 15
-#define V_QID(x) ((x) << S_QID)
-
-#define S_PIDX 0
-#define V_PIDX(x) ((x) << S_PIDX)
-
-#define M_LP_COUNT 0x7ffU
-#define S_LP_COUNT 0
-#define G_LP_COUNT(x) (((x) >> S_LP_COUNT) & M_LP_COUNT)
-
-#define M_HP_COUNT 0x7ffU
-#define S_HP_COUNT 16
-#define G_HP_COUNT(x) (((x) >> S_HP_COUNT) & M_HP_COUNT)
-
-#define A_SGE_INT_ENABLE3 0x1040
-
-#define S_DBFIFO_HP_INT 8
-#define V_DBFIFO_HP_INT(x) ((x) << S_DBFIFO_HP_INT)
-#define F_DBFIFO_HP_INT V_DBFIFO_HP_INT(1U)
-
-#define S_DBFIFO_LP_INT 7
-#define V_DBFIFO_LP_INT(x) ((x) << S_DBFIFO_LP_INT)
-#define F_DBFIFO_LP_INT V_DBFIFO_LP_INT(1U)
-
#define S_DROPPED_DB 0
#define V_DROPPED_DB(x) ((x) << S_DROPPED_DB)
#define F_DROPPED_DB V_DROPPED_DB(1U)
+#define A_SGE_DOORBELL_CONTROL 0x10a8
-#define S_ERR_DROPPED_DB 18
-#define V_ERR_DROPPED_DB(x) ((x) << S_ERR_DROPPED_DB)
-#define F_ERR_DROPPED_DB V_ERR_DROPPED_DB(1U)
-
-#define A_PCIE_MEM_ACCESS_OFFSET 0x306c
-
-#define M_HP_INT_THRESH 0xfU
-#define M_LP_INT_THRESH 0xfU
+#define A_SGE_CTXT_CMD 0x11fc
+#define A_SGE_DBQ_CTXT_BADDR 0x1084
#define PCIE_PF_CLI 0x44
#define PCIE_INT_CAUSE 0x3004
--
1.7.1
^ permalink raw reply related
* [net-next PATCH] be2net: use PCIe AER capability
From: Sathya Perla @ 2012-09-05 11:56 UTC (permalink / raw)
To: netdev; +Cc: Sathya Perla
This patch allows code to handle the PCIe AER capability.
The PCI callbacks for error handling/reset/recovery already exist in be2net
and have been tested with EEH/ppc.
This patch has been tested using the aer-inject tool.
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
---
drivers/net/ethernet/emulex/benet/be_main.c | 8 ++++++++
1 files changed, 8 insertions(+), 0 deletions(-)
diff --git a/drivers/net/ethernet/emulex/benet/be_main.c b/drivers/net/ethernet/emulex/benet/be_main.c
index 6d139d6..cc2947d 100644
--- a/drivers/net/ethernet/emulex/benet/be_main.c
+++ b/drivers/net/ethernet/emulex/benet/be_main.c
@@ -20,6 +20,7 @@
#include "be.h"
#include "be_cmds.h"
#include <asm/div64.h>
+#include <linux/aer.h>
MODULE_VERSION(DRV_VER);
MODULE_DEVICE_TABLE(pci, be_dev_ids);
@@ -3551,6 +3552,8 @@ static void __devexit be_remove(struct pci_dev *pdev)
be_ctrl_cleanup(adapter);
+ pci_disable_pcie_error_reporting(pdev);
+
pci_set_drvdata(pdev, NULL);
pci_release_regions(pdev);
pci_disable_device(pdev);
@@ -3844,6 +3847,10 @@ static int __devinit be_probe(struct pci_dev *pdev,
}
}
+ status = pci_enable_pcie_error_reporting(pdev);
+ if (status)
+ dev_err(&pdev->dev, "Could not use PCIe error reporting\n");
+
status = be_ctrl_init(adapter);
if (status)
goto free_netdev;
@@ -4066,6 +4073,7 @@ static pci_ers_result_t be_eeh_reset(struct pci_dev *pdev)
if (status)
return PCI_ERS_RESULT_DISCONNECT;
+ pci_cleanup_aer_uncorrect_error_status(pdev);
return PCI_ERS_RESULT_RECOVERED;
}
--
1.7.4
^ permalink raw reply related
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox