* Re: [PATCH 10/14] drivers/isdn: delete double assignment
From: walter harms @ 2010-10-26 11:55 UTC (permalink / raw)
To: Julia Lawall; +Cc: Karsten Keil, kernel-janitors, netdev, linux-kernel
In-Reply-To: <1288088743-3725-11-git-send-email-julia@diku.dk>
Julia Lawall schrieb:
> From: Julia Lawall <julia@diku.dk>
>
> Delete successive assignments to the same location. In the first case, the
> hscx array has two elements, so change the assignment to initialize the
> second one. In the second case, the two assignments are simply identical.
>
> A simplified version of the semantic match that finds this problem is as
> follows: (http://coccinelle.lip6.fr/)
>
> // <smpl>
> @@
> expression i;
> @@
>
> *i = ...;
> i = ...;
> // </smpl>
>
> Signed-off-by: Julia Lawall <julia@diku.dk>
>
> ---
> In the first case, the patch changes the semantics and has not been tested.
>
> drivers/isdn/hardware/mISDN/mISDNinfineon.c | 2 +-
> drivers/isdn/hisax/l3_1tr6.c | 2 --
> 2 files changed, 1 insertion(+), 3 deletions(-)
>
> diff --git a/drivers/isdn/hardware/mISDN/mISDNinfineon.c b/drivers/isdn/hardware/mISDN/mISDNinfineon.c
> index af25e1f..e90db88 100644
> --- a/drivers/isdn/hardware/mISDN/mISDNinfineon.c
> +++ b/drivers/isdn/hardware/mISDN/mISDNinfineon.c
> @@ -563,7 +563,7 @@ reset_inf(struct inf_hw *hw)
> mdelay(10);
> hw->ipac.isac.adf2 = 0x87;
> hw->ipac.hscx[0].slot = 0x1f;
> - hw->ipac.hscx[0].slot = 0x23;
> + hw->ipac.hscx[1].slot = 0x23;
> break;
> case INF_GAZEL_R753:
> val = inl((u32)hw->cfg.start + GAZEL_CNTRL);
> diff --git a/drivers/isdn/hisax/l3_1tr6.c b/drivers/isdn/hisax/l3_1tr6.c
> index b0554f8..a5c76fc 100644
> --- a/drivers/isdn/hisax/l3_1tr6.c
> +++ b/drivers/isdn/hisax/l3_1tr6.c
> @@ -164,8 +164,6 @@ l3_1tr6_setup(struct l3_process *pc, u_char pr, void *arg)
> char tmp[80];
> struct sk_buff *skb = arg;
>
> - p = skb->data;
> -
> /* Channel Identification */
> p = skb->data;
> if ((p = findie(p, skb->len, WE0_chanID, 0))) {
>
perhaps you can move the next assignment out of if also ?
p = findie(skb->data, skb->len, WE0_chanID, 0);
if (p) { ....
re,
wh
> --
> To unsubscribe from this list: send the line "unsubscribe kernel-janitors" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
>
^ permalink raw reply
* Re: [PATCH 7/14] drivers/net/typhoon.c: delete double assignment
From: David Dillow @ 2010-10-26 11:52 UTC (permalink / raw)
To: Julia Lawall; +Cc: kernel-janitors, netdev, linux-kernel
In-Reply-To: <1288088743-3725-8-git-send-email-julia@diku.dk>
On Tue, 2010-10-26 at 12:25 +0200, Julia Lawall wrote:
> From: Julia Lawall <julia@diku.dk>
>
> Delete successive assignments to the same location. The current definition
> does not initialize the respRing structure, which has the same type as the
> cmdRing structure, so initialize that one instead.
> Signed-off-by: Julia Lawall <julia@diku.dk>
Acked-by: David Dillow <dave@thedillows.org>
> ---
> This changes the semantics and has not been tested.
tp->respRing.lastWrite is only an artifact of using a common struct for
the rings and is not otherwise used, so no change to semantics.
^ permalink raw reply
* Reply Needed
From: Dr Carl Lee @ 2010-10-26 13:21 UTC (permalink / raw)
To: netdev
Hello,
I have a proposition for you, this however is not mandatory nor will I
in any manner compel you to honor against your will.Let me start by
introducing myself. I am Dr.Carl Lee, Director of Operations
of the Hang Seng Bank Ltd,Sai Wan Ho Branch. I have a mutual beneficial
business suggestion for you.
1. Can you handle this project?
2. Can I give you this trust ?
Absolute confidentiality is required from you.Besides,I will use my connection
to get some documents to back up the fund so that the fund can not be
question by any authority.
More information await you in my next response to your email message.
Treat as very urgent.
Yours Faithfully,
Dr. Carl Lee.
^ permalink raw reply
* Re: [v3 RFC PATCH 0/4] Implement multiqueue virtio-net
From: Michael S. Tsirkin @ 2010-10-26 11:09 UTC (permalink / raw)
To: Krishna Kumar2
Cc: anthony, arnd, avi, davem, eric.dumazet, kvm, netdev,
netdev-owner, rusty
In-Reply-To: <OFBE9E2C99.648EFB19-ON652577C8.0035FEDD-652577C8.0036AD51@in.ibm.com>
On Tue, Oct 26, 2010 at 03:31:39PM +0530, Krishna Kumar2 wrote:
> > "Michael S. Tsirkin" <mst@redhat.com>
> >
> > On Tue, Oct 26, 2010 at 02:38:53PM +0530, Krishna Kumar2 wrote:
> > > Results for UDP BW tests (unidirectional, sum across
> > > 3 iterations, each iteration of 45 seconds, default
> > > netperf, vhosts bound to cpus 0-3; no other tuning):
> >
> > Is binding vhost threads to CPUs really required?
> > What happens if we let the scheduler do its job?
>
> Nothing drastic, I remember BW% and SD% both improved a
> bit as a result of binding.
If there's a significant improvement this would mean that
we need to rethink the vhost-net interaction with the scheduler.
> I started binding vhost thread
> after Avi suggested it in response to my v1 patch (he
> suggested some more that I haven't done), and have been
> doing only this tuning ever since. This is part of his
> mail for the tuning:
>
> > vhost:
> > thread #0: CPU0
> > thread #1: CPU1
> > thread #2: CPU2
> > thread #3: CPU3
>
> I simply bound each thread to CPU0-3 instead.
>
> Thanks,
>
> - KK
^ permalink raw reply
* Re: [PATCH 3/14] drivers/vhost/vhost.c: delete double assignment
From: Michael S. Tsirkin @ 2010-10-26 11:06 UTC (permalink / raw)
To: Julia Lawall; +Cc: virtualization, kernel-janitors, linux-kernel, kvm, netdev
In-Reply-To: <1288088743-3725-4-git-send-email-julia@diku.dk>
On Tue, Oct 26, 2010 at 12:25:32PM +0200, Julia Lawall wrote:
> From: Julia Lawall <julia@diku.dk>
>
> Delete successive assignments to the same location.
>
> A simplified version of the semantic match that finds this problem is as
> follows: (http://coccinelle.lip6.fr/)
>
> // <smpl>
> @@
> expression i;
> @@
>
> *i = ...;
> i = ...;
> // </smpl>
>
> Signed-off-by: Julia Lawall <julia@diku.dk>
Thanks, applied.
> ---
> drivers/vhost/vhost.c | 1 -
> 1 file changed, 1 deletion(-)
>
> diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
> index 94701ff..ed27727 100644
> --- a/drivers/vhost/vhost.c
> +++ b/drivers/vhost/vhost.c
> @@ -157,7 +157,6 @@ static void vhost_vq_reset(struct vhost_dev *dev,
> vq->avail_idx = 0;
> vq->last_used_idx = 0;
> vq->used_flags = 0;
> - vq->used_flags = 0;
> vq->log_used = false;
> vq->log_addr = -1ull;
> vq->vhost_hlen = 0;
^ permalink raw reply
* RE: [PATCH] qlcnic: dma address align check
From: Amit Salecha @ 2010-10-26 10:52 UTC (permalink / raw)
To: Eric Dumazet
Cc: davem@davemloft.net, netdev@vger.kernel.org, Ameen Rahman,
Anirban Chakraborty
In-Reply-To: <1288089033.3169.73.camel@edumazet-laptop>
> > Is pci_alloc_consistent guarantee to give PAGE align dma address ?
>
> I believe so
>
> By the way, you should use dma_alloc_coherent() instead of
> pci_alloc_consistent() so that you can use GFP_KERNEL allocations
> instead of GFP_ATOMIC, it might help in low memory conditions (if you
> dont hold a spinlock at this point)
>
>
> If TX_DESC_RINGSIZE(tx_ring) is not a power of two, then yes you could
> probably add 64 bytes and avoid allocating a full page only for the u32
> field ;)
>
>
TX_DESC_RINGSIZE is power of two.
If pci_alloc_consistent guarantee to give PAGE align dma address, then this patch is not required.
>
^ permalink raw reply
* RE: [PATCH] qlcnic: dma address align check
From: Eric Dumazet @ 2010-10-26 10:30 UTC (permalink / raw)
To: Amit Salecha
Cc: davem@davemloft.net, netdev@vger.kernel.org, Ameen Rahman,
Anirban Chakraborty
In-Reply-To: <99737F4847ED0A48AECC9F4A1974A4B80F871A817A@MNEXMB2.qlogic.org>
Le mardi 26 octobre 2010 à 05:04 -0500, Amit Salecha a écrit :
> Is it ? I am not aware about such calculation.
Yes it is
> Is pci_alloc_consistent guarantee to give PAGE align dma address ?
I believe so
By the way, you should use dma_alloc_coherent() instead of
pci_alloc_consistent() so that you can use GFP_KERNEL allocations
instead of GFP_ATOMIC, it might help in low memory conditions (if you
dont hold a spinlock at this point)
If TX_DESC_RINGSIZE(tx_ring) is not a power of two, then yes you could
probably add 64 bytes and avoid allocating a full page only for the u32
field ;)
^ permalink raw reply
* [PATCH 10/14] drivers/isdn: delete double assignment
From: Julia Lawall @ 2010-10-26 10:25 UTC (permalink / raw)
To: Karsten Keil; +Cc: kernel-janitors, netdev, linux-kernel
In-Reply-To: <1288088743-3725-1-git-send-email-julia@diku.dk>
From: Julia Lawall <julia@diku.dk>
Delete successive assignments to the same location. In the first case, the
hscx array has two elements, so change the assignment to initialize the
second one. In the second case, the two assignments are simply identical.
A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
@@
expression i;
@@
*i = ...;
i = ...;
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
---
In the first case, the patch changes the semantics and has not been tested.
drivers/isdn/hardware/mISDN/mISDNinfineon.c | 2 +-
drivers/isdn/hisax/l3_1tr6.c | 2 --
2 files changed, 1 insertion(+), 3 deletions(-)
diff --git a/drivers/isdn/hardware/mISDN/mISDNinfineon.c b/drivers/isdn/hardware/mISDN/mISDNinfineon.c
index af25e1f..e90db88 100644
--- a/drivers/isdn/hardware/mISDN/mISDNinfineon.c
+++ b/drivers/isdn/hardware/mISDN/mISDNinfineon.c
@@ -563,7 +563,7 @@ reset_inf(struct inf_hw *hw)
mdelay(10);
hw->ipac.isac.adf2 = 0x87;
hw->ipac.hscx[0].slot = 0x1f;
- hw->ipac.hscx[0].slot = 0x23;
+ hw->ipac.hscx[1].slot = 0x23;
break;
case INF_GAZEL_R753:
val = inl((u32)hw->cfg.start + GAZEL_CNTRL);
diff --git a/drivers/isdn/hisax/l3_1tr6.c b/drivers/isdn/hisax/l3_1tr6.c
index b0554f8..a5c76fc 100644
--- a/drivers/isdn/hisax/l3_1tr6.c
+++ b/drivers/isdn/hisax/l3_1tr6.c
@@ -164,8 +164,6 @@ l3_1tr6_setup(struct l3_process *pc, u_char pr, void *arg)
char tmp[80];
struct sk_buff *skb = arg;
- p = skb->data;
-
/* Channel Identification */
p = skb->data;
if ((p = findie(p, skb->len, WE0_chanID, 0))) {
^ permalink raw reply related
* [PATCH 7/14] drivers/net/typhoon.c: delete double assignment
From: Julia Lawall @ 2010-10-26 10:25 UTC (permalink / raw)
To: David Dillow; +Cc: kernel-janitors, netdev, linux-kernel
In-Reply-To: <1288088743-3725-1-git-send-email-julia@diku.dk>
From: Julia Lawall <julia@diku.dk>
Delete successive assignments to the same location. The current definition
does not initialize the respRing structure, which has the same type as the
cmdRing structure, so initialize that one instead.
A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
@@
expression i;
@@
*i = ...;
i = ...;
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
---
This changes the semantics and has not been tested.
drivers/net/typhoon.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/typhoon.c b/drivers/net/typhoon.c
index 1cc6713..fc014eb 100644
--- a/drivers/net/typhoon.c
+++ b/drivers/net/typhoon.c
@@ -1328,7 +1328,7 @@ typhoon_init_rings(struct typhoon *tp)
tp->rxHiRing.lastWrite = 0;
tp->rxBuffRing.lastWrite = 0;
tp->cmdRing.lastWrite = 0;
- tp->cmdRing.lastWrite = 0;
+ tp->respRing.lastWrite = 0;
tp->txLoRing.lastRead = 0;
tp->txHiRing.lastRead = 0;
^ permalink raw reply related
* [PATCH 5/14] drivers/net/sb1000.c: delete double assignment
From: Julia Lawall @ 2010-10-26 10:25 UTC (permalink / raw)
To: netdev; +Cc: kernel-janitors, linux-kernel
In-Reply-To: <1288088743-3725-1-git-send-email-julia@diku.dk>
From: Julia Lawall <julia@diku.dk>
The other code around these duplicated assignments initializes the 0 1 2
and 3 elements of an array, so change the initialization of the
rx_session_id array to do the same.
A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
@@
expression i;
@@
*i = ...;
i = ...;
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
---
This changes the semantics and has not been tested.
drivers/net/sb1000.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/net/sb1000.c b/drivers/net/sb1000.c
index a9ae505..66c2f1a 100644
--- a/drivers/net/sb1000.c
+++ b/drivers/net/sb1000.c
@@ -961,9 +961,9 @@ sb1000_open(struct net_device *dev)
lp->rx_error_count = 0;
lp->rx_error_dpc_count = 0;
lp->rx_session_id[0] = 0x50;
- lp->rx_session_id[0] = 0x48;
- lp->rx_session_id[0] = 0x44;
- lp->rx_session_id[0] = 0x42;
+ lp->rx_session_id[1] = 0x48;
+ lp->rx_session_id[2] = 0x44;
+ lp->rx_session_id[3] = 0x42;
lp->rx_frame_id[0] = 0;
lp->rx_frame_id[1] = 0;
lp->rx_frame_id[2] = 0;
^ permalink raw reply related
* [PATCH 3/14] drivers/vhost/vhost.c: delete double assignment
From: Julia Lawall @ 2010-10-26 10:25 UTC (permalink / raw)
To: Michael S. Tsirkin
Cc: virtualization, kernel-janitors, linux-kernel, kvm, netdev
In-Reply-To: <1288088743-3725-1-git-send-email-julia@diku.dk>
From: Julia Lawall <julia@diku.dk>
Delete successive assignments to the same location.
A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
@@
expression i;
@@
*i = ...;
i = ...;
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
---
drivers/vhost/vhost.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index 94701ff..ed27727 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -157,7 +157,6 @@ static void vhost_vq_reset(struct vhost_dev *dev,
vq->avail_idx = 0;
vq->last_used_idx = 0;
vq->used_flags = 0;
- vq->used_flags = 0;
vq->log_used = false;
vq->log_addr = -1ull;
vq->vhost_hlen = 0;
^ permalink raw reply related
* RE: [PATCH] qlcnic: dma address align check
From: Amit Salecha @ 2010-10-26 10:04 UTC (permalink / raw)
To: Eric Dumazet
Cc: davem@davemloft.net, netdev@vger.kernel.org, Ameen Rahman,
Anirban Chakraborty
In-Reply-To: <1288086894.3169.53.camel@edumazet-laptop>
> -----Original Message-----
> From: Eric Dumazet [mailto:eric.dumazet@gmail.com]
> Sent: Tuesday, October 26, 2010 3:25 PM
> To: Amit Salecha
> Cc: davem@davemloft.net; netdev@vger.kernel.org; Ameen Rahman; Anirban
> Chakraborty
> Subject: Re: [PATCH] qlcnic: dma address align check
>
> Le mardi 26 octobre 2010 à 02:38 -0700, Amit Kumar Salecha a écrit :
> > Device requires tx_hw_cosnumer to be 64 byte aligned.
> > Tx desc size is 64 byte, alloc tx_hw_consumer with tx desc.
> >
> > Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
> > ---
> > drivers/net/qlcnic/qlcnic_ctx.c | 35 +++++++++++++++--------------
> ------
> > 1 files changed, 15 insertions(+), 20 deletions(-)
> >
> > diff --git a/drivers/net/qlcnic/qlcnic_ctx.c
> b/drivers/net/qlcnic/qlcnic_ctx.c
> > index 1cdc05d..21c9c28 100644
> > --- a/drivers/net/qlcnic/qlcnic_ctx.c
> > +++ b/drivers/net/qlcnic/qlcnic_ctx.c
> > @@ -418,18 +418,9 @@ int qlcnic_alloc_hw_resources(struct
> qlcnic_adapter *adapter)
> > recv_ctx = &adapter->recv_ctx;
> > tx_ring = adapter->tx_ring;
> >
> > - tx_ring->hw_consumer = (__le32 *)pci_alloc_consistent(pdev,
> sizeof(u32),
> > - &tx_ring->hw_cons_phys_addr);
> > - if (tx_ring->hw_consumer == NULL) {
> > - dev_err(&pdev->dev, "failed to allocate tx consumer\n");
> > - return -ENOMEM;
> > - }
> > - *(tx_ring->hw_consumer) = 0;
> > -
> > /* cmd desc ring */
> > - addr = pci_alloc_consistent(pdev, TX_DESC_RINGSIZE(tx_ring),
> > - &tx_ring->phys_addr);
> > -
> > + addr = pci_alloc_consistent(pdev, TX_DESC_RINGSIZE(tx_ring) +
> > + sizeof(u32), &tx_ring->phys_addr);
>
> Wont this use twice memory than before, due to power-of-two
> allocations ?
>
> Allocating 65536 + 4 bytes gives you 131072 bytes.
>
Is it ? I am not aware about such calculation.
Is pci_alloc_consistent guarantee to give PAGE align dma address ?
-Amit
^ permalink raw reply
* Re: [v3 RFC PATCH 0/4] Implement multiqueue virtio-net
From: Krishna Kumar2 @ 2010-10-26 10:01 UTC (permalink / raw)
To: Michael S. Tsirkin
Cc: anthony, arnd, avi, davem, eric.dumazet, kvm, netdev,
netdev-owner, rusty
In-Reply-To: <20101026093846.GA6766@redhat.com>
> "Michael S. Tsirkin" <mst@redhat.com>
>
> On Tue, Oct 26, 2010 at 02:38:53PM +0530, Krishna Kumar2 wrote:
> > Results for UDP BW tests (unidirectional, sum across
> > 3 iterations, each iteration of 45 seconds, default
> > netperf, vhosts bound to cpus 0-3; no other tuning):
>
> Is binding vhost threads to CPUs really required?
> What happens if we let the scheduler do its job?
Nothing drastic, I remember BW% and SD% both improved a
bit as a result of binding. I started binding vhost thread
after Avi suggested it in response to my v1 patch (he
suggested some more that I haven't done), and have been
doing only this tuning ever since. This is part of his
mail for the tuning:
> vhost:
> thread #0: CPU0
> thread #1: CPU1
> thread #2: CPU2
> thread #3: CPU3
I simply bound each thread to CPU0-3 instead.
Thanks,
- KK
^ permalink raw reply
* [PATCH] inetpeer: __rcu annotations
From: Eric Dumazet @ 2010-10-26 9:55 UTC (permalink / raw)
To: David Miller; +Cc: netdev
Adds __rcu annotations to inetpeer
(struct inet_peer)->avl_left
(struct inet_peer)->avl_right
This is a tedious cleanup, but removes one smp_wmb() from link_to_pool()
since we now use more self documenting rcu_assign_pointer().
Note the use of RCU_INIT_POINTER() instead of rcu_assign_pointer() in
all cases we dont need a memory barrier.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
---
include/net/inetpeer.h | 2
net/ipv4/inetpeer.c | 138 ++++++++++++++++++++++-----------------
2 files changed, 81 insertions(+), 59 deletions(-)
diff --git a/include/net/inetpeer.h b/include/net/inetpeer.h
index 417d0c8..fe239bf 100644
--- a/include/net/inetpeer.h
+++ b/include/net/inetpeer.h
@@ -15,7 +15,7 @@
struct inet_peer {
/* group together avl_left,avl_right,v4daddr to speedup lookups */
- struct inet_peer *avl_left, *avl_right;
+ struct inet_peer __rcu *avl_left, *avl_right;
__be32 v4daddr; /* peer's address */
__u32 avl_height;
struct list_head unused;
diff --git a/net/ipv4/inetpeer.c b/net/ipv4/inetpeer.c
index 9ffa24b..ccf01c9 100644
--- a/net/ipv4/inetpeer.c
+++ b/net/ipv4/inetpeer.c
@@ -72,18 +72,19 @@ static struct kmem_cache *peer_cachep __read_mostly;
#define node_height(x) x->avl_height
#define peer_avl_empty ((struct inet_peer *)&peer_fake_node)
+#define peer_avl_empty_rcu ((struct inet_peer __rcu __force *)&peer_fake_node)
static const struct inet_peer peer_fake_node = {
- .avl_left = peer_avl_empty,
- .avl_right = peer_avl_empty,
+ .avl_left = peer_avl_empty_rcu,
+ .avl_right = peer_avl_empty_rcu,
.avl_height = 0
};
static struct {
- struct inet_peer *root;
+ struct inet_peer __rcu *root;
spinlock_t lock;
int total;
} peers = {
- .root = peer_avl_empty,
+ .root = peer_avl_empty_rcu,
.lock = __SPIN_LOCK_UNLOCKED(peers.lock),
.total = 0,
};
@@ -156,11 +157,14 @@ static void unlink_from_unused(struct inet_peer *p)
*/
#define lookup(_daddr, _stack) \
({ \
- struct inet_peer *u, **v; \
+ struct inet_peer *u; \
+ struct inet_peer __rcu **v; \
\
stackptr = _stack; \
*stackptr++ = &peers.root; \
- for (u = peers.root; u != peer_avl_empty; ) { \
+ for (u = rcu_dereference_protected(peers.root, \
+ lockdep_is_held(&peers.lock)); \
+ u != peer_avl_empty; ) { \
if (_daddr == u->v4daddr) \
break; \
if ((__force __u32)_daddr < (__force __u32)u->v4daddr) \
@@ -168,7 +172,8 @@ static void unlink_from_unused(struct inet_peer *p)
else \
v = &u->avl_right; \
*stackptr++ = v; \
- u = *v; \
+ u = rcu_dereference_protected(*v, \
+ lockdep_is_held(&peers.lock)); \
} \
u; \
})
@@ -209,13 +214,17 @@ static struct inet_peer *lookup_rcu_bh(__be32 daddr)
/* Called with local BH disabled and the pool lock held. */
#define lookup_rightempty(start) \
({ \
- struct inet_peer *u, **v; \
+ struct inet_peer *u; \
+ struct inet_peer __rcu **v; \
*stackptr++ = &start->avl_left; \
v = &start->avl_left; \
- for (u = *v; u->avl_right != peer_avl_empty; ) { \
+ for (u = rcu_dereference_protected(*v, \
+ lockdep_is_held(&peers.lock)); \
+ u->avl_right != peer_avl_empty_rcu; ) { \
v = &u->avl_right; \
*stackptr++ = v; \
- u = *v; \
+ u = rcu_dereference_protected(*v, \
+ lockdep_is_held(&peers.lock)); \
} \
u; \
})
@@ -224,74 +233,86 @@ static struct inet_peer *lookup_rcu_bh(__be32 daddr)
* Variable names are the proof of operation correctness.
* Look into mm/map_avl.c for more detail description of the ideas.
*/
-static void peer_avl_rebalance(struct inet_peer **stack[],
- struct inet_peer ***stackend)
+static void peer_avl_rebalance(struct inet_peer __rcu **stack[],
+ struct inet_peer __rcu ***stackend)
{
- struct inet_peer **nodep, *node, *l, *r;
+ struct inet_peer __rcu **nodep;
+ struct inet_peer *node, *l, *r;
int lh, rh;
while (stackend > stack) {
nodep = *--stackend;
- node = *nodep;
- l = node->avl_left;
- r = node->avl_right;
+ node = rcu_dereference_protected(*nodep,
+ lockdep_is_held(&peers.lock));
+ l = rcu_dereference_protected(node->avl_left,
+ lockdep_is_held(&peers.lock));
+ r = rcu_dereference_protected(node->avl_right,
+ lockdep_is_held(&peers.lock));
lh = node_height(l);
rh = node_height(r);
if (lh > rh + 1) { /* l: RH+2 */
struct inet_peer *ll, *lr, *lrl, *lrr;
int lrh;
- ll = l->avl_left;
- lr = l->avl_right;
+ ll = rcu_dereference_protected(l->avl_left,
+ lockdep_is_held(&peers.lock));
+ lr = rcu_dereference_protected(l->avl_right,
+ lockdep_is_held(&peers.lock));
lrh = node_height(lr);
if (lrh <= node_height(ll)) { /* ll: RH+1 */
- node->avl_left = lr; /* lr: RH or RH+1 */
- node->avl_right = r; /* r: RH */
+ RCU_INIT_POINTER(node->avl_left, lr); /* lr: RH or RH+1 */
+ RCU_INIT_POINTER(node->avl_right, r); /* r: RH */
node->avl_height = lrh + 1; /* RH+1 or RH+2 */
- l->avl_left = ll; /* ll: RH+1 */
- l->avl_right = node; /* node: RH+1 or RH+2 */
+ RCU_INIT_POINTER(l->avl_left, ll); /* ll: RH+1 */
+ RCU_INIT_POINTER(l->avl_right, node); /* node: RH+1 or RH+2 */
l->avl_height = node->avl_height + 1;
- *nodep = l;
+ RCU_INIT_POINTER(*nodep, l);
} else { /* ll: RH, lr: RH+1 */
- lrl = lr->avl_left; /* lrl: RH or RH-1 */
- lrr = lr->avl_right; /* lrr: RH or RH-1 */
- node->avl_left = lrr; /* lrr: RH or RH-1 */
- node->avl_right = r; /* r: RH */
+ lrl = rcu_dereference_protected(lr->avl_left,
+ lockdep_is_held(&peers.lock)); /* lrl: RH or RH-1 */
+ lrr = rcu_dereference_protected(lr->avl_right,
+ lockdep_is_held(&peers.lock)); /* lrr: RH or RH-1 */
+ RCU_INIT_POINTER(node->avl_left, lrr); /* lrr: RH or RH-1 */
+ RCU_INIT_POINTER(node->avl_right, r); /* r: RH */
node->avl_height = rh + 1; /* node: RH+1 */
- l->avl_left = ll; /* ll: RH */
- l->avl_right = lrl; /* lrl: RH or RH-1 */
+ RCU_INIT_POINTER(l->avl_left, ll); /* ll: RH */
+ RCU_INIT_POINTER(l->avl_right, lrl); /* lrl: RH or RH-1 */
l->avl_height = rh + 1; /* l: RH+1 */
- lr->avl_left = l; /* l: RH+1 */
- lr->avl_right = node; /* node: RH+1 */
+ RCU_INIT_POINTER(lr->avl_left, l); /* l: RH+1 */
+ RCU_INIT_POINTER(lr->avl_right, node); /* node: RH+1 */
lr->avl_height = rh + 2;
- *nodep = lr;
+ RCU_INIT_POINTER(*nodep, lr);
}
} else if (rh > lh + 1) { /* r: LH+2 */
struct inet_peer *rr, *rl, *rlr, *rll;
int rlh;
- rr = r->avl_right;
- rl = r->avl_left;
+ rr = rcu_dereference_protected(r->avl_right,
+ lockdep_is_held(&peers.lock));
+ rl = rcu_dereference_protected(r->avl_left,
+ lockdep_is_held(&peers.lock));
rlh = node_height(rl);
if (rlh <= node_height(rr)) { /* rr: LH+1 */
- node->avl_right = rl; /* rl: LH or LH+1 */
- node->avl_left = l; /* l: LH */
+ RCU_INIT_POINTER(node->avl_right, rl); /* rl: LH or LH+1 */
+ RCU_INIT_POINTER(node->avl_left, l); /* l: LH */
node->avl_height = rlh + 1; /* LH+1 or LH+2 */
- r->avl_right = rr; /* rr: LH+1 */
- r->avl_left = node; /* node: LH+1 or LH+2 */
+ RCU_INIT_POINTER(r->avl_right, rr); /* rr: LH+1 */
+ RCU_INIT_POINTER(r->avl_left, node); /* node: LH+1 or LH+2 */
r->avl_height = node->avl_height + 1;
- *nodep = r;
+ RCU_INIT_POINTER(*nodep, r);
} else { /* rr: RH, rl: RH+1 */
- rlr = rl->avl_right; /* rlr: LH or LH-1 */
- rll = rl->avl_left; /* rll: LH or LH-1 */
- node->avl_right = rll; /* rll: LH or LH-1 */
- node->avl_left = l; /* l: LH */
+ rlr = rcu_dereference_protected(rl->avl_right,
+ lockdep_is_held(&peers.lock)); /* rlr: LH or LH-1 */
+ rll = rcu_dereference_protected(rl->avl_left,
+ lockdep_is_held(&peers.lock)); /* rll: LH or LH-1 */
+ RCU_INIT_POINTER(node->avl_right, rll); /* rll: LH or LH-1 */
+ RCU_INIT_POINTER(node->avl_left, l); /* l: LH */
node->avl_height = lh + 1; /* node: LH+1 */
- r->avl_right = rr; /* rr: LH */
- r->avl_left = rlr; /* rlr: LH or LH-1 */
+ RCU_INIT_POINTER(r->avl_right, rr); /* rr: LH */
+ RCU_INIT_POINTER(r->avl_left, rlr); /* rlr: LH or LH-1 */
r->avl_height = lh + 1; /* r: LH+1 */
- rl->avl_right = r; /* r: LH+1 */
- rl->avl_left = node; /* node: LH+1 */
+ RCU_INIT_POINTER(rl->avl_right, r); /* r: LH+1 */
+ RCU_INIT_POINTER(rl->avl_left, node); /* node: LH+1 */
rl->avl_height = lh + 2;
- *nodep = rl;
+ RCU_INIT_POINTER(*nodep, rl);
}
} else {
node->avl_height = (lh > rh ? lh : rh) + 1;
@@ -303,10 +324,10 @@ static void peer_avl_rebalance(struct inet_peer **stack[],
#define link_to_pool(n) \
do { \
n->avl_height = 1; \
- n->avl_left = peer_avl_empty; \
- n->avl_right = peer_avl_empty; \
- smp_wmb(); /* lockless readers can catch us now */ \
- **--stackptr = n; \
+ n->avl_left = peer_avl_empty_rcu; \
+ n->avl_right = peer_avl_empty_rcu; \
+ /* lockless readers can catch us now */ \
+ rcu_assign_pointer(**--stackptr, n); \
peer_avl_rebalance(stack, stackptr); \
} while (0)
@@ -330,24 +351,25 @@ static void unlink_from_pool(struct inet_peer *p)
* We use refcnt=-1 to alert lockless readers this entry is deleted.
*/
if (atomic_cmpxchg(&p->refcnt, 1, -1) == 1) {
- struct inet_peer **stack[PEER_MAXDEPTH];
- struct inet_peer ***stackptr, ***delp;
+ struct inet_peer __rcu **stack[PEER_MAXDEPTH];
+ struct inet_peer __rcu ***stackptr, ***delp;
if (lookup(p->v4daddr, stack) != p)
BUG();
delp = stackptr - 1; /* *delp[0] == p */
- if (p->avl_left == peer_avl_empty) {
+ if (p->avl_left == peer_avl_empty_rcu) {
*delp[0] = p->avl_right;
--stackptr;
} else {
/* look for a node to insert instead of p */
struct inet_peer *t;
t = lookup_rightempty(p);
- BUG_ON(*stackptr[-1] != t);
+ BUG_ON(rcu_dereference_protected(*stackptr[-1],
+ lockdep_is_held(&peers.lock)) != t);
**--stackptr = t->avl_left;
/* t is removed, t->v4daddr > x->v4daddr for any
* x in p->avl_left subtree.
* Put t in the old place of p. */
- *delp[0] = t;
+ RCU_INIT_POINTER(*delp[0], t);
t->avl_left = p->avl_left;
t->avl_right = p->avl_right;
t->avl_height = p->avl_height;
@@ -414,7 +436,7 @@ static int cleanup_once(unsigned long ttl)
struct inet_peer *inet_getpeer(__be32 daddr, int create)
{
struct inet_peer *p;
- struct inet_peer **stack[PEER_MAXDEPTH], ***stackptr;
+ struct inet_peer __rcu **stack[PEER_MAXDEPTH], ***stackptr;
/* Look up for the address quickly, lockless.
* Because of a concurrent writer, we might not find an existing entry.
^ permalink raw reply related
* Re: [PATCH] qlcnic: dma address align check
From: Eric Dumazet @ 2010-10-26 9:54 UTC (permalink / raw)
To: Amit Kumar Salecha; +Cc: davem, netdev, ameen.rahman, anirban.chakraborty
In-Reply-To: <1288085882-11988-5-git-send-email-amit.salecha@qlogic.com>
Le mardi 26 octobre 2010 à 02:38 -0700, Amit Kumar Salecha a écrit :
> Device requires tx_hw_cosnumer to be 64 byte aligned.
> Tx desc size is 64 byte, alloc tx_hw_consumer with tx desc.
>
> Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
> ---
> drivers/net/qlcnic/qlcnic_ctx.c | 35 +++++++++++++++--------------------
> 1 files changed, 15 insertions(+), 20 deletions(-)
>
> diff --git a/drivers/net/qlcnic/qlcnic_ctx.c b/drivers/net/qlcnic/qlcnic_ctx.c
> index 1cdc05d..21c9c28 100644
> --- a/drivers/net/qlcnic/qlcnic_ctx.c
> +++ b/drivers/net/qlcnic/qlcnic_ctx.c
> @@ -418,18 +418,9 @@ int qlcnic_alloc_hw_resources(struct qlcnic_adapter *adapter)
> recv_ctx = &adapter->recv_ctx;
> tx_ring = adapter->tx_ring;
>
> - tx_ring->hw_consumer = (__le32 *)pci_alloc_consistent(pdev, sizeof(u32),
> - &tx_ring->hw_cons_phys_addr);
> - if (tx_ring->hw_consumer == NULL) {
> - dev_err(&pdev->dev, "failed to allocate tx consumer\n");
> - return -ENOMEM;
> - }
> - *(tx_ring->hw_consumer) = 0;
> -
> /* cmd desc ring */
> - addr = pci_alloc_consistent(pdev, TX_DESC_RINGSIZE(tx_ring),
> - &tx_ring->phys_addr);
> -
> + addr = pci_alloc_consistent(pdev, TX_DESC_RINGSIZE(tx_ring) +
> + sizeof(u32), &tx_ring->phys_addr);
Wont this use twice memory than before, due to power-of-two
allocations ?
Allocating 65536 + 4 bytes gives you 131072 bytes.
^ permalink raw reply
* Re: [v3 RFC PATCH 0/4] Implement multiqueue virtio-net
From: Michael S. Tsirkin @ 2010-10-26 9:38 UTC (permalink / raw)
To: Krishna Kumar2
Cc: anthony, arnd, avi, davem, eric.dumazet, kvm, netdev, rusty
In-Reply-To: <OFBB51646E.01CC3630-ON652577C8.0030D6EF-652577C8.0031DA3C@in.ibm.com>
On Tue, Oct 26, 2010 at 02:38:53PM +0530, Krishna Kumar2 wrote:
> Results for UDP BW tests (unidirectional, sum across
> 3 iterations, each iteration of 45 seconds, default
> netperf, vhosts bound to cpus 0-3; no other tuning):
Is binding vhost threads to CPUs really required?
What happens if we let the scheduler do its job?
--
MST
^ permalink raw reply
* [PATCH] qlcnic: define valid vlan id range
From: Amit Kumar Salecha @ 2010-10-26 9:38 UTC (permalink / raw)
To: davem; +Cc: netdev, ameen.rahman, anirban.chakraborty, Sony Chacko
In-Reply-To: <1288085882-11988-1-git-send-email-amit.salecha@qlogic.com>
From: Sony Chacko <sony.chacko@qlogic.com>
4095 vlan id is reserved and should not be use.
Signed-off-by: Sony Chacko <sony.chacko@qlogic.com>
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
---
drivers/net/qlcnic/qlcnic.h | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/net/qlcnic/qlcnic.h b/drivers/net/qlcnic/qlcnic.h
index 6400e6a..8ecc170 100644
--- a/drivers/net/qlcnic/qlcnic.h
+++ b/drivers/net/qlcnic/qlcnic.h
@@ -1134,7 +1134,7 @@ struct qlcnic_eswitch {
#define MAX_RX_QUEUES 4
#define DEFAULT_MAC_LEARN 1
-#define IS_VALID_VLAN(vlan) (vlan >= MIN_VLAN_ID && vlan <= MAX_VLAN_ID)
+#define IS_VALID_VLAN(vlan) (vlan >= MIN_VLAN_ID && vlan < MAX_VLAN_ID)
#define IS_VALID_BW(bw) (bw >= MIN_BW && bw <= MAX_BW)
#define IS_VALID_TX_QUEUES(que) (que > 0 && que <= MAX_TX_QUEUES)
#define IS_VALID_RX_QUEUES(que) (que > 0 && que <= MAX_RX_QUEUES)
--
1.6.0.2
^ permalink raw reply related
* [PATCH] qlcnic: reduce RX ring size for nic partition
From: Amit Kumar Salecha @ 2010-10-26 9:38 UTC (permalink / raw)
To: davem; +Cc: netdev, ameen.rahman, anirban.chakraborty, Sony Chacko
In-Reply-To: <1288085882-11988-1-git-send-email-amit.salecha@qlogic.com>
From: Sony Chacko <sony.chacko@qlogic.com>
For NIC Partition adapter, rcv ring size can be reduce, as
physical port is partition-ed.
Signed-off-by: Sony Chacko <sony.chacko@qlogic.com>
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
---
drivers/net/qlcnic/qlcnic.h | 4 ++++
drivers/net/qlcnic/qlcnic_ethtool.c | 23 +++++------------------
drivers/net/qlcnic/qlcnic_main.c | 14 ++++++++++++--
3 files changed, 21 insertions(+), 20 deletions(-)
diff --git a/drivers/net/qlcnic/qlcnic.h b/drivers/net/qlcnic/qlcnic.h
index a60ff17..6400e6a 100644
--- a/drivers/net/qlcnic/qlcnic.h
+++ b/drivers/net/qlcnic/qlcnic.h
@@ -146,11 +146,13 @@
#define MAX_CMD_DESCRIPTORS 1024
#define MAX_RCV_DESCRIPTORS_1G 4096
#define MAX_RCV_DESCRIPTORS_10G 8192
+#define MAX_RCV_DESCRIPTORS_VF 2048
#define MAX_JUMBO_RCV_DESCRIPTORS_1G 512
#define MAX_JUMBO_RCV_DESCRIPTORS_10G 1024
#define DEFAULT_RCV_DESCRIPTORS_1G 2048
#define DEFAULT_RCV_DESCRIPTORS_10G 4096
+#define DEFAULT_RCV_DESCRIPTORS_VF 1024
#define MAX_RDS_RINGS 2
#define get_next_index(index, length) \
@@ -971,6 +973,8 @@ struct qlcnic_adapter {
u16 num_txd;
u16 num_rxd;
u16 num_jumbo_rxd;
+ u16 max_rxd;
+ u16 max_jumbo_rxd;
u8 max_rds_rings;
u8 max_sds_rings;
diff --git a/drivers/net/qlcnic/qlcnic_ethtool.c b/drivers/net/qlcnic/qlcnic_ethtool.c
index 25e93a5..ec21d24 100644
--- a/drivers/net/qlcnic/qlcnic_ethtool.c
+++ b/drivers/net/qlcnic/qlcnic_ethtool.c
@@ -437,14 +437,8 @@ qlcnic_get_ringparam(struct net_device *dev,
ring->rx_jumbo_pending = adapter->num_jumbo_rxd;
ring->tx_pending = adapter->num_txd;
- if (adapter->ahw.port_type == QLCNIC_GBE) {
- ring->rx_max_pending = MAX_RCV_DESCRIPTORS_1G;
- ring->rx_jumbo_max_pending = MAX_JUMBO_RCV_DESCRIPTORS_1G;
- } else {
- ring->rx_max_pending = MAX_RCV_DESCRIPTORS_10G;
- ring->rx_jumbo_max_pending = MAX_JUMBO_RCV_DESCRIPTORS_10G;
- }
-
+ ring->rx_max_pending = adapter->max_rxd;
+ ring->rx_jumbo_max_pending = adapter->max_jumbo_rxd;
ring->tx_max_pending = MAX_CMD_DESCRIPTORS;
ring->rx_mini_max_pending = 0;
@@ -472,24 +466,17 @@ qlcnic_set_ringparam(struct net_device *dev,
struct ethtool_ringparam *ring)
{
struct qlcnic_adapter *adapter = netdev_priv(dev);
- u16 max_rcv_desc = MAX_RCV_DESCRIPTORS_10G;
- u16 max_jumbo_desc = MAX_JUMBO_RCV_DESCRIPTORS_10G;
u16 num_rxd, num_jumbo_rxd, num_txd;
-
if (ring->rx_mini_pending)
return -EOPNOTSUPP;
- if (adapter->ahw.port_type == QLCNIC_GBE) {
- max_rcv_desc = MAX_RCV_DESCRIPTORS_1G;
- max_jumbo_desc = MAX_JUMBO_RCV_DESCRIPTORS_10G;
- }
-
num_rxd = qlcnic_validate_ringparam(ring->rx_pending,
- MIN_RCV_DESCRIPTORS, max_rcv_desc, "rx");
+ MIN_RCV_DESCRIPTORS, adapter->max_rxd, "rx");
num_jumbo_rxd = qlcnic_validate_ringparam(ring->rx_jumbo_pending,
- MIN_JUMBO_DESCRIPTORS, max_jumbo_desc, "rx jumbo");
+ MIN_JUMBO_DESCRIPTORS, adapter->max_jumbo_rxd,
+ "rx jumbo");
num_txd = qlcnic_validate_ringparam(ring->tx_pending,
MIN_CMD_DESCRIPTORS, MAX_CMD_DESCRIPTORS, "tx");
diff --git a/drivers/net/qlcnic/qlcnic_main.c b/drivers/net/qlcnic/qlcnic_main.c
index 5a3ce08..7a298cd 100644
--- a/drivers/net/qlcnic/qlcnic_main.c
+++ b/drivers/net/qlcnic/qlcnic_main.c
@@ -656,13 +656,23 @@ qlcnic_check_options(struct qlcnic_adapter *adapter)
dev_info(&pdev->dev, "firmware v%d.%d.%d\n",
fw_major, fw_minor, fw_build);
-
if (adapter->ahw.port_type == QLCNIC_XGBE) {
- adapter->num_rxd = DEFAULT_RCV_DESCRIPTORS_10G;
+ if (adapter->flags & QLCNIC_ESWITCH_ENABLED) {
+ adapter->num_rxd = DEFAULT_RCV_DESCRIPTORS_VF;
+ adapter->max_rxd = MAX_RCV_DESCRIPTORS_VF;
+ } else {
+ adapter->num_rxd = DEFAULT_RCV_DESCRIPTORS_10G;
+ adapter->max_rxd = MAX_RCV_DESCRIPTORS_10G;
+ }
+
adapter->num_jumbo_rxd = MAX_JUMBO_RCV_DESCRIPTORS_10G;
+ adapter->max_jumbo_rxd = MAX_JUMBO_RCV_DESCRIPTORS_10G;
+
} else if (adapter->ahw.port_type == QLCNIC_GBE) {
adapter->num_rxd = DEFAULT_RCV_DESCRIPTORS_1G;
adapter->num_jumbo_rxd = MAX_JUMBO_RCV_DESCRIPTORS_1G;
+ adapter->max_jumbo_rxd = MAX_JUMBO_RCV_DESCRIPTORS_1G;
+ adapter->max_rxd = MAX_RCV_DESCRIPTORS_1G;
}
adapter->msix_supported = !!use_msi_x;
--
1.6.0.2
^ permalink raw reply related
* [PATCH] qlcnic: fix mac learning
From: Amit Kumar Salecha @ 2010-10-26 9:37 UTC (permalink / raw)
To: davem; +Cc: netdev, ameen.rahman, anirban.chakraborty
In-Reply-To: <1288085882-11988-1-git-send-email-amit.salecha@qlogic.com>
In failover bonding case, same mac address can be programmed on other slave function.
Fw will delete old entry (original func) associated with that mac address.
Need to reporgram mac address, if failover again happen to original function.
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
---
drivers/net/qlcnic/qlcnic.h | 1 +
drivers/net/qlcnic/qlcnic_main.c | 5 +++++
2 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/drivers/net/qlcnic/qlcnic.h b/drivers/net/qlcnic/qlcnic.h
index 26c37d3..a60ff17 100644
--- a/drivers/net/qlcnic/qlcnic.h
+++ b/drivers/net/qlcnic/qlcnic.h
@@ -942,6 +942,7 @@ struct qlcnic_ipaddr {
#define QLCNIC_LOOPBACK_TEST 2
#define QLCNIC_FILTER_AGE 80
+#define QLCNIC_READD_AGE 20
#define QLCNIC_LB_MAX_FILTERS 64
struct qlcnic_filter {
diff --git a/drivers/net/qlcnic/qlcnic_main.c b/drivers/net/qlcnic/qlcnic_main.c
index f047c7c..5a3ce08 100644
--- a/drivers/net/qlcnic/qlcnic_main.c
+++ b/drivers/net/qlcnic/qlcnic_main.c
@@ -1860,6 +1860,11 @@ qlcnic_send_filter(struct qlcnic_adapter *adapter,
hlist_for_each_entry_safe(tmp_fil, tmp_hnode, n, head, fnode) {
if (!memcmp(tmp_fil->faddr, &src_addr, ETH_ALEN) &&
tmp_fil->vlan_id == vlan_id) {
+
+ if (jiffies >
+ (QLCNIC_READD_AGE * HZ + tmp_fil->ftime))
+ qlcnic_change_filter(adapter, src_addr, vlan_id,
+ tx_ring);
tmp_fil->ftime = jiffies;
return;
}
--
1.6.0.2
^ permalink raw reply related
* [PATCH] qlcnic: dma address align check
From: Amit Kumar Salecha @ 2010-10-26 9:38 UTC (permalink / raw)
To: davem; +Cc: netdev, ameen.rahman, anirban.chakraborty
In-Reply-To: <1288085882-11988-1-git-send-email-amit.salecha@qlogic.com>
Device requires tx_hw_cosnumer to be 64 byte aligned.
Tx desc size is 64 byte, alloc tx_hw_consumer with tx desc.
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
---
drivers/net/qlcnic/qlcnic_ctx.c | 35 +++++++++++++++--------------------
1 files changed, 15 insertions(+), 20 deletions(-)
diff --git a/drivers/net/qlcnic/qlcnic_ctx.c b/drivers/net/qlcnic/qlcnic_ctx.c
index 1cdc05d..21c9c28 100644
--- a/drivers/net/qlcnic/qlcnic_ctx.c
+++ b/drivers/net/qlcnic/qlcnic_ctx.c
@@ -418,18 +418,9 @@ int qlcnic_alloc_hw_resources(struct qlcnic_adapter *adapter)
recv_ctx = &adapter->recv_ctx;
tx_ring = adapter->tx_ring;
- tx_ring->hw_consumer = (__le32 *)pci_alloc_consistent(pdev, sizeof(u32),
- &tx_ring->hw_cons_phys_addr);
- if (tx_ring->hw_consumer == NULL) {
- dev_err(&pdev->dev, "failed to allocate tx consumer\n");
- return -ENOMEM;
- }
- *(tx_ring->hw_consumer) = 0;
-
/* cmd desc ring */
- addr = pci_alloc_consistent(pdev, TX_DESC_RINGSIZE(tx_ring),
- &tx_ring->phys_addr);
-
+ addr = pci_alloc_consistent(pdev, TX_DESC_RINGSIZE(tx_ring) +
+ sizeof(u32), &tx_ring->phys_addr);
if (addr == NULL) {
dev_err(&pdev->dev, "failed to allocate tx desc ring\n");
err = -ENOMEM;
@@ -437,6 +428,17 @@ int qlcnic_alloc_hw_resources(struct qlcnic_adapter *adapter)
}
tx_ring->desc_head = (struct cmd_desc_type0 *)addr;
+ tx_ring->hw_consumer = (__le32 *)(((char *)addr) +
+ TX_DESC_RINGSIZE(tx_ring));
+ tx_ring->hw_cons_phys_addr = (dma_addr_t)(((char *)tx_ring->phys_addr) +
+ TX_DESC_RINGSIZE(tx_ring));
+ if (tx_ring->hw_cons_phys_addr & 0x3F) {
+ dev_err(&pdev->dev, "Device requires 64 byte aligned dma addr"
+ ". dma_addr=%p\n", (void *)tx_ring->hw_cons_phys_addr);
+ err = -ENOMEM;
+ goto err_out_free;
+ }
+ *(tx_ring->hw_consumer) = 0;
for (ring = 0; ring < adapter->max_rds_rings; ring++) {
rds_ring = &recv_ctx->rds_rings[ring];
@@ -516,19 +518,12 @@ void qlcnic_free_hw_resources(struct qlcnic_adapter *adapter)
recv_ctx = &adapter->recv_ctx;
tx_ring = adapter->tx_ring;
- if (tx_ring->hw_consumer != NULL) {
- pci_free_consistent(adapter->pdev,
- sizeof(u32),
- tx_ring->hw_consumer,
- tx_ring->hw_cons_phys_addr);
- tx_ring->hw_consumer = NULL;
- }
-
if (tx_ring->desc_head != NULL) {
pci_free_consistent(adapter->pdev,
- TX_DESC_RINGSIZE(tx_ring),
+ TX_DESC_RINGSIZE(tx_ring) + sizeof(u32),
tx_ring->desc_head, tx_ring->phys_addr);
tx_ring->desc_head = NULL;
+ tx_ring->hw_consumer = NULL;
}
for (ring = 0; ring < adapter->max_rds_rings; ring++) {
--
1.6.0.2
^ permalink raw reply related
* [PATCH 0/4]qlcnic: bug fixes
From: Amit Kumar Salecha @ 2010-10-26 9:37 UTC (permalink / raw)
To: davem; +Cc: netdev, ameen.rahman, anirban.chakraborty
Hi
Series of 4 bug fixes. Apply them on net-2.6 branch.
-Amit
^ permalink raw reply
* RE: [PATCH v2 11/14] bnx2x: Update bnx2x to use new vlan accleration.
From: Vladislav Zolotarov @ 2010-10-26 9:14 UTC (permalink / raw)
To: Jesse Gross
Cc: David Miller, netdev@vger.kernel.org, Hao Zheng, Eilon Greenstein
In-Reply-To: <AANLkTimjTZU_GWgBNADgAtw7ViztZD-cdhHLQYremz2A@mail.gmail.com>
>
> Even though the fix prevents the panic, I'm still a little concerned
> that you ran into it at all when vlan stripping was disabled. That
> function should only be called when a tag was received by the card.
> Is it possible that __vlan_hwaccel_put_tag is being called even in
> cases when no tag was stripped?
Correct, and that's because u've patched the driver to put it the
following way:
if (le16_to_cpu(cqe->fast_path_cqe.pars_flags.flags) &
PARSING_FLAGS_VLAN)
__vlan_hwaccel_put_tag(skb, le16_to_cpu(cqe->fast_path_cqe.vlan_tag));
The condition above will be TRUE regardless VLAN stripping is enabled or
disabled as the parsing flags come from our PARSER HW block and simply
indicates whether this frame has a VLAN header or not. When I disabled a VLAN
stripping I should have fixed this lines too but I think I didn't... ;)
> Maybe we made a mistake when converting the driver?
Since the driver always configures the VLAN stripping now, regardless the kernel
configuration, I think your patch was just fine. ;)
Thanks,
vlad
^ permalink raw reply
* Re: [v3 RFC PATCH 0/4] Implement multiqueue virtio-net
From: Krishna Kumar2 @ 2010-10-26 9:08 UTC (permalink / raw)
To: Krishna Kumar2
Cc: anthony, arnd, avi, davem, eric.dumazet, kvm, Michael S. Tsirkin,
netdev, rusty
In-Reply-To: <OF5C53E9CF.FFDF2CE7-ON652577C8.00191D14-652577C8.001C2154@LocalDomain>
Krishna Kumar2/India/IBM wrote on 10/26/2010 10:40:35 AM:
> > I am trying to wrap my head around kernel/user interface here.
> > E.g., will we need another incompatible change when we add multiple RX
> > queues?
>
> Though I added a 'mq' option to qemu, there shouldn't be
> any incompatibility between old and new qemu's wrt vhost
> and virtio-net drivers. So the old qemu will run new host
> and new guest without issues, and new qemu can also run
> old host and old guest. Multiple RXQ will also not add
> any incompatibility.
>
> With MQ RX, I will be able to remove the hueristic (idea
> from David Stevens). The idea is: Guest sends out packets
> on, say TXQ#2, vhost#2 processes the packets but packets
> going out from host to guest might be sent out on a
> different RXQ, say RXQ#4. Guest receives the packet on
> RXQ#4, and all future responses on that connection are sent
> on TXQ#4. Now vhost#4 processes both RX and TX packets for
> this connection. Without needing to hash on the connection,
> guest can make sure that the same vhost thread will handle
> a single connection.
>
> > Also need to think about how robust our single stream heuristic is,
> > e.g. what are the chances it will misdetect a bidirectional
> > UDP stream as a single TCP?
> I think it should not happen. The hueristic code gets
> called for handling just the transmit packets, packets
> that vhost sends out to the guest skip this path.
>
> I tested unidirectional and bidirectional UDP to confirm:
>
> 8 iterations of iperf tests, each iteration of 15 secs,
> result is the sum of all 8 iterations in Gbits/sec
> __________________________________________
> Uni-directional Bi-directional
> Org New Org New
> __________________________________________
> 71.78 71.77 71.74 72.07
> __________________________________________
Results for UDP BW tests (unidirectional, sum across
3 iterations, each iteration of 45 seconds, default
netperf, vhosts bound to cpus 0-3; no other tuning):
------ numtxqs=8, vhosts=5 ---------
# BW% CPU% SD%
------------------------------------
1 .49 1.07 0
2 23.51 52.51 26.66
4 75.17 72.43 8.57
8 86.54 80.21 27.85
16 92.37 85.99 6.27
24 91.37 84.91 8.41
32 89.78 82.90 3.31
48 89.85 79.95 -3.57
64 85.83 80.28 2.22
80 88.90 79.47 -23.18
96 90.12 79.98 14.71
128 86.13 80.60 4.42
------------------------------------
BW: 71.3%, CPU: 80.4%, SD: 1.2%
------ numtxqs=16, vhosts=5 --------
# BW% CPU% SD%
------------------------------------
1 1.80 0 0
2 19.81 50.68 26.66
4 57.31 52.77 8.57
8 108.44 88.19 -5.21
16 106.09 85.03 -4.44
24 102.34 84.23 -.82
32 102.77 82.71 -5.81
48 100.00 79.62 -7.29
64 96.86 79.75 -6.10
80 99.26 79.82 -27.34
96 94.79 80.02 -5.08
128 98.14 81.15 -15.25
------------------------------------
BW: 77.9%, CPU: 80.4%, SD: -13.6%
Thanks,
- KK
^ permalink raw reply
* Re: [v3 RFC PATCH 0/4] Implement multiqueue virtio-net
From: Michael S. Tsirkin @ 2010-10-26 8:57 UTC (permalink / raw)
To: Krishna Kumar2
Cc: anthony, arnd, avi, davem, eric.dumazet, kvm, netdev, rusty
In-Reply-To: <OF8735A2A4.74B4EE91-ON652577C5.0040D90B-652577C7.0056B99D@in.ibm.com>
On Mon, Oct 25, 2010 at 09:20:38PM +0530, Krishna Kumar2 wrote:
> > Krishna Kumar2/India/IBM@IBMIN wrote on 10/20/2010 02:24:52 PM:
>
> Any feedback, comments, objections, issues or bugs about the
> patches? Please let me know if something needs to be done.
>
> Some more test results:
> _____________________________________________________
> Host->Guest BW (numtxqs=2)
> # BW% CPU% RCPU% SD% RSD%
> _____________________________________________________
I think we discussed the need for external to guest testing
over 10G. For large messages we should not see any change
but you should be able to get better numbers for small messages
assuming a MQ NIC card.
--
MST
^ permalink raw reply
* [PATCH net-next-2.6] be2net: Schedule/Destroy worker thread in probe()/remove() rather than open()/close()
From: Somnath Kotur @ 2010-10-26 9:01 UTC (permalink / raw)
To: netdev
When async mcc compls are rcvd on an i/f that is down (and so interrupts are disabled)
they just lie unprocessed in the compl queue.The compl queue can eventually get filled
up and cause the BE to lock up.The fix is to use be_worker to reap mcc compls when the
i/f is down.be_worker is now launched in be_probe() and canceled in be_remove().
Signed-off-by: Somnath Kotur <somnath.kotur@emulex.com>
---
drivers/net/benet/be_main.c | 22 ++++++++++++++++++----
1 files changed, 18 insertions(+), 4 deletions(-)
diff --git a/drivers/net/benet/be_main.c b/drivers/net/benet/be_main.c
index 45b1f66..deb546e 100644
--- a/drivers/net/benet/be_main.c
+++ b/drivers/net/benet/be_main.c
@@ -1806,6 +1806,20 @@ static void be_worker(struct work_struct *work)
struct be_rx_obj *rxo;
int i;
+ /* when interrupts are not yet enabled, just reap any pending
+ * mcc completions */
+ if (!netif_running(adapter->netdev)) {
+ int mcc_compl, status = 0;
+
+ mcc_compl = be_process_mcc(adapter, &status);
+
+ if (mcc_compl) {
+ struct be_mcc_obj *mcc_obj = &adapter->mcc_obj;
+ be_cq_notify(adapter, mcc_obj->cq.id, false, mcc_compl);
+ }
+ goto reschedule;
+ }
+
if (!adapter->stats_ioctl_sent)
be_cmd_get_stats(adapter, &adapter->stats_cmd);
@@ -1824,6 +1838,7 @@ static void be_worker(struct work_struct *work)
if (!adapter->ue_detected)
be_detect_dump_ue(adapter);
+reschedule:
schedule_delayed_work(&adapter->work, msecs_to_jiffies(1000));
}
@@ -2019,8 +2034,6 @@ static int be_close(struct net_device *netdev)
struct be_eq_obj *tx_eq = &adapter->tx_eq;
int vec, i;
- cancel_delayed_work_sync(&adapter->work);
-
be_async_mcc_disable(adapter);
netif_stop_queue(netdev);
@@ -2085,8 +2098,6 @@ static int be_open(struct net_device *netdev)
/* Now that interrupts are on we can process async mcc */
be_async_mcc_enable(adapter);
- schedule_delayed_work(&adapter->work, msecs_to_jiffies(100));
-
status = be_cmd_link_status_query(adapter, &link_up, &mac_speed,
&link_speed);
if (status)
@@ -2715,6 +2726,8 @@ static void __devexit be_remove(struct pci_dev *pdev)
if (!adapter)
return;
+ cancel_delayed_work_sync(&adapter->work);
+
unregister_netdev(adapter->netdev);
be_clear(adapter);
@@ -2870,6 +2883,7 @@ static int __devinit be_probe(struct pci_dev *pdev,
goto unsetup;
dev_info(&pdev->dev, "%s port %d\n", nic_name(pdev), adapter->port_num);
+ schedule_delayed_work(&adapter->work, msecs_to_jiffies(100));
return 0;
unsetup:
--
1.5.6.1
^ 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