* [patch 1/1] fix "netpoll: Allow netpoll_setup/cleanup recursion"
From: akpm @ 2010-06-25 3:54 UTC (permalink / raw)
To: davem; +Cc: netdev, akpm, herbert
From: Andrew Morton <akpm@linux-foundation.org>
Remove rtnl_unlock() which had no corresponding rtnl_lock().
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---
net/core/netpoll.c | 1 -
1 file changed, 1 deletion(-)
diff -puN net/core/netpoll.c~fix-netpoll-allow-netpoll_setup-cleanup-recursion net/core/netpoll.c
--- a/net/core/netpoll.c~fix-netpoll-allow-netpoll_setup-cleanup-recursion
+++ a/net/core/netpoll.c
@@ -748,7 +748,6 @@ int __netpoll_setup(struct netpoll *np)
/* last thing to do is link it to the net device structure */
rcu_assign_pointer(ndev->npinfo, npinfo);
- rtnl_unlock();
return 0;
_
^ permalink raw reply
* Re: [PATCH 6/8] netpoll: Allow netpoll_setup/cleanup recursion
From: Andrew Morton @ 2010-06-25 3:50 UTC (permalink / raw)
To: David Miller
Cc: herbert, mst, frzhang, netdev, amwang, shemminger, mpm, paulmck,
a.p.zijlstra, mingo
In-Reply-To: <20100624.203006.35035648.davem@davemloft.net>
On Thu, 24 Jun 2010 20:30:06 -0700 (PDT) David Miller <davem@davemloft.net> wrote:
> From: Andrew Morton <akpm@linux-foundation.org>
> Date: Thu, 24 Jun 2010 18:21:23 -0700
>
> > Dave, the fixup should be folded into the original patch please -
> > otherwise we'll have a machine-hangs-up bisection hole which spans two
> > weeks work of commits.
>
> I'm not unwrapping my tree that far back, and there's already been
> a dependent tree pull or two which depend upon the tree past that
> commit, one of it was a rather large wireless merge.
>
> It's going to screw over too many people.
>
> We'll just merge the bug fix in, and be done with it.
Sigh. That really sucks.
What happens if you want to actually *drop* a patch from net-next?
Surely that happens?
^ permalink raw reply
* Re: [PATCH 6/8] netpoll: Allow netpoll_setup/cleanup recursion
From: David Miller @ 2010-06-25 3:30 UTC (permalink / raw)
To: akpm
Cc: herbert, mst, frzhang, netdev, amwang, shemminger, mpm, paulmck,
a.p.zijlstra, mingo
In-Reply-To: <20100624182123.45264dfe.akpm@linux-foundation.org>
From: Andrew Morton <akpm@linux-foundation.org>
Date: Thu, 24 Jun 2010 18:21:23 -0700
> Dave, the fixup should be folded into the original patch please -
> otherwise we'll have a machine-hangs-up bisection hole which spans two
> weeks work of commits.
I'm not unwrapping my tree that far back, and there's already been
a dependent tree pull or two which depend upon the tree past that
commit, one of it was a rather large wireless merge.
It's going to screw over too many people.
We'll just merge the bug fix in, and be done with it.
^ permalink raw reply
* Re: [PATCH 6/8] netpoll: Allow netpoll_setup/cleanup recursion
From: Herbert Xu @ 2010-06-25 3:01 UTC (permalink / raw)
To: Andrew Morton
Cc: Michael S. Tsirkin, Qianfeng Zhang, David S. Miller, netdev,
WANG Cong, Stephen Hemminger, Matt Mackall, Paul E. McKenney,
Peter Zijlstra, Ingo Molnar
In-Reply-To: <20100624182123.45264dfe.akpm@linux-foundation.org>
On Thu, Jun 24, 2010 at 06:21:23PM -0700, Andrew Morton wrote:
>
> I just removed the rtnl_unlock() - I couldn't see much in there which
> needed rtnl_locking..
Yes that's the correct fix. I forgot to remove it from the
recursive setup path after moving the RTNL lock out. Sorry!
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply
* Re: + fix-netpoll-allow-netpoll_setup-cleanup-recursion.patch added to -mm tree
From: Herbert Xu @ 2010-06-25 3:00 UTC (permalink / raw)
To: akpm; +Cc: mm-commits, davem, netdev
In-Reply-To: <201006250121.o5P1La69008693@imap1.linux-foundation.org>
On Thu, Jun 24, 2010 at 05:52:47PM -0700, akpm@linux-foundation.org wrote:
>
> The patch titled
> fix "netpoll: Allow netpoll_setup/cleanup recursion"
> has been added to the -mm tree. Its filename is
> fix-netpoll-allow-netpoll_setup-cleanup-recursion.patch
>
> Before you just go and hit "reply", please:
> a) Consider who else should be cc'ed
> b) Prefer to cc a suitable mailing list as well
> c) Ideally: find the original patch on the mailing list and do a
> reply-to-all to that, adding suitable additional cc's
>
> *** Remember to use Documentation/SubmitChecklist when testing your code ***
>
> See http://userweb.kernel.org/~akpm/stuff/added-to-mm.txt to find
> out what to do about this
>
> The current -mm tree may be found at http://userweb.kernel.org/~akpm/mmotm/
>
> ------------------------------------------------------
> Subject: fix "netpoll: Allow netpoll_setup/cleanup recursion"
> From: Andrew Morton <akpm@linux-foundation.org>
>
> Remove rtnl_unlock() which had no corresponding rtnl_lock().
>
> Cc: Herbert Xu <herbert@gondor.apana.org.au>
> Cc: David S. Miller <davem@davemloft.net>
> Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Thanks for catching this Andrew!
> ---
>
> net/core/netpoll.c | 1 -
> 1 file changed, 1 deletion(-)
>
> diff -puN net/core/netpoll.c~fix-netpoll-allow-netpoll_setup-cleanup-recursion net/core/netpoll.c
> --- a/net/core/netpoll.c~fix-netpoll-allow-netpoll_setup-cleanup-recursion
> +++ a/net/core/netpoll.c
> @@ -748,7 +748,6 @@ int __netpoll_setup(struct netpoll *np)
>
> /* last thing to do is link it to the net device structure */
> rcu_assign_pointer(ndev->npinfo, npinfo);
> - rtnl_unlock();
>
> return 0;
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply
* [PATCH v2] fragment: add fast path
From: Changli Gao @ 2010-06-25 2:54 UTC (permalink / raw)
To: David Miller
Cc: Alexey Kuznetsov, Pekka Savola (ipv6), James Morris,
Hideaki YOSHIFUJI, Patrick McHardy, Eric Dumazet, netdev,
Mitchell Erblich, Changli Gao
add fast path
As the fragments are sent in order in most of OSes, such as Windows, Darwin and
FreeBSD, it is likely the new fragments are at the end of the inet_frag_queue.
In the fast path, we check if the skb at the end of the inet_frag_queue is the
prev we expect.
Signed-off-by: Changli Gao <xiaosuo@gmail.com>
----
include/net/inet_frag.h | 1 +
net/ipv4/ip_fragment.c | 12 ++++++++++++
net/ipv6/reassembly.c | 11 +++++++++++
3 files changed, 24 insertions(+)
diff --git a/include/net/inet_frag.h b/include/net/inet_frag.h
index 39f2dc9..16ff29a 100644
--- a/include/net/inet_frag.h
+++ b/include/net/inet_frag.h
@@ -20,6 +20,7 @@ struct inet_frag_queue {
atomic_t refcnt;
struct timer_list timer; /* when will this queue expire? */
struct sk_buff *fragments; /* list of received fragments */
+ struct sk_buff *fragments_tail;
ktime_t stamp;
int len; /* total length of orig datagram */
int meat;
diff --git a/net/ipv4/ip_fragment.c b/net/ipv4/ip_fragment.c
index 858d346..dbe8999 100644
--- a/net/ipv4/ip_fragment.c
+++ b/net/ipv4/ip_fragment.c
@@ -314,6 +314,7 @@ static int ip_frag_reinit(struct ipq *qp)
qp->q.len = 0;
qp->q.meat = 0;
qp->q.fragments = NULL;
+ qp->q.fragments_tail = NULL;
qp->iif = 0;
return 0;
@@ -386,6 +387,11 @@ static int ip_frag_queue(struct ipq *qp, struct sk_buff *skb)
* in the chain of fragments so far. We must know where to put
* this fragment, right?
*/
+ prev = qp->q.fragments_tail;
+ if (!prev || FRAG_CB(prev)->offset < offset) {
+ next = NULL;
+ goto found;
+ }
prev = NULL;
for (next = qp->q.fragments; next != NULL; next = next->next) {
if (FRAG_CB(next)->offset >= offset)
@@ -393,6 +399,7 @@ static int ip_frag_queue(struct ipq *qp, struct sk_buff *skb)
prev = next;
}
+found:
/* We found where to put this one. Check for overlap with
* preceding fragment, and, if needed, align things so that
* any overlaps are eliminated.
@@ -451,6 +458,8 @@ static int ip_frag_queue(struct ipq *qp, struct sk_buff *skb)
/* Insert this fragment in the chain of fragments. */
skb->next = next;
+ if (!next)
+ qp->q.fragments_tail = skb;
if (prev)
prev->next = skb;
else
@@ -504,6 +513,8 @@ static int ip_frag_reasm(struct ipq *qp, struct sk_buff *prev,
goto out_nomem;
fp->next = head->next;
+ if (!fp->next)
+ qp->q.fragments_tail = fp;
prev->next = fp;
skb_morph(head, qp->q.fragments);
@@ -574,6 +585,7 @@ static int ip_frag_reasm(struct ipq *qp, struct sk_buff *prev,
iph->tot_len = htons(len);
IP_INC_STATS_BH(net, IPSTATS_MIB_REASMOKS);
qp->q.fragments = NULL;
+ qp->q.fragments_tail = NULL;
return 0;
out_nomem:
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index 0b97230..b832f7b 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -333,6 +333,11 @@ static int ip6_frag_queue(struct frag_queue *fq, struct sk_buff *skb,
* in the chain of fragments so far. We must know where to put
* this fragment, right?
*/
+ prev = fq->q.fragments_tail;
+ if (!prev || FRAG6_CB(prev)->offset < offset) {
+ next = NULL;
+ goto found;
+ }
prev = NULL;
for(next = fq->q.fragments; next != NULL; next = next->next) {
if (FRAG6_CB(next)->offset >= offset)
@@ -340,6 +345,7 @@ static int ip6_frag_queue(struct frag_queue *fq, struct sk_buff *skb,
prev = next;
}
+found:
/* We found where to put this one. Check for overlap with
* preceding fragment, and, if needed, align things so that
* any overlaps are eliminated.
@@ -397,6 +403,8 @@ static int ip6_frag_queue(struct frag_queue *fq, struct sk_buff *skb,
/* Insert this fragment in the chain of fragments. */
skb->next = next;
+ if (!next)
+ fq->q.fragments_tail = skb;
if (prev)
prev->next = skb;
else
@@ -463,6 +471,8 @@ static int ip6_frag_reasm(struct frag_queue *fq, struct sk_buff *prev,
goto out_oom;
fp->next = head->next;
+ if (!fp->next)
+ fq->q.fragments_tail = fp;
prev->next = fp;
skb_morph(head, fq->q.fragments);
@@ -549,6 +559,7 @@ static int ip6_frag_reasm(struct frag_queue *fq, struct sk_buff *prev,
IP6_INC_STATS_BH(net, __in6_dev_get(dev), IPSTATS_MIB_REASMOKS);
rcu_read_unlock();
fq->q.fragments = NULL;
+ fq->q.fragments_tail = NULL;
return 1;
out_oversize:
^ permalink raw reply related
* Re: [PATCH net-next] MAINTAINERS: b43 updates
From: Larry Finger @ 2010-06-25 2:41 UTC (permalink / raw)
To: Joe Perches
Cc: John W. Linville, Stefano Brivio,
linux-wireless-u79uwXL29TY76Z2rM5mHXA,
netdev-u79uwXL29TY76Z2rM5mHXA,
linux-kernel-u79uwXL29TY76Z2rM5mHXA, mb-fseUSCV1ubazQB+pC5nmwQ,
zajec5-Re5JQEeQqe8AvxtiuMwx3w, Gábor Stefanik,
Rafał Miłecki
In-Reply-To: <1277423829.1654.111.camel-AkRN8/LKpobuYGix6ZUp1Q@public.gmane.org>
On 06/24/2010 06:57 PM, Joe Perches wrote:
> On Thu, 2010-06-24 at 18:53 -0500, Larry Finger wrote:
>> I never said that b43 is unmaintained. It is definitely getting much
>> more than "odd fixes".
>
> If Stefano isn't maintaining it, who is?
We are working on that.
Larry
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" 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 v3] act_mirred: don't clone skb when skb isn't shared
From: Changli Gao @ 2010-06-25 2:25 UTC (permalink / raw)
To: Jamal Hadi Salim; +Cc: David S. Miller, netdev, Changli Gao
don't clone skb when skb isn't shared
When the tcf_action is TC_ACT_STOLEN, and the skb isn't shared, we don't need
to clone a new skb. As the skb will be freed after this function returns, we
can use it freely once we get a reference to it.
Signed-off-by: Changli Gao <xiaosuo@gmail.com>
----
include/net/sch_generic.h | 11 +++++++++--
net/sched/act_mirred.c | 6 +++---
2 files changed, 12 insertions(+), 5 deletions(-)
diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
index b35301b..977ec06 100644
--- a/include/net/sch_generic.h
+++ b/include/net/sch_generic.h
@@ -594,9 +594,16 @@ static inline u32 qdisc_l2t(struct qdisc_rate_table* rtab, unsigned int pktlen)
}
#ifdef CONFIG_NET_CLS_ACT
-static inline struct sk_buff *skb_act_clone(struct sk_buff *skb, gfp_t gfp_mask)
+static inline struct sk_buff *skb_act_clone(struct sk_buff *skb, gfp_t gfp_mask,
+ int action)
{
- struct sk_buff *n = skb_clone(skb, gfp_mask);
+ struct sk_buff *n;
+
+ if ((action == TC_ACT_STOLEN || action == TC_ACT_QUEUED) &&
+ !skb_shared(skb))
+ n = skb_get(skb);
+ else
+ n = skb_clone(skb, gfp_mask);
if (n) {
n->tc_verd = SET_TC_VERD(n->tc_verd, 0);
diff --git a/net/sched/act_mirred.c b/net/sched/act_mirred.c
index c0b6863..2e9a7b9 100644
--- a/net/sched/act_mirred.c
+++ b/net/sched/act_mirred.c
@@ -169,13 +169,13 @@ static int tcf_mirred(struct sk_buff *skb, struct tc_action *a,
goto out;
}
- skb2 = skb_act_clone(skb, GFP_ATOMIC);
+ at = G_TC_AT(skb->tc_verd);
+ skb2 = skb_act_clone(skb, GFP_ATOMIC, m->tcf_action);
if (skb2 == NULL)
goto out;
m->tcf_bstats.bytes += qdisc_pkt_len(skb2);
m->tcf_bstats.packets++;
- at = G_TC_AT(skb->tc_verd);
if (!(at & AT_EGRESS)) {
if (m->tcfm_ok_push)
skb_push(skb2, skb2->dev->hard_header_len);
@@ -185,8 +185,8 @@ static int tcf_mirred(struct sk_buff *skb, struct tc_action *a,
if (m->tcfm_eaction != TCA_EGRESS_MIRROR)
skb2->tc_verd = SET_TC_FROM(skb2->tc_verd, at);
- skb2->dev = dev;
skb2->skb_iif = skb->dev->ifindex;
+ skb2->dev = dev;
dev_queue_xmit(skb2);
err = 0;
^ permalink raw reply related
* RE: [RFC PATCH v7 01/19] Add a new structure for skb buffer from external.
From: Xin, Xiaohui @ 2010-06-25 2:07 UTC (permalink / raw)
To: Herbert Xu, Dong, Eddie
Cc: Stephen Hemminger, netdev@vger.kernel.org, kvm@vger.kernel.org,
linux-kernel@vger.kernel.org, mst@redhat.com, mingo@elte.hu,
davem@davemloft.net, jdike@linux.intel.com
In-Reply-To: <20100624100836.GA16220@gondor.apana.org.au>
Herbert,
That's why I have sent you the patch for guest virtio-net driver. I reserved 512 bytes in each page, then I can always have the space to copy and avoid the backend memory used up issue.
Thanks
Xiaohui
>-----Original Message-----
>From: Herbert Xu [mailto:herbert@gondor.apana.org.au]
>Sent: Thursday, June 24, 2010 6:09 PM
>To: Dong, Eddie
>Cc: Xin, Xiaohui; Stephen Hemminger; netdev@vger.kernel.org; kvm@vger.kernel.org;
>linux-kernel@vger.kernel.org; mst@redhat.com; mingo@elte.hu; davem@davemloft.net;
>jdike@linux.intel.com
>Subject: Re: [RFC PATCH v7 01/19] Add a new structure for skb buffer from external.
>
>On Wed, Jun 23, 2010 at 06:05:41PM +0800, Dong, Eddie wrote:
>>
>> I mean once the frontend side driver post the buffers to the backend driver, the backend
>driver will "immediately" use that buffers to compose skb or gro_frags and post them to the
>assigned host NIC driver as receive buffers. In that case, if the backend driver recieves a
>packet from the NIC that requires to do copy, it may be unable to find additional free guest
>buffer because all of them are already used by the NIC driver. We have to reserve some guest
>buffers for the possible copy even if the buffer address is not identified by original skb :(
>
>OK I see what you mean. Can you tell me how does Xiaohui's
>previous patch-set deal with this problem?
>
>Thanks,
>--
>Visit Openswan at http://www.openswan.org/
>Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
>Home Page: http://gondor.apana.org.au/~herbert/
>PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply
* Re: [PATCH] smsc95xx: Add module parameter to override MAC address
From: Simon Horman @ 2010-06-25 1:25 UTC (permalink / raw)
To: Sebastien Jan; +Cc: Steve Glendinning, netdev, linux-omap
In-Reply-To: <1277367254-9345-1-git-send-email-s-jan@ti.com>
On Thu, Jun 24, 2010 at 10:14:14AM +0200, Sebastien Jan wrote:
> Define a new module parameter 'macaddr' to override the MAC address
> fetched either from eeprom, or randomly generated.
>
> The expected MAC address shall be in the 01:23:45:67:89:AB format.
I'm confused as to why this is desirable when the mac address
can already be configured after module insertion via
smsc95xx_netdev_ops.eth_mac_addr().
^ permalink raw reply
* Re: [PATCH 6/8] netpoll: Allow netpoll_setup/cleanup recursion
From: Andrew Morton @ 2010-06-25 1:21 UTC (permalink / raw)
To: Herbert Xu
Cc: Michael S. Tsirkin, Qianfeng Zhang, David S. Miller, netdev,
WANG Cong, Stephen Hemminger, Matt Mackall, Paul E. McKenney,
Peter Zijlstra, Ingo Molnar
In-Reply-To: <E1OMtjk-0006P2-6r@gondolin.me.apana.org.au>
On Fri, 11 Jun 2010 12:12:48 +1000
Herbert Xu <herbert@gondor.hengli.com.au> wrote:
> netpoll: Allow netpoll_setup/cleanup recursion
>
> This patch adds the functions __netpoll_setup/__netpoll_cleanup
> which is designed to be called recursively through ndo_netpoll_seutp.
>
> They must be called with RTNL held, and the caller must initialise
> np->dev and ensure that it has a valid reference count.
>
> ...
>
> -int netpoll_setup(struct netpoll *np)
> +int __netpoll_setup(struct netpoll *np)
> {
> - struct net_device *ndev = NULL;
> - struct in_device *in_dev;
> + struct net_device *ndev = np->dev;
> struct netpoll_info *npinfo;
> const struct net_device_ops *ops;
> unsigned long flags;
> int err;
>
> + if ((ndev->priv_flags & IFF_DISABLE_NETPOLL) ||
> + !ndev->netdev_ops->ndo_poll_controller) {
> + printk(KERN_ERR "%s: %s doesn't support polling, aborting.\n",
> + np->name, np->dev_name);
> + err = -ENOTSUPP;
> + goto out;
> + }
> +
> + if (!ndev->npinfo) {
> + npinfo = kmalloc(sizeof(*npinfo), GFP_KERNEL);
> + if (!npinfo) {
> + err = -ENOMEM;
> + goto out;
> + }
> +
> + npinfo->rx_flags = 0;
> + INIT_LIST_HEAD(&npinfo->rx_np);
> +
> + spin_lock_init(&npinfo->rx_lock);
> + skb_queue_head_init(&npinfo->arp_tx);
> + skb_queue_head_init(&npinfo->txq);
> + INIT_DELAYED_WORK(&npinfo->tx_work, queue_process);
> +
> + atomic_set(&npinfo->refcnt, 1);
> +
> + ops = np->dev->netdev_ops;
> + if (ops->ndo_netpoll_setup) {
> + err = ops->ndo_netpoll_setup(ndev, npinfo);
> + if (err)
> + goto free_npinfo;
> + }
> + } else {
> + npinfo = ndev->npinfo;
> + atomic_inc(&npinfo->refcnt);
> + }
> +
> + npinfo->netpoll = np;
> +
> + if (np->rx_hook) {
> + spin_lock_irqsave(&npinfo->rx_lock, flags);
> + npinfo->rx_flags |= NETPOLL_RX_ENABLED;
> + list_add_tail(&np->rx, &npinfo->rx_np);
> + spin_unlock_irqrestore(&npinfo->rx_lock, flags);
> + }
> +
> + /* last thing to do is link it to the net device structure */
> + rcu_assign_pointer(ndev->npinfo, npinfo);
> + rtnl_unlock();
And there it is, an unbalanced rtnl_unlock().
This stupid little thing took me over a day's work to find - it's just
been awful.
The user-visible symptom was that a bug in the netpoll code causes the
machine to hang after loading ipv6 (!), because
addrconf_fixup_forwarding()'s rtnl_trylock() kept on failing, and the
restart_syscall() kept on getting restarted, so an initscripts procfs
write just kept banging its head against the excessively-unlocked
mutex.
The mutex code handles an excessively-unlocked mutex (mutex.count==2)
really badly. Some API functions say "its locked", others say "it
isn't", etc.
Maybe it's better with mutex debugging enabled - didn't try that.
Things get pretty user-unfriendly when there's a bug within the
netconsole code itself.
Enabling lockdep simply made the bug cure itself - I suspect the mutex
code's handling of mutexes is different if lockdep is enabled. That
would be pretty bad behaviour from the lockdep code.
I just removed the rtnl_unlock() - I couldn't see much in there which
needed rtnl_locking..
Dave, the fixup should be folded into the original patch please -
otherwise we'll have a machine-hangs-up bisection hole which spans two
weeks work of commits.
^ permalink raw reply
* RE: [RFC PATCH v7 01/19] Add a new structure for skb buffer from external.
From: Dong, Eddie @ 2010-06-25 1:03 UTC (permalink / raw)
To: Herbert Xu
Cc: Xin, Xiaohui, Stephen Hemminger, netdev@vger.kernel.org,
kvm@vger.kernel.org, linux-kernel@vger.kernel.org, mst@redhat.com,
mingo@elte.hu, davem@davemloft.net, jdike@linux.intel.com,
Dong, Eddie, Dong, Eddie
In-Reply-To: <20100624100836.GA16220@gondor.apana.org.au>
Herbert Xu wrote:
> On Wed, Jun 23, 2010 at 06:05:41PM +0800, Dong, Eddie wrote:
>>
>> I mean once the frontend side driver post the buffers to the backend
>> driver, the backend driver will "immediately" use that buffers to
>> compose skb or gro_frags and post them to the assigned host NIC
>> driver as receive buffers. In that case, if the backend driver
>> recieves a packet from the NIC that requires to do copy, it may be
>> unable to find additional free guest buffer because all of them are
>> already used by the NIC driver. We have to reserve some guest
>> buffers for the possible copy even if the buffer address is not
>> identified by original skb :(
>
> OK I see what you mean. Can you tell me how does Xiaohui's
> previous patch-set deal with this problem?
>
> Thanks,
In current patch, each SKB for the assigned device (SRIOV VF or NIC or a complete queue pairs) uses the buffer from guest, so it eliminates copy completely in software and requires hardware to do so. If we can have an additonal place to store the buffer per skb (may cause copy later on), we can do copy later on or re-post the buffer to assigned NIC driver later on. But that may be not very clean either :(
BTW, some hardware may require certain level of packet copy such as for broadcast packets in very old VMDq device, which is not addressed in previous Xiaohui's patch yet. We may address this by implementing an additional virtqueue between guest and host for slow path (broadcast packets only here) with additinal complexity in FE/BE driver.
Thx, Eddie
^ permalink raw reply
* [PATCH net-next 5/6] cnic: Further unify kcq handling code.
From: Michael Chan @ 2010-06-25 0:58 UTC (permalink / raw)
To: davem; +Cc: netdev
In-Reply-To: <1277427522-17306-4-git-send-email-mchan@broadcom.com>
This eliminates some of the duplicate code for the various devices
that require the same basic kcq handling.
Signed-off-by: Michael Chan <mchan@broadcom.com>
---
drivers/net/cnic.c | 71 +++++++++++++++++++++++++---------------------------
1 files changed, 34 insertions(+), 37 deletions(-)
diff --git a/drivers/net/cnic.c b/drivers/net/cnic.c
index 8c41fe0..c0cc0d8 100644
--- a/drivers/net/cnic.c
+++ b/drivers/net/cnic.c
@@ -2223,16 +2223,12 @@ static void cnic_chk_pkt_rings(struct cnic_local *cp)
clear_bit(CNIC_LCL_FL_L2_WAIT, &cp->cnic_local_flags);
}
-static int cnic_service_bnx2(void *data, void *status_blk)
+static u32 cnic_service_bnx2_queues(struct cnic_dev *dev)
{
- struct cnic_dev *dev = data;
struct cnic_local *cp = dev->cnic_priv;
- u32 status_idx = *cp->kcq1.status_idx_ptr;
+ u32 status_idx = (u16) *cp->kcq1.status_idx_ptr;
int kcqe_cnt;
- if (unlikely(!test_bit(CNIC_F_CNIC_UP, &dev->flags)))
- return status_idx;
-
cp->kwq_con_idx = *cp->kwq_con_idx_ptr;
while ((kcqe_cnt = cnic_get_kcqes(dev, &cp->kcq1))) {
@@ -2242,7 +2238,7 @@ static int cnic_service_bnx2(void *data, void *status_blk)
/* Tell compiler that status_blk fields can change. */
barrier();
if (status_idx != *cp->kcq1.status_idx_ptr) {
- status_idx = *cp->kcq1.status_idx_ptr;
+ status_idx = (u16) *cp->kcq1.status_idx_ptr;
cp->kwq_con_idx = *cp->kwq_con_idx_ptr;
} else
break;
@@ -2251,37 +2247,29 @@ static int cnic_service_bnx2(void *data, void *status_blk)
CNIC_WR16(dev, cp->kcq1.io_addr, cp->kcq1.sw_prod_idx);
cnic_chk_pkt_rings(cp);
+
return status_idx;
}
-static void cnic_service_bnx2_msix(unsigned long data)
+static int cnic_service_bnx2(void *data, void *status_blk)
{
- struct cnic_dev *dev = (struct cnic_dev *) data;
+ struct cnic_dev *dev = data;
struct cnic_local *cp = dev->cnic_priv;
- struct status_block_msix *status_blk = cp->status_blk.bnx2;
u32 status_idx = *cp->kcq1.status_idx_ptr;
- int kcqe_cnt;
-
- cp->kwq_con_idx = status_blk->status_cmd_consumer_index;
- while ((kcqe_cnt = cnic_get_kcqes(dev, &cp->kcq1))) {
-
- service_kcqes(dev, kcqe_cnt);
+ if (unlikely(!test_bit(CNIC_F_CNIC_UP, &dev->flags)))
+ return status_idx;
- /* Tell compiler that status_blk fields can change. */
- barrier();
- if (status_idx != *cp->kcq1.status_idx_ptr) {
- status_idx = *cp->kcq1.status_idx_ptr;
- cp->kwq_con_idx = status_blk->status_cmd_consumer_index;
- } else
- break;
- }
+ return cnic_service_bnx2_queues(dev);
+}
- CNIC_WR16(dev, cp->kcq1.io_addr, cp->kcq1.sw_prod_idx);
+static void cnic_service_bnx2_msix(unsigned long data)
+{
+ struct cnic_dev *dev = (struct cnic_dev *) data;
+ struct cnic_local *cp = dev->cnic_priv;
- cnic_chk_pkt_rings(cp);
+ cp->last_status_idx = cnic_service_bnx2_queues(dev);
- cp->last_status_idx = status_idx;
CNIC_WR(dev, BNX2_PCICFG_INT_ACK_CMD, cp->int_num |
BNX2_PCICFG_INT_ACK_CMD_INDEX_VALID | cp->last_status_idx);
}
@@ -2338,27 +2326,35 @@ static void cnic_ack_bnx2x_msix(struct cnic_dev *dev)
IGU_INT_DISABLE, 0);
}
-static void cnic_service_bnx2x_bh(unsigned long data)
+static u32 cnic_service_bnx2x_kcq(struct cnic_dev *dev, struct kcq_info *info)
{
- struct cnic_dev *dev = (struct cnic_dev *) data;
- struct cnic_local *cp = dev->cnic_priv;
- u32 status_idx = *cp->kcq1.status_idx_ptr;
+ u32 last_status = *info->status_idx_ptr;
int kcqe_cnt;
- if (unlikely(!test_bit(CNIC_F_CNIC_UP, &dev->flags)))
- return;
-
- while ((kcqe_cnt = cnic_get_kcqes(dev, &cp->kcq1))) {
+ while ((kcqe_cnt = cnic_get_kcqes(dev, info))) {
service_kcqes(dev, kcqe_cnt);
/* Tell compiler that sblk fields can change. */
barrier();
- if (status_idx == *cp->kcq1.status_idx_ptr)
+ if (last_status == *info->status_idx_ptr)
break;
- status_idx = *cp->kcq1.status_idx_ptr;
+ last_status = *info->status_idx_ptr;
}
+ return last_status;
+}
+
+static void cnic_service_bnx2x_bh(unsigned long data)
+{
+ struct cnic_dev *dev = (struct cnic_dev *) data;
+ struct cnic_local *cp = dev->cnic_priv;
+ u32 status_idx;
+
+ if (unlikely(!test_bit(CNIC_F_CNIC_UP, &dev->flags)))
+ return;
+
+ status_idx = cnic_service_bnx2x_kcq(dev, &cp->kcq1);
CNIC_WR16(dev, cp->kcq1.io_addr, cp->kcq1.sw_prod_idx + MAX_KCQ_IDX);
cnic_ack_bnx2x_int(dev, cp->status_blk_num, CSTORM_ID,
@@ -3786,6 +3782,7 @@ static int cnic_start_bnx2_hw(struct cnic_dev *dev)
cp->kcq1.hw_prod_idx_ptr =
(u16 *) &msblk->status_completion_producer_index;
cp->kcq1.status_idx_ptr = (u16 *) &msblk->status_idx;
+ cp->kwq_con_idx_ptr = (u16 *) &msblk->status_cmd_consumer_index;
cp->int_num = sb_id << BNX2_PCICFG_INT_ACK_CMD_INT_NUM_SHIFT;
cnic_ctx_wr(dev, kwq_cid_addr, L5_KRNLQ_HOST_QIDX, sb);
cnic_ctx_wr(dev, kcq_cid_addr, L5_KRNLQ_HOST_QIDX, sb);
--
1.6.4.GIT
^ permalink raw reply related
* [PATCH net-next 4/6] cnic: Restructure kcq processing.
From: Michael Chan @ 2010-06-25 0:58 UTC (permalink / raw)
To: davem; +Cc: netdev
In-Reply-To: <1277427522-17306-3-git-send-email-mchan@broadcom.com>
By doing more work in the common function cnic_get_kcqes(), and
making full use of the kcq_info structure.
Signed-off-by: Michael Chan <mchan@broadcom.com>
---
drivers/net/cnic.c | 76 +++++++++++++++-------------------------------------
1 files changed, 22 insertions(+), 54 deletions(-)
diff --git a/drivers/net/cnic.c b/drivers/net/cnic.c
index 948b25b..8c41fe0 100644
--- a/drivers/net/cnic.c
+++ b/drivers/net/cnic.c
@@ -2141,18 +2141,20 @@ static u16 cnic_bnx2x_hw_idx(u16 idx)
return idx;
}
-static int cnic_get_kcqes(struct cnic_dev *dev, u16 hw_prod, u16 *sw_prod)
+static int cnic_get_kcqes(struct cnic_dev *dev, struct kcq_info *info)
{
struct cnic_local *cp = dev->cnic_priv;
- u16 i, ri, last;
+ u16 i, ri, hw_prod, last;
struct kcqe *kcqe;
int kcqe_cnt = 0, last_cnt = 0;
- i = ri = last = *sw_prod;
+ i = ri = last = info->sw_prod_idx;
ri &= MAX_KCQ_IDX;
+ hw_prod = *info->hw_prod_idx_ptr;
+ hw_prod = cp->hw_idx(hw_prod);
while ((i != hw_prod) && (kcqe_cnt < MAX_COMPLETED_KCQE)) {
- kcqe = &cp->kcq1.kcq[KCQ_PG(ri)][KCQ_IDX(ri)];
+ kcqe = &info->kcq[KCQ_PG(ri)][KCQ_IDX(ri)];
cp->completed_kcq[kcqe_cnt++] = kcqe;
i = cp->next_idx(i);
ri = i & MAX_KCQ_IDX;
@@ -2162,7 +2164,7 @@ static int cnic_get_kcqes(struct cnic_dev *dev, u16 hw_prod, u16 *sw_prod)
}
}
- *sw_prod = last;
+ info->sw_prod_idx = last;
return last_cnt;
}
@@ -2224,10 +2226,8 @@ static void cnic_chk_pkt_rings(struct cnic_local *cp)
static int cnic_service_bnx2(void *data, void *status_blk)
{
struct cnic_dev *dev = data;
- struct status_block *sblk = status_blk;
struct cnic_local *cp = dev->cnic_priv;
- u32 status_idx = sblk->status_idx;
- u16 hw_prod, sw_prod;
+ u32 status_idx = *cp->kcq1.status_idx_ptr;
int kcqe_cnt;
if (unlikely(!test_bit(CNIC_F_CNIC_UP, &dev->flags)))
@@ -2235,29 +2235,20 @@ static int cnic_service_bnx2(void *data, void *status_blk)
cp->kwq_con_idx = *cp->kwq_con_idx_ptr;
- hw_prod = sblk->status_completion_producer_index;
- sw_prod = cp->kcq1.sw_prod_idx;
- while (sw_prod != hw_prod) {
- kcqe_cnt = cnic_get_kcqes(dev, hw_prod, &sw_prod);
- if (kcqe_cnt == 0)
- goto done;
+ while ((kcqe_cnt = cnic_get_kcqes(dev, &cp->kcq1))) {
service_kcqes(dev, kcqe_cnt);
/* Tell compiler that status_blk fields can change. */
barrier();
- if (status_idx != sblk->status_idx) {
- status_idx = sblk->status_idx;
+ if (status_idx != *cp->kcq1.status_idx_ptr) {
+ status_idx = *cp->kcq1.status_idx_ptr;
cp->kwq_con_idx = *cp->kwq_con_idx_ptr;
- hw_prod = sblk->status_completion_producer_index;
} else
break;
}
-done:
- CNIC_WR16(dev, cp->kcq1.io_addr, sw_prod);
-
- cp->kcq1.sw_prod_idx = sw_prod;
+ CNIC_WR16(dev, cp->kcq1.io_addr, cp->kcq1.sw_prod_idx);
cnic_chk_pkt_rings(cp);
return status_idx;
@@ -2268,34 +2259,25 @@ static void cnic_service_bnx2_msix(unsigned long data)
struct cnic_dev *dev = (struct cnic_dev *) data;
struct cnic_local *cp = dev->cnic_priv;
struct status_block_msix *status_blk = cp->status_blk.bnx2;
- u32 status_idx = status_blk->status_idx;
- u16 hw_prod, sw_prod;
+ u32 status_idx = *cp->kcq1.status_idx_ptr;
int kcqe_cnt;
cp->kwq_con_idx = status_blk->status_cmd_consumer_index;
- hw_prod = status_blk->status_completion_producer_index;
- sw_prod = cp->kcq1.sw_prod_idx;
- while (sw_prod != hw_prod) {
- kcqe_cnt = cnic_get_kcqes(dev, hw_prod, &sw_prod);
- if (kcqe_cnt == 0)
- goto done;
+ while ((kcqe_cnt = cnic_get_kcqes(dev, &cp->kcq1))) {
service_kcqes(dev, kcqe_cnt);
/* Tell compiler that status_blk fields can change. */
barrier();
- if (status_idx != status_blk->status_idx) {
- status_idx = status_blk->status_idx;
+ if (status_idx != *cp->kcq1.status_idx_ptr) {
+ status_idx = *cp->kcq1.status_idx_ptr;
cp->kwq_con_idx = status_blk->status_cmd_consumer_index;
- hw_prod = status_blk->status_completion_producer_index;
} else
break;
}
-done:
- CNIC_WR16(dev, cp->kcq1.io_addr, sw_prod);
- cp->kcq1.sw_prod_idx = sw_prod;
+ CNIC_WR16(dev, cp->kcq1.io_addr, cp->kcq1.sw_prod_idx);
cnic_chk_pkt_rings(cp);
@@ -2360,41 +2342,27 @@ static void cnic_service_bnx2x_bh(unsigned long data)
{
struct cnic_dev *dev = (struct cnic_dev *) data;
struct cnic_local *cp = dev->cnic_priv;
- u16 hw_prod, sw_prod;
- struct cstorm_status_block_c *sblk =
- &cp->status_blk.bnx2x->c_status_block;
- u32 status_idx = sblk->status_block_index;
+ u32 status_idx = *cp->kcq1.status_idx_ptr;
int kcqe_cnt;
if (unlikely(!test_bit(CNIC_F_CNIC_UP, &dev->flags)))
return;
- hw_prod = sblk->index_values[HC_INDEX_C_ISCSI_EQ_CONS];
- hw_prod = cp->hw_idx(hw_prod);
- sw_prod = cp->kcq1.sw_prod_idx;
- while (sw_prod != hw_prod) {
- kcqe_cnt = cnic_get_kcqes(dev, hw_prod, &sw_prod);
- if (kcqe_cnt == 0)
- goto done;
+ while ((kcqe_cnt = cnic_get_kcqes(dev, &cp->kcq1))) {
service_kcqes(dev, kcqe_cnt);
/* Tell compiler that sblk fields can change. */
barrier();
- if (status_idx == sblk->status_block_index)
+ if (status_idx == *cp->kcq1.status_idx_ptr)
break;
- status_idx = sblk->status_block_index;
- hw_prod = sblk->index_values[HC_INDEX_C_ISCSI_EQ_CONS];
- hw_prod = cp->hw_idx(hw_prod);
+ status_idx = *cp->kcq1.status_idx_ptr;
}
-done:
- CNIC_WR16(dev, cp->kcq1.io_addr, sw_prod + MAX_KCQ_IDX);
+ CNIC_WR16(dev, cp->kcq1.io_addr, cp->kcq1.sw_prod_idx + MAX_KCQ_IDX);
cnic_ack_bnx2x_int(dev, cp->status_blk_num, CSTORM_ID,
status_idx, IGU_INT_ENABLE, 1);
-
- cp->kcq1.sw_prod_idx = sw_prod;
}
static int cnic_service_bnx2x(void *data, void *status_blk)
--
1.6.4.GIT
^ permalink raw reply related
* [PATCH net-next 1/6] cnic: Fine-tune CID memory space calculation.
From: Michael Chan @ 2010-06-25 0:58 UTC (permalink / raw)
To: davem; +Cc: netdev
The current code makes assumptions about the CID (context ID) memory
space and starting CID that may not be always correct when firmware
changes. In particular, BNX2_ISCSI_START_CID may not always be fixed.
We now calculate cp->max_cid_space and cp->iscsi_start_cid dynamically
instead of using fixed constants. The unused cp->max_iscsi_conn is also
eliminated.
Signed-off-by: Michael Chan <mchan@broadcom.com>
---
drivers/net/cnic.c | 32 ++++++++++++++++++--------------
drivers/net/cnic.h | 4 +++-
2 files changed, 21 insertions(+), 15 deletions(-)
diff --git a/drivers/net/cnic.c b/drivers/net/cnic.c
index e5539f0..23395fa 100644
--- a/drivers/net/cnic.c
+++ b/drivers/net/cnic.c
@@ -257,7 +257,7 @@ static int cnic_get_l5_cid(struct cnic_local *cp, u32 cid, u32 *l5_cid)
{
u32 i;
- for (i = 0; i < MAX_ISCSI_TBL_SZ; i++) {
+ for (i = 0; i < cp->max_cid_space; i++) {
if (cp->ctx_tbl[i].cid == cid) {
*l5_cid = i;
return 0;
@@ -981,17 +981,10 @@ error:
static int cnic_alloc_bnx2x_context(struct cnic_dev *dev)
{
struct cnic_local *cp = dev->cnic_priv;
- struct cnic_eth_dev *ethdev = cp->ethdev;
int ctx_blk_size = cp->ethdev->ctx_blk_size;
- int total_mem, blks, i, cid_space;
-
- if (BNX2X_ISCSI_START_CID < ethdev->starting_cid)
- return -EINVAL;
-
- cid_space = MAX_ISCSI_TBL_SZ +
- (BNX2X_ISCSI_START_CID - ethdev->starting_cid);
+ int total_mem, blks, i;
- total_mem = BNX2X_CONTEXT_MEM_SIZE * cid_space;
+ total_mem = BNX2X_CONTEXT_MEM_SIZE * cp->max_cid_space;
blks = total_mem / ctx_blk_size;
if (total_mem % ctx_blk_size)
blks++;
@@ -1035,16 +1028,27 @@ static int cnic_alloc_bnx2x_context(struct cnic_dev *dev)
static int cnic_alloc_bnx2x_resc(struct cnic_dev *dev)
{
struct cnic_local *cp = dev->cnic_priv;
+ struct cnic_eth_dev *ethdev = cp->ethdev;
+ u32 start_cid = ethdev->starting_cid;
int i, j, n, ret, pages;
struct cnic_dma *kwq_16_dma = &cp->kwq_16_data_info;
+ cp->max_cid_space = MAX_ISCSI_TBL_SZ;
+ cp->iscsi_start_cid = start_cid;
+ if (start_cid < BNX2X_ISCSI_START_CID) {
+ u32 delta = BNX2X_ISCSI_START_CID - start_cid;
+
+ cp->iscsi_start_cid = BNX2X_ISCSI_START_CID;
+ cp->max_cid_space += delta;
+ }
+
cp->iscsi_tbl = kzalloc(sizeof(struct cnic_iscsi) * MAX_ISCSI_TBL_SZ,
GFP_KERNEL);
if (!cp->iscsi_tbl)
goto error;
cp->ctx_tbl = kzalloc(sizeof(struct cnic_context) *
- MAX_CNIC_L5_CONTEXT, GFP_KERNEL);
+ cp->max_cid_space, GFP_KERNEL);
if (!cp->ctx_tbl)
goto error;
@@ -1053,7 +1057,7 @@ static int cnic_alloc_bnx2x_resc(struct cnic_dev *dev)
cp->ctx_tbl[i].ulp_proto_id = CNIC_ULP_ISCSI;
}
- pages = PAGE_ALIGN(MAX_CNIC_L5_CONTEXT * CNIC_KWQ16_DATA_SIZE) /
+ pages = PAGE_ALIGN(cp->max_cid_space * CNIC_KWQ16_DATA_SIZE) /
PAGE_SIZE;
ret = cnic_alloc_dma(dev, kwq_16_dma, pages, 0);
@@ -1061,7 +1065,7 @@ static int cnic_alloc_bnx2x_resc(struct cnic_dev *dev)
return -ENOMEM;
n = PAGE_SIZE / CNIC_KWQ16_DATA_SIZE;
- for (i = 0, j = 0; i < MAX_ISCSI_TBL_SZ; i++) {
+ for (i = 0, j = 0; i < cp->max_cid_space; i++) {
long off = CNIC_KWQ16_DATA_SIZE * (i % n);
cp->ctx_tbl[i].kwqe_data = kwq_16_dma->pg_arr[j] + off;
@@ -4116,7 +4120,7 @@ static int cnic_start_bnx2x_hw(struct cnic_dev *dev)
u8 sb_id = cp->status_blk_num;
ret = cnic_init_id_tbl(&cp->cid_tbl, MAX_ISCSI_TBL_SZ,
- BNX2X_ISCSI_START_CID);
+ cp->iscsi_start_cid);
if (ret)
return -ENOMEM;
diff --git a/drivers/net/cnic.h b/drivers/net/cnic.h
index 08b1235..b7e2f7f 100644
--- a/drivers/net/cnic.h
+++ b/drivers/net/cnic.h
@@ -248,8 +248,10 @@ struct cnic_local {
struct cnic_iscsi *iscsi_tbl;
struct cnic_context *ctx_tbl;
struct cnic_id_tbl cid_tbl;
- int max_iscsi_conn;
atomic_t iscsi_conn;
+ u32 iscsi_start_cid;
+
+ u32 max_cid_space;
/* per connection parameters */
int num_iscsi_tasks;
--
1.6.4.GIT
^ permalink raw reply related
* [PATCH net-next 6/6] cnic: Update version to 2.1.3.
From: Michael Chan @ 2010-06-25 0:58 UTC (permalink / raw)
To: davem; +Cc: netdev
In-Reply-To: <1277427522-17306-5-git-send-email-mchan@broadcom.com>
---
drivers/net/cnic_if.h | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/cnic_if.h b/drivers/net/cnic_if.h
index 0c55177..344c842 100644
--- a/drivers/net/cnic_if.h
+++ b/drivers/net/cnic_if.h
@@ -12,8 +12,8 @@
#ifndef CNIC_IF_H
#define CNIC_IF_H
-#define CNIC_MODULE_VERSION "2.1.2"
-#define CNIC_MODULE_RELDATE "May 26, 2010"
+#define CNIC_MODULE_VERSION "2.1.3"
+#define CNIC_MODULE_RELDATE "June 24, 2010"
#define CNIC_ULP_RDMA 0
#define CNIC_ULP_ISCSI 1
--
1.6.4.GIT
^ permalink raw reply related
* [PATCH net-next 3/6] cnic: Unify kcq allocation for all devices.
From: Michael Chan @ 2010-06-25 0:58 UTC (permalink / raw)
To: davem; +Cc: netdev
In-Reply-To: <1277427522-17306-2-git-send-email-mchan@broadcom.com>
By creating a common data stucture kcq_info for all devices, the kcq
(kernel completion queue) for all devices can be allocated by common
code.
Signed-off-by: Michael Chan <mchan@broadcom.com>
---
drivers/net/cnic.c | 143 +++++++++++++++++++++++++++++++--------------------
drivers/net/cnic.h | 19 ++++---
2 files changed, 98 insertions(+), 64 deletions(-)
diff --git a/drivers/net/cnic.c b/drivers/net/cnic.c
index bd92a24..948b25b 100644
--- a/drivers/net/cnic.c
+++ b/drivers/net/cnic.c
@@ -804,7 +804,7 @@ static void cnic_free_resc(struct cnic_dev *dev)
cnic_free_dma(dev, &cp->conn_buf_info);
cnic_free_dma(dev, &cp->kwq_info);
cnic_free_dma(dev, &cp->kwq_16_data_info);
- cnic_free_dma(dev, &cp->kcq_info);
+ cnic_free_dma(dev, &cp->kcq1.dma);
kfree(cp->iscsi_tbl);
cp->iscsi_tbl = NULL;
kfree(cp->ctx_tbl);
@@ -863,6 +863,37 @@ static int cnic_alloc_context(struct cnic_dev *dev)
return 0;
}
+static int cnic_alloc_kcq(struct cnic_dev *dev, struct kcq_info *info)
+{
+ int err, i, is_bnx2 = 0;
+ struct kcqe **kcq;
+
+ if (test_bit(CNIC_F_BNX2_CLASS, &dev->flags))
+ is_bnx2 = 1;
+
+ err = cnic_alloc_dma(dev, &info->dma, KCQ_PAGE_CNT, is_bnx2);
+ if (err)
+ return err;
+
+ kcq = (struct kcqe **) info->dma.pg_arr;
+ info->kcq = kcq;
+
+ if (is_bnx2)
+ return 0;
+
+ for (i = 0; i < KCQ_PAGE_CNT; i++) {
+ struct bnx2x_bd_chain_next *next =
+ (struct bnx2x_bd_chain_next *) &kcq[i][MAX_KCQE_CNT];
+ int j = i + 1;
+
+ if (j >= KCQ_PAGE_CNT)
+ j = 0;
+ next->addr_hi = (u64) info->dma.pg_map_arr[j] >> 32;
+ next->addr_lo = info->dma.pg_map_arr[j] & 0xffffffff;
+ }
+ return 0;
+}
+
static int cnic_alloc_l2_rings(struct cnic_dev *dev, int pages)
{
struct cnic_local *cp = dev->cnic_priv;
@@ -954,10 +985,9 @@ static int cnic_alloc_bnx2_resc(struct cnic_dev *dev)
goto error;
cp->kwq = (struct kwqe **) cp->kwq_info.pg_arr;
- ret = cnic_alloc_dma(dev, &cp->kcq_info, KCQ_PAGE_CNT, 1);
+ ret = cnic_alloc_kcq(dev, &cp->kcq1);
if (ret)
goto error;
- cp->kcq = (struct kcqe **) cp->kcq_info.pg_arr;
ret = cnic_alloc_context(dev);
if (ret)
@@ -1076,22 +1106,9 @@ static int cnic_alloc_bnx2x_resc(struct cnic_dev *dev)
j++;
}
- ret = cnic_alloc_dma(dev, &cp->kcq_info, KCQ_PAGE_CNT, 0);
+ ret = cnic_alloc_kcq(dev, &cp->kcq1);
if (ret)
goto error;
- cp->kcq = (struct kcqe **) cp->kcq_info.pg_arr;
-
- for (i = 0; i < KCQ_PAGE_CNT; i++) {
- struct bnx2x_bd_chain_next *next =
- (struct bnx2x_bd_chain_next *)
- &cp->kcq[i][MAX_KCQE_CNT];
- int j = i + 1;
-
- if (j >= KCQ_PAGE_CNT)
- j = 0;
- next->addr_hi = (u64) cp->kcq_info.pg_map_arr[j] >> 32;
- next->addr_lo = cp->kcq_info.pg_map_arr[j] & 0xffffffff;
- }
pages = PAGE_ALIGN(BNX2X_ISCSI_NUM_CONNECTIONS *
BNX2X_ISCSI_CONN_BUF_SIZE) / PAGE_SIZE;
@@ -2135,7 +2152,7 @@ static int cnic_get_kcqes(struct cnic_dev *dev, u16 hw_prod, u16 *sw_prod)
ri &= MAX_KCQ_IDX;
while ((i != hw_prod) && (kcqe_cnt < MAX_COMPLETED_KCQE)) {
- kcqe = &cp->kcq[KCQ_PG(ri)][KCQ_IDX(ri)];
+ kcqe = &cp->kcq1.kcq[KCQ_PG(ri)][KCQ_IDX(ri)];
cp->completed_kcq[kcqe_cnt++] = kcqe;
i = cp->next_idx(i);
ri = i & MAX_KCQ_IDX;
@@ -2219,7 +2236,7 @@ static int cnic_service_bnx2(void *data, void *status_blk)
cp->kwq_con_idx = *cp->kwq_con_idx_ptr;
hw_prod = sblk->status_completion_producer_index;
- sw_prod = cp->kcq_prod_idx;
+ sw_prod = cp->kcq1.sw_prod_idx;
while (sw_prod != hw_prod) {
kcqe_cnt = cnic_get_kcqes(dev, hw_prod, &sw_prod);
if (kcqe_cnt == 0)
@@ -2238,9 +2255,9 @@ static int cnic_service_bnx2(void *data, void *status_blk)
}
done:
- CNIC_WR16(dev, cp->kcq_io_addr, sw_prod);
+ CNIC_WR16(dev, cp->kcq1.io_addr, sw_prod);
- cp->kcq_prod_idx = sw_prod;
+ cp->kcq1.sw_prod_idx = sw_prod;
cnic_chk_pkt_rings(cp);
return status_idx;
@@ -2258,7 +2275,7 @@ static void cnic_service_bnx2_msix(unsigned long data)
cp->kwq_con_idx = status_blk->status_cmd_consumer_index;
hw_prod = status_blk->status_completion_producer_index;
- sw_prod = cp->kcq_prod_idx;
+ sw_prod = cp->kcq1.sw_prod_idx;
while (sw_prod != hw_prod) {
kcqe_cnt = cnic_get_kcqes(dev, hw_prod, &sw_prod);
if (kcqe_cnt == 0)
@@ -2277,8 +2294,8 @@ static void cnic_service_bnx2_msix(unsigned long data)
}
done:
- CNIC_WR16(dev, cp->kcq_io_addr, sw_prod);
- cp->kcq_prod_idx = sw_prod;
+ CNIC_WR16(dev, cp->kcq1.io_addr, sw_prod);
+ cp->kcq1.sw_prod_idx = sw_prod;
cnic_chk_pkt_rings(cp);
@@ -2290,11 +2307,11 @@ done:
static void cnic_doirq(struct cnic_dev *dev)
{
struct cnic_local *cp = dev->cnic_priv;
- u16 prod = cp->kcq_prod_idx & MAX_KCQ_IDX;
+ u16 prod = cp->kcq1.sw_prod_idx & MAX_KCQ_IDX;
if (likely(test_bit(CNIC_F_CNIC_UP, &dev->flags))) {
prefetch(cp->status_blk.gen);
- prefetch(&cp->kcq[KCQ_PG(prod)][KCQ_IDX(prod)]);
+ prefetch(&cp->kcq1.kcq[KCQ_PG(prod)][KCQ_IDX(prod)]);
tasklet_schedule(&cp->cnic_irq_task);
}
@@ -2354,7 +2371,7 @@ static void cnic_service_bnx2x_bh(unsigned long data)
hw_prod = sblk->index_values[HC_INDEX_C_ISCSI_EQ_CONS];
hw_prod = cp->hw_idx(hw_prod);
- sw_prod = cp->kcq_prod_idx;
+ sw_prod = cp->kcq1.sw_prod_idx;
while (sw_prod != hw_prod) {
kcqe_cnt = cnic_get_kcqes(dev, hw_prod, &sw_prod);
if (kcqe_cnt == 0)
@@ -2373,11 +2390,11 @@ static void cnic_service_bnx2x_bh(unsigned long data)
}
done:
- CNIC_WR16(dev, cp->kcq_io_addr, sw_prod + MAX_KCQ_IDX);
+ CNIC_WR16(dev, cp->kcq1.io_addr, sw_prod + MAX_KCQ_IDX);
cnic_ack_bnx2x_int(dev, cp->status_blk_num, CSTORM_ID,
status_idx, IGU_INT_ENABLE, 1);
- cp->kcq_prod_idx = sw_prod;
+ cp->kcq1.sw_prod_idx = sw_prod;
}
static int cnic_service_bnx2x(void *data, void *status_blk)
@@ -3711,7 +3728,7 @@ static int cnic_start_bnx2_hw(struct cnic_dev *dev)
struct cnic_local *cp = dev->cnic_priv;
struct cnic_eth_dev *ethdev = cp->ethdev;
struct status_block *sblk = cp->status_blk.gen;
- u32 val;
+ u32 val, kcq_cid_addr, kwq_cid_addr;
int err;
cnic_set_bnx2_mac(dev);
@@ -3736,7 +3753,7 @@ static int cnic_start_bnx2_hw(struct cnic_dev *dev)
cnic_init_context(dev, KWQ_CID);
cnic_init_context(dev, KCQ_CID);
- cp->kwq_cid_addr = GET_CID_ADDR(KWQ_CID);
+ kwq_cid_addr = GET_CID_ADDR(KWQ_CID);
cp->kwq_io_addr = MB_GET_CID_ADDR(KWQ_CID) + L5_KRNLQ_HOST_QIDX;
cp->max_kwq_idx = MAX_KWQ_IDX;
@@ -3752,50 +3769,58 @@ static int cnic_start_bnx2_hw(struct cnic_dev *dev)
/* Initialize the kernel work queue context. */
val = KRNLQ_TYPE_TYPE_KRNLQ | KRNLQ_SIZE_TYPE_SIZE |
(BCM_PAGE_BITS - 8) | KRNLQ_FLAGS_QE_SELF_SEQ;
- cnic_ctx_wr(dev, cp->kwq_cid_addr, L5_KRNLQ_TYPE, val);
+ cnic_ctx_wr(dev, kwq_cid_addr, L5_KRNLQ_TYPE, val);
val = (BCM_PAGE_SIZE / sizeof(struct kwqe) - 1) << 16;
- cnic_ctx_wr(dev, cp->kwq_cid_addr, L5_KRNLQ_QE_SELF_SEQ_MAX, val);
+ cnic_ctx_wr(dev, kwq_cid_addr, L5_KRNLQ_QE_SELF_SEQ_MAX, val);
val = ((BCM_PAGE_SIZE / sizeof(struct kwqe)) << 16) | KWQ_PAGE_CNT;
- cnic_ctx_wr(dev, cp->kwq_cid_addr, L5_KRNLQ_PGTBL_NPAGES, val);
+ cnic_ctx_wr(dev, kwq_cid_addr, L5_KRNLQ_PGTBL_NPAGES, val);
val = (u32) ((u64) cp->kwq_info.pgtbl_map >> 32);
- cnic_ctx_wr(dev, cp->kwq_cid_addr, L5_KRNLQ_PGTBL_HADDR_HI, val);
+ cnic_ctx_wr(dev, kwq_cid_addr, L5_KRNLQ_PGTBL_HADDR_HI, val);
val = (u32) cp->kwq_info.pgtbl_map;
- cnic_ctx_wr(dev, cp->kwq_cid_addr, L5_KRNLQ_PGTBL_HADDR_LO, val);
+ cnic_ctx_wr(dev, kwq_cid_addr, L5_KRNLQ_PGTBL_HADDR_LO, val);
+
+ kcq_cid_addr = GET_CID_ADDR(KCQ_CID);
+ cp->kcq1.io_addr = MB_GET_CID_ADDR(KCQ_CID) + L5_KRNLQ_HOST_QIDX;
- cp->kcq_cid_addr = GET_CID_ADDR(KCQ_CID);
- cp->kcq_io_addr = MB_GET_CID_ADDR(KCQ_CID) + L5_KRNLQ_HOST_QIDX;
+ cp->kcq1.sw_prod_idx = 0;
+ cp->kcq1.hw_prod_idx_ptr =
+ (u16 *) &sblk->status_completion_producer_index;
- cp->kcq_prod_idx = 0;
+ cp->kcq1.status_idx_ptr = (u16 *) &sblk->status_idx;
/* Initialize the kernel complete queue context. */
val = KRNLQ_TYPE_TYPE_KRNLQ | KRNLQ_SIZE_TYPE_SIZE |
(BCM_PAGE_BITS - 8) | KRNLQ_FLAGS_QE_SELF_SEQ;
- cnic_ctx_wr(dev, cp->kcq_cid_addr, L5_KRNLQ_TYPE, val);
+ cnic_ctx_wr(dev, kcq_cid_addr, L5_KRNLQ_TYPE, val);
val = (BCM_PAGE_SIZE / sizeof(struct kcqe) - 1) << 16;
- cnic_ctx_wr(dev, cp->kcq_cid_addr, L5_KRNLQ_QE_SELF_SEQ_MAX, val);
+ cnic_ctx_wr(dev, kcq_cid_addr, L5_KRNLQ_QE_SELF_SEQ_MAX, val);
val = ((BCM_PAGE_SIZE / sizeof(struct kcqe)) << 16) | KCQ_PAGE_CNT;
- cnic_ctx_wr(dev, cp->kcq_cid_addr, L5_KRNLQ_PGTBL_NPAGES, val);
+ cnic_ctx_wr(dev, kcq_cid_addr, L5_KRNLQ_PGTBL_NPAGES, val);
- val = (u32) ((u64) cp->kcq_info.pgtbl_map >> 32);
- cnic_ctx_wr(dev, cp->kcq_cid_addr, L5_KRNLQ_PGTBL_HADDR_HI, val);
+ val = (u32) ((u64) cp->kcq1.dma.pgtbl_map >> 32);
+ cnic_ctx_wr(dev, kcq_cid_addr, L5_KRNLQ_PGTBL_HADDR_HI, val);
- val = (u32) cp->kcq_info.pgtbl_map;
- cnic_ctx_wr(dev, cp->kcq_cid_addr, L5_KRNLQ_PGTBL_HADDR_LO, val);
+ val = (u32) cp->kcq1.dma.pgtbl_map;
+ cnic_ctx_wr(dev, kcq_cid_addr, L5_KRNLQ_PGTBL_HADDR_LO, val);
cp->int_num = 0;
if (ethdev->drv_state & CNIC_DRV_STATE_USING_MSIX) {
+ struct status_block_msix *msblk = cp->status_blk.bnx2;
u32 sb_id = cp->status_blk_num;
u32 sb = BNX2_L2CTX_L5_STATUSB_NUM(sb_id);
+ cp->kcq1.hw_prod_idx_ptr =
+ (u16 *) &msblk->status_completion_producer_index;
+ cp->kcq1.status_idx_ptr = (u16 *) &msblk->status_idx;
cp->int_num = sb_id << BNX2_PCICFG_INT_ACK_CMD_INT_NUM_SHIFT;
- cnic_ctx_wr(dev, cp->kwq_cid_addr, L5_KRNLQ_HOST_QIDX, sb);
- cnic_ctx_wr(dev, cp->kcq_cid_addr, L5_KRNLQ_HOST_QIDX, sb);
+ cnic_ctx_wr(dev, kwq_cid_addr, L5_KRNLQ_HOST_QIDX, sb);
+ cnic_ctx_wr(dev, kcq_cid_addr, L5_KRNLQ_HOST_QIDX, sb);
}
/* Enable Commnad Scheduler notification when we write to the
@@ -4132,28 +4157,34 @@ static int cnic_start_bnx2x_hw(struct cnic_dev *dev)
if (ret)
return -ENOMEM;
- cp->kcq_io_addr = BAR_CSTRORM_INTMEM +
+ cp->kcq1.io_addr = BAR_CSTRORM_INTMEM +
CSTORM_ISCSI_EQ_PROD_OFFSET(func, 0);
- cp->kcq_prod_idx = 0;
+ cp->kcq1.sw_prod_idx = 0;
+
+ cp->kcq1.hw_prod_idx_ptr =
+ &cp->status_blk.bnx2x->c_status_block.index_values[
+ HC_INDEX_C_ISCSI_EQ_CONS];
+ cp->kcq1.status_idx_ptr =
+ &cp->status_blk.bnx2x->c_status_block.status_block_index;
cnic_get_bnx2x_iscsi_info(dev);
/* Only 1 EQ */
- CNIC_WR16(dev, cp->kcq_io_addr, MAX_KCQ_IDX);
+ CNIC_WR16(dev, cp->kcq1.io_addr, MAX_KCQ_IDX);
CNIC_WR(dev, BAR_CSTRORM_INTMEM +
CSTORM_ISCSI_EQ_CONS_OFFSET(func, 0), 0);
CNIC_WR(dev, BAR_CSTRORM_INTMEM +
CSTORM_ISCSI_EQ_NEXT_PAGE_ADDR_OFFSET(func, 0),
- cp->kcq_info.pg_map_arr[1] & 0xffffffff);
+ cp->kcq1.dma.pg_map_arr[1] & 0xffffffff);
CNIC_WR(dev, BAR_CSTRORM_INTMEM +
CSTORM_ISCSI_EQ_NEXT_PAGE_ADDR_OFFSET(func, 0) + 4,
- (u64) cp->kcq_info.pg_map_arr[1] >> 32);
+ (u64) cp->kcq1.dma.pg_map_arr[1] >> 32);
CNIC_WR(dev, BAR_CSTRORM_INTMEM +
CSTORM_ISCSI_EQ_NEXT_EQE_ADDR_OFFSET(func, 0),
- cp->kcq_info.pg_map_arr[0] & 0xffffffff);
+ cp->kcq1.dma.pg_map_arr[0] & 0xffffffff);
CNIC_WR(dev, BAR_CSTRORM_INTMEM +
CSTORM_ISCSI_EQ_NEXT_EQE_ADDR_OFFSET(func, 0) + 4,
- (u64) cp->kcq_info.pg_map_arr[0] >> 32);
+ (u64) cp->kcq1.dma.pg_map_arr[0] >> 32);
CNIC_WR8(dev, BAR_CSTRORM_INTMEM +
CSTORM_ISCSI_EQ_NEXT_PAGE_ADDR_VALID_OFFSET(func, 0), 1);
CNIC_WR16(dev, BAR_CSTRORM_INTMEM +
@@ -4381,7 +4412,7 @@ static void cnic_stop_bnx2x_hw(struct cnic_dev *dev)
0);
CNIC_WR(dev, BAR_CSTRORM_INTMEM +
CSTORM_ISCSI_EQ_CONS_OFFSET(cp->func, 0), 0);
- CNIC_WR16(dev, cp->kcq_io_addr, 0);
+ CNIC_WR16(dev, cp->kcq1.io_addr, 0);
cnic_free_resc(dev);
}
diff --git a/drivers/net/cnic.h b/drivers/net/cnic.h
index b7e2f7f..275c361 100644
--- a/drivers/net/cnic.h
+++ b/drivers/net/cnic.h
@@ -169,6 +169,16 @@ struct cnic_context {
} proto;
};
+struct kcq_info {
+ struct cnic_dma dma;
+ struct kcqe **kcq;
+
+ u16 *hw_prod_idx_ptr;
+ u16 sw_prod_idx;
+ u16 *status_idx_ptr;
+ u32 io_addr;
+};
+
struct cnic_local {
spinlock_t cnic_ulp_lock;
@@ -202,9 +212,6 @@ struct cnic_local {
u16 rx_cons;
u16 tx_cons;
- u32 kwq_cid_addr;
- u32 kcq_cid_addr;
-
struct cnic_dma kwq_info;
struct kwqe **kwq;
@@ -218,11 +225,7 @@ struct cnic_local {
u16 *kwq_con_idx_ptr;
u16 kwq_con_idx;
- struct cnic_dma kcq_info;
- struct kcqe **kcq;
-
- u16 kcq_prod_idx;
- u32 kcq_io_addr;
+ struct kcq_info kcq1;
union {
void *gen;
--
1.6.4.GIT
^ permalink raw reply related
* [PATCH net-next 2/6] cnic: Unify IRQ code for all hardware types.
From: Michael Chan @ 2010-06-25 0:58 UTC (permalink / raw)
To: davem; +Cc: netdev
In-Reply-To: <1277427522-17306-1-git-send-email-mchan@broadcom.com>
By creating a common cnic_doirq().
Signed-off-by: Michael Chan <mchan@broadcom.com>
---
drivers/net/cnic.c | 33 ++++++++++++++++++++-------------
1 files changed, 20 insertions(+), 13 deletions(-)
diff --git a/drivers/net/cnic.c b/drivers/net/cnic.c
index 23395fa..bd92a24 100644
--- a/drivers/net/cnic.c
+++ b/drivers/net/cnic.c
@@ -2188,6 +2188,9 @@ static void cnic_chk_pkt_rings(struct cnic_local *cp)
u16 tx_cons = *cp->tx_cons_ptr;
int comp = 0;
+ if (!test_bit(CNIC_F_CNIC_UP, &cp->dev->flags))
+ return;
+
if (cp->tx_cons != tx_cons || cp->rx_cons != rx_cons) {
if (test_bit(CNIC_LCL_FL_L2_WAIT, &cp->cnic_local_flags))
comp = cnic_l2_completion(cp);
@@ -2284,20 +2287,28 @@ done:
BNX2_PCICFG_INT_ACK_CMD_INDEX_VALID | cp->last_status_idx);
}
+static void cnic_doirq(struct cnic_dev *dev)
+{
+ struct cnic_local *cp = dev->cnic_priv;
+ u16 prod = cp->kcq_prod_idx & MAX_KCQ_IDX;
+
+ if (likely(test_bit(CNIC_F_CNIC_UP, &dev->flags))) {
+ prefetch(cp->status_blk.gen);
+ prefetch(&cp->kcq[KCQ_PG(prod)][KCQ_IDX(prod)]);
+
+ tasklet_schedule(&cp->cnic_irq_task);
+ }
+}
+
static irqreturn_t cnic_irq(int irq, void *dev_instance)
{
struct cnic_dev *dev = dev_instance;
struct cnic_local *cp = dev->cnic_priv;
- u16 prod = cp->kcq_prod_idx & MAX_KCQ_IDX;
if (cp->ack_int)
cp->ack_int(dev);
- prefetch(cp->status_blk.gen);
- prefetch(&cp->kcq[KCQ_PG(prod)][KCQ_IDX(prod)]);
-
- if (likely(test_bit(CNIC_F_CNIC_UP, &dev->flags)))
- tasklet_schedule(&cp->cnic_irq_task);
+ cnic_doirq(dev);
return IRQ_HANDLED;
}
@@ -2373,15 +2384,11 @@ static int cnic_service_bnx2x(void *data, void *status_blk)
{
struct cnic_dev *dev = data;
struct cnic_local *cp = dev->cnic_priv;
- u16 prod = cp->kcq_prod_idx & MAX_KCQ_IDX;
- if (likely(test_bit(CNIC_F_CNIC_UP, &dev->flags))) {
- prefetch(cp->status_blk.bnx2x);
- prefetch(&cp->kcq[KCQ_PG(prod)][KCQ_IDX(prod)]);
+ if (!(cp->ethdev->drv_state & CNIC_DRV_STATE_USING_MSIX))
+ cnic_doirq(dev);
- tasklet_schedule(&cp->cnic_irq_task);
- cnic_chk_pkt_rings(cp);
- }
+ cnic_chk_pkt_rings(cp);
return 0;
}
--
1.6.4.GIT
^ permalink raw reply related
* Re: [RFC PATCH v2 1/5] irq: add tracepoint to softirq_raise
From: Koki Sanagi @ 2010-06-25 0:49 UTC (permalink / raw)
To: rostedt
Cc: netdev, davem, scott.a.mcmillan, kaneshige.kenji, izumi.taku,
linux-kernel
In-Reply-To: <1277406914.9181.67.camel@gandalf.stny.rr.com>
Hi Steven,
Sorry, I submitted other patches to only netdev.
I'll resubmit soon all patches to linux-kernel.
(2010/06/25 4:15), Steven Rostedt wrote:
> Hi Koki,
>
> Your subject says 1/5 but I do not see any other patches.
>
>
> On Thu, 2010-06-24 at 17:16 +0900, Koki Sanagi wrote:
>> This patch adds a tracepoint to raising of softirq.
>> This is useful if you want to detect which hard interrupt raise softirq
>> and lets you know a time between raising softirq and performing softirq.
>> Combinating with other tracepoint, it lets us know a process of packets
>> (See patch 0/5).
>>
>> <idle>-0 [001] 241229.957184: softirq_raise: vec=3 [action=NET_RX]
>> <idle>-0 [000] 241229.993399: softirq_raise: vec=1 [action=TIMER]
>> <idle>-0 [000] 241229.993400: softirq_raise: vec=9 [action=RCU]
>>
>> This is a same patch Lai Jiangshan submitted.
>> http://marc.info/?l=linux-kernel&m=126026122728732&w=2
>>
>> Signed-off-by: Koki Sanagi <sanagi.koki@jp.fujitsu.com>
>> ---
>> include/linux/interrupt.h | 8 +++++++-
>> include/trace/events/irq.h | 34 +++++++++++++++++++++++++++++++---
>> 2 files changed, 38 insertions(+), 4 deletions(-)
>>
>> diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h
>> index c233113..1cb5726 100644
>> --- a/include/linux/interrupt.h
>> +++ b/include/linux/interrupt.h
>> @@ -18,6 +18,7 @@
>> #include <asm/atomic.h>
>> #include <asm/ptrace.h>
>> #include <asm/system.h>
>> +#include <trace/events/irq.h>
>>
>> /*
>> * These correspond to the IORESOURCE_IRQ_* defines in
>> @@ -402,7 +403,12 @@ asmlinkage void do_softirq(void);
>> asmlinkage void __do_softirq(void);
>> extern void open_softirq(int nr, void (*action)(struct softirq_action *));
>> extern void softirq_init(void);
>> -#define __raise_softirq_irqoff(nr) do { or_softirq_pending(1UL << (nr)); } while (0)
>> +static inline void __raise_softirq_irqoff(unsigned int nr)
>> +{
>> + trace_softirq_raise(nr);
>> + or_softirq_pending(1UL << nr);
>> +}
>> +
>> extern void raise_softirq_irqoff(unsigned int nr);
>> extern void raise_softirq(unsigned int nr);
>> extern void wakeup_softirqd(void);
>> diff --git a/include/trace/events/irq.h b/include/trace/events/irq.h
>> index 0e4cfb6..7cb7435 100644
>> --- a/include/trace/events/irq.h
>> +++ b/include/trace/events/irq.h
>> @@ -5,7 +5,9 @@
>> #define _TRACE_IRQ_H
>>
>> #include <linux/tracepoint.h>
>> -#include <linux/interrupt.h>
>> +
>> +struct irqaction;
>> +struct softirq_action;
>>
>> #define softirq_name(sirq) { sirq##_SOFTIRQ, #sirq }
>> #define show_softirq_name(val) \
>> @@ -82,6 +84,32 @@ TRACE_EVENT(irq_handler_exit,
>> __entry->irq, __entry->ret ? "handled" : "unhandled")
>> );
>>
>> +/**
>> + * softirq_raise - called immediately when a softirq is raised
>> + * @nr: softirq vector number
>> + *
>> + * Tracepoint for tracing when softirq action is raised.
>> + * Also, when used in combination with the softirq_entry tracepoint
>> + * we can determine the softirq raise latency.
>> + */
>> +TRACE_EVENT(softirq_raise,
>> +
>> + TP_PROTO(unsigned int nr),
>> +
>> + TP_ARGS(nr),
>> +
>> + TP_STRUCT__entry(
>> + __field( unsigned int, vec )
>> + ),
>> +
>> + TP_fast_assign(
>> + __entry->vec = nr;
>> + ),
>> +
>> + TP_printk("vec=%d [action=%s]", __entry->vec,
>> + show_softirq_name(__entry->vec))
>
> Hmm, is there a way to reuse a DECLARE_EVENT_CLASS here?
>
>> +);
>> +
>> DECLARE_EVENT_CLASS(softirq,
>>
>> TP_PROTO(struct softirq_action *h, struct softirq_action *vec),
>> @@ -89,11 +117,11 @@ DECLARE_EVENT_CLASS(softirq,
>> TP_ARGS(h, vec),
>>
>> TP_STRUCT__entry(
>> - __field( int, vec )
>> + __field( unsigned int, vec )
>> ),
>>
>> TP_fast_assign(
>> - __entry->vec = (int)(h - vec);
>> + __entry->vec = (unsigned int)(h - vec);
>
> Just curious, did you see the original as a bug?
>
> -- Steve
>
>> ),
>>
>> TP_printk("vec=%d [action=%s]", __entry->vec,
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" 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 net-next-2.6 2/2] 3c59x: Use fine-grained locks for MII and windowed register access
From: Ben Hutchings @ 2010-06-25 0:45 UTC (permalink / raw)
To: Steffen Klassert; +Cc: David Miller, netdev, Chase Douglas, Arne Nordmark
In-Reply-To: <20100624140057.GJ5570@secunet.com>
[-- Attachment #1: Type: text/plain, Size: 1846 bytes --]
On Thu, 2010-06-24 at 16:00 +0200, Steffen Klassert wrote:
> On Thu, Jun 24, 2010 at 01:57:19PM +0100, Ben Hutchings wrote:
> > >
> > > This adds a lot of calls to spin_lock_irqsave/spin_unlock_irqrestore to many
> > > places where this is not necessary at all. For example during device probe and
> > > device open, window_read/window_write are called multiple times, each time
> > > disabling the interrupts. I'd suggest to have unlocked, locked and irqsave
> > > versions of window_read/window_write and use them in appropriate places.
> >
> > So what? These are not speed-critical. The fast-path functions do
> > acquire the lock just once.
> >
>
> The point is that we should not disable the interrupts if we don't need to
> do so. It is not speed critical for the 3c59x driver but disabling the
> interrupts should be avoided whenever possible. For example during device
> probe and device open we can't race against an interrupt handler because
> the device is not yet running.
>
> An example from vortex_probe1() is:
>
> for (i = 0; i < 6; i++)
> window_write8(vp, dev->dev_addr[i], 2, i);
>
> which expands to someting like:
>
> for (i = 0; i < 6; i++) {
> unsigned long flags;
> spin_lock_irqsave(&vp->window_lock, flags);
> window_set(vp, window);
> iowrite8(dev->dev_addr[i], vp->ioaddr + i);
> spin_unlock_irqrestore(&vp->window_lock, flags);
> return ret;
> }
[...]
I still fail to see why this matters.
The sfc driver which I look after in my day job also uses
spin_lock_irqsave() for each CSR update, when this could be avoided in
the initialisation path. None of the many customers using rt kernels
has ever complained about this. It's just not an important issue.
Ben.
--
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.
[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]
^ permalink raw reply
* Re: [PATCH net-next] MAINTAINERS: b43 updates
From: Joe Perches @ 2010-06-24 23:57 UTC (permalink / raw)
To: Larry Finger
Cc: John W. Linville, Stefano Brivio, linux-wireless, netdev,
linux-kernel, mb, zajec5
In-Reply-To: <4C23EFDE.3000702@lwfinger.net>
On Thu, 2010-06-24 at 18:53 -0500, Larry Finger wrote:
> I never said that b43 is unmaintained. It is definitely getting much
> more than "odd fixes".
If Stefano isn't maintaining it, who is?
^ permalink raw reply
* Re: [PATCH net-next] MAINTAINERS: b43 updates
From: Larry Finger @ 2010-06-24 23:53 UTC (permalink / raw)
To: Joe Perches
Cc: John W. Linville, Stefano Brivio, linux-wireless, netdev,
linux-kernel, mb, zajec5
In-Reply-To: <1277423102.1654.107.camel@Joe-Laptop.home>
On 06/24/2010 06:45 PM, Joe Perches wrote:
> On Thu, 2010-06-24 at 14:56 -0500, Larry Finger wrote:
>> The primary arbitrator for patches to the subtle parts of b43 is Michael
>> Buesch; however, he is no longer an official MAINTAINER. I can ACK some
>> things; however, any changes that are associated with my reverse
>> engineering of the Broadcom drivers are off limits.
>
> Add Stefano Brivio's name and email address to CREDITS
> Removed Stefano Brivio's from B43 MAINTAINERS
> Added the b43 development mailing lists
> Changed B43 status to "Odd Fixes"
>
> Signed-off-by: Joe Perches <joe@perches.com>
> ---
> CREDITS | 4 ++++
> MAINTAINERS | 5 +++--
> 2 files changed, 7 insertions(+), 2 deletions(-)
>
> diff --git a/CREDITS b/CREDITS
> index 72b4878..30215f0 100644
> --- a/CREDITS
> +++ b/CREDITS
> @@ -495,6 +495,10 @@ S: Kopmansg 2
> S: 411 13 Goteborg
> S: Sweden
>
> +N: Stefano Brivio
> +E: stefano.brivio@polimi.it
> +D: B43 wireless driver
> +
> N: Paul Bristow
> E: paul@paulbristow.net
> W: http://paulbristow.net/linux/idefloppy.html
> diff --git a/MAINTAINERS b/MAINTAINERS
> index 6d119c9..5bc296c 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -1225,16 +1225,17 @@ F: include/net/ax25.h
> F: net/ax25/
>
> B43 WIRELESS DRIVER
> -M: Stefano Brivio <stefano.brivio@polimi.it>
> L: linux-wireless@vger.kernel.org
> +L: b43-dev@lists.infradead.org
> W: http://linuxwireless.org/en/users/Drivers/b43
> -S: Maintained
> +S: Odd Fixes
> F: drivers/net/wireless/b43/
>
> B43LEGACY WIRELESS DRIVER
> M: Larry Finger <Larry.Finger@lwfinger.net>
> M: Stefano Brivio <stefano.brivio@polimi.it>
> L: linux-wireless@vger.kernel.org
> +L: b43-dev@lists.infradead.org
> W: http://linuxwireless.org/en/users/Drivers/b43
> S: Maintained
> F: drivers/net/wireless/b43legacy/
NACK. I never said that b43 is unmaintained. It is definitely getting much
more than "odd fixes".
Larry
^ permalink raw reply
* [PATCH] s2io: read rx_packets count from the hardware stats
From: Michal Schmidt @ 2010-06-24 23:32 UTC (permalink / raw)
To: netdev; +Cc: Ramkrishna Vepa, Sivakumar Subramani, Sreenivasa Honnur
Most of the statistics the s2io driver provides in /proc/net/dev
it reads directly from the hardware counters. For some reason it does
not do that for rx_packets. It counts rx_packets purely in software.
A customer reported a bug where in /proc/net/dev the 'multicast' counter
was increasing faster than 'packets' ( = rx_packets in the source code).
This confuses userspace, especially snmpd.
The hardware provides a counter for the total number of received
frames (RMAC_VLD_FRMS) which the driver can use for the rx_packets
statistic. By reading both statistics from the hardware it makes sure
that all multicast frames are included in the total.
The customer tested a patch like this (only modified for RHEL5) with
S2io Inc. Xframe II 10Gbps Ethernet (rev 02)
and it fixed the problem.
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
---
drivers/net/s2io.c | 11 ++++++-----
drivers/net/s2io.h | 1 -
2 files changed, 6 insertions(+), 6 deletions(-)
diff --git a/drivers/net/s2io.c b/drivers/net/s2io.c
index 668327c..eefc4b2 100644
--- a/drivers/net/s2io.c
+++ b/drivers/net/s2io.c
@@ -4919,6 +4919,10 @@ static struct net_device_stats *s2io_get_stats(struct net_device *dev)
sp->stats.tx_packets;
sp->stats.tx_packets = le32_to_cpu(stats->tmac_frms);
+ dev->stats.rx_packets += le32_to_cpu(stats->rmac_vld_frms) -
+ sp->stats.rx_packets;
+ sp->stats.rx_packets = le32_to_cpu(stats->rmac_vld_frms);
+
dev->stats.tx_errors += le32_to_cpu(stats->tmac_any_err_frms) -
sp->stats.tx_errors;
sp->stats.tx_errors = le32_to_cpu(stats->tmac_any_err_frms);
@@ -4935,12 +4939,11 @@ static struct net_device_stats *s2io_get_stats(struct net_device *dev)
sp->stats.rx_length_errors;
sp->stats.rx_length_errors = le64_to_cpu(stats->rmac_long_frms);
- /* collect per-ring rx_packets and rx_bytes */
- dev->stats.rx_packets = dev->stats.rx_bytes = 0;
+ /* collect per-ring rx_bytes */
+ dev->stats.rx_bytes = 0;
for (i = 0; i < config->rx_ring_num; i++) {
struct ring_info *ring = &mac_control->rings[i];
- dev->stats.rx_packets += ring->rx_packets;
dev->stats.rx_bytes += ring->rx_bytes;
}
@@ -7455,8 +7458,6 @@ static int rx_osm_handler(struct ring_info *ring_data, struct RxD_t * rxdp)
}
}
- /* Updating statistics */
- ring_data->rx_packets++;
rxdp->Host_Control = 0;
if (sp->rxd_mode == RXD_MODE_1) {
int len = RXD_GET_BUFFER0_SIZE_1(rxdp->Control_2);
diff --git a/drivers/net/s2io.h b/drivers/net/s2io.h
index 47c36e0..4da9ab8 100644
--- a/drivers/net/s2io.h
+++ b/drivers/net/s2io.h
@@ -747,7 +747,6 @@ struct ring_info {
struct buffAdd **ba;
/* per-Ring statistics */
- unsigned long rx_packets;
unsigned long rx_bytes;
} ____cacheline_aligned;
^ permalink raw reply related
* [PATCH net-next] MAINTAINERS: b43 updates
From: Joe Perches @ 2010-06-24 23:45 UTC (permalink / raw)
To: Larry Finger
Cc: John W. Linville, Stefano Brivio, linux-wireless, netdev,
linux-kernel, mb, zajec5
In-Reply-To: <4C23B877.4060209@lwfinger.net>
On Thu, 2010-06-24 at 14:56 -0500, Larry Finger wrote:
> The primary arbitrator for patches to the subtle parts of b43 is Michael
> Buesch; however, he is no longer an official MAINTAINER. I can ACK some
> things; however, any changes that are associated with my reverse
> engineering of the Broadcom drivers are off limits.
Add Stefano Brivio's name and email address to CREDITS
Removed Stefano Brivio's from B43 MAINTAINERS
Added the b43 development mailing lists
Changed B43 status to "Odd Fixes"
Signed-off-by: Joe Perches <joe@perches.com>
---
CREDITS | 4 ++++
MAINTAINERS | 5 +++--
2 files changed, 7 insertions(+), 2 deletions(-)
diff --git a/CREDITS b/CREDITS
index 72b4878..30215f0 100644
--- a/CREDITS
+++ b/CREDITS
@@ -495,6 +495,10 @@ S: Kopmansg 2
S: 411 13 Goteborg
S: Sweden
+N: Stefano Brivio
+E: stefano.brivio@polimi.it
+D: B43 wireless driver
+
N: Paul Bristow
E: paul@paulbristow.net
W: http://paulbristow.net/linux/idefloppy.html
diff --git a/MAINTAINERS b/MAINTAINERS
index 6d119c9..5bc296c 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -1225,16 +1225,17 @@ F: include/net/ax25.h
F: net/ax25/
B43 WIRELESS DRIVER
-M: Stefano Brivio <stefano.brivio@polimi.it>
L: linux-wireless@vger.kernel.org
+L: b43-dev@lists.infradead.org
W: http://linuxwireless.org/en/users/Drivers/b43
-S: Maintained
+S: Odd Fixes
F: drivers/net/wireless/b43/
B43LEGACY WIRELESS DRIVER
M: Larry Finger <Larry.Finger@lwfinger.net>
M: Stefano Brivio <stefano.brivio@polimi.it>
L: linux-wireless@vger.kernel.org
+L: b43-dev@lists.infradead.org
W: http://linuxwireless.org/en/users/Drivers/b43
S: Maintained
F: drivers/net/wireless/b43legacy/
^ permalink raw reply related
* Re: [PATCH 3/3] vhost: apply cpumask and cgroup to vhost pollers
From: Sridhar Samudrala @ 2010-06-24 22:45 UTC (permalink / raw)
To: Michael S. Tsirkin
Cc: Tejun Heo, Oleg Nesterov, netdev, lkml, kvm@vger.kernel.org,
Andrew Morton, Dmitri Vorobiev, Jiri Kosina, Thomas Gleixner,
Ingo Molnar, Andi Kleen
In-Reply-To: <20100624081135.GA937@redhat.com>
On Thu, 2010-06-24 at 11:11 +0300, Michael S. Tsirkin wrote:
> On Sun, May 30, 2010 at 10:25:01PM +0200, Tejun Heo wrote:
> > Apply the cpumask and cgroup of the initializing task to the created
> > vhost poller.
> >
> > Based on Sridhar Samudrala's patch.
> >
> > Cc: Michael S. Tsirkin <mst@redhat.com>
> > Cc: Sridhar Samudrala <samudrala.sridhar@gmail.com>
>
>
> I wanted to apply this, but modpost fails:
> ERROR: "sched_setaffinity" [drivers/vhost/vhost_net.ko] undefined!
> ERROR: "sched_getaffinity" [drivers/vhost/vhost_net.ko] undefined!
>
> Did you try building as a module?
In my original implementation, i had these calls in workqueue.c.
Now that these are moved to vhost.c which can be built as a module,
these symbols need to be exported.
The following patch fixes the build issue with vhost as a module.
Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
diff --git a/kernel/sched.c b/kernel/sched.c
index 3c2a54f..15a0c6f 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -4837,6 +4837,7 @@ out_put_task:
put_online_cpus();
return retval;
}
+EXPORT_SYMBOL_GPL(sched_setaffinity);
static int get_user_cpu_mask(unsigned long __user *user_mask_ptr,
unsigned len,
struct cpumask *new_mask)
@@ -4900,6 +4901,7 @@ out_unlock:
return retval;
}
+EXPORT_SYMBOL_GPL(sched_getaffinity);
/**
* sys_sched_getaffinity - get the cpu affinity of a process
> > ---
> > drivers/vhost/vhost.c | 36 +++++++++++++++++++++++++++++++-----
> > 1 file changed, 31 insertions(+), 5 deletions(-)
> >
> > Index: work/drivers/vhost/vhost.c
> > ===================================================================
> > --- work.orig/drivers/vhost/vhost.c
> > +++ work/drivers/vhost/vhost.c
> > @@ -23,6 +23,7 @@
> > #include <linux/highmem.h>
> > #include <linux/slab.h>
> > #include <linux/kthread.h>
> > +#include <linux/cgroup.h>
> >
> > #include <linux/net.h>
> > #include <linux/if_packet.h>
> > @@ -176,12 +177,30 @@ repeat:
> > long vhost_dev_init(struct vhost_dev *dev,
> > struct vhost_virtqueue *vqs, int nvqs)
> > {
> > - struct task_struct *poller;
> > - int i;
> > + struct task_struct *poller = NULL;
> > + cpumask_var_t mask;
> > + int i, ret = -ENOMEM;
> > +
> > + if (!alloc_cpumask_var(&mask, GFP_KERNEL))
> > + goto out;
> >
> > poller = kthread_create(vhost_poller, dev, "vhost-%d", current->pid);
> > - if (IS_ERR(poller))
> > - return PTR_ERR(poller);
> > + if (IS_ERR(poller)) {
> > + ret = PTR_ERR(poller);
> > + goto out;
> > + }
> > +
> > + ret = sched_getaffinity(current->pid, mask);
> > + if (ret)
> > + goto out;
> > +
> > + ret = sched_setaffinity(poller->pid, mask);
> > + if (ret)
> > + goto out;
> > +
> > + ret = cgroup_attach_task_current_cg(poller);
> > + if (ret)
> > + goto out;
> >
> > dev->vqs = vqs;
> > dev->nvqs = nvqs;
> > @@ -202,7 +221,14 @@ long vhost_dev_init(struct vhost_dev *de
> > vhost_poll_init(&dev->vqs[i].poll,
> > dev->vqs[i].handle_kick, POLLIN, dev);
> > }
> > - return 0;
> > +
> > + wake_up_process(poller); /* avoid contributing to loadavg */
> > + ret = 0;
> > +out:
> > + if (ret)
> > + kthread_stop(poller);
> > + free_cpumask_var(mask);
> > + return ret;
> > }
> >
> > /* Caller should have device mutex */
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" 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 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