* Re: [PATCH] ipv6: remove the unnecessary statement in find_match()
From: Duan Jiong @ 2013-10-31 11:09 UTC (permalink / raw)
To: David Miller; +Cc: netdev, hannes
In-Reply-To: <20131031084537.GA17210@order.stressinduktion.org>
于 2013年10月31日 16:45, Hannes Frederic Sowa 写道:
> On Thu, Oct 31, 2013 at 02:02:11PM +0800, Duan Jiong wrote:
>> 于 2013年10月31日 12:22, David Miller 写道:
>>> From: Hannes Frederic Sowa <hannes@stressinduktion.org>
>>> Date: Wed, 30 Oct 2013 22:11:57 +0100
>>>
>>>> On Wed, Oct 30, 2013 at 05:08:37PM -0400, David Miller wrote:
>>>>> From: Duan Jiong <duanj.fnst@cn.fujitsu.com>
>>>>> Date: Wed, 30 Oct 2013 15:39:26 +0800
>>>>>
>>>>>>
>>>>>> After reading the function rt6_check_neigh(), we can
>>>>>> know that the RT6_NUD_FAIL_SOFT can be returned only
>>>>>> when the IS_ENABLE(CONFIG_IPV6_ROUTER_PREF) is false.
>>>>>> so in function find_match(), there is no need to execute
>>>>>> the statement !IS_ENABLED(CONFIG_IPV6_ROUTER_PREF).
>>>>>>
>>>>>> Signed-off-by: Duan Jiong <duanj.fnst@cn.fujitsu.com>
>>>>>
>>>>> Applied to net-next, thanks.
>>>>>
>>>>> CONFIG_IPV6_ROUTER_PREF is another good candidate for Kconfig
>>>>> removal. I know we've had several bugs that only apply when
>>>>> this option is on vs. off. We're maintaining two different
>>>>> code paths, for really no good reason.
>>>>
>>>> I agree and actually thought about that yesterday. Do you think a sysctl
>>>> is a good option?
>>>
>>> Every distribution ships with the Kconfig option on, and no sysctl
>>> exists currently to control it.
>>>
>>> So I'd say it's not necessary at all, or at the very least let's have
>>> someone come forward with a real rather than theoretical use case for
>>> such a feature before adding it.
>>>
>>> Actually, if RFC 4191 has the usual language like "there SHOULD be
>>> an administrative mechanism to disable blah blah blah" I could
>>> be convinced to add it now. Can someone take a look?
>>
>> It seems that there is no such an administrative mechanism in RFC 4191.
>>
>> By the way, if the sysctl is used, we are still maintaining two different
>> code paths, isn't it? so i think David's idea is good.
>
> Makes life easier, no objections from me.
>
Removing CONFIG_IPV6_ROUTER_PREF means that the Router Preference is always
on, is this understanding right?
If that's is correct, i think compatibility issues will arise. For example, when the
CONFIG_IPV6_ROUTER_PREF option is on, the kernel should not do round-robin during
default router selection, but when the CONFIG_IPV6_ROUTER_PREF option is off, the
kernel should do it.
Thanks,
Duan
> Greetings,
>
> Hannes
>
> --
> 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 1/2] net: introduce skb_coalesce_rx_frag()
From: Jason Wang @ 2013-10-31 11:34 UTC (permalink / raw)
To: Kmindg G
Cc: mwdalton, mst, netdev, linux-kernel, virtualization, edumazet,
davem
In-Reply-To: <CAOjA20XtG80ePVKKiTHWvy7m99wo5sYC+64RTJN2tWNEopRiSQ@mail.gmail.com>
On 10/31/2013 07:05 PM, Kmindg G wrote:
> On Thu, Oct 31, 2013 at 6:28 PM, Jason Wang <jasowang@redhat.com> wrote:
>> > Sometimes we need to coalesce the rx frags to avoid frag list. One example is
>> > virtio-net driver which tries to use small frags for both MTU sized packet and
>> > GSO packet. So this patch introduce skb_coalesce_rx_frag() to do this.
>> >
>> > Cc: Rusty Russell <rusty@rustcorp.com.au>
>> > Cc: Michael S. Tsirkin <mst@redhat.com>
>> > Cc: Michael Dalton <mwdalton@google.com>
>> > Cc: Eric Dumazet <edumazet@google.com>
>> > Signed-off-by: Jason Wang <jasowang@redhat.com>
>> > ---
>> > include/linux/skbuff.h | 3 +++
>> > net/core/skbuff.c | 13 +++++++++++++
>> > 2 files changed, 16 insertions(+)
>> >
>> > diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
>> > index 2c15497..e34652b 100644
>> > --- a/include/linux/skbuff.h
>> > +++ b/include/linux/skbuff.h
>> > @@ -1372,6 +1372,9 @@ static inline void skb_fill_page_desc(struct sk_buff *skb, int i,
>> > void skb_add_rx_frag(struct sk_buff *skb, int i, struct page *page, int off,
>> > int size, unsigned int truesize);
>> >
>> > +void skb_coalesce_rx_frag(struct sk_buff *skb, int i, int off, int size,
>> > + unsigned int truesize);
>> > +
>> > #define SKB_PAGE_ASSERT(skb) BUG_ON(skb_shinfo(skb)->nr_frags)
>> > #define SKB_FRAG_ASSERT(skb) BUG_ON(skb_has_frag_list(skb))
>> > #define SKB_LINEAR_ASSERT(skb) BUG_ON(skb_is_nonlinear(skb))
>> > diff --git a/net/core/skbuff.c b/net/core/skbuff.c
>> > index 0ab32fa..fdef994 100644
>> > --- a/net/core/skbuff.c
>> > +++ b/net/core/skbuff.c
>> > @@ -476,6 +476,19 @@ void skb_add_rx_frag(struct sk_buff *skb, int i, struct page *page, int off,
>> > }
>> > EXPORT_SYMBOL(skb_add_rx_frag);
>> >
>> > +void skb_coalesce_rx_frag(struct sk_buff *skb, int i, int off, int size,
>> > + unsigned int truesize)
>> > +{
>> > + skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
>> > +
>> > + skb_frag_size_add(frag, size);
>> > + skb->len += size;
>> > + skb->data_len += size;
>> > + skb->truesize += truesize;
>> > + skb_frag_unref(skb, i);
>> > +}
> I didn't see you use "off" in skb_coalesce_rx_frag.
>
Ture, it's useless. Will post V2.
Thanks
^ permalink raw reply
* Re: [PATCH net-next 1/2] net: introduce skb_coalesce_rx_frag()
From: Michael S. Tsirkin @ 2013-10-31 11:41 UTC (permalink / raw)
To: Jason Wang
Cc: mwdalton, netdev, linux-kernel, virtualization, edumazet, davem
In-Reply-To: <1383215313-23651-1-git-send-email-jasowang@redhat.com>
On Thu, Oct 31, 2013 at 06:28:32PM +0800, Jason Wang wrote:
> Sometimes we need to coalesce the rx frags to avoid frag list. One example is
> virtio-net driver which tries to use small frags for both MTU sized packet and
> GSO packet. So this patch introduce skb_coalesce_rx_frag() to do this.
>
> Cc: Rusty Russell <rusty@rustcorp.com.au>
> Cc: Michael S. Tsirkin <mst@redhat.com>
> Cc: Michael Dalton <mwdalton@google.com>
> Cc: Eric Dumazet <edumazet@google.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
> ---
> include/linux/skbuff.h | 3 +++
> net/core/skbuff.c | 13 +++++++++++++
> 2 files changed, 16 insertions(+)
>
> diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
> index 2c15497..e34652b 100644
> --- a/include/linux/skbuff.h
> +++ b/include/linux/skbuff.h
> @@ -1372,6 +1372,9 @@ static inline void skb_fill_page_desc(struct sk_buff *skb, int i,
> void skb_add_rx_frag(struct sk_buff *skb, int i, struct page *page, int off,
> int size, unsigned int truesize);
>
> +void skb_coalesce_rx_frag(struct sk_buff *skb, int i, int off, int size,
> + unsigned int truesize);
> +
> #define SKB_PAGE_ASSERT(skb) BUG_ON(skb_shinfo(skb)->nr_frags)
> #define SKB_FRAG_ASSERT(skb) BUG_ON(skb_has_frag_list(skb))
> #define SKB_LINEAR_ASSERT(skb) BUG_ON(skb_is_nonlinear(skb))
> diff --git a/net/core/skbuff.c b/net/core/skbuff.c
> index 0ab32fa..fdef994 100644
> --- a/net/core/skbuff.c
> +++ b/net/core/skbuff.c
> @@ -476,6 +476,19 @@ void skb_add_rx_frag(struct sk_buff *skb, int i, struct page *page, int off,
> }
> EXPORT_SYMBOL(skb_add_rx_frag);
>
> +void skb_coalesce_rx_frag(struct sk_buff *skb, int i, int off, int size,
> + unsigned int truesize)
> +{
> + skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
> +
> + skb_frag_size_add(frag, size);
> + skb->len += size;
> + skb->data_len += size;
> + skb->truesize += truesize;
> + skb_frag_unref(skb, i);
> +}
> +EXPORT_SYMBOL(skb_coalesce_rx_frag);
> +
> static void skb_drop_list(struct sk_buff **listp)
> {
> kfree_skb_list(*listp);
> --
> 1.8.1.2
^ permalink raw reply
* Re: [PATCH net-next 2/2] virtio-net: coalesce rx frags when possible during rx
From: Michael S. Tsirkin @ 2013-10-31 11:41 UTC (permalink / raw)
To: Jason Wang
Cc: mwdalton, netdev, linux-kernel, virtualization, edumazet, davem
In-Reply-To: <1383215313-23651-2-git-send-email-jasowang@redhat.com>
On Thu, Oct 31, 2013 at 06:28:33PM +0800, Jason Wang wrote:
> Commit 2613af0ed18a11d5c566a81f9a6510b73180660a (virtio_net: migrate mergeable
> rx buffers to page frag allocators) try to increase the payload/truesize for
> MTU-sized traffic. But this will introduce the extra overhead for GSO packets
> received because of the frag list. This commit tries to reduce this issue by
> coalesce the possible rx frags when possible during rx. Test result shows the
> about 15% improvement on full size GSO packet receiving (and even better than
> commit 2613af0ed18a11d5c566a81f9a6510b73180660a).
>
> Before this commit:
> ./netperf -H 192.168.100.4
> MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.100.4
> () port 0 AF_INET : demo
> Recv Send Send
> Socket Socket Message Elapsed
> Size Size Size Time Throughput
> bytes bytes bytes secs. 10^6bits/sec
>
> 87380 16384 16384 10.00 20303.87
>
> After this commit:
> ./netperf -H 192.168.100.4
> MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.100.4 () port 0 AF_INET : demo
> Recv Send Send
> Socket Socket Message Elapsed
> Size Size Size Time Throughput
> bytes bytes bytes secs. 10^6bits/sec
>
> 87380 16384 16384 10.00 23841.26
>
> Cc: Rusty Russell <rusty@rustcorp.com.au>
> Cc: Michael S. Tsirkin <mst@redhat.com>
> Cc: Michael Dalton <mwdalton@google.com>
> Cc: Eric Dumazet <edumazet@google.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
Good idea.
Acked-by: Michael S. Tsirkin <mst@redhat.com>
> ---
> drivers/net/virtio_net.c | 14 ++++++++++----
> 1 file changed, 10 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 113ee93..4ff4f78 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -305,7 +305,7 @@ static int receive_mergeable(struct receive_queue *rq, struct sk_buff *head_skb)
> struct sk_buff *curr_skb = head_skb;
> char *buf;
> struct page *page;
> - int num_buf, len;
> + int num_buf, len, offset;
>
> num_buf = hdr->mhdr.num_buffers;
> while (--num_buf) {
> @@ -342,9 +342,15 @@ static int receive_mergeable(struct receive_queue *rq, struct sk_buff *head_skb)
> head_skb->truesize += MAX_PACKET_LEN;
> }
> page = virt_to_head_page(buf);
> - skb_add_rx_frag(curr_skb, num_skb_frags, page,
> - buf - (char *)page_address(page), len,
> - MAX_PACKET_LEN);
> + offset = buf - (char *)page_address(page);
> + if (skb_can_coalesce(curr_skb, num_skb_frags, page, offset)) {
> + skb_coalesce_rx_frag(curr_skb, num_skb_frags - 1,
> + offset, len, MAX_PACKET_LEN);
> + } else {
> + skb_add_rx_frag(curr_skb, num_skb_frags, page,
> + offset, len,
> + MAX_PACKET_LEN);
> + }
> --rq->num;
> }
> return 0;
> --
> 1.8.1.2
^ permalink raw reply
* Re: [PATCH 19/51] DMA-API: media: dt3155v4l: replace dma_set_mask()+dma_set_coherent_mask() with new helper
From: Mauro Carvalho Chehab @ 2013-10-31 11:46 UTC (permalink / raw)
To: Hans Verkuil
Cc: Russell King, alsa-devel-K7yf7f+aM1XWsZ/bQMPhNw,
b43-dev-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r,
devel-gWbeCf7V1WCQmaza687I9mD2FQJk+8+b,
devicetree-u79uwXL29TY76Z2rM5mHXA,
dri-devel-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW,
e1000-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f,
linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r,
linux-crypto-u79uwXL29TY76Z2rM5mHXA,
linux-doc-u79uwXL29TY76Z2rM5mHXA,
linux-fbdev-u79uwXL29TY76Z2rM5mHXA,
linux-ide-u79uwXL29TY76Z2rM5mHXA,
linux-media-u79uwXL29TY76Z2rM5mHXA,
linux-mmc-u79uwXL29TY76Z2rM5mHXA,
linux-nvme-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r,
linux-omap-u79uwXL29TY76Z2rM5mHXA,
linuxppc-dev-uLR06cmDAlY/bJ5BZ2RsiQ,
linux-samsung-soc-u79uwXL29TY76Z2rM5mHXA,
linux-scsi-u79uwXL29TY76Z2rM5mHXA,
linux-tegra-u79uwXL29TY76Z2rM5mHXA,
linux-usb-u79uwXL29TY76Z2rM5mHXA,
linux-wireless-u79uwXL29TY76Z2rM5mHXA,
netdev-u79uwXL29TY76Z2rM5mHXA, Solarflare linux maintainers,
uclinux-dist-devel-ZG0+EudsQA8dtHy/vicBwGD2FQJk+8+b,
Greg Kroah-Hartman, Russell King
In-Reply-To: <5249673B.5020705-qWit8jRvyhVmR6Xm/wNWPw@public.gmane.org>
Hi Russell,
Em Mon, 30 Sep 2013 13:57:47 +0200
Hans Verkuil <hverkuil-qWit8jRvyhVmR6Xm/wNWPw@public.gmane.org> escreveu:
> On 09/19/2013 11:44 PM, Russell King wrote:
> > Replace the following sequence:
> >
> > dma_set_mask(dev, mask);
> > dma_set_coherent_mask(dev, mask);
> >
> > with a call to the new helper dma_set_mask_and_coherent().
> >
> > Signed-off-by: Russell King <rmk+kernel-lFZ/pmaqli7XmaaqVzeoHQ@public.gmane.org>
>
> Acked-by: Hans Verkuil <hans.verkuil-FYB4Gu1CFyUAvxtiuMwx3w@public.gmane.org>
Somehow, I lost your original post (I got unsubscribed on a few days
from all vger mailing lists at the end of september).
I suspect that you want to sent this via your tree, right?
If so:
Acked-by: Mauro Carvalho Chehab <m.chehab-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org>
>
> Regards,
>
> Hans
>
> > ---
> > drivers/staging/media/dt3155v4l/dt3155v4l.c | 5 +----
> > 1 files changed, 1 insertions(+), 4 deletions(-)
> >
> > diff --git a/drivers/staging/media/dt3155v4l/dt3155v4l.c b/drivers/staging/media/dt3155v4l/dt3155v4l.c
> > index 90d6ac4..081407b 100644
> > --- a/drivers/staging/media/dt3155v4l/dt3155v4l.c
> > +++ b/drivers/staging/media/dt3155v4l/dt3155v4l.c
> > @@ -901,10 +901,7 @@ dt3155_probe(struct pci_dev *pdev, const struct pci_device_id *id)
> > int err;
> > struct dt3155_priv *pd;
> >
> > - err = dma_set_mask(&pdev->dev, DMA_BIT_MASK(32));
> > - if (err)
> > - return -ENODEV;
> > - err = dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(32));
> > + err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
> > if (err)
> > return -ENODEV;
> > pd = kzalloc(sizeof(*pd), GFP_KERNEL);
> >
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-media" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
Cheers,
Mauro
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" 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 net-next V2 1/2] net: introduce skb_coalesce_rx_frag()
From: Jason Wang @ 2013-10-31 11:47 UTC (permalink / raw)
To: davem, edumazet, linux-kernel, netdev, rusty, mst, mwdalton,
virtualization
Cc: kmindg
Sometimes we need to coalesce the rx frags to avoid frag list. One example is
virtio-net driver which tries to use small frags for both MTU sized packet and
GSO packet. So this patch introduce skb_coalesce_rx_frag() to do this.
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Michael Dalton <mwdalton@google.com>
Cc: Eric Dumazet <edumazet@google.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
Changes from V1:
- remove the useless off parameter.
---
include/linux/skbuff.h | 3 +++
net/core/skbuff.c | 13 +++++++++++++
2 files changed, 16 insertions(+)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 2c15497..fffaeaf 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -1372,6 +1372,9 @@ static inline void skb_fill_page_desc(struct sk_buff *skb, int i,
void skb_add_rx_frag(struct sk_buff *skb, int i, struct page *page, int off,
int size, unsigned int truesize);
+void skb_coalesce_rx_frag(struct sk_buff *skb, int i, int size,
+ unsigned int truesize);
+
#define SKB_PAGE_ASSERT(skb) BUG_ON(skb_shinfo(skb)->nr_frags)
#define SKB_FRAG_ASSERT(skb) BUG_ON(skb_has_frag_list(skb))
#define SKB_LINEAR_ASSERT(skb) BUG_ON(skb_is_nonlinear(skb))
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 0ab32fa..87670e1 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -476,6 +476,19 @@ void skb_add_rx_frag(struct sk_buff *skb, int i, struct page *page, int off,
}
EXPORT_SYMBOL(skb_add_rx_frag);
+void skb_coalesce_rx_frag(struct sk_buff *skb, int i, int size,
+ unsigned int truesize)
+{
+ skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
+
+ skb_frag_size_add(frag, size);
+ skb->len += size;
+ skb->data_len += size;
+ skb->truesize += truesize;
+ skb_frag_unref(skb, i);
+}
+EXPORT_SYMBOL(skb_coalesce_rx_frag);
+
static void skb_drop_list(struct sk_buff **listp)
{
kfree_skb_list(*listp);
--
1.8.1.2
^ permalink raw reply related
* [PATCH net-next V2 2/2] virtio-net: coalesce rx frags when possible during rx
From: Jason Wang @ 2013-10-31 11:47 UTC (permalink / raw)
To: davem, edumazet, linux-kernel, netdev, rusty, mst, mwdalton,
virtualization
Cc: kmindg
In-Reply-To: <1383220027-12278-1-git-send-email-jasowang@redhat.com>
Commit 2613af0ed18a11d5c566a81f9a6510b73180660a (virtio_net: migrate mergeable
rx buffers to page frag allocators) try to increase the payload/truesize for
MTU-sized traffic. But this will introduce the extra overhead for GSO packets
received because of the frag list. This commit tries to reduce this issue by
coalesce the possible rx frags when possible during rx. Test result shows the
about 15% improvement on full size GSO packet receiving (and even better than
commit 2613af0ed18a11d5c566a81f9a6510b73180660a).
Before this commit:
./netperf -H 192.168.100.4
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.100.4
() port 0 AF_INET : demo
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/sec
87380 16384 16384 10.00 20303.87
After this commit:
./netperf -H 192.168.100.4
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.100.4 () port 0 AF_INET : demo
Recv Send Send
Socket Socket Message Elapsed
Size Size Size Time Throughput
bytes bytes bytes secs. 10^6bits/sec
87380 16384 16384 10.00 23841.26
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Michael Dalton <mwdalton@google.com>
Cc: Eric Dumazet <edumazet@google.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
---
drivers/net/virtio_net.c | 14 ++++++++++----
1 file changed, 10 insertions(+), 4 deletions(-)
diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 113ee93..5dc0de0 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -305,7 +305,7 @@ static int receive_mergeable(struct receive_queue *rq, struct sk_buff *head_skb)
struct sk_buff *curr_skb = head_skb;
char *buf;
struct page *page;
- int num_buf, len;
+ int num_buf, len, offset;
num_buf = hdr->mhdr.num_buffers;
while (--num_buf) {
@@ -342,9 +342,15 @@ static int receive_mergeable(struct receive_queue *rq, struct sk_buff *head_skb)
head_skb->truesize += MAX_PACKET_LEN;
}
page = virt_to_head_page(buf);
- skb_add_rx_frag(curr_skb, num_skb_frags, page,
- buf - (char *)page_address(page), len,
- MAX_PACKET_LEN);
+ offset = buf - (char *)page_address(page);
+ if (skb_can_coalesce(curr_skb, num_skb_frags, page, offset)) {
+ skb_coalesce_rx_frag(curr_skb, num_skb_frags - 1,
+ len, MAX_PACKET_LEN);
+ } else {
+ skb_add_rx_frag(curr_skb, num_skb_frags, page,
+ offset, len,
+ MAX_PACKET_LEN);
+ }
--rq->num;
}
return 0;
--
1.8.1.2
^ permalink raw reply related
* Re: [PATCH] ipv6: remove the unnecessary statement in find_match()
From: Hannes Frederic Sowa @ 2013-10-31 11:57 UTC (permalink / raw)
To: Duan Jiong; +Cc: David Miller, netdev
In-Reply-To: <52723A84.9080308@cn.fujitsu.com>
On Thu, Oct 31, 2013 at 07:09:56PM +0800, Duan Jiong wrote:
> 于 2013年10月31日 16:45, Hannes Frederic Sowa 写道:
> > On Thu, Oct 31, 2013 at 02:02:11PM +0800, Duan Jiong wrote:
> >> 于 2013年10月31日 12:22, David Miller 写道:
> >>> From: Hannes Frederic Sowa <hannes@stressinduktion.org>
> >>> Date: Wed, 30 Oct 2013 22:11:57 +0100
> >>>
> >>>> On Wed, Oct 30, 2013 at 05:08:37PM -0400, David Miller wrote:
> >>>>> From: Duan Jiong <duanj.fnst@cn.fujitsu.com>
> >>>>> Date: Wed, 30 Oct 2013 15:39:26 +0800
> >>>>>
> >>>>>>
> >>>>>> After reading the function rt6_check_neigh(), we can
> >>>>>> know that the RT6_NUD_FAIL_SOFT can be returned only
> >>>>>> when the IS_ENABLE(CONFIG_IPV6_ROUTER_PREF) is false.
> >>>>>> so in function find_match(), there is no need to execute
> >>>>>> the statement !IS_ENABLED(CONFIG_IPV6_ROUTER_PREF).
> >>>>>>
> >>>>>> Signed-off-by: Duan Jiong <duanj.fnst@cn.fujitsu.com>
> >>>>>
> >>>>> Applied to net-next, thanks.
> >>>>>
> >>>>> CONFIG_IPV6_ROUTER_PREF is another good candidate for Kconfig
> >>>>> removal. I know we've had several bugs that only apply when
> >>>>> this option is on vs. off. We're maintaining two different
> >>>>> code paths, for really no good reason.
> >>>>
> >>>> I agree and actually thought about that yesterday. Do you think a sysctl
> >>>> is a good option?
> >>>
> >>> Every distribution ships with the Kconfig option on, and no sysctl
> >>> exists currently to control it.
> >>>
> >>> So I'd say it's not necessary at all, or at the very least let's have
> >>> someone come forward with a real rather than theoretical use case for
> >>> such a feature before adding it.
> >>>
> >>> Actually, if RFC 4191 has the usual language like "there SHOULD be
> >>> an administrative mechanism to disable blah blah blah" I could
> >>> be convinced to add it now. Can someone take a look?
> >>
> >> It seems that there is no such an administrative mechanism in RFC 4191.
> >>
> >> By the way, if the sysctl is used, we are still maintaining two different
> >> code paths, isn't it? so i think David's idea is good.
> >
> > Makes life easier, no objections from me.
> >
>
> Removing CONFIG_IPV6_ROUTER_PREF means that the Router Preference is always
> on, is this understanding right?
If we drop support for one of the routing scoring algorithms it will be the
other one. So yes, it will always be on.
> If that's is correct, i think compatibility issues will arise. For example, when the
> CONFIG_IPV6_ROUTER_PREF option is on, the kernel should not do round-robin during
> default router selection, but when the CONFIG_IPV6_ROUTER_PREF option is off, the
> kernel should do it.
Exactly, but all people use ROUTER_PREF activated given it is enabled
by default by most distributions. I also know about no best practices
or IX-policies where they want a router to deliberately turned router
preference support off.
We can talk about doing RR in the same preference level, I'll have to do more
research on this.
But given that someone wants RR, it is better done by using ECMP routes.
^ permalink raw reply
* [PATCH net-next 02/10] net/mlx4_core: Fix register/unreg vlan flow
From: Or Gerlitz @ 2013-10-31 12:29 UTC (permalink / raw)
To: davem; +Cc: netdev, amirv, roland, jackm, Or Gerlitz
In-Reply-To: <1383222569-10930-1-git-send-email-ogerlitz@mellanox.com>
From: Jack Morgenstein <jackm@dev.mellanox.co.il>
The reg/unreg vlan code was broken:
1. a wrapped function called another wrapped function, causing a deadlock.
2. unregister_vlan called cmd_box instead of cmd_box_imm, leading to
incorrectly passed parameters.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/port.c | 21 ++++++++++-----------
1 files changed, 10 insertions(+), 11 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/port.c b/drivers/net/ethernet/mellanox/mlx4/port.c
index 946e0af..d3d3106 100644
--- a/drivers/net/ethernet/mellanox/mlx4/port.c
+++ b/drivers/net/ethernet/mellanox/mlx4/port.c
@@ -284,7 +284,7 @@ static int mlx4_set_port_vlan_table(struct mlx4_dev *dev, u8 port,
memcpy(mailbox->buf, entries, MLX4_VLAN_TABLE_SIZE);
in_mod = MLX4_SET_PORT_VLAN_TABLE << 8 | port;
err = mlx4_cmd(dev, mailbox->dma, in_mod, 1, MLX4_CMD_SET_PORT,
- MLX4_CMD_TIME_CLASS_B, MLX4_CMD_WRAPPED);
+ MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
mlx4_free_cmd_mailbox(dev, mailbox);
@@ -370,6 +370,9 @@ int mlx4_register_vlan(struct mlx4_dev *dev, u8 port, u16 vlan, int *index)
u64 out_param = 0;
int err;
+ if (vlan > 4095)
+ return -EINVAL;
+
if (mlx4_is_mfunc(dev)) {
set_param_l(&out_param, port);
err = mlx4_cmd_imm(dev, vlan, &out_param, RES_VLAN,
@@ -412,18 +415,14 @@ out:
void mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, int index)
{
- u64 in_param = 0;
- int err;
+ u64 out_param = 0;
if (mlx4_is_mfunc(dev)) {
- set_param_l(&in_param, port);
- err = mlx4_cmd(dev, in_param, RES_VLAN, RES_OP_RESERVE_AND_MAP,
- MLX4_CMD_FREE_RES, MLX4_CMD_TIME_CLASS_A,
- MLX4_CMD_WRAPPED);
- if (!err)
- mlx4_warn(dev, "Failed freeing vlan at index:%d\n",
- index);
-
+ set_param_l(&out_param, port);
+ (void) mlx4_cmd_imm(dev, index, &out_param, RES_VLAN,
+ RES_OP_RESERVE_AND_MAP,
+ MLX4_CMD_FREE_RES, MLX4_CMD_TIME_CLASS_A,
+ MLX4_CMD_WRAPPED);
return;
}
__mlx4_unregister_vlan(dev, port, index);
--
1.7.1
^ permalink raw reply related
* [PATCH net-next 05/10] net/mlx4_core: Resource tracker for reg/unreg vlans
From: Or Gerlitz @ 2013-10-31 12:29 UTC (permalink / raw)
To: davem; +Cc: netdev, amirv, roland, jackm, Or Gerlitz
In-Reply-To: <1383222569-10930-1-git-send-email-ogerlitz@mellanox.com>
From: Jack Morgenstein <jackm@dev.mellanox.co.il>
Add resource tracker support for reg/unreg vlans calls done by VFs.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
.../net/ethernet/mellanox/mlx4/resource_tracker.c | 127 +++++++++++++++++++-
1 files changed, 121 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
index 993a2ef..e18bfff 100644
--- a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
+++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
@@ -55,6 +55,14 @@ struct mac_res {
u8 port;
};
+struct vlan_res {
+ struct list_head list;
+ u16 vlan;
+ int ref_count;
+ int vlan_index;
+ u8 port;
+};
+
struct res_common {
struct list_head list;
struct rb_node node;
@@ -266,6 +274,7 @@ static const char *ResourceType(enum mlx4_resource rt)
case RES_MPT: return "RES_MPT";
case RES_MTT: return "RES_MTT";
case RES_MAC: return "RES_MAC";
+ case RES_VLAN: return "RES_VLAN";
case RES_EQ: return "RES_EQ";
case RES_COUNTER: return "RES_COUNTER";
case RES_FS_RULE: return "RES_FS_RULE";
@@ -274,6 +283,7 @@ static const char *ResourceType(enum mlx4_resource rt)
};
}
+static void rem_slave_vlans(struct mlx4_dev *dev, int slave);
int mlx4_init_resource_tracker(struct mlx4_dev *dev)
{
struct mlx4_priv *priv = mlx4_priv(dev);
@@ -309,11 +319,18 @@ void mlx4_free_resource_tracker(struct mlx4_dev *dev,
int i;
if (priv->mfunc.master.res_tracker.slave_list) {
- if (type != RES_TR_FREE_STRUCTS_ONLY)
- for (i = 0 ; i < dev->num_slaves; i++)
+ if (type != RES_TR_FREE_STRUCTS_ONLY) {
+ for (i = 0; i < dev->num_slaves; i++) {
if (type == RES_TR_FREE_ALL ||
dev->caps.function != i)
mlx4_delete_all_resources_for_slave(dev, i);
+ }
+ /* free master's vlans */
+ i = dev->caps.function;
+ mutex_lock(&priv->mfunc.master.res_tracker.slave_list[i].mutex);
+ rem_slave_vlans(dev, i);
+ mutex_unlock(&priv->mfunc.master.res_tracker.slave_list[i].mutex);
+ }
if (type != RES_TR_FREE_SLAVES_ONLY) {
kfree(priv->mfunc.master.res_tracker.slave_list);
@@ -1469,12 +1486,96 @@ static int mac_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
return err;
}
-static int vlan_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
- u64 in_param, u64 *out_param, int port)
+static int vlan_add_to_slave(struct mlx4_dev *dev, int slave, u16 vlan,
+ int port, int vlan_index)
{
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+ struct list_head *vlan_list =
+ &tracker->slave_list[slave].res_list[RES_VLAN];
+ struct vlan_res *res, *tmp;
+
+ list_for_each_entry_safe(res, tmp, vlan_list, list) {
+ if (res->vlan == vlan && res->port == (u8) port) {
+ /* vlan found. update ref count */
+ ++res->ref_count;
+ return 0;
+ }
+ }
+
+ res = kzalloc(sizeof(*res), GFP_KERNEL);
+ if (!res)
+ return -ENOMEM;
+ res->vlan = vlan;
+ res->port = (u8) port;
+ res->vlan_index = vlan_index;
+ res->ref_count = 1;
+ list_add_tail(&res->list,
+ &tracker->slave_list[slave].res_list[RES_VLAN]);
return 0;
}
+
+static void vlan_del_from_slave(struct mlx4_dev *dev, int slave, u16 vlan,
+ int port)
+{
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+ struct list_head *vlan_list =
+ &tracker->slave_list[slave].res_list[RES_VLAN];
+ struct vlan_res *res, *tmp;
+
+ list_for_each_entry_safe(res, tmp, vlan_list, list) {
+ if (res->vlan == vlan && res->port == (u8) port) {
+ if (!--res->ref_count) {
+ list_del(&res->list);
+ kfree(res);
+ }
+ break;
+ }
+ }
+}
+
+static void rem_slave_vlans(struct mlx4_dev *dev, int slave)
+{
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
+ struct list_head *vlan_list =
+ &tracker->slave_list[slave].res_list[RES_VLAN];
+ struct vlan_res *res, *tmp;
+ int i;
+
+ list_for_each_entry_safe(res, tmp, vlan_list, list) {
+ list_del(&res->list);
+ /* dereference the vlan the num times the slave referenced it */
+ for (i = 0; i < res->ref_count; i++)
+ __mlx4_unregister_vlan(dev, res->port, res->vlan);
+ kfree(res);
+ }
+}
+
+static int vlan_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
+ u64 in_param, u64 *out_param, int port)
+{
+ int err;
+ u16 vlan;
+ int vlan_index;
+
+ if (!port || op != RES_OP_RESERVE_AND_MAP)
+ return -EINVAL;
+
+ vlan = (u16) in_param;
+
+ err = __mlx4_register_vlan(dev, port, vlan, &vlan_index);
+ if (!err) {
+ set_param_l(out_param, (u32) vlan_index);
+ err = vlan_add_to_slave(dev, slave, vlan, port, vlan_index);
+ if (err)
+ __mlx4_unregister_vlan(dev, port, vlan);
+ }
+ return err;
+}
+
static int counter_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
u64 in_param, u64 *out_param)
{
@@ -1755,7 +1856,21 @@ static int mac_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
static int vlan_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
u64 in_param, u64 *out_param, int port)
{
- return 0;
+ int err = 0;
+
+ switch (op) {
+ case RES_OP_RESERVE_AND_MAP:
+ if (!port)
+ return -EINVAL;
+ vlan_del_from_slave(dev, slave, in_param, port);
+ __mlx4_unregister_vlan(dev, port, in_param);
+ break;
+ default:
+ err = -EINVAL;
+ break;
+ }
+
+ return err;
}
static int counter_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
@@ -3968,7 +4083,7 @@ void mlx4_delete_all_resources_for_slave(struct mlx4_dev *dev, int slave)
struct mlx4_priv *priv = mlx4_priv(dev);
mutex_lock(&priv->mfunc.master.res_tracker.slave_list[slave].mutex);
- /*VLAN*/
+ rem_slave_vlans(dev, slave);
rem_slave_macs(dev, slave);
rem_slave_fs_rule(dev, slave);
rem_slave_qps(dev, slave);
--
1.7.1
^ permalink raw reply related
* [PATCH net-next 07/10] net/mlx4_core: Fix checking order in MR table init
From: Or Gerlitz @ 2013-10-31 12:29 UTC (permalink / raw)
To: davem; +Cc: netdev, amirv, roland, jackm, Or Gerlitz
In-Reply-To: <1383222569-10930-1-git-send-email-ogerlitz@mellanox.com>
From: Jack Morgenstein <jackm@dev.mellanox.co.il>
In procedure mlx4_init_mr_table(), slaves should do no processing,
but should return success. This initialization is hypervisor-only.
However, the check for num_mpts being a power-of-2 was performed
before the check to return immediately if the driver is for a slave.
This resulted in spurious failures.
The order of performing the checks is reversed, so that if the
driver is for a slave, no processing is done and success is returned.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/mr.c | 6 +++---
1 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/mr.c b/drivers/net/ethernet/mellanox/mlx4/mr.c
index f91719a..63391a1 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mr.c
+++ b/drivers/net/ethernet/mellanox/mlx4/mr.c
@@ -755,14 +755,14 @@ int mlx4_init_mr_table(struct mlx4_dev *dev)
struct mlx4_mr_table *mr_table = &priv->mr_table;
int err;
- if (!is_power_of_2(dev->caps.num_mpts))
- return -EINVAL;
-
/* Nothing to do for slaves - all MR handling is forwarded
* to the master */
if (mlx4_is_slave(dev))
return 0;
+ if (!is_power_of_2(dev->caps.num_mpts))
+ return -EINVAL;
+
err = mlx4_bitmap_init(&mr_table->mpt_bitmap, dev->caps.num_mpts,
~0, dev->caps.reserved_mrws, 0);
if (err)
--
1.7.1
^ permalink raw reply related
* [PATCH net-next 01/10] net/mlx4_core: Fix call to __mlx4_unregister_mac
From: Or Gerlitz @ 2013-10-31 12:29 UTC (permalink / raw)
To: davem; +Cc: netdev, amirv, roland, jackm, Or Gerlitz
In-Reply-To: <1383222569-10930-1-git-send-email-ogerlitz@mellanox.com>
From: Jack Morgenstein <jackm@dev.mellanox.co.il>
In function mlx4_master_deactivate_admin_state() __mlx4_unregister_mac was
called using the MAC index. It should be called with the value of the MAC itself.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/cmd.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
index 735765c..ae8eb4c 100644
--- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
@@ -1691,7 +1691,7 @@ static void mlx4_master_deactivate_admin_state(struct mlx4_priv *priv, int slave
vp_oper->vlan_idx = NO_INDX;
}
if (NO_INDX != vp_oper->mac_idx) {
- __mlx4_unregister_mac(&priv->dev, port, vp_oper->mac_idx);
+ __mlx4_unregister_mac(&priv->dev, port, vp_oper->state.mac);
vp_oper->mac_idx = NO_INDX;
}
}
--
1.7.1
^ permalink raw reply related
* [PATCH net-next 06/10] net/mlx4_core: Don't fail reg/unreg vlan for older guests
From: Or Gerlitz @ 2013-10-31 12:29 UTC (permalink / raw)
To: davem; +Cc: netdev, amirv, roland, jackm, Or Gerlitz
In-Reply-To: <1383222569-10930-1-git-send-email-ogerlitz@mellanox.com>
From: Jack Morgenstein <jackm@dev.mellanox.co.il>
In upstream kernels under SRIOV, the vlan register/unregister calls
were NOPs (doing nothing and returning OK). We detect these old
calls from guests (via the comm channel), since previously the
port number in mlx4_register_vlan was passed (improperly) in the
out_param. This has been corrected so that the port number is now
passed in bits 8..15 of the in_modifier field.
For old calls, these bits will be zero, so if the passed port
number is zero, we can still look at the out_param field to see
if it contains a valid port number. If yes, the VM is running
an old driver.
Since for old drivers, the register/unregister_vlan wrappers were
NOPs, we continue this policy -- the reason being that upstream
had an additional bug in eth driver running on guests (where
procedure mlx4_en_vlan_rx_kill_vid() had the following code:
if (!mlx4_find_cached_vlan(mdev->dev, priv->port, vid, &idx))
mlx4_unregister_vlan(mdev->dev, priv->port, idx);
else
en_err(priv, "could not find vid %d in cache\n", vid);
On a VM, mlx4_find_cached_vlan() will always fail, since the
vlan cache is located on the Hypervisor; on guests it is empty.
Therefore, if we allow upstream guests to register vlans, we will
have vlan leakage since the unregister will never be performed.
Leaving vlan reg/unreg for old guest drivers as a NOP is not a
feature regression, since in upstream the register/unregister
vlan wrapper is a NOP.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/cmd.c | 1 +
drivers/net/ethernet/mellanox/mlx4/mlx4.h | 1 +
.../net/ethernet/mellanox/mlx4/resource_tracker.c | 17 ++++++++++++++++-
3 files changed, 18 insertions(+), 1 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
index 887d625..65d41b7 100644
--- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
@@ -1718,6 +1718,7 @@ static void mlx4_master_do_cmd(struct mlx4_dev *dev, int slave, u8 cmd,
if (cmd == MLX4_COMM_CMD_RESET) {
mlx4_warn(dev, "Received reset from slave:%d\n", slave);
slave_state[slave].active = false;
+ slave_state[slave].old_vlan_api = false;
mlx4_master_deactivate_admin_state(priv, slave);
for (i = 0; i < MLX4_EVENT_TYPES_NUM; ++i) {
slave_state[slave].event_eq[i].eqn = -1;
diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4.h b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
index f2ad4f6..9794126 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mlx4.h
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
@@ -455,6 +455,7 @@ struct mlx4_slave_state {
u8 last_cmd;
u8 init_port_mask;
bool active;
+ bool old_vlan_api;
u8 function;
dma_addr_t vhcr_dma;
u16 mtu[MLX4_MAX_PORTS + 1];
diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
index e18bfff..3586388 100644
--- a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
+++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
@@ -1555,15 +1555,26 @@ static void rem_slave_vlans(struct mlx4_dev *dev, int slave)
}
static int vlan_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
- u64 in_param, u64 *out_param, int port)
+ u64 in_param, u64 *out_param, int in_port)
{
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ struct mlx4_slave_state *slave_state = priv->mfunc.master.slave_state;
int err;
u16 vlan;
int vlan_index;
+ int port;
+
+ port = !in_port ? get_param_l(out_param) : in_port;
if (!port || op != RES_OP_RESERVE_AND_MAP)
return -EINVAL;
+ /* upstream kernels had NOP for reg/unreg vlan. Continue this. */
+ if (!in_port && port > 0 && port <= dev->caps.num_ports) {
+ slave_state[slave].old_vlan_api = true;
+ return 0;
+ }
+
vlan = (u16) in_param;
err = __mlx4_register_vlan(dev, port, vlan, &vlan_index);
@@ -1856,10 +1867,14 @@ static int mac_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
static int vlan_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
u64 in_param, u64 *out_param, int port)
{
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ struct mlx4_slave_state *slave_state = priv->mfunc.master.slave_state;
int err = 0;
switch (op) {
case RES_OP_RESERVE_AND_MAP:
+ if (slave_state[slave].old_vlan_api)
+ return 0;
if (!port)
return -EINVAL;
vlan_del_from_slave(dev, slave, in_param, port);
--
1.7.1
^ permalink raw reply related
* [PATCH net-next 04/10] net/mlx4_en: Use vlan id instead of vlan index for unregistration
From: Or Gerlitz @ 2013-10-31 12:29 UTC (permalink / raw)
To: davem; +Cc: netdev, amirv, roland, jackm, Or Gerlitz
In-Reply-To: <1383222569-10930-1-git-send-email-ogerlitz@mellanox.com>
From: Jack Morgenstein <jackm@dev.mellanox.co.il>
Use of vlan_index created problems unregistering vlans on guests.
In addition, tools delete vlan by tag, not by index, lets follow that.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/cmd.c | 2 +-
drivers/net/ethernet/mellanox/mlx4/en_netdev.c | 6 +---
drivers/net/ethernet/mellanox/mlx4/mlx4.h | 2 +-
drivers/net/ethernet/mellanox/mlx4/port.c | 27 +++++++++++---------
.../net/ethernet/mellanox/mlx4/resource_tracker.c | 2 +-
include/linux/mlx4/device.h | 2 +-
6 files changed, 20 insertions(+), 21 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
index ae8eb4c..887d625 100644
--- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
@@ -1687,7 +1687,7 @@ static void mlx4_master_deactivate_admin_state(struct mlx4_priv *priv, int slave
vp_oper = &priv->mfunc.master.vf_oper[slave].vport[port];
if (NO_INDX != vp_oper->vlan_idx) {
__mlx4_unregister_vlan(&priv->dev,
- port, vp_oper->vlan_idx);
+ port, vp_oper->state.default_vlan);
vp_oper->vlan_idx = NO_INDX;
}
if (NO_INDX != vp_oper->mac_idx) {
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_netdev.c b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
index 85d9166..b555412 100644
--- a/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
@@ -417,7 +417,6 @@ static int mlx4_en_vlan_rx_kill_vid(struct net_device *dev,
struct mlx4_en_priv *priv = netdev_priv(dev);
struct mlx4_en_dev *mdev = priv->mdev;
int err;
- int idx;
en_dbg(HW, priv, "Killing VID:%d\n", vid);
@@ -425,10 +424,7 @@ static int mlx4_en_vlan_rx_kill_vid(struct net_device *dev,
/* Remove VID from port VLAN filter */
mutex_lock(&mdev->state_lock);
- if (!mlx4_find_cached_vlan(mdev->dev, priv->port, vid, &idx))
- mlx4_unregister_vlan(mdev->dev, priv->port, idx);
- else
- en_dbg(HW, priv, "could not find vid %d in cache\n", vid);
+ mlx4_unregister_vlan(mdev->dev, priv->port, vid);
if (mdev->device_up && priv->port_up) {
err = mlx4_SET_VLAN_FLTR(mdev->dev, priv);
diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4.h b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
index 348bb8c..f2ad4f6 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mlx4.h
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
@@ -1111,7 +1111,7 @@ int mlx4_change_port_types(struct mlx4_dev *dev,
void mlx4_init_mac_table(struct mlx4_dev *dev, struct mlx4_mac_table *table);
void mlx4_init_vlan_table(struct mlx4_dev *dev, struct mlx4_vlan_table *table);
-void __mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, int index);
+void __mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, u16 vlan);
int __mlx4_register_vlan(struct mlx4_dev *dev, u8 port, u16 vlan, int *index);
int mlx4_SET_PORT(struct mlx4_dev *dev, u8 port, int pkey_tbl_sz);
diff --git a/drivers/net/ethernet/mellanox/mlx4/port.c b/drivers/net/ethernet/mellanox/mlx4/port.c
index 9433c1f..caaa154 100644
--- a/drivers/net/ethernet/mellanox/mlx4/port.c
+++ b/drivers/net/ethernet/mellanox/mlx4/port.c
@@ -406,23 +406,26 @@ int mlx4_register_vlan(struct mlx4_dev *dev, u8 port, u16 vlan, int *index)
}
EXPORT_SYMBOL_GPL(mlx4_register_vlan);
-void __mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, int index)
+void __mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, u16 vlan)
{
struct mlx4_vlan_table *table = &mlx4_priv(dev)->port[port].vlan_table;
+ int index;
- if (index < MLX4_VLAN_REGULAR) {
- mlx4_warn(dev, "Trying to free special vlan index %d\n", index);
- return;
+ mutex_lock(&table->mutex);
+ if (mlx4_find_cached_vlan(dev, port, vlan, &index)) {
+ mlx4_warn(dev, "vlan 0x%x is not in the vlan table\n", vlan);
+ goto out;
}
- mutex_lock(&table->mutex);
- if (!table->refs[index]) {
- mlx4_warn(dev, "No vlan entry for index %d\n", index);
+ if (index < MLX4_VLAN_REGULAR) {
+ mlx4_warn(dev, "Trying to free special vlan index %d\n", index);
goto out;
}
+
if (--table->refs[index]) {
- mlx4_dbg(dev, "Have more references for index %d,"
- "no need to modify vlan table\n", index);
+ mlx4_dbg(dev, "Have %d more references for index %d,"
+ "no need to modify vlan table\n", table->refs[index],
+ index);
goto out;
}
table->entries[index] = 0;
@@ -432,19 +435,19 @@ out:
mutex_unlock(&table->mutex);
}
-void mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, int index)
+void mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, u16 vlan)
{
u64 out_param = 0;
if (mlx4_is_mfunc(dev)) {
- (void) mlx4_cmd_imm(dev, index, &out_param,
+ (void) mlx4_cmd_imm(dev, vlan, &out_param,
((u32) port) << 8 | (u32) RES_VLAN,
RES_OP_RESERVE_AND_MAP,
MLX4_CMD_FREE_RES, MLX4_CMD_TIME_CLASS_A,
MLX4_CMD_WRAPPED);
return;
}
- __mlx4_unregister_vlan(dev, port, index);
+ __mlx4_unregister_vlan(dev, port, vlan);
}
EXPORT_SYMBOL_GPL(mlx4_unregister_vlan);
diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
index a5aa3be..993a2ef 100644
--- a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
+++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
@@ -4085,7 +4085,7 @@ void mlx4_vf_immed_vlan_work_handler(struct work_struct *_work)
if (work->flags & MLX4_VF_IMMED_VLAN_FLAG_VLAN && !errors &&
NO_INDX != work->orig_vlan_ix)
__mlx4_unregister_vlan(&work->priv->dev, work->port,
- work->orig_vlan_ix);
+ work->orig_vlan_id);
out:
kfree(work);
return;
diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index 297a163..e2e9288 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -1079,7 +1079,7 @@ int mlx4_SET_PORT_SCHEDULER(struct mlx4_dev *dev, u8 port, u8 *tc_tx_bw,
u8 *pg, u16 *ratelimit);
int mlx4_find_cached_vlan(struct mlx4_dev *dev, u8 port, u16 vid, int *idx);
int mlx4_register_vlan(struct mlx4_dev *dev, u8 port, u16 vlan, int *index);
-void mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, int index);
+void mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, u16 vlan);
int mlx4_map_phys_fmr(struct mlx4_dev *dev, struct mlx4_fmr *fmr, u64 *page_list,
int npages, u64 iova, u32 *lkey, u32 *rkey);
--
1.7.1
^ permalink raw reply related
* [PATCH net-next 08/10] mlx4: Structures and init/teardown for VF resource quotas
From: Or Gerlitz @ 2013-10-31 12:29 UTC (permalink / raw)
To: davem; +Cc: netdev, amirv, roland, jackm, Or Gerlitz
In-Reply-To: <1383222569-10930-1-git-send-email-ogerlitz@mellanox.com>
From: Jack Morgenstein <jackm@dev.mellanox.co.il>
This is step #1 for implementing SRIOV resource quotas for VFs.
Quotas are implemented per resource type for VFs and the PF, to prevent
any entity from simply grabbing all the resources for itself and leaving
the other entities unable to obtain such resources.
Resources which are allocated using quotas: QPs, CQs, SRQs, MPTs, MTTs, MAC,
VLAN, and Counters.
The quota system works as follows:
Each entity (VF or PF) is given a max number of a given resource (its quota),
and a guaranteed minimum number for each resource (starvation prevention).
For QPs, CQs, SRQs, MPTs and MTTs:
50% of the available quantity for the resource is divided equally among
the PF and all the active VFs (i.e., the number of VFs in the mlx4_core module
parameter "num_vfs"). This 50% represents the "guaranteed minimum" pool.
The other 50% is the "free pool", allocated on a first-come-first-serve basis.
For each VF/PF, resources are first allocated from its "guaranteed-minimum"
pool. When that pool is exhausted, the driver attempts to allocate from
the resource "free-pool".
The quota (i.e., max) for the VFs and the PF is:
The free-pool amount (50% of the real max) + the guaranteed minimum
For MACs:
Guarantee 2 MACs per VF/PF per port. As a result, since we have only
128 MACs per port, reduce the allowable number of VFs from 64 to 63.
Any remaining MACs are put into a free pool.
For VLANs:
For the PF, the per-port quota is 128 and guarantee is 64
(to allow the PF to register at least a VLAN per VF in VST mode).
For the VFs, the per-port quota is 64 and the guarantee is 0.
We assume that VGT VFs are trusted not to abuse the VLAN resource.
For Counters:
For all functions (PF and VFs), the quota is 128 and the guarantee is 0.
In this patch, we define the needed structures, which are added to the
resource-tracker struct. In addition, we do initialization
for the resource quota, and adjust the query_device response to use quotas
rather than resource maxima.
As part of the implementation, we introduce a new field in
mlx4_dev: quotas. This field holds the resource quotas used
to report maxima to the upper layers (ib_core, via query_device).
The HCA maxima of these values are passed to the VFs (via
QUERY_HCA) so that they may continue to use these in handling
QPs, CQs, SRQs and MPTs.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/infiniband/hw/mlx4/main.c | 8 +-
drivers/net/ethernet/mellanox/mlx4/fw.c | 11 +-
drivers/net/ethernet/mellanox/mlx4/main.c | 32 +++--
drivers/net/ethernet/mellanox/mlx4/mlx4.h | 17 ++
drivers/net/ethernet/mellanox/mlx4/qp.c | 3 +-
.../net/ethernet/mellanox/mlx4/resource_tracker.c | 157 +++++++++++++++++++-
include/linux/mlx4/device.h | 17 ++
7 files changed, 222 insertions(+), 23 deletions(-)
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index f061264..7567437 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -177,18 +177,18 @@ static int mlx4_ib_query_device(struct ib_device *ibdev,
props->max_mr_size = ~0ull;
props->page_size_cap = dev->dev->caps.page_size_cap;
- props->max_qp = dev->dev->caps.num_qps - dev->dev->caps.reserved_qps;
+ props->max_qp = dev->dev->quotas.qp;
props->max_qp_wr = dev->dev->caps.max_wqes - MLX4_IB_SQ_MAX_SPARE;
props->max_sge = min(dev->dev->caps.max_sq_sg,
dev->dev->caps.max_rq_sg);
- props->max_cq = dev->dev->caps.num_cqs - dev->dev->caps.reserved_cqs;
+ props->max_cq = dev->dev->quotas.cq;
props->max_cqe = dev->dev->caps.max_cqes;
- props->max_mr = dev->dev->caps.num_mpts - dev->dev->caps.reserved_mrws;
+ props->max_mr = dev->dev->quotas.mpt;
props->max_pd = dev->dev->caps.num_pds - dev->dev->caps.reserved_pds;
props->max_qp_rd_atom = dev->dev->caps.max_qp_dest_rdma;
props->max_qp_init_rd_atom = dev->dev->caps.max_qp_init_rdma;
props->max_res_rd_atom = props->max_qp_rd_atom * props->max_qp;
- props->max_srq = dev->dev->caps.num_srqs - dev->dev->caps.reserved_srqs;
+ props->max_srq = dev->dev->quotas.srq;
props->max_srq_wr = dev->dev->caps.max_srq_wqes - 1;
props->max_srq_sge = dev->dev->caps.max_srq_sge;
props->max_fast_reg_page_list_len = MLX4_MAX_FAST_REG_PAGES;
diff --git a/drivers/net/ethernet/mellanox/mlx4/fw.c b/drivers/net/ethernet/mellanox/mlx4/fw.c
index c151e7a..f8c88c3 100644
--- a/drivers/net/ethernet/mellanox/mlx4/fw.c
+++ b/drivers/net/ethernet/mellanox/mlx4/fw.c
@@ -177,6 +177,7 @@ int mlx4_QUERY_FUNC_CAP_wrapper(struct mlx4_dev *dev, int slave,
struct mlx4_cmd_mailbox *outbox,
struct mlx4_cmd_info *cmd)
{
+ struct mlx4_priv *priv = mlx4_priv(dev);
u8 field;
u32 size;
int err = 0;
@@ -250,13 +251,13 @@ int mlx4_QUERY_FUNC_CAP_wrapper(struct mlx4_dev *dev, int slave,
field = 0; /* protected FMR support not available as yet */
MLX4_PUT(outbox->buf, field, QUERY_FUNC_CAP_FMR_OFFSET);
- size = dev->caps.num_qps;
+ size = priv->mfunc.master.res_tracker.res_alloc[RES_QP].quota[slave];
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_QP_QUOTA_OFFSET);
- size = dev->caps.num_srqs;
+ size = priv->mfunc.master.res_tracker.res_alloc[RES_SRQ].quota[slave];
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET);
- size = dev->caps.num_cqs;
+ size = priv->mfunc.master.res_tracker.res_alloc[RES_CQ].quota[slave];
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_CQ_QUOTA_OFFSET);
size = dev->caps.num_eqs;
@@ -265,10 +266,10 @@ int mlx4_QUERY_FUNC_CAP_wrapper(struct mlx4_dev *dev, int slave,
size = dev->caps.reserved_eqs;
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_RESERVED_EQ_OFFSET);
- size = dev->caps.num_mpts;
+ size = priv->mfunc.master.res_tracker.res_alloc[RES_MPT].quota[slave];
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MPT_QUOTA_OFFSET);
- size = dev->caps.num_mtts;
+ size = priv->mfunc.master.res_tracker.res_alloc[RES_MTT].quota[slave];
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MTT_QUOTA_OFFSET);
size = dev->caps.num_mgms + dev->caps.num_amgms;
diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
index 179d267..7d2628d 100644
--- a/drivers/net/ethernet/mellanox/mlx4/main.c
+++ b/drivers/net/ethernet/mellanox/mlx4/main.c
@@ -562,13 +562,17 @@ static int mlx4_slave_cap(struct mlx4_dev *dev)
}
dev->caps.num_ports = func_cap.num_ports;
- dev->caps.num_qps = func_cap.qp_quota;
- dev->caps.num_srqs = func_cap.srq_quota;
- dev->caps.num_cqs = func_cap.cq_quota;
- dev->caps.num_eqs = func_cap.max_eq;
- dev->caps.reserved_eqs = func_cap.reserved_eq;
- dev->caps.num_mpts = func_cap.mpt_quota;
- dev->caps.num_mtts = func_cap.mtt_quota;
+ dev->quotas.qp = func_cap.qp_quota;
+ dev->quotas.srq = func_cap.srq_quota;
+ dev->quotas.cq = func_cap.cq_quota;
+ dev->quotas.mpt = func_cap.mpt_quota;
+ dev->quotas.mtt = func_cap.mtt_quota;
+ dev->caps.num_qps = 1 << hca_param.log_num_qps;
+ dev->caps.num_srqs = 1 << hca_param.log_num_srqs;
+ dev->caps.num_cqs = 1 << hca_param.log_num_cqs;
+ dev->caps.num_mpts = 1 << hca_param.log_mpt_sz;
+ dev->caps.num_eqs = func_cap.max_eq;
+ dev->caps.reserved_eqs = func_cap.reserved_eq;
dev->caps.num_pds = MLX4_NUM_PDS;
dev->caps.num_mgms = 0;
dev->caps.num_amgms = 0;
@@ -2102,9 +2106,15 @@ static int __mlx4_init_one(struct pci_dev *pdev, int pci_dev_data)
"aborting.\n");
return err;
}
- if (num_vfs > MLX4_MAX_NUM_VF) {
- printk(KERN_ERR "There are more VF's (%d) than allowed(%d)\n",
- num_vfs, MLX4_MAX_NUM_VF);
+
+ /* Due to requirement that all VFs and the PF are *guaranteed* 2 MACS
+ * per port, we must limit the number of VFs to 63 (since their are
+ * 128 MACs)
+ */
+ if (num_vfs >= MLX4_MAX_NUM_VF) {
+ dev_err(&pdev->dev,
+ "Requested more VF's (%d) than allowed (%d)\n",
+ num_vfs, MLX4_MAX_NUM_VF - 1);
return -EINVAL;
}
@@ -2322,6 +2332,8 @@ slave_start:
if (err)
goto err_steer;
+ mlx4_init_quotas(dev);
+
for (port = 1; port <= dev->caps.num_ports; port++) {
err = mlx4_init_port_info(dev, port);
if (err)
diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4.h b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
index 9794126..e7eb86e 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mlx4.h
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
@@ -504,12 +504,27 @@ struct slave_list {
struct list_head res_list[MLX4_NUM_OF_RESOURCE_TYPE];
};
+struct resource_allocator {
+ union {
+ int res_reserved;
+ int res_port_rsvd[MLX4_MAX_PORTS];
+ };
+ union {
+ int res_free;
+ int res_port_free[MLX4_MAX_PORTS];
+ };
+ int *quota;
+ int *allocated;
+ int *guaranteed;
+};
+
struct mlx4_resource_tracker {
spinlock_t lock;
/* tree for each resources */
struct rb_root res_tree[MLX4_NUM_OF_RESOURCE_TYPE];
/* num_of_slave's lists, one per slave */
struct slave_list *slave_list;
+ struct resource_allocator res_alloc[MLX4_NUM_OF_RESOURCE_TYPE];
};
#define SLAVE_EVENT_EQ_SIZE 128
@@ -1253,4 +1268,6 @@ static inline spinlock_t *mlx4_tlock(struct mlx4_dev *dev)
void mlx4_vf_immed_vlan_work_handler(struct work_struct *_work);
+void mlx4_init_quotas(struct mlx4_dev *dev);
+
#endif /* MLX4_H */
diff --git a/drivers/net/ethernet/mellanox/mlx4/qp.c b/drivers/net/ethernet/mellanox/mlx4/qp.c
index e891b05..2715e61 100644
--- a/drivers/net/ethernet/mellanox/mlx4/qp.c
+++ b/drivers/net/ethernet/mellanox/mlx4/qp.c
@@ -480,8 +480,7 @@ int mlx4_init_qp_table(struct mlx4_dev *dev)
*/
err = mlx4_bitmap_init(&qp_table->bitmap, dev->caps.num_qps,
- (1 << 23) - 1, dev->phys_caps.base_sqpn + 8 +
- 16 * MLX4_MFUNC_MAX * !!mlx4_is_master(dev),
+ (1 << 23) - 1, mlx4_num_reserved_sqps(dev),
reserved_from_top);
if (err)
return err;
diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
index 3586388..cc5d6d0 100644
--- a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
+++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
@@ -284,10 +284,59 @@ static const char *ResourceType(enum mlx4_resource rt)
}
static void rem_slave_vlans(struct mlx4_dev *dev, int slave);
+static inline void initialize_res_quotas(struct mlx4_dev *dev,
+ struct resource_allocator *res_alloc,
+ enum mlx4_resource res_type,
+ int vf, int num_instances)
+{
+ res_alloc->guaranteed[vf] = num_instances / (2 * (dev->num_vfs + 1));
+ res_alloc->quota[vf] = (num_instances / 2) + res_alloc->guaranteed[vf];
+ if (vf == mlx4_master_func_num(dev)) {
+ res_alloc->res_free = num_instances;
+ if (res_type == RES_MTT) {
+ /* reserved mtts will be taken out of the PF allocation */
+ res_alloc->res_free += dev->caps.reserved_mtts;
+ res_alloc->guaranteed[vf] += dev->caps.reserved_mtts;
+ res_alloc->quota[vf] += dev->caps.reserved_mtts;
+ }
+ }
+}
+
+void mlx4_init_quotas(struct mlx4_dev *dev)
+{
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ int pf;
+
+ /* quotas for VFs are initialized in mlx4_slave_cap */
+ if (mlx4_is_slave(dev))
+ return;
+
+ if (!mlx4_is_mfunc(dev)) {
+ dev->quotas.qp = dev->caps.num_qps - dev->caps.reserved_qps -
+ mlx4_num_reserved_sqps(dev);
+ dev->quotas.cq = dev->caps.num_cqs - dev->caps.reserved_cqs;
+ dev->quotas.srq = dev->caps.num_srqs - dev->caps.reserved_srqs;
+ dev->quotas.mtt = dev->caps.num_mtts - dev->caps.reserved_mtts;
+ dev->quotas.mpt = dev->caps.num_mpts - dev->caps.reserved_mrws;
+ return;
+ }
+
+ pf = mlx4_master_func_num(dev);
+ dev->quotas.qp =
+ priv->mfunc.master.res_tracker.res_alloc[RES_QP].quota[pf];
+ dev->quotas.cq =
+ priv->mfunc.master.res_tracker.res_alloc[RES_CQ].quota[pf];
+ dev->quotas.srq =
+ priv->mfunc.master.res_tracker.res_alloc[RES_SRQ].quota[pf];
+ dev->quotas.mtt =
+ priv->mfunc.master.res_tracker.res_alloc[RES_MTT].quota[pf];
+ dev->quotas.mpt =
+ priv->mfunc.master.res_tracker.res_alloc[RES_MPT].quota[pf];
+}
int mlx4_init_resource_tracker(struct mlx4_dev *dev)
{
struct mlx4_priv *priv = mlx4_priv(dev);
- int i;
+ int i, j;
int t;
priv->mfunc.master.res_tracker.slave_list =
@@ -308,8 +357,104 @@ int mlx4_init_resource_tracker(struct mlx4_dev *dev)
for (i = 0 ; i < MLX4_NUM_OF_RESOURCE_TYPE; i++)
priv->mfunc.master.res_tracker.res_tree[i] = RB_ROOT;
+ for (i = 0; i < MLX4_NUM_OF_RESOURCE_TYPE; i++) {
+ struct resource_allocator *res_alloc =
+ &priv->mfunc.master.res_tracker.res_alloc[i];
+ res_alloc->quota = kmalloc((dev->num_vfs + 1) * sizeof(int), GFP_KERNEL);
+ res_alloc->guaranteed = kmalloc((dev->num_vfs + 1) * sizeof(int), GFP_KERNEL);
+ if (i == RES_MAC || i == RES_VLAN)
+ res_alloc->allocated = kzalloc(MLX4_MAX_PORTS *
+ (dev->num_vfs + 1) * sizeof(int),
+ GFP_KERNEL);
+ else
+ res_alloc->allocated = kzalloc((dev->num_vfs + 1) * sizeof(int), GFP_KERNEL);
+
+ if (!res_alloc->quota || !res_alloc->guaranteed ||
+ !res_alloc->allocated)
+ goto no_mem_err;
+
+ for (t = 0; t < dev->num_vfs + 1; t++) {
+ switch (i) {
+ case RES_QP:
+ initialize_res_quotas(dev, res_alloc, RES_QP,
+ t, dev->caps.num_qps -
+ dev->caps.reserved_qps -
+ mlx4_num_reserved_sqps(dev));
+ break;
+ case RES_CQ:
+ initialize_res_quotas(dev, res_alloc, RES_CQ,
+ t, dev->caps.num_cqs -
+ dev->caps.reserved_cqs);
+ break;
+ case RES_SRQ:
+ initialize_res_quotas(dev, res_alloc, RES_SRQ,
+ t, dev->caps.num_srqs -
+ dev->caps.reserved_srqs);
+ break;
+ case RES_MPT:
+ initialize_res_quotas(dev, res_alloc, RES_MPT,
+ t, dev->caps.num_mpts -
+ dev->caps.reserved_mrws);
+ break;
+ case RES_MTT:
+ initialize_res_quotas(dev, res_alloc, RES_MTT,
+ t, dev->caps.num_mtts -
+ dev->caps.reserved_mtts);
+ break;
+ case RES_MAC:
+ if (t == mlx4_master_func_num(dev)) {
+ res_alloc->quota[t] = MLX4_MAX_MAC_NUM;
+ res_alloc->guaranteed[t] = 2;
+ for (j = 0; j < MLX4_MAX_PORTS; j++)
+ res_alloc->res_port_free[j] = MLX4_MAX_MAC_NUM;
+ } else {
+ res_alloc->quota[t] = MLX4_MAX_MAC_NUM;
+ res_alloc->guaranteed[t] = 2;
+ }
+ break;
+ case RES_VLAN:
+ if (t == mlx4_master_func_num(dev)) {
+ res_alloc->quota[t] = MLX4_MAX_VLAN_NUM;
+ res_alloc->guaranteed[t] = MLX4_MAX_VLAN_NUM / 2;
+ for (j = 0; j < MLX4_MAX_PORTS; j++)
+ res_alloc->res_port_free[j] =
+ res_alloc->quota[t];
+ } else {
+ res_alloc->quota[t] = MLX4_MAX_VLAN_NUM / 2;
+ res_alloc->guaranteed[t] = 0;
+ }
+ break;
+ case RES_COUNTER:
+ res_alloc->quota[t] = dev->caps.max_counters;
+ res_alloc->guaranteed[t] = 0;
+ if (t == mlx4_master_func_num(dev))
+ res_alloc->res_free = res_alloc->quota[t];
+ break;
+ default:
+ break;
+ }
+ if (i == RES_MAC || i == RES_VLAN) {
+ for (j = 0; j < MLX4_MAX_PORTS; j++)
+ res_alloc->res_port_rsvd[j] +=
+ res_alloc->guaranteed[t];
+ } else {
+ res_alloc->res_reserved += res_alloc->guaranteed[t];
+ }
+ }
+ }
spin_lock_init(&priv->mfunc.master.res_tracker.lock);
- return 0 ;
+ return 0;
+
+no_mem_err:
+ for (i = 0; i < MLX4_NUM_OF_RESOURCE_TYPE; i++) {
+ kfree(priv->mfunc.master.res_tracker.res_alloc[i].allocated);
+ priv->mfunc.master.res_tracker.res_alloc[i].allocated = NULL;
+ kfree(priv->mfunc.master.res_tracker.res_alloc[i].guaranteed);
+ priv->mfunc.master.res_tracker.res_alloc[i].guaranteed = NULL;
+ kfree(priv->mfunc.master.res_tracker.res_alloc[i].quota);
+ priv->mfunc.master.res_tracker.res_alloc[i].quota = NULL;
+ }
+ return -ENOMEM;
}
void mlx4_free_resource_tracker(struct mlx4_dev *dev,
@@ -333,6 +478,14 @@ void mlx4_free_resource_tracker(struct mlx4_dev *dev,
}
if (type != RES_TR_FREE_SLAVES_ONLY) {
+ for (i = 0; i < MLX4_NUM_OF_RESOURCE_TYPE; i++) {
+ kfree(priv->mfunc.master.res_tracker.res_alloc[i].allocated);
+ priv->mfunc.master.res_tracker.res_alloc[i].allocated = NULL;
+ kfree(priv->mfunc.master.res_tracker.res_alloc[i].guaranteed);
+ priv->mfunc.master.res_tracker.res_alloc[i].guaranteed = NULL;
+ kfree(priv->mfunc.master.res_tracker.res_alloc[i].quota);
+ priv->mfunc.master.res_tracker.res_alloc[i].quota = NULL;
+ }
kfree(priv->mfunc.master.res_tracker.slave_list);
priv->mfunc.master.res_tracker.slave_list = NULL;
}
diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index e2e9288..f6f5927 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -641,12 +641,23 @@ struct mlx4_counter {
__be64 tx_bytes;
};
+struct mlx4_quotas {
+ int qp;
+ int cq;
+ int srq;
+ int mpt;
+ int mtt;
+ int counter;
+ int xrcd;
+};
+
struct mlx4_dev {
struct pci_dev *pdev;
unsigned long flags;
unsigned long num_slaves;
struct mlx4_caps caps;
struct mlx4_phys_caps phys_caps;
+ struct mlx4_quotas quotas;
struct radix_tree_root qp_table_tree;
u8 rev_id;
char board_id[MLX4_BOARD_ID_LEN];
@@ -772,6 +783,12 @@ static inline int mlx4_is_master(struct mlx4_dev *dev)
return dev->flags & MLX4_FLAG_MASTER;
}
+static inline int mlx4_num_reserved_sqps(struct mlx4_dev *dev)
+{
+ return dev->phys_caps.base_sqpn + 8 +
+ 16 * MLX4_MFUNC_MAX * !!mlx4_is_master(dev);
+}
+
static inline int mlx4_is_qp_reserved(struct mlx4_dev *dev, u32 qpn)
{
return (qpn < dev->phys_caps.base_sqpn + 8 +
--
1.7.1
^ permalink raw reply related
* [PATCH net-next 03/10] net/mlx4_core: Fix reg/unreg vlan/mac to conform to the firmware spec
From: Or Gerlitz @ 2013-10-31 12:29 UTC (permalink / raw)
To: davem; +Cc: netdev, amirv, roland, jackm, Or Gerlitz
In-Reply-To: <1383222569-10930-1-git-send-email-ogerlitz@mellanox.com>
From: Jack Morgenstein <jackm@dev.mellanox.co.il>
The functions mlx4_register_vlan, mlx4_unregister_vlan, mlx4_register_mac,
mlx4_unregister_mac all made illegal use of the out_param in multifunc mode
to pass the port number. The firmware spec specifies that the port number
should be passed in bits 8..15 of the input-modifier field for ALLOC_RES and
FREE_RES (sections 20.15.1 and 20.15.2).
For MAC register/unregister, this patch contains workarounds so that guests
running previous kernels continue to work on a new Hypervisor, and guests
running the new kernel will continue to work on old hypervisors.
Vlan registeration capability is still not operational in multifunction mode,
since the vlan wrapper functions are not implemented in this patch.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/port.c | 45 ++++++++++++++------
.../net/ethernet/mellanox/mlx4/resource_tracker.c | 28 +++++++-----
include/linux/mlx4/device.h | 1 +
3 files changed, 49 insertions(+), 25 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/port.c b/drivers/net/ethernet/mellanox/mlx4/port.c
index d3d3106..9433c1f 100644
--- a/drivers/net/ethernet/mellanox/mlx4/port.c
+++ b/drivers/net/ethernet/mellanox/mlx4/port.c
@@ -178,13 +178,24 @@ EXPORT_SYMBOL_GPL(__mlx4_register_mac);
int mlx4_register_mac(struct mlx4_dev *dev, u8 port, u64 mac)
{
u64 out_param = 0;
- int err;
+ int err = -EINVAL;
if (mlx4_is_mfunc(dev)) {
- set_param_l(&out_param, port);
- err = mlx4_cmd_imm(dev, mac, &out_param, RES_MAC,
- RES_OP_RESERVE_AND_MAP, MLX4_CMD_ALLOC_RES,
- MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+ if (!(dev->flags & MLX4_FLAG_OLD_REG_MAC)) {
+ err = mlx4_cmd_imm(dev, mac, &out_param,
+ ((u32) port) << 8 | (u32) RES_MAC,
+ RES_OP_RESERVE_AND_MAP, MLX4_CMD_ALLOC_RES,
+ MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+ }
+ if (err && err == -EINVAL && mlx4_is_slave(dev)) {
+ /* retry using old REG_MAC format */
+ set_param_l(&out_param, port);
+ err = mlx4_cmd_imm(dev, mac, &out_param, RES_MAC,
+ RES_OP_RESERVE_AND_MAP, MLX4_CMD_ALLOC_RES,
+ MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+ if (!err)
+ dev->flags |= MLX4_FLAG_OLD_REG_MAC;
+ }
if (err)
return err;
@@ -231,10 +242,18 @@ void mlx4_unregister_mac(struct mlx4_dev *dev, u8 port, u64 mac)
u64 out_param = 0;
if (mlx4_is_mfunc(dev)) {
- set_param_l(&out_param, port);
- (void) mlx4_cmd_imm(dev, mac, &out_param, RES_MAC,
- RES_OP_RESERVE_AND_MAP, MLX4_CMD_FREE_RES,
- MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+ if (!(dev->flags & MLX4_FLAG_OLD_REG_MAC)) {
+ (void) mlx4_cmd_imm(dev, mac, &out_param,
+ ((u32) port) << 8 | (u32) RES_MAC,
+ RES_OP_RESERVE_AND_MAP, MLX4_CMD_FREE_RES,
+ MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+ } else {
+ /* use old unregister mac format */
+ set_param_l(&out_param, port);
+ (void) mlx4_cmd_imm(dev, mac, &out_param, RES_MAC,
+ RES_OP_RESERVE_AND_MAP, MLX4_CMD_FREE_RES,
+ MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
+ }
return;
}
__mlx4_unregister_mac(dev, port, mac);
@@ -374,8 +393,8 @@ int mlx4_register_vlan(struct mlx4_dev *dev, u8 port, u16 vlan, int *index)
return -EINVAL;
if (mlx4_is_mfunc(dev)) {
- set_param_l(&out_param, port);
- err = mlx4_cmd_imm(dev, vlan, &out_param, RES_VLAN,
+ err = mlx4_cmd_imm(dev, vlan, &out_param,
+ ((u32) port) << 8 | (u32) RES_VLAN,
RES_OP_RESERVE_AND_MAP, MLX4_CMD_ALLOC_RES,
MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
if (!err)
@@ -418,8 +437,8 @@ void mlx4_unregister_vlan(struct mlx4_dev *dev, u8 port, int index)
u64 out_param = 0;
if (mlx4_is_mfunc(dev)) {
- set_param_l(&out_param, port);
- (void) mlx4_cmd_imm(dev, index, &out_param, RES_VLAN,
+ (void) mlx4_cmd_imm(dev, index, &out_param,
+ ((u32) port) << 8 | (u32) RES_VLAN,
RES_OP_RESERVE_AND_MAP,
MLX4_CMD_FREE_RES, MLX4_CMD_TIME_CLASS_A,
MLX4_CMD_WRAPPED);
diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
index dd68763..a5aa3be 100644
--- a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
+++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
@@ -1443,7 +1443,7 @@ static void rem_slave_macs(struct mlx4_dev *dev, int slave)
}
static int mac_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
- u64 in_param, u64 *out_param)
+ u64 in_param, u64 *out_param, int in_port)
{
int err = -EINVAL;
int port;
@@ -1452,7 +1452,7 @@ static int mac_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
if (op != RES_OP_RESERVE_AND_MAP)
return err;
- port = get_param_l(out_param);
+ port = !in_port ? get_param_l(out_param) : in_port;
mac = in_param;
err = __mlx4_register_mac(dev, port, mac);
@@ -1470,7 +1470,7 @@ static int mac_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
}
static int vlan_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
- u64 in_param, u64 *out_param)
+ u64 in_param, u64 *out_param, int port)
{
return 0;
}
@@ -1528,7 +1528,7 @@ int mlx4_ALLOC_RES_wrapper(struct mlx4_dev *dev, int slave,
int err;
int alop = vhcr->op_modifier;
- switch (vhcr->in_modifier) {
+ switch (vhcr->in_modifier & 0xFF) {
case RES_QP:
err = qp_alloc_res(dev, slave, vhcr->op_modifier, alop,
vhcr->in_param, &vhcr->out_param);
@@ -1556,12 +1556,14 @@ int mlx4_ALLOC_RES_wrapper(struct mlx4_dev *dev, int slave,
case RES_MAC:
err = mac_alloc_res(dev, slave, vhcr->op_modifier, alop,
- vhcr->in_param, &vhcr->out_param);
+ vhcr->in_param, &vhcr->out_param,
+ (vhcr->in_modifier >> 8) & 0xFF);
break;
case RES_VLAN:
err = vlan_alloc_res(dev, slave, vhcr->op_modifier, alop,
- vhcr->in_param, &vhcr->out_param);
+ vhcr->in_param, &vhcr->out_param,
+ (vhcr->in_modifier >> 8) & 0xFF);
break;
case RES_COUNTER:
@@ -1730,14 +1732,14 @@ static int srq_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
}
static int mac_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
- u64 in_param, u64 *out_param)
+ u64 in_param, u64 *out_param, int in_port)
{
int port;
int err = 0;
switch (op) {
case RES_OP_RESERVE_AND_MAP:
- port = get_param_l(out_param);
+ port = !in_port ? get_param_l(out_param) : in_port;
mac_del_from_slave(dev, slave, in_param, port);
__mlx4_unregister_mac(dev, port, in_param);
break;
@@ -1751,7 +1753,7 @@ static int mac_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
}
static int vlan_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
- u64 in_param, u64 *out_param)
+ u64 in_param, u64 *out_param, int port)
{
return 0;
}
@@ -1803,7 +1805,7 @@ int mlx4_FREE_RES_wrapper(struct mlx4_dev *dev, int slave,
int err = -EINVAL;
int alop = vhcr->op_modifier;
- switch (vhcr->in_modifier) {
+ switch (vhcr->in_modifier & 0xFF) {
case RES_QP:
err = qp_free_res(dev, slave, vhcr->op_modifier, alop,
vhcr->in_param);
@@ -1831,12 +1833,14 @@ int mlx4_FREE_RES_wrapper(struct mlx4_dev *dev, int slave,
case RES_MAC:
err = mac_free_res(dev, slave, vhcr->op_modifier, alop,
- vhcr->in_param, &vhcr->out_param);
+ vhcr->in_param, &vhcr->out_param,
+ (vhcr->in_modifier >> 8) & 0xFF);
break;
case RES_VLAN:
err = vlan_free_res(dev, slave, vhcr->op_modifier, alop,
- vhcr->in_param, &vhcr->out_param);
+ vhcr->in_param, &vhcr->out_param,
+ (vhcr->in_modifier >> 8) & 0xFF);
break;
case RES_COUNTER:
diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index 9ad0c18..297a163 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -54,6 +54,7 @@ enum {
MLX4_FLAG_MASTER = 1 << 2,
MLX4_FLAG_SLAVE = 1 << 3,
MLX4_FLAG_SRIOV = 1 << 4,
+ MLX4_FLAG_OLD_REG_MAC = 1 << 6,
};
enum {
--
1.7.1
^ permalink raw reply related
* [PATCH net-next 00/10] Mellanox driver updates Oct 31 2013
From: Or Gerlitz @ 2013-10-31 12:29 UTC (permalink / raw)
To: davem; +Cc: netdev, amirv, roland, jackm, Or Gerlitz
Hi Dave,
This patch set from Jack Morgenstein does the following:
1. Fix MAC/VLAN SRIOV implementation, and add wrapper functions for VLAN allocation
and de-allocation (patches 1-6).
2. Implements resource quotas when running under SRIOV (patches 7-10).
Patch 7 is a small bug fix, and patches 8-10 implement the quotas.
Quotas are implemented per resource type for VFs and the PF, to prevent
any entity from simply grabbing all the resources for itself and leaving
the other entities unable to obtain such resources.
The series is against net-next commit ba48650 "ipv6: remove the unnecessary statement in find_match()"
Patch #1 is bug fix which needs to go into -stable of kernels >= 3.10
Or.
Jack Morgenstein (10):
net/mlx4_core: Fix call to __mlx4_unregister_mac
net/mlx4_core: Fix register/unreg vlan flow
net/mlx4_core: Fix reg/unreg vlan/mac to conform to the firmware spec
net/mlx4_en: Use vlan id instead of vlan index for unregistration
net/mlx4_core: Resource tracker for reg/unreg vlans
net/mlx4_core: Don't fail reg/unreg vlan for older guests
net/mlx4_core: Fix checking order in MR table init
mlx4: Structures and init/teardown for VF resource quotas
net/mlx4_core: Fix quota handling in the QUERY_FUNC_CAP wrapper
net/mlx4_core: Implement resource quota enforcement
drivers/infiniband/hw/mlx4/main.c | 8 +-
drivers/net/ethernet/mellanox/mlx4/cmd.c | 5 +-
drivers/net/ethernet/mellanox/mlx4/en_netdev.c | 6 +-
drivers/net/ethernet/mellanox/mlx4/fw.c | 99 +++--
drivers/net/ethernet/mellanox/mlx4/main.c | 32 +-
drivers/net/ethernet/mellanox/mlx4/mlx4.h | 21 +-
drivers/net/ethernet/mellanox/mlx4/mr.c | 6 +-
drivers/net/ethernet/mellanox/mlx4/port.c | 87 ++--
drivers/net/ethernet/mellanox/mlx4/qp.c | 3 +-
.../net/ethernet/mellanox/mlx4/resource_tracker.c | 510 ++++++++++++++++++--
include/linux/mlx4/device.h | 20 +-
11 files changed, 676 insertions(+), 121 deletions(-)
^ permalink raw reply
* [PATCH net-next 09/10] net/mlx4_core: Fix quota handling in the QUERY_FUNC_CAP wrapper
From: Or Gerlitz @ 2013-10-31 12:29 UTC (permalink / raw)
To: davem; +Cc: netdev, amirv, roland, jackm, Or Gerlitz
In-Reply-To: <1383222569-10930-1-git-send-email-ogerlitz@mellanox.com>
From: Jack Morgenstein <jackm@dev.mellanox.co.il>
In current kernels, the mlx4 driver running on a VM does not
differentiate between max resource numbers for the HCA and
max quotas -- it simply takes the quota values passed to it
as max-resource values.
However, the driver actually requires the VFs to be aware of
the actual number of resources that the HCA was initialized with,
for QPs, CQs, SRQs and MPTs.
For QPs, CQs and SRQs, the reason is that in completion handling
the driver must know which of the 24 bits are the actual resource
number, and which are "padding" bits.
For MPTs, also, the driver assumes knowledge of the number of MPTs
in the system.
The previous commit fixes the quota logic on the VM for the quota values
passed to it by QUERY_FUNC_CAPS.
For QPs, CQs, SRQs, and MPTs, it takes the max resource numbers
from QUERY_HCA (and not QUERY_FUNC_CAPS). The quotas passed
in QUERY_FUNC_CAPS are used to report max resource number values
in the response to ib_query_device.
However, the Hypervisor driver must consider that VMs
may be running previous kernels, and compatibility must be preserved.
To resolve the incompatibility with previous kernels running on VMs,
we deprecated the quota fields in mlx4_QUERY_FUNC_CAP. In the
deprecated fields, we pass the max-resource values from INIT_HCA
The quota fields are moved to a new location, and the current kernel
driver takes the proper values from that location. There is
also a new flag in dword 0, bit 28 of the mlx4_QUERY_FUNC_CAP mailbox;
if this flag is set, the (VM) driver takes the quota values from the
new location.
VMs running previous kernels will work properly, except that the max resource
numbers reported in ib_query_device for these resources will be
too high. The Hypervisor driver will, however, enforce the quotas
for these VMs.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/fw.c | 88 +++++++++++++++++++++++--------
1 files changed, 65 insertions(+), 23 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/fw.c b/drivers/net/ethernet/mellanox/mlx4/fw.c
index f8c88c3..c3e70bc 100644
--- a/drivers/net/ethernet/mellanox/mlx4/fw.c
+++ b/drivers/net/ethernet/mellanox/mlx4/fw.c
@@ -186,18 +186,26 @@ int mlx4_QUERY_FUNC_CAP_wrapper(struct mlx4_dev *dev, int slave,
#define QUERY_FUNC_CAP_NUM_PORTS_OFFSET 0x1
#define QUERY_FUNC_CAP_PF_BHVR_OFFSET 0x4
#define QUERY_FUNC_CAP_FMR_OFFSET 0x8
-#define QUERY_FUNC_CAP_QP_QUOTA_OFFSET 0x10
-#define QUERY_FUNC_CAP_CQ_QUOTA_OFFSET 0x14
-#define QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET 0x18
-#define QUERY_FUNC_CAP_MPT_QUOTA_OFFSET 0x20
-#define QUERY_FUNC_CAP_MTT_QUOTA_OFFSET 0x24
-#define QUERY_FUNC_CAP_MCG_QUOTA_OFFSET 0x28
+#define QUERY_FUNC_CAP_QP_QUOTA_OFFSET_DEP 0x10
+#define QUERY_FUNC_CAP_CQ_QUOTA_OFFSET_DEP 0x14
+#define QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET_DEP 0x18
+#define QUERY_FUNC_CAP_MPT_QUOTA_OFFSET_DEP 0x20
+#define QUERY_FUNC_CAP_MTT_QUOTA_OFFSET_DEP 0x24
+#define QUERY_FUNC_CAP_MCG_QUOTA_OFFSET_DEP 0x28
#define QUERY_FUNC_CAP_MAX_EQ_OFFSET 0x2c
#define QUERY_FUNC_CAP_RESERVED_EQ_OFFSET 0x30
+#define QUERY_FUNC_CAP_QP_QUOTA_OFFSET 0x50
+#define QUERY_FUNC_CAP_CQ_QUOTA_OFFSET 0x54
+#define QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET 0x58
+#define QUERY_FUNC_CAP_MPT_QUOTA_OFFSET 0x60
+#define QUERY_FUNC_CAP_MTT_QUOTA_OFFSET 0x64
+#define QUERY_FUNC_CAP_MCG_QUOTA_OFFSET 0x68
+
#define QUERY_FUNC_CAP_FMR_FLAG 0x80
#define QUERY_FUNC_CAP_FLAG_RDMA 0x40
#define QUERY_FUNC_CAP_FLAG_ETH 0x80
+#define QUERY_FUNC_CAP_FLAG_QUOTAS 0x10
/* when opcode modifier = 1 */
#define QUERY_FUNC_CAP_PHYS_PORT_OFFSET 0x3
@@ -238,8 +246,9 @@ int mlx4_QUERY_FUNC_CAP_wrapper(struct mlx4_dev *dev, int slave,
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_QP1_PROXY);
} else if (vhcr->op_modifier == 0) {
- /* enable rdma and ethernet interfaces */
- field = (QUERY_FUNC_CAP_FLAG_ETH | QUERY_FUNC_CAP_FLAG_RDMA);
+ /* enable rdma and ethernet interfaces, and new quota locations */
+ field = (QUERY_FUNC_CAP_FLAG_ETH | QUERY_FUNC_CAP_FLAG_RDMA |
+ QUERY_FUNC_CAP_FLAG_QUOTAS);
MLX4_PUT(outbox->buf, field, QUERY_FUNC_CAP_FLAGS_OFFSET);
field = dev->caps.num_ports;
@@ -253,12 +262,18 @@ int mlx4_QUERY_FUNC_CAP_wrapper(struct mlx4_dev *dev, int slave,
size = priv->mfunc.master.res_tracker.res_alloc[RES_QP].quota[slave];
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_QP_QUOTA_OFFSET);
+ size = dev->caps.num_qps;
+ MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_QP_QUOTA_OFFSET_DEP);
size = priv->mfunc.master.res_tracker.res_alloc[RES_SRQ].quota[slave];
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET);
+ size = dev->caps.num_srqs;
+ MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET_DEP);
size = priv->mfunc.master.res_tracker.res_alloc[RES_CQ].quota[slave];
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_CQ_QUOTA_OFFSET);
+ size = dev->caps.num_cqs;
+ MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_CQ_QUOTA_OFFSET_DEP);
size = dev->caps.num_eqs;
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MAX_EQ_OFFSET);
@@ -268,12 +283,17 @@ int mlx4_QUERY_FUNC_CAP_wrapper(struct mlx4_dev *dev, int slave,
size = priv->mfunc.master.res_tracker.res_alloc[RES_MPT].quota[slave];
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MPT_QUOTA_OFFSET);
+ size = dev->caps.num_mpts;
+ MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MPT_QUOTA_OFFSET_DEP);
size = priv->mfunc.master.res_tracker.res_alloc[RES_MTT].quota[slave];
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MTT_QUOTA_OFFSET);
+ size = dev->caps.num_mtts;
+ MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MTT_QUOTA_OFFSET_DEP);
size = dev->caps.num_mgms + dev->caps.num_amgms;
MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MCG_QUOTA_OFFSET);
+ MLX4_PUT(outbox->buf, size, QUERY_FUNC_CAP_MCG_QUOTA_OFFSET_DEP);
} else
err = -EINVAL;
@@ -288,7 +308,7 @@ int mlx4_QUERY_FUNC_CAP(struct mlx4_dev *dev, u32 gen_or_port,
u32 *outbox;
u8 field, op_modifier;
u32 size;
- int err = 0;
+ int err = 0, quotas = 0;
op_modifier = !!gen_or_port; /* 0 = general, 1 = logical port */
@@ -312,6 +332,7 @@ int mlx4_QUERY_FUNC_CAP(struct mlx4_dev *dev, u32 gen_or_port,
goto out;
}
func_cap->flags = field;
+ quotas = !!(func_cap->flags & QUERY_FUNC_CAP_FLAG_QUOTAS);
MLX4_GET(field, outbox, QUERY_FUNC_CAP_NUM_PORTS_OFFSET);
func_cap->num_ports = field;
@@ -319,29 +340,50 @@ int mlx4_QUERY_FUNC_CAP(struct mlx4_dev *dev, u32 gen_or_port,
MLX4_GET(size, outbox, QUERY_FUNC_CAP_PF_BHVR_OFFSET);
func_cap->pf_context_behaviour = size;
- MLX4_GET(size, outbox, QUERY_FUNC_CAP_QP_QUOTA_OFFSET);
- func_cap->qp_quota = size & 0xFFFFFF;
+ if (quotas) {
+ MLX4_GET(size, outbox, QUERY_FUNC_CAP_QP_QUOTA_OFFSET);
+ func_cap->qp_quota = size & 0xFFFFFF;
+
+ MLX4_GET(size, outbox, QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET);
+ func_cap->srq_quota = size & 0xFFFFFF;
+
+ MLX4_GET(size, outbox, QUERY_FUNC_CAP_CQ_QUOTA_OFFSET);
+ func_cap->cq_quota = size & 0xFFFFFF;
+
+ MLX4_GET(size, outbox, QUERY_FUNC_CAP_MPT_QUOTA_OFFSET);
+ func_cap->mpt_quota = size & 0xFFFFFF;
+
+ MLX4_GET(size, outbox, QUERY_FUNC_CAP_MTT_QUOTA_OFFSET);
+ func_cap->mtt_quota = size & 0xFFFFFF;
+
+ MLX4_GET(size, outbox, QUERY_FUNC_CAP_MCG_QUOTA_OFFSET);
+ func_cap->mcg_quota = size & 0xFFFFFF;
- MLX4_GET(size, outbox, QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET);
- func_cap->srq_quota = size & 0xFFFFFF;
+ } else {
+ MLX4_GET(size, outbox, QUERY_FUNC_CAP_QP_QUOTA_OFFSET_DEP);
+ func_cap->qp_quota = size & 0xFFFFFF;
- MLX4_GET(size, outbox, QUERY_FUNC_CAP_CQ_QUOTA_OFFSET);
- func_cap->cq_quota = size & 0xFFFFFF;
+ MLX4_GET(size, outbox, QUERY_FUNC_CAP_SRQ_QUOTA_OFFSET_DEP);
+ func_cap->srq_quota = size & 0xFFFFFF;
+ MLX4_GET(size, outbox, QUERY_FUNC_CAP_CQ_QUOTA_OFFSET_DEP);
+ func_cap->cq_quota = size & 0xFFFFFF;
+
+ MLX4_GET(size, outbox, QUERY_FUNC_CAP_MPT_QUOTA_OFFSET_DEP);
+ func_cap->mpt_quota = size & 0xFFFFFF;
+
+ MLX4_GET(size, outbox, QUERY_FUNC_CAP_MTT_QUOTA_OFFSET_DEP);
+ func_cap->mtt_quota = size & 0xFFFFFF;
+
+ MLX4_GET(size, outbox, QUERY_FUNC_CAP_MCG_QUOTA_OFFSET_DEP);
+ func_cap->mcg_quota = size & 0xFFFFFF;
+ }
MLX4_GET(size, outbox, QUERY_FUNC_CAP_MAX_EQ_OFFSET);
func_cap->max_eq = size & 0xFFFFFF;
MLX4_GET(size, outbox, QUERY_FUNC_CAP_RESERVED_EQ_OFFSET);
func_cap->reserved_eq = size & 0xFFFFFF;
- MLX4_GET(size, outbox, QUERY_FUNC_CAP_MPT_QUOTA_OFFSET);
- func_cap->mpt_quota = size & 0xFFFFFF;
-
- MLX4_GET(size, outbox, QUERY_FUNC_CAP_MTT_QUOTA_OFFSET);
- func_cap->mtt_quota = size & 0xFFFFFF;
-
- MLX4_GET(size, outbox, QUERY_FUNC_CAP_MCG_QUOTA_OFFSET);
- func_cap->mcg_quota = size & 0xFFFFFF;
goto out;
}
--
1.7.1
^ permalink raw reply related
* [PATCH net-next 10/10] net/mlx4_core: Implement resource quota enforcement
From: Or Gerlitz @ 2013-10-31 12:29 UTC (permalink / raw)
To: davem; +Cc: netdev, amirv, roland, jackm, Or Gerlitz
In-Reply-To: <1383222569-10930-1-git-send-email-ogerlitz@mellanox.com>
From: Jack Morgenstein <jackm@dev.mellanox.co.il>
Implements resource quota grant decision when resources are requested,
for the following resources: QPs, CQs, SRQs, MPTs, MTTs, vlans, MACs,
and Counters.
When granting a resource, the quota system increases the allocated-count
for that slave.
When the slave later frees the resource, its allocated-count is reduced.
A spinlock is used to protect the integrity of each resource's free-pool counter.
(One slave may be in the process of being granted a resource while another
slave has crashed, initiating cleanup of that slave's resource quotas).
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
drivers/net/ethernet/mellanox/mlx4/mlx4.h | 1 +
.../net/ethernet/mellanox/mlx4/resource_tracker.c | 185 ++++++++++++++++++--
2 files changed, 173 insertions(+), 13 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4.h b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
index e7eb86e..e582a41 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mlx4.h
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
@@ -505,6 +505,7 @@ struct slave_list {
};
struct resource_allocator {
+ spinlock_t alloc_lock; /* protect quotas */
union {
int res_reserved;
int res_port_rsvd[MLX4_MAX_PORTS];
diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
index cc5d6d0..b1603e2 100644
--- a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
+++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
@@ -284,6 +284,85 @@ static const char *ResourceType(enum mlx4_resource rt)
}
static void rem_slave_vlans(struct mlx4_dev *dev, int slave);
+static inline int mlx4_grant_resource(struct mlx4_dev *dev, int slave,
+ enum mlx4_resource res_type, int count,
+ int port)
+{
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ struct resource_allocator *res_alloc =
+ &priv->mfunc.master.res_tracker.res_alloc[res_type];
+ int err = -EINVAL;
+ int allocated, free, reserved, guaranteed, from_free;
+
+ if (slave > dev->num_vfs)
+ return -EINVAL;
+
+ spin_lock(&res_alloc->alloc_lock);
+ allocated = (port > 0) ?
+ res_alloc->allocated[(port - 1) * (dev->num_vfs + 1) + slave] :
+ res_alloc->allocated[slave];
+ free = (port > 0) ? res_alloc->res_port_free[port - 1] :
+ res_alloc->res_free;
+ reserved = (port > 0) ? res_alloc->res_port_rsvd[port - 1] :
+ res_alloc->res_reserved;
+ guaranteed = res_alloc->guaranteed[slave];
+
+ if (allocated + count > res_alloc->quota[slave])
+ goto out;
+
+ if (allocated + count <= guaranteed) {
+ err = 0;
+ } else {
+ /* portion may need to be obtained from free area */
+ if (guaranteed - allocated > 0)
+ from_free = count - (guaranteed - allocated);
+ else
+ from_free = count;
+
+ if (free - from_free > reserved)
+ err = 0;
+ }
+
+ if (!err) {
+ /* grant the request */
+ if (port > 0) {
+ res_alloc->allocated[(port - 1) * (dev->num_vfs + 1) + slave] += count;
+ res_alloc->res_port_free[port - 1] -= count;
+ } else {
+ res_alloc->allocated[slave] += count;
+ res_alloc->res_free -= count;
+ }
+ }
+
+out:
+ spin_unlock(&res_alloc->alloc_lock);
+ return err;
+}
+
+static inline void mlx4_release_resource(struct mlx4_dev *dev, int slave,
+ enum mlx4_resource res_type, int count,
+ int port)
+{
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ struct resource_allocator *res_alloc =
+ &priv->mfunc.master.res_tracker.res_alloc[res_type];
+
+ if (slave > dev->num_vfs)
+ return;
+
+ spin_lock(&res_alloc->alloc_lock);
+ if (port > 0) {
+ res_alloc->allocated[(port - 1) * (dev->num_vfs + 1) + slave] -= count;
+ res_alloc->res_port_free[port - 1] += count;
+ } else {
+ res_alloc->allocated[slave] -= count;
+ res_alloc->res_free += count;
+ }
+
+ spin_unlock(&res_alloc->alloc_lock);
+ return;
+}
+
static inline void initialize_res_quotas(struct mlx4_dev *dev,
struct resource_allocator *res_alloc,
enum mlx4_resource res_type,
@@ -373,6 +452,7 @@ int mlx4_init_resource_tracker(struct mlx4_dev *dev)
!res_alloc->allocated)
goto no_mem_err;
+ spin_lock_init(&res_alloc->alloc_lock);
for (t = 0; t < dev->num_vfs + 1; t++) {
switch (i) {
case RES_QP:
@@ -1399,12 +1479,19 @@ static int qp_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
case RES_OP_RESERVE:
count = get_param_l(&in_param);
align = get_param_h(&in_param);
- err = __mlx4_qp_reserve_range(dev, count, align, &base);
+ err = mlx4_grant_resource(dev, slave, RES_QP, count, 0);
if (err)
return err;
+ err = __mlx4_qp_reserve_range(dev, count, align, &base);
+ if (err) {
+ mlx4_release_resource(dev, slave, RES_QP, count, 0);
+ return err;
+ }
+
err = add_res_range(dev, slave, base, count, RES_QP, 0);
if (err) {
+ mlx4_release_resource(dev, slave, RES_QP, count, 0);
__mlx4_qp_release_range(dev, base, count);
return err;
}
@@ -1452,15 +1539,24 @@ static int mtt_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
return err;
order = get_param_l(&in_param);
+
+ err = mlx4_grant_resource(dev, slave, RES_MTT, 1 << order, 0);
+ if (err)
+ return err;
+
base = __mlx4_alloc_mtt_range(dev, order);
- if (base == -1)
+ if (base == -1) {
+ mlx4_release_resource(dev, slave, RES_MTT, 1 << order, 0);
return -ENOMEM;
+ }
err = add_res_range(dev, slave, base, 1, RES_MTT, order);
- if (err)
+ if (err) {
+ mlx4_release_resource(dev, slave, RES_MTT, 1 << order, 0);
__mlx4_free_mtt_range(dev, base, order);
- else
+ } else {
set_param_l(out_param, base);
+ }
return err;
}
@@ -1475,13 +1571,20 @@ static int mpt_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
switch (op) {
case RES_OP_RESERVE:
+ err = mlx4_grant_resource(dev, slave, RES_MPT, 1, 0);
+ if (err)
+ break;
+
index = __mlx4_mpt_reserve(dev);
- if (index == -1)
+ if (index == -1) {
+ mlx4_release_resource(dev, slave, RES_MPT, 1, 0);
break;
+ }
id = index & mpt_mask(dev);
err = add_res_range(dev, slave, id, 1, RES_MPT, index);
if (err) {
+ mlx4_release_resource(dev, slave, RES_MPT, 1, 0);
__mlx4_mpt_release(dev, index);
break;
}
@@ -1515,12 +1618,19 @@ static int cq_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
switch (op) {
case RES_OP_RESERVE_AND_MAP:
- err = __mlx4_cq_alloc_icm(dev, &cqn);
+ err = mlx4_grant_resource(dev, slave, RES_CQ, 1, 0);
if (err)
break;
+ err = __mlx4_cq_alloc_icm(dev, &cqn);
+ if (err) {
+ mlx4_release_resource(dev, slave, RES_CQ, 1, 0);
+ break;
+ }
+
err = add_res_range(dev, slave, cqn, 1, RES_CQ, 0);
if (err) {
+ mlx4_release_resource(dev, slave, RES_CQ, 1, 0);
__mlx4_cq_free_icm(dev, cqn);
break;
}
@@ -1543,12 +1653,19 @@ static int srq_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
switch (op) {
case RES_OP_RESERVE_AND_MAP:
- err = __mlx4_srq_alloc_icm(dev, &srqn);
+ err = mlx4_grant_resource(dev, slave, RES_SRQ, 1, 0);
if (err)
break;
+ err = __mlx4_srq_alloc_icm(dev, &srqn);
+ if (err) {
+ mlx4_release_resource(dev, slave, RES_SRQ, 1, 0);
+ break;
+ }
+
err = add_res_range(dev, slave, srqn, 1, RES_SRQ, 0);
if (err) {
+ mlx4_release_resource(dev, slave, RES_SRQ, 1, 0);
__mlx4_srq_free_icm(dev, srqn);
break;
}
@@ -1569,9 +1686,13 @@ static int mac_add_to_slave(struct mlx4_dev *dev, int slave, u64 mac, int port)
struct mlx4_resource_tracker *tracker = &priv->mfunc.master.res_tracker;
struct mac_res *res;
+ if (mlx4_grant_resource(dev, slave, RES_MAC, 1, port))
+ return -EINVAL;
res = kzalloc(sizeof *res, GFP_KERNEL);
- if (!res)
+ if (!res) {
+ mlx4_release_resource(dev, slave, RES_MAC, 1, port);
return -ENOMEM;
+ }
res->mac = mac;
res->port = (u8) port;
list_add_tail(&res->list,
@@ -1591,6 +1712,7 @@ static void mac_del_from_slave(struct mlx4_dev *dev, int slave, u64 mac,
list_for_each_entry_safe(res, tmp, mac_list, list) {
if (res->mac == mac && res->port == (u8) port) {
list_del(&res->list);
+ mlx4_release_resource(dev, slave, RES_MAC, 1, port);
kfree(res);
break;
}
@@ -1608,6 +1730,7 @@ static void rem_slave_macs(struct mlx4_dev *dev, int slave)
list_for_each_entry_safe(res, tmp, mac_list, list) {
list_del(&res->list);
__mlx4_unregister_mac(dev, res->port, res->mac);
+ mlx4_release_resource(dev, slave, RES_MAC, 1, res->port);
kfree(res);
}
}
@@ -1656,9 +1779,13 @@ static int vlan_add_to_slave(struct mlx4_dev *dev, int slave, u16 vlan,
}
}
+ if (mlx4_grant_resource(dev, slave, RES_VLAN, 1, port))
+ return -EINVAL;
res = kzalloc(sizeof(*res), GFP_KERNEL);
- if (!res)
+ if (!res) {
+ mlx4_release_resource(dev, slave, RES_VLAN, 1, port);
return -ENOMEM;
+ }
res->vlan = vlan;
res->port = (u8) port;
res->vlan_index = vlan_index;
@@ -1682,6 +1809,8 @@ static void vlan_del_from_slave(struct mlx4_dev *dev, int slave, u16 vlan,
if (res->vlan == vlan && res->port == (u8) port) {
if (!--res->ref_count) {
list_del(&res->list);
+ mlx4_release_resource(dev, slave, RES_VLAN,
+ 1, port);
kfree(res);
}
break;
@@ -1703,6 +1832,7 @@ static void rem_slave_vlans(struct mlx4_dev *dev, int slave)
/* dereference the vlan the num times the slave referenced it */
for (i = 0; i < res->ref_count; i++)
__mlx4_unregister_vlan(dev, res->port, res->vlan);
+ mlx4_release_resource(dev, slave, RES_VLAN, 1, res->port);
kfree(res);
}
}
@@ -1749,15 +1879,23 @@ static int counter_alloc_res(struct mlx4_dev *dev, int slave, int op, int cmd,
if (op != RES_OP_RESERVE)
return -EINVAL;
- err = __mlx4_counter_alloc(dev, &index);
+ err = mlx4_grant_resource(dev, slave, RES_COUNTER, 1, 0);
if (err)
return err;
+ err = __mlx4_counter_alloc(dev, &index);
+ if (err) {
+ mlx4_release_resource(dev, slave, RES_COUNTER, 1, 0);
+ return err;
+ }
+
err = add_res_range(dev, slave, index, 1, RES_COUNTER, 0);
- if (err)
+ if (err) {
__mlx4_counter_free(dev, index);
- else
+ mlx4_release_resource(dev, slave, RES_COUNTER, 1, 0);
+ } else {
set_param_l(out_param, index);
+ }
return err;
}
@@ -1864,6 +2002,7 @@ static int qp_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
err = rem_res_range(dev, slave, base, count, RES_QP, 0);
if (err)
break;
+ mlx4_release_resource(dev, slave, RES_QP, count, 0);
__mlx4_qp_release_range(dev, base, count);
break;
case RES_OP_MAP_ICM:
@@ -1901,8 +2040,10 @@ static int mtt_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
base = get_param_l(&in_param);
order = get_param_h(&in_param);
err = rem_res_range(dev, slave, base, 1, RES_MTT, order);
- if (!err)
+ if (!err) {
+ mlx4_release_resource(dev, slave, RES_MTT, 1 << order, 0);
__mlx4_free_mtt_range(dev, base, order);
+ }
return err;
}
@@ -1927,6 +2068,7 @@ static int mpt_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
err = rem_res_range(dev, slave, id, 1, RES_MPT, 0);
if (err)
break;
+ mlx4_release_resource(dev, slave, RES_MPT, 1, 0);
__mlx4_mpt_release(dev, index);
break;
case RES_OP_MAP_ICM:
@@ -1961,6 +2103,7 @@ static int cq_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
if (err)
break;
+ mlx4_release_resource(dev, slave, RES_CQ, 1, 0);
__mlx4_cq_free_icm(dev, cqn);
break;
@@ -1985,6 +2128,7 @@ static int srq_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
if (err)
break;
+ mlx4_release_resource(dev, slave, RES_SRQ, 1, 0);
__mlx4_srq_free_icm(dev, srqn);
break;
@@ -2056,6 +2200,7 @@ static int counter_free_res(struct mlx4_dev *dev, int slave, int op, int cmd,
return err;
__mlx4_counter_free(dev, index);
+ mlx4_release_resource(dev, slave, RES_COUNTER, 1, 0);
return err;
}
@@ -3785,6 +3930,11 @@ static void rem_slave_qps(struct mlx4_dev *dev, int slave)
&tracker->res_tree[RES_QP]);
list_del(&qp->com.list);
spin_unlock_irq(mlx4_tlock(dev));
+ if (!valid_reserved(dev, slave, qpn)) {
+ __mlx4_qp_release_range(dev, qpn, 1);
+ mlx4_release_resource(dev, slave,
+ RES_QP, 1, 0);
+ }
kfree(qp);
state = 0;
break;
@@ -3856,6 +4006,8 @@ static void rem_slave_srqs(struct mlx4_dev *dev, int slave)
&tracker->res_tree[RES_SRQ]);
list_del(&srq->com.list);
spin_unlock_irq(mlx4_tlock(dev));
+ mlx4_release_resource(dev, slave,
+ RES_SRQ, 1, 0);
kfree(srq);
state = 0;
break;
@@ -3922,6 +4074,8 @@ static void rem_slave_cqs(struct mlx4_dev *dev, int slave)
&tracker->res_tree[RES_CQ]);
list_del(&cq->com.list);
spin_unlock_irq(mlx4_tlock(dev));
+ mlx4_release_resource(dev, slave,
+ RES_CQ, 1, 0);
kfree(cq);
state = 0;
break;
@@ -3985,6 +4139,8 @@ static void rem_slave_mrs(struct mlx4_dev *dev, int slave)
&tracker->res_tree[RES_MPT]);
list_del(&mpt->com.list);
spin_unlock_irq(mlx4_tlock(dev));
+ mlx4_release_resource(dev, slave,
+ RES_MPT, 1, 0);
kfree(mpt);
state = 0;
break;
@@ -4054,6 +4210,8 @@ static void rem_slave_mtts(struct mlx4_dev *dev, int slave)
&tracker->res_tree[RES_MTT]);
list_del(&mtt->com.list);
spin_unlock_irq(mlx4_tlock(dev));
+ mlx4_release_resource(dev, slave, RES_MTT,
+ 1 << mtt->order, 0);
kfree(mtt);
state = 0;
break;
@@ -4212,6 +4370,7 @@ static void rem_slave_counters(struct mlx4_dev *dev, int slave)
list_del(&counter->com.list);
kfree(counter);
__mlx4_counter_free(dev, index);
+ mlx4_release_resource(dev, slave, RES_COUNTER, 1, 0);
}
}
spin_unlock_irq(mlx4_tlock(dev));
--
1.7.1
^ permalink raw reply related
* Re: [PATCH net] net: sctp: do not trigger BUG_ON in sctp_cmd_delete_tcb
From: Neil Horman @ 2013-10-31 13:42 UTC (permalink / raw)
To: Daniel Borkmann; +Cc: davem, netdev, linux-sctp, Vlad Yasevich
In-Reply-To: <1383207212-21658-1-git-send-email-dborkman@redhat.com>
On Thu, Oct 31, 2013 at 09:13:32AM +0100, Daniel Borkmann wrote:
> Introduced in f9e42b853523 ("net: sctp: sideeffect: throw BUG if
> primary_path is NULL"), we intended to find a buggy assoc that's
> part of the assoc hash table with a primary_path that is NULL.
> However, we better remove the BUG_ON for now and find a more
> suitable place to assert for these things as Mark reports that
> this also triggers the bug when duplication cookie processing
> happens, and the assoc is not part of the hash table (so all
> good in this case). Such a situation can for example easily be
> reproduced by:
>
> tc qdisc add dev eth0 root handle 1: prio bands 2 priomap 1 1 1 1 1 1
> tc qdisc add dev eth0 parent 1:2 handle 20: netem loss 20%
> tc filter add dev eth0 protocol ip parent 1: prio 2 u32 match ip \
> protocol 132 0xff match u8 0x0b 0xff at 32 flowid 1:2
>
> This drops 20% of COOKIE-ACK packets. After some follow-up
> discussion with Vlad we came to the conclusion that for now we
> should still better remove this BUG_ON() assertion, and come up
> with two follow-ups later on, that is, i) find a more suitable
> place for this assertion, and possibly ii) have a special
> allocator/initializer for such kind of temporary assocs.
>
> Reported-by: Mark Thomas <Mark.Thomas@metaswitch.com>
> Signed-off-by: Vlad Yasevich <vyasevich@gmail.com>
> Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
> ---
> net/sctp/sm_sideeffect.c | 1 -
> 1 file changed, 1 deletion(-)
>
> diff --git a/net/sctp/sm_sideeffect.c b/net/sctp/sm_sideeffect.c
> index 666c668..1a6eef3 100644
> --- a/net/sctp/sm_sideeffect.c
> +++ b/net/sctp/sm_sideeffect.c
> @@ -860,7 +860,6 @@ static void sctp_cmd_delete_tcb(sctp_cmd_seq_t *cmds,
> (!asoc->temp) && (sk->sk_shutdown != SHUTDOWN_MASK))
> return;
>
> - BUG_ON(asoc->peer.primary_path == NULL);
> sctp_unhash_established(asoc);
> sctp_association_free(asoc);
> }
> --
> 1.8.3.1
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-sctp" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
^ permalink raw reply
* Re: [PATCH net-next V2 2/2] virtio-net: coalesce rx frags when possible during rx
From: Eric Dumazet @ 2013-10-31 13:57 UTC (permalink / raw)
To: Jason Wang
Cc: mwdalton, mst, netdev, linux-kernel, virtualization, kmindg,
edumazet, davem
In-Reply-To: <1383220027-12278-2-git-send-email-jasowang@redhat.com>
On Thu, 2013-10-31 at 19:47 +0800, Jason Wang wrote:
> Commit 2613af0ed18a11d5c566a81f9a6510b73180660a (virtio_net: migrate mergeable
> rx buffers to page frag allocators) try to increase the payload/truesize for
> MTU-sized traffic. But this will introduce the extra overhead for GSO packets
> received because of the frag list. This commit tries to reduce this issue by
> coalesce the possible rx frags when possible during rx. Test result shows the
> about 15% improvement on full size GSO packet receiving (and even better than
> commit 2613af0ed18a11d5c566a81f9a6510b73180660a).
>
> Before this commit:
> ./netperf -H 192.168.100.4
> MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.100.4
> () port 0 AF_INET : demo
> Recv Send Send
> Socket Socket Message Elapsed
> Size Size Size Time Throughput
> bytes bytes bytes secs. 10^6bits/sec
>
> 87380 16384 16384 10.00 20303.87
>
> After this commit:
> ./netperf -H 192.168.100.4
> MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 192.168.100.4 () port 0 AF_INET : demo
> Recv Send Send
> Socket Socket Message Elapsed
> Size Size Size Time Throughput
> bytes bytes bytes secs. 10^6bits/sec
>
> 87380 16384 16384 10.00 23841.26
>
> Cc: Rusty Russell <rusty@rustcorp.com.au>
> Cc: Michael S. Tsirkin <mst@redhat.com>
> Cc: Michael Dalton <mwdalton@google.com>
> Cc: Eric Dumazet <edumazet@google.com>
> Acked-by: Michael S. Tsirkin <mst@redhat.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
Excellent !
We now have 2 or 3 frags per skb, like tcp stack manages to do on output
path.
Michael Dalton is also working on a autotuning patch, using an EWMA, so
that the size of individual sg blocks can vary from 1500 to 4096, this
might show even better throughput, we'll see.
Acked-by: Eric Dumazet <edumazet@google.com>
^ permalink raw reply
* Re: [net-next PATCH] net: codel: Avoid undefined behavior from signed overflow
From: Jesper Dangaard Brouer @ 2013-10-31 14:15 UTC (permalink / raw)
To: Eric Dumazet; +Cc: Jesper Dangaard Brouer, netdev, Paul E. McKenney, Dave Taht
In-Reply-To: <1383156104.4857.49.camel@edumazet-glaptop.roam.corp.google.com>
On Wed, 30 Oct 2013 11:01:44 -0700
Eric Dumazet <eric.dumazet@gmail.com> wrote:
> On Wed, 2013-10-30 at 18:23 +0100, Jesper Dangaard Brouer wrote:
> > From: Jesper Dangaard Brouer <netoptimizer@brouer.com>
> >
> > As described in commit 5a581b367 (jiffies: Avoid undefined
> > behavior from signed overflow), according to the C standard
> > 3.4.3p3, overflow of a signed integer results in undefined
> > behavior.
> >
> > To fix this, do as the above commit, and do an unsigned
> > subtraction, and interpreting the result as a signed
> > two's-complement number. This is based on the theory from
> > RFC 1982 and is nicely described in wikipedia here:
> > https://en.wikipedia.org/wiki/Serial_number_arithmetic#General_Solution
> >
> > A side-note, I have seen practical issues with the previous logic
> > when dealing with 16-bit, on a 64-bit machine (gcc version
> > 4.4.5). This were 32-bit, which I have not observed issues with.
> >
> > Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> > Signed-off-by: Jesper Dangaard Brouer <netoptimizer@brouer.com>
> > ---
> >
> > include/net/codel.h | 8 ++++----
> > 1 files changed, 4 insertions(+), 4 deletions(-)
> >
> > diff --git a/include/net/codel.h b/include/net/codel.h
> > index 389cf62..700fcdf 100644
> > --- a/include/net/codel.h
> > +++ b/include/net/codel.h
> > @@ -72,10 +72,10 @@ static inline codel_time_t codel_get_time(void)
> > return ns >> CODEL_SHIFT;
> > }
> >
> > -#define codel_time_after(a, b) ((s32)(a) - (s32)(b) > 0)
> > -#define codel_time_after_eq(a, b) ((s32)(a) - (s32)(b) >= 0)
> > -#define codel_time_before(a, b) ((s32)(a) - (s32)(b) < 0)
> > -#define codel_time_before_eq(a, b) ((s32)(a) - (s32)(b) <= 0)
> > +#define codel_time_after(a, b) ((s32)((a) - (b)) > 0)
> > +#define codel_time_after_eq(a, b) ((s32)((a) - (b)) >= 0)
> > +#define codel_time_before(a, b) ((s32)((a) - (b)) < 0)
> > +#define codel_time_before_eq(a, b) ((s32)((a) - (b)) <= 0)
> >
>
> I see nothing enforcing an unsigned subtraction as claimed in your
> changelog.
>
> a / b could be signed.
>
> Paul commit 5a581b367b5 was OK because of existing typecheck(unsigned
> long, ....)
Okay, I'll cook up another patch, after work.
Adding all the typecheck() stuff, just bloats the code.
Would it be better/okay just to do?:
(s32)((u32)(a) - (u32)(b)) > 0)
--
Best regards,
Jesper Dangaard Brouer
MSc.CS, Sr. Network Kernel Developer at Red Hat
Author of http://www.iptv-analyzer.org
LinkedIn: http://www.linkedin.com/in/brouer
^ permalink raw reply
* Re: [PATCH net-next V2 1/2] net: introduce skb_coalesce_rx_frag()
From: Eric Dumazet @ 2013-10-31 14:26 UTC (permalink / raw)
To: Jason Wang
Cc: mwdalton, mst, netdev, linux-kernel, virtualization, kmindg,
edumazet, davem
In-Reply-To: <1383220027-12278-1-git-send-email-jasowang@redhat.com>
On Thu, 2013-10-31 at 19:47 +0800, Jason Wang wrote:
> Sometimes we need to coalesce the rx frags to avoid frag list. One example is
> virtio-net driver which tries to use small frags for both MTU sized packet and
> GSO packet. So this patch introduce skb_coalesce_rx_frag() to do this.
>
> Cc: Rusty Russell <rusty@rustcorp.com.au>
> Cc: Michael S. Tsirkin <mst@redhat.com>
> Cc: Michael Dalton <mwdalton@google.com>
> Cc: Eric Dumazet <edumazet@google.com>
> Acked-by: Michael S. Tsirkin <mst@redhat.com>
> Signed-off-by: Jason Wang <jasowang@redhat.com>
> ---
> Changes from V1:
> - remove the useless off parameter.
> ---
> include/linux/skbuff.h | 3 +++
> net/core/skbuff.c | 13 +++++++++++++
> 2 files changed, 16 insertions(+)
>
> diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
> index 2c15497..fffaeaf 100644
> --- a/include/linux/skbuff.h
> +++ b/include/linux/skbuff.h
> @@ -1372,6 +1372,9 @@ static inline void skb_fill_page_desc(struct sk_buff *skb, int i,
> void skb_add_rx_frag(struct sk_buff *skb, int i, struct page *page, int off,
> int size, unsigned int truesize);
>
> +void skb_coalesce_rx_frag(struct sk_buff *skb, int i, int size,
> + unsigned int truesize);
> +
> #define SKB_PAGE_ASSERT(skb) BUG_ON(skb_shinfo(skb)->nr_frags)
> #define SKB_FRAG_ASSERT(skb) BUG_ON(skb_has_frag_list(skb))
> #define SKB_LINEAR_ASSERT(skb) BUG_ON(skb_is_nonlinear(skb))
> diff --git a/net/core/skbuff.c b/net/core/skbuff.c
> index 0ab32fa..87670e1 100644
> --- a/net/core/skbuff.c
> +++ b/net/core/skbuff.c
> @@ -476,6 +476,19 @@ void skb_add_rx_frag(struct sk_buff *skb, int i, struct page *page, int off,
> }
> EXPORT_SYMBOL(skb_add_rx_frag);
>
> +void skb_coalesce_rx_frag(struct sk_buff *skb, int i, int size,
> + unsigned int truesize)
> +{
> + skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
> +
> + skb_frag_size_add(frag, size);
> + skb->len += size;
> + skb->data_len += size;
> + skb->truesize += truesize;
> + skb_frag_unref(skb, i);
This unref is not logical, or should at least be
__skb_frag_unref(frag);
But I do think this is best done in the caller.
In virtio_net this would be a :
put_page(page);
In tcp stack we do almost the same, but we take the reference on the
page if we could not coalesce with prio frag, instead of doing a get and
put in the other case.
if (can_coalesce) {
skb_frag_size_add(&skb_shinfo(skb)->frags[i - 1], copy);
} else {
get_page(page);
skb_fill_page_desc(skb, i, page, offset, copy);
}
> +}
> +EXPORT_SYMBOL(skb_coalesce_rx_frag);
> +
> static void skb_drop_list(struct sk_buff **listp)
> {
> kfree_skb_list(*listp);
^ permalink raw reply
* Re: [PATCH] x86: Run checksumming in parallel accross multiple alu's
From: Neil Horman @ 2013-10-31 14:33 UTC (permalink / raw)
To: Ingo Molnar
Cc: Eric Dumazet, linux-kernel, sebastien.dugue, Thomas Gleixner,
Ingo Molnar, H. Peter Anvin, x86, netdev
In-Reply-To: <20131031102200.GA10098@gmail.com>
On Thu, Oct 31, 2013 at 11:22:00AM +0100, Ingo Molnar wrote:
>
> * Neil Horman <nhorman@tuxdriver.com> wrote:
>
> > > etc. For such short runtimes make sure the last column displays
> > > close to 100%, so that the PMU results become trustable.
> > >
> > > A nehalem+ PMU will allow 2-4 events to be measured in parallel,
> > > plus generics like 'cycles', 'instructions' can be added 'for free'
> > > because they get counted in a separate (fixed purpose) PMU register.
> > >
> > > The last colum tells you what percentage of the runtime that
> > > particular event was actually active. 100% (or empty last column)
> > > means it was active all the time.
> > >
> > > Thanks,
> > >
> > > Ingo
> > >
> >
> > Hmm,
> >
> > I ran this test:
> >
> > for i in `seq 0 1 3`
> > do
> > echo $i > /sys/module/csum_test/parameters/module_test_mode
> > taskset -c 0 perf stat --repeat 20 -C 0 -e L1-dcache-load-misses -e L1-dcache-prefetches -e cycles -e instructions -ddd ./test.sh
> > done
>
> You need to remove '-ddd' which is a shortcut for a ton of useful
> events, but here you want to use fewer events, to increase the
> precision of the measurement.
>
> Thanks,
>
> Ingo
>
Thank you ingo, that fixed it. I'm trying some other variants of the csum
algorithm that Doug and I discussed last night, but FWIW, the relative
performance of the 4 test cases (base/prefetch/parallel/both) remains unchanged.
I'm starting to feel like at this point, theres very little point in doing
parallel alu operations (unless we can find a way to break the dependency on the
carry flag, which is what I'm tinkering with now).
Neil
^ permalink raw reply
* Re: [PATCH 19/51] DMA-API: media: dt3155v4l: replace dma_set_mask()+dma_set_coherent_mask() with new helper
From: Russell King - ARM Linux @ 2013-10-31 14:49 UTC (permalink / raw)
To: Mauro Carvalho Chehab
Cc: Hans Verkuil, alsa-devel, linux-doc, linux-mmc, linux-fbdev,
linux-nvme, linux-ide, devel, linux-samsung-soc, linux-scsi,
e1000-devel, b43-dev, linux-media, devicetree, dri-devel,
linux-tegra, linux-omap, linux-arm-kernel,
Solarflare linux maintainers, netdev, linux-usb, linux-wireless,
linux-crypto, Greg Kroah-Hartman, uclinux-dist-devel,
linuxppc-dev
In-Reply-To: <20131031094640.205840a2@samsung.com>
On Thu, Oct 31, 2013 at 09:46:40AM -0200, Mauro Carvalho Chehab wrote:
> Hi Russell,
>
> Em Mon, 30 Sep 2013 13:57:47 +0200
> Hans Verkuil <hverkuil@xs4all.nl> escreveu:
>
> > On 09/19/2013 11:44 PM, Russell King wrote:
> > > Replace the following sequence:
> > >
> > > dma_set_mask(dev, mask);
> > > dma_set_coherent_mask(dev, mask);
> > >
> > > with a call to the new helper dma_set_mask_and_coherent().
> > >
> > > Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
> >
> > Acked-by: Hans Verkuil <hans.verkuil@cisco.com>
>
> Somehow, I lost your original post (I got unsubscribed on a few days
> from all vger mailing lists at the end of september).
>
> I suspect that you want to sent this via your tree, right?
Yes please.
> If so:
>
> Acked-by: Mauro Carvalho Chehab <m.chehab@samsung.com>
Added, thanks.
> > > - err = dma_set_mask(&pdev->dev, DMA_BIT_MASK(32));
> > > - if (err)
> > > - return -ENODEV;
> > > - err = dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(32));
> > > + err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
> > > if (err)
> > > return -ENODEV;
One thing I've just noticed is that return should be "return err" not
"return -ENODEV" - are you okay for me to change that in this patch?
Thanks.
^ permalink raw reply
page: next (older) | prev (newer) | latest
- recent:[subjects (threaded)|topics (new)|topics (active)]
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox