* Re: [PATCH net-next] sfc: stop the TX queue before pushing new buffers
From: Edward Cree @ 2018-05-23 15:28 UTC (permalink / raw)
To: Martin Habets, linux-net-drivers, davem; +Cc: netdev, jarod
In-Reply-To: <152706844446.27257.6312747433904122379.stgit@mh-desktop.uk.solarflarecom.com>
On 23/05/18 10:41, Martin Habets wrote:
> efx_enqueue_skb() can push new buffers for the xmit_more functionality.
> We must stops the TX queue before this or else the TX queue does not get
> restarted and we get a netdev watchdog.
>
> In the error handling we may now need to unwind more than 1 packet, and
> we may need to push the new buffers onto the partner queue.
>
> Fixes: e9117e5099ea ("sfc: Firmware-Assisted TSO version 2")
> Reported-by: Jarod Wilson <jarod@redhat.com>
> Signed-off-by: Martin Habets <mhabets@solarflare.com>
> ---
>
> Dave, could you please also queue up this patch for stable?
>
> drivers/net/ethernet/sfc/tx.c | 31 +++++++++++++++++++++++--------
> 1 file changed, 23 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/net/ethernet/sfc/tx.c b/drivers/net/ethernet/sfc/tx.c
> index cece961f2e82..17e0697f42d5 100644
> --- a/drivers/net/ethernet/sfc/tx.c
> +++ b/drivers/net/ethernet/sfc/tx.c
> @@ -435,17 +435,18 @@ static int efx_tx_map_data(struct efx_tx_queue *tx_queue, struct sk_buff *skb,
> } while (1);
> }
>
> -/* Remove buffers put into a tx_queue. None of the buffers must have
> - * an skb attached.
> +/* Remove buffers put into a tx_queue for the current packet.
> + * None of the buffers must have an skb attached.
> */
> -static void efx_enqueue_unwind(struct efx_tx_queue *tx_queue)
> +static void efx_enqueue_unwind(struct efx_tx_queue *tx_queue,
> + unsigned int insert_count)
> {
> struct efx_tx_buffer *buffer;
> unsigned int bytes_compl = 0;
> unsigned int pkts_compl = 0;
>
> /* Work backwards until we hit the original insert pointer value */
> - while (tx_queue->insert_count != tx_queue->write_count) {
> + while (tx_queue->insert_count != insert_count) {
> --tx_queue->insert_count;
> buffer = __efx_tx_queue_get_insert_buffer(tx_queue);
> efx_dequeue_buffer(tx_queue, buffer, &pkts_compl, &bytes_compl);
> @@ -504,6 +505,8 @@ static int efx_tx_tso_fallback(struct efx_tx_queue *tx_queue,
> */
> netdev_tx_t efx_enqueue_skb(struct efx_tx_queue *tx_queue, struct sk_buff *skb)
> {
> + unsigned int old_insert_count = tx_queue->insert_count;
> + bool xmit_more = skb->xmit_more;
> bool data_mapped = false;
> unsigned int segments;
> unsigned int skb_len;
> @@ -553,8 +556,10 @@ netdev_tx_t efx_enqueue_skb(struct efx_tx_queue *tx_queue, struct sk_buff *skb)
> /* Update BQL */
> netdev_tx_sent_queue(tx_queue->core_txq, skb_len);
>
> + efx_tx_maybe_stop_queue(tx_queue);
> +
> /* Pass off to hardware */
> - if (!skb->xmit_more || netif_xmit_stopped(tx_queue->core_txq)) {
> + if (!xmit_more || netif_xmit_stopped(tx_queue->core_txq)) {
> struct efx_tx_queue *txq2 = efx_tx_queue_partner(tx_queue);
>
> /* There could be packets left on the partner queue if those
> @@ -577,14 +582,24 @@ netdev_tx_t efx_enqueue_skb(struct efx_tx_queue *tx_queue, struct sk_buff *skb)
> tx_queue->tx_packets++;
> }
>
> - efx_tx_maybe_stop_queue(tx_queue);
> -
> return NETDEV_TX_OK;
>
>
> err:
> - efx_enqueue_unwind(tx_queue);
> + efx_enqueue_unwind(tx_queue, old_insert_count);
> dev_kfree_skb_any(skb);
> +
> + /* If we're not expecting another transmit and we had something to push
> + * on this queue or a partner queue then we need to push here to get the
> + * previous packets out.
> + */
> + if (!xmit_more) {
> + struct efx_tx_queue *txq2 = efx_tx_queue_partner(tx_queue);
> +
> + if (txq2->xmit_more_available)
> + efx_nic_push_buffers(txq2);
> + }
This only pushes the partner queue, it needs to also push this queue if
tx_queue->xmit_more_available (as your comment just above mentions).
Apart from this, LGTM.
-Ed
> +
> return NETDEV_TX_OK;
> }
>
>
^ permalink raw reply
* Re: [net-next 1/6] net/dcb: Add dcbnl buffer attribute
From: Huy Nguyen @ 2018-05-23 15:27 UTC (permalink / raw)
To: Jakub Kicinski
Cc: Saeed Mahameed, David S. Miller, netdev, Jiri Pirko, Or Gerlitz,
Parav Pandit, Ido Schimmel
In-Reply-To: <20180523022314.783e47fa@cakuba>
On 5/23/2018 4:23 AM, Jakub Kicinski wrote:
> >From patch description it seems like your default setup is shared
> buffer split 50% (lossy)/50% (all prios) and the example you give
> changes that to 25% (lossy)/25%x3 prio groups.
>
> With existing devlink API could this be modelled by three ingress pools
> with 2 TCs bound each?
Yes, possible. When you map prio to tc. Please be careful with prio term
in switch
since switch has more than 8 prio.
^ permalink raw reply
* Re: [bpf-next V4 PATCH 6/8] xdp: change ndo_xdp_xmit API to support bulking
From: Jesper Dangaard Brouer @ 2018-05-23 15:27 UTC (permalink / raw)
To: John Fastabend
Cc: netdev, Daniel Borkmann, Alexei Starovoitov, Christoph Hellwig,
BjörnTöpel, Magnus Karlsson, makita.toshiaki, brouer
In-Reply-To: <c9f0c7b0-26a8-815d-71bf-4e5ccdaa74aa@gmail.com>
On Wed, 23 May 2018 07:42:14 -0700
John Fastabend <john.fastabend@gmail.com> wrote:
> On 05/18/2018 06:35 AM, Jesper Dangaard Brouer wrote:
> [...]
>
> Couple suggestions for some optimizations/improvements but otherwise
> looks good to me.
>
> Thanks,
> John
>
[...]
> > diff --git a/drivers/net/ethernet/intel/i40e/i40e_txrx.c b/drivers/net/ethernet/intel/i40e/i40e_txrx.c
> > index 5efa68de935b..9b698c5acd05 100644
> > --- a/drivers/net/ethernet/intel/i40e/i40e_txrx.c
> > +++ b/drivers/net/ethernet/intel/i40e/i40e_txrx.c
> > @@ -3664,14 +3664,19 @@ netdev_tx_t i40e_lan_xmit_frame(struct sk_buff *skb, struct net_device *netdev)
> > * @dev: netdev
> > * @xdp: XDP buffer
> > *
> > - * Returns Zero if sent, else an error code
> > + * Returns number of frames successfully sent. Frames that fail are
> > + * free'ed via XDP return API.
> > + *
> > + * For error cases, a negative errno code is returned and no-frames
> > + * are transmitted (caller must handle freeing frames).
> > **/
> > -int i40e_xdp_xmit(struct net_device *dev, struct xdp_frame *xdpf)
> > +int i40e_xdp_xmit(struct net_device *dev, int n, struct xdp_frame **frames)
> > {
> > struct i40e_netdev_priv *np = netdev_priv(dev);
> > unsigned int queue_index = smp_processor_id();
> > struct i40e_vsi *vsi = np->vsi;
> > - int err;
> > + int drops = 0;
> > + int i;
> >
> > if (test_bit(__I40E_VSI_DOWN, vsi->state))
> > return -ENETDOWN;
> > @@ -3679,11 +3684,18 @@ int i40e_xdp_xmit(struct net_device *dev, struct xdp_frame *xdpf)
> > if (!i40e_enabled_xdp_vsi(vsi) || queue_index >= vsi->num_queue_pairs)
> > return -ENXIO;
> >
> > - err = i40e_xmit_xdp_ring(xdpf, vsi->xdp_rings[queue_index]);
> > - if (err != I40E_XDP_TX)
> > - return -ENOSPC;
> > + for (i = 0; i < n; i++) {
> > + struct xdp_frame *xdpf = frames[i];
> > + int err;
> >
> > - return 0;
> > + err = i40e_xmit_xdp_ring(xdpf, vsi->xdp_rings[queue_index]);
> > + if (err != I40E_XDP_TX) {
> > + xdp_return_frame_rx_napi(xdpf);
> > + drops++;
> > + }
> > + }
> > +
>
> Perhaps not needed in this series, but I wonder how useful it is to hit
> the ring with the remaining frames after the first error. The error will
> typically be due to the TX queue being full. In this case it might make
> more sense to just drop the entire train of frames vs beating the up a
> full queue.
Yes, that would be an optimization, but I think we can do that in
another series.
This patcheset actually open up for a lot of followup optimizations.
I imagine/plan to optimize i40e_xmit_xdp_ring() further, e.g. by
implementing a bulking variant (e.g check I40E_DESC_UNUSED+update
xdp_ring->next_to_use once, map several DMA pages in one call, only
call smp_wmb() once per bulk, etc.),
> > + return n - drops;
> > }
--
Best regards,
Jesper Dangaard Brouer
MSc.CS, Principal Kernel Engineer at Red Hat
LinkedIn: http://www.linkedin.com/in/brouer
^ permalink raw reply
* Re: [net-next] i40iw/i40e: Remove link dependency on i40e
From: Jason Gunthorpe @ 2018-05-23 15:18 UTC (permalink / raw)
To: Alexander Duyck
Cc: Christoph Hellwig, Jeff Kirsher, David Miller, Doug Ledford,
Sindhu Devale, Netdev, linux-rdma, Neil Horman, Stefan Assmann,
John Greene, Shiraz Saleem
In-Reply-To: <CAKgT0UeZRZ8O3PQWSz2Wod8HNjwc_bcuqqb70A7qypPn5kzW+Q@mail.gmail.com>
On Wed, May 23, 2018 at 08:03:44AM -0700, Alexander Duyck wrote:
> On Tue, May 22, 2018 at 11:19 PM, Christoph Hellwig <hch@infradead.org> wrote:
> > On Tue, May 22, 2018 at 02:04:06PM -0700, Jeff Kirsher wrote:
> >> > Why would you want to do this? The rdma driver is non-functional
> >> > without the ethernet driver, so why on earth would we want to defeat
> >> > the module dependency mechanism?
> >>
> >> This change is driven by the OSV's like Red Hat, where customer's were
> >> updating the i40e driver, which in turn broke i40iw.
> >
> > Doctor it hurts when I do this..
> >
> > There is no reason to make a mess of our drivers because people are
> > doing things they should haver never done and that aren't supported
> > in Linux.
> >
> > If Intel didn;t offer any out of tree drivers I'm pretty sure no
> > customer would even attempt this. So fix this where the problem is.
>
> Are you serious? You are never going to see out-of-tree drivers go
> away. They exist for the simple reason that most customers/OSVs are
> slow to upgrade their kernels so we have people running on a 3.10
> something kernel on their RHEL 7.X and want to use the latest greatest
> hardware.
So provide the i40iw module when providing the i40e upgrade module?
I still can't understand why this is a problem that needs to be
solved in mainline, or why it deserves a special and unique fix to
i40e, or even what the *actual* problem is..
Jason
^ permalink raw reply
* Re: [net-next 1/6] net/dcb: Add dcbnl buffer attribute
From: Huy Nguyen @ 2018-05-23 15:08 UTC (permalink / raw)
To: Jakub Kicinski
Cc: Saeed Mahameed, David S. Miller, netdev, Jiri Pirko, Or Gerlitz,
Parav Pandit, Ido Schimmel
In-Reply-To: <20180523022314.783e47fa@cakuba>
> I hope that is not true, since we (Netronome) are trying to use it for
> NIC configuration, too. We should generalize the API if need be.
Yes, it is up to your company. devlink is static tool. DCBNL are intended to
be dynamically configured by switch. In real world, not many people
configure NIC's qos
from host.
^ permalink raw reply
* Re: [bpf-next V4 PATCH 3/8] xdp: add tracepoint for devmap like cpumap have
From: Jesper Dangaard Brouer @ 2018-05-23 15:04 UTC (permalink / raw)
To: John Fastabend
Cc: netdev, Daniel Borkmann, Alexei Starovoitov, Christoph Hellwig,
BjörnTöpel, Magnus Karlsson, makita.toshiaki, brouer
In-Reply-To: <4345d6a2-935d-e206-e309-b63825f880b8@gmail.com>
On Wed, 23 May 2018 07:24:03 -0700
John Fastabend <john.fastabend@gmail.com> wrote:
> > @@ -219,8 +221,8 @@ void __dev_map_insert_ctx(struct bpf_map *map, u32 bit)
> > static int bq_xmit_all(struct bpf_dtab_netdev *obj,
> > struct xdp_bulk_queue *bq)
> > {
> > - unsigned int processed = 0, drops = 0;
> > struct net_device *dev = obj->dev;
> > + int sent = 0, drops = 0;
> > int i;
> >
> > if (unlikely(!bq->count))
> > @@ -241,10 +243,13 @@ static int bq_xmit_all(struct bpf_dtab_netdev *obj,
> > drops++;
> > xdp_return_frame(xdpf);
> > }
> > - processed++;
> > + sent++;
>
> Do 'dropped' frames also get counted as 'sent' frames? This seems a bit
> counter-intuitive to me. Should it be 'drops+sent = total frames'
> instead?
Again, sorry for mixing this up when spliting up the patch. The
patchset does end up with: 'drops+sent = total frames'. (The
"processed" counter is a copy-paste of code from cpumap, which have
another semantics). I'll clean it up in V5, to help reviewers.
--
Best regards,
Jesper Dangaard Brouer
MSc.CS, Principal Kernel Engineer at Red Hat
LinkedIn: http://www.linkedin.com/in/brouer
^ permalink raw reply
* Re: [net-next] i40iw/i40e: Remove link dependency on i40e
From: Alexander Duyck @ 2018-05-23 15:03 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Jeff Kirsher, Jason Gunthorpe, David Miller, Doug Ledford,
Sindhu Devale, Netdev, linux-rdma, Neil Horman, Stefan Assmann,
John Greene, Shiraz Saleem
In-Reply-To: <20180523061922.GA4753@infradead.org>
On Tue, May 22, 2018 at 11:19 PM, Christoph Hellwig <hch@infradead.org> wrote:
> On Tue, May 22, 2018 at 02:04:06PM -0700, Jeff Kirsher wrote:
>> > Why would you want to do this? The rdma driver is non-functional
>> > without the ethernet driver, so why on earth would we want to defeat
>> > the module dependency mechanism?
>>
>> This change is driven by the OSV's like Red Hat, where customer's were
>> updating the i40e driver, which in turn broke i40iw.
>
> Doctor it hurts when I do this..
>
> There is no reason to make a mess of our drivers because people are
> doing things they should haver never done and that aren't supported
> in Linux.
>
> If Intel didn;t offer any out of tree drivers I'm pretty sure no
> customer would even attempt this. So fix this where the problem is.
Are you serious? You are never going to see out-of-tree drivers go
away. They exist for the simple reason that most customers/OSVs are
slow to upgrade their kernels so we have people running on a 3.10
something kernel on their RHEL 7.X and want to use the latest greatest
hardware.
I find it ridiculous that you expect us to support a product with
in-kernel only and it is pretty short sighted to insist that that is
the only way to support a product.
With that said it probably wouldn't hurt to throw a WARN_ONCE in here
somewhere that basically informs the user they are doing something
stupid and essentially disabling the i40iw driver if they remove i40e.
- Alex
^ permalink raw reply
* Re: [PATCH v2 net] stmmac: strip vlan tag on reception only for 8021q tagged frames
From: Elad Nachman @ 2018-05-23 15:00 UTC (permalink / raw)
To: Jose Abreu, Florian Fainelli, David Miller
Cc: makita.toshiaki, netdev, peppe.cavallaro, alexandre.torgue
In-Reply-To: <09c5b6b3-faa5-fd99-76bd-72a107122f2a@synopsys.com>
Jose,
I am not sure which drivers you have checked. I guess most non-networking embedded drivers never use 802.1AD
so they stay broken unknowingly.
Specifically, I have tested Intel e1000e based card which works correctly versus stmmac which works incorrectly.
If you check netdev.c in e1000e then the vlan strip is conditional upon checking of 802.1Q bit in the descriptor,
which does not happen in stmmac_main.c - the vlan stripping happens based on any tag header check.
Once I applied my patch things started working - did not have to patch e1000e. The problem is that ip link allows to create 802.1ad devices which are not 0x8100 tagged but stmmac and probably other drivers handles the non 0x8100 tag incorrectly and the vlan slave discards the frame.
Regarding the getting the tag from the descriptor - this is of course a possibility. My original patch handled stripping of all tags but then I was told here that I cannot strip 802.1ad tags without the NETIF_F_HW_VLAN_STAG_RX feature, which is probably correct regardless of the source of the vlan tag - skb or descriptor.
I can also add the NETIF_F_HW_VLAN_STAG_RX feature plus the following original v1 patch:
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c 2018-04-11 17:04:00.586057300 +0300
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c 2018-04-11 17:05:33.601992400 +0300
@@ -3293,17 +3293,19 @@ dma_map_err:
static void stmmac_rx_vlan(struct net_device *dev, struct sk_buff *skb)
{
- struct ethhdr *ehdr;
+ struct vlan_ethhdr *veth;
u16 vlanid;
+ __be16 vlan_proto;
if ((dev->features & NETIF_F_HW_VLAN_CTAG_RX) ==
NETIF_F_HW_VLAN_CTAG_RX &&
!__vlan_get_tag(skb, &vlanid)) {
/* pop the vlan tag */
- ehdr = (struct ethhdr *)skb->data;
- memmove(skb->data + VLAN_HLEN, ehdr, ETH_ALEN * 2);
+ veth = (struct vlan_ethhdr *)skb->data;
+ vlan_proto = veth->h_vlan_proto;
+ memmove(skb->data + VLAN_HLEN, veth, ETH_ALEN * 2);
skb_pull(skb, VLAN_HLEN);
- __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), vlanid);
+ __vlan_hwaccel_put_tag(skb, vlan_proto, vlanid);
}
}
There are three reasons why I prefer using this approach rather than the descriptor approach:
A. It is inline with the original driver code.
B. Using descriptor vlan will require to completely rewrite stmmac_rx_vlan and will result in at least 2-3 times line counts in the patch.
C. I have no idea if first silicon implementations of DWMAC IP had some kind of HW bug (for example AXI clock glitch) which caused sampling of the VLAN tag in the descriptors to be unstable, and since I have no access to such hardware for regression I risk breaking stmmac for such old SOCs in case they decide to jump kernel versions to the latest.
Thanks,
Elad.
On 22/05/18 11:56, Jose Abreu wrote:
> On 21-05-2018 17:42, Florian Fainelli wrote:
>> On 05/21/2018 08:48 AM, David Miller wrote:
>>> From: David Miller <davem@davemloft.net>
>>> Date: Thu, 17 May 2018 12:43:56 -0400 (EDT)
>>>
>>>> Giuseppe and Alexandre, please review this patch.
>>> If nobody thinks this patch is important enough to actually
>>> review, I'm tossing it.
>>>
>>> Sorry.
>>>
>> How about looping in Jose?
>
> Thanks for the cc Florian!
>
> Elad,
>
> Looking at this patch I have a couple of questions and suggestions:
>
> I see that most drivers use this pattern so, are they all broken?
> or is this a special case?
>
> You can also get the inner/outer vlan tag by reading desc0 of
> receive descriptor. Which can make this completely agnostic of
> VLAN tag.
>
> Thanks and Best Regards,
> Jose Miguel Abreu
>
^ permalink raw reply
* Re: [PATCH V3 8/8] dt-bindings: stm32: add compatible for syscon
From: Rob Herring @ 2018-05-23 14:47 UTC (permalink / raw)
To: Christophe ROULLIER
Cc: mark.rutland@arm.com, mcoquelin.stm32@gmail.com, Alexandre TORGUE,
Peppe CAVALLARO, devicetree@vger.kernel.org,
linux-arm-kernel@lists.infradead.org, netdev@vger.kernel.org,
andrew@lunn.ch
In-Reply-To: <4f6941a0-4b41-f2db-52b1-d45aa7287fa2@st.com>
On Wed, May 23, 2018 at 4:32 AM, Christophe ROULLIER
<christophe.roullier@st.com> wrote:
> On 05/22/2018 07:22 PM, Rob Herring wrote:
>> On Mon, May 21, 2018 at 10:07:26AM +0200, Christophe Roullier wrote:
>>> This patch describes syscon DT bindings.
>>>
>>> Signed-off-by: Christophe Roullier <christophe.roullier@st.com>
>>> ---
>>> Documentation/devicetree/bindings/arm/stm32.txt | 5 +++++
>>> 1 file changed, 5 insertions(+)
>>>
>>> diff --git a/Documentation/devicetree/bindings/arm/stm32.txt b/Documentation/devicetree/bindings/arm/stm32.txt
>>> index 6808ed9..e46ebad 100644
>>> --- a/Documentation/devicetree/bindings/arm/stm32.txt
>>> +++ b/Documentation/devicetree/bindings/arm/stm32.txt
>>> @@ -8,3 +8,8 @@ using one of the following compatible strings:
>>> st,stm32f746
>>> st,stm32h743
>>> st,stm32mp157
>>> +
>>> +Required nodes:
>>> +- syscon: the soc bus node must have a system controller node pointing to the
>>> + global control registers, with the compatible string
>>> + "st,stm32mp157-syscfg", "syscon";
>>
>> Please don't mix soc/board bindings with other nodes. So perhaps
>> stm32-syscon.txt.
>>
>> Rob
>>
>
> Hi Rob,
>
> Is it ok for you with this tree file:
Yes, one nit below.
>
> Documentation/devicetree/bindings/arm/stm32/stm32.txt
> Documentation/devicetree/bindings/arm/stm32/stm32-syscon.txt
> With stm32-syscon.txt:
>
> ---------------------------------------------------
> STMicroelectronics STM32 Platforms System Controller
>
> Properties:
> - compatible : should contain two values. First value must be :
> - " st,stm32mp157-syscfg " - for stm32mp157 based SoCs,
> second value must be always "syscon".
> - reg : offset and length of the register set.
>
> Example:
> syscfg: system-config@50020000 {
syscon@...
> compatible = "st,stm32mp157-syscfg", "syscon";
> reg = <0x50020000 0x400>;
> };
> ---------------------------------------------------
>
> Do we need to update also all MCU family (stm32f4, stm32h7, stm32f7)
> property to be coherent ?
Yes, if they all have the same or similar syscfg block.
>
> Thanks for your feedback.
>
> Christophe.
>
^ permalink raw reply
* Re: [bpf-next V4 PATCH 6/8] xdp: change ndo_xdp_xmit API to support bulking
From: John Fastabend @ 2018-05-23 14:42 UTC (permalink / raw)
To: Jesper Dangaard Brouer, netdev, Daniel Borkmann,
Alexei Starovoitov
Cc: Christoph Hellwig, BjörnTöpel, Magnus Karlsson,
makita.toshiaki
In-Reply-To: <152665050207.21055.17828855357297094854.stgit@firesoul>
On 05/18/2018 06:35 AM, Jesper Dangaard Brouer wrote:
> This patch change the API for ndo_xdp_xmit to support bulking
> xdp_frames.
>
> When kernel is compiled with CONFIG_RETPOLINE, XDP sees a huge slowdown.
> Most of the slowdown is caused by DMA API indirect function calls, but
> also the net_device->ndo_xdp_xmit() call.
>
> Benchmarked patch with CONFIG_RETPOLINE, using xdp_redirect_map with
> single flow/core test (CPU E5-1650 v4 @ 3.60GHz), showed
> performance improved:
> for driver ixgbe: 6,042,682 pps -> 6,853,768 pps = +811,086 pps
> for driver i40e : 6,187,169 pps -> 6,724,519 pps = +537,350 pps
>
> With frames avail as a bulk inside the driver ndo_xdp_xmit call,
> further optimizations are possible, like bulk DMA-mapping for TX.
>
> Testing without CONFIG_RETPOLINE show the same performance for
> physical NIC drivers.
>
> The virtual NIC driver tun sees a huge performance boost, as it can
> avoid doing per frame producer locking, but instead amortize the
> locking cost over the bulk.
>
> V2: Fix compile errors reported by kbuild test robot <lkp@intel.com>
> V4: Isolated ndo, driver changes and callers.
>
> Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Couple suggestions for some optimizations/improvements but otherwise
looks good to me.
Thanks,
John
> ---
> drivers/net/ethernet/intel/i40e/i40e_txrx.c | 26 +++++++---
> drivers/net/ethernet/intel/i40e/i40e_txrx.h | 2 -
> drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 21 ++++++--
> drivers/net/tun.c | 37 +++++++++-----
> drivers/net/virtio_net.c | 66 +++++++++++++++++++------
> include/linux/netdevice.h | 14 +++--
> kernel/bpf/devmap.c | 31 ++++++++----
> net/core/filter.c | 8 ++-
> 8 files changed, 141 insertions(+), 64 deletions(-)
>
> diff --git a/drivers/net/ethernet/intel/i40e/i40e_txrx.c b/drivers/net/ethernet/intel/i40e/i40e_txrx.c
> index 5efa68de935b..9b698c5acd05 100644
> --- a/drivers/net/ethernet/intel/i40e/i40e_txrx.c
> +++ b/drivers/net/ethernet/intel/i40e/i40e_txrx.c
> @@ -3664,14 +3664,19 @@ netdev_tx_t i40e_lan_xmit_frame(struct sk_buff *skb, struct net_device *netdev)
> * @dev: netdev
> * @xdp: XDP buffer
> *
> - * Returns Zero if sent, else an error code
> + * Returns number of frames successfully sent. Frames that fail are
> + * free'ed via XDP return API.
> + *
> + * For error cases, a negative errno code is returned and no-frames
> + * are transmitted (caller must handle freeing frames).
> **/
> -int i40e_xdp_xmit(struct net_device *dev, struct xdp_frame *xdpf)
> +int i40e_xdp_xmit(struct net_device *dev, int n, struct xdp_frame **frames)
> {
> struct i40e_netdev_priv *np = netdev_priv(dev);
> unsigned int queue_index = smp_processor_id();
> struct i40e_vsi *vsi = np->vsi;
> - int err;
> + int drops = 0;
> + int i;
>
> if (test_bit(__I40E_VSI_DOWN, vsi->state))
> return -ENETDOWN;
> @@ -3679,11 +3684,18 @@ int i40e_xdp_xmit(struct net_device *dev, struct xdp_frame *xdpf)
> if (!i40e_enabled_xdp_vsi(vsi) || queue_index >= vsi->num_queue_pairs)
> return -ENXIO;
>
> - err = i40e_xmit_xdp_ring(xdpf, vsi->xdp_rings[queue_index]);
> - if (err != I40E_XDP_TX)
> - return -ENOSPC;
> + for (i = 0; i < n; i++) {
> + struct xdp_frame *xdpf = frames[i];
> + int err;
>
> - return 0;
> + err = i40e_xmit_xdp_ring(xdpf, vsi->xdp_rings[queue_index]);
> + if (err != I40E_XDP_TX) {
> + xdp_return_frame_rx_napi(xdpf);
> + drops++;
> + }
> + }
> +
Perhaps not needed in this series, but I wonder how useful it is to hit
the ring with the remaining frames after the first error. The error will
typically be due to the TX queue being full. In this case it might make
more sense to just drop the entire train of frames vs beating the up a
full queue.
> + return n - drops;
> }
>
> /**
> diff --git a/drivers/net/ethernet/intel/i40e/i40e_txrx.h b/drivers/net/ethernet/intel/i40e/i40e_txrx.h
> index fdd2c55f03a6..eb8804b3d7b6 100644
> --- a/drivers/net/ethernet/intel/i40e/i40e_txrx.h
> +++ b/drivers/net/ethernet/intel/i40e/i40e_txrx.h
> @@ -487,7 +487,7 @@ u32 i40e_get_tx_pending(struct i40e_ring *ring, bool in_sw);
> void i40e_detect_recover_hung(struct i40e_vsi *vsi);
> int __i40e_maybe_stop_tx(struct i40e_ring *tx_ring, int size);
> bool __i40e_chk_linearize(struct sk_buff *skb);
> -int i40e_xdp_xmit(struct net_device *dev, struct xdp_frame *xdpf);
> +int i40e_xdp_xmit(struct net_device *dev, int n, struct xdp_frame **frames);
> void i40e_xdp_flush(struct net_device *dev);
>
> /**
> diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> index 6652b201df5b..9645619f7729 100644
> --- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> +++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> @@ -10017,11 +10017,13 @@ static int ixgbe_xdp(struct net_device *dev, struct netdev_bpf *xdp)
> }
> }
>
> -static int ixgbe_xdp_xmit(struct net_device *dev, struct xdp_frame *xdpf)
> +static int ixgbe_xdp_xmit(struct net_device *dev, int n,
> + struct xdp_frame **frames)
> {
> struct ixgbe_adapter *adapter = netdev_priv(dev);
> struct ixgbe_ring *ring;
> - int err;
> + int drops = 0;
> + int i;
>
> if (unlikely(test_bit(__IXGBE_DOWN, &adapter->state)))
> return -ENETDOWN;
> @@ -10033,11 +10035,18 @@ static int ixgbe_xdp_xmit(struct net_device *dev, struct xdp_frame *xdpf)
> if (unlikely(!ring))
> return -ENXIO;
>
> - err = ixgbe_xmit_xdp_ring(adapter, xdpf);
> - if (err != IXGBE_XDP_TX)
> - return -ENOSPC;
> + for (i = 0; i < n; i++) {
> + struct xdp_frame *xdpf = frames[i];
> + int err;
>
> - return 0;
> + err = ixgbe_xmit_xdp_ring(adapter, xdpf);
> + if (err != IXGBE_XDP_TX) {
> + xdp_return_frame_rx_napi(xdpf);
> + drops++;
> + }
> + }
Same as above, how about just aborting if we get an error?
> +
> + return n - drops;
> }
>
> static void ixgbe_xdp_flush(struct net_device *dev)
> diff --git a/drivers/net/tun.c b/drivers/net/tun.c
> index 44d4f3d25350..d3dcfcb1c4b3 100644
> --- a/drivers/net/tun.c
> +++ b/drivers/net/tun.c
> @@ -70,6 +70,7 @@
> #include <net/netns/generic.h>
> #include <net/rtnetlink.h>
> #include <net/sock.h>
> +#include <net/xdp.h>
> #include <linux/seq_file.h>
> #include <linux/uio.h>
> #include <linux/skb_array.h>
> @@ -1290,34 +1291,44 @@ static const struct net_device_ops tun_netdev_ops = {
> .ndo_get_stats64 = tun_net_get_stats64,
> };
>
> -static int tun_xdp_xmit(struct net_device *dev, struct xdp_frame *frame)
> +static int tun_xdp_xmit(struct net_device *dev, int n, struct xdp_frame **frames)
> {
> struct tun_struct *tun = netdev_priv(dev);
> struct tun_file *tfile;
> u32 numqueues;
> - int ret = 0;
> + int drops = 0;
> + int cnt = n;
> + int i;
>
> rcu_read_lock();
>
> numqueues = READ_ONCE(tun->numqueues);
> if (!numqueues) {
> - ret = -ENOSPC;
> - goto out;
> + rcu_read_unlock();
> + return -ENXIO; /* Caller will free/return all frames */
> }
>
> tfile = rcu_dereference(tun->tfiles[smp_processor_id() %
> numqueues]);
> - /* Encode the XDP flag into lowest bit for consumer to differ
> - * XDP buffer from sk_buff.
> - */
> - if (ptr_ring_produce(&tfile->tx_ring, tun_xdp_to_ptr(frame))) {
> - this_cpu_inc(tun->pcpu_stats->tx_dropped);
> - ret = -ENOSPC;
> +
> + spin_lock(&tfile->tx_ring.producer_lock);
> + for (i = 0; i < n; i++) {
> + struct xdp_frame *xdp = frames[i];
> + /* Encode the XDP flag into lowest bit for consumer to differ
> + * XDP buffer from sk_buff.
> + */
> + void *frame = tun_xdp_to_ptr(xdp);
> +
> + if (__ptr_ring_produce(&tfile->tx_ring, frame)) {
> + this_cpu_inc(tun->pcpu_stats->tx_dropped);
> + xdp_return_frame_rx_napi(xdp);
> + drops++;
> + }
We have a ptr_ring_consume_batched() API it seems we should also have a
ptr_ring_produce_batched() now as well. Could be a follow up patch but
I think its probably worth considering.
> }
> + spin_unlock(&tfile->tx_ring.producer_lock);
>
> -out:
> rcu_read_unlock();
> - return ret;
> + return cnt - drops;
> }
>
> static int tun_xdp_tx(struct net_device *dev, struct xdp_buff *xdp)
> @@ -1327,7 +1338,7 @@ static int tun_xdp_tx(struct net_device *dev, struct xdp_buff *xdp)
> if (unlikely(!frame))
> return -EOVERFLOW;
>
> - return tun_xdp_xmit(dev, frame);
> + return tun_xdp_xmit(dev, 1, &frame);
> }
>
> static void tun_xdp_flush(struct net_device *dev)
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index f34794a76c4d..39a0783d1cde 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -419,23 +419,13 @@ static void virtnet_xdp_flush(struct net_device *dev)
> virtqueue_kick(sq->vq);
> }
>
> -static int __virtnet_xdp_xmit(struct virtnet_info *vi,
> - struct xdp_frame *xdpf)
> +static int __virtnet_xdp_xmit_one(struct virtnet_info *vi,
> + struct send_queue *sq,
> + struct xdp_frame *xdpf)
> {
> struct virtio_net_hdr_mrg_rxbuf *hdr;
> - struct xdp_frame *xdpf_sent;
> - struct send_queue *sq;
> - unsigned int len;
> - unsigned int qp;
> int err;
>
> - qp = vi->curr_queue_pairs - vi->xdp_queue_pairs + smp_processor_id();
> - sq = &vi->sq[qp];
> -
> - /* Free up any pending old buffers before queueing new ones. */
> - while ((xdpf_sent = virtqueue_get_buf(sq->vq, &len)) != NULL)
> - xdp_return_frame(xdpf_sent);
> -
> /* virtqueue want to use data area in-front of packet */
> if (unlikely(xdpf->metasize > 0))
> return -EOPNOTSUPP;
> @@ -459,11 +449,40 @@ static int __virtnet_xdp_xmit(struct virtnet_info *vi,
> return 0;
> }
>
> -static int virtnet_xdp_xmit(struct net_device *dev, struct xdp_frame *xdpf)
> +static int __virtnet_xdp_tx_xmit(struct virtnet_info *vi,
> + struct xdp_frame *xdpf)
> +{
> + struct xdp_frame *xdpf_sent;
> + struct send_queue *sq;
> + unsigned int len;
> + unsigned int qp;
> +
> + qp = vi->curr_queue_pairs - vi->xdp_queue_pairs + smp_processor_id();
> + sq = &vi->sq[qp];
> +
> + /* Free up any pending old buffers before queueing new ones. */
> + while ((xdpf_sent = virtqueue_get_buf(sq->vq, &len)) != NULL)
> + xdp_return_frame(xdpf_sent);
> +
> + return __virtnet_xdp_xmit_one(vi, sq, xdpf);
> +}
> +
> +static int virtnet_xdp_xmit(struct net_device *dev,
> + int n, struct xdp_frame **frames)
> {
> struct virtnet_info *vi = netdev_priv(dev);
> struct receive_queue *rq = vi->rq;
> + struct xdp_frame *xdpf_sent;
> struct bpf_prog *xdp_prog;
> + struct send_queue *sq;
> + unsigned int len;
> + unsigned int qp;
> + int drops = 0;
> + int err;
> + int i;
> +
> + qp = vi->curr_queue_pairs - vi->xdp_queue_pairs + smp_processor_id();
> + sq = &vi->sq[qp];
>
> /* Only allow ndo_xdp_xmit if XDP is loaded on dev, as this
> * indicate XDP resources have been successfully allocated.
> @@ -472,7 +491,20 @@ static int virtnet_xdp_xmit(struct net_device *dev, struct xdp_frame *xdpf)
> if (!xdp_prog)
> return -ENXIO;
>
> - return __virtnet_xdp_xmit(vi, xdpf);
> + /* Free up any pending old buffers before queueing new ones. */
> + while ((xdpf_sent = virtqueue_get_buf(sq->vq, &len)) != NULL)
> + xdp_return_frame(xdpf_sent);
> +
> + for (i = 0; i < n; i++) {
> + struct xdp_frame *xdpf = frames[i];
> +
> + err = __virtnet_xdp_xmit_one(vi, sq, xdpf);
> + if (err) {
> + xdp_return_frame_rx_napi(xdpf);
> + drops++;
> + }
> + }
> + return n - drops;
> }
>
> static unsigned int virtnet_get_headroom(struct virtnet_info *vi)
> @@ -616,7 +648,7 @@ static struct sk_buff *receive_small(struct net_device *dev,
> xdpf = convert_to_xdp_frame(&xdp);
> if (unlikely(!xdpf))
> goto err_xdp;
> - err = __virtnet_xdp_xmit(vi, xdpf);
> + err = __virtnet_xdp_tx_xmit(vi, xdpf);
> if (unlikely(err)) {
> trace_xdp_exception(vi->dev, xdp_prog, act);
> goto err_xdp;
> @@ -779,7 +811,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
> xdpf = convert_to_xdp_frame(&xdp);
> if (unlikely(!xdpf))
> goto err_xdp;
> - err = __virtnet_xdp_xmit(vi, xdpf);
> + err = __virtnet_xdp_tx_xmit(vi, xdpf);
> if (unlikely(err)) {
> trace_xdp_exception(vi->dev, xdp_prog, act);
> if (unlikely(xdp_page != page))
> diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
> index 03ed492c4e14..debdb6286170 100644
> --- a/include/linux/netdevice.h
> +++ b/include/linux/netdevice.h
> @@ -1185,9 +1185,13 @@ struct dev_ifalias {
> * This function is used to set or query state related to XDP on the
> * netdevice and manage BPF offload. See definition of
> * enum bpf_netdev_command for details.
> - * int (*ndo_xdp_xmit)(struct net_device *dev, struct xdp_frame *xdp);
> - * This function is used to submit a XDP packet for transmit on a
> - * netdevice.
> + * int (*ndo_xdp_xmit)(struct net_device *dev, int n, struct xdp_frame **xdp);
> + * This function is used to submit @n XDP packets for transmit on a
> + * netdevice. Returns number of frames successfully transmitted, frames
> + * that got dropped are freed/returned via xdp_return_frame().
> + * Returns negative number, means general error invoking ndo, meaning
> + * no frames were xmit'ed and core-caller will free all frames.
> + * TODO: Consider add flag to allow sending flush operation.
> * void (*ndo_xdp_flush)(struct net_device *dev);
> * This function is used to inform the driver to flush a particular
> * xdp tx queue. Must be called on same CPU as xdp_xmit.
> @@ -1375,8 +1379,8 @@ struct net_device_ops {
> int needed_headroom);
> int (*ndo_bpf)(struct net_device *dev,
> struct netdev_bpf *bpf);
> - int (*ndo_xdp_xmit)(struct net_device *dev,
> - struct xdp_frame *xdp);
> + int (*ndo_xdp_xmit)(struct net_device *dev, int n,
> + struct xdp_frame **xdp);
> void (*ndo_xdp_flush)(struct net_device *dev);
> };
>
> diff --git a/kernel/bpf/devmap.c b/kernel/bpf/devmap.c
> index 6f84100723b0..1317629662ae 100644
> --- a/kernel/bpf/devmap.c
> +++ b/kernel/bpf/devmap.c
> @@ -222,7 +222,7 @@ static int bq_xmit_all(struct bpf_dtab_netdev *obj,
> struct xdp_bulk_queue *bq)
> {
> struct net_device *dev = obj->dev;
> - int sent = 0, drops = 0;
> + int sent = 0, drops = 0, err = 0;
> int i;
>
> if (unlikely(!bq->count))
> @@ -234,23 +234,32 @@ static int bq_xmit_all(struct bpf_dtab_netdev *obj,
> prefetch(xdpf);
> }
>
> - for (i = 0; i < bq->count; i++) {
> - struct xdp_frame *xdpf = bq->q[i];
> - int err;
> -
> - err = dev->netdev_ops->ndo_xdp_xmit(dev, xdpf);
> - if (err) {
> - drops++;
> - xdp_return_frame(xdpf);
> - }
> - sent++;
> + sent = dev->netdev_ops->ndo_xdp_xmit(dev, bq->count, bq->q);
> + if (sent < 0) {
> + err = sent;
> + sent = 0;
> + goto error;
> }
> + drops = bq->count - sent;
> +out:
> bq->count = 0;
>
> trace_xdp_devmap_xmit(&obj->dtab->map, obj->bit,
> sent, drops, bq->dev_rx, dev);
> bq->dev_rx = NULL;
> return 0;
> +error:
> + /* If ndo_xdp_xmit fails with an errno, no frames have been
> + * xmit'ed and it's our responsibility to them free all.
> + */
> + for (i = 0; i < bq->count; i++) {
> + struct xdp_frame *xdpf = bq->q[i];
> +
> + /* RX path under NAPI protection, can return frames faster */
> + xdp_return_frame_rx_napi(xdpf);
> + drops++;
> + }
> + goto out;
> }
>
> /* __dev_map_flush is called from xdp_do_flush_map() which _must_ be signaled
> diff --git a/net/core/filter.c b/net/core/filter.c
> index 4a93423cc5ea..19504b7f4959 100644
> --- a/net/core/filter.c
> +++ b/net/core/filter.c
> @@ -3035,7 +3035,7 @@ static int __bpf_tx_xdp(struct net_device *dev,
> u32 index)
> {
> struct xdp_frame *xdpf;
> - int err;
> + int sent;
>
> if (!dev->netdev_ops->ndo_xdp_xmit) {
> return -EOPNOTSUPP;
> @@ -3045,9 +3045,9 @@ static int __bpf_tx_xdp(struct net_device *dev,
> if (unlikely(!xdpf))
> return -EOVERFLOW;
>
> - err = dev->netdev_ops->ndo_xdp_xmit(dev, xdpf);
> - if (err)
> - return err;
> + sent = dev->netdev_ops->ndo_xdp_xmit(dev, 1, &xdpf);
> + if (sent <= 0)
> + return sent;
> dev->netdev_ops->ndo_xdp_flush(dev);
> return 0;
> }
>
^ permalink raw reply
* [PATCH net-next 4/4] net/smc: longer delay when freeing client link groups
From: Ursula Braun @ 2018-05-23 14:38 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, schwidefsky, heiko.carstens, raspl, ubraun
In-Reply-To: <20180523143812.25824-1-ubraun@linux.ibm.com>
Client link group creation always follows the server linkgroup creation.
If peer creates a new server link group, client has to create a new
client link group. If peer reuses a server link group for a new
connection, client has to reuse its client link group as well. To
avoid out-of-sync conditions for link groups a longer delay for
for client link group removal is defined to make sure this link group
still exists, once the peer decides to reuse a server link group.
Currently the client link group delay time is just 10 jiffies larger
than the server link group delay time. This patch increases the delay
difference to 10 seconds to have a better protection against
out-of-sync link groups.
Signed-off-by: Ursula Braun <ubraun@linux.ibm.com>
---
net/smc/smc_core.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/net/smc/smc_core.c b/net/smc/smc_core.c
index 2bf138e7d3ec..add82b0266f3 100644
--- a/net/smc/smc_core.c
+++ b/net/smc/smc_core.c
@@ -28,7 +28,7 @@
#define SMC_LGR_NUM_INCR 256
#define SMC_LGR_FREE_DELAY_SERV (600 * HZ)
-#define SMC_LGR_FREE_DELAY_CLNT (SMC_LGR_FREE_DELAY_SERV + 10)
+#define SMC_LGR_FREE_DELAY_CLNT (SMC_LGR_FREE_DELAY_SERV + 10 * HZ)
static struct smc_lgr_list smc_lgr_list = { /* established link groups */
.lock = __SPIN_LOCK_UNLOCKED(smc_lgr_list.lock),
--
2.16.3
^ permalink raw reply related
* [PATCH net-next 3/4] net/smc: urgent data support
From: Ursula Braun @ 2018-05-23 14:38 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, schwidefsky, heiko.carstens, raspl, ubraun
In-Reply-To: <20180523143812.25824-1-ubraun@linux.ibm.com>
From: Stefan Raspl <raspl@linux.ibm.com>
Add support for out of band data send and receive.
Signed-off-by: Stefan Raspl <raspl@linux.ibm.com>
Signed-off-by: Ursula Braun <ubraun@linux.ibm.com>
---
net/smc/af_smc.c | 24 ++++++++++-
net/smc/smc.h | 15 +++++++
net/smc/smc_cdc.c | 44 ++++++++++++++++++--
net/smc/smc_cdc.h | 13 ++++++
net/smc/smc_core.c | 1 +
net/smc/smc_rx.c | 120 +++++++++++++++++++++++++++++++++++++++++++++++------
net/smc/smc_tx.c | 55 ++++++++++++++++--------
net/smc/smc_tx.h | 2 +-
8 files changed, 238 insertions(+), 36 deletions(-)
diff --git a/net/smc/af_smc.c b/net/smc/af_smc.c
index f2d925921d81..2c369d4bb1c1 100644
--- a/net/smc/af_smc.c
+++ b/net/smc/af_smc.c
@@ -8,8 +8,6 @@
*
* Initial restrictions:
* - support for alternate links postponed
- * - partial support for non-blocking sockets only
- * - support for urgent data postponed
*
* Copyright IBM Corp. 2016, 2018
*
@@ -1338,6 +1336,8 @@ static __poll_t smc_poll(struct file *file, struct socket *sock,
if (sk->sk_state == SMC_APPCLOSEWAIT1)
mask |= EPOLLIN;
}
+ if (smc->conn.urg_state == SMC_URG_VALID)
+ mask |= EPOLLPRI;
}
release_sock(sk);
@@ -1477,10 +1477,13 @@ static int smc_getsockopt(struct socket *sock, int level, int optname,
static int smc_ioctl(struct socket *sock, unsigned int cmd,
unsigned long arg)
{
+ union smc_host_cursor cons, urg;
+ struct smc_connection *conn;
struct smc_sock *smc;
int answ;
smc = smc_sk(sock->sk);
+ conn = &smc->conn;
if (smc->use_fallback) {
if (!smc->clcsock)
return -EBADF;
@@ -1517,6 +1520,23 @@ static int smc_ioctl(struct socket *sock, unsigned int cmd,
else
answ = smc_tx_prepared_sends(&smc->conn);
break;
+ case SIOCATMARK:
+ if (smc->sk.sk_state == SMC_LISTEN)
+ return -EINVAL;
+ if (smc->sk.sk_state == SMC_INIT ||
+ smc->sk.sk_state == SMC_CLOSED) {
+ answ = 0;
+ } else {
+ smc_curs_write(&cons,
+ smc_curs_read(&conn->local_tx_ctrl.cons, conn),
+ conn);
+ smc_curs_write(&urg,
+ smc_curs_read(&conn->urg_curs, conn),
+ conn);
+ answ = smc_curs_diff(conn->rmb_desc->len,
+ &cons, &urg) == 1;
+ }
+ break;
default:
return -ENOIOCTLCMD;
}
diff --git a/net/smc/smc.h b/net/smc/smc.h
index a1467e411645..51ae1f10d81a 100644
--- a/net/smc/smc.h
+++ b/net/smc/smc.h
@@ -114,6 +114,12 @@ struct smc_host_cdc_msg { /* Connection Data Control message */
u8 reserved[18];
} __aligned(8);
+enum smc_urg_state {
+ SMC_URG_VALID, /* data present */
+ SMC_URG_NOTYET, /* data pending */
+ SMC_URG_READ /* data was already read */
+};
+
struct smc_connection {
struct rb_node alert_node;
struct smc_link_group *lgr; /* link group of connection */
@@ -160,6 +166,15 @@ struct smc_connection {
union smc_host_cursor rx_curs_confirmed; /* confirmed to peer
* source of snd_una ?
*/
+ union smc_host_cursor urg_curs; /* points at urgent byte */
+ enum smc_urg_state urg_state;
+ bool urg_tx_pend; /* urgent data staged */
+ bool urg_rx_skip_pend;
+ /* indicate urgent oob data
+ * read, but previous regular
+ * data still pending
+ */
+ char urg_rx_byte; /* urgent byte */
atomic_t bytes_to_rcv; /* arrived data,
* not yet received
*/
diff --git a/net/smc/smc_cdc.c b/net/smc/smc_cdc.c
index 8d2c079c87b0..a7e8d63fc8ae 100644
--- a/net/smc/smc_cdc.c
+++ b/net/smc/smc_cdc.c
@@ -164,6 +164,28 @@ static inline bool smc_cdc_before(u16 seq1, u16 seq2)
return (s16)(seq1 - seq2) < 0;
}
+static void smc_cdc_handle_urg_data_arrival(struct smc_sock *smc,
+ int *diff_prod)
+{
+ struct smc_connection *conn = &smc->conn;
+ char *base;
+
+ /* new data included urgent business */
+ smc_curs_write(&conn->urg_curs,
+ smc_curs_read(&conn->local_rx_ctrl.prod, conn),
+ conn);
+ conn->urg_state = SMC_URG_VALID;
+ if (!sock_flag(&smc->sk, SOCK_URGINLINE))
+ /* we'll skip the urgent byte, so don't account for it */
+ (*diff_prod)--;
+ base = (char *)conn->rmb_desc->cpu_addr;
+ if (conn->urg_curs.count)
+ conn->urg_rx_byte = *(base + conn->urg_curs.count - 1);
+ else
+ conn->urg_rx_byte = *(base + conn->rmb_desc->len - 1);
+ sk_send_sigurg(&smc->sk);
+}
+
static void smc_cdc_msg_recv_action(struct smc_sock *smc,
struct smc_cdc_msg *cdc)
{
@@ -194,15 +216,25 @@ static void smc_cdc_msg_recv_action(struct smc_sock *smc,
diff_prod = smc_curs_diff(conn->rmb_desc->len, &prod_old,
&conn->local_rx_ctrl.prod);
if (diff_prod) {
+ if (conn->local_rx_ctrl.prod_flags.urg_data_present)
+ smc_cdc_handle_urg_data_arrival(smc, &diff_prod);
/* bytes_to_rcv is decreased in smc_recvmsg */
smp_mb__before_atomic();
atomic_add(diff_prod, &conn->bytes_to_rcv);
/* guarantee 0 <= bytes_to_rcv <= rmb_desc->len */
smp_mb__after_atomic();
smc->sk.sk_data_ready(&smc->sk);
- } else if ((conn->local_rx_ctrl.prod_flags.write_blocked) ||
- (conn->local_rx_ctrl.prod_flags.cons_curs_upd_req)) {
- smc->sk.sk_data_ready(&smc->sk);
+ } else {
+ if (conn->local_rx_ctrl.prod_flags.write_blocked ||
+ conn->local_rx_ctrl.prod_flags.cons_curs_upd_req ||
+ conn->local_rx_ctrl.prod_flags.urg_data_pending) {
+ if (conn->local_rx_ctrl.prod_flags.urg_data_pending)
+ conn->urg_state = SMC_URG_NOTYET;
+ /* force immediate tx of current consumer cursor, but
+ * under send_lock to guarantee arrival in seqno-order
+ */
+ smc_tx_sndbuf_nonempty(conn);
+ }
}
/* piggy backed tx info */
@@ -212,6 +244,12 @@ static void smc_cdc_msg_recv_action(struct smc_sock *smc,
/* trigger socket release if connection closed */
smc_close_wake_tx_prepared(smc);
}
+ if (diff_cons && conn->urg_tx_pend &&
+ atomic_read(&conn->peer_rmbe_space) == conn->peer_rmbe_size) {
+ /* urg data confirmed by peer, indicate we're ready for more */
+ conn->urg_tx_pend = false;
+ smc->sk.sk_write_space(&smc->sk);
+ }
if (conn->local_rx_ctrl.conn_state_flags.peer_conn_abort) {
smc->sk.sk_err = ECONNRESET;
diff --git a/net/smc/smc_cdc.h b/net/smc/smc_cdc.h
index d2012fd22100..f60082fee5b8 100644
--- a/net/smc/smc_cdc.h
+++ b/net/smc/smc_cdc.h
@@ -146,6 +146,19 @@ static inline int smc_curs_diff(unsigned int size,
return max_t(int, 0, (new->count - old->count));
}
+/* calculate cursor difference between old and new - returns negative
+ * value in case old > new
+ */
+static inline int smc_curs_comp(unsigned int size,
+ union smc_host_cursor *old,
+ union smc_host_cursor *new)
+{
+ if (old->wrap > new->wrap ||
+ (old->wrap == new->wrap && old->count > new->count))
+ return -smc_curs_diff(size, new, old);
+ return smc_curs_diff(size, old, new);
+}
+
static inline void smc_host_cursor_to_cdc(union smc_cdc_cursor *peer,
union smc_host_cursor *local,
struct smc_connection *conn)
diff --git a/net/smc/smc_core.c b/net/smc/smc_core.c
index 21c244f53b0a..2bf138e7d3ec 100644
--- a/net/smc/smc_core.c
+++ b/net/smc/smc_core.c
@@ -544,6 +544,7 @@ int smc_conn_create(struct smc_sock *smc,
}
conn->local_tx_ctrl.common.type = SMC_CDC_MSG_TYPE;
conn->local_tx_ctrl.len = SMC_WR_TX_SIZE;
+ conn->urg_state = SMC_URG_READ;
#ifndef KERNEL_HAS_ATOMIC64
spin_lock_init(&conn->acurs_lock);
#endif
diff --git a/net/smc/smc_rx.c b/net/smc/smc_rx.c
index 290a434471d1..3d77b383cccd 100644
--- a/net/smc/smc_rx.c
+++ b/net/smc/smc_rx.c
@@ -47,16 +47,59 @@ static void smc_rx_wake_up(struct sock *sk)
* @conn connection to update
* @cons consumer cursor
* @len number of Bytes consumed
+ * Returns:
+ * 1 if we should end our receive, 0 otherwise
*/
-static void smc_rx_update_consumer(struct smc_connection *conn,
- union smc_host_cursor cons, size_t len)
+static int smc_rx_update_consumer(struct smc_sock *smc,
+ union smc_host_cursor cons, size_t len)
{
+ struct smc_connection *conn = &smc->conn;
+ struct sock *sk = &smc->sk;
+ bool force = false;
+ int diff, rc = 0;
+
smc_curs_add(conn->rmb_desc->len, &cons, len);
+
+ /* did we process urgent data? */
+ if (conn->urg_state == SMC_URG_VALID || conn->urg_rx_skip_pend) {
+ diff = smc_curs_comp(conn->rmb_desc->len, &cons,
+ &conn->urg_curs);
+ if (sock_flag(sk, SOCK_URGINLINE)) {
+ if (diff == 0) {
+ force = true;
+ rc = 1;
+ conn->urg_state = SMC_URG_READ;
+ }
+ } else {
+ if (diff == 1) {
+ /* skip urgent byte */
+ force = true;
+ smc_curs_add(conn->rmb_desc->len, &cons, 1);
+ conn->urg_rx_skip_pend = false;
+ } else if (diff < -1)
+ /* we read past urgent byte */
+ conn->urg_state = SMC_URG_READ;
+ }
+ }
+
smc_curs_write(&conn->local_tx_ctrl.cons, smc_curs_read(&cons, conn),
conn);
+
/* send consumer cursor update if required */
/* similar to advertising new TCP rcv_wnd if required */
- smc_tx_consumer_update(conn);
+ smc_tx_consumer_update(conn, force);
+
+ return rc;
+}
+
+static void smc_rx_update_cons(struct smc_sock *smc, size_t len)
+{
+ struct smc_connection *conn = &smc->conn;
+ union smc_host_cursor cons;
+
+ smc_curs_write(&cons, smc_curs_read(&conn->local_tx_ctrl.cons, conn),
+ conn);
+ smc_rx_update_consumer(smc, cons, len);
}
struct smc_spd_priv {
@@ -70,7 +113,6 @@ static void smc_rx_pipe_buf_release(struct pipe_inode_info *pipe,
struct smc_spd_priv *priv = (struct smc_spd_priv *)buf->private;
struct smc_sock *smc = priv->smc;
struct smc_connection *conn;
- union smc_host_cursor cons;
struct sock *sk = &smc->sk;
if (sk->sk_state == SMC_CLOSED ||
@@ -79,9 +121,7 @@ static void smc_rx_pipe_buf_release(struct pipe_inode_info *pipe,
goto out;
conn = &smc->conn;
lock_sock(sk);
- smc_curs_write(&cons, smc_curs_read(&conn->local_tx_ctrl.cons, conn),
- conn);
- smc_rx_update_consumer(conn, cons, priv->len);
+ smc_rx_update_cons(smc, priv->len);
release_sock(sk);
if (atomic_sub_and_test(priv->len, &conn->splice_pending))
smc_rx_wake_up(sk);
@@ -184,6 +224,52 @@ int smc_rx_wait(struct smc_sock *smc, long *timeo,
return rc;
}
+static int smc_rx_recv_urg(struct smc_sock *smc, struct msghdr *msg, int len,
+ int flags)
+{
+ struct smc_connection *conn = &smc->conn;
+ union smc_host_cursor cons;
+ struct sock *sk = &smc->sk;
+ int rc = 0;
+
+ if (sock_flag(sk, SOCK_URGINLINE) ||
+ !(conn->urg_state == SMC_URG_VALID) ||
+ conn->urg_state == SMC_URG_READ)
+ return -EINVAL;
+
+ if (conn->urg_state == SMC_URG_VALID) {
+ if (!(flags & MSG_PEEK))
+ smc->conn.urg_state = SMC_URG_READ;
+ msg->msg_flags |= MSG_OOB;
+ if (len > 0) {
+ if (!(flags & MSG_TRUNC))
+ rc = memcpy_to_msg(msg, &conn->urg_rx_byte, 1);
+ len = 1;
+ smc_curs_write(&cons,
+ smc_curs_read(&conn->local_tx_ctrl.cons,
+ conn),
+ conn);
+ if (smc_curs_diff(conn->rmb_desc->len, &cons,
+ &conn->urg_curs) > 1)
+ conn->urg_rx_skip_pend = true;
+ /* Urgent Byte was already accounted for, but trigger
+ * skipping the urgent byte in non-inline case
+ */
+ if (!(flags & MSG_PEEK))
+ smc_rx_update_consumer(smc, cons, 0);
+ } else {
+ msg->msg_flags |= MSG_TRUNC;
+ }
+
+ return rc ? -EFAULT : len;
+ }
+
+ if (sk->sk_state == SMC_CLOSED || sk->sk_shutdown & RCV_SHUTDOWN)
+ return 0;
+
+ return -EAGAIN;
+}
+
/* smc_rx_recvmsg - receive data from RMBE
* @msg: copy data to receive buffer
* @pipe: copy data to pipe if set - indicates splice() call
@@ -209,12 +295,12 @@ int smc_rx_recvmsg(struct smc_sock *smc, struct msghdr *msg,
if (unlikely(flags & MSG_ERRQUEUE))
return -EINVAL; /* future work for sk.sk_family == AF_SMC */
- if (flags & MSG_OOB)
- return -EINVAL; /* future work */
sk = &smc->sk;
if (sk->sk_state == SMC_LISTEN)
return -ENOTCONN;
+ if (flags & MSG_OOB)
+ return smc_rx_recv_urg(smc, msg, len, flags);
timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
target = sock_rcvlowat(sk, flags & MSG_WAITALL, len);
@@ -227,6 +313,9 @@ int smc_rx_recvmsg(struct smc_sock *smc, struct msghdr *msg,
if (atomic_read(&conn->bytes_to_rcv))
goto copy;
+ else if (conn->urg_state == SMC_URG_VALID)
+ /* we received a single urgent Byte - skip */
+ smc_rx_update_cons(smc, 0);
if (sk->sk_shutdown & RCV_SHUTDOWN ||
smc_cdc_rxed_any_close_or_senddone(conn) ||
@@ -281,14 +370,18 @@ int smc_rx_recvmsg(struct smc_sock *smc, struct msghdr *msg,
continue;
}
- /* not more than what user space asked for */
- copylen = min_t(size_t, read_remaining, readable);
smc_curs_write(&cons,
smc_curs_read(&conn->local_tx_ctrl.cons, conn),
conn);
/* subsequent splice() calls pick up where previous left */
if (splbytes)
smc_curs_add(conn->rmb_desc->len, &cons, splbytes);
+ if (conn->urg_state == SMC_URG_VALID &&
+ sock_flag(&smc->sk, SOCK_URGINLINE) &&
+ readable > 1)
+ readable--; /* always stop at urgent Byte */
+ /* not more than what user space asked for */
+ copylen = min_t(size_t, read_remaining, readable);
/* determine chunks where to read from rcvbuf */
/* either unwrapped case, or 1st chunk of wrapped case */
chunk_len = min_t(size_t, copylen, conn->rmb_desc->len -
@@ -333,8 +426,8 @@ int smc_rx_recvmsg(struct smc_sock *smc, struct msghdr *msg,
atomic_sub(copylen, &conn->bytes_to_rcv);
/* guarantee 0 <= bytes_to_rcv <= rmb_desc->len */
smp_mb__after_atomic();
- if (msg)
- smc_rx_update_consumer(conn, cons, copylen);
+ if (msg && smc_rx_update_consumer(smc, cons, copylen))
+ goto out;
}
} while (read_remaining);
out:
@@ -346,4 +439,5 @@ void smc_rx_init(struct smc_sock *smc)
{
smc->sk.sk_data_ready = smc_rx_wake_up;
atomic_set(&smc->conn.splice_pending, 0);
+ smc->conn.urg_state = SMC_URG_READ;
}
diff --git a/net/smc/smc_tx.c b/net/smc/smc_tx.c
index 1f4a38b857f0..cee666400752 100644
--- a/net/smc/smc_tx.c
+++ b/net/smc/smc_tx.c
@@ -32,7 +32,7 @@
/***************************** sndbuf producer *******************************/
/* callback implementation for sk.sk_write_space()
- * to wakeup sndbuf producers that blocked with smc_tx_wait_memory().
+ * to wakeup sndbuf producers that blocked with smc_tx_wait().
* called under sk_socket lock.
*/
static void smc_tx_write_space(struct sock *sk)
@@ -56,7 +56,7 @@ static void smc_tx_write_space(struct sock *sk)
}
}
-/* Wakeup sndbuf producers that blocked with smc_tx_wait_memory().
+/* Wakeup sndbuf producers that blocked with smc_tx_wait().
* Cf. tcp_data_snd_check()=>tcp_check_space()=>tcp_new_space().
*/
void smc_tx_sndbuf_nonfull(struct smc_sock *smc)
@@ -66,8 +66,10 @@ void smc_tx_sndbuf_nonfull(struct smc_sock *smc)
smc->sk.sk_write_space(&smc->sk);
}
-/* blocks sndbuf producer until at least one byte of free space available */
-static int smc_tx_wait_memory(struct smc_sock *smc, int flags)
+/* blocks sndbuf producer until at least one byte of free space available
+ * or urgent Byte was consumed
+ */
+static int smc_tx_wait(struct smc_sock *smc, int flags)
{
DEFINE_WAIT_FUNC(wait, woken_wake_function);
struct smc_connection *conn = &smc->conn;
@@ -103,14 +105,15 @@ static int smc_tx_wait_memory(struct smc_sock *smc, int flags)
break;
}
sk_clear_bit(SOCKWQ_ASYNC_NOSPACE, sk);
- if (atomic_read(&conn->sndbuf_space))
- break; /* at least 1 byte of free space available */
+ if (atomic_read(&conn->sndbuf_space) && !conn->urg_tx_pend)
+ break; /* at least 1 byte of free & no urgent data */
set_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
sk_wait_event(sk, &timeo,
sk->sk_err ||
(sk->sk_shutdown & SEND_SHUTDOWN) ||
smc_cdc_rxed_any_close(conn) ||
- atomic_read(&conn->sndbuf_space),
+ (atomic_read(&conn->sndbuf_space) &&
+ !conn->urg_tx_pend),
&wait);
}
remove_wait_queue(sk_sleep(sk), &wait);
@@ -157,8 +160,11 @@ int smc_tx_sendmsg(struct smc_sock *smc, struct msghdr *msg, size_t len)
if (smc_cdc_rxed_any_close(conn))
return send_done ?: -ECONNRESET;
- if (!atomic_read(&conn->sndbuf_space)) {
- rc = smc_tx_wait_memory(smc, msg->msg_flags);
+ if (msg->msg_flags & MSG_OOB)
+ conn->local_tx_ctrl.prod_flags.urg_data_pending = 1;
+
+ if (!atomic_read(&conn->sndbuf_space) || conn->urg_tx_pend) {
+ rc = smc_tx_wait(smc, msg->msg_flags);
if (rc) {
if (send_done)
return send_done;
@@ -168,7 +174,7 @@ int smc_tx_sendmsg(struct smc_sock *smc, struct msghdr *msg, size_t len)
}
/* initialize variables for 1st iteration of subsequent loop */
- /* could be just 1 byte, even after smc_tx_wait_memory above */
+ /* could be just 1 byte, even after smc_tx_wait above */
writespace = atomic_read(&conn->sndbuf_space);
/* not more than what user space asked for */
copylen = min_t(size_t, send_remaining, writespace);
@@ -218,6 +224,8 @@ int smc_tx_sendmsg(struct smc_sock *smc, struct msghdr *msg, size_t len)
/* since we just produced more new data into sndbuf,
* trigger sndbuf consumer: RDMA write into peer RMBE and CDC
*/
+ if ((msg->msg_flags & MSG_OOB) && !send_remaining)
+ conn->urg_tx_pend = true;
if ((msg->msg_flags & MSG_MORE || smc_tx_is_corked(smc)) &&
(atomic_read(&conn->sndbuf_space) >
(conn->sndbuf_desc->len >> 1)))
@@ -299,6 +307,7 @@ static int smc_tx_rdma_writes(struct smc_connection *conn)
union smc_host_cursor sent, prep, prod, cons;
struct ib_sge sges[SMC_IB_MAX_SEND_SGE];
struct smc_link_group *lgr = conn->lgr;
+ struct smc_cdc_producer_flags *pflags;
int to_send, rmbespace;
struct smc_link *link;
dma_addr_t dma_addr;
@@ -326,7 +335,8 @@ static int smc_tx_rdma_writes(struct smc_connection *conn)
conn);
/* if usable snd_wnd closes ask peer to advertise once it opens again */
- conn->local_tx_ctrl.prod_flags.write_blocked = (to_send >= rmbespace);
+ pflags = &conn->local_tx_ctrl.prod_flags;
+ pflags->write_blocked = (to_send >= rmbespace);
/* cf. usable snd_wnd */
len = min(to_send, rmbespace);
@@ -391,6 +401,8 @@ static int smc_tx_rdma_writes(struct smc_connection *conn)
src_len_sum = src_len;
}
+ if (conn->urg_tx_pend && len == to_send)
+ pflags->urg_data_present = 1;
smc_tx_advance_cursors(conn, &prod, &sent, len);
/* update connection's cursors with advanced local cursors */
smc_curs_write(&conn->local_tx_ctrl.prod,
@@ -410,6 +422,7 @@ static int smc_tx_rdma_writes(struct smc_connection *conn)
*/
int smc_tx_sndbuf_nonempty(struct smc_connection *conn)
{
+ struct smc_cdc_producer_flags *pflags;
struct smc_cdc_tx_pend *pend;
struct smc_wr_buf *wr_buf;
int rc;
@@ -433,14 +446,21 @@ int smc_tx_sndbuf_nonempty(struct smc_connection *conn)
goto out_unlock;
}
- rc = smc_tx_rdma_writes(conn);
- if (rc) {
- smc_wr_tx_put_slot(&conn->lgr->lnk[SMC_SINGLE_LINK],
- (struct smc_wr_tx_pend_priv *)pend);
- goto out_unlock;
+ if (!conn->local_tx_ctrl.prod_flags.urg_data_present) {
+ rc = smc_tx_rdma_writes(conn);
+ if (rc) {
+ smc_wr_tx_put_slot(&conn->lgr->lnk[SMC_SINGLE_LINK],
+ (struct smc_wr_tx_pend_priv *)pend);
+ goto out_unlock;
+ }
}
rc = smc_cdc_msg_send(conn, wr_buf, pend);
+ pflags = &conn->local_tx_ctrl.prod_flags;
+ if (!rc && pflags->urg_data_present) {
+ pflags->urg_data_pending = 0;
+ pflags->urg_data_present = 0;
+ }
out_unlock:
spin_unlock_bh(&conn->send_lock);
@@ -473,7 +493,7 @@ void smc_tx_work(struct work_struct *work)
release_sock(&smc->sk);
}
-void smc_tx_consumer_update(struct smc_connection *conn)
+void smc_tx_consumer_update(struct smc_connection *conn, bool force)
{
union smc_host_cursor cfed, cons;
int to_confirm;
@@ -487,6 +507,7 @@ void smc_tx_consumer_update(struct smc_connection *conn)
to_confirm = smc_curs_diff(conn->rmb_desc->len, &cfed, &cons);
if (conn->local_rx_ctrl.prod_flags.cons_curs_upd_req ||
+ force ||
((to_confirm > conn->rmbe_update_limit) &&
((to_confirm > (conn->rmb_desc->len / 2)) ||
conn->local_rx_ctrl.prod_flags.write_blocked))) {
diff --git a/net/smc/smc_tx.h b/net/smc/smc_tx.h
index 44d077942976..9d2238909fa0 100644
--- a/net/smc/smc_tx.h
+++ b/net/smc/smc_tx.h
@@ -32,6 +32,6 @@ void smc_tx_init(struct smc_sock *smc);
int smc_tx_sendmsg(struct smc_sock *smc, struct msghdr *msg, size_t len);
int smc_tx_sndbuf_nonempty(struct smc_connection *conn);
void smc_tx_sndbuf_nonfull(struct smc_sock *smc);
-void smc_tx_consumer_update(struct smc_connection *conn);
+void smc_tx_consumer_update(struct smc_connection *conn, bool force);
#endif /* SMC_TX_H */
--
2.16.3
^ permalink raw reply related
* [PATCH net-next 2/4] net/smc: lock smc_lgr_list in port_terminate()
From: Ursula Braun @ 2018-05-23 14:38 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, schwidefsky, heiko.carstens, raspl, ubraun
In-Reply-To: <20180523143812.25824-1-ubraun@linux.ibm.com>
From: Hans Wippel <hwippel@linux.ibm.com>
Currently, smc_port_terminate() is not holding the lock of the lgr list
while it is traversing the list. This patch adds locking to this
function and changes smc_lgr_terminate() accordingly.
Signed-off-by: Hans Wippel <hwippel@linux.ibm.com>
Signed-off-by: Ursula Braun <ubraun@linux.ibm.com>
---
net/smc/smc_core.c | 16 +++++++++++++---
1 file changed, 13 insertions(+), 3 deletions(-)
diff --git a/net/smc/smc_core.c b/net/smc/smc_core.c
index 1e5c0e90a706..21c244f53b0a 100644
--- a/net/smc/smc_core.c
+++ b/net/smc/smc_core.c
@@ -346,7 +346,7 @@ void smc_lgr_forget(struct smc_link_group *lgr)
}
/* terminate linkgroup abnormally */
-void smc_lgr_terminate(struct smc_link_group *lgr)
+static void __smc_lgr_terminate(struct smc_link_group *lgr)
{
struct smc_connection *conn;
struct smc_sock *smc;
@@ -355,7 +355,8 @@ void smc_lgr_terminate(struct smc_link_group *lgr)
if (lgr->terminating)
return; /* lgr already terminating */
lgr->terminating = 1;
- smc_lgr_forget(lgr);
+ if (!list_empty(&lgr->list)) /* forget lgr */
+ list_del_init(&lgr->list);
smc_llc_link_inactive(&lgr->lnk[SMC_SINGLE_LINK]);
write_lock_bh(&lgr->conns_lock);
@@ -377,16 +378,25 @@ void smc_lgr_terminate(struct smc_link_group *lgr)
smc_lgr_schedule_free_work(lgr);
}
+void smc_lgr_terminate(struct smc_link_group *lgr)
+{
+ spin_lock_bh(&smc_lgr_list.lock);
+ __smc_lgr_terminate(lgr);
+ spin_unlock_bh(&smc_lgr_list.lock);
+}
+
/* Called when IB port is terminated */
void smc_port_terminate(struct smc_ib_device *smcibdev, u8 ibport)
{
struct smc_link_group *lgr, *l;
+ spin_lock_bh(&smc_lgr_list.lock);
list_for_each_entry_safe(lgr, l, &smc_lgr_list.list, list) {
if (lgr->lnk[SMC_SINGLE_LINK].smcibdev == smcibdev &&
lgr->lnk[SMC_SINGLE_LINK].ibport == ibport)
- smc_lgr_terminate(lgr);
+ __smc_lgr_terminate(lgr);
}
+ spin_unlock_bh(&smc_lgr_list.lock);
}
/* Determine vlan of internal TCP socket.
--
2.16.3
^ permalink raw reply related
* [PATCH net-next 1/4] net/smc: return 0 for ioctl calls in states INIT and CLOSED
From: Ursula Braun @ 2018-05-23 14:38 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, schwidefsky, heiko.carstens, raspl, ubraun
In-Reply-To: <20180523143812.25824-1-ubraun@linux.ibm.com>
A connected SMC-socket contains addresses of descriptors for the
send buffer and the rmb (receive buffer). Fields of these descriptors
are used to determine the answer for certain ioctl requests.
Add extra handling for unconnected SMC socket states without valid
buffer descriptor addresses.
Signed-off-by: Ursula Braun <ubraun@linux.ibm.com>
Reported-by: syzbot+e6714328fda813fc670f@syzkaller.appspotmail.com
---
net/smc/af_smc.c | 18 +++++++++++++++---
1 file changed, 15 insertions(+), 3 deletions(-)
diff --git a/net/smc/af_smc.c b/net/smc/af_smc.c
index 48530dab5c94..f2d925921d81 100644
--- a/net/smc/af_smc.c
+++ b/net/smc/af_smc.c
@@ -1490,20 +1490,32 @@ static int smc_ioctl(struct socket *sock, unsigned int cmd,
case SIOCINQ: /* same as FIONREAD */
if (smc->sk.sk_state == SMC_LISTEN)
return -EINVAL;
- answ = atomic_read(&smc->conn.bytes_to_rcv);
+ if (smc->sk.sk_state == SMC_INIT ||
+ smc->sk.sk_state == SMC_CLOSED)
+ answ = 0;
+ else
+ answ = atomic_read(&smc->conn.bytes_to_rcv);
break;
case SIOCOUTQ:
/* output queue size (not send + not acked) */
if (smc->sk.sk_state == SMC_LISTEN)
return -EINVAL;
- answ = smc->conn.sndbuf_desc->len -
+ if (smc->sk.sk_state == SMC_INIT ||
+ smc->sk.sk_state == SMC_CLOSED)
+ answ = 0;
+ else
+ answ = smc->conn.sndbuf_desc->len -
atomic_read(&smc->conn.sndbuf_space);
break;
case SIOCOUTQNSD:
/* output queue size (not send only) */
if (smc->sk.sk_state == SMC_LISTEN)
return -EINVAL;
- answ = smc_tx_prepared_sends(&smc->conn);
+ if (smc->sk.sk_state == SMC_INIT ||
+ smc->sk.sk_state == SMC_CLOSED)
+ answ = 0;
+ else
+ answ = smc_tx_prepared_sends(&smc->conn);
break;
default:
return -ENOIOCTLCMD;
--
2.16.3
^ permalink raw reply related
* [PATCH net-next 0/4] patches 2018-05-23
From: Ursula Braun @ 2018-05-23 14:38 UTC (permalink / raw)
To: davem; +Cc: netdev, linux-s390, schwidefsky, heiko.carstens, raspl, ubraun
Dave,
here are more smc-patches for net-next:
Patch 1 fixes an ioctl problem detected by syzbot.
Patch 2 improves smc_lgr_list locking in case of abnormal link
group termination. If you want to receive a version for the net-tree,
please let me know. It would look somewhat different, since the port
terminate code has been moved to smc_core.c on net-next.
Patch 3 enables SMC to deal with urgent data.
Patch 4 is a minor improvement to avoid out-of-sync linkgroups
between 2 peers.
Thanks, Ursula
Hans Wippel (1):
net/smc: lock smc_lgr_list in port_terminate()
Stefan Raspl (1):
net/smc: urgent data support
Ursula Braun (2):
net/smc: return 0 for ioctl calls in states INIT and CLOSED
net/smc: longer delay when freeing client link groups
net/smc/af_smc.c | 42 ++++++++++++++++---
net/smc/smc.h | 15 +++++++
net/smc/smc_cdc.c | 44 ++++++++++++++++++--
net/smc/smc_cdc.h | 13 ++++++
net/smc/smc_core.c | 19 +++++++--
net/smc/smc_rx.c | 120 +++++++++++++++++++++++++++++++++++++++++++++++------
net/smc/smc_tx.c | 55 ++++++++++++++++--------
net/smc/smc_tx.h | 2 +-
8 files changed, 267 insertions(+), 43 deletions(-)
--
2.16.3
^ permalink raw reply
* [PATCH net-next 2/2] cxgb4: do L1 config when module is inserted
From: Ganesh Goudar @ 2018-05-23 14:33 UTC (permalink / raw)
To: netdev, davem
Cc: nirranjan, indranil, venkatesh, linux-scsi, varun, Ganesh Goudar,
Casey Leedom
trigger an L1 configure operation when a transceiver module
is inserted in order to cause current "sticky" options like
Requested Forward Error Correction to be reapplied.
Signed-off-by: Casey Leedom <leedom@chelsio.com>
Signed-off-by: Ganesh Goudar <ganeshgr@chelsio.com>
---
drivers/net/ethernet/chelsio/cxgb4/cxgb4.h | 26 ++++++++++++++--
drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c | 11 +++++--
drivers/net/ethernet/chelsio/cxgb4/t4_hw.c | 40 +++++++++++++++++++++----
3 files changed, 65 insertions(+), 12 deletions(-)
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
index 211086b..0f305d9 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
@@ -491,6 +491,9 @@ struct link_config {
unsigned char link_ok; /* link up? */
unsigned char link_down_rc; /* link down reason */
+
+ bool new_module; /* ->OS Transceiver Module inserted */
+ bool redo_l1cfg; /* ->CC redo current "sticky" L1 CFG */
};
#define FW_LEN16(fw_struct) FW_CMD_LEN16_V(sizeof(fw_struct) / 16)
@@ -1324,7 +1327,7 @@ static inline unsigned int qtimer_val(const struct adapter *adap,
extern char cxgb4_driver_name[];
extern const char cxgb4_driver_version[];
-void t4_os_portmod_changed(const struct adapter *adap, int port_id);
+void t4_os_portmod_changed(struct adapter *adap, int port_id);
void t4_os_link_changed(struct adapter *adap, int port_id, int link_stat);
void t4_free_sge_resources(struct adapter *adap);
@@ -1505,8 +1508,25 @@ void t4_intr_disable(struct adapter *adapter);
int t4_slow_intr_handler(struct adapter *adapter);
int t4_wait_dev_ready(void __iomem *regs);
-int t4_link_l1cfg(struct adapter *adap, unsigned int mbox, unsigned int port,
- struct link_config *lc);
+
+int t4_link_l1cfg_core(struct adapter *adap, unsigned int mbox,
+ unsigned int port, struct link_config *lc,
+ bool sleep_ok, int timeout);
+
+static inline int t4_link_l1cfg(struct adapter *adapter, unsigned int mbox,
+ unsigned int port, struct link_config *lc)
+{
+ return t4_link_l1cfg_core(adapter, mbox, port, lc,
+ true, FW_CMD_MAX_TIMEOUT);
+}
+
+static inline int t4_link_l1cfg_ns(struct adapter *adapter, unsigned int mbox,
+ unsigned int port, struct link_config *lc)
+{
+ return t4_link_l1cfg_core(adapter, mbox, port, lc,
+ false, FW_CMD_MAX_TIMEOUT);
+}
+
int t4_restart_aneg(struct adapter *adap, unsigned int mbox, unsigned int port);
u32 t4_read_pcie_cfg4(struct adapter *adap, int reg);
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
index 130d1ee..513e1d3 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
@@ -301,14 +301,14 @@ void t4_os_link_changed(struct adapter *adapter, int port_id, int link_stat)
}
}
-void t4_os_portmod_changed(const struct adapter *adap, int port_id)
+void t4_os_portmod_changed(struct adapter *adap, int port_id)
{
static const char *mod_str[] = {
NULL, "LR", "SR", "ER", "passive DA", "active DA", "LRM"
};
- const struct net_device *dev = adap->port[port_id];
- const struct port_info *pi = netdev_priv(dev);
+ struct net_device *dev = adap->port[port_id];
+ struct port_info *pi = netdev_priv(dev);
if (pi->mod_type == FW_PORT_MOD_TYPE_NONE)
netdev_info(dev, "port module unplugged\n");
@@ -325,6 +325,11 @@ void t4_os_portmod_changed(const struct adapter *adap, int port_id)
else
netdev_info(dev, "%s: unknown module type %d inserted\n",
dev->name, pi->mod_type);
+
+ /* If the interface is running, then we'll need any "sticky" Link
+ * Parameters redone with a new Transceiver Module.
+ */
+ pi->link_cfg.redo_l1cfg = netif_running(dev);
}
int dbfifo_int_thresh = 10; /* 10 == 640 entry threshold */
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
index 537ed07..704f696 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
@@ -4058,14 +4058,16 @@ static inline fw_port_cap32_t cc_to_fwcap_fec(enum cc_fec cc_fec)
* - If auto-negotiation is off set the MAC to the proper speed/duplex/FC,
* otherwise do it later based on the outcome of auto-negotiation.
*/
-int t4_link_l1cfg(struct adapter *adapter, unsigned int mbox,
- unsigned int port, struct link_config *lc)
+int t4_link_l1cfg_core(struct adapter *adapter, unsigned int mbox,
+ unsigned int port, struct link_config *lc,
+ bool sleep_ok, int timeout)
{
unsigned int fw_caps = adapter->params.fw_caps_support;
- struct fw_port_cmd cmd;
- unsigned int fw_mdi = FW_PORT_CAP32_MDI_V(FW_PORT_CAP32_MDI_AUTO);
fw_port_cap32_t fw_fc, cc_fec, fw_fec, rcap;
+ struct fw_port_cmd cmd;
+ unsigned int fw_mdi;
+ fw_mdi = (FW_PORT_CAP32_MDI_V(FW_PORT_CAP32_MDI_AUTO) & lc->pcaps);
/* Convert driver coding of Pause Frame Flow Control settings into the
* Firmware's API.
*/
@@ -4087,7 +4089,7 @@ int t4_link_l1cfg(struct adapter *adapter, unsigned int mbox,
/* Figure out what our Requested Port Capabilities are going to be.
*/
if (!(lc->pcaps & FW_PORT_CAP32_ANEG)) {
- rcap = (lc->pcaps & ADVERT_MASK) | fw_fc | fw_fec;
+ rcap = lc->acaps | fw_fc | fw_fec;
lc->fc = lc->requested_fc & ~PAUSE_AUTONEG;
lc->fec = cc_fec;
} else if (lc->autoneg == AUTONEG_DISABLE) {
@@ -4113,7 +4115,8 @@ int t4_link_l1cfg(struct adapter *adapter, unsigned int mbox,
cmd.u.l1cfg.rcap = cpu_to_be32(fwcaps32_to_caps16(rcap));
else
cmd.u.l1cfg32.rcap32 = cpu_to_be32(rcap);
- return t4_wr_mbox(adapter, mbox, &cmd, sizeof(cmd), NULL);
+ return t4_wr_mbox_meat_timeout(adapter, mbox, &cmd, sizeof(cmd), NULL,
+ sleep_ok, timeout);
}
/**
@@ -8335,6 +8338,9 @@ void t4_handle_get_port_info(struct port_info *pi, const __be64 *rpl)
fc = fwcap_to_cc_pause(linkattr);
speed = fwcap_to_speed(linkattr);
+ lc->new_module = false;
+ lc->redo_l1cfg = false;
+
if (mod_type != pi->mod_type) {
/* With the newer SFP28 and QSFP28 Transceiver Module Types,
* various fundamental Port Capabilities which used to be
@@ -8369,6 +8375,8 @@ void t4_handle_get_port_info(struct port_info *pi, const __be64 *rpl)
pi->port_type = port_type;
pi->mod_type = mod_type;
+
+ lc->new_module = t4_is_inserted_mod_type(mod_type);
t4_os_portmod_changed(adapter, pi->port_id);
}
@@ -8401,6 +8409,26 @@ void t4_handle_get_port_info(struct port_info *pi, const __be64 *rpl)
t4_os_link_changed(adapter, pi->port_id, link_ok);
}
+
+ if (lc->new_module && lc->redo_l1cfg) {
+ struct link_config old_lc;
+ int ret;
+
+ /* Save the current L1 Configuration and restore it if an
+ * error occurs. We probably should fix the l1_cfg*()
+ * routines not to change the link_config when an error
+ * occurs ...
+ */
+ old_lc = *lc;
+ ret = t4_link_l1cfg_ns(adapter, adapter->mbox, pi->lport, lc);
+ if (ret) {
+ *lc = old_lc;
+ dev_warn(adapter->pdev_dev,
+ "Attempt to update new Transceiver Module settings failed\n");
+ }
+ }
+ lc->new_module = false;
+ lc->redo_l1cfg = false;
}
/**
--
2.1.0
^ permalink raw reply related
* [PATCH net-next 1/2] cxgb4: change the port capability bits definition
From: Ganesh Goudar @ 2018-05-23 14:32 UTC (permalink / raw)
To: netdev, davem
Cc: nirranjan, indranil, venkatesh, linux-scsi, varun, Ganesh Goudar,
Casey Leedom
MDI Port Capabilities bit definitions were inconsistent with
regard to the MDI enum values. 2 bits used to define MDI in
the port capabilities are not really separable, it's a 2-bit
field with 4 different values. Change the port capability bit
definitions to be "AUTO" and "STRAIGHT" in order to get them
to line up with the enum's.
Signed-off-by: Casey Leedom <leedom@chelsio.com>
Signed-off-by: Ganesh Goudar <ganeshgr@chelsio.com>
---
drivers/net/ethernet/chelsio/cxgb4/t4_hw.c | 4 ++--
drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h | 8 ++++----
drivers/net/ethernet/chelsio/cxgb4vf/t4vf_hw.c | 2 +-
drivers/scsi/csiostor/csio_hw.c | 2 +-
4 files changed, 8 insertions(+), 8 deletions(-)
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
index df5e7c7..537ed07 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
@@ -3941,8 +3941,8 @@ static fw_port_cap32_t fwcaps16_to_caps32(fw_port_cap16_t caps16)
CAP16_TO_CAP32(FC_RX);
CAP16_TO_CAP32(FC_TX);
CAP16_TO_CAP32(ANEG);
- CAP16_TO_CAP32(MDIX);
CAP16_TO_CAP32(MDIAUTO);
+ CAP16_TO_CAP32(MDISTRAIGHT);
CAP16_TO_CAP32(FEC_RS);
CAP16_TO_CAP32(FEC_BASER_RS);
CAP16_TO_CAP32(802_3_PAUSE);
@@ -3982,8 +3982,8 @@ static fw_port_cap16_t fwcaps32_to_caps16(fw_port_cap32_t caps32)
CAP32_TO_CAP16(802_3_PAUSE);
CAP32_TO_CAP16(802_3_ASM_DIR);
CAP32_TO_CAP16(ANEG);
- CAP32_TO_CAP16(MDIX);
CAP32_TO_CAP16(MDIAUTO);
+ CAP32_TO_CAP16(MDISTRAIGHT);
CAP32_TO_CAP16(FEC_RS);
CAP32_TO_CAP16(FEC_BASER_RS);
diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h b/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h
index e6b2e95..2d91480 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4fw_api.h
@@ -2471,8 +2471,8 @@ enum fw_port_cap {
FW_PORT_CAP_FC_RX = 0x0040,
FW_PORT_CAP_FC_TX = 0x0080,
FW_PORT_CAP_ANEG = 0x0100,
- FW_PORT_CAP_MDIX = 0x0200,
- FW_PORT_CAP_MDIAUTO = 0x0400,
+ FW_PORT_CAP_MDIAUTO = 0x0200,
+ FW_PORT_CAP_MDISTRAIGHT = 0x0400,
FW_PORT_CAP_FEC_RS = 0x0800,
FW_PORT_CAP_FEC_BASER_RS = 0x1000,
FW_PORT_CAP_FEC_RESERVED = 0x2000,
@@ -2515,8 +2515,8 @@ enum fw_port_mdi {
#define FW_PORT_CAP32_802_3_PAUSE 0x00040000UL
#define FW_PORT_CAP32_802_3_ASM_DIR 0x00080000UL
#define FW_PORT_CAP32_ANEG 0x00100000UL
-#define FW_PORT_CAP32_MDIX 0x00200000UL
-#define FW_PORT_CAP32_MDIAUTO 0x00400000UL
+#define FW_PORT_CAP32_MDIAUTO 0x00200000UL
+#define FW_PORT_CAP32_MDISTRAIGHT 0x00400000UL
#define FW_PORT_CAP32_FEC_RS 0x00800000UL
#define FW_PORT_CAP32_FEC_BASER_RS 0x01000000UL
#define FW_PORT_CAP32_FEC_RESERVED1 0x02000000UL
diff --git a/drivers/net/ethernet/chelsio/cxgb4vf/t4vf_hw.c b/drivers/net/ethernet/chelsio/cxgb4vf/t4vf_hw.c
index 798695b..3017f78 100644
--- a/drivers/net/ethernet/chelsio/cxgb4vf/t4vf_hw.c
+++ b/drivers/net/ethernet/chelsio/cxgb4vf/t4vf_hw.c
@@ -341,8 +341,8 @@ static fw_port_cap32_t fwcaps16_to_caps32(fw_port_cap16_t caps16)
CAP16_TO_CAP32(FC_RX);
CAP16_TO_CAP32(FC_TX);
CAP16_TO_CAP32(ANEG);
- CAP16_TO_CAP32(MDIX);
CAP16_TO_CAP32(MDIAUTO);
+ CAP16_TO_CAP32(MDISTRAIGHT);
CAP16_TO_CAP32(FEC_RS);
CAP16_TO_CAP32(FEC_BASER_RS);
CAP16_TO_CAP32(802_3_PAUSE);
diff --git a/drivers/scsi/csiostor/csio_hw.c b/drivers/scsi/csiostor/csio_hw.c
index 96bbb82..a10cf25 100644
--- a/drivers/scsi/csiostor/csio_hw.c
+++ b/drivers/scsi/csiostor/csio_hw.c
@@ -1500,8 +1500,8 @@ fw_port_cap32_t fwcaps16_to_caps32(fw_port_cap16_t caps16)
CAP16_TO_CAP32(FC_RX);
CAP16_TO_CAP32(FC_TX);
CAP16_TO_CAP32(ANEG);
- CAP16_TO_CAP32(MDIX);
CAP16_TO_CAP32(MDIAUTO);
+ CAP16_TO_CAP32(MDISTRAIGHT);
CAP16_TO_CAP32(FEC_RS);
CAP16_TO_CAP32(FEC_BASER_RS);
CAP16_TO_CAP32(802_3_PAUSE);
--
2.1.0
^ permalink raw reply related
* Re: [bpf-next V4 PATCH 3/8] xdp: add tracepoint for devmap like cpumap have
From: John Fastabend @ 2018-05-23 14:24 UTC (permalink / raw)
To: Jesper Dangaard Brouer, netdev, Daniel Borkmann,
Alexei Starovoitov
Cc: Christoph Hellwig, BjörnTöpel, Magnus Karlsson,
makita.toshiaki
In-Reply-To: <152665048683.21055.2555532949856555388.stgit@firesoul>
On 05/18/2018 06:34 AM, Jesper Dangaard Brouer wrote:
> Notice how this allow us get XDP statistic without affecting the XDP
> performance, as tracepoint is no-longer activated on a per packet basis.
>
> Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
> ---
[...]
> #include <trace/define_trace.h>
> diff --git a/kernel/bpf/devmap.c b/kernel/bpf/devmap.c
> index cab72c100bb5..6f84100723b0 100644
> --- a/kernel/bpf/devmap.c
> +++ b/kernel/bpf/devmap.c
> @@ -50,6 +50,7 @@
> #include <linux/bpf.h>
> #include <net/xdp.h>
> #include <linux/filter.h>
> +#include <trace/events/xdp.h>
>
> #define DEV_CREATE_FLAG_MASK \
> (BPF_F_NUMA_NODE | BPF_F_RDONLY | BPF_F_WRONLY)
> @@ -57,6 +58,7 @@
> #define DEV_MAP_BULK_SIZE 16
> struct xdp_bulk_queue {
> struct xdp_frame *q[DEV_MAP_BULK_SIZE];
> + struct net_device *dev_rx;
> unsigned int count;
> };
>
> @@ -219,8 +221,8 @@ void __dev_map_insert_ctx(struct bpf_map *map, u32 bit)
> static int bq_xmit_all(struct bpf_dtab_netdev *obj,
> struct xdp_bulk_queue *bq)
> {
> - unsigned int processed = 0, drops = 0;
> struct net_device *dev = obj->dev;
> + int sent = 0, drops = 0;
> int i;
>
> if (unlikely(!bq->count))
> @@ -241,10 +243,13 @@ static int bq_xmit_all(struct bpf_dtab_netdev *obj,
> drops++;
> xdp_return_frame(xdpf);
> }
> - processed++;
> + sent++;
Do 'dropped' frames also get counted as 'sent' frames? This seems a bit
counter-intuitive to me. Should it be 'drops+sent = total frames'
instead?
> }
> bq->count = 0;
>
> + trace_xdp_devmap_xmit(&obj->dtab->map, obj->bit,
> + sent, drops, bq->dev_rx, dev);
> + bq->dev_rx = NULL;
> return 0;
> }
>
> @@ -301,18 +306,28 @@ struct bpf_dtab_netdev *__dev_map_lookup_elem(struct bpf_map *map, u32 key)
> /* Runs under RCU-read-side, plus in softirq under NAPI protection.
> * Thus, safe percpu variable access.
> */
> -static int bq_enqueue(struct bpf_dtab_netdev *obj, struct xdp_frame *xdpf)
> +static int bq_enqueue(struct bpf_dtab_netdev *obj, struct xdp_frame *xdpf,
> + struct net_device *dev_rx)
> +
> {
> struct xdp_bulk_queue *bq = this_cpu_ptr(obj->bulkq);
>
> if (unlikely(bq->count == DEV_MAP_BULK_SIZE))
> bq_xmit_all(obj, bq);
>
> + /* Ingress dev_rx will be the same for all xdp_frame's in
> + * bulk_queue, because bq stored per-CPU and must be flushed
> + * from net_device drivers NAPI func end.
> + */
> + if (!bq->dev_rx)
> + bq->dev_rx = dev_rx;
> +
> bq->q[bq->count++] = xdpf;
> return 0;
> }
>
> -int dev_map_enqueue(struct bpf_dtab_netdev *dst, struct xdp_buff *xdp)
> +int dev_map_enqueue(struct bpf_dtab_netdev *dst, struct xdp_buff *xdp,
> + struct net_device *dev_rx)
> {
> struct net_device *dev = dst->dev;
> struct xdp_frame *xdpf;
> @@ -325,7 +340,7 @@ int dev_map_enqueue(struct bpf_dtab_netdev *dst, struct xdp_buff *xdp)
> if (unlikely(!xdpf))
> return -EOVERFLOW;
>
> - err = bq_enqueue(dst, xdpf);
> + err = bq_enqueue(dst, xdpf, dev_rx);
> if (err)
> return err;
>
> diff --git a/net/core/filter.c b/net/core/filter.c
> index 1447ec94ef74..4a93423cc5ea 100644
> --- a/net/core/filter.c
> +++ b/net/core/filter.c
> @@ -3063,7 +3063,7 @@ static int __bpf_tx_xdp_map(struct net_device *dev_rx, void *fwd,
> case BPF_MAP_TYPE_DEVMAP: {
> struct bpf_dtab_netdev *dst = fwd;
>
> - err = dev_map_enqueue(dst, xdp);
> + err = dev_map_enqueue(dst, xdp, dev_rx);
> if (err)
> return err;
> __dev_map_insert_ctx(map, index);
>
^ permalink raw reply
* Re: [PATCH bpf-next v3 10/10] tools: bpftool: add delimiters to multi-function JITed dumps
From: Sandipan Das @ 2018-05-23 13:59 UTC (permalink / raw)
To: Daniel Borkmann
Cc: Jakub Kicinski, ast, netdev, linuxppc-dev, mpe, naveen.n.rao,
Quentin Monnet
In-Reply-To: <024d9f55-2382-8080-c0c2-e3acfbfb9590@iogearbox.net>
On 05/23/2018 07:20 PM, Daniel Borkmann wrote:
> On 05/23/2018 12:37 PM, Sandipan Das wrote:
> [...]
>> Other than that, for powerpc64, there is a problem with the way the
>> binutils disassembler code (in "opcodes/ppc-dis.c") passes arguments
>> to the callback fprintf_json().
>>
>> In fprintf_json(), we always expect the va_list elements to resolve
>> to strings (char *). But for powerpc64, the register or immediate
>> operands are always passed as integers. So, when the code attempts
>> to resolve these operands using va_arg(ap, char *), bpftool crashes.
>> For now, I am using a workaround based on vsnprintf() but this does
>> not get the semantics correct for memory operands. You can probably
>> see that for the store instructions in the JSON dump above this.
>>
>> Daniel,
>>
>> Would it be okay if I send out a fix for this in a different series?
>
> I'm fine either way with regards to the fix. Feels like a portability bug
> in the binutils disassembler?
>
> We could probably have a feature test like in test-disassembler-four-args
> and select a workaround in bpftool based on that outcome.
>
> Thanks Sandipan!
>
> [1] tools/build/feature/test-disassembler-four-args.c
>
Cool. Thanks for the tip!
- Sandipan
^ permalink raw reply
* Re: [PATCH] ppp: remove the PPPIOCDETACH ioctl
From: Guillaume Nault @ 2018-05-23 13:57 UTC (permalink / raw)
To: Eric Biggers
Cc: linux-ppp, Paul Mackerras, netdev, linux-fsdevel, linux-kernel,
syzkaller-bugs, Eric Biggers
In-Reply-To: <20180523035952.25768-1-ebiggers3@gmail.com>
On Tue, May 22, 2018 at 08:59:52PM -0700, Eric Biggers wrote:
> From: Eric Biggers <ebiggers@google.com>
>
> The PPPIOCDETACH ioctl effectively tries to "close" the given ppp file
> before f_count has reached 0, which is fundamentally a bad idea. It
> does check 'f_count < 2', which excludes concurrent operations on the
> file since they would only be possible with a shared fd table, in which
> case each fdget() would take a file reference. However, it fails to
> account for the fact that even with 'f_count == 1' the file can still be
> linked into epoll instances. As reported by syzbot, this can trivially
> be used to cause a use-after-free.
>
> Yet, the only known user of PPPIOCDETACH is pppd versions older than
> ppp-2.4.2, which was released almost 15 years ago (November 2003).
> Also, PPPIOCDETACH apparently stopped working reliably at around the
> same time, when the f_count check was added to the kernel, e.g. see
> https://lkml.org/lkml/2002/12/31/83. Also, the current 'f_count < 2'
> check makes PPPIOCDETACH only work in single-threaded applications; it
> always fails if called from a multithreaded application.
>
> All pppd versions released in the last 15 years just close() the file
> descriptor instead.
>
> Therefore, instead of hacking around this bug by exporting epoll
> internals to modules, and probably missing other related bugs, just
> remove the PPPIOCDETACH ioctl and see if anyone actually notices.
>
> Reported-by: syzbot+16363c99d4134717c05b@syzkaller.appspotmail.com
> Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
> Cc: stable@vger.kernel.org
> Signed-off-by: Eric Biggers <ebiggers@google.com>
> ---
> Documentation/networking/ppp_generic.txt | 6 -----
> drivers/net/ppp/ppp_generic.c | 29 ------------------------
> fs/compat_ioctl.c | 1 -
> include/uapi/linux/ppp-ioctl.h | 1 -
> 4 files changed, 37 deletions(-)
>
> diff --git a/Documentation/networking/ppp_generic.txt b/Documentation/networking/ppp_generic.txt
> index 091d20273dcb..61daf4b39600 100644
> --- a/Documentation/networking/ppp_generic.txt
> +++ b/Documentation/networking/ppp_generic.txt
> @@ -300,12 +300,6 @@ unattached instance are:
> The ioctl calls available on an instance of /dev/ppp attached to a
> channel are:
>
> -* PPPIOCDETACH detaches the instance from the channel. This ioctl is
> - deprecated since the same effect can be achieved by closing the
> - instance. In order to prevent possible races this ioctl will fail
> - with an EINVAL error if more than one file descriptor refers to this
> - instance (i.e. as a result of dup(), dup2() or fork()).
> -
> * PPPIOCCONNECT connects this channel to a PPP interface. The
> argument should point to an int containing the interface unit
> number. It will return an EINVAL error if the channel is already
> diff --git a/drivers/net/ppp/ppp_generic.c b/drivers/net/ppp/ppp_generic.c
> index dc7c7ec43202..dce8812fe802 100644
> --- a/drivers/net/ppp/ppp_generic.c
> +++ b/drivers/net/ppp/ppp_generic.c
> @@ -603,35 +603,6 @@ static long ppp_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
> goto out;
> }
>
> - if (cmd == PPPIOCDETACH) {
> - /*
> - * We have to be careful here... if the file descriptor
> - * has been dup'd, we could have another process in the
> - * middle of a poll using the same file *, so we had
> - * better not free the interface data structures -
> - * instead we fail the ioctl. Even in this case, we
> - * shut down the interface if we are the owner of it.
> - * Actually, we should get rid of PPPIOCDETACH, userland
> - * (i.e. pppd) could achieve the same effect by closing
> - * this fd and reopening /dev/ppp.
> - */
> - err = -EINVAL;
> - if (pf->kind == INTERFACE) {
> - ppp = PF_TO_PPP(pf);
> - rtnl_lock();
> - if (file == ppp->owner)
> - unregister_netdevice(ppp->dev);
> - rtnl_unlock();
> - }
> - if (atomic_long_read(&file->f_count) < 2) {
> - ppp_release(NULL, file);
> - err = 0;
> - } else
> - pr_warn("PPPIOCDETACH file->f_count=%ld\n",
> - atomic_long_read(&file->f_count));
> - goto out;
> - }
> -
I'd rather add
+ if (cmd == PPPIOCDETACH) {
+ err = -EINVAL;
+ goto out;
+ }
Making PPPIOCDETACH unknown to ppp_generic means that the ioctl would
be handled by the underlying channel when pf->kind == CHANNEL (see the
chan->ops->ioctl() call further down). That shouldn't be a problem per
se, but even though PPPIOCDETACH is unsupported, I feel that it should
remain a ppp_generic thing. I don't really want its value to be reused
for other purposes in the future or have different behaviour depending
on the underlying channel.
Also PPPIOCDETACH can already fail with -EINVAL. Therefore, if ever
there really were programs out there using this call, they'd already
have to handle this case. Unconditionally returning -EINVAL would
further minimise possibilities for breakage.
^ permalink raw reply
* Re: [net-next 1/6] net/dcb: Add dcbnl buffer attribute
From: John Fastabend @ 2018-05-23 13:52 UTC (permalink / raw)
To: Jiri Pirko, Jakub Kicinski
Cc: Saeed Mahameed, David S. Miller, netdev, Huy Nguyen, Or Gerlitz
In-Reply-To: <20180523094331.GC3046@nanopsycho>
On 05/23/2018 02:43 AM, Jiri Pirko wrote:
> Tue, May 22, 2018 at 07:20:26AM CEST, jakub.kicinski@netronome.com wrote:
>> On Mon, 21 May 2018 14:04:57 -0700, Saeed Mahameed wrote:
>>> From: Huy Nguyen <huyn@mellanox.com>
>>>
>>> In this patch, we add dcbnl buffer attribute to allow user
>>> change the NIC's buffer configuration such as priority
>>> to buffer mapping and buffer size of individual buffer.
>>>
>>> This attribute combined with pfc attribute allows advance user to
>>> fine tune the qos setting for specific priority queue. For example,
>>> user can give dedicated buffer for one or more prirorities or user
>>> can give large buffer to certain priorities.
>>>
>>> We present an use case scenario where dcbnl buffer attribute configured
>>> by advance user helps reduce the latency of messages of different sizes.
>>>
>>> Scenarios description:
>>> On ConnectX-5, we run latency sensitive traffic with
>>> small/medium message sizes ranging from 64B to 256KB and bandwidth sensitive
>>> traffic with large messages sizes 512KB and 1MB. We group small, medium,
>>> and large message sizes to their own pfc enables priorities as follow.
>>> Priorities 1 & 2 (64B, 256B and 1KB)
>>> Priorities 3 & 4 (4KB, 8KB, 16KB, 64KB, 128KB and 256KB)
>>> Priorities 5 & 6 (512KB and 1MB)
>>>
>>> By default, ConnectX-5 maps all pfc enabled priorities to a single
>>> lossless fixed buffer size of 50% of total available buffer space. The
>>> other 50% is assigned to lossy buffer. Using dcbnl buffer attribute,
>>> we create three equal size lossless buffers. Each buffer has 25% of total
>>> available buffer space. Thus, the lossy buffer size reduces to 25%. Priority
>>> to lossless buffer mappings are set as follow.
>>> Priorities 1 & 2 on lossless buffer #1
>>> Priorities 3 & 4 on lossless buffer #2
>>> Priorities 5 & 6 on lossless buffer #3
>>>
>>> We observe improvements in latency for small and medium message sizes
>>> as follows. Please note that the large message sizes bandwidth performance is
>>> reduced but the total bandwidth remains the same.
>>> 256B message size (42 % latency reduction)
>>> 4K message size (21% latency reduction)
>>> 64K message size (16% latency reduction)
>>>
>>> Signed-off-by: Huy Nguyen <huyn@mellanox.com>
>>> Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
>>
>> On a cursory look this bares a lot of resemblance to devlink shared
>> buffer configuration ABI. Did you look into using that?
>>
>> Just to be clear devlink shared buffer ABIs don't require representors
>> and "switchdev mode".
>
> If the CX5 buffer they are trying to utilize here is per port and not a
> shared one, it would seem ok for me to not have it in "devlink sb".
>
+1 I think its probably reasonable to let devlink manage the global
(device layer) buffers and then have dcbnl partition the buffer up
further per netdev. Notice there is already a partitioning of the
buffers happening when DCB is enabled and/or parameters are changed.
So giving explicit control over this seems OK to me.
It would be nice though if the API gave us some hint on max/min/stride
of allowed values. Could the get API return these along with current
value? Presumably the allowed max size could change with devlink buffer
changes in how the global buffer is divided up as well.
The argument against allowing this API is it doesn't have anything to
do with the 802.1Q standard, but that is fine IMO.
.John
^ permalink raw reply
* Re: [PATCH bpf-next v3 10/10] tools: bpftool: add delimiters to multi-function JITed dumps
From: Daniel Borkmann @ 2018-05-23 13:50 UTC (permalink / raw)
To: Sandipan Das, Jakub Kicinski
Cc: ast, netdev, linuxppc-dev, mpe, naveen.n.rao, Quentin Monnet
In-Reply-To: <7142939b-515e-50ac-bc0b-50444bf9cc97@linux.vnet.ibm.com>
On 05/23/2018 12:37 PM, Sandipan Das wrote:
[...]
> Other than that, for powerpc64, there is a problem with the way the
> binutils disassembler code (in "opcodes/ppc-dis.c") passes arguments
> to the callback fprintf_json().
>
> In fprintf_json(), we always expect the va_list elements to resolve
> to strings (char *). But for powerpc64, the register or immediate
> operands are always passed as integers. So, when the code attempts
> to resolve these operands using va_arg(ap, char *), bpftool crashes.
> For now, I am using a workaround based on vsnprintf() but this does
> not get the semantics correct for memory operands. You can probably
> see that for the store instructions in the JSON dump above this.
>
> Daniel,
>
> Would it be okay if I send out a fix for this in a different series?
I'm fine either way with regards to the fix. Feels like a portability bug
in the binutils disassembler?
We could probably have a feature test like in test-disassembler-four-args
and select a workaround in bpftool based on that outcome.
Thanks Sandipan!
[1] tools/build/feature/test-disassembler-four-args.c
^ permalink raw reply
* Re: [PATCH net-next] tipc: eliminate complaint of KMSAN uninit-value in tipc_conn_rcv_sub
From: Ying Xue @ 2018-05-23 13:38 UTC (permalink / raw)
To: David Miller; +Cc: netdev, syzkaller-bugs, tipc-discussion, dvyukov
In-Reply-To: <20180519.230021.538446373514892322.davem@davemloft.net>
On 05/20/2018 11:00 AM, David Miller wrote:
> From: Ying Xue <ying.xue@windriver.com>
> Date: Fri, 18 May 2018 19:50:55 +0800
>
>> As variable s of struct tipc_subscr type is not initialized
>> in tipc_conn_rcv_from_sock() before it is used in tipc_conn_rcv_sub(),
>> KMSAN reported the following uninit-value type complaint:
>
> I agree with others that the short read is the bug.
In this error case, Dmitry is right. A short read happened in
tipc_recvmsg() especially when the size of skb received from a socket of
user space was smaller than the msg_iter.count of struct msghdr (ie,
tipc_subscr object size) passed by tipc_conn_rcv_from_sock() in kernel
space.
But when tipc_recvmsg() copied the data of skb to msg_iter.kvec of
struct msghdr with skb data length rather than msg_iter.count, it means
the part of space (ie, msg_iter.count - skb data length) of
msg_iter.kvec was not initialized. For the detailed info, please refer
to its relevant code:
tipc_recvmsg()
{
...
if (likely(!err)) {
copy = min_t(int, dlen, buflen);
if (unlikely(copy != dlen))
m->msg_flags |= MSG_TRUNC;
rc = skb_copy_datagram_msg(skb, hlen, m, copy);
...
}
If we receive a skb message with recvmsg() in user space, it seems no
problem even if the length of msg_iter.iov is bigger than skb data size.
But under the same situation, if we receive a message through
sock_recvmsg() in kernel space, it might be a problem.
I have checked the receive functions of other stacks like TCP and UDP,
as a result, when msg_iter.count is bigger than skb->len, they never use
memset() to initialize the remaining area of msg_iter.kvec (ie,
msg_iter.count - skb->len) no matter whether we receive a message
through sock_recvmsg() in user space or kernel space.
>
> You need to decide what should happen if not a full tipc_subscr object
> is obtained from the sock_recvmsg() call.
>
> Proceeding to pass it on to tipc_conn_rcv_sub() cannot possibly be
> correct.
If we do not conduct a full read for tipc_subscr object through
sock_recvmsg() call, some fields of tipc_subscr object might be
incorrect. But I can confirm that the incorrect fields of tipc_subscr
object any fatal error except that we might wrongly add one subscription.
>
> You're not getting what you are expecting from the peer, the memset()
> you are adding doesn't change that.
Before tipc_subscr object address is passed to msg_iter.kvec pointer, we
have initialized the whole tipc_subscr object with memset(). Just in
this case, this method should kill the uninit-value complaint.
If we initialize tipc_subscr object in tipc_recvmsg(), we have to
conduct initialization behavior for the msg_iter.kvec/msg_iter.vio area
(msg_iter.count - skb->len) no matter whether a message is received from
user space or kernel space. Particularly when the caller of recvmsg() is
in user space, the initialization seems no necessary.
>
> And once you get this badly sized read, what does that do to
> the stream of subsequent recvmsg calls here?
>
Even if we get a bad sized read for tipc_subscr object in
tipc_conn_rcv_sub(), it doesn't cause any fatal impact on system or TIPC
stack.
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
^ permalink raw reply
* Re: [PATCH v2] packet: track ring entry use using a shadow ring to prevent RX ring overrun
From: Willem de Bruijn @ 2018-05-23 13:37 UTC (permalink / raw)
To: Jon Rosen (jrosen)
Cc: David S. Miller, Willem de Bruijn, Eric Dumazet, Kees Cook,
David Windsor, Rosen, Rami, Reshetova, Elena, Mike Maloney,
Benjamin Poirier, Thomas Gleixner, Greg Kroah-Hartman,
open list:NETWORKING [GENERAL], open list
In-Reply-To: <9cbf19b5765746b88af6dadc71b99d78@XCH-RTP-016.cisco.com>
On Wed, May 23, 2018 at 7:54 AM, Jon Rosen (jrosen) <jrosen@cisco.com> wrote:
>> > For the ring, there is no requirement to allocate exactly the amount
>> > specified by the user request. Safer than relying on shared memory
>> > and simpler than the extra allocation in this patch would be to allocate
>> > extra shadow memory at the end of the ring (and not mmap that).
>> >
>> > That still leaves an extra cold cacheline vs using tp_padding.
>>
>> Given my lack of experience and knowledge in writing kernel code
>> it was easier for me to allocate the shadow ring as a separate
>> structure. Of course it's not about me and my skills so if it's
>> more appropriate to allocate at the tail of the existing ring
>> then certainly I can look at doing that.
>
> The memory for the ring is not one contiguous block, it's an array of
> blocks of pages (or 'order' sized blocks of pages). I don't think
> increasing the size of each of the blocks to provided storage would be
> such a good idea as it will risk spilling over into the next order and
> wasting lots of memory. I suspect it's also more complex than a single
> shadow ring to do both the allocation and the access.
>
> It could be tacked onto the end of the pg_vec[] used to store the
> pointers to the blocks. The challenge with that is that a pg_vec[] is
> created for each of RX and TX rings so either it would have to
> allocate unnecessary storage for TX or the caller will have to say if
> extra space should be allocated or not. E.g.:
>
> static struct pgv *alloc_pg_vec(struct tpacket_req *req, int order, int scratch, void **scratch_p)
>
> I'm not sure avoiding the extra allocation and moving it to the
> pg_vec[] for the RX ring is going to get the simplification you were
> hoping for. Is there another way of storing the shadow ring which
> I should consider?
I did indeed mean attaching extra pages to pg_vec[]. It should be
simpler than a separate structure, but I may be wrong.
Either way, I still would prefer to avoid the shadow buffer completely.
It incurs complexity and cycle cost on all users because of only the
rare (non-existent?) consumer that overwrites the padding bytes.
Perhaps we can use padding yet avoid deadlock by writing a
timed value. The simplest would be jiffies >> N. Then only a
process that writes this exact value would be subject to drops and
then still only for a limited period.
Instead of depending on wall clock time, like jiffies, another option
would be to keep a percpu array of values. Each cpu has a zero
entry if it is not writing, nonzero if it is. If a writer encounters a
number in padding that is > num_cpus, then the state is garbage
from userspace. If <= num_cpus, it is adhered to only until that cpu
clears its entry, which is guaranteed to happen eventually.
Just a quick thought. This might not fly at all upon closer scrutiny.
^ permalink raw reply
* Re: KASAN: use-after-free Read in remove_wait_queue (2)
From: Guillaume Nault @ 2018-05-23 13:26 UTC (permalink / raw)
To: Eric Biggers
Cc: linux-ppp, Paul Mackerras, netdev, linux-kernel, syzkaller-bugs,
syzbot, viro
In-Reply-To: <20180523032958.GE658@sol.localdomain>
On Tue, May 22, 2018 at 08:29:58PM -0700, Eric Biggers wrote:
> On Fri, May 18, 2018 at 06:02:23PM +0200, Guillaume Nault wrote:
> > On Sun, May 13, 2018 at 11:11:55PM -0700, Eric Biggers wrote:
> > > [+ppp list and maintainer]
> > >
> > > This is a bug in ppp_generic.c; it still happens on Linus' tree and it's easily
> > > reproducible, see program below. The bug is that the PPPIOCDETACH ioctl doesn't
> > > consider that the file can still be attached to epoll instances even when
> > > ->f_count == 1.
> >
> > Right. What would it take to remove the file for the epoll instances?
> > Sorry for the naive question, but I'm not familiar with VFS and didn't
> > find a helper function we could call.
> >
>
> There is eventpoll_release_file(), but it's not exported to modules. It might
> work to call it, but it seems like a hack.
>
> > > Also, the reproducer doesn't test this but I think ppp_poll(),
> > > ppp_read(), and ppp_write() can all race with PPPIOCDETACH, causing
> > > use-after-frees as well.
> >
> > I also believe so. ppp_release() resets ->private_data, and even though
> > functions like ppp_read() test ->private_data before executing, there's
> > no synchronisation mechanism to ensure that the update is visible
> > before the unit or channel is destroyed. Is that the kind of race you
> > had in mind?
>
> Yes, though after looking into it more I *think* these additional races aren't
> actually possible, due to the 'f_count < 2' check. These races could only
> happen with a shared fd table, but in that case fdget() would increment f_count
> for the duration of each operation, resulting in 'f_count >= 2' if both ioctl()
> and something else is running on the same file concurrently.
>
> Note that this also means PPPIOCDETACH doesn't work at all if called from a
> multithreaded application...
>
> >
> > > Any chance that PPPIOCDETACH can simply be removed,
> > > given that it's apparently been "deprecated" for 16 years?
> > > Does anyone use it?
> >
> > The only users I'm aware of are pppd versions older than ppp-2.4.2
> > (released in November 2003). But even at that time, there were issues
> > with PPPIOCDETACH as pppd didn't seem to react properly when this call
> > failed. An Internet search on the "PPPIOCDETACH file->f_count=" kernel
> > log string, or on the "Couldn't release PPP unit: Invalid argument"
> > error message of pppd, returns several related bug reports.
> >
> > Originally, PPPIOCDETACH never failed, but testing ->f_count was
> > later introduced to fix crashes when the file descriptor had been
> > duplicated. It seems that this was motivated by polling issues too.
> >
> > Long story short, it looks like PPPIOCDETACH never has worked well
> > and we have at least two more bugs to fix. Given how it has proven
> > fragile, I wouldn't be surprised if there were even more lurking
> > around. I'd say that it's probably safer to drop it than to add more
> > workarounds and playing wack-a-mole with those bugs.
>
> IMO, if we can get away with removing it without any users noticing, that would
> be much better than trying to fix it with a VFS-level hack, and probably missing
> some cases. I'll send a patch to get things started...
>
Yes, I fully agree. That looks much safer, and given the track record
of this ioctl I very much doubt anyone would depend on it.
^ 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