* Re: next: arm64: LTP sendto01 test causes system crash in ilp32 mode
From: Yury Norov @ 2017-10-11 19:43 UTC (permalink / raw)
To: Eric Dumazet
Cc: linux-kernel, linux-arm-kernel, netdev, Catalin Marinas,
David S. Miller, Florian Westphal
In-Reply-To: <1507747717.31614.28.camel@edumazet-glaptop3.roam.corp.google.com>
Hi Eric,
On Wed, Oct 11, 2017 at 11:48:37AM -0700, Eric Dumazet wrote:
> On Wed, 2017-10-11 at 11:41 -0700, Eric Dumazet wrote:
> > On Wed, 2017-10-11 at 21:35 +0300, Yury Norov wrote:
> > > Hi all,
> > >
> > > It seems like next-20171009 with ilp32 patches crashes on LTP sendto01 test
> > > in sys_sendto() path, like this:
> >
> > Thanks for the report.
> > Probably caused by one of my recent patches, so I am taking a look.
>
> Yes, this was silly.
>
> Please test this fix :
>
> diff --git a/include/net/tcp.h b/include/net/tcp.h
> index 5a95e5886b55e03e4a8bfeac3506c657a4f97dde..15163454174babdcb465904f725b919268dd1bc7 100644
> --- a/include/net/tcp.h
> +++ b/include/net/tcp.h
> @@ -1712,6 +1712,7 @@ static inline void tcp_insert_write_queue_before(struct sk_buff *new,
>
> static inline void tcp_unlink_write_queue(struct sk_buff *skb, struct sock *sk)
> {
> + tcp_skb_tsorted_anchor_cleanup(skb);
> __skb_unlink(skb, &sk->sk_write_queue);
> }
The fix works for me, thanks.
Tested-by: Yury Norov <ynorov@caviumnetworks.com>
^ permalink raw reply
* Re: BUG:af_packet fails to TX TSO frames
From: Anton Ivanov @ 2017-10-11 19:39 UTC (permalink / raw)
To: Willem de Bruijn; +Cc: Anton Ivanov, Network Development, David Miller
In-Reply-To: <CAF=yD-+MCLAoP5=orUyG0tQjXykde++QrOdE7v44srR54yP32Q@mail.gmail.com>
On 11/10/17 19:57, Willem de Bruijn wrote:
> On Wed, Oct 11, 2017 at 2:39 PM, Anton Ivanov
> <anton.ivanov@kot-begemot.co.uk> wrote:
>> The check as now insists that the actual driver supports GSO_ROBUST, because
>> we have marked the skb dodgy.
>>
>> The specific bit which does this check is in net_gso_ok()
>>
>> Now, lets's see how many Ethernet drivers set GSO_ROBUST.
>>
>> find drivers/net/ethernet -type f -name "*.[c,h]" -exec grep -H GSO_ROBUST
>> {} \;
>>
>> That returns nothing in 4.x
>>
>> IMHO - af_packet allocates the skb, does all checks (and extra may be added)
>> on the gso, why is this set dodgy in the first place?
> It is set when the header has to be validated.
>
> The segmentation logic will validate and fixup gso_segs. See for
> instance tcp_gso_segment:
>
> if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
> /* Packet is from an untrusted source, reset gso_segs. */
>
> skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
>
> segs = NULL;
> goto out;
> }
>
> If the device would have the robust bit set and otherwise supports the
> required features, fix up gso_segs and pass the large packet to the
> device.
>
> Else it continues to the software gso path.
>
> Large packets generated with psock_txring_vnet.c pass this test. I
That test is indeed a different path - this goes via the tpacket_snd
which allocs via sock_alloc_send_skb. That results in a non-fragged skb
as it calls pskb after that with data_len = 0 asking for a contiguous one.
My stuff is using sendmmsg which ends up via packet_snd which allocs
via sock_alloc_send_pskb which is invoked in a way which always creates
2 segments - one for the linear section and one for the rest (and more
if needed). It is faster than tpacket by the way (several times).
As a comparison tap and other virtual drivers use sock_alloc_send_pskb
with non-zero data length which results in multiple frags. The code in
packet_snd is in fact identical with tap (+/- some cosmetic differences).
That is the difference between the tests and that is why your test works
and mine fails.
Now, alloc-ing a 64k contiguous skb every time IMHO is wrong.
So the logic in the xmit check at present works only because it is given
only a very corner case for a GSO frame and tested versus it. It should
work with the generic case which is what comes out of
sock_alloc_send_pskb (same as in tap).
A.
> suspect that there is a subtle difference in the virtio_net_hdr fields
> that that generates vs. your program.
>
^ permalink raw reply
* [PATCH] vxge: Clean up unused variables in vxge-traffic
From: Christos Gkekas @ 2017-10-11 19:26 UTC (permalink / raw)
To: Jon Mason, netdev, linux-kernel; +Cc: Christos Gkekas
Delete unused channel variables in vxge-traffic.
Signed-off-by: Christos Gkekas <chris.gekas@gmail.com>
---
drivers/net/ethernet/neterion/vxge/vxge-traffic.c | 19 -------------------
1 file changed, 19 deletions(-)
diff --git a/drivers/net/ethernet/neterion/vxge/vxge-traffic.c b/drivers/net/ethernet/neterion/vxge/vxge-traffic.c
index 5f630a2..0c3b5de 100644
--- a/drivers/net/ethernet/neterion/vxge/vxge-traffic.c
+++ b/drivers/net/ethernet/neterion/vxge/vxge-traffic.c
@@ -1209,9 +1209,6 @@ void vxge_hw_ring_rxd_pre_post(struct __vxge_hw_ring *ring, void *rxdh)
void vxge_hw_ring_rxd_post_post(struct __vxge_hw_ring *ring, void *rxdh)
{
struct vxge_hw_ring_rxd_1 *rxdp = (struct vxge_hw_ring_rxd_1 *)rxdh;
- struct __vxge_hw_channel *channel;
-
- channel = &ring->channel;
rxdp->control_0 = VXGE_HW_RING_RXD_LIST_OWN_ADAPTER;
@@ -1359,11 +1356,8 @@ enum vxge_hw_status vxge_hw_ring_rxd_next_completed(
enum vxge_hw_status vxge_hw_ring_handle_tcode(
struct __vxge_hw_ring *ring, void *rxdh, u8 t_code)
{
- struct __vxge_hw_channel *channel;
enum vxge_hw_status status = VXGE_HW_OK;
- channel = &ring->channel;
-
/* If the t_code is not supported and if the
* t_code is other than 0x5 (unparseable packet
* such as unknown UPV6 header), Drop it !!!
@@ -1399,10 +1393,6 @@ enum vxge_hw_status vxge_hw_ring_handle_tcode(
static void __vxge_hw_non_offload_db_post(struct __vxge_hw_fifo *fifo,
u64 txdl_ptr, u32 num_txds, u32 no_snoop)
{
- struct __vxge_hw_channel *channel;
-
- channel = &fifo->channel;
-
writeq(VXGE_HW_NODBW_TYPE(VXGE_HW_NODBW_TYPE_NODBW) |
VXGE_HW_NODBW_LAST_TXD_NUMBER(num_txds) |
VXGE_HW_NODBW_GET_NO_SNOOP(no_snoop),
@@ -1506,9 +1496,6 @@ void vxge_hw_fifo_txdl_buffer_set(struct __vxge_hw_fifo *fifo,
{
struct __vxge_hw_fifo_txdl_priv *txdl_priv;
struct vxge_hw_fifo_txd *txdp, *txdp_last;
- struct __vxge_hw_channel *channel;
-
- channel = &fifo->channel;
txdl_priv = __vxge_hw_fifo_txdl_priv(fifo, txdlh);
txdp = (struct vxge_hw_fifo_txd *)txdlh + txdl_priv->frags;
@@ -1554,9 +1541,6 @@ void vxge_hw_fifo_txdl_post(struct __vxge_hw_fifo *fifo, void *txdlh)
struct __vxge_hw_fifo_txdl_priv *txdl_priv;
struct vxge_hw_fifo_txd *txdp_last;
struct vxge_hw_fifo_txd *txdp_first;
- struct __vxge_hw_channel *channel;
-
- channel = &fifo->channel;
txdl_priv = __vxge_hw_fifo_txdl_priv(fifo, txdlh);
txdp_first = txdlh;
@@ -1672,10 +1656,7 @@ enum vxge_hw_status vxge_hw_fifo_handle_tcode(struct __vxge_hw_fifo *fifo,
void *txdlh,
enum vxge_hw_fifo_tcode t_code)
{
- struct __vxge_hw_channel *channel;
-
enum vxge_hw_status status = VXGE_HW_OK;
- channel = &fifo->channel;
if (((t_code & 0x7) < 0) || ((t_code & 0x7) > 0x4)) {
status = VXGE_HW_ERR_INVALID_TCODE;
--
2.7.4
^ permalink raw reply related
* Re: BUG:af_packet fails to TX TSO frames
From: Anton Ivanov @ 2017-10-11 18:39 UTC (permalink / raw)
To: Willem de Bruijn, Anton Ivanov; +Cc: Network Development, David Miller
In-Reply-To: <CAF=yD-+JaS8JwEC2TyUG=Ebaho-MM+gjumrbHYM-H0EK1Hu62Q@mail.gmail.com>
The check as now insists that the actual driver supports GSO_ROBUST,
because we have marked the skb dodgy.
The specific bit which does this check is in net_gso_ok()
Now, lets's see how many Ethernet drivers set GSO_ROBUST.
find drivers/net/ethernet -type f -name "*.[c,h]" -exec grep -H
GSO_ROBUST {} \;
That returns nothing in 4.x
IMHO - af_packet allocates the skb, does all checks (and extra may be
added) on the gso, why is this set dodgy in the first place?
A.
On 11/10/17 17:26, Willem de Bruijn wrote:
> On Wed, Oct 11, 2017 at 11:54 AM, Anton Ivanov
> <anton.ivanov@cambridgegreys.com> wrote:
>> It is that patch.
>>
>> I rolled it back and immediately got it to work correctly on a Broadcom
>> Tigon. I can test on all other scenarios, I have tried, I suspect it will
>> come back alive on all of them.
>>
>> I am going to try to trace it through and see exactly where it drops a skb
>> which the card has no issues in accepting.
> It might be in the initialization of gso_type and csum. The virtio_net_hdr
> can encode various combinations of flags that are not allowed by the
> validation logic.
>
^ permalink raw reply
* Re: [PATCH] net: ftgmac100: Request clock and set speed
From: David Miller @ 2017-10-11 19:08 UTC (permalink / raw)
To: joel; +Cc: benh, netdev, linux-kernel, andrew
In-Reply-To: <20171010044925.21078-1-joel@jms.id.au>
From: Joel Stanley <joel@jms.id.au>
Date: Tue, 10 Oct 2017 15:19:25 +1030
> According to the ASPEED datasheet, gigabit speeds require a clock of
> 100MHz or higher. Other speeds require 25MHz or higher.
>
> Signed-off-by: Joel Stanley <joel@jms.id.au>
Hey Joel, it seems that Benjamin would like you to guard this new
logic with whether we have an aspeed configuration or not.
Thank you.
^ permalink raw reply
* act_skbedit unable to check queue_mapping at init time
From: Florian Fainelli @ 2017-10-11 19:02 UTC (permalink / raw)
To: netdev, jhs, jiri, alexander.h.duyck, xiyou.wangcong
Hi all,
It is perfectly possible for the following to happen:
# ls /sys/class/net/gphy/queues
rx-0 tx-0 tx-1 tx-2 tx-3
# tc qdisc add dev gphy root handle 1: multiq
# tc filter add dev gphy parent 1: protocol ip prio 1 u32
match ip dst 192.168.1.1 action skbedit queue_mapping 5
that specifies a queue_mapping value that exceeds what the underlying
device supports. There is a check for dev->real_num_tx_queue >
d->queue_mapping in tcf_skbedit() which makes sure we won't overflow the
queue space, but this will essentially render the action inactive.
I can't find a way to get a reference on a network device from a
tc_action_ops->init function in order to validate that the queue_mapping
is valid at ->init() time and return an appropriate error if we are
exceeding what the device does. This is an action chained to a match and
a filter, so somehow one would think it would be possible to obtain a
reference on the network device the filter was installed on...
Any clues how we could fix that?
--
Florian
^ permalink raw reply
* Re: BUG:af_packet fails to TX TSO frames
From: Willem de Bruijn @ 2017-10-11 18:57 UTC (permalink / raw)
To: Anton Ivanov; +Cc: Anton Ivanov, Network Development, David Miller
In-Reply-To: <23ace6d6-afa7-9a3a-aa61-1245ee6c0498@kot-begemot.co.uk>
On Wed, Oct 11, 2017 at 2:39 PM, Anton Ivanov
<anton.ivanov@kot-begemot.co.uk> wrote:
> The check as now insists that the actual driver supports GSO_ROBUST, because
> we have marked the skb dodgy.
>
> The specific bit which does this check is in net_gso_ok()
>
> Now, lets's see how many Ethernet drivers set GSO_ROBUST.
>
> find drivers/net/ethernet -type f -name "*.[c,h]" -exec grep -H GSO_ROBUST
> {} \;
>
> That returns nothing in 4.x
>
> IMHO - af_packet allocates the skb, does all checks (and extra may be added)
> on the gso, why is this set dodgy in the first place?
It is set when the header has to be validated.
The segmentation logic will validate and fixup gso_segs. See for
instance tcp_gso_segment:
if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
/* Packet is from an untrusted source, reset gso_segs. */
skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
segs = NULL;
goto out;
}
If the device would have the robust bit set and otherwise supports the
required features, fix up gso_segs and pass the large packet to the
device.
Else it continues to the software gso path.
Large packets generated with psock_txring_vnet.c pass this test. I
suspect that there is a subtle difference in the virtio_net_hdr fields
that that generates vs. your program.
^ permalink raw reply
* Re: next: arm64: LTP sendto01 test causes system crash in ilp32 mode
From: Eric Dumazet @ 2017-10-11 18:48 UTC (permalink / raw)
To: Yury Norov
Cc: linux-kernel, linux-arm-kernel, netdev, Catalin Marinas,
David S. Miller, Florian Westphal
In-Reply-To: <1507747313.31614.27.camel@edumazet-glaptop3.roam.corp.google.com>
On Wed, 2017-10-11 at 11:41 -0700, Eric Dumazet wrote:
> On Wed, 2017-10-11 at 21:35 +0300, Yury Norov wrote:
> > Hi all,
> >
> > It seems like next-20171009 with ilp32 patches crashes on LTP sendto01 test
> > in sys_sendto() path, like this:
>
> Thanks for the report.
> Probably caused by one of my recent patches, so I am taking a look.
Yes, this was silly.
Please test this fix :
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 5a95e5886b55e03e4a8bfeac3506c657a4f97dde..15163454174babdcb465904f725b919268dd1bc7 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -1712,6 +1712,7 @@ static inline void tcp_insert_write_queue_before(struct sk_buff *new,
static inline void tcp_unlink_write_queue(struct sk_buff *skb, struct sock *sk)
{
+ tcp_skb_tsorted_anchor_cleanup(skb);
__skb_unlink(skb, &sk->sk_write_queue);
}
^ permalink raw reply related
* Re: next: arm64: LTP sendto01 test causes system crash in ilp32 mode
From: Eric Dumazet @ 2017-10-11 18:41 UTC (permalink / raw)
To: Yury Norov
Cc: linux-kernel, linux-arm-kernel, netdev, Catalin Marinas,
David S. Miller, Florian Westphal
In-Reply-To: <20171011183522.tm2xxgzbdspjtsax@yury-thinkpad>
On Wed, 2017-10-11 at 21:35 +0300, Yury Norov wrote:
> Hi all,
>
> It seems like next-20171009 with ilp32 patches crashes on LTP sendto01 test
> in sys_sendto() path, like this:
Thanks for the report.
Probably caused by one of my recent patches, so I am taking a look.
^ permalink raw reply
* Re: [Patch net-next] tcp: add a tracepoint for tcp_retransmit_skb()
From: Eric Dumazet @ 2017-10-11 18:38 UTC (permalink / raw)
To: Alexei Starovoitov
Cc: Hannes Frederic Sowa, Cong Wang, netdev, Eric Dumazet,
Yuchung Cheng, Neal Cardwell, Martin KaFai Lau, Brendan Gregg,
Song Liu
In-Reply-To: <20171011182113.rdxgrebygmiijpqs@ast-mbp>
On Wed, 2017-10-11 at 11:21 -0700, Alexei Starovoitov wrote:
> On Wed, Oct 11, 2017 at 06:31:45AM -0700, Eric Dumazet wrote:
> > On Tue, 2017-10-10 at 19:56 -0700, Alexei Starovoitov wrote:
> >
> > > actually we hit that too for completely different tracing use case.
> > > Indeed would be good to generate socket cookie unconditionally
> > > for all sockets. I don't think there is any harm.
> > >
> >
> > Simply call sock_gen_cookie() when needed.
> >
> > If a tracepoint needs the cookie and the cookie was not yet generated,
> > it will be generated at this point.
>
> we already have bpf_get_socket_cookie() that will call it,
> but this helper is for bpf socket filters, clsact and other
> networking related prog types, whereas all of tracing is
> read-only and side-effect-free, so we cannot use
> bpf_get_socket_cookie() there.
> Hence for tracing in kprobe-bpf we use raw sk pointer
> as map key and full tuple when passing the socket info to user
> space. If we could use socket cookie vs full tuple it would
> make a nice difference.
Since this sock_gen_cookie() is lock-free and IRQ ready, it should be
not be a problem to pretend it works with a const socket.
I am a bit unsure about revealing in socket cookie a precise count of
sockets created on a netns. Some attackers might use this in a side
channel attack.
^ permalink raw reply
* next: arm64: LTP sendto01 test causes system crash in ilp32 mode
From: Yury Norov @ 2017-10-11 18:35 UTC (permalink / raw)
To: linux-kernel, linux-arm-kernel, netdev
Cc: Catalin Marinas, David S. Miller, Florian Westphal
[-- Attachment #1: Type: text/plain, Size: 5837 bytes --]
Hi all,
It seems like next-20171009 with ilp32 patches crashes on LTP sendto01 test
in sys_sendto() path, like this:
[ 554.034021] [<ffff80003ccd5a58>] 0xffff80003ccd5a58
[ 554.034156] [<ffff00000888fd34>] skb_release_all+0x14/0x30
[ 554.034288] [<ffff00000888fd64>] __kfree_skb+0x14/0x28
[ 554.034409] [<ffff0000088ece6c>] tcp_sendmsg_locked+0x4dc/0xcc8
[ 554.034541] [<ffff0000088ed68c>] tcp_sendmsg+0x34/0x58
[ 554.034659] [<ffff000008919fd4>] inet_sendmsg+0x2c/0xf8
[ 554.034783] [<ffff0000088842e8>] sock_sendmsg+0x18/0x30
[ 554.034928] [<ffff0000088861fc>] SyS_sendto+0x84/0xf8
I cannot reproduce it in lp64 mode, and test is passed in ilp32 mode
if I run it alone, even in infinite loop. But in ltplite scenario the
fail is always reproducible.
The brief analisys of dump shows that kernel crashes due to bad value
in ->destructor field of struct sk_buff, when tries to call
skb->destructor() in skb_release_all(). It looks very unusual,
comparing to typical ilp32 ABI bugs, and I suspect that here is generic
issue - maybe some race condition?
Kernel v4.14-rc4 works well. If no ideas, I'll bisect it a bit later.
Ooops log is below. Config is attached, and kernel sources are:
https://github.com/norov/linux/tree/ilp32-20171009
Yury
[ 554.026522] Unable to handle kernel read from unreadable memory at virtual address ffff80003ccd5a58
[ 554.027005] Mem abort info:
[ 554.027124] Exception class = IABT (current EL), IL = 32 bits
[ 554.027292] SET = 0, FnV = 0
[ 554.027378] EA = 0, S1PTW = 0
[ 554.027537] swapper pgtable: 4k pages, 48-bit VAs, pgd = ffff000009069000
[ 554.027732] [ffff80003ccd5a58] *pgd=000000007eff7003, *pud=000000007eff6003, *pmd=00f800007cc00711
[ 554.028128] Internal error: Oops: 8600000e [#1] PREEMPT SMP
[ 554.028308] Modules linked in:
[ 554.028480] CPU: 1 PID: 6388 Comm: send01 Not tainted 4.14.0-rc4-next-20171009-00025-g6229c950955a #256
[ 554.028684] Hardware name: linux,dummy-virt (DT)
[ 554.028797] task: ffff80003b6d0e80 task.stack: ffff000009d70000
[ 554.028959] PC is at 0xffff80003ccd5a58
[ 554.029272] LR is at skb_release_head_state+0x5c/0xf8
[ 554.029406] pc : [<ffff80003ccd5a58>] lr : [<ffff00000888fc84>] pstate: 40000145
[ 554.029676] sp : ffff000009d73c00
[ 554.029806] x29: ffff000009d73c00 x28: ffff800039a86c80
[ 554.030021] x27: ffff800039a86dd8 x26: 00000000fffffff2
[ 554.030139] x25: ffff80003ccd5a00 x24: 0000000000000000
[ 554.030258] x23: ffff000009d73de8 x22: 0000000000000000
[ 554.030375] x21: ffff000009d73df8 x20: 0000000000000000
[ 554.030490] x19: ffff80003ccd5a00 x18: 00000000f7e73df8
[ 554.030606] x17: 00000000f7f40320 x16: ffff000008886178
[ 554.030721] x15: 0000000000000126 x14: 00000000f7fea700
[ 554.030840] x13: 00000000f7e75b8c x12: 00000000f7e7e43c
[ 554.030959] x11: 6f732064696c6176 x10: 0101010101010101
[ 554.031060] x9 : 206d305b1b535341 x8 : 0000000000005555
[ 554.031159] x7 : ffff80003b6d0e80 x6 : ffff80003c0aa910
[ 554.031256] x5 : ffff80003c0aad10 x4 : 0000000000000000
[ 554.031354] x3 : 000000010000f809 x2 : 0000000000000700
[ 554.031452] x1 : ffff80003ccd5a58 x0 : ffff80003ccd5a00
[ 554.031566] Process send01 (pid: 6388, stack limit = 0xffff000009d70000)
[ 554.031753] Call trace:
[ 554.031870] Exception stack(0xffff000009d73ac0 to 0xffff000009d73c00)
[ 554.032064] 3ac0: ffff80003ccd5a00 ffff80003ccd5a58 0000000000000700 000000010000f809
[ 554.032224] 3ae0: 0000000000000000 ffff80003c0aad10 ffff80003c0aa910 ffff80003b6d0e80
[ 554.032380] 3b00: 0000000000005555 206d305b1b535341 0101010101010101 6f732064696c6176
[ 554.032584] 3b20: 00000000f7e7e43c 00000000f7e75b8c 00000000f7fea700 0000000000000126
[ 554.032732] 3b40: ffff000008886178 00000000f7f40320 00000000f7e73df8 ffff80003ccd5a00
[ 554.032883] 3b60: 0000000000000000 ffff000009d73df8 0000000000000000 ffff000009d73de8
[ 554.033066] 3b80: 0000000000000000 ffff80003ccd5a00 00000000fffffff2 ffff800039a86dd8
[ 554.033233] 3ba0: ffff800039a86c80 ffff000009d73c00 ffff00000888fc84 ffff000009d73c00
[ 554.033386] 3bc0: ffff80003ccd5a58 0000000040000145 ffff0000089a2a64 0000000000000145
[ 554.033656] 3be0: 0001000000000000 ffff00000888fd08 ffff000009d73c00 ffff80003ccd5a58
[ 554.034021] [<ffff80003ccd5a58>] 0xffff80003ccd5a58
[ 554.034156] [<ffff00000888fd34>] skb_release_all+0x14/0x30
[ 554.034288] [<ffff00000888fd64>] __kfree_skb+0x14/0x28
[ 554.034409] [<ffff0000088ece6c>] tcp_sendmsg_locked+0x4dc/0xcc8
[ 554.034541] [<ffff0000088ed68c>] tcp_sendmsg+0x34/0x58
[ 554.034659] [<ffff000008919fd4>] inet_sendmsg+0x2c/0xf8
[ 554.034783] [<ffff0000088842e8>] sock_sendmsg+0x18/0x30
[ 554.034928] [<ffff0000088861fc>] SyS_sendto+0x84/0xf8
[ 554.035046] Exception stack(0xffff000009d73ec0 to 0xffff000009d74000)
[ 554.035186] 3ec0: 0000000000000004 00000000ffffffff 0000000000000400 0000000000000000
[ 554.035334] 3ee0: 0000000000000000 0000000000000000 20203130646e6573 1b20203220202020
[ 554.035503] 3f00: 00000000000000ce 206d305b1b535341 0101010101010101 6f732064696c6176
[ 554.035657] 3f20: 00000000f7e7e43c 00000000f7e75b8c 00000000f7fea700 0000000000000126
[ 554.035825] 3f40: 00000000004240e0 00000000f7f40320 00000000f7e73df8 000000000040e000
[ 554.035981] 3f60: 00000000f7feaea0 0000000000424000 0000000000424000 0000000000447000
[ 554.036148] 3f80: 0000000000447000 000000000040e000 000000000000002c 000000000040ee28
[ 554.036315] 3fa0: 0000000000447450 00000000fffef5b0 0000000000402748 00000000fffef5b0
[ 554.036520] 3fc0: 00000000f7f40348 0000000000000000 0000000000000004 00000000000000ce
[ 554.036683] 3fe0: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[ 554.036853] [<ffff0000080837dc>] el0_svc_naked+0x20/0x24
[ 554.037052] Code: 00000000 00000000 00000000 00000000 (00000000)
[ 554.037369] ---[ end trace c38823b11ae81586 ]---
[-- Attachment #2: config.gz --]
[-- Type: application/gzip, Size: 36439 bytes --]
^ permalink raw reply
* Re: [Patch net-next] tcp: add a tracepoint for tcp_retransmit_skb()
From: Alexei Starovoitov @ 2017-10-11 18:21 UTC (permalink / raw)
To: Eric Dumazet
Cc: Hannes Frederic Sowa, Cong Wang, netdev, Eric Dumazet,
Yuchung Cheng, Neal Cardwell, Martin KaFai Lau, Brendan Gregg,
Song Liu
In-Reply-To: <1507728705.31614.24.camel@edumazet-glaptop3.roam.corp.google.com>
On Wed, Oct 11, 2017 at 06:31:45AM -0700, Eric Dumazet wrote:
> On Tue, 2017-10-10 at 19:56 -0700, Alexei Starovoitov wrote:
>
> > actually we hit that too for completely different tracing use case.
> > Indeed would be good to generate socket cookie unconditionally
> > for all sockets. I don't think there is any harm.
> >
>
> Simply call sock_gen_cookie() when needed.
>
> If a tracepoint needs the cookie and the cookie was not yet generated,
> it will be generated at this point.
we already have bpf_get_socket_cookie() that will call it,
but this helper is for bpf socket filters, clsact and other
networking related prog types, whereas all of tracing is
read-only and side-effect-free, so we cannot use
bpf_get_socket_cookie() there.
Hence for tracing in kprobe-bpf we use raw sk pointer
as map key and full tuple when passing the socket info to user
space. If we could use socket cookie vs full tuple it would
make a nice difference.
^ permalink raw reply
* Re: [PATCH][bpf-next] bpf: remove redundant variable old_flags
From: Alexei Starovoitov @ 2017-10-11 18:15 UTC (permalink / raw)
To: Colin King
Cc: Alexei Starovoitov, Daniel Borkmann, netdev, kernel-janitors,
linux-kernel
In-Reply-To: <20171011105623.19998-1-colin.king@canonical.com>
On Wed, Oct 11, 2017 at 11:56:23AM +0100, Colin King wrote:
> From: Colin Ian King <colin.king@canonical.com>
>
> Variable old_flags is being assigned but is never read; it is redundant
> and can be removed.
>
> Cleans up clang warning: Value stored to 'old_flags' is never read
>
> Signed-off-by: Colin Ian King <colin.king@canonical.com>
> ---
> kernel/bpf/cgroup.c | 2 --
> 1 file changed, 2 deletions(-)
>
> diff --git a/kernel/bpf/cgroup.c b/kernel/bpf/cgroup.c
> index e88abc0865d5..3db5a17fcfe8 100644
> --- a/kernel/bpf/cgroup.c
> +++ b/kernel/bpf/cgroup.c
> @@ -192,7 +192,6 @@ int __cgroup_bpf_attach(struct cgroup *cgrp, struct bpf_prog *prog,
> struct cgroup_subsys_state *css;
> struct bpf_prog_list *pl;
> bool pl_was_allocated;
> - u32 old_flags;
> int err;
>
> if ((flags & BPF_F_ALLOW_OVERRIDE) && (flags & BPF_F_ALLOW_MULTI))
> @@ -239,7 +238,6 @@ int __cgroup_bpf_attach(struct cgroup *cgrp, struct bpf_prog *prog,
> pl->prog = prog;
> }
>
> - old_flags = cgrp->bpf.flags[type];
Acked-by: Alexei Starovoitov <ast@kernel.org>
thanks for the cleanup. That was a leftover of one of the previous
experiments I did.
The patch is for net-next.
^ permalink raw reply
* [PATCH net-next] sched: tc_mirred: Remove whitespaces
From: Florian Fainelli @ 2017-10-11 18:14 UTC (permalink / raw)
To: netdev; +Cc: Florian Fainelli, Jamal Hadi Salim, Cong Wang, Jiri Pirko,
open list
This file contains unnecessary whitespaces as newlines, remove them,
found by looking at what struct tc_mirred looks like.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
include/uapi/linux/tc_act/tc_mirred.h | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/include/uapi/linux/tc_act/tc_mirred.h b/include/uapi/linux/tc_act/tc_mirred.h
index 3d7a2b352a62..69038c29e8a9 100644
--- a/include/uapi/linux/tc_act/tc_mirred.h
+++ b/include/uapi/linux/tc_act/tc_mirred.h
@@ -9,13 +9,13 @@
#define TCA_EGRESS_MIRROR 2 /* mirror packet to EGRESS */
#define TCA_INGRESS_REDIR 3 /* packet redirect to INGRESS*/
#define TCA_INGRESS_MIRROR 4 /* mirror packet to INGRESS */
-
+
struct tc_mirred {
tc_gen;
int eaction; /* one of IN/EGRESS_MIRROR/REDIR */
__u32 ifindex; /* ifindex of egress port */
};
-
+
enum {
TCA_MIRRED_UNSPEC,
TCA_MIRRED_TM,
@@ -24,5 +24,5 @@ enum {
__TCA_MIRRED_MAX
};
#define TCA_MIRRED_MAX (__TCA_MIRRED_MAX - 1)
-
+
#endif
--
2.9.3
^ permalink raw reply related
* Re: [PATCH iproute2 2/2] ss: print MD5 signature keys configured on TCP sockets
From: Stephen Hemminger @ 2017-10-11 18:06 UTC (permalink / raw)
To: Ivan Delalande; +Cc: netdev
In-Reply-To: <20171006234820.27567-2-colona@arista.com>
On Fri, 6 Oct 2017 16:48:20 -0700
Ivan Delalande <colona@arista.com> wrote:
> These keys are reported by kernel 4.14 and later under the
> INET_DIAG_MD5SIG attribute, when INET_DIAG_INFO is requested (ss -i)
> and we have CAP_NET_ADMIN. The additional output looks like:
>
> md5keys:fe80::/64=signing_key,10.1.2.0/24=foobar,::1/128=Test
>
> Signed-off-by: Ivan Delalande <colona@arista.com>
Sure makes sense applied.
^ permalink raw reply
* Re: [PATCH v3] lib: fix multiple strlcpy definition
From: Stephen Hemminger @ 2017-10-11 18:03 UTC (permalink / raw)
To: Baruch Siach; +Cc: netdev, Phil Sutter
In-Reply-To: <9a00dc2dca0650efd7e169db9bb15ae1ec043c08.1507528184.git.baruch@tkos.co.il>
On Mon, 9 Oct 2017 08:49:44 +0300
Baruch Siach <baruch@tkos.co.il> wrote:
> Some C libraries, like uClibc and musl, provide BSD compatible
> strlcpy(). Add check_strlcpy() to configure, and avoid defining strlcpy
> and strlcat when the C library provides them.
>
> This fixes the following static link error with uClibc-ng:
>
> .../sysroot/usr/lib/libc.a(strlcpy.os): In function `strlcpy':
> strlcpy.c:(.text+0x0): multiple definition of `strlcpy'
> ../lib/libutil.a(utils.o):utils.c:(.text+0x1ddc): first defined here
> collect2: error: ld returned 1 exit status
>
> Acked-by: Phil Sutter <phil@nwl.cc>
> Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Thanks for fixing. Most people never use other versions of libc
so things get broken rather often.
^ permalink raw reply
* Re: [PATCH] tests: Remove bashisms (s/source/.)
From: Randy Dunlap @ 2017-10-11 18:02 UTC (permalink / raw)
To: Stephen Hemminger, Petr Vorel; +Cc: netdev
In-Reply-To: <20171011110115.13739488@xeon-e3>
On 10/11/17 11:01, Stephen Hemminger wrote:
> On Sun, 8 Oct 2017 16:39:16 +0200
> Petr Vorel <petr.vorel@gmail.com> wrote:
>
>> Signed-off-by: Petr Vorel <petr.vorel@gmail.com>
>
> Ok, applied. But iproute2 is really limited to Linux and bash is lingua franca on Linux
>
no French, please. Some distros use dash, but being POSIX is usually good.
--
~Randy
^ permalink raw reply
* Re: [PATCH] tests: Remove bashisms (s/source/.)
From: Stephen Hemminger @ 2017-10-11 18:01 UTC (permalink / raw)
To: Petr Vorel; +Cc: netdev
In-Reply-To: <20171008143916.21813-1-petr.vorel@gmail.com>
On Sun, 8 Oct 2017 16:39:16 +0200
Petr Vorel <petr.vorel@gmail.com> wrote:
> Signed-off-by: Petr Vorel <petr.vorel@gmail.com>
Ok, applied. But iproute2 is really limited to Linux and bash is lingua franca on Linux
^ permalink raw reply
* [PATCH net-next 5/5] net: systemport: Turn on ACB at the SYSTEMPORT level
From: Florian Fainelli @ 2017-10-11 17:57 UTC (permalink / raw)
To: netdev; +Cc: davem, andrew, vivien.didelot, Florian Fainelli
In-Reply-To: <20171011175752.22030-1-f.fainelli@gmail.com>
Now that we have established the queue mapping between the switch port
egress queues and the SYSTEMPORT egress queues, we can turn on Advanced
Congestion Buffering (ACB) at the SYSTEMPORT level. This enables the
Ethernet MAC controller to get out of band flow control information
directly from the switch port and queue that it monitors such that its
internal TDMA can be appropriately backpressured.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
drivers/net/ethernet/broadcom/bcmsysport.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/broadcom/bcmsysport.c b/drivers/net/ethernet/broadcom/bcmsysport.c
index 78bed9a84e81..dafc26690555 100644
--- a/drivers/net/ethernet/broadcom/bcmsysport.c
+++ b/drivers/net/ethernet/broadcom/bcmsysport.c
@@ -1422,10 +1422,14 @@ static int bcm_sysport_init_tx_ring(struct bcm_sysport_priv *priv,
reg &= ~(RING_QID_MASK | RING_PORT_ID_MASK << RING_PORT_ID_SHIFT);
reg |= ring->switch_queue & RING_QID_MASK;
reg |= ring->switch_port << RING_PORT_ID_SHIFT;
- reg |= RING_IGNORE_STATUS;
tdma_writel(priv, reg, TDMA_DESC_RING_MAPPING(index));
tdma_writel(priv, 0, TDMA_DESC_RING_PCP_DEI_VID(index));
+ /* Enable ACB algorithm 2 */
+ reg = tdma_readl(priv, TDMA_CONTROL);
+ reg |= tdma_control_bit(priv, ACB_ALGO);
+ tdma_writel(priv, reg, TDMA_CONTROL);
+
/* Do not use tdma_control_bit() here because TSB_SWAP1 collides
* with the original definition of ACB_ALGO
*/
--
2.9.3
^ permalink raw reply related
* [PATCH net-next 4/5] net: dsa: bcm_sf2: Turn on ACB at the switch level
From: Florian Fainelli @ 2017-10-11 17:57 UTC (permalink / raw)
To: netdev; +Cc: davem, andrew, vivien.didelot, Florian Fainelli
In-Reply-To: <20171011175752.22030-1-f.fainelli@gmail.com>
Turn on the out of band Advanced Congestion Buffering (ACB) mechanism at
the switch level now that we have properly established the queue mapping
between the switch egress queues and the SYSTEMPORT egress queues. This
allows the switch to correctly backpressure the host system when one of
its queue drops below the configured thresholds.
This is also helping achieve so called "lossless" behavior by adapting
the TX interrupt pacing to the actual speed and capacity of the switch
port.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
drivers/net/dsa/bcm_sf2.c | 30 ++++++++++++++++++++++++++++++
drivers/net/dsa/bcm_sf2_regs.h | 23 +++++++++++++++++++++++
2 files changed, 53 insertions(+)
diff --git a/drivers/net/dsa/bcm_sf2.c b/drivers/net/dsa/bcm_sf2.c
index 7aecc98d0a18..32025b990437 100644
--- a/drivers/net/dsa/bcm_sf2.c
+++ b/drivers/net/dsa/bcm_sf2.c
@@ -205,6 +205,19 @@ static int bcm_sf2_port_setup(struct dsa_switch *ds, int port,
if (port == priv->moca_port)
bcm_sf2_port_intr_enable(priv, port);
+ /* Set per-queue pause threshold to 32 */
+ core_writel(priv, 32, CORE_TXQ_THD_PAUSE_QN_PORT(port));
+
+ /* Set ACB threshold to 24 */
+ for (i = 0; i < SF2_NUM_EGRESS_QUEUES; i++) {
+ reg = acb_readl(priv, ACB_QUEUE_CFG(port *
+ SF2_NUM_EGRESS_QUEUES + i));
+ reg &= ~XOFF_THRESHOLD_MASK;
+ reg |= 24;
+ acb_writel(priv, reg, ACB_QUEUE_CFG(port *
+ SF2_NUM_EGRESS_QUEUES + i));
+ }
+
return b53_enable_port(ds, port, phy);
}
@@ -613,6 +626,20 @@ static void bcm_sf2_sw_fixed_link_update(struct dsa_switch *ds, int port,
status->pause = 1;
}
+static void bcm_sf2_enable_acb(struct dsa_switch *ds)
+{
+ struct bcm_sf2_priv *priv = bcm_sf2_to_priv(ds);
+ u32 reg;
+
+ /* Enable ACB globally */
+ reg = acb_readl(priv, ACB_CONTROL);
+ reg |= (ACB_FLUSH_MASK << ACB_FLUSH_SHIFT);
+ acb_writel(priv, reg, ACB_CONTROL);
+ reg &= ~(ACB_FLUSH_MASK << ACB_FLUSH_SHIFT);
+ reg |= ACB_EN | ACB_ALGORITHM;
+ acb_writel(priv, reg, ACB_CONTROL);
+}
+
static int bcm_sf2_sw_suspend(struct dsa_switch *ds)
{
struct bcm_sf2_priv *priv = bcm_sf2_to_priv(ds);
@@ -655,6 +682,8 @@ static int bcm_sf2_sw_resume(struct dsa_switch *ds)
bcm_sf2_imp_setup(ds, port);
}
+ bcm_sf2_enable_acb(ds);
+
return 0;
}
@@ -766,6 +795,7 @@ static int bcm_sf2_sw_setup(struct dsa_switch *ds)
}
bcm_sf2_sw_configure_vlan(ds);
+ bcm_sf2_enable_acb(ds);
return 0;
}
diff --git a/drivers/net/dsa/bcm_sf2_regs.h b/drivers/net/dsa/bcm_sf2_regs.h
index d8b8074a47b9..d1596dfca323 100644
--- a/drivers/net/dsa/bcm_sf2_regs.h
+++ b/drivers/net/dsa/bcm_sf2_regs.h
@@ -115,6 +115,24 @@ enum bcm_sf2_reg_offs {
#define P7_IRQ_OFF 0
#define P_IRQ_OFF(x) ((6 - (x)) * P_NUM_IRQ)
+/* Register set relative to 'ACB' */
+#define ACB_CONTROL 0x00
+#define ACB_EN (1 << 0)
+#define ACB_ALGORITHM (1 << 1)
+#define ACB_FLUSH_SHIFT 2
+#define ACB_FLUSH_MASK 0x3
+
+#define ACB_QUEUE_0_CFG 0x08
+#define XOFF_THRESHOLD_MASK 0x7ff
+#define XON_EN (1 << 11)
+#define TOTAL_XOFF_THRESHOLD_SHIFT 12
+#define TOTAL_XOFF_THRESHOLD_MASK 0x7ff
+#define TOTAL_XOFF_EN (1 << 23)
+#define TOTAL_XON_EN (1 << 24)
+#define PKTLEN_SHIFT 25
+#define PKTLEN_MASK 0x3f
+#define ACB_QUEUE_CFG(x) (ACB_QUEUE_0_CFG + ((x) * 0x4))
+
/* Register set relative to 'CORE' */
#define CORE_G_PCTL_PORT0 0x00000
#define CORE_G_PCTL_PORT(x) (CORE_G_PCTL_PORT0 + (x * 0x4))
@@ -237,6 +255,11 @@ enum bcm_sf2_reg_offs {
#define CORE_PORT_VLAN_CTL_PORT(x) (0xc400 + ((x) * 0x8))
#define PORT_VLAN_CTRL_MASK 0x1ff
+#define CORE_TXQ_THD_PAUSE_QN_PORT_0 0x2c80
+#define TXQ_PAUSE_THD_MASK 0x7ff
+#define CORE_TXQ_THD_PAUSE_QN_PORT(x) (CORE_TXQ_THD_PAUSE_QN_PORT_0 + \
+ (x) * 0x8)
+
#define CORE_DEFAULT_1Q_TAG_P(x) (0xd040 + ((x) * 8))
#define CFI_SHIFT 12
#define PRI_SHIFT 13
--
2.9.3
^ permalink raw reply related
* [PATCH net-next 3/5] net: systemport: Establish lower/upper queue mapping
From: Florian Fainelli @ 2017-10-11 17:57 UTC (permalink / raw)
To: netdev; +Cc: davem, andrew, vivien.didelot, Florian Fainelli
In-Reply-To: <20171011175752.22030-1-f.fainelli@gmail.com>
Establish a queue mapping between the DSA slave network device queues
created that correspond to switch port queues, and the transmit queue
that SYSTEMPORT manages.
We need to configure the SYSTEMPORT transmit queue with the switch port number
and switch port queue number in order for the switch and SYSTEMPORT hardware to
utilize the out of band congestion notification. This hardware mechanism works
by looking at the switch port egress queue and determines whether there is
enough buffers for this queue, with that class of service for a successful
transmission and if not, backpressures the SYSTEMPORT queue that is being used.
For this to work, we implement a notifier which looks at the
DSA_PORT_REGISTER event. When DSA network devices are registered, the
framework calls the DSA notifiers when that happens, extracts the number
of queues for these devices and their associated port number, remembers
that in the driver private structure and linearly maps those queues to
TX rings/queues that we manage.
This scheme works because DSA slave network deviecs always transmit
through SYSTEMPORT so when DSA slave network devices are
destroyed/brought down, the corresponding SYSTEMPORT queues are no
longer used. Also, by design of the DSA framework, the master network
device (SYSTEMPORT) is registered first.
For faster lookups we use an array of up to DSA_MAX_PORTS * number of
queues per port, and then map pointers to bcm_sysport_tx_ring such that
our ndo_select_queue() implementation can just index into that array to
locate the corresponding ring index.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
drivers/net/ethernet/broadcom/bcmsysport.c | 115 ++++++++++++++++++++++++++++-
drivers/net/ethernet/broadcom/bcmsysport.h | 11 ++-
2 files changed, 121 insertions(+), 5 deletions(-)
diff --git a/drivers/net/ethernet/broadcom/bcmsysport.c b/drivers/net/ethernet/broadcom/bcmsysport.c
index 83eec9a8c275..78bed9a84e81 100644
--- a/drivers/net/ethernet/broadcom/bcmsysport.c
+++ b/drivers/net/ethernet/broadcom/bcmsysport.c
@@ -1416,7 +1416,14 @@ static int bcm_sysport_init_tx_ring(struct bcm_sysport_priv *priv,
tdma_writel(priv, 0, TDMA_DESC_RING_COUNT(index));
tdma_writel(priv, 1, TDMA_DESC_RING_INTR_CONTROL(index));
tdma_writel(priv, 0, TDMA_DESC_RING_PROD_CONS_INDEX(index));
- tdma_writel(priv, RING_IGNORE_STATUS, TDMA_DESC_RING_MAPPING(index));
+
+ /* Configure QID and port mapping */
+ reg = tdma_readl(priv, TDMA_DESC_RING_MAPPING(index));
+ reg &= ~(RING_QID_MASK | RING_PORT_ID_MASK << RING_PORT_ID_SHIFT);
+ reg |= ring->switch_queue & RING_QID_MASK;
+ reg |= ring->switch_port << RING_PORT_ID_SHIFT;
+ reg |= RING_IGNORE_STATUS;
+ tdma_writel(priv, reg, TDMA_DESC_RING_MAPPING(index));
tdma_writel(priv, 0, TDMA_DESC_RING_PCP_DEI_VID(index));
/* Do not use tdma_control_bit() here because TSB_SWAP1 collides
@@ -1447,8 +1454,9 @@ static int bcm_sysport_init_tx_ring(struct bcm_sysport_priv *priv,
napi_enable(&ring->napi);
netif_dbg(priv, hw, priv->netdev,
- "TDMA cfg, size=%d, desc_cpu=%p\n",
- ring->size, ring->desc_cpu);
+ "TDMA cfg, size=%d, desc_cpu=%p switch q=%d,port=%d\n",
+ ring->size, ring->desc_cpu, ring->switch_queue,
+ ring->switch_port);
return 0;
}
@@ -2011,6 +2019,92 @@ static const struct ethtool_ops bcm_sysport_ethtool_ops = {
.set_link_ksettings = phy_ethtool_set_link_ksettings,
};
+static u16 bcm_sysport_select_queue(struct net_device *dev, struct sk_buff *skb,
+ void *accel_priv,
+ select_queue_fallback_t fallback)
+{
+ struct bcm_sysport_priv *priv = netdev_priv(dev);
+ u16 queue = skb_get_queue_mapping(skb);
+ struct bcm_sysport_tx_ring *tx_ring;
+ unsigned int q, port;
+
+ if (!netdev_uses_dsa(dev))
+ return fallback(dev, skb);
+
+ /* DSA tagging layer will have configured the correct queue */
+ q = BRCM_TAG_GET_QUEUE(queue);
+ port = BRCM_TAG_GET_PORT(queue);
+ tx_ring = priv->ring_map[q + port * priv->per_port_num_tx_queues];
+
+ return tx_ring->index;
+}
+
+static int bcm_sysport_map_queues(struct net_device *dev,
+ struct dsa_notifier_register_info *info)
+{
+ struct bcm_sysport_priv *priv = netdev_priv(dev);
+ struct bcm_sysport_tx_ring *ring;
+ struct net_device *slave_dev;
+ unsigned int num_tx_queues;
+ unsigned int q, start, port;
+
+ /* We can't be setting up queue inspection for non directly attached
+ * switches
+ */
+ if (info->switch_number)
+ return 0;
+
+ port = info->port_number;
+ slave_dev = info->info.dev;
+
+ /* On SYSTEMPORT Lite we have twice as less queues, so we cannot do a
+ * 1:1 mapping, we can only do a 2:1 mapping. By reducing the number of
+ * per-port (slave_dev) network devices queue, we achieve just that.
+ * This need to happen now before any slave network device is used such
+ * it accurately reflects the number of real TX queues.
+ */
+ if (priv->is_lite)
+ netif_set_real_num_tx_queues(slave_dev,
+ slave_dev->num_tx_queues / 2);
+ num_tx_queues = slave_dev->real_num_tx_queues;
+
+ if (priv->per_port_num_tx_queues &&
+ priv->per_port_num_tx_queues != num_tx_queues)
+ netdev_warn(slave_dev, "asymetric number of per-port queues\n");
+
+ priv->per_port_num_tx_queues = num_tx_queues;
+
+ start = find_first_zero_bit(&priv->queue_bitmap, dev->num_tx_queues);
+ for (q = 0; q < num_tx_queues; q++) {
+ ring = &priv->tx_rings[q + start];
+
+ /* Just remember the mapping actual programming done
+ * during bcm_sysport_init_tx_ring
+ */
+ ring->switch_queue = q;
+ ring->switch_port = port;
+ priv->ring_map[q + port * num_tx_queues] = ring;
+
+ /* Set all queues as being used now */
+ set_bit(q + start, &priv->queue_bitmap);
+ }
+
+ return 0;
+}
+
+static int bcm_sysport_dsa_notifier(struct notifier_block *unused,
+ unsigned long event, void *ptr)
+{
+ struct dsa_notifier_register_info *info;
+
+ if (event != DSA_PORT_REGISTER)
+ return NOTIFY_DONE;
+
+ info = ptr;
+
+ return notifier_from_errno(bcm_sysport_map_queues(info->master, info));
+}
+
static const struct net_device_ops bcm_sysport_netdev_ops = {
.ndo_start_xmit = bcm_sysport_xmit,
.ndo_tx_timeout = bcm_sysport_tx_timeout,
@@ -2023,6 +2117,7 @@ static const struct net_device_ops bcm_sysport_netdev_ops = {
.ndo_poll_controller = bcm_sysport_poll_controller,
#endif
.ndo_get_stats64 = bcm_sysport_get_stats64,
+ .ndo_select_queue = bcm_sysport_select_queue,
};
#define REV_FMT "v%2x.%02x"
@@ -2172,10 +2267,18 @@ static int bcm_sysport_probe(struct platform_device *pdev)
u64_stats_init(&priv->syncp);
+ priv->dsa_notifier.notifier_call = bcm_sysport_dsa_notifier;
+
+ ret = register_dsa_notifier(&priv->dsa_notifier);
+ if (ret) {
+ dev_err(&pdev->dev, "failed to register DSA notifier\n");
+ goto err_deregister_fixed_link;
+ }
+
ret = register_netdev(dev);
if (ret) {
dev_err(&pdev->dev, "failed to register net_device\n");
- goto err_deregister_fixed_link;
+ goto err_deregister_notifier;
}
priv->rev = topctrl_readl(priv, REV_CNTL) & REV_MASK;
@@ -2188,6 +2291,8 @@ static int bcm_sysport_probe(struct platform_device *pdev)
return 0;
+err_deregister_notifier:
+ unregister_dsa_notifier(&priv->dsa_notifier);
err_deregister_fixed_link:
if (of_phy_is_fixed_link(dn))
of_phy_deregister_fixed_link(dn);
@@ -2199,11 +2304,13 @@ static int bcm_sysport_probe(struct platform_device *pdev)
static int bcm_sysport_remove(struct platform_device *pdev)
{
struct net_device *dev = dev_get_drvdata(&pdev->dev);
+ struct bcm_sysport_priv *priv = netdev_priv(dev);
struct device_node *dn = pdev->dev.of_node;
/* Not much to do, ndo_close has been called
* and we use managed allocations
*/
+ unregister_dsa_notifier(&priv->dsa_notifier);
unregister_netdev(dev);
if (of_phy_is_fixed_link(dn))
of_phy_deregister_fixed_link(dn);
diff --git a/drivers/net/ethernet/broadcom/bcmsysport.h b/drivers/net/ethernet/broadcom/bcmsysport.h
index 82e401df199e..82f70a6783cb 100644
--- a/drivers/net/ethernet/broadcom/bcmsysport.h
+++ b/drivers/net/ethernet/broadcom/bcmsysport.h
@@ -404,7 +404,7 @@ struct bcm_rsb {
#define RING_CONS_INDEX_MASK 0xffff
#define RING_MAPPING 0x14
-#define RING_QID_MASK 0x3
+#define RING_QID_MASK 0x7
#define RING_PORT_ID_SHIFT 3
#define RING_PORT_ID_MASK 0x7
#define RING_IGNORE_STATUS (1 << 6)
@@ -712,6 +712,8 @@ struct bcm_sysport_tx_ring {
struct bcm_sysport_priv *priv; /* private context backpointer */
unsigned long packets; /* packets statistics */
unsigned long bytes; /* bytes statistics */
+ unsigned int switch_queue; /* switch port queue number */
+ unsigned int switch_port; /* switch port queue number */
};
/* Driver private structure */
@@ -765,5 +767,12 @@ struct bcm_sysport_priv {
/* For atomic update generic 64bit value on 32bit Machine */
struct u64_stats_sync syncp;
+
+ /* map information between switch port queues and local queues */
+ struct notifier_block dsa_notifier;
+ unsigned int per_port_num_tx_queues;
+ unsigned long queue_bitmap;
+ struct bcm_sysport_tx_ring *ring_map[DSA_MAX_PORTS * 8];
+
};
#endif /* __BCM_SYSPORT_H */
--
2.9.3
^ permalink raw reply related
* [PATCH net-next 2/5] net: dsa: tag_brcm: Indicate to master netdevice port + queue
From: Florian Fainelli @ 2017-10-11 17:57 UTC (permalink / raw)
To: netdev; +Cc: davem, andrew, vivien.didelot, Florian Fainelli
In-Reply-To: <20171011175752.22030-1-f.fainelli@gmail.com>
We need to tell the DSA master network device doing the actual
transmission what the desired switch port and queue number is for it to
resolve that to the internal transmit queue it is mapped to.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
include/net/dsa.h | 5 +++++
net/dsa/tag_brcm.c | 6 ++++++
2 files changed, 11 insertions(+)
diff --git a/include/net/dsa.h b/include/net/dsa.h
index 40a709a0754d..ce1d622734d7 100644
--- a/include/net/dsa.h
+++ b/include/net/dsa.h
@@ -516,4 +516,9 @@ static inline int call_dsa_notifiers(unsigned long val, struct net_device *dev,
}
#endif
+/* Broadcom tag specific helpers to insert and extract queue/port number */
+#define BRCM_TAG_SET_PORT_QUEUE(p, q) ((p) << 8 | q)
+#define BRCM_TAG_GET_PORT(v) ((v) >> 8)
+#define BRCM_TAG_GET_QUEUE(v) ((v) & 0xff)
+
#endif
diff --git a/net/dsa/tag_brcm.c b/net/dsa/tag_brcm.c
index 8e4bdb9d9ae3..cc4f472fbd77 100644
--- a/net/dsa/tag_brcm.c
+++ b/net/dsa/tag_brcm.c
@@ -86,6 +86,12 @@ static struct sk_buff *brcm_tag_xmit(struct sk_buff *skb, struct net_device *dev
brcm_tag[2] = BRCM_IG_DSTMAP2_MASK;
brcm_tag[3] = (1 << p->dp->index) & BRCM_IG_DSTMAP1_MASK;
+ /* Now tell the master network device about the desired output queue
+ * as well
+ */
+ skb_set_queue_mapping(skb, BRCM_TAG_SET_PORT_QUEUE(p->dp->index,
+ queue));
+
return skb;
}
--
2.9.3
^ permalink raw reply related
* [PATCH net-next 1/5] net: dsa: Add support for DSA specific notifiers
From: Florian Fainelli @ 2017-10-11 17:57 UTC (permalink / raw)
To: netdev; +Cc: davem, andrew, vivien.didelot, Florian Fainelli
In-Reply-To: <20171011175752.22030-1-f.fainelli@gmail.com>
In preparation for communicating a given DSA network device's port
number and switch index, create a specialized DSA notifier and two
events: DSA_PORT_REGISTER and DSA_PORT_UNREGISTER that communicate: the
slave network device (slave_dev), port number and switch number in the
tree.
This will be later used for network device drivers like bcmsysport which
needs to cooperate with its DSA network devices to set-up queue mapping
and scheduling.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
include/net/dsa.h | 45 +++++++++++++++++++++++++++++++++++++++++++++
net/dsa/dsa.c | 23 +++++++++++++++++++++++
net/dsa/slave.c | 13 +++++++++++++
3 files changed, 81 insertions(+)
diff --git a/include/net/dsa.h b/include/net/dsa.h
index 10dceccd9ce8..40a709a0754d 100644
--- a/include/net/dsa.h
+++ b/include/net/dsa.h
@@ -471,4 +471,49 @@ static inline int dsa_switch_resume(struct dsa_switch *ds)
}
#endif /* CONFIG_PM_SLEEP */
+enum dsa_notifier_type {
+ DSA_PORT_REGISTER,
+ DSA_PORT_UNREGISTER,
+};
+
+struct dsa_notifier_info {
+ struct net_device *dev;
+};
+
+struct dsa_notifier_register_info {
+ struct dsa_notifier_info info; /* must be first */
+ struct net_device *master;
+ unsigned int port_number;
+ unsigned int switch_number;
+};
+
+static inline struct net_device *
+dsa_notifier_info_to_dev(const struct dsa_notifier_info *info)
+{
+ return info->dev;
+}
+
+#if IS_ENABLED(CONFIG_NET_DSA)
+int register_dsa_notifier(struct notifier_block *nb);
+int unregister_dsa_notifier(struct notifier_block *nb);
+int call_dsa_notifiers(unsigned long val, struct net_device *dev,
+ struct dsa_notifier_info *info);
+#else
+static inline int register_dsa_notifier(struct notifier_block *nb)
+{
+ return 0;
+}
+
+static inline int unregister_dsa_notifier(struct notifier_block *nb)
+{
+ return 0;
+}
+
+static inline int call_dsa_notifiers(unsigned long val, struct net_device *dev,
+ struct dsa_notifier_info *info)
+{
+ return NOTIFY_DONE;
+}
+#endif
+
#endif
diff --git a/net/dsa/dsa.c b/net/dsa/dsa.c
index 51ca2a524a27..832c659ff993 100644
--- a/net/dsa/dsa.c
+++ b/net/dsa/dsa.c
@@ -14,6 +14,7 @@
#include <linux/platform_device.h>
#include <linux/slab.h>
#include <linux/module.h>
+#include <linux/notifier.h>
#include <linux/of.h>
#include <linux/of_mdio.h>
#include <linux/of_platform.h>
@@ -261,6 +262,28 @@ bool dsa_schedule_work(struct work_struct *work)
return queue_work(dsa_owq, work);
}
+static ATOMIC_NOTIFIER_HEAD(dsa_notif_chain);
+
+int register_dsa_notifier(struct notifier_block *nb)
+{
+ return atomic_notifier_chain_register(&dsa_notif_chain, nb);
+}
+EXPORT_SYMBOL_GPL(register_dsa_notifier);
+
+int unregister_dsa_notifier(struct notifier_block *nb)
+{
+ return atomic_notifier_chain_unregister(&dsa_notif_chain, nb);
+}
+EXPORT_SYMBOL_GPL(unregister_dsa_notifier);
+
+int call_dsa_notifiers(unsigned long val, struct net_device *dev,
+ struct dsa_notifier_info *info)
+{
+ info->dev = dev;
+ return atomic_notifier_call_chain(&dsa_notif_chain, val, info);
+}
+EXPORT_SYMBOL_GPL(call_dsa_notifiers);
+
static int __init dsa_init_module(void)
{
int rc;
diff --git a/net/dsa/slave.c b/net/dsa/slave.c
index fb2954ff198c..45f4ea845c07 100644
--- a/net/dsa/slave.c
+++ b/net/dsa/slave.c
@@ -1116,6 +1116,7 @@ int dsa_slave_resume(struct net_device *slave_dev)
int dsa_slave_create(struct dsa_port *port, const char *name)
{
+ struct dsa_notifier_register_info rinfo = { };
struct dsa_switch *ds = port->ds;
struct net_device *master;
struct net_device *slave_dev;
@@ -1177,6 +1178,12 @@ int dsa_slave_create(struct dsa_port *port, const char *name)
goto out_free;
}
+ rinfo.info.dev = slave_dev;
+ rinfo.master = master;
+ rinfo.port_number = p->dp->index;
+ rinfo.switch_number = p->dp->ds->index;
+ call_dsa_notifiers(DSA_PORT_REGISTER, slave_dev, &rinfo.info);
+
ret = register_netdev(slave_dev);
if (ret) {
netdev_err(master, "error %d registering interface %s\n",
@@ -1200,6 +1207,7 @@ int dsa_slave_create(struct dsa_port *port, const char *name)
void dsa_slave_destroy(struct net_device *slave_dev)
{
struct dsa_slave_priv *p = netdev_priv(slave_dev);
+ struct dsa_notifier_register_info rinfo = { };
struct device_node *port_dn;
port_dn = p->dp->dn;
@@ -1211,6 +1219,11 @@ void dsa_slave_destroy(struct net_device *slave_dev)
if (of_phy_is_fixed_link(port_dn))
of_phy_deregister_fixed_link(port_dn);
}
+ rinfo.info.dev = slave_dev;
+ rinfo.master = p->dp->cpu_dp->netdev;
+ rinfo.port_number = p->dp->index;
+ rinfo.switch_number = p->dp->ds->index;
+ call_dsa_notifiers(DSA_PORT_UNREGISTER, slave_dev, &rinfo.info);
unregister_netdev(slave_dev);
free_percpu(p->stats64);
free_netdev(slave_dev);
--
2.9.3
^ permalink raw reply related
* [PATCH net-next 0/5] Enable ACB for bcm_sf2 and bcmsysport
From: Florian Fainelli @ 2017-10-11 17:57 UTC (permalink / raw)
To: netdev; +Cc: davem, andrew, vivien.didelot, Florian Fainelli
Hi all,
This patch series enables Broadcom's Advanced Congestion Buffering mechanism
which requires cooperation between the CPU/Management Ethernet MAC controller
and the switch.
I took the notifier approach because ultimately the information we need to
carry to the master network device is DSA specific and I saw little room for
generalizing beyond what DSA requires. Chances are that this is highly specific
to the Broadcom HW as I don't know of any HW out there that supports something
nearly similar for similar or identical needs.
Florian Fainelli (5):
net: dsa: Add support for DSA specific notifiers
net: dsa: tag_brcm: Indicate to master netdevice port + queue
net: systemport: Establish lower/upper queue mapping
net: dsa: bcm_sf2: Turn on ACB at the switch level
net: systemport: Turn on ACB at the SYSTEMPORT level
drivers/net/dsa/bcm_sf2.c | 30 ++++++++
drivers/net/dsa/bcm_sf2_regs.h | 23 ++++++
drivers/net/ethernet/broadcom/bcmsysport.c | 119 ++++++++++++++++++++++++++++-
drivers/net/ethernet/broadcom/bcmsysport.h | 11 ++-
include/net/dsa.h | 50 ++++++++++++
net/dsa/dsa.c | 23 ++++++
net/dsa/slave.c | 13 ++++
net/dsa/tag_brcm.c | 6 ++
8 files changed, 270 insertions(+), 5 deletions(-)
--
2.9.3
^ permalink raw reply
* Re: [PATCH iproute2] iplink: new option to set neigh suppression on a bridge port
From: Stephen Hemminger @ 2017-10-11 17:58 UTC (permalink / raw)
To: Roopa Prabhu; +Cc: netdev, nikolay
In-Reply-To: <1507610533-18724-1-git-send-email-roopa@cumulusnetworks.com>
On Mon, 9 Oct 2017 21:42:13 -0700
Roopa Prabhu <roopa@cumulusnetworks.com> wrote:
> From: Roopa Prabhu <roopa@cumulusnetworks.com>
>
> neigh suppression can be used to suppress arp and nd flood
> to bridge ports. It maps to the recently added
> kernel support for bridge port flag IFLA_BRPORT_NEIGH_SUPPRESS.
>
> Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Applied to net-next branch.
^ 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