Since hclgevf_reset_wait() is used to wait for the hardware to complete
the reset, it is not necessary to hold the rtnl_lock during
hclgevf_reset_wait(). So this patch releases the lock for the duration
of hclgevf_reset_wait().
Fixes: 6988eb2a9b77 ("net: hns3: Add support to reset the enet/ring
The HEAD pointer of the hardware command queue maybe equal to the command
queue's next_to_use in the driver, so that does not belong to the invalid
HEAD pointer, since the hardware may not process the command in time,
causing the HEAD pointer to be too late to update. The variables' name
in this
In a multi-core machine, the mailbox service and reset service
will be executed at the same time. The reset service will re-initialize
the command queue, before that, the mailbox handler can only get some
invalid messages.
The HCLGE_STATE_CMD_DISABLE flag means that the command queue is not
Since hclge_reset_wait() is used to wait for the hardware to complete
the reset, it is not necessary to hold the rtnl_lock during
hclge_reset_wait(). So this patch releases the lock for the duration
of hclge_reset_wait().
Fixes: 6d4fab39533f ("net: hns3: Reset net device with rtnl_lock")
This patch series include bugfix for the HNS3 ethernet
controller driver.
Change log:
V2->V3:
Fixes comments from Sergei Shtylyov
V1->V2:
Fixes the compilation break reported by kbuild test robot
http://patchwork.ozlabs.org/patch/989818/
Huazhong Tan (11):
net: hns3:
When hns3_nic_init_vector_data() failed for mapping ring to vector,
it should cancel the netif_napi_add() that have been successfully done
and then exit.
Fixes: 76ad4f0ee747 ("net: hns3: Add support of HNS3 Ethernet Driver for hip08
SoC")
Signed-off-by: Huazhong Tan
---
The current driver supports handling two vector0 interrupts, reset and
mailbox. When the hardware reports an interrupt of another type of
interrupt source, if the driver does not process the interrupt and
enables the interrupt, the hardware will repeatedly report the unknown
interrupt.
Therefore,
When there is a PHY, the driver needs to complete some operations through
MDIO during reset reinitialization, so HCLGE_STATE_CMD_DISABLE is more
suitable than HCLGE_STATE_RST_HANDLING to prevent the MDIO operation from
being sent during the hardware reset.
Fixes: b50ae26c57cb ("net: hns3: never
There are some functions that, when they fail to send the command,
need to return the corresponding error value to its caller.
Fixes: 46a3df9f9718 ("net: hns3: Add HNS3 Acceleration Engine & Compatibility
Layer Support")
Fixes: 681ec3999b3d ("net: hns3: fix for vlan table lost problem when
It is not necessary to reset the queue in the hns3_uninit_all_ring(),
since the queue is stopped in the down operation, and will be reset
in the up operaton. And the judgment of the HCLGE_STATE_RST_HANDLING
flag in the hclge_reset_tqp() is not correct, because we need to reset
tqp during pf reset,
When hns3_get_ring_config()/hns3_queue_to_ring() failed during resetting,
the allocated memory has not been freed before hns3_get_ring_config() and
hns3_queue_to_ring() return. So this patch fixes the buffer not freeing
problem during resetting.
Fixes: 76ad4f0ee747 ("net: hns3: Add support of
The spin lock of the command queue only needs to be initialized once
when the driver initializes the command queue. It is not necessary to
initialize the spin lock when resetting. At the same time, the
modification of the queue member should be performed after acquiring
the lock.
Fixes:
On 2018/10/28 3:05, Sergei Shtylyov wrote:
On 27.10.2018 5:41, Huazhong Tan wrote:
In a multi-core machine, the mailbox service and reset service
will be executed at the same time. The reset server will re-initialize
the commond queue, before that, the mailbox handler can only get some
On 2018/10/28 3:02, Sergei Shtylyov wrote:
Hello!
On 27.10.2018 5:41, Huazhong Tan wrote:
It is not necessary to reset the queue in the hns3_uninit_all_ring(),
since the queue is stopped in the down operation, and will be resetted
s/resetted/reset/.
OK, thanks.
in the up
icmp6_send() function is expensive on systems with a large number of
interfaces. Every time it’s called, it has to verify that the source
address does not correspond to an existing anycast address by looping
through every device and every anycast address on the device. This can
result in
You are right, David...I mistook the refcount_dec_and_test() in
aca_put() as being for the fib6_info, but it's for the aca_refcnt.
Thanks! I'll submit a corrected patch.
On Sat, Oct 27, 2018 at 7:39 PM David Ahern wrote:
>
> On 10/27/18 12:02 PM, Jeff Barnhill wrote:
> > @@ -275,6 +356,11 @@ int
On 10/8/2018 11:14 AM, Jason Wang wrote:
On 2018年09月29日 23:41, Jianfeng Tan wrote:
When we use raw socket as the vhost backend, a packet from virito with
gso offloading information, cannot be sent out in later validaton at
xmit path, as we did not set correct skb->protocol which is further
On 10/27/18 12:02 PM, Jeff Barnhill wrote:
> @@ -275,6 +356,11 @@ int __ipv6_dev_ac_inc(struct inet6_dev *idev, const
> struct in6_addr *addr)
> err = -ENOMEM;
> goto out;
> }
> + err = ipv6_add_acaddr_hash(dev_net(idev->dev), addr);
> + if (err) {
> +
Tobias Jungel wrote:
>The attribute IFLA_BOND_AD_ACTOR_SYSTEM is sent to user space having the
>length of sizeof(bond->params.ad_actor_system) which is 8 byte. This
>patch aligns the length to ETH_ALEN to have the same MAC address exposed
>as using sysfs.
>
>fixes f87fda00b6ed2
>
>Signed-off-by:
On 26.10.2018 13:45, Andre Tomt wrote:
On 25.10.2018 19:38, Eric Dumazet wrote:
On 10/24/2018 12:41 PM, Andre Tomt wrote:
It eventually showed up again with mlx4, on 4.18.16 + fix and also on
4.19. I still do not have a useful packet capture.
It is running a torrent client serving up
When an FDB entry is configured, the address is validated to have the
length of an Ethernet address, but the device for which the address is
configured can be of any type.
The above can result in the use of uninitialized memory when the address
is later compared against existing addresses since
Hi,
just wanted to report, thet after switching to 4.19 (fro 4.14.x, so maybe
the problem appeared somewhere between), I'm getting tons of similar
messages:
Oct 27 09:06:27 xxx kernel: br501: hw csum failure
Oct 27 09:06:27 xxx kernel: CPU: 8 PID: 0 Comm: swapper/8 Tainted: G
E
On 27.10.2018 5:41, Huazhong Tan wrote:
In a multi-core machine, the mailbox service and reset service
will be executed at the same time. The reset server will re-initialize
the commond queue, before that, the mailbox handler can only get some
Command?
invalid messages.
The
Hello!
On 27.10.2018 5:41, Huazhong Tan wrote:
It is not necessary to reset the queue in the hns3_uninit_all_ring(),
since the queue is stopped in the down operation, and will be resetted
s/resetted/reset/.
in the up operaton. And the judgment of the HCLGE_STATE_RST_HANDLING
flag in the
icmp6_send() function is expensive on systems with a large number of
interfaces. Every time it’s called, it has to verify that the source
address does not correspond to an existing anycast address by looping
through every device and every anycast address on the device. This can
result in
q_{etf,taprio}.c uses CLOCK_TAI, which isn't exposed by glibc < 2.21 or
uClibc, breaking the build. Provide a fallback definition like it is done
for IPPROTO_MPLS and others.
Signed-off-by: Peter Korsgaard
---
include/utils.h | 4
1 file changed, 4 insertions(+)
diff --git
> What would you think of a flag on the receiving end like,
> "CHECKSUM_INVALID_BUT_UNNECESSARY"? It would be treated as
> CHECKSUM_UNNECESSARY in the case that the the packet is locally
> received. But if the packet is going to be forwarded instead, then
> skb_checksum_help is called on it before
The attribute IFLA_BOND_AD_ACTOR_SYSTEM is sent to user space having the
length of sizeof(bond->params.ad_actor_system) which is 8 byte. This
patch aligns the length to ETH_ALEN to have the same MAC address exposed
as using sysfs.
fixes f87fda00b6ed2
Signed-off-by: Tobias Jungel
---
Previouly, virtio net header size is hardcoded to be 10, which makes
the feature mrg_rxbuf not available.
We redefine PACKET_VNET_HDR ioctl which treats user input as boolean,
but now as int, 0, 10, 12, or everything else be treated as 10.
There will be one case which is treated differently: if
If the hi3110 shares the SPI bus with another traffic-intensive device
and packets are received in high volume (by a separate machine sending
with "cangen -g 0 -i -x"), reception stops after a few minutes and the
counter in /proc/interrupts stops incrementing. Bus state is "active".
Bringing the
When hns3_nic_init_vector_data() failed for mapping ring to vector,
it should cancel the netif_napi_add() that have been successfully done
and then exit.
Fixes: 76ad4f0ee747 ("net: hns3: Add support of HNS3 Ethernet Driver for hip08
SoC")
Signed-off-by: Huazhong Tan
---
The spin lock of the command queue only needs to be initialized once
when the driver initializes the command queue. It is not necessary to
initialize the spin lock when resetting. At the same time, the
modification of the queue member should be performed after acquiring
the lock.
Fixes:
When hns3_get_ring_config()/hns3_queue_to_ring() failed during resetting,
the allocated memory has not been freed before hns3_get_ring_config() and
hns3_queue_to_ring() return. So this patch fixes the buffer not freeing
problem during resetting.
Fixes: 76ad4f0ee747 ("net: hns3: Add support of
Since hclge_reset_wait() is used to wait for the hardware to complete
the reset, it is not necessary to hold the rtnl_lock during
hclge_reset_wait(). So this patch release the lock for the duration
of hclge_reset_wait().
Fixes: 6d4fab39533f ("net: hns3: Reset net device with rtnl_lock")
This patch series include bugfix for the HNS3 ethernet
controller driver.
Change log:
V1->V2:
Fixes the compilation break reported by kbuild test robot
http://patchwork.ozlabs.org/patch/989818/
Huazhong Tan (11):
net: hns3: add error handler for hns3_nic_init_vector_data()
The current driver supports handling two vector0 interrupts, reset and
mailbox. When the hardware reports an interrupt of another type of
interrupt source, if the driver does not process the interrupt and
enables the interrupt, the hardware will repeatedly report the unknown
interrupt.
Therefore,
When there is a PHY, the driver needs to complete some operations through
MDIO during reset reinitialization, so HCLGE_STATE_CMD_DISABLE is more
suitable than HCLGE_STATE_RST_HANDLING to prevent the MDIO operation from
being sent during the hardware reset.
Fixes: b50ae26c57cb ("net: hns3: never
There are some functions that, when they fail to execute a send command,
need to return the corresponding error value to its caller.
Fixes: 46a3df9f9718 ("net: hns3: Add HNS3 Acceleration Engine & Compatibility
Layer Support")
Fixes: 681ec3999b3d ("net: hns3: fix for vlan table lost problem when
The HEAD pointer of the hardware command queue maybe equal to the command
queue's next_to_use the driver, so that does not belong to the invalid
HEAD pointer, since the hardware may not process the command in time,
causing the HEAD pointer to be too late to update. The variables' name
in this
It is not necessary to reset the queue in the hns3_uninit_all_ring(),
since the queue is stopped in the down operation, and will be resetted
in the up operaton. And the judgment of the HCLGE_STATE_RST_HANDLING
flag in the hclge_reset_tqp() is not correct, because we need to reset
tqp during pf
In a multi-core machine, the mailbox service and reset service
will be executed at the same time. The reset server will re-initialize
the commond queue, before that, the mailbox handler can only get some
invalid messages.
The HCLGE_STATE_CMD_DISABLE flag means that the command queue is not
Since hclgevf_reset_wait() is used to wait for the hardware to complete
the reset, it is not necessary to hold the rtnl_lock during
hclgevf_reset_wait(). So this patch release the lock for the duration
of hclgevf_reset_wait().
Fixes: 6988eb2a9b77 ("net: hns3: Add support to reset the enet/ring
On 27/10/2018 03:50, kbuild test robot wrote:
> tree: https://git.kernel.org/pub/scm/linux/kernel/git/davem/net.git master
> head: aab456dfa404f3a16d6f1780e62a6a8533c4d008
> commit: 5a2de63fd1a59c30c02526d427bc014b98adf508 [17/19] bridge: do not add
> port to router list when receives query
xfrm_output_one() does not return a error code when there is
no dst_entry attached to the skb, it is still possible crash
with a NULL pointer dereference in xfrm_output_resume(). Fix
it by return error code -EHOSTUNREACH.
Fixes: 9e1437937807 ("xfrm: Fix NULL pointer dereference when skb_dst_force
44 matches
Mail list logo
