16.11.3 (LTS) patches review and test

16.11.3 (LTS) patches review and test

[Yuanhan Liu]

Hi all,

Here is a list of patches targeted for LTS release 16.11.3. Please
help review and test. The planned date for the final release is 30th,
Aug. Before that, please let me know if anyone has objections with
these patches being applied.

These patches are located at branch 16.11 of dpdk-stable repo:
    http://dpdk.org/browse/dpdk-stable/

Thanks.

        --yliu

---
Aaron Conole (1):
      net/enic: fix crash when freeing 0 packet to mempool

Ajit Khaparde (9):
      net/bnxt: fix reporting of link status
      net/bnxt: fix get link config
      net/bnxt: fix autoneg on 10GBase-T links
      net/bnxt: fix set link config
      net/bnxt: fix set link config
      net/bnxt: check invalid L2 filter id
      net/bnxt: free filter before reusing it
      net/bnxt: enable default VNIC allocation
      net/bnxt: fix vnic cleanup

Allain Legacy (1):
      net/virtio-user: fix crash when detaching device

Arek Kusztal (1):
      crypto/qat: fix NULL authentication hang

Beilei Xing (1):
      net/i40e: fix ethertype filter for new FW

Chas Williams (1):
      net/vmxnet3: fix filtering on promiscuous disabling

Cian Ferriter (1):
      mbuf: fix VXLAN port in comment

Dahir Osman (1):
      ip_frag: free mbufs on reassembly table destroy

Daniel Mrzyglod (2):
      test/bonding: fix mode 4 names
      test/bonding: fix namespace of the RSS tests

Dariusz Stojaczyk (1):
      vhost: fix guest pages memory leak

Declan Doherty (2):
      net/bonding: change link status check to no-wait
      net/bonding: fix number of bonding Tx/Rx queues

Ferruh Yigit (3):
      kni: fix build with gcc 7.1
      net/enic: fix build with gcc 7.1
      net/mlx5: fix build with gcc 7.1

Gowrishankar Muthukrishnan (1):
      net/mlx5: fix inconsistent link status query

Hemant Agrawal (1):
      examples/qos_sched: fix build for less lcores

Herakliusz Lipiec (1):
      net/i40e: add return value checks

Herbert Guan (2):
      test/bonding: fix memory corruptions
      test/bonding: fix parameters of a balance Tx

Ivan Dyukov (1):
      vhost: fix checking of device features

Jamie Lavigne (1):
      mem: fix malloc element resize with padding

Jeff Guo (1):
      net/i40e: fix link down and negotiation

Jianfeng Tan (3):
      eal: fix config file path when checking process
      vhost: fix TCP checksum
      vhost: fix IP checksum

Jingjing Wu (1):
      net/e1000: fix LSC interrupt

Kirill Rybalchenko (1):
      cryptodev: fix device stop function

Mandeep Rohilla (1):
      net/vmxnet3: fix receive queue memory leak

Markus Theil (1):
      net/igb: fix add/delete of flex filters

Matan Azrad (1):
      net/mlx4: fix probe failure report

Michal Krawczyk (1):
      net/ena: fix cleanup of the Tx bufs

Pablo de Lara (12):
      examples/l2fwd-crypto: fix option parsing
      examples/l2fwd-crypto: fix application help
      test/crypto: fix overflow
      examples/l2fwd-crypto: fix padding
      crypto/aesni_mb: fix HMAC supported key sizes
      crypto/openssl: fix HMAC supported key sizes
      mbuf: fix doxygen comment of bulk alloc
      crypto/qat: fix HMAC supported key sizes
      crypto/qat: fix SHA384-HMAC block size
      doc: remove incorrect limitation on AESNI-MB PMD
      doc: remove incorrect limitation on QAT PMD
      test/crypto: fix wrong AAD setting

Qi Zhang (5):
      net/i40e: exclude internal packet's byte count
      net/i40e: fix VF statistics
      net/ixgbe: fix mirror rule index overflow
      net/i40e: fix Rx data segment buffer length
      examples/l3fwd: fix IPv6 packet type parse

Qiming Yang (1):
      net/ixgbe: fix Rx/Tx queue interrupt for x550 devices

Rahul Lakkireddy (1):
      net/cxgbe: fix port statistics

Rasesh Mody (1):
      net/qede: fix chip details print

Sha Zhang (1):
      net/bonding: fix when NTT flag updated

Shahaf Shuler (1):
      net/mlx5: fix completion buffer size

Stephen Hemminger (2):
      net/virtio: do not falsely claim to do IP checksum
      net/virtio: do not claim to support LRO

Tiwei Bie (4):
      net/virtio: zero the whole memory zone
      contigmem: free allocated memory on error
      contigmem: do not zero pages during each mmap
      vhost: make page logging atomic

Vasily Philipov (2):
      mbuf: fix debug checks for headroom and tailroom
      net/mlx4: fix mbuf poisoning in debug code

Wei Dai (1):
      lpm: fix index of tbl8

Wei Zhao (2):
      net/igb: fix checksum valid flags
      net/igb: fix flex filter length

Wenzhuo Lu (1):
      net/i40e/base: fix Tx error stats on VF

Xiao Wang (1):
      net/fm10k: initialize link status in device start

Yong Wang (1):
      net/i40e: fix division by 0

Yongseok Koh (2):
      net/mlx5: fix exception handling
      net/mlx5: fix redundant free of Tx buffer

Re: [dpdk-stable] 16.11.3 (LTS) patches review and test

[Kevin Traynor]

On 08/18/2017 11:45 AM, Yuanhan Liu wrote:
> Hi all,
> 
> Here is a list of patches targeted for LTS release 16.11.3. Please
> help review and test. The planned date for the final release is 30th,
> Aug. Before that, please let me know if anyone has objections with
> these patches being applied.
> 
> These patches are located at branch 16.11 of dpdk-stable repo:
>     http://dpdk.org/browse/dpdk-stable/
> 
> Thanks.
> 
>         --yliu
> 

Hi Yuanhan,

I think this one should go in too, as OVS hits this and writes back the
wrong watermark value to the shared register which can cause problems
for other ports. I've applied and tested it with DPDK 16.11.

commit 0e61ab56e01655f02bfe4e6249e032e864b0f5dd
Author: Qi Zhang <qi.z.zhang@intel.com>
Date:   Thu Aug 10 18:48:07 2017 +0800

    net/i40e: fix flow control watermark mismatch

    Flow control watermark is not read out correctly,
    that may cause an application who not intend to change
    watermark but does change it with a rte_eth_dev_flow_ctrl_set
    call right after rte_eth_dev_flow_ctrl_get.


thanks,
Kevin.

> ---
> Aaron Conole (1):
>       net/enic: fix crash when freeing 0 packet to mempool
> 
> Ajit Khaparde (9):
>       net/bnxt: fix reporting of link status
>       net/bnxt: fix get link config
>       net/bnxt: fix autoneg on 10GBase-T links
>       net/bnxt: fix set link config
>       net/bnxt: fix set link config
>       net/bnxt: check invalid L2 filter id
>       net/bnxt: free filter before reusing it
>       net/bnxt: enable default VNIC allocation
>       net/bnxt: fix vnic cleanup
> 
> Allain Legacy (1):
>       net/virtio-user: fix crash when detaching device
> 
> Arek Kusztal (1):
>       crypto/qat: fix NULL authentication hang
> 
> Beilei Xing (1):
>       net/i40e: fix ethertype filter for new FW
> 
> Chas Williams (1):
>       net/vmxnet3: fix filtering on promiscuous disabling
> 
> Cian Ferriter (1):
>       mbuf: fix VXLAN port in comment
> 
> Dahir Osman (1):
>       ip_frag: free mbufs on reassembly table destroy
> 
> Daniel Mrzyglod (2):
>       test/bonding: fix mode 4 names
>       test/bonding: fix namespace of the RSS tests
> 
> Dariusz Stojaczyk (1):
>       vhost: fix guest pages memory leak
> 
> Declan Doherty (2):
>       net/bonding: change link status check to no-wait
>       net/bonding: fix number of bonding Tx/Rx queues
> 
> Ferruh Yigit (3):
>       kni: fix build with gcc 7.1
>       net/enic: fix build with gcc 7.1
>       net/mlx5: fix build with gcc 7.1
> 
> Gowrishankar Muthukrishnan (1):
>       net/mlx5: fix inconsistent link status query
> 
> Hemant Agrawal (1):
>       examples/qos_sched: fix build for less lcores
> 
> Herakliusz Lipiec (1):
>       net/i40e: add return value checks
> 
> Herbert Guan (2):
>       test/bonding: fix memory corruptions
>       test/bonding: fix parameters of a balance Tx
> 
> Ivan Dyukov (1):
>       vhost: fix checking of device features
> 
> Jamie Lavigne (1):
>       mem: fix malloc element resize with padding
> 
> Jeff Guo (1):
>       net/i40e: fix link down and negotiation
> 
> Jianfeng Tan (3):
>       eal: fix config file path when checking process
>       vhost: fix TCP checksum
>       vhost: fix IP checksum
> 
> Jingjing Wu (1):
>       net/e1000: fix LSC interrupt
> 
> Kirill Rybalchenko (1):
>       cryptodev: fix device stop function
> 
> Mandeep Rohilla (1):
>       net/vmxnet3: fix receive queue memory leak
> 
> Markus Theil (1):
>       net/igb: fix add/delete of flex filters
> 
> Matan Azrad (1):
>       net/mlx4: fix probe failure report
> 
> Michal Krawczyk (1):
>       net/ena: fix cleanup of the Tx bufs
> 
> Pablo de Lara (12):
>       examples/l2fwd-crypto: fix option parsing
>       examples/l2fwd-crypto: fix application help
>       test/crypto: fix overflow
>       examples/l2fwd-crypto: fix padding
>       crypto/aesni_mb: fix HMAC supported key sizes
>       crypto/openssl: fix HMAC supported key sizes
>       mbuf: fix doxygen comment of bulk alloc
>       crypto/qat: fix HMAC supported key sizes
>       crypto/qat: fix SHA384-HMAC block size
>       doc: remove incorrect limitation on AESNI-MB PMD
>       doc: remove incorrect limitation on QAT PMD
>       test/crypto: fix wrong AAD setting
> 
> Qi Zhang (5):
>       net/i40e: exclude internal packet's byte count
>       net/i40e: fix VF statistics
>       net/ixgbe: fix mirror rule index overflow
>       net/i40e: fix Rx data segment buffer length
>       examples/l3fwd: fix IPv6 packet type parse
> 
> Qiming Yang (1):
>       net/ixgbe: fix Rx/Tx queue interrupt for x550 devices
> 
> Rahul Lakkireddy (1):
>       net/cxgbe: fix port statistics
> 
> Rasesh Mody (1):
>       net/qede: fix chip details print
> 
> Sha Zhang (1):
>       net/bonding: fix when NTT flag updated
> 
> Shahaf Shuler (1):
>       net/mlx5: fix completion buffer size
> 
> Stephen Hemminger (2):
>       net/virtio: do not falsely claim to do IP checksum
>       net/virtio: do not claim to support LRO
> 
> Tiwei Bie (4):
>       net/virtio: zero the whole memory zone
>       contigmem: free allocated memory on error
>       contigmem: do not zero pages during each mmap
>       vhost: make page logging atomic
> 
> Vasily Philipov (2):
>       mbuf: fix debug checks for headroom and tailroom
>       net/mlx4: fix mbuf poisoning in debug code
> 
> Wei Dai (1):
>       lpm: fix index of tbl8
> 
> Wei Zhao (2):
>       net/igb: fix checksum valid flags
>       net/igb: fix flex filter length
> 
> Wenzhuo Lu (1):
>       net/i40e/base: fix Tx error stats on VF
> 
> Xiao Wang (1):
>       net/fm10k: initialize link status in device start
> 
> Yong Wang (1):
>       net/i40e: fix division by 0
> 
> Yongseok Koh (2):
>       net/mlx5: fix exception handling
>       net/mlx5: fix redundant free of Tx buffer
> 

Re: [dpdk-stable] 16.11.3 (LTS) patches review and test

[Yuanhan Liu]

On Fri, Aug 18, 2017 at 06:29:02PM +0100, Kevin Traynor wrote:
> On 08/18/2017 11:45 AM, Yuanhan Liu wrote:
> > Hi all,
> > 
> > Here is a list of patches targeted for LTS release 16.11.3. Please
> > help review and test. The planned date for the final release is 30th,
> > Aug. Before that, please let me know if anyone has objections with
> > these patches being applied.
> > 
> > These patches are located at branch 16.11 of dpdk-stable repo:
> >     http://dpdk.org/browse/dpdk-stable/
> > 
> > Thanks.
> > 
> >         --yliu
> > 
> 
> Hi Yuanhan,
> 
> I think this one should go in too, as OVS hits this and writes back the
> wrong watermark value to the shared register which can cause problems
> for other ports. I've applied and tested it with DPDK 16.11.
> 
> commit 0e61ab56e01655f02bfe4e6249e032e864b0f5dd
> Author: Qi Zhang <qi.z.zhang@intel.com>
> Date:   Thu Aug 10 18:48:07 2017 +0800
> 
>     net/i40e: fix flow control watermark mismatch
> 
>     Flow control watermark is not read out correctly,
>     that may cause an application who not intend to change
>     watermark but does change it with a rte_eth_dev_flow_ctrl_set
>     call right after rte_eth_dev_flow_ctrl_get.

Weird, I couldn't find this commit:

    $ git show 0e61ab56e01655f02bfe4e6249e032e864b0f5dd
    fatal: bad object 0e61ab56e01655f02bfe4e6249e032e864b0f5dd

	--yliu

Re: [dpdk-stable] 16.11.3 (LTS) patches review and test

[Ferruh Yigit]

On 8/21/2017 4:40 AM, Yuanhan Liu wrote:
> On Fri, Aug 18, 2017 at 06:29:02PM +0100, Kevin Traynor wrote:
>> On 08/18/2017 11:45 AM, Yuanhan Liu wrote:
>>> Hi all,
>>>
>>> Here is a list of patches targeted for LTS release 16.11.3. Please
>>> help review and test. The planned date for the final release is 30th,
>>> Aug. Before that, please let me know if anyone has objections with
>>> these patches being applied.
>>>
>>> These patches are located at branch 16.11 of dpdk-stable repo:
>>>     http://dpdk.org/browse/dpdk-stable/
>>>
>>> Thanks.
>>>
>>>         --yliu
>>>
>>
>> Hi Yuanhan,
>>
>> I think this one should go in too, as OVS hits this and writes back the
>> wrong watermark value to the shared register which can cause problems
>> for other ports. I've applied and tested it with DPDK 16.11.
>>
>> commit 0e61ab56e01655f02bfe4e6249e032e864b0f5dd
>> Author: Qi Zhang <qi.z.zhang@intel.com>
>> Date:   Thu Aug 10 18:48:07 2017 +0800
>>
>>     net/i40e: fix flow control watermark mismatch
>>
>>     Flow control watermark is not read out correctly,
>>     that may cause an application who not intend to change
>>     watermark but does change it with a rte_eth_dev_flow_ctrl_set
>>     call right after rte_eth_dev_flow_ctrl_get.
> 
> Weird, I couldn't find this commit:
> 
>     $ git show 0e61ab56e01655f02bfe4e6249e032e864b0f5dd
>     fatal: bad object 0e61ab56e01655f02bfe4e6249e032e864b0f5dd

The commit is in next-net. Not sure how to proceed, any idea?

> 
> 	--yliu
> 

Re: [dpdk-stable] 16.11.3 (LTS) patches review and test

[Yuanhan Liu]

On Mon, Aug 21, 2017 at 09:23:24AM +0100, Ferruh Yigit wrote:
> >> I think this one should go in too, as OVS hits this and writes back the
> >> wrong watermark value to the shared register which can cause problems
> >> for other ports. I've applied and tested it with DPDK 16.11.
> >>
> >> commit 0e61ab56e01655f02bfe4e6249e032e864b0f5dd
> >> Author: Qi Zhang <qi.z.zhang@intel.com>
> >> Date:   Thu Aug 10 18:48:07 2017 +0800
> >>
> >>     net/i40e: fix flow control watermark mismatch
> >>
> >>     Flow control watermark is not read out correctly,
> >>     that may cause an application who not intend to change
> >>     watermark but does change it with a rte_eth_dev_flow_ctrl_set
> >>     call right after rte_eth_dev_flow_ctrl_get.
> > 
> > Weird, I couldn't find this commit:
> > 
> >     $ git show 0e61ab56e01655f02bfe4e6249e032e864b0f5dd
> >     fatal: bad object 0e61ab56e01655f02bfe4e6249e032e864b0f5dd
> 
> The commit is in next-net. Not sure how to proceed, any idea?

I see. Thanks. So it's not merged to Thomas tree yet. I normally take
commits from there. Because I add a tag like following every time I
pick a commit:

    [ upstream commit 5b9b65b14e05c106bb8229c0fe0b347315da7d00 ]

If I take commits directly from next-* tree, the commit could have
been changed while merging to Thomas tree, for two reasons:

- the next-* could have done a rebase
- we don't do "git merge" at Thomas tree


However, if that commit is urgent, I think I could make it for 16.11.3
release with following tag:

    [ next-net commit 0e61ab56e01655f02bfe4e6249e032e864b0f5dd ]

If not, I will postone it to 16.11.4.

	--yliu

Re: [dpdk-stable] 16.11.3 (LTS) patches review and test

[Kevin Traynor]

On 08/21/2017 10:23 AM, Yuanhan Liu wrote:
> On Mon, Aug 21, 2017 at 09:23:24AM +0100, Ferruh Yigit wrote:
>>>> I think this one should go in too, as OVS hits this and writes back the
>>>> wrong watermark value to the shared register which can cause problems
>>>> for other ports. I've applied and tested it with DPDK 16.11.
>>>>
>>>> commit 0e61ab56e01655f02bfe4e6249e032e864b0f5dd
>>>> Author: Qi Zhang <qi.z.zhang@intel.com>
>>>> Date:   Thu Aug 10 18:48:07 2017 +0800
>>>>
>>>>     net/i40e: fix flow control watermark mismatch
>>>>
>>>>     Flow control watermark is not read out correctly,
>>>>     that may cause an application who not intend to change
>>>>     watermark but does change it with a rte_eth_dev_flow_ctrl_set
>>>>     call right after rte_eth_dev_flow_ctrl_get.
>>>
>>> Weird, I couldn't find this commit:
>>>
>>>     $ git show 0e61ab56e01655f02bfe4e6249e032e864b0f5dd
>>>     fatal: bad object 0e61ab56e01655f02bfe4e6249e032e864b0f5dd
>>
>> The commit is in next-net. Not sure how to proceed, any idea?
> 
> I see. Thanks. So it's not merged to Thomas tree yet. I normally take
> commits from there. Because I add a tag like following every time I
> pick a commit:
> 
>     [ upstream commit 5b9b65b14e05c106bb8229c0fe0b347315da7d00 ]
>

oops, I didn't think about that :/

> If I take commits directly from next-* tree, the commit could have
> been changed while merging to Thomas tree, for two reasons:
> 
> - the next-* could have done a rebase
> - we don't do "git merge" at Thomas tree
> 
> 
> However, if that commit is urgent, I think I could make it for 16.11.3
> release with following tag:
> 
>     [ next-net commit 0e61ab56e01655f02bfe4e6249e032e864b0f5dd ]
> 
> If not, I will postone it to 16.11.4.
> 
> 	--yliu
> 

Given that it took 9 months for the bug to be found, I think postponing
to DPDK 16.11.4 is better than upsetting the dpdk-stable commit log to
reference multiple repos.

thanks,
Kevin.