NACK: [SRU][F:linux-bluefield][PATCH 0/8] Fix fragmentation support for TC connection tracking

Tim Gardner tim.gardner at canonical.com
Mon Aug 30 11:43:22 UTC 2021


So, you're going to send a v2 of this patch set with correct 
'backported' descriptions in patches [1] and [2], right ?

rtg

On 8/30/21 5:15 AM, Maor Dickman wrote:
> Linux-bluefield branch contains two patches [1][2] which cause a conflict when I
> tried to cherry-pick the patch and forced me to do small modification.
> 
> Patch [1] was submitted to upstream after this patch ("Fix fragmentation support
> for TC connection tracking")
> 
> Patch [2] is wrong so I revert it in later patch.
>   
> [1] 7a73ed569d90 net/sched: act_ct: Fix skb double-free in tcf_ct_handle_fragments() error flow
> [2] f797fab7280c net/sched: act_ct: Fix skb double-free in tcf_ct_handle_fragments() error flow
> 
>> -----Original Message-----
>> From: Tim Gardner <tim.gardner at canonical.com>
>> Sent: Tuesday, August 24, 2021 2:47 PM
>> To: Bodong Wang <bodong at nvidia.com>; kernel-team at lists.ubuntu.com
>> Cc: Maor Dickman <maord at nvidia.com>; Majd Dibbiny <majd at nvidia.com>
>> Subject: NACK: [SRU][F:linux-bluefield][PATCH 0/8] Fix fragmentation
>> support for TC connection tracking
>>
>> External email: Use caution opening links or attachments
>>
>>
>> While these patches do apply correctly, PATCH 1/8 is not a cherry-pick.
>> Even if you did the revert first ("Revert "net/sched: act_ct: Fix skb double-
>> free in tcf_ct_handle_fragments() error flow"), its still not a clean cherry-pick
>> though it is a simpler backport.
>>
>> rtg
>>
>> On 8/23/21 4:36 PM, Bodong Wang wrote:
>>> When using OVS with tc to offload connection tracking flows, sending
>>> udp/icmp fragmented traffic will cause call trace with NULL dereference.
>>>
>>> This series contains 7 patches from upstream which fix act_ct handling
>>> of fragmented packets. And revert a patch which is covered by the 7
>>> upstream patches.
>>>
>>> Davide Caratti (1):
>>>     net/sched: act_ct: fix wild memory access when clearing fragments
>>>
>>> Maor Dickman (1):
>>>     Revert "net/sched: act_ct: Fix skb double-free in
>>>       tcf_ct_handle_fragments() error flow"
>>>
>>> wenxu (6):
>>>     net/sched: act_ct: fix restore the qdisc_skb_cb after defrag
>>>     net/sched: act_ct: fix miss set mru for ovs after defrag in act_ct
>>>     net/sched: fix miss init the mru in qdisc_skb_cb
>>>     net/sched: act_mirred: refactor the handle of xmit
>>>     net/sched: sch_frag: add generic packet fragment support.
>>>     ipv6: add ipv6_fragment hook in ipv6_stub
>>>
>>>    include/linux/skbuff.h    |   1 +
>>>    include/net/act_api.h     |   6 ++
>>>    include/net/ipv6_stubs.h  |   2 +
>>>    include/net/sch_generic.h |   8 +--
>>>    net/core/dev.c            |   2 +
>>>    net/ipv6/addrconf_core.c  |   8 +++
>>>    net/ipv6/af_inet6.c       |   1 +
>>>    net/openvswitch/flow.c    |   1 +
>>>    net/sched/Makefile        |   1 +
>>>    net/sched/act_api.c       |  16 +++++
>>>    net/sched/act_ct.c        |  29 +++++++--
>>>    net/sched/act_mirred.c    |  21 +++++--
>>>    net/sched/cls_api.c       |   1 +
>>>    net/sched/sch_frag.c      | 150
>> ++++++++++++++++++++++++++++++++++++++++++++++
>>>    14 files changed, 231 insertions(+), 16 deletions(-)
>>>    create mode 100644 net/sched/sch_frag.c
>>>
>>
>> --
>> -----------
>> Tim Gardner
>> Canonical, Inc

-- 
-----------
Tim Gardner
Canonical, Inc



More information about the kernel-team mailing list