On Mon, Feb 17, 2020 at 04:17:53PM +0000, Ferruh Yigit wrote:
> On 2/17/2020 3:30 PM, Harman Kalra wrote:
> > On Mon, Feb 17, 2020 at 03:19:53PM +0000, Ferruh Yigit wrote:
> >> On 2/17/2020 3:06 PM, Harman Kalra wrote:
> >>> On Mon, Feb 17, 2020 at 02:54:13PM +0000, Ferruh Yigit wrote:
> >>>> On 2/17/2020 2:35 PM, Harman Kalra wrote:
> >>>>> On Mon, Feb 17, 2020 at 01:21:13PM +0000, Ferruh Yigit wrote:
> >>>>>> External Email
> >>>>>>
> >>>>>> ----------------------------------------------------------------------
> >>>>>> On 2/15/2020 8:35 AM, Harman Kalra wrote:
> >>>>>>> PTP functionality has been broken after a change in kernel
> >>>>>>> where enum npc_kpu_lc_ltype is change to allow adjustment of
> >>>>>>> LTYPE_MASK to detect all types of IP headers.
> >>>>>>> Syncing the required changes in DPDK to fix the issue.
> >>>>>>
> >>>>>> Hi Harman,
> >>>>>>
> >>>>>> Can you please provide references for,
> >>>>>> - What is the kernel commit causing the mentioned change?
> >>>>>> - After this change which version of the kernel supported by the PMD?
> >>>>>
> >>>>> Hi Ferruh
> >>>>> The kernel change I was refering to was in our internel kernel repo
> >>>>> and those kernel changes have not been upstreamed yet. Currently its 
> >>>>> part
> >>>>> of our marvell SDK version SDK-10.3.1.x.
> >>>>
> >>>> Got it, what is the dependency of the PMD to the Linux kernel? And does 
> >>>> the PMD
> >>>> works with upstream kernels?
> >>>
> >>> Yes, for basic functionality PMD works fine with upstreamed kernel but for
> >>> supporting some advanced features like PTP, kernel changes are yet to be
> >>> upstreamed. Our kernel driver for octeontx2 PMD:
> >>> https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_torvalds_linux_tree_master_drivers_net_ethernet_marvell_octeontx2_af&d=DwIDaQ&c=nKjWec2b6R0mOyPaz7xtfQ&r=5ESHPj7V-7JdkxT_Z_SU6RrS37ys4UXudBQ_rrS5LRo&m=pXArrQOu5hyDcQfkwc34h1CkBsAQuEnZDcanbaM-GCQ&s=c7WM8a2aLkuXH5P5HpQofM1nmCAXufFYva5YVyEiYcM&e=
> >>>  
> >>> Kernel upstreaming activity is not in same pace as DPDK.
> >>
> >> Is the marvell SDK code in a public repo? If so can you provide the 
> >> details of
> >> the kernel changes in that repo? It is good to have some reference on the 
> >> matter.
> > 
> > Marvel SDK is not in any public repo. But for reference internal kernel viz 
> > part of
> > SDK version SDK-10.3.1.x with commit id 
> > "08705a860e2d3aebce44d6c357ed1c0ae71d3fd9"
> > onwards will have the required changes.
> 
> If it is not public, the commit id won't help much. Would you be OK if I add
> something like following while merging:
> 
> The kernel side changes are exists in marvell SDK version SDK-10.3.1.x.

I am sorry, one correction kernel side changes exists in SDK version 
SDK-10.3.2.x
and not in SDK-10.3.1.x
Please feel free to mention SDK version SDK-10.3.2.x

> 
> >>
> >>>
> >>> Thanks
> >>> Harman
> >>>>
> >>>>>
> >>>>> Thanks
> >>>>> Harman
> >>>>>
> >>>>>>
> >>>>>>>
> >>>>>>> Fixes: b5dc3140448e ("net/octeontx2: support base PTP")
> >>>>>>> Cc: sta...@dpdk.org
> >>>>>>>
> >>>>>>> Signed-off-by: Harman Kalra <hka...@marvell.com>
> >>>>>>> ---
> >>>>>>>  drivers/common/octeontx2/hw/otx2_npc.h | 4 ++--
> >>>>>>>  1 file changed, 2 insertions(+), 2 deletions(-)
> >>>>>>>
> >>>>>>> diff --git a/drivers/common/octeontx2/hw/otx2_npc.h 
> >>>>>>> b/drivers/common/octeontx2/hw/otx2_npc.h
> >>>>>>> index a0536e0ae..3dfc137a3 100644
> >>>>>>> --- a/drivers/common/octeontx2/hw/otx2_npc.h
> >>>>>>> +++ b/drivers/common/octeontx2/hw/otx2_npc.h
> >>>>>>> @@ -201,7 +201,8 @@ enum npc_kpu_lb_ltype {
> >>>>>>>  };
> >>>>>>>  
> >>>>>>>  enum npc_kpu_lc_ltype {
> >>>>>>> -     NPC_LT_LC_IP = 1,
> >>>>>>> +     NPC_LT_LC_PTP = 1,
> >>>>>>> +     NPC_LT_LC_IP,
> >>>>>>>       NPC_LT_LC_IP_OPT,
> >>>>>>>       NPC_LT_LC_IP6,
> >>>>>>>       NPC_LT_LC_IP6_EXT,
> >>>>>>> @@ -209,7 +210,6 @@ enum npc_kpu_lc_ltype {
> >>>>>>>       NPC_LT_LC_RARP,
> >>>>>>>       NPC_LT_LC_MPLS,
> >>>>>>>       NPC_LT_LC_NSH,
> >>>>>>> -     NPC_LT_LC_PTP,
> >>>>>>>       NPC_LT_LC_FCOE,
> >>>>>>>  };
> >>>>>>>  
> >>>>>>>
> >>>>>>
> >>>>
> >>
> 

Reply via email to