UPGRADE YOUR BROWSER

We have detected your current browser version is not the latest one. Xilinx.com uses the latest web technologies to bring you the best online experience possible. Please upgrade to a Xilinx.com supported browser:Chrome, Firefox, Internet Explorer 11, Safari. Thank you!

cancel
Showing results for 
Search instead for 
Did you mean: 
Observer ajroth
Observer
8,466 Views
Registered: ‎04-14-2015

duplicative TX pkts out of VC707 TEMAC to 1000BASE-X PCS/PMA or SGMII v14.2 setup - PCS/PMA

Jump to solution

The mentioned reference design http://www.xilinx.com/support/answers/59968.html after many mods has ended up getting me further than prior efforts following documentation,etc. for setting up SGMII for VC707 using Vivado 2014.2.

 

However I get repeat tx packets.  For every tx pkt I look at in wireshark there is a duplicative packet with the only difference is an added leading x3c x00 xc3 xff  - 4 leading bytes before the packet thus shifting everything.  Anyone know what these are?

 

Everything else looks great, I throw in an ILA all the way to the ports of the tx data into the trancsiever core right before the GTX then sends out the txn and txp serial data, and I don't ever see these leading bytes coming from the TEMAC or higher up in the PCS/PMA core to be sent out the transciever, yet wireshark always sees 2 packets, 1 good, 1 with the 4 leading crap bytes?

 

Any ideas why the transciever would be duplicating packets and throwing in these leading bytes?

 

THANKS!

 

(I posted this also on my old thread http://forums.xilinx.com/t5/Connectivity/VC707-TEMAC-to-1000BASE-X-PCS-PMA-or-SGMII-v14-2-setup-PCS-PMA/m-p/593080#M7869)

0 Kudos
1 Solution

Accepted Solutions
Highlighted
Observer ajroth
Observer
15,183 Views
Registered: ‎04-14-2015

Re: duplicative TX pkts out of VC707 TEMAC to 1000BASE-X PCS/PMA or SGMII v14.2 setup - PCS/PMA

Jump to solution

ISSUE RESOLVED!

 

RESOLUTION

 

ISSUE WASN'T WITH PCS/PMA CORE OR ANYTHING IN MY DESIGN - THE TEST SETUP WAS USING A Gigabit USB 2.0 Network Adapter that we assumed worked fine.  The adapter was causing the issue!!  Plugging directly into another test laoptop ethernet port showed no duplicate packets, with USB adapter on that laptop duplication occured, tried identical BELKIN Gigabit USB 2.0 network adapter and it had the same issue.  Possible issue with the linux drivers / NIC.. who knows...

 

THANKS / SORRY!

0 Kudos
6 Replies
Xilinx Employee
Xilinx Employee
8,453 Views
Registered: ‎02-06-2013

Re: duplicative TX pkts out of VC707 TEMAC to 1000BASE-X PCS/PMA or SGMII v14.2 setup - PCS/PMA

Jump to solution

Hi

 

There won't be any logic in the transceiver to duplicate the packets unless it is driven from the parall tx data input.

 

Can you attach the ILA data at the TX data input of transceiver that shows the packets and wireshark caputures.

 

Also confirm if you loopback the PHY or set the Transceiver in PMA loopback,do you still see the duplicate packets in the ILA capture in the RX Path.

Regards,

Satish

--------------------------------------------------​--------------------------------------------
Kindly note- Please mark the Answer as "Accept as solution" if information provided is helpful.

Give Kudos to a post which you think is helpful.
--------------------------------------------------​-------------------------------------------
0 Kudos
Observer ajroth
Observer
8,438 Views
Registered: ‎04-14-2015

Re: duplicative TX pkts out of VC707 TEMAC to 1000BASE-X PCS/PMA or SGMII v14.2 setup - PCS/PMA

Jump to solution
Thanks Satish - I'll try those things, I noticed today It looks like the leading 8 bytes on the duplicated packet are actual total length of packets --- for the ARP packets I get x3c00c3ff which first 4 bytes = last 4 bytes inverted, then swap first 2 bytes = x003c = 60 bytes which is pkt length (without these leading bytes). Then I have test packets with x3204cdfb which is x0432 = 1074 = total length of the test tx packet minus the leading 4 bytes again.... so for some reason the duplicated packets include the packet length appended to the front in a weird manner. I'll post wireshark captures and/or ILA screen shots sometime soon if needed.. thanks!
0 Kudos
Observer ajroth
Observer
8,432 Views
Registered: ‎04-14-2015

Re: duplicative TX pkts out of VC707 TEMAC to 1000BASE-X PCS/PMA or SGMII v14.2 setup - PCS/PMA

Jump to solution

Attached is a wire shark capture -

 

Setup is fpga ip 10.11.12.13 with src MAC 5A0102030405 connected to test pc is 10.11.12.2 with MAC 002275d70344 from a belkin USB to ethernet adaptor (this couldn't duplicat pkts, and don't think the test pc NIC, or wireshark would duplicate pkts and add these 8 bytes..?)

 

ignore the system sent micromuse-lm packets, filter with !(udp.dstport==1534)

 

What you'll see in wireshark is arp requests from fpga that resolves, followed by a lengthy number of UDP pkts with udp data field size 1032 bytes=8256bits,  udp data section has [8bytes=64bits udp pkt stream sequence counter for burst |  1024Bytes PRBS data]

You'll see for most every tx packet from the fpga 10.11.12.13 there is an immediate following exact same tx pkt with the 8 bytes that relate to size added to the front as explained.  It's verified these pkts are identical minus the leading 8 bytes.

 

 

probes are set within gig_ethernet_pcs_pma_0_transceiver.vhd of the pcs/pma xilinx IP core, tx data between pcs/pma main core and it's gig_ethernet_pcs_pma_0_GTWIZARD block.  I may probe down within the GTWIZARD block but I don't think that could duplicate pkts or do this for tx pkts, within the GTWIZARD block does it buffer up the pkts because first look it just appeared to pass tx data straight down to the gtxe2_i : GTXE2_CHANNEL instantiation ...

 

  ila_pcs_pma_transceiver_125 : ila_1p256w4096d_probe
  port map (
    clk                   => usrclk2,
    probe0(15 downto 0)   => txdata_double,
    probe0(23 downto 16)  => txdata_reg,
    probe0(31 downto 24)  => txdata,
    probe0(255 downto 32) => (others => '0')
  );
 
  ila_pcs_pma_transceiver_64_5 : ila_1p256w4096d_probe
  port map (
    clk  => usrclk,
    probe0(15 downto 0)  => txdata_int,
    probe0(255 downto 16) => (others => '0')
  );

 

I try to trigger on the duplicated data header bytes but they never appear..

0 Kudos
Observer ajroth
Observer
8,426 Views
Registered: ‎04-14-2015

Re: duplicative TX pkts out of VC707 TEMAC to 1000BASE-X PCS/PMA or SGMII v14.2 setup - PCS/PMA

Jump to solution
note- when viewing the ILA - it triggers on x5A, 10 in a row, but some are just the PRBS data, you can see the initial pkts are those that follow the idle x50bc sequence, and then you can zoom in see its the 5A0102030405 MAC, and look further in time on that pkt you can see we don't ever get duplicate pkts at this test point like we see in wireshark. Looking at the first 8 bytes of data field in these you will be able to follow see the sequence counter increments by 1 each new pkt. In the attached ila waveform I'm not sure if you see but the pkts begin after there's a good tx arp return from FPGA to testpc..
0 Kudos
Observer ajroth
Observer
8,422 Views
Registered: ‎04-14-2015

Re: duplicative TX pkts out of VC707 TEMAC to 1000BASE-X PCS/PMA or SGMII v14.2 setup - PCS/PMA

Jump to solution

attached is in loopback mode - don't see duplicates coming out...

goint to change my test pc and wireshark etc setup to confirm its then for sure in the GT_WIZARD block down to the GTXE2_CHANNEL and/or physical PHY ... which not sure how those blocks/hw could duplicate a whole packet and buffer them up??  I thought it was just down to transmitting bits down there... but the PCS PMA loopback confirmed not dups.. Maybe the test setup..tbd.

loopback note pcs/pma PG047 for v14.2 - pg 269-270 - loopback before GTX in pcspma core

0 Kudos
Highlighted
Observer ajroth
Observer
15,184 Views
Registered: ‎04-14-2015

Re: duplicative TX pkts out of VC707 TEMAC to 1000BASE-X PCS/PMA or SGMII v14.2 setup - PCS/PMA

Jump to solution

ISSUE RESOLVED!

 

RESOLUTION

 

ISSUE WASN'T WITH PCS/PMA CORE OR ANYTHING IN MY DESIGN - THE TEST SETUP WAS USING A Gigabit USB 2.0 Network Adapter that we assumed worked fine.  The adapter was causing the issue!!  Plugging directly into another test laoptop ethernet port showed no duplicate packets, with USB adapter on that laptop duplication occured, tried identical BELKIN Gigabit USB 2.0 network adapter and it had the same issue.  Possible issue with the linux drivers / NIC.. who knows...

 

THANKS / SORRY!

0 Kudos