Hey Carolyn:
exactly, i do *not* ever get a real timestamp valid condition, I
can't imagine why.
:-) thanks for your respondence.
-----邮件原件-----
发件人: Wyborny, Carolyn [mailto:[email protected]]
发送时间: 2010年10月14日 23:25
收件人: 李昂; [email protected]
抄送: '王龙'; '和志华'
主题: RE: [E1000-devel] question about TX hardware timestamping, it doesn't
work
Hello,
I see, I misunderstood your question. 8-) To confirm: you do *not* ever
get a timestamp valid condition, so you are always creating one with your
sync code?
I will continue investigating the issue.
Thanks,
Carolyn
>-----Original Message-----
>From: 李昂 [mailto:[email protected]]
>Sent: Friday, October 08, 2010 6:29 PM
>To: Wyborny, Carolyn; [email protected]
>Cc: '王龙'; '和志华'
>Subject: 答复: [E1000-devel] question about TX hardware
>timestamping, it doesn't work
>
>Hello Carolyn:
> I'm sure that the timestamp is appearing on a packet outside the
>transmitting system, when PHY timestamping the packet, the tsbits is
>asserted, driver should check it in the interrupt routine, the
>timestamp of
>outside packet should be pushed in the TXSTMPL & TXSTMPH
>registers(lower &
>upper).
> The description information of bit0 in the TSYNCTXCTL
>register could
>be found in 82574gbe datasheet, 10.2.9.8 on page 369:
> " TXTT -- Tx time stamp valid. Equals 1b when a
>valid value
>for Tx timestamp is captured in the Tx time stamp register.
>Cleared by read
>of Tx time stamp register TXSTMPH."
>
> Thanks for your respondence
>
>>
>>Hello,
>>
>>I've briefly reviewed your code. When you check for the tsbits in the
>transmitted packet, are you checking from a network sniffer,
>on the wire,
>or from a system receiving the transmitted packets? I am not
>sure that the
>bits can >be read from the driver on the system doing the transmit. It
>would be more accurate to check the packet on the wire or received on
>another system, similar to checksum.
>>
>>I will continue reviewing your code. Please confirm if you
>the timestamp
>is appearing on a packet outside the transmitting system.
>>
>>Thanks,
>>
>>Carolyn
>>
>-----邮件原件-----
>发件人: Wyborny, Carolyn [mailto:[email protected]]
>发送时间: 2010年10月9日 1:41
>收件人: 李昂; [email protected]
>抄送: 王龙; 和志华
>主题: RE: [E1000-devel] question about TX hardware timestamping,
>it doesn't
>work
>
>
>
>>-----Original Message-----
>>From: 李昂 [mailto:[email protected]]
>>Sent: Tuesday, September 28, 2010 11:20 PM
>>To: Wyborny, Carolyn; [email protected]
>>Cc: 王龙; 和志华
>>Subject: re: [E1000-devel] question about TX hardware
>>timestamping, it doesn't work
>>
>>Hey Carolyn:
>>
>> thanks for your respondence, the attachment is the
>>package of e1000e
>>driver(I have modified netdev.c), searching "li.ang" in the
>>netdev.c file,
>>you will find modification, it including initialization code &
>>registers I'm
>>writing to enable the hwtstamp feature.
>>
>>
>>
>>----- Original Message -----
>>From: Wyborny, Carolyn [mailto:[email protected]]
>>Sent: 2010年9月28日 1:35
>>To:??; [email protected]
>>Cc: ??; ???
>>Subject: RE: [E1000-devel] question about TX hardware timestamping, it
>>doesn't work
>>
>>Hello,
>>
>>I apologize for the delay in responding. We do not officially
>>support the
>>time stamp feature on the 82574 device but we do know that people have
>>gotten it working. If the fields in the descriptor are zero,
>>its likely its
>>not being intialized correctly. I am not an expert on this
>>feature, but
>>I'll need some more information to try and assist.
>>
>>If you open a bug on Source Forge on this issue, we can store the data
>>gathered there.
>>
>>Can you send me your initialization code, or at least the
>>registers you are
>>writing to enable the feature and the values you are writing
>>to them. Also,
>>the exact fields you are writing in the tx descriptor and the
>values.
>>
>>Thanks,
>>
>>Carolyn
>>Carolyn Wyborny
>>Linux Development
>>LAN Access Division
>>Intel Corporation
>>
>>
>>
>>>-----Original Message-----
>>>From: ?? [mailto:[email protected]]
>>>Sent: Sunday, September 19, 2010 2:04 AM
>>>To: [email protected]
>>>Cc: ??; ???
>>>Subject: [E1000-devel] question about TX hardware
>>>timestamping, it doesn't work
>>>
>>>Dear All:
>>>
>>>
>>>
>>> On 82574L Chipset, e1000e work well when I need RX
>>>hwtimestamp, but
>>>it doesn't work when I need TX hwtimestamp
>>>
>>>
>>>
>>> Legacy Transmit Descriptor Format is used, and I set
>>>the TS bit of
>>>ExtCMD to indicates to the 82574 to put a time stamp on the packet
>>>designated by the descriptor, of course, I also set the bit4
>>>of TSYNCTXCTL
>>>to enable TX timestamp.
>>>
>>>
>>>
>>> But, it doesn't work, when I read bit0 of TSYNCTXCTL(TX
>>>timestamp
>>>valid bit) in the TX interrupt service routine(packet has been
>>>transmitted),
>>>I got zero, and TXSTMPL/H are also filled zero.
>>>
>>>
>>>
>>> Any advise to solve this problem is welcome
>>>
>>>
>>>
>>> [email protected] li.ang
>>>
>>>
>>>
>>> Thanks a lot
>>>
>>>
>>
>
>
------------------------------------------------------------------------------
Download new Adobe(R) Flash(R) Builder(TM) 4
The new Adobe(R) Flex(R) 4 and Flash(R) Builder(TM) 4 (formerly
Flex(R) Builder(TM)) enable the development of rich applications that run
across multiple browsers and platforms. Download your free trials today!
http://p.sf.net/sfu/adobe-dev2dev
_______________________________________________
E1000-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel® Ethernet, visit
http://communities.intel.com/community/wired
