Hi Ted,
I was going through the function e1000g_receive() in
e1000g_rx.c, and I noticed that there is a variable 'pkt_count', and it
looks like that it is being used to count the number of packets for every
interrupt. Can you please confirm this. The thing is that I am doing some
very controlled experiments using UDP, wherein a single message will fit
into a single UDP packet. If you think that pkt_count does hold the number
of packets per interrupt, then can you please suggest a way to capture this
information as well.
Thank you,
Vishal
On Tue, Sep 8, 2009 at 7:16 PM, Ted You <[email protected]> wrote:
> Hi Vhiz,
>
> You can check the source file of uts/common/io/e1000g/e1000g_main.c,
> search the variables of intr_throttling_rate, rx_intr_delay, and
> rx_intr_abs_delay. However, in the e1000g driver, the code just provides
> a method for users to configure those parameters. It does not explicitly
> explain how they affect the rx interrupt. The Intel document AP450 has
> more details about it.
>
> And you can also check the file uts/common/io/e1000g/e1000g_rx.c. The
> function e1000g_receive() in e1000g_rx.c, and the functions
> e1000g_intr_work() / e1000g_poll_ring() in e1000g_main.c are the major
> packet receiving routines of the driver.
>
> Hope this helps.
>
> Regards,
> Ted
>
>
>
> Thank you for that! I had a look at the document, but
>>> I could not find anything useful for my problem. Will
>>> it help if I ponder at the driver code? If yes, can
>>> someone please point out any specific parts which I
>>> should look at.
>>>
>>> Thanks.
>>>
>>
>> Can someone please help here..
>>
>
_______________________________________________
driver-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/driver-discuss
