By delaying the delivery (that is, the indicating down) of a packet, you
can put guarantee that the overall rate of delivery to the NIC will no
greater than X. But you cannot be sure that it will be X or even know
with certainty that it will be, say, 90% of X, or 80%, or whatever. This
is so because there are one or more drivers beneath you, and how they
behave is hard to know, perhaps even essentially unpredictable.
In short, you can guarantee that the NIC is in the fastest case a 10M
NIC. But you’re not able to guarantee 10M, only a 10M ceiling.
–
If replying by e-mail, please remove “nospam.” from the address.
James Antognini
Windows DDK MVP
> ----------
From:
xxxxx@mindspring.nospam.com[SMTP:xxxxx@mindspring.nospam.com]
Reply To: xxxxx@lists.osr.com
Sent: Wednesday, July 23, 2003 6:55 AM
To: xxxxx@lists.osr.com
Subject: [ntdev] Re: How to control net card’s speed?
I do quote when I think the reference is uncertain.
Personally I often miss context of your responses and it seems I’m not the
only one. It’s a pity because they are usually valuable. It would be nice if
you can quote relevant parts of previous messages. Consider please there are
people like me who read list as bunch of separate mails once or twice daily
and don’t use mailer which is able to group messages to threads.
Best regards,
Michal Vodicka
STMicroelectronics Design and Application s.r.o.
[michal.vodicka@st.com, http:://www.st.com]
OK, you’re not the first to complain. I’m so used to my news reader, which
threads, that I have had a hard time in seeing the point of quoting. But I’ll
start doing so.
Michal Vodicka wrote:
Personally I often miss context of your responses and it seems I’m not the
only one. It’s a pity because they are usually valuable. It would be nice if
you can quote relevant parts of previous messages. Consider please there are
people like me who read list as bunch of separate mails once or twice daily
and don’t use mailer which is able to group messages to threads.
–
If replying by e-mail, please remove “nospam.” from the address.
James Antognini
Windows DDK MVP
Even if I query these packets, I have delayed them, right?
So I choose a simple way to handle them — just delay certain microsecond
to pass them, and it works well.
“James Antognini” ??? news:xxxxx@ntdev…
>
> By delaying the delivery (that is, the indicating down) of a packet, you
> can put guarantee that the overall rate of delivery to the NIC will no
> greater than X. But you cannot be sure that it will be X or even know
> with certainty that it will be, say, 90% of X, or 80%, or whatever. This
> is so because there are one or more drivers beneath you, and how they
> behave is hard to know, perhaps even essentially unpredictable.
>
> In short, you can guarantee that the NIC is in the fastest case a 10M
> NIC. But you’re not able to guarantee 10M, only a 10M ceiling.
>
> –
> If replying by e-mail, please remove “nospam.” from the address.
>
> James Antognini
> Windows DDK MVP
>
>
>
>
No, querying (inspecting) will delay at most only by a very small amount. The
passthru example is pure inspection, and the only delay is the cost of executing
passthru’s instructions and its API calls.
By delaying, I meant that in ProtocolReceive, for example, you do not indicate
up the packet – it’s unfortunate that the DDK uses “packet” when that really
means the descriptor of the packet payload that will or has gone over the
network – at that point, from that routine. Rather, you save necessary
information, by copying it to a convenient place, and return from the routine;
so far as NDIS is concerned, you have handled the packet, in that you did
something immediately or made it possible to do something later. Later, in some
other place, with the information you saved, you indicate the packet (or header
Anthony wrote:
Even if I query these packets, I have delayed them, right?
–
If replying by e-mail, please remove “nospam.” from the address.
James Antognini
Windows DDK MVP
> ----------
From:
xxxxx@mindspring.nospam.com[SMTP:xxxxx@mindspring.nospam.com]
Reply To: xxxxx@lists.osr.com
Sent: Thursday, July 24, 2003 1:02 AM
To: xxxxx@lists.osr.com
Subject: [ntdev] Re: How to control net card’s speed?
OK, you’re not the first to complain. I’m so used to my news reader, which
threads, that I have had a hard time in seeing the point of quoting. But
I’ll
start doing so.
Thanks. The positive side effect can be easier searching for future data
miners. Currently I often use google groups when have to do some w9x kernel
development and years old discussions are really helpful. Sometimes (not
unusual) google has problems with threading and it isn’t able to find other
messages in the thread. Quoting is invaluable there.
Best regards,
Michal Vodicka
STMicroelectronics Design and Application s.r.o.
[michal.vodicka@st.com, http:://www.st.com]
Hello James,
As I have said, I have done in my way successfully.
Yep, the system maybe have to re-send certain packets because of the
dalay or the resousce invalid. But it will not effect the communication at
all. Please make it clear, to slow down my netcard is my single aim, and I
pay no attention to so-called re-send or resource puzzles.
Anthony
“James Antognini” ??? news:xxxxx@ntdev…
>
> No, querying (inspecting) will delay at most only by a very small amount.
The
> passthru example is pure inspection, and the only delay is the cost of
executing
> passthru’s instructions and its API calls.
>
> By delaying, I meant that in ProtocolReceive, for example, you do not
indicate
> up the packet – it’s unfortunate that the DDK uses “packet” when that
really
> means the descriptor of the packet payload that will or has gone over the
> network – at that point, from that routine. Rather, you save necessary
> information, by copying it to a convenient place, and return from the
routine;
> so far as NDIS is concerned, you have handled the packet, in that you did
> something immediately or made it possible to do something later. Later, in
some
> other place, with the information you saved, you indicate the packet (or
header
> + lookahead) up.
>
> Anthony wrote:
>
> > Even if I query these packets, I have delayed them, right?
>
> –
> If replying by e-mail, please remove “nospam.” from the address.
>
> James Antognini
> Windows DDK MVP
>
>
>
>