On Oct 29, 2007, at 2:29 , Garrett D'Amore wrote:
> 1) Great job! This driver looks really clean, and well written.
> I'm fairly impressed.
Thank you. I am glad to have the burden of completion lifted :-)
> 2) The "vfe" driver is being worked for inclusion into Nevada
> someday soon. Its too bad this duplication of effort exists,
> because, your driver looks quite nice.
I understand. I decided long time ago to do this and kept doing even
when the need for it went away.
It was also an educational exercise. Besides that; power-efficiency.
Do as much as possible with a low power board. At some point in the
future, my efforts will be justified. (at least; that is what I like
to believe..)
>
> 3) FYI, the ndd ioctls which you've figured out are going to be
> defunct really soon, thanks to PSARC 2007/429, aka Brussels.
Hehe, from brussels.pdf:
The “ndd compatibility” component of the umbrella case for PSARC
2007/429 will provide legacy
support of existing ndd usage using the methods described below.
Anyway, when brussels arrives I will integrate vr with it.
> 4) I wonder if the common interrupt blanking feature of Nemo/GLDv3
> is a better solution to using a separate task to handle your
> polling. Crossbow should fix that, in any case.
Possibly. I'll re-look into this. I sort-of-assumed that GLD3
blanking required hardware support, which this card doesn't do.
However, I think the current implementation achieves the same:
intr_avoided 1217317
intr_claimed 2306603
This a snapshot of my EPIA. After (roughly) half of the claimed
interrupts, there was more work to do. Now, it would be really cheap
to use the (soft) interrupt thread for these subsequent tasks (a la
linux max_interrupt_work), but that doesn't play nice with interrupt
sharing, thus the task.
Regards, Joost
>
> -- Garrett
>
> Joost Mulders wrote:
>> Hi all,
>>
>> I am happy to announce the availability of yet another driver
>> ("vr") for a few
>> of the VIA Rhine fast Ethernet controllers:
>>
>> VIA Rhine VT6102 Fast Ethernet, pci1106,3065
>> VIA Rhine VT6105 Fast Ethernet, pci1106,3106
>>
>> A package with both sparc and intel binaries is at
>> http://joostm.nl/solaris/vr/VR_1_09/via-rhine-ethernet.pkg
>> Source is at
>> http://joostm.nl/solaris/vr/VR_1_09/
>>
>> The reason for existance of this driver is a bit of a story:
>> A long time ago, I bought an VIA EPIA board for use as a low-
>> power home server.
>> At that time, rh didn't work too well for me, so I installed BSD
>> and decided
>> (being naive) to write a new driver as an educational excersise.
>> I took a luxury amount of time for that ....
>> Meanwhile rh (now vfe) works very well. I completed vr just for
>> the sake of
>> completion.
>>
>> vr is a GLD3 driver created with an eye on CPU utilization,
>> because my EPIA
>> doesn't have too much of that :-)
>>
>> Feedback appreciated.
>>
>> Enjoy, Joost
>>
>
--
Joost Mulders + email: [EMAIL PROTECTED]
Technical Specialist + phone: +31-33-45-15701
Client Solutions + fax: +31-33-45-15734
Sun Microsystems + mobile: +31-6-5198-7268
-= Anything not done right, has to be done again =-
_______________________________________________
driver-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/driver-discuss
