Paul Durrant wrote:
On 4/21/07, Joost Mulders <[EMAIL PROTECTED]> wrote:
As for hme/qfe, I have good news for you. hme and qfe are similar
MAC's. The reason for having different sources for the two was the
trunking capability for qfe. So, if you have hme, you'll cover qfe as
well.
They're not just similar they're *identical*. The qfe NIC is just a
bridge and 4 cheerios, whereas the hme is no bridge and single
cheerio.
It used to be the case the the PCI hotplug bus enumerator even had
specific code in to recognize a bridge + 4 cheerios and modify the
'compatible' property on the created device nodes so as to bind to the
qfe driver rather than the hme!
Paul
Okay, I'd assumed there were some specific bug fixes in qfe that might
not be in the hme driver.
But if "hme" is ported to GLDv3, would that make it possible for the qfe
driver, as such, to just go away? (By virtue of the fact that GLDv3
offers 802.3ad link aggregation?)
I'm thinking it would then be fairly straight-forward to convert the
code to support a x86 hardware as well, which seems to be a popular
request for qfe.
-- Garrett
_______________________________________________
networking-discuss mailing list
[email protected]
