Stale PCI ID

[Gleb Smirnoff]

  Jack,

On Fri, Sep 15, 2006 at 07:01:25PM -0700, Jack Vogel wrote:
J> >J> In the last attempt to merge community CVS with Intel internal code I
J> >J> came across an issue I'd like to bring up.
J> >J>
J> >J> There is an ancient e1000 card, pci id 1000, an 82542, that we
J> >J> don't have in our source, yet community cvs still does. Support
J> >J> for this was removed from Linux long ago because the card did
J> >J> not even work.
J> >J>
J> >J> I just had our test group hunt up one of these and test, and sure
J> >J> enough, the driver recognizes it, but it does NOT pass traffic.
J> >J>
J> >J> Clearly no one is using these, at least not with STABLE :) and
J> >J> as Intel does not want to support this I would recommend removing
J> >J> the ID from the driver.
J> >J>
J> >J> Comments?
J> >
J> >It should be removed then. When merging the Intel versions of driver to
J> >FreeBSD, I've noticed that some PCI IDs disappeared from vendor driver.
J> >Since I couldn't obtain any comments from the release tarball, I decided
J> >to be on safe side and leave these PCI IDs in the driver untouched.
J> >
J> >Which exact one are you speaking about? The E1000_DEV_ID_82542, that is
J> >equal to 0x1000 is supported by em-6.1.4 vendor's driver.
J> 
J> I AM the vendor :)  And I believe that the 0x1000 ID crept into my code
J> from merging with your tree :) Our test group came back to me and
J> said 'hey, this is old and we don't think these boards even work' , so
J> I told them to test, they did, and sure enough, it wouldnt even pass
J> traffic.

AFAIK, the mega merge with FreeBSD code was made in version 6.0.5. But the
E1000_DEV_ID_82542 is also present in em-3.2.18 and em-5.1.5 drivers.
That's why I am unsure that it is the one that needs to be removed.

J> There is actually 3 different rev boards, and rumor has it that the
J> last revision would work, but I think its better to just drop the thing.

-- 
Totus tuus, Glebius.
GLEBIUS-RIPN GLEB-RIPE