perry@piermont.com wrote:

> We don't have the fix FreeBSD did for the checksum issue -- turning
> off the checksum assists is not a fix.

If this PR specifies only the checksum issue I'll handle it,
but it doesn't.

> If you can't trust
> Bill Paul to be right on this, how could you trust him to have written
> the correct driver in the first place? It is his code.

By code inspection?

I'd like to pull only code which I could confirm (by code inspection
or on real hardware), and I won't pull any fixes blindly without
proper approvals. For example, his old drivers had a bunch of
vtophys() and didn't use bus_dma() at all, then people who pulled
such code blindly were often blamed.

> If you want to close the PR with "NetBSD doesn't want to fix problems
> fixed in FreeBSD", please, go ahead.

I just say this PR can't be assinged to me even if I will work
some of such problems unless this PR mentions particular issues.

If there is someone who will handle this PR, there is no problem.

> This whole conversation has been useless from the beginning. I sent
> that PR so I wouldn't forget to try to pull in the FreeBSD fixes, and
> since then you've been sending mail saying "it is wrong of you to want
> to update the driver". Well, I don't see why it is wrong.

I'd say "it is wrong of you to want to update the driver blindly,"
as we always say "it doesn't work unless it's right."
---
Izumi Tsutsui