[Top] [All Lists]

Re: kern/34935: if_re_pci.c needs updating

Subject: Re: kern/34935: if_re_pci.c needs updating
From: "Perry E. Metzger"
Date: Mon, 30 Oct 2006 01:30:02 +0000 UTC
The following reply was made to PR kern/34935; it has been noted by GNATS.

From: "Perry E. Metzger" <perry@xxxxxxxxxxxx>
To: Izumi Tsutsui <tsutsui@xxxxxxxxxxxxxxx>
Cc: gnats-bugs@xxxxxxxxxx, kern-bug-people@xxxxxxxxxx,
        gnats-admin@xxxxxxxxxx, netbsd-bugs@xxxxxxxxxx
Subject: Re: kern/34935: if_re_pci.c needs updating
Date: Sun, 29 Oct 2006 20:22:30 -0500

 Izumi Tsutsui <tsutsui@xxxxxxxxxxxxxxx> writes:
 > perry@xxxxxxxxxxxx wrote:
 >> >  Please post individual bugs/fixes which you can confirm.
 >> I can't confirm them without doing the port, at which point why would
 >> I need to post them? :)
 > Otherwise this PR will never be closed and makes no sense, I'm afraid.
 Clearly the PR makes plenty of sense. There are documented issues that
 FreeBSD has fixed. Again, I can't absolutely confirm that their fixes
 work without actually trying them, but I see no reason that I need to
 have fixed a problem before filing a PR.
 >> Clearly the fix that prevents certain versions of the card from
 >> screwing up hardware assisted checksums is important.
 > Only this? Or more other 8139C+ fixes?
 > Is it a good idea to pull all fixes blindly?
 I suspect that most of their fixes should be pulled "blindly" -- we
 pulled their driver "blindly" after all, assuming that it was
 correctly written.
 > Anyway please also see recent posts on tech-kern and tech-net.
 > yamt said that hw ip4csum bug only happens with
 > very small packet (less than ETHER_PAD_LEN) so FreeBSD's
 > fix is not right.
 FreeBSD's fix is for precisely that problem.
 > Furthermore, they choose to copy small packets
 > to a new mbuf for padding, and several people object it.
 When and if they find a better fix they can implement it.
 Perry E. Metzger               perry@xxxxxxxxxxxx

<Prev in Thread] Current Thread [Next in Thread>