Date: Sat, 1 Nov 1997 12:17:29 +0100 From: Stefan Esser <se@FreeBSD.ORG> To: John-Mark Gurney <gurney_j@resnet.uoregon.edu> Cc: Chris Csanady <ccsanady@bob.scl.ameslab.gov>, Chris Vance <cvance@tis.com>, freebsd-hackers@FreeBSD.ORG Subject: Re: LKM Interface Message-ID: <19971101121729.47401@mi.uni-koeln.de> In-Reply-To: <19971101010707.25996@hydrogen.nike.efn.org>; from John-Mark Gurney on Sat, Nov 01, 1997 at 01:07:07AM -0800 References: <19971030180558.63267@hydrogen.nike.efn.org> <199710310308.VAA01889@bob.scl.ameslab.gov> <19971030194044.15908@hydrogen.nike.efn.org> <19971031184937.23539@mi.uni-koeln.de> <19971101010707.25996@hydrogen.nike.efn.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 1997-11-01 01:07 -0800, John-Mark Gurney <gurney_j@efn.org> wrote:
> Stefan Esser scribbled this message on Oct 31:
> > If you need it, it is easy to put back. The API was that the
> > PCI LKM just called
> >
> > pci_register_lkm(struct pci_device *p, int rev)
> >
> > with the parameter of the DATA_SET and rev=0.
> >
> > The rescan was automatically performed from within that LKM
> > registration function.
>
> so, did the pci_register_lkm do this? or was this another function
> that would do this?
The pci_regsister_lkm() function initiated the rescan.
Only cards with no driver were looked at.
> > Sorry, I'm in a hurry, now. If there is any need for PCI LKM
> > support, I may be able to put back that feature over the weekend.
>
> if you could, just send me the patches... I'll make sure that they
> work with the new system...
You'll find everything in the last version of pci.c
before my rewrite (rev. 1.60 did not have LKM support,
and rev 1.64 is the last one where I made a change to
the implementation, so you may want to look at a diff
from 1.60 to 1.64).
But that code relied on the probed PCI devices to be
registered (was function pci_remember()), but I want
to have a completely different data structure in the
future (you already know that ...).
> > I also prepared a simple shell script that could be run from
> > /etc/rc and which searched for PCI devices that had no driver
> > compiled in, and tried to modload the corresponding driver
> > (actually used for one ATM card driver, so far).
>
> interesting... could you include that too?
Sorry, but seems I did fail to make me understood in
more than one sentence of the previous reply ...
The Zeitnet ATM driver was written by
"Jin Guojun[ITG]" <jin@george.lbl.gov>
and contains the PCI LKM as patches (relative to
FreeBSD-2.2.0, I guess). Information is available
from the Zeitnet web site, according to the author
of the driver.
Regards, STefan
PS: I'll try to find time to reply top your previous
mail regarding the data device structure later
this weekend ...
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19971101121729.47401>