I am responsible for updating a port of the Pico Embedded
development system host software for OpenBSD 3.9
to current.

If humanly possible I would like to be able to provide the Pico
target Cardbus/PCMCIA drivers are a loadable module rather
than as a patch to the OpenBSD kernel source.

Much googling has resulting n a limited amount of resources for
developing simple OpenBSD character device modules,
but nothing about adapting them to Cardbus or PCMCIA (or in the near
future express bus).

I have experience with systems development on a number of platforms,
but I am fairly new to OpenBSD.

Pointers to more advanced driver/module development resources, even
just general driver development resources
would be greatly appreciated.

I appologize if somehow in my searching I have missed something obvious.

On Fri, Aug 31, 2007 at 01:50:06PM -0600, Chris Kuethe wrote:

I have the same experience on my Dell D620, with different numbers.

On 8/31/07, Chris Kuethe gmail.com> wrote:

On 8/29/07, Jonathan Gray wrote:

fadt hdr_rev: 1, reset_reg.addr: 0x0 flags 0x4a5
ACPI reboot method not supported

yet the machine still reboots. Gigabyte M61P-S3

* Jeremy C. Reed reedmedia.net> [2007-08-30 03:30]:

Right, you had mentioned to this a couple of months ago to me. While
I feel for you, it's an OpenBSD list and I chase problems on OpenBSD,
where I haven't seen this corruption - so please avoid posting about
non-openbsd bugs to an openbsd mailing list in the future. Thanks.

-Bob

On Aug 31, 2007, at 7:16 AM, Theo de Raadt wrote:

From a purely short-term pragmatic standpoint, i have found that
rejecting bad helos has a huge impact on the number of mails i have
to do username lookups on or run through amavisd-new.

on Saturday, for example, i rejected 5015 bad helos ( gibberish:
90%%, my hostname/ip: 8%%, non-fully-qualified hostnames: 2%% ) and
passed ~ 2000 legitimate messages.

actually, it is the practice of permitting errors in a protocol
that leads to the protocol being increasingly violated.

and then, as this progresses at the protocol, it becomes harder
and harder to spot right from wrong at other levels.

please, fight standard breakage.

hmm, on Thu, Aug 30, 2007 at 12:56:58PM -0700, Marco S Hyman said that

There's a counter in the kernel for counting the number of preemption
events but it's not being used. It would be useful to have this as
another way of detecting failovers and also for detecting instability,
etc.

There's two other counters that exist that aren't being used: ostates and
badaddrs. I can't really see where those two would fit in; they can
probably be removed from struct carpstats.

If people are interested in detecting carp failovers, they should also
look at PR 5513 which offers a patch to add logging of carp state
transitions.

It is not my job to educate clueless admins. If their server is
mis-configured it is not, by my definition, a legitimate server.
If they want to contact this site by email they will have to
either fix their configuration, use a different server, or print
out their message and send it via postal mail. Or give up. I
don't much care which they do.

In practice this has not been a problem. Small companies without
knowledgable admins tend to use their ISP mail server. Of course,
the only valid email I'm aware of rejecting in the last year came from
YAHOO mail servers. That person found another way to contact me.

// marc