Subject: Re: issues with screenblank and wscons
To: Kailash Sethuraman <>
From: Luke Mewburn <>
List: tech-userlevel
Date: 11/25/2004 20:54:49
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline

On Thu, Nov 25, 2004 at 03:28:52PM +0800, Kailash Sethuraman wrote:
  | I have been poking at screenblank in an attempt to get it to
  | work with wscons. In its current state, stat's on the devices
  | dont correctly detect activity. Calling open on wskbd or any ws*
  | related device detaches it from the parent mux, hence any approach
  | involving this is also not feasible. So from what I can gather,
  | the only approach seemed to be opening the ttyE* devices and
  | periodically checking them for activity. Using kqueues and applying
  | the EVFILT_READ on these devices seem to correctly detect input ,
  | while ignoring output. I have attempted on a patch that implements
  | this, its at Its
  | against screenblank.c distributed with NetBSD 2.0RC1. This mostly
  | works for all normal console activity. However this approach is
  | also flawed as it can only detect chars going to the tty, and not
  | incoming through the keyboard. For example, this method fails when
  | I ssh into another box from the console. This has led me to believe
  | that kernel modification would be necessary to fix screenblank.
  | Perhaps adding an ioctl to wsdisplay or something higher to return
  | the time of the last wsevent? Is the need for this justfied?

This looks like PR 11905:

I think the problem is that the mtime and/or atime isn't updated
on the wscons ttyE? correctly, or at least, as it would be on a
serial or traditional workstation console.   I would like to see
this fixed correctly.

Content-Type: application/pgp-signature
Content-Disposition: inline

Version: GnuPG v1.2.6 (NetBSD)