[Lcdproc] CwLnx display corruption - possible cause and fix

Peter Marschall peter@adpm.de
Mon Apr 30 13:57:01 2007


Hi Dave,

On Monday, 30. April 2007 07:48, Dave Platt wrote:
> Here's the results of a few days of fiddling.  I've
>
> -  Fixed the nonblocking-write problem
>
> -  Optimized the optimized framebuffer-update code to eliminate
>    excessive move-insertion-point commands
>
> -  Cleaned up (I think) the initialization sequence so it'll actually
>    talk at 19200 if requested to do so
>
> -  Added code to the framebuffer-flushing routine to actually propagate
>    changes in the backlight status and brightness to the device.  The
>    backlight now goes off at low system-load levels (if lcdproc is
>    configured to enable this) and the blink/flash feature is working
>    (my digital-audio-recorder client uses this to signal the end of a
>    recording).
>
> -  Made "SEAMLESS_HBAR" mode work on the 12232.  This required adding
>    an alternative set_char routine which doesn't limit the hbar
>    characters to 5 columns on this device.  [I'm tempted to add
>    support for the native draw-bar commands in the 1602 and 12232 but
>    that would involve making the framebuffer management quite a lot
>    more complex.]
>
> -  Added some newlines to the Debian init scripts' PRINTF commands
>    so that they can be run interactively without leaving the subsequent
>    prompt on the same line as the script output.
>
> -  I put the soft-reset-BIOS support code back in, but am not actually
>    using it in the code at the moment.
>
> The patch is against the CVS tree as of a day or so before 0.5.2 was
> released... ought to apply fairly easily, I think.

Patch aplied witout a hitch.
Changes committed to CVS (with a few indenting changes to keep indentation ;-)

They will be in tonight's nightly tar ball.

Anybody with CwLnx displays please test: this is a major fix for CwLnx.


Thanks for supporting LCDproc
Peter

-- 
Peter Marschall
peter@adpm.de