Subject: Re: Fixes for TGA driver in XFree86
To: None <firstname.lastname@example.org>
From: None <email@example.com>
Date: 11/26/2004 09:25:46
I have a TGA2 that came with my alphastation 500. It's got what appears to be some slightly bad RAM on it (always-white pixels here and there). Other than that, the card works just fine (the xdecnet server works fine with it).
As such, anyone who wants to play with it, just lemme know and cover the cost of s/h and i'll send it on it's way to you. I gotta ask for the s/h cause can't take food out of baby's mouth :( .
Oh, I'm in Omaha, NE for shipping/pickup references.
> From: "Michael Kukat" <firstname.lastname@example.org>
> Date: 2004/11/26 Fri AM 02:48:39 EST
> To: email@example.com
> Subject: Fixes for TGA driver in XFree86
> okay, i started fiddling around now. Current first improvement in the tga
> driver in XFree86 (tested on NetBSD/alpha 2.0RC4):
> For now, a table of modes for the selections of the rotary switch existed in
> some strange order, with parts enabled/disabled by defines. As the programming
> of the clock chip doesn't seem to be really known, this table with known
> settings is scanned according to the resulting modeline. This scan just checks
> for V/H resolution, and due to the nature of this scan funktion and the table,
> any 1280x1024 mode results in 72Hz VSync selection. This is suboptimal on
> TFTs, which are better used with 60Hz. Yesterday evening i completely
> re-ordered this code, now the full mode table is available, and the scan
> routine also uses the integer value of the pixel clock to find a matching
> mode. The output now correctly finds mode 2, which is 1280x1024@60Hz and
> matches the correct rotary switch setting 2. You now have to get a modeline
> with correct V/H-resolution and correct pixel clock (at least the integer MHz
> must match). Due to this should not really be a problem, the routine should be
> a bit better now.
> The next issue is the mentioned too bright picture on TFTs, i'll try to play
> around with some RAMDAC registers and hopefully get this fixed by this.
> If this really works, and if i find the time, i even would try to add working
> support for the PowerStorm 4D20, which is 24bit, and currently "just" has
> palette problems, but i didn't check further. Besides those evil color
> problems, the card seems to work okay. And as all those TGAs are based on the
> same chipset, i assume that fixing those DAC problems should enable using all
> TGA models. I can just test with old TGA, 8bit TGA2 (PBXGB-AA) and 24bit TGA2
> But now for the place to address these patches. First, does anyone use these
> cards, maybe with TFTs and maybe having the same problems with too high black
> levels? Meaning - is there any interest for improved TGA support?
> And second... should the fixes flow in at XFree86 directly (where i have to
> find out where to post them :), or can i just send a cvs diff for xsrc here
> and someone has contact to the XFree86 team (Mel Kravitz did lots of works
> here with NetBSD/alpha and xsrc) and can feed the diffs there?
> Maybe in a later stage i'll try to get some cards running which are currently
> not running. Matrox MilII support would be nice, as the cards are cheap and
> Besides this: Mel, good work! You gave a base to check lots of cards on
> different Miatas recently, where the results not yet went here i think.
> http://www.unixiron.org/ Home Powered by: (Net|Open|Free)BSD IRIX NonStop-UX
> Solaris AIX HP-UX Tru64 MUNIX Ultrix VMS SINIX Dolphin_Unix OpenStep MacOS A/UX