Thread (1 message) 1 message, 1 author, 2003-11-21

Re: tridentfb questions -- tridentfb for Kernel-2.6

From: Yi-Shin Li <hidden>
Date: 2003-11-21 13:50:57

Hello Jani,

I started work on this tridentfb at my spare time. I finally
made some small progress to have it work with kernel 2.6 in
my platform.

Attached(tridentfb.tar.bz2) is my modified tridentfb module.

Here is the README in my attached file:
================================================================================
Yi-Shin Li "ysli _AT_ ee.columbia.edu"
http://joulee.dyndns.org/index.pl/tridentfb

Why this tridentfb?
===================

My Trident CyberBlade/i1 which comes with VIA's EPIA-V10K doesn't
work with the tridentfb driver from linux-2.4.22 or linux-2.6.0-test8;
nor the epiafb from http://epiafb.sourceforge.net/.

I can load above drivers. However, the mode I got is "640x400@70".
It is weird and nothing can be displayed after loading above modules.
Could possibly because that I am using CRT but not LCD or TVout.

Therefore, I decide to give myself a chance to learn how to write
framebuffer driver for Linux 2.6 kernel. I refered tdfxfb, radeonfb,
and skeletonfb and sent a mail to Jani Monoses about my concern for
this driver. Then, I started work on this.


What it can do so far:
======================

Jani considered comprehensively for various types of Trident Chips
on different kind of displays modes(CRT/FP). I cannot do that since
I just have one CyberBlade/i1 on my hand. Here is the list for what
the driver can do:

-- Work with non-accelerated 640x480-8bpp mode for CRT display.
   (tested with mplayer on vesa output mode)

-- Can be loaded into and removed from kernel.

-- Can only be load as kernel module.


Todo list:
==========

I am not a kernel expert. There are a lot stuffs for me to learn.

-- Figure out the mechanism of fb_find_mode() and fb_check_var(),
   so that we can pass "mode_option" as module parameter.

-- Figure out how to enable hardware acceleration mode.

-- Figure out how to integrate this driver into kernel.
==end of README=================================================================

Regards,
Yi-Shin


On Wed, 5 Nov 2003, Jani Monoses wrote:
quoted
Dear Jani,

Thanks for your work on the tridentfb driver. I traced your code 
because it can not probe the correct display mode on my EPIA platform
(with cyberblade/i1, CRT). My kernel version was 2.4.22-openmosix. 

I found that you have new version(0.7.8-NEWAPI) in 2.6.0-test7. I will
try it after work today.

Q1: In "trident_pci_remove()", you use "release_region()" for mmio.
    Shouldn't it be release_mem_region()? since we requested that
    region with "request_mem_region()".

Q2: In "trident_pci_probe()", should we release_mem_region() for both
    mmio and smem before returning -1 after requesting the above
    region? In radeonfb.c, they did this.
You're right on these, I had not time to update the sources lately.
The 2.6 version is not yet accelerated although it works well with
generic accel from fbdev.
quoted
Q3: Is there any forum/mailing-list for tridentfb?
No, tridentfb is not a hot topic.
quoted
Q4: Where could I get most updated fbdev information?
    the linux-fbdev.sf.net is kind of outdated.
The sources :)

Jani
--
Yi-Shin Li [off-list ref]

Attachments

Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help