Re: mmap() fails for V_RAM and V_BIOS

[KIYOHARA Takashi <kiyohara%kk.iij4u.or.jp@localhost>]

Hi! Robert,


From: Robert Swindells <rjs%fdy2.demon.co.uk@localhost>
Date: Sat,  8 Nov 2008 11:45:04 +0000 (GMT)

> 
> KIYOHARA Takashi wrote:
> >From: Martin Husemann <martin%duskware.de@localhost>
> >Date: Sat, 8 Nov 2008 10:28:56 +0100
> >
> >> On Sat, Nov 08, 2008 at 11:34:11AM +0900, KIYOHARA Takashi wrote:
> >> > I loaded libint10.o with s3_driver.o with the PowerPC machine (bebox).
> >> > libint10.o tries mmap() of V_RAM (0xa0000-0xbffff) and V_BIOS (0xc0000).
> >> > However, I do not think that it has this space on PowerPC.  Should this
> >> > allocate the space somewhere for compulsion?
> >> 
> >> The V_RAM space is usually described by a memory BAR of your vga - no idea
> >> about the vga cards ROM though.
> 
> >My S3's card (S3 Trio64) has a memory space.  It uses as frame buffer,
> >mapped on 0xc0000000 by BAR0 against from CPU.
> 
> On ibmnws the S3 Trio64 has BAR0 at 0x00000000 too, with the
> framebuffer at the start and the device registers at 0x01000000 within
> the same bar.

I am setting 0x0x00100000 to BAR0 of Trio64 now.
Call extent_create("pcimem", 0x00100000,,,).


> Attempting to mmap() the bar through /dev/mem fails, but works if I
> use the virtual console device.

What is 'virtual console device'?  The genfb use you too?


> I'm trying to use KDrive as the server instead of Xorg/XFree86. My
> current problem is that the KDrive s3 driver uses memory operations on
> the device registers with little-endian constants instead of using
> inw()/outw() calls like the XFree86 driver.

I am very interested in your work.  ;-)
I continue and examine the memory map of int10.

Thanks,
--
kiyohara