Holger,
aha! This is _exactly_ the problem that I'm looking at now.
I'm using Russell's 2.4.1 patch. What I'm doing is moving from
2.4.0-pre* to the 'latest'. The problem, basically, is around the
use by the frame buffer memory (fbmem.c) driver using
the io_remap_page_range() macro. On Integrator (an ARM
development board) that macro tries to map a PCI memory address
whereas the macro is expecting a physical address. My workaround
is for there to be a board specific io_remap_page_range macro which
maps the address from PCI memory land into a board's physical address.
So, for example, Integrator's physical PCI memory address is
0x40000000 and PCI setup code maps in stuff at PCI memory
addresses between 0 and 0x20000000.
Dave
Holger Waechtler wrote:
> Hi Russel, hi everybody else,
>
> are there known problems with mmap'd framebuffer on the ARM ?
> We use a Matrox G200 PCI in ARM's Integrator board.
> Read/write access works fine, the penguin logo and graphical console show
> up. But as soon an application mmap's the framebuffer and tries to write,
> the program either segfaults or the machine locks up.
>
> Any idea ?
> Thanks in advance,
>
> - Holger
>
> _______________________________________________
> http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm
> Please visit the above address for information on this list.
--
----------------------------------------------------------------------
David A Rusling Consulting Engineer
ARM Limited
Liberty House,
Moorbridge Road,
Maidenhead, SL6 8LT
Tel: UK-(0)1628-427754
Fax: UK-(0)1628-780551
e-mail: [EMAIL PROTECTED]
----------------------------------------------------------------------
_______________________________________________
http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm
Please visit the above address for information on this list.
