On Tuesday 07 July 2009 15:09:51 Rafał Miłecki wrote:
> Last time we ended with discovering it's openChrome register operation
> that affects Broadcom's PCI bus. The most obvious was we have bug
> causing writing over our pci region. Problem is quite well described
> in http://www.openchrome.org/trac/ticket/288
>
> We were able to discover exact part of driver causing this problem.
> It's about writing to our 0x2A register. That's well documented in one
> comment: http://www.openchrome.org/trac/ticket/288#comment:20 and
> confirmed by some patches disabling 0x2A writing that are known to
> workaround it.
>
> Problem is that 0x2A doesn't seem to be anything outside our pci
> region. We do many other operations on even higher ports (from 0x00 up
> to 0x4C) and it doesn't seem to break anything.
>
> Do you have any idea what can be wrong with this 0x2A? I can't imagine
> any connection between GPU and WiFi done on purpose. Can that just
> mean broken motherboard by design? Or maybe some BIOS issue?
>
> As far as I know this happens only on very one notebook model, that is
> HP 2133 (but many cases confirmed that, it's not just one broken
> case).
>
> I know it's not really your business, but any help/tips would be appreciated.
>
Your register accesses seem to be indirect:
r...@mini:> inb 0x3C4 #check which register is active
0x00
r...@mini:> outb 0x3C4 0x2a #i want to acess register 2A
Can you explain how the register access works exactly? Is this MMIO or is this
x86 in/out instructions (PIO)?
What does lcpci -vvnn show on your system (for wifi and GPU)
--
Greetings, Michael.
_______________________________________________
Bcm43xx-dev mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/bcm43xx-dev
