Mailinglist Archive: radeonhd (427 mails)

[garyhunt@xxxxxxxx]

Matthias Hopf wrote:

> Wow.
> Seems like some RS740 don't like their memory controller
to be
> programmed via PCI space. Could be that programming via
the memory
> mapped register region is more stable for other chips as
well.
> This is an area where Egbert has much much much more
experience, so
> maybe he can comment on that.
>
> In the meantime, please try (not only you, but others as
well) the
> attached patch - it refactors the whole MC programming
logic to use the
> memory mapped registers. Before committing something as
severe as that
> it has to be validated on quite a number of chips...
> It actually simplifies the logic quite a lot (makes it
independent of
> libpciaccess), so if it actually works I would like to see
this
> committed, even if it is not necessary on other chips :-)
>
> > Let me know if there's anyway I can help with this.
>
> Thanks for your findings, and especially the patch!
>
> Matthias


Worked here for the first time, no MC idle issues.

 RADEONHD: version 1.2.5, built from git branch master,
commit dd287015 + changes


 II) RADEONHD(0): Unknown card detected:
0x796E:0x1025:0x0155.

 (--) RADEONHD(0): Detected an RS740 on an unidentified card
 (II) RADEONHD(0): Mapped IO @ 0xfe7f0000 to 0x7f2738876000
(size 0x00010000)
 (II) RADEONHD(0): Getting BIOS copy from legacy VBIOS
location
 (II) RADEONHD(0): ATOM BIOS Rom: 
        SubsystemVendorID: 0x1002 SubsystemID: 0x796e
        IOBaseAddress: 0xd000
        Filename: Bengal_740_3


Crashes on restarting X server, 

Gary

-- 
To unsubscribe, e-mail: radeonhd+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: radeonhd+help@xxxxxxxxxxxx