[Nouveau] [Nouveau-cvs] xf86-video-nv: Branch 'master'

[Ben Skeggs]

On Tue, 2009-04-07 at 11:56 +0200, Maarten Maathuis wrote:
> Do you have strong feelings against making these ports different for
> 8xxx and 9xxx?
No, that sounds like it's the best guess we have so far as to which
cards use which sets of ports.  So, this is probably a good plan for the
moment.

Ben.
> 
> Maarten.
> 
> On Fri, Apr 3, 2009 at 7:12 PM, Maarten Maathuis <madman2003 at gmail.com> wrote:
> > The change was prompted by a personal email i recieved regarding a bug
> > on a 9200 (which included the right ports), a mmio-trace then
> > confirmed on some cards (maybe only 9xxx) that these i2c ports were
> > being init'ed. So i know for fact that they are true for at least one
> > card. xf86-video-nv is practically unmaintained and it's starting to
> > show. On a sidenote it's also true that SOR2 init is incomplete, so
> > without a bios init it won't work.
> >
> > My personal guess is the i2c ports are different for 8xxx and 9xxx.
> >
> > Maarten.
> >
> > On Fri, Apr 3, 2009 at 12:13 PM, Ben Skeggs <skeggsb at gmail.com> wrote:
> >> On Fri, 2009-04-03 at 10:43 +0200, Maarten Maathuis wrote:
> >>> You need these ports for LVDS?
> >>>
> >>> Because i know these ports are correct for dvi and friends.
> >>>
> >>> So if you changed them for lvds, we need to seperate port 4 and 5 into
> >>> lvds and non-lvds.
> >> Why doesn't xf86-video-nv do this?  There's a RH bug (492289) where the
> >> ddx doesn't get any modes (on DVI), but kms (which has the old values)
> >> does.  I didn't actually hear from the reporter yet, but seeing as nv
> >> uses the old values I figured that could be a likely culprit...
> >>
> >> Did your change actually fix anything?  Your commit message was a vague
> >>  "some minor modesetting changes" seemingly without reason.
> >>
> >> Ben.
> >>>
> >>> Maarten.
> >>>
> >>> On Fri, Apr 3, 2009 at 6:26 AM, Ben Skeggs
> >>> <darktama at kemper.freedesktop.org> wrote:
> >>> >  src/nv50reg.h |    4 ++--
> >>> >  1 file changed, 2 insertions(+), 2 deletions(-)
> >>> >
> >>> > New commits:
> >>> > commit 11be9a982073d66a68cd3db2bfc611eb58d3ea81
> >>> > Author: Ben Skeggs <skeggsb at gmail.com>
> >>> > Date:   Fri Apr 3 12:22:04 2009 +1000
> >>> >
> >>> >    nv50: fix i2c port addresses
> >>> >
> >>> > diff --git a/src/nv50reg.h b/src/nv50reg.h
> >>> > index b3125ce..4e21220 100644
> >>> > --- a/src/nv50reg.h
> >>> > +++ b/src/nv50reg.h
> >>> > @@ -50,8 +50,8 @@
> >>> >  #define NV50_PCONNECTOR_I2C_PORT_1                     0x0000e150
> >>> >  #define NV50_PCONNECTOR_I2C_PORT_2                     0x0000e168
> >>> >  #define NV50_PCONNECTOR_I2C_PORT_3                     0x0000e180
> >>> > -#define NV50_PCONNECTOR_I2C_PORT_4                     0x0000e254
> >>> > -#define NV50_PCONNECTOR_I2C_PORT_5                     0x0000e274
> >>> > +#define NV50_PCONNECTOR_I2C_PORT_4                     0x0000e240
> >>> > +#define NV50_PCONNECTOR_I2C_PORT_5                     0x0000e258
> >>> >
> >>> >  /* 0x00610024 is the state register to read, all it's bits also exist in 0x0061002C in the form of interrupt switches. */
> >>> >  #define NV50_DISPLAY_SUPERVISOR                0x00610024
> >>> >
> >>> > ------------------------------------------------------------------------------
> >>> > _______________________________________________
> >>> > Nouveau-cvs mailing list
> >>> > Nouveau-cvs at lists.sourceforge.net
> >>> > https://lists.sourceforge.net/lists/listinfo/nouveau-cvs
> >>> >
> >>
> >>
> >