EDID for radeonfb

EDID for radeonfb

[Kronos]

i2c-algo-bit.o: read 128 bytes.
needed 0 jiffies
read_EDID: i2c_transfer = 2
0x0 0xff 0xff 0xff 0xff 0xff 0xff
EDID v1 signature!
RADEON_GPIO_Y_1: hi
EDID checksum passed.
Dumping data:
0x0 0xff 0xff 0xff 0xff 0xff 0xff 0x0 0x41 0xc 0x18 0x8 0xa3 0x6 0x0 0x0
0x14 0xd 0x1 0x3 0x1f 0x22 0x1b 0x78 0xea 0xc 0x95 0xa2 0x56 0x4c 0x96 0x25
0x1a 0x50 0x54 0xbf 0xef 0x80 0x71 0x4a 0x71 0x4f 0x81 0x40 0x81 0x80 0x1 0x1 
0x1 0x1 0x1 0x1 0x1 0x1 0x30 0x2a 0x0 0x98 0x51 0x0 0x2a 0x40 0x30 0x70
0x13 0x0 0x52 0xe 0x11 0x0 0x0 0x1e 0x0 0x0 0x0 0xff 0x0 0x20 0x48 0x44
0x20 0x20 0x30 0x30 0x31 0x36 0x39 0x39 0xa 0x20 0x0 0x0 0x0 0xfc 0x0 0x50
0x68 0x69 0x6c 0x69 0x70 0x73 0x20 0x31 0x37 0x30 0x53 0x34 0x0 0x0 0x0 0xfd 
0x0 0x38 0x4c 0x1e 0x52 0xe 0x0 0xa 0x20 0x20 0x20 0x20 0x20 0x20 0x0 0x22

Ok, I've a working implementation of the I2C code for radeonfb :)
The bug was very stupid: 0x50 became 0x20 and I spent a couple of weeks
wondering why it didn't work...

I've tested the code on a R300 board but it should work on any radeon. I
need a couple of day to merge  it into radeonfb and to test. I'll send a
patch ASAP.

Luca
-- 
Reply-To: kronos@kronoz.cjb.net
Home: http://kronoz.cjb.net
Runtime error 6D at f000:a12f : user incompetente


-------------------------------------------------------
This SF.net email is sponsored by Dice.com.
Did you know that Dice has over 25,000 tech jobs available today? From
careers in IT to Engineering to Tech Sales, Dice has tech jobs from the
best hiring companies. http://www.dice.com/index.epl?rel_code=104

Re: EDID for radeonfb

[Benjamin Herrenschmidt]

> Ok, I've a working implementation of the I2C code for radeonfb :)
> The bug was very stupid: 0x50 became 0x20 and I spent a couple of weeks
> wondering why it didn't work...
> 
> I've tested the code on a R300 board but it should work on any radeon. I
> need a couple of day to merge  it into radeonfb and to test. I'll send a
> patch ASAP.

Hi !

It would be real cool if you could start from the radeonfb version
that is at bk://ppc.bkbits.net/linuxppc-2.5-benh (It needs my previously
posted Power Management patch to fbdev to build, but you can just
comment out 2 lines to make it build & run without it).

I'm currently having maintainership of this driver and I'll have an
update sent to James rsn based on that tree, so that would simplify
the merging significantly.

Ben.



-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines
at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0

Re: EDID for radeonfb

[Jon Smirl]

The EDID code will allow the driver to ignore the BIOS copy in low RAM and use
the ROM instead. Ben, could you make sure that the code to enable/access the
ROM works on your Macs since I don't own one? Did you finish merging what was
left of the Rage/Radeon PCI IDs from my patch?

EDID plus ROM will solve both problems: multiple adpaters and Mac portables
needing EDID.

--- Benjamin Herrenschmidt <benh@kernel.crashing.org> wrote:
> 
> > Ok, I've a working implementation of the I2C code for radeonfb :)
> > The bug was very stupid: 0x50 became 0x20 and I spent a couple of weeks
> > wondering why it didn't work...
> > 
> > I've tested the code on a R300 board but it should work on any radeon. I
> > need a couple of day to merge  it into radeonfb and to test. I'll send a
> > patch ASAP.
> 
> Hi !
> 
> It would be real cool if you could start from the radeonfb version
> that is at bk://ppc.bkbits.net/linuxppc-2.5-benh (It needs my previously
> posted Power Management patch to fbdev to build, but you can just
> comment out 2 lines to make it build & run without it).
> 
> I'm currently having maintainership of this driver and I'll have an
> update sent to James rsn based on that tree, so that would simplify
> the merging significantly.
> 
> Ben.
> 


=====
Jon Smirl
jonsmirl@yahoo.com

__________________________________
Do you Yahoo!?
Yahoo! SiteBuilder - Free, easy-to-use web site design software
http://sitebuilder.yahoo.com


-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines
at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0

Re: EDID for radeonfb

[Benjamin Herrenschmidt]

On Thu, 2003-08-21 at 22:39, Jon Smirl wrote:
> The EDID code will allow the driver to ignore the BIOS copy in low RAM and use
> the ROM instead. Ben, could you make sure that the code to enable/access the
> ROM works on your Macs since I don't own one? Did you finish merging what was
> left of the Rage/Radeon PCI IDs from my patch?

The "Mac" cards don't have an x86 BIOS in ROM, they have an Open
Firmware instead, we need to fix proper detection of that, but that's
something I can try to do. I didn't finish merging it yet, I've been
quite busy with other drivers this week, radeonfb is on my todolist
for this week-end.

> EDID plus ROM will solve both problems: multiple adpaters and Mac portables
> needing EDID.

Yup (Note: the need for EDID in low RAM happened with x86 configs, on Macs,
I get it via the Open Firmware device tree).

Ben.




-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines
at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0

Re: EDID for radeonfb

[Jon Smirl]

Is GregKH designing the EDID thorough sysfs scheme? I haven't seen a proposal
on how it is going to work. We need to coordinate with that so that we don't
end up with a different EDID scheme for every card.

--- Benjamin Herrenschmidt <benh@kernel.crashing.org> wrote:
> On Thu, 2003-08-21 at 22:39, Jon Smirl wrote:
> > The EDID code will allow the driver to ignore the BIOS copy in low RAM and
> use
> > the ROM instead. Ben, could you make sure that the code to enable/access
> the
> > ROM works on your Macs since I don't own one? Did you finish merging what
> was
> > left of the Rage/Radeon PCI IDs from my patch?
> 
> The "Mac" cards don't have an x86 BIOS in ROM, they have an Open
> Firmware instead, we need to fix proper detection of that, but that's
> something I can try to do. I didn't finish merging it yet, I've been
> quite busy with other drivers this week, radeonfb is on my todolist
> for this week-end.
> 
> > EDID plus ROM will solve both problems: multiple adpaters and Mac portables
> > needing EDID.
> 
> Yup (Note: the need for EDID in low RAM happened with x86 configs, on Macs,
> I get it via the Open Firmware device tree).
> 
> Ben.
> 
> 


=====
Jon Smirl
jonsmirl@yahoo.com

__________________________________
Do you Yahoo!?
Yahoo! SiteBuilder - Free, easy-to-use web site design software
http://sitebuilder.yahoo.com


-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines
at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0

Re: EDID for radeonfb

[Benjamin Herrenschmidt]

On Thu, 2003-08-21 at 22:51, Jon Smirl wrote:
> Is GregKH designing the EDID thorough sysfs scheme? I haven't seen a proposal
> on how it is going to work. We need to coordinate with that so that we don't
> end up with a different EDID scheme for every card.

Imho, the fbdev should be a separate device from the monitor:

fbdev -> fbmon -> EDID

The EDID should be a property of the monitor

So fbdev should instanciate a monitor object for each head after
detection. It's still unclear for me how to deal with the various
types of monitors though

 - EDID available
 - Mac old style "sense code" available
 - fixed mode known based on motherboard (old mac laptops among others,
   probably some embedded)
 - unknown

The EDID shall be used to build a modedb of available modes for this
display, the driver doing validation based on that, though a flag to
set_var would still be useful to force a mode that doesn't pass
validation (EDIDs can be wrong...)


Ben.



-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines
at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0

Re: EDID for radeonfb

[Jon Smirl]

The list of modes supported by the board has to get worked into this too. Who
should do the diff'ing to determine valid modes? The list of legal monitor
modes might be from EDID or the fbmode database.

Should sysfs have one entry for legal board modes and one for legal monitor
modes? Or should it just have a single entry for legal combos?

--- Benjamin Herrenschmidt <benh@kernel.crashing.org> wrote:
> On Thu, 2003-08-21 at 22:51, Jon Smirl wrote:
> > Is GregKH designing the EDID thorough sysfs scheme? I haven't seen a
> proposal
> > on how it is going to work. We need to coordinate with that so that we
> don't
> > end up with a different EDID scheme for every card.
> 
> Imho, the fbdev should be a separate device from the monitor:
> 
> fbdev -> fbmon -> EDID
> 
> The EDID should be a property of the monitor
> 
> So fbdev should instanciate a monitor object for each head after
> detection. It's still unclear for me how to deal with the various
> types of monitors though
> 
>  - EDID available
>  - Mac old style "sense code" available
>  - fixed mode known based on motherboard (old mac laptops among others,
>    probably some embedded)
>  - unknown
> 
> The EDID shall be used to build a modedb of available modes for this
> display, the driver doing validation based on that, though a flag to
> set_var would still be useful to force a mode that doesn't pass
> validation (EDIDs can be wrong...)
> 
> 
> Ben.
> 


=====
Jon Smirl
jonsmirl@yahoo.com

__________________________________
Do you Yahoo!?
Yahoo! SiteBuilder - Free, easy-to-use web site design software
http://sitebuilder.yahoo.com


-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines
at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0

Re: EDID for radeonfb

[Benjamin Herrenschmidt]

On Thu, 2003-08-21 at 23:21, Jon Smirl wrote:
> The list of modes supported by the board has to get worked into this too. Who
> should do the diff'ing to determine valid modes? The list of legal monitor
> modes might be from EDID or the fbmode database.
> 
> Should sysfs have one entry for legal board modes and one for legal monitor
> modes? Or should it just have a single entry for legal combos?

Good point. Looking at what things like MacOS do, I'd say fbmon provides
a list to radeonfb, which filters it out and creates the "real" list.

I think it's not time to do a full blown fbmon layer now, we need more
experience with it and it's too late in the 2.6 process. I'll probably
just keep fbmon as a pool of "utility" functions and do most from
radeonfb (and aty128fb if I get a chance to port the i2c code) for now.

One thing I'm hesitating to do is to dive into a driver split &
rearchitecture now or not... 

Ben.



-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines
at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0