Chameleon

Issue 191: ATI.C of all branches seemd to have a mistake

Reported by ugo kind, Oct 9, 2011

I see 
{ 0x6760,	0x00000000, CHIP_FAMILY_RV730,		"AMD Radeon HD 
6470M",				kNull		},

but must report 2 info:
my card ATI radeon 6940M is 0x6760 (and is possible to have 2 cards 
with same ID??)
and it has seymour GPU (southern island)
i Inform that the chip family is CAICOS.. starting from jan2011

but .. strange.. everywhere in the net I see that 6470M is also 
Caicos chipset .. and no RV730

so why all the branches have the same ati.c with wrong data?

I don't want to take the ownership of that change.. please could you 
do a verify?

in the end.. I'll test the boot with my mod.. to see if I have 
better run

Comment 1 by MacMan, Oct 9, 2011

Yes it is possible to have multiple cards with the same ID. For 
example, the 5770 and some 6770's share the same ID of 0x68B8, the 
only way to differentiate them is by the subsys_id.

Comment 2 by MacMan, Oct 10, 2011

In reviewing 
http://developer.amd.com/gpu_assets/ATI_Device_ID_List_July_2011.txt 
this looks like an error.

Comment 3 by ugo kind, Oct 10, 2011

ok.so in the trunk effectively i see a new entry 6490m
but are you sure it fit RV chipset?

i tried to boot with this untouched trunk and to compile also a 
variant..but none are good in lion with graphicsenabler= yes

i see always unsupported card!

the secondary problem is the ability to catch data from the bus.

i tried to use radeonpci and radrondump.

(1st those tools run only in 32bit
but the dump return the bios of the hd3000!!
even if lion detect and load ati kects and framebuffer, there is no 
way to force switching.
also ioreg reports.a lot of data.

Comment 4 by Cosmosis Jones, Oct 18, 2011

post your bdmesg and an ioreg -lw0

Comment 5 by ugo kind, Oct 18, 2011

here you are

Comment 6 by Cosmosis Jones, Oct 18, 2011

that was booted w/ GraphicsEnabler=YES?

Comment 7 by Cosmosis Jones, Oct 18, 2011

neermind i see you're using ATY_Init.
I need a bdmesg and ioreg from you using GraphicsEnabler=YES as 
well.
Thanks

Comment 8 by Cosmosis Jones, Oct 18, 2011

Owner: cosmo1t

Comment 9 by ugo kind, Oct 19, 2011

ok sorry I can write only at this time everyday
so I had atyinit for a test.. because if I use or not ATYinit I have 
the same output -> no info about any part of ATI display
Because of dualgpu.. everywhere is wrotten to use pcirootid=1 and 
GraophicsEnabler=No..

So I attach the reports in GE Enabled yes and in GE Enabled NO. BUT 
with yes ort no I have the same.-> no ati 

believeme.. my philosophy is to leave as untouched as possible the 
S/L/E.. and my HPDV6 (recent HW sandybridge i7) and others similar 
are able to start with less patches.
Then.. SATA/USB/CAM/PWRB/SLEEP are Ootb (sleep only if I patch 
dsdt/speedstep appleintelcpupowermanagement)

remains:audio-video

video is HD3000 (I see it fine in mac, always in the 16 combinations 
of :GEnabler:PCIrootid:Drivers:VBIOS)
and the utility igfxxxx detects it fine.

the only problem is that with ATI.C at boot (modded or last trunk) 
report : unsupported video card

as I told above
bye
ugo

Comment 10 by Cosmosis Jones, Oct 21, 2011

ah this is not just an ati.c issue.. This is the entire graphics 
enabler when u have more than 1 card. Just like when people have 2 
nvdias, or 1 ati and 1 nvidia, they have to do an EFI string or a 
DSDT patch for the graphics..

We'll still keep this open..

Comment 11 by ugo kind, Oct 21, 2011

k
i see
even if i consider it 
true/untrue
i mean: in windows.the ati catalyst driver has a newer and 
fundamental option: the power management: you can force using ati 
instead of intel hd3000
that mean..i suppose..that ati can drive bith
nd this is universal in all newer
 laptops..also apple
but
to complete the picture, there are models with thunderbolt 
engine..models with 6gb pcie and an external slot (ex an express 
card) and thosecheaper without any graphic expansion

my opinion is that there are no 2 vga..but 2 engines switchables
and it confirms the ioreg

bye

Comment 12 by MacMan, Oct 24, 2011

CHIP_FAMILY_RV730 for 6490M and 6470M was changed to 
CHIP_FAMILY_CAICOS in revision 1657.

Comment 13 by Cosmosis Jones, May 5, 2012

Status: Fixed

Created: 2 years 10 months ago by ugo kind

Updated: 2 years 3 months ago

Status: Fixed

Owner: Cosmosis Jones

Followed by: 2 persons

Labels:
Priority:Medium
Type:Defect