--Signature_Tue__6_Dec_2005_07_23_25_-0500_9T9=t10mDDs6VUO1
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable

Hello,

> (II) resource ranges after preInit:
>  [0] 0 0 0x90000000 - 0x90003fff (0x4000) MS[B]
>  [1] 0 0 0x94000000 - 0x97ffffff (0x4000000) MS[B]
>  [2] -1 0 0xffffffff - 0xffffffff (0x1) MX[B]
>  [3] -1 0 0x00000000 - 0x00000000 (0x1) MX[B]
>  [4] -1 0 0x90020000 - 0x9003ffff (0x20000) MX[B](B)
>  [5] -1 0 0x90000000 - 0x90003fff (0x4000) MX[B](B)
>  [6] -1 0 0x94000000 - 0x97ffffff (0x4000000) MX[B](B)
>  [7] 0 0 0x000a0000 - 0x000affff (0x10000) MS[B](OprU)
>  [8] 0 0 0x000b0000 - 0x000b7fff (0x8000) MS[B](OprU)
>  [9] 0 0 0x000b8000 - 0x000bffff (0x8000) MS[B](OprU)
>  [10] 0 0 0x00000400 - 0x000004ff (0x100) IS[B]
>  [11] -1 0 0x0000ffff - 0x0000ffff (0x1) IX[B]
>  [12] -1 0 0x00000000 - 0x00000000 (0x1) IX[B]
>  [13] -1 0 0x00000400 - 0x000004ff (0x100) IX[B](B)
>  [14] 0 0 0x000003b0 - 0x000003bb (0xc) IS[B](OprU)
>  [15] 0 0 0x000003c0 - 0x000003df (0x20) IS[B](OprU)

> (WW) mapVidMem 90000000, 4000, 4 fd =3D 5
> (II) mapVidMem effe2000
> (WW) mapVidMem 94000000, 800000, 4 fd =3D 5
> (II) mapVidMem ef640000

> 17136 XFree86  CALL  mmap(0,0x800000,3,1,7,0,0,0x94000000)
> 17136 XFree86  RET   mmap -278659072/0xef640000
> 17136 XFree86  CALL  write(0,0x19f6fc4,0x17)
> 17136 XFree86  GIO   fd 0 wrote 23 bytes
>        "(II) mapVidMem ef640000"
> 17136 XFree86  RET   write 23/0x17
> 17136 XFree86  PSIG  SIGBUS caught handler=3D0x1847f34 mask=3D():
> code=3DBUS_OBJERR, addr=3D0x20ad0b8, trap=3D512) 17136 XFree86  CALL=20
> __sigaction_sigtramp(SIGBUS,0xffffdb00,0xffffdb20,0xefecc234,1) 17136
> XFree86  RET   __sigaction_sigtramp 0 17136 XFree86  CALL=20
> write(2,0x19f6fc4,0x70) 17136 XFree86  GIO   fd 2 wrote 112 bytes

So it dies right after mapping registers and the framebuffer.
Option "iBookHacks" is for Radeons only, it won't do anything with a
Rage 128.
As far as I can tell the SIGBUS happens when the driver is almost done
setting things up. You won't get a useful stack trace from the Xserver,
at least not from driver code - we need a monolithic Xserver for that.
The easiest way to build one is to put "#define DoLoadableServer NO"
into xsrc/xfree/xc/config/cf/host.def and run make World from
xsrc/xfree/xc. This will produce a single, big XFree86 binary containing
all the drivers and modules so gdb can resolve symbols in all of them.
Install it ( make install from xc/programs/Xserver is enough ) and see
if you can get a stack trace from it. We need to know where exactly it
borked.

I suspect the driver tries to read data from the BIOS and doesn't expect
an OpenFirmware image but that's really just a guess. Since I don't have
any hardware that's remotely similar I can't test that myself.

have fun
Michael

PS: let's take this to port-macppc, I'm pretty sure the issue is
mac-specific.

--Signature_Tue__6_Dec_2005_07_23_25_-0500_9T9=t10mDDs6VUO1
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (NetBSD)

iQEVAwUBQ5WCwcpnzkX8Yg2nAQIFogf/Vqhf7W/4PXYhFkYh8owJNFcmfSyOpb5I
Ha0/qm5SkPBo+nPFnBXr+1MbBLsByuVpqQPRFEbClv9RW0K8B0ouXz5cEEliECCa
2rXY/2INwyN6EPs4SwYY4mCxjGwbJorHhh5Twr9IB292paWRU64E87IVrmTLsHAE
1YW6STb6cj8bZ9e+RyM6qkR/BCgvjS1NRVOkKFHQqphgws/UnwzTPyTbnCa3aFLv
I9G35bQh92JIYIERp0GXHIJ9hl9kUH1UXWWDrwwwjgY46yTXRd6Jk81bQBBW4yUd
mS1Ql8zT3ucUkpJUXqcd0ERftZRLc/hVyehF68vhPRXNhiGYfQ9JbA==
=js5i
-----END PGP SIGNATURE-----

--Signature_Tue__6_Dec_2005_07_23_25_-0500_9T9=t10mDDs6VUO1--