2.6.14-mm1 doesnt bootup on PPC64

[linas]

On Tue, Nov 08, 2005 at 11:58:54AM +0530, Srivatsa Vaddagiri was heard to remark:
> On Tue, Nov 08, 2005 at 12:16:33AM +1030, Alan Modra wrote:
> > Compiler version?  http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24644
> > might be relevant.
> 
> I tried with both gcc 3.4.3 (RHEL 4) and gcc 3.3.3 (SLES 9). Both result
> in a non-booting kernel. This is with ppc64's 'defconfig' btw.

Not the only problem.

-- If pci hotplug for ppc64 is enabled, the kernel won't compile,
   misc symbols missing related to of_pci_whatever()
 
   (I don't have the messages in front of me; I notified John Rose)

-- w/ hotplug disabled, the kernel boots, but spews a bunch of these:

[360420.236575] Brought up 4 CPUs
[360420.236578] softlockup thread 3 started up.
[360420.238334] NET: Registered protocol family 16
[360420.239497] PCI: Probing PCI hardware
<3>Badness in kref_get at lib/kref.c:32
[360420.239622] Badness in kref_get at lib/kref.c:32
[360420.239632] Call Trace:
[360420.239640] [C000000002503720] [C00000000002F2AC] .show_stack+0x5c/0x1cc (unreliable)
[360420.239665] [C0000000025037D0] [C00000000043D1D4] .program_check_exception+0x34c/0x62c
[360420.239687] [C000000002503880] [C000000000004348] program_check_common+0xc8/0x100
[360420.239706] --- Exception: 700 at .kref_get+0xc/0x24
[360420.239721]     LR = .kobject_get+0x24/0x40
[360420.239731] [C000000002503B70] [0000000000000000] 0x0 (unreliable)
[360420.239749] [C000000002503BF0] [C000000000287CB8] .get_device+0x24/0x40
[360420.239767] [C000000002503C70] [C000000000204458] .pci_dev_get+0x24/0x40
[360420.239785] [C000000002503CF0] [C00000000020038C] .pci_device_add+0x40/0xc4
[360420.239803] [C000000002503D80] [C000000000017420] .of_scan_bus+0x2f8/0x698
[360420.239822] [C000000002503E50] [C0000000004FE14C] .pcibios_init+0x360/0x364
[360420.239842] [C000000002503F00] [C000000000009584] .init+0x1e8/0x44c
[360420.239859] [C000000002503F90] [C00000000000A4E4] .kernel_thread+0x4c/0x68
<3>Badness in kref_get at lib/kref.c:32
[360420.240093] Badness in kref_get at lib/kref.c:32

and it later crashes with 

cpu 0x0: Vector: 300 (Data Access) at [c000000002503770]
    pc: c00000000006eb30: .add_wait_queue_exclusive+0x4c/0x74
    lr: c00000000006eb10: .add_wait_queue_exclusive+0x2c/0x74
    sp: c0000000025039f0
   msr: 8000000000001032
   dar: 0
 dsisr: 42000000
  current = 0xc000000073fb87e0
  paca    = 0xc000000000565000
    pid   = 1, comm = swapper
enter ? for help
0:mon> bt
   type            address
0:mon> t
[c0000000025039f0] c0000000000a2f1c .alloc_page_interleave+0x3c/0xb4
(unreliable)
[c000000002503a80] c00000000043899c .__down+0x5c/0x11c
[c000000002503b50] c00000000028a988 .device_attach+0xd4/0xe8
[c000000002503be0] c000000000289b08 .bus_add_device+0x60/0x1c8
[c000000002503c90] c000000000288110 .device_add+0x140/0x204
[c000000002503d30] c0000000001ff228 .pci_bus_add_device+0x24/0x8c
[c000000002503dc0] c0000000001ff2f4 .pci_bus_add_devices+0x64/0x164
[c000000002503e50] c0000000004fdf40 .pcibios_init+0x154/0x364
[c000000002503f00] c000000000009584 .init+0x1e8/0x44c
[c000000002503f90] c00000000000a4e4 .kernel_thread+0x4c/0x68
0:mon>

I'm punting on this for today; I'm guessing there are 
some missing or misapplied patches somewhere.

--linas