gentoo-2.6.15-r1 - Kernel oops when removing then reinstalling DVD drive into Pismo mediabay

[Benjamin Herrenschmidt]

On Fri, 2006-05-12 at 21:09 -0400, Jordan Herring wrote:
> This is using the 2.6.15 kernel with the Gentoo patches. The 2.6.16
> kernel also exhibits this behavior.

Looks like something broke/changed in the IDE code that cause it to blow
when the media-bay re-registers the interface. I'll need to dig in that
code, it's not trivial. Will let you know.

> Mar  4 08:48:50 zetagundam mediabay0: switching to 3
> Mar  4 08:48:50 zetagundam mediabay0: powering up
> Mar  4 08:48:50 zetagundam mediabay0: enabling (kind:3)
> Mar  4 08:48:50 zetagundam mediabay0: waiting reset (kind:3)
> Mar  4 08:48:51 zetagundam mediabay0: waiting IDE reset (kind:3)
> Mar  4 08:48:52 zetagundam mediabay0: waiting IDE ready (kind:3)
> Mar  4 08:48:52 zetagundam mediabay 0, registering IDE...
> Mar  4 08:48:52 zetagundam Probing IDE interface ide2...
> Mar  4 08:48:52 zetagundam hde: MATSHITADVD-ROM SR-8173, ATAPI CD/DVD-ROM drive 
> Mar  4 08:48:52 zetagundam Oops: kernel access of bad area, sig: 11 [#1] 
> Mar  4 08:48:52 zetagundam NIP: C010EF6C LR: C010E1F4 SP: CFEE5EB0 REGS: cfee5e00 TRAP: 0300    Not tainted 
> Mar  4 08:48:52 zetagundam MSR: 00009032 EE: 1 PR: 0 FP: 0 ME: 1 IR/DR: 11
> Mar  4 08:48:52 zetagundam DAR: 220000A2, DSISR: 40000000 
> Mar  4 08:48:52 zetagundam TASK = cfe387f0[721] 'media-bay' THREAD: cfee4000
> Mar  4 08:48:52 zetagundam Last syscall: -1 
> Mar  4 08:48:52 zetagundam GPR00: C016BC14 CFEE5EB0 CFE387F0 220000A2 C0373D50 C016BD14 C0373D4C 00000000 
> Mar  4 08:48:52 zetagundam GPR08: C02DAC53 FFFFFFFF 00004000 C0373DC0 82004088 00000000 00000000 00000000 
> Mar  4 08:48:52 zetagundam GPR16: 00000000 00000000 00000000 00000000 00230000 00000000 00000000 00009032 
> Mar  4 08:48:52 zetagundam GPR24: C0310000 00000014 00000000 FFFFFFEA C0373768 C0373D4C C0373D18 2200008A 
> Mar  4 08:48:52 zetagundam NIP [c010ef6c] kref_get+0x0/0x24 
> Mar  4 08:48:52 zetagundam LR [c010e1f4] kobject_get+0x20/0x38 
> Mar  4 08:48:52 zetagundam Call trace:
> Mar  4 08:48:52 zetagundam [c016bc14] get_device+0x20/0x38 
> Mar  4 08:48:52 zetagundam [c016bde4] device_add+0x48/0x170 
> Mar  4 08:48:52 zetagundam [c019663c] probe_hwif+0x600/0x780 
> Mar  4 08:48:52 zetagundam [c0197374] probe_hwif_init_with_fixup+0x1c/0xac 
> Mar  4 08:48:52 zetagundam [c018f688] ide_register_hw_with_fixup+0x198/0x1dc 
> Mar  4 08:48:52 zetagundam [c0184e24] media_bay_step+0x3b0/0x54c 
> Mar  4 08:48:52 zetagundam [c0185054] media_bay_task+0x94/0xf8 
> Mar  4 08:48:52 zetagundam [c00069bc] kernel_thread+0x44/0x60
> 
> Basically, what happens is that if I pull the drive and then reinsert
> it (doesn't matter how long) it immediately does this and the keyboard
> seizes up. If I ssh in and kill the process that was on that console,
> the kernel will continue to run, if I recall correctly. (Of course,
> being a laptop, this is rather impractical most of the time.) Usually
> when this happens, the drive is *not* mounted at all.