2013-07-10 09:57:38

by Markus Trippelsdorf

Subject: EDAC lockdep splat

While debugging an unrelated issue I came across the following lockdep
splat:

...
Jul 10 11:38:00 x4 kernel: AMD64 EDAC driver v3.4.0
Jul 10 11:38:00 x4 kernel: EDAC amd64: DRAM ECC enabled.
Jul 10 11:38:00 x4 kernel: EDAC amd64: F10h detected (node 0).
Jul 10 11:38:00 x4 kernel: EDAC MC: DCT0 chip selects:
Jul 10 11:38:00 x4 kernel: EDAC amd64: MC: 0: 1024MB 1: 1024MB
Jul 10 11:38:00 x4 kernel: EDAC amd64: MC: 2: 1024MB 3: 1024MB
Jul 10 11:38:00 x4 kernel: EDAC amd64: MC: 4: 0MB 5: 0MB
Jul 10 11:38:00 x4 kernel: EDAC amd64: MC: 6: 0MB 7: 0MB
Jul 10 11:38:00 x4 kernel: EDAC MC: DCT1 chip selects:
Jul 10 11:38:00 x4 kernel: EDAC amd64: MC: 0: 1024MB 1: 1024MB
Jul 10 11:38:00 x4 kernel: EDAC amd64: MC: 2: 1024MB 3: 1024MB
Jul 10 11:38:00 x4 kernel: EDAC amd64: MC: 4: 0MB 5: 0MB
Jul 10 11:38:00 x4 kernel: EDAC amd64: MC: 6: 0MB 7: 0MB
Jul 10 11:38:00 x4 kernel: EDAC amd64: using x4 syndromes.
Jul 10 11:38:00 x4 kernel: EDAC amd64: MCT channel count: 2
Jul 10 11:38:00 x4 kernel: EDAC amd64: CS0: Unbuffered DDR3 RAM
Jul 10 11:38:00 x4 kernel: EDAC amd64: CS1: Unbuffered DDR3 RAM
Jul 10 11:38:00 x4 kernel: EDAC amd64: CS2: Unbuffered DDR3 RAM
Jul 10 11:38:00 x4 kernel: EDAC amd64: CS3: Unbuffered DDR3 RAM
Jul 10 11:38:00 x4 kernel: BUG: key ffff8802169693f8 not in .data!
Jul 10 11:38:00 x4 kernel: ------------[ cut here ]------------
Jul 10 11:38:00 x4 kernel: WARNING: CPU: 3 PID: 1 at kernel/lockdep.c:2987 lockdep_init_map+0x45e/0x490()
Jul 10 11:38:00 x4 kernel: sda: unknown partition table
Jul 10 11:38:00 x4 kernel: DEBUG_LOCKS_WARN_ON(1)
Jul 10 11:38:00 x4 kernel: CPU: 3 PID: 1 Comm: swapper/0 Not tainted 3.10.0-08587-g496322b #34
Jul 10 11:38:00 x4 kernel: Hardware name: System manufacturer System Product Name/M4A78T-E, BIOS 3503 04/13/2011
Jul 10 11:38:00 x4 kernel: 0000000000000009 ffff880216885ac8 ffffffff815a9e66 ffff880216885b10
Jul 10 11:38:00 x4 kernel: ffff880216885b00 ffffffff810627ad ffff8802155b1910 ffff8802169693f8
Jul 10 11:38:00 x4 kernel: sd 1:0:0:0: [sda] Attached SCSI disk
Jul 10 11:38:00 x4 kernel: 0000000000000000 0000000000000000 ffff88021559c400 ffff880216885b60
Jul 10 11:38:00 x4 kernel: Call Trace:
Jul 10 11:38:00 x4 kernel: [<ffffffff815a9e66>] dump_stack+0x45/0x56
Jul 10 11:38:00 x4 kernel: [<ffffffff810627ad>] warn_slowpath_common+0x7d/0xb0
Jul 10 11:38:00 x4 kernel: [<ffffffff81062857>] warn_slowpath_fmt+0x47/0x50
Jul 10 11:38:00 x4 kernel: [<ffffffff810a588e>] lockdep_init_map+0x45e/0x490
Jul 10 11:38:00 x4 kernel: [<ffffffff810a3096>] debug_mutex_init+0x36/0x50
Jul 10 11:38:00 x4 kernel: [<ffffffff8108526d>] __mutex_init+0x5d/0x70
Jul 10 11:38:00 x4 kernel: [<ffffffff813c8235>] bus_register+0x175/0x2e0
Jul 10 11:38:00 x4 kernel: [<ffffffff8146c3d3>] edac_create_sysfs_mci_device+0x53/0x530
Jul 10 11:38:00 x4 kernel: [<ffffffff8146a303>] edac_mc_add_mc+0xf3/0x250
Jul 10 11:38:00 x4 kernel: [<ffffffff81471424>] amd64_probe_one_instance+0x734/0xfa0
Jul 10 11:38:00 x4 kernel: [<ffffffff8126182c>] pci_device_probe+0x9c/0xe0
Jul 10 11:38:00 x4 kernel: [<ffffffff813c90a1>] driver_probe_device+0x71/0x230
Jul 10 11:38:00 x4 kernel: [<ffffffff813c9363>] __driver_attach+0xa3/0xb0
Jul 10 11:38:00 x4 kernel: [<ffffffff813c92c0>] ? __device_attach+0x60/0x60
Jul 10 11:38:00 x4 kernel: [<ffffffff813c7403>] bus_for_each_dev+0x63/0xa0
Jul 10 11:38:00 x4 kernel: [<ffffffff813c8fb9>] driver_attach+0x19/0x20
Jul 10 11:38:00 x4 kernel: [<ffffffff813c7ee0>] bus_add_driver+0x1d0/0x260
Jul 10 11:38:00 x4 kernel: [<ffffffff81ab62f6>] ? mce_amd_init+0x15d/0x15d
Jul 10 11:38:00 x4 kernel: [<ffffffff813c9921>] driver_register+0x71/0x150
Jul 10 11:38:00 x4 kernel: [<ffffffff81255206>] ? __raw_spin_lock_init+0x36/0x60
Jul 10 11:38:00 x4 kernel: [<ffffffff81ab62f6>] ? mce_amd_init+0x15d/0x15d
Jul 10 11:38:00 x4 kernel: [<ffffffff8126190b>] __pci_register_driver+0x5b/0x60
Jul 10 11:38:00 x4 kernel: [<ffffffff81ab63a5>] amd64_edac_init+0xaf/0x191
Jul 10 11:38:00 x4 kernel: [<ffffffff81ab62f6>] ? mce_amd_init+0x15d/0x15d
Jul 10 11:38:00 x4 kernel: [<ffffffff81a95d61>] do_one_initcall+0x7f/0x10d
Jul 10 11:38:00 x4 kernel: [<ffffffff81080838>] ? parse_args+0x1f8/0x330
Jul 10 11:38:00 x4 kernel: [<ffffffff81a95efc>] kernel_init_freeable+0x10d/0x18d
Jul 10 11:38:00 x4 kernel: [<ffffffff81a957ad>] ? do_early_param+0x88/0x88
Jul 10 11:38:00 x4 kernel: [<ffffffff815a1cb0>] ? rest_init+0xc0/0xc0
Jul 10 11:38:00 x4 kernel: [<ffffffff815a1cb9>] kernel_init+0x9/0x180
Jul 10 11:38:00 x4 kernel: [<ffffffff815b20ac>] ret_from_fork+0x7c/0xb0
Jul 10 11:38:00 x4 kernel: [<ffffffff815a1cb0>] ? rest_init+0xc0/0xc0
Jul 10 11:38:00 x4 kernel: ---[ end trace 7de9b5e7ff84ed1b ]---
Jul 10 11:38:00 x4 kernel: EDAC MC0: Giving out device to 'amd64_edac' 'F10h': DEV 0000:00:18.2
Jul 10 11:38:00 x4 kernel: EDAC PCI0: Giving out device to module 'amd64_edac' controller 'EDAC PCI controller': DEV '0000:00:18.2' (POLLED)
...

--
Markus

2013-07-10 10:39:33

by Borislav Petkov

Subject: Re: EDAC lockdep splat

On Wed, Jul 10, 2013 at 11:57:33AM +0200, Markus Trippelsdorf wrote:
> While debugging an unrelated issue I came across the following lockdep
> splat:

…

> Jul 10 11:38:00 x4 kernel: BUG: key ffff8802169693f8 not in .data!
> Jul 10 11:38:00 x4 kernel: ------------[ cut here ]------------
> Jul 10 11:38:00 x4 kernel: WARNING: CPU: 3 PID: 1 at kernel/lockdep.c:2987 lockdep_init_map+0x45e/0x490()
> Jul 10 11:38:00 x4 kernel: sda: unknown partition table
> Jul 10 11:38:00 x4 kernel: DEBUG_LOCKS_WARN_ON(1)

I know: http://marc.info/?l=linux-kernel&m=136809709202740&w=2

I was expecting Mauro to fix it since his patches grew this out:

commit 7a623c039075e4ea21648d88133fafa6dcfd113d
Author: Mauro Carvalho Chehab <[email protected]>
Date: Mon Apr 16 16:41:11 2012 -0300

edac: rewrite the sysfs code to use struct device

adds struct bus_type to struct mem_ctl_info which is dynamically
allocated and that's why lockdep complains.

bus_register() needs to get a statically allocated struct bus_type as
the rest of the kernel does it.

--
Regards/Gruss,
Boris.

Sent from a fat crate under my desk. Formatting is fine.
--

2013-07-10 10:41:32

by Borislav Petkov

Subject: Re: EDAC lockdep splat

Forgot to CC him.

On Wed, Jul 10, 2013 at 12:39:15PM +0200, Borislav Petkov wrote:
> On Wed, Jul 10, 2013 at 11:57:33AM +0200, Markus Trippelsdorf wrote:
> > While debugging an unrelated issue I came across the following lockdep
> > splat:
>
> …
>
> > Jul 10 11:38:00 x4 kernel: BUG: key ffff8802169693f8 not in .data!
> > Jul 10 11:38:00 x4 kernel: ------------[ cut here ]------------
> > Jul 10 11:38:00 x4 kernel: WARNING: CPU: 3 PID: 1 at kernel/lockdep.c:2987 lockdep_init_map+0x45e/0x490()
> > Jul 10 11:38:00 x4 kernel: sda: unknown partition table
> > Jul 10 11:38:00 x4 kernel: DEBUG_LOCKS_WARN_ON(1)
>
> I know: http://marc.info/?l=linux-kernel&m=136809709202740&w=2
>
> I was expecting Mauro to fix it since his patches grew this out:
>
> commit 7a623c039075e4ea21648d88133fafa6dcfd113d
> Author: Mauro Carvalho Chehab <[email protected]>
> Date: Mon Apr 16 16:41:11 2012 -0300
>
> edac: rewrite the sysfs code to use struct device
>
> adds struct bus_type to struct mem_ctl_info which is dynamically
> allocated and that's why lockdep complains.
>
> bus_register() needs to get a statically allocated struct bus_type as
> the rest of the kernel does it.

--
Regards/Gruss,
Boris.

Sent from a fat crate under my desk. Formatting is fine.
--

2013-07-10 11:07:00

by Mauro Carvalho Chehab

Subject: Re: EDAC lockdep splat

Hi Boris/Markus,

Em Wed, 10 Jul 2013 12:41:15 +0200
Borislav Petkov <[email protected]> escreveu:

> Forgot to CC him.
>
> On Wed, Jul 10, 2013 at 12:39:15PM +0200, Borislav Petkov wrote:
> > On Wed, Jul 10, 2013 at 11:57:33AM +0200, Markus Trippelsdorf wrote:
> > > While debugging an unrelated issue I came across the following lockdep
> > > splat:
> >
> > …
> >
> > > Jul 10 11:38:00 x4 kernel: BUG: key ffff8802169693f8 not in .data!
> > > Jul 10 11:38:00 x4 kernel: ------------[ cut here ]------------
> > > Jul 10 11:38:00 x4 kernel: WARNING: CPU: 3 PID: 1 at kernel/lockdep.c:2987 lockdep_init_map+0x45e/0x490()
> > > Jul 10 11:38:00 x4 kernel: sda: unknown partition table
> > > Jul 10 11:38:00 x4 kernel: DEBUG_LOCKS_WARN_ON(1)
> >
> > I know: http://marc.info/?l=linux-kernel&m=136809709202740&w=2
> >
> > I was expecting Mauro to fix it since his patches grew this out:
> >
> > commit 7a623c039075e4ea21648d88133fafa6dcfd113d
> > Author: Mauro Carvalho Chehab <[email protected]>
> > Date: Mon Apr 16 16:41:11 2012 -0300
> >
> > edac: rewrite the sysfs code to use struct device
> >
> > adds struct bus_type to struct mem_ctl_info which is dynamically
> > allocated and that's why lockdep complains.
> >
> > bus_register() needs to get a statically allocated struct bus_type as
> > the rest of the kernel does it.
>

I'm currently without any systems where I could try to reproduce it, as
I moved to a new job at Samsung, and I'm still in the process of getting
access to the corporate facilities and to get the needed hardware.
The hardware I currently own doesn't support EDAC.

So, I doubt that I would be able to touch on that in the next couple
weeks or so.

If you're urging for a fix, feel free to propose a patch.
Otherwise, I'll take care of it as soon as I receive the hardware.

Regards,
Mauro