Prev: cris: gpio: do not call copy_to_user()/copy_from_user() while holding spinlocks
Next: when does a program become processed ? static void unknown_nmi_error
From: FUJITA Tomonori on 2 Aug 2010 08:00
CC'ed to dma-debug maintainer.

On Mon, 02 Aug 2010 16:21:09 +0530
Subrata Modak <subrata(a)linux.vnet.ibm.com> wrote:

> Hi,
>
> On boot, Badness at lib/dma-debug.c:902, Call Trace & Instruction Dump
> are recorded at /var/log/messages:
>
> ================================================================
> udev: starting version 151
> ses 0:8:0:0: Attached Enclosure device
> IBM eHEA ethernet device driver (Release EHEA_0105)
> mlx4_core: Mellanox ConnectX core driver v0.01 (May 1, 2007)
> mlx4_core: Initializing 0002:01:00.0
> mlx4_core 0002:01:00.0: enabling device (0140 -> 0142)
> ehea: eth0: Jumbo frames are enabled
> ehea: eth0 -> logical port id #1
> ehea: eth1: Jumbo frames are enabled
> ehea: eth1 -> logical port id #2
> mlx4_core 0002:01:00.0: Requested 17 vectors, but only 8 MSI-X vectors
> available, trying again
> mlx4_core 0002:01:00.0: DMA-API: device driver tries to sync DMA memory it has
> not allocated [device address=0x0000000060f22000] [size=4096 bytes]
> ------------[ cut here ]------------
> Badness at lib/dma-debug.c:902
> NIP: c0000000003fdfa0 LR: c0000000003fdf9c CTR: 0000000000000001
> REGS: c000000f35bfad00 TRAP: 0700 Not tainted (2.6.35)
> MSR: 8000000000029032 <EE,ME,CE,IR,DR> CR: 48002482 XER: 20000010
> TASK = c000000f35c08000[502] 'modprobe' THREAD: c000000f35bf8000 CPU: 8
> GPR00: c0000000003fdf9c c000000f35bfaf80 c000000001464c48 0000000000000096
> GPR04: 0000000000000001 c0000000000c19b0 0000000000000000 0000000000000002
> GPR08: 0000000000000000 c000000f35c08000 00000000000043b3 0000000000000001
> GPR12: 7669636520616464 c000000003fa5800 0000000000000000 0000000010016628
> GPR16: c000000f32a02000 c000000f32402000 c000000f33604648 c000000f35bfb1c0
> GPR20: c000000f38002120 0000000060f22000 0000000000000200 c000000001f4b000
> GPR24: 0000000000000001 0000000000000001 c000000001f47900 c000000f35bfb0b0
> GPR28: 0000000000000000 c000000f38002120 c0000000013eaea8 c000000f35bfaf80
> NIP [c0000000003fdfa0] .check_sync+0x108/0x52c
> LR [c0000000003fdf9c] .check_sync+0x104/0x52c
> Call Trace:
> [c000000f35bfaf80] [c0000000003fdf9c] .check_sync+0x104/0x52c (unreliable)
> [c000000f35bfb040] [c0000000003fe8d4] .debug_dma_sync_single_for_cpu+0x58/0x70

I guess that this driver does a partial sync with
dma_sync_single_for_* API. dma-debug can't handle it properly. It's
likely that this is a false warning.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Roedel, Joerg on 4 Aug 2010 09:20
On Mon, Aug 02, 2010 at 07:55:03AM -0400, FUJITA Tomonori wrote:
> I guess that this driver does a partial sync with
> dma_sync_single_for_* API. dma-debug can't handle it properly. It's
> likely that this is a false warning.

If this turns out to be true it is not trivial to fix. I prepare a patch
to test for you.

Joerg

--
AMD Operating System Research Center

Advanced Micro Devices GmbH Einsteinring 24 85609 Dornach
General Managers: Alberto Bozzo, Andrew Bowd
Registration: Dornach, Landkr. Muenchen; Registerger. Muenchen, HRB Nr. 43632

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: FUJITA Tomonori on 4 Aug 2010 10:20
On Wed, 4 Aug 2010 15:16:34 +0200
"Roedel, Joerg" <Joerg.Roedel(a)amd.com> wrote:

> On Mon, Aug 02, 2010 at 07:55:03AM -0400, FUJITA Tomonori wrote:
> > I guess that this driver does a partial sync with
> > dma_sync_single_for_* API. dma-debug can't handle it properly. It's
> > likely that this is a false warning.
>
> If this turns out to be true it is not trivial to fix. I prepare a patch
> to test for you.

I've not looked at the details of this driver, but there are drivers
that do such. So dma-debug needs to be fixed anyway; you can't assume
that a DMA address that dma_map_single returned is passed to
dma_sync_single_for API.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
 | 
Pages: 1
Prev: cris: gpio: do not call copy_to_user()/copy_from_user() while holding spinlocks
Next: when does a program become processed ? static void unknown_nmi_error