Bug 16122

Well, it's complaining that the underlying device's driver didn't call the secret, undocumented bdi_register().

Please describe the stack which underlies that ext3 filesystem so we can go hunt down the offending code.

My disk controller from lspci -nnk is

00:06.0 IDE interface [0101]: nVidia Corporation MCP67 IDE Controller [10de:0560] (rev a1)
        Subsystem: Hewlett-Packard Company Device [103c:30d6]
        Kernel driver in use: AMD_IDE

Is this what you need?

OK, so it's a simple ext3-on-ide setup.  No DM or loop or anything like that involved?

I do not think openSUSE 11.3 M7 uses a loop on boot. The dmesg output indicates that device-mapper was initialized 17 secs before the WARNING; however, I'm not using it. I will blacklist the DM modules and see if I still get the WARNING.

The demesg output just prior to the WARNING is

   11.996365] ssb: Sonics Silicon Backplane found on PCI device 0000:04:00.0
[   12.148321] hda: host max PIO5 wanted PIO255(auto-tune) selected PIO4
[   12.148661] hda: MWDMA2 mode selected
[   12.149177] ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
[   12.150932] ACPI: PCI Interrupt Link [LAZA] enabled at IRQ 20
[   12.150948] HDA Intel 0000:00:07.0: PCI INT A -> Link[LAZA] -> GSI 20 (level, low) -> IRQ 20
[   12.150953] hda_intel: Disable MSI for Nvidia chipset
[   12.151034] HDA Intel 0000:00:07.0: setting latency timer to 64
[   12.290805] cfg80211: Calling CRDA to update world regulatory domain
[   12.301154] cfg80211: World regulatory domain updated:
[   12.301159]     (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[   12.301163]     (2402000 KHz - 2472000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[   12.301166]     (2457000 KHz - 2482000 KHz @ 20000 KHz), (300 mBi, 2000 mBm)
[   12.301169]     (2474000 KHz - 2494000 KHz @ 20000 KHz), (300 mBi, 2000 mBm)
[   12.301171]     (5170000 KHz - 5250000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[   12.301174]     (5735000 KHz - 5835000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[   12.878617] b43-phy0: Broadcom 4312 WLAN found (core revision 15)
[   12.924045] b43-phy0 debug: Found PHY: Analog 6, Type 5, Revision 1
[   12.924063] b43-phy0 debug: Found Radio: Manuf 0x17F, Version 0x2062, Revision 2
[   13.064213] ide-cd driver 5.00
[   13.067134] ide-cd: hda: ATAPI 24X DVD-ROM DVD-R/RAM CD-R/RW drive, 2048kB Cache
[   13.067143] Uniform CD-ROM driver Revision: 3.20
[   13.080677] phy0: Selected rate control algorithm 'minstrel'
[   13.084263] Registered led device: b43-phy0::tx
[   13.084435] Registered led device: b43-phy0::rx
[   13.084603] Registered led device: b43-phy0::radio
[   13.086455] Broadcom 43xx driver loaded [ Features: PMNLS, Firmware-ID: FW13 ]
[   13.275196] Adding 2104448k swap on /dev/sda5.  Priority:-1 extents:1 across:2104448k 
[   13.885398] device-mapper: ioctl: 4.17.0-ioctl (2010-03-05) initialised: dm-devel@redhat.com
[   30.302902] ------------[ cut here ]------------
[   30.302912] WARNING: at fs/fs-writeback.c:1142 __mark_inode_dirty+0x103/0x170()

This boot was using the 2.6.35-rc1 kernel from Linville's wireless-testing tree.

What does it say after the WARN trace?  This:


                                printk(KERN_ERR "bdi-%s not registered\n",
                                                                bdi->name);

It is "block". The messages are as follows:

[   30.303036] ---[ end trace a80f1425532f94da ]---
[   30.303039] bdi-block not registered
[   30.325691] loop: module loaded
[   30.427469] EXT3-fs: barriers not enabled
[   30.438818] kjournald starting.  Commit interval 5 seconds
[   30.439197] EXT3-fs (sda7): using internal journal

Well that's not much use - that's just the default value and all it tells us is that nobody called bdi_register().  Bah.

That add_disk() fails to check the bdi_register_dev() return value doesn't help.

Blah, over to you Jens!

I'm willing to work on this, but don't know where to start.

Is there a simple explanation of when and where bdi_register() should be called?

I put a test of the returned value of bdi_register_dev() in add_disk(), but never got a non-zero result.

BTW, is this snippet in bdi_register() correct?

    if (bdi->dev)   /* The driver needs to use separate queues per device */
           goto exit;

This smells like an error, but returns zero.

On Wednesday 09 June 2010, Larry Finger wrote:
> On 06/08/2010 05:10 PM, Rafael J. Wysocki wrote:
> > This message has been generated automatically as a part of a summary report
> > of recent regressions.
> > 
> > The following bug entry is on the current list of known regressions
> > from 2.6.34.  Please verify if it still should be listed and let the
> tracking team
> > know (either way).
> > 
> > 
> > Bug-Entry   : http://bugzilla.kernel.org/show_bug.cgi?id=16122
> > Subject             : 2.6.35-rc1: WARNING at fs/fs-writeback.c:1142
> __mark_inode_dirty+0x103/0x170
> > Submitter   : Larry Finger <Larry.Finger@lwfinger.net>
> > Date                : 2010-06-04 13:18 (5 days old)
> 
> Still present in 2.6.35-rc2.

This puzzles me... Larry, can you try .34? I don't think this is a regression from .34.

I have the 2.6.34 kernel from wireless-testing, which should be the same a mainline, except for the wireless part. I never see this WARNING with that kernel, only from 2.6.35-rc1 onward.

On Monday, June 21, 2010, Larry Finger wrote:
> On 06/20/2010 05:17 PM, Rafael J. Wysocki wrote:
> > This message has been generated automatically as a part of a summary report
> > of recent regressions.
> > 
> > The following bug entry is on the current list of known regressions
> > from 2.6.34.  Please verify if it still should be listed and let the
> tracking team
> > know (either way).
> > 
> > 
> > Bug-Entry   : http://bugzilla.kernel.org/show_bug.cgi?id=16122
> > Subject             : 2.6.35-rc1: WARNING at fs/fs-writeback.c:1142
> __mark_inode_dirty+0x103/0x170
> > Submitter   : Larry Finger <Larry.Finger@lwfinger.net>
> > Date                : 2010-06-04 13:18 (17 days old)
> 
> This is still present in rc3.

Handled-By : Jens Axboe <axboe@kernel.dk>

On Friday, July 09, 2010, Larry Finger wrote:
> On 07/08/2010 06:33 PM, Rafael J. Wysocki wrote:
> > This message has been generated automatically as a part of a summary report
> > of recent regressions.
> >
> > The following bug entry is on the current list of known regressions
> > from 2.6.34.  Please verify if it still should be listed and let the
> tracking team
> > know (either way).
> >
> >
> > Bug-Entry   : http://bugzilla.kernel.org/show_bug.cgi?id=16122
> > Subject             : 2.6.35-rc1: WARNING at fs/fs-writeback.c:1142
> __mark_inode_dirty+0x103/0x170
> > Submitter   : Larry Finger<Larry.Finger@lwfinger.net>
> > Date                : 2010-06-04 13:18 (35 days old)
> 
> This bug is still present in 2.6.35-rc4.

*** This bug has been marked as a duplicate of bug 16312 ***