Date:	Tue, 28 Oct 2008 13:38:11 -0400 (EDT)
From:	Alan Stern <stern@...land.harvard.edu>
To:	Luciano Rocha <luciano@...otux.com>
cc:	James Bottomley <James.Bottomley@...senPartnership.com>,
	"Rafael J. Wysocki" <rjw@...k.pl>,
	Linux-Kernel <linux-kernel@...r.kernel.org>,
	USB list <linux-usb@...r.kernel.org>,
	SCSI development list <linux-scsi@...r.kernel.org>
Subject: Re: usb hdd problems with 2.6.27.2

On Tue, 28 Oct 2008, Luciano Rocha wrote:

> > 
> > Okay, I tried the same sort of experiment in 2.6.27.4 and got the same
> > result as you did.  I was able to fix it by applying the 8bfa24727
> > commit mentioned earlier together with the patch below.
> 
> I tried it, but I get some I/O errors with it.

That's to be expected.  These patches aren't supposed to prevent I/O 
errors, they are merely supposed to prevent the errors from continuing 
indefinitely.

> With 8bfa24727 and this patch, I get:
> 
> [   94.289819] usb 1-2.1.1: new high speed USB device using ehci_hcd and address 8
> [   94.493891] usb 1-2.1.1: configuration #1 chosen from 1 choice
> [   94.537884] usbcore: registered new interface driver libusual
> [   94.611888] Initializing USB Mass Storage driver...
> [   94.612704] scsi2 : SCSI emulation for USB Mass Storage devices
> [   94.619835] usbcore: registered new interface driver usb-storage
> [   94.619891] USB Mass Storage support registered.
> [   94.622190] usb-storage: device found at 8
> [   94.622232] usb-storage: waiting for device to settle before scanning
> [   99.622933] scsi 2:0:0:0: Direct-Access     SAMSUNG  HD400LD          WQ10 PQ: 0 ANSI: 0
> [   99.627569] usb-storage: device scan complete
> [   99.657746] Driver 'sd' needs updating - please use bus_type methods
> [   99.661428] sd 2:0:0:0: [sda] 781422769 512-byte hardware sectors (400088 MB)
> [   99.662551] sd 2:0:0:0: [sda] Write Protect is off
> [   99.662662] sd 2:0:0:0: [sda] Mode Sense: 03 00 00 00
> [   99.662702] sd 2:0:0:0: [sda] Assuming drive cache: write through
> [   99.666582] sd 2:0:0:0: [sda] 781422769 512-byte hardware sectors (400088 MB)
> [   99.672877] sd 2:0:0:0: [sda] Write Protect is off
> [   99.672884] sd 2:0:0:0: [sda] Mode Sense: 03 00 00 00
> [   99.672886] sd 2:0:0:0: [sda] Assuming drive cache: write through
> [   99.672890]  sda: sda1
> [   99.683592] sd 2:0:0:0: [sda] Attached SCSI disk
> [   99.704160] sd 2:0:0:0: Attached scsi generic sg0 type 0
> [   99.808029] sd 2:0:0:0: [sda] Sense Key : No Sense [current] 
> [   99.808040] sd 2:0:0:0: [sda] Add. Sense: No additional sense information
> [   99.808051] end_request: I/O error, dev sda, sector 781422768
> [   99.808056] Buffer I/O error on device sda, logical block 781422768
> [   99.837771] sd 2:0:0:0: [sda] Sense Key : No Sense [current] 
> [   99.837780] sd 2:0:0:0: [sda] Add. Sense: No additional sense information
> [   99.837790] end_request: I/O error, dev sda, sector 781422768
> [   99.837794] Buffer I/O error on device sda, logical block 781422768
> [   99.838775] end_request: I/O error, dev sda, sector 781422761
> [   99.838781] Buffer I/O error on device sda, logical block 781422761
> [   99.838784] Buffer I/O error on device sda, logical block 781422762
> [   99.838786] Buffer I/O error on device sda, logical block 781422763
> [   99.838789] Buffer I/O error on device sda, logical block 781422764
> [   99.838791] Buffer I/O error on device sda, logical block 781422765
> [   99.838793] Buffer I/O error on device sda, logical block 781422766
> [   99.838796] Buffer I/O error on device sda, logical block 781422767
> [   99.839525] end_request: I/O error, dev sda, sector 781422762
> [   99.839530] Buffer I/O error on device sda, logical block 781422762
> [   99.847527] sd 2:0:0:0: [sda] Sense Key : No Sense [current] 
> [   99.847539] sd 2:0:0:0: [sda] Add. Sense: No additional sense information
> [   99.847554] end_request: I/O error, dev sda, sector 781422768
> [   99.864923] sd 2:0:0:0: [sda] Sense Key : No Sense [current] 
> [   99.864942] sd 2:0:0:0: [sda] Add. Sense: No additional sense information
> [   99.864967] end_request: I/O error, dev sda, sector 781422768
> [   99.883038] sd 2:0:0:0: [sda] Sense Key : No Sense [current] 
> [   99.883047] sd 2:0:0:0: [sda] Add. Sense: No additional sense information
> [   99.883058] end_request: I/O error, dev sda, sector 781422768
> [   99.884239] end_request: I/O error, dev sda, sector 781422763
> [   99.901645] sd 2:0:0:0: [sda] Sense Key : No Sense [current] 
> [   99.901655] sd 2:0:0:0: [sda] Add. Sense: No additional sense information
> [   99.901665] end_request: I/O error, dev sda, sector 781422768
> [   99.925027] sd 2:0:0:0: [sda] Sense Key : No Sense [current] 
> [   99.925036] sd 2:0:0:0: [sda] Add. Sense: No additional sense information
> [   99.925046] end_request: I/O error, dev sda, sector 781422768
> [  119.957384] kjournald starting.  Commit interval 5 seconds
> [  119.962246] EXT3 FS on dm-5, internal journal
> [  119.962267] EXT3-fs: mounted filesystem with ordered data mode.

And the errors did stop, so the patch did what it was supposed to do.

> Oddly enough, with the other patch you sent me fixing the reported
> number of sectors (*), on top of this one, I also get I/O errors:
> 
> 
> [  127.047323] usb 1-2.1.1: new high speed USB device using ehci_hcd and address 13
> [  127.240694] usb 1-2.1.1: configuration #1 chosen from 1 choice
> [  127.241178] scsi3 : SCSI emulation for USB Mass Storage devices
> [  127.241752] usb-storage: device found at 13
> [  127.241924] usb-storage: waiting for device to settle before scanning
> [  132.242710] scsi 3:0:0:0: Direct-Access     SAMSUNG  HD400LD          WQ10 PQ: 0 ANSI: 0
> [  132.244195] sd 3:0:0:0: [sda] 781422768 512-byte hardware sectors (400088 MB)
> [  132.245070] sd 3:0:0:0: [sda] Write Protect is off
> [  132.245162] sd 3:0:0:0: [sda] Mode Sense: 03 00 00 00
> [  132.245190] sd 3:0:0:0: [sda] Assuming drive cache: write through
> [  132.245929] sd 3:0:0:0: [sda] 781422768 512-byte hardware sectors (400088 MB)
> [  132.246686] sd 3:0:0:0: [sda] Write Protect is off
> [  132.246691] sd 3:0:0:0: [sda] Mode Sense: 03 00 00 00
> [  132.246693] sd 3:0:0:0: [sda] Assuming drive cache: write through
> [  132.246696]  sda: sda1
> [  132.272924] sd 3:0:0:0: [sda] Attached SCSI disk
> [  132.273058] sd 3:0:0:0: Attached scsi generic sg0 type 0
> [  132.273297] usb-storage: device scan complete
> [  132.325802] end_request: I/O error, dev sda, sector 781422761
> [  132.325815] __ratelimit: 6 callbacks suppressed
> [  132.325818] Buffer I/O error on device sda, logical block 97677845
> [  132.327676] end_request: I/O error, dev sda, sector 781422761
> [  132.327689] Buffer I/O error on device sda, logical block 97677845
> [  132.328302] end_request: I/O error, dev sda, sector 781422761
> [  132.328314] Buffer I/O error on device sda, logical block 97677845
> [  132.328926] end_request: I/O error, dev sda, sector 781422761
> [  132.328938] Buffer I/O error on device sda, logical block 97677845
> [  132.329551] end_request: I/O error, dev sda, sector 781422761
> [  132.329557] Buffer I/O error on device sda, logical block 97677845
> [  132.330176] end_request: I/O error, dev sda, sector 781422761
> [  132.330188] Buffer I/O error on device sda, logical block 97677845
> [  132.330800] end_request: I/O error, dev sda, sector 781422761
> [  132.330807] Buffer I/O error on device sda, logical block 97677845
> [  132.345053] end_request: I/O error, dev sda, sector 781422761
> [  132.345065] Buffer I/O error on device sda, logical block 97677845
> [  132.345926] end_request: I/O error, dev sda, sector 781422761
> [  132.345938] Buffer I/O error on device sda, logical block 97677845
> [  136.743629] kjournald starting.  Commit interval 5 seconds
> [  136.744951] EXT3 FS on dm-5, internal journal
> [  136.745039] EXT3-fs: mounted filesystem with ordered data mode.

That's a little strange, but not unheard of.

> Without this patch, but still with the (*) one, everything seems fine:
> 
> [   90.680299] usb 1-2.1.1: new high speed USB device using ehci_hcd and address 8
> [   90.873674] usb 1-2.1.1: configuration #1 chosen from 1 choice
> [   91.493054] usbcore: registered new interface driver libusual
> [   91.582207] Initializing USB Mass Storage driver...
> [   91.582696] scsi2 : SCSI emulation for USB Mass Storage devices
> [   91.584473] usbcore: registered new interface driver usb-storage
> [   91.584582] USB Mass Storage support registered.
> [   91.585032] usb-storage: device found at 8
> [   91.585039] usb-storage: waiting for device to settle before scanning
> [   96.584772] scsi 2:0:0:0: Direct-Access     SAMSUNG  HD400LD          WQ10 PQ: 0 ANSI: 0
> [   96.585058] usb-storage: device scan complete
> [   96.623106] Driver 'sd' needs updating - please use bus_type methods
> [   96.624120] sd 2:0:0:0: [sda] 781422768 512-byte hardware sectors (400088 MB)
> [   96.624869] sd 2:0:0:0: [sda] Write Protect is off
> [   96.624876] sd 2:0:0:0: [sda] Mode Sense: 03 00 00 00
> [   96.624880] sd 2:0:0:0: [sda] Assuming drive cache: write through
> [   96.625615] sd 2:0:0:0: [sda] 781422768 512-byte hardware sectors (400088 MB)
> [   96.626365] sd 2:0:0:0: [sda] Write Protect is off
> [   96.626371] sd 2:0:0:0: [sda] Mode Sense: 03 00 00 00
> [   96.626375] sd 2:0:0:0: [sda] Assuming drive cache: write through
> [   96.626379]  sda: sda1
> [   96.638218] sd 2:0:0:0: [sda] Attached SCSI disk
> [   96.662312] sd 2:0:0:0: Attached scsi generic sg0 type 0
> [  107.078870] kjournald starting.  Commit interval 5 seconds
> [  107.080071] EXT3 FS on dm-5, internal journal
> [  107.080082] EXT3-fs: mounted filesystem with ordered data mode.

Now that is truly bizarre.  It means that the drive responded 
differently even though it received the same sequence of commands as 
before.

> Do you wish me to test anything else?

Yes.  Please post a usbmon trace for each of the last two tests.  That 
is, with the unusual_devs update only, and with the unusual_devs update 
plus the invoke_transport update and the scsi_requeue_command update.

Alan Stern

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists