On Fri, 2014-06-27 at 17:34 +0200, Michael Büsch wrote:
> I tried the following patch:
>
> Index: linux/drivers/scsi/sd.c
> ===================================================================
> --- linux.orig/drivers/scsi/sd.c 2014-06-26 18:40:39.214696552 +0200
> +++ linux/drivers/scsi/sd.c 2014-06-27 15:52:30.776159456 +0200
> @@ -2440,7 +2440,7 @@ sd_read_cache_type(struct scsi_disk *sdk
> sdkp->RCD = 0;
> }
>
> - sdkp->DPOFUA = (data.device_specific & 0x10) != 0;
> + sdkp->DPOFUA = 0;
> if (sdkp->DPOFUA && !sdkp->device->use_10_for_rw) {
> sd_first_printk(KERN_NOTICE, sdkp,
> "Uses READ/WRITE(6), disabling FUA\n");
>
>
> This obviously is not the correct thing to do, but it makes the disk usable:
>
> Jun 27 17:26:50 marge kernel: [ 523.909815] usb 2-1: new SuperSpeed USB device number 2 using xhci_hcd
> Jun 27 17:26:50 marge kernel: [ 523.929246] usb 2-1: New USB device found, idVendor=152d, idProduct=0567
> Jun 27 17:26:50 marge kernel: [ 523.929258] usb 2-1: New USB device strings: Mfr=1, Product=2, SerialNumber=3
> Jun 27 17:26:50 marge kernel: [ 523.929265] usb 2-1: Product: USB to ATA/ATAPI Bridge
> Jun 27 17:26:50 marge kernel: [ 523.929271] usb 2-1: Manufacturer: JMicron
> Jun 27 17:26:50 marge kernel: [ 523.929276] usb 2-1: SerialNumber: xxx
> Jun 27 17:26:50 marge kernel: [ 523.930999] usb-storage 2-1:1.0: USB Mass Storage device detected
> Jun 27 17:26:50 marge kernel: [ 523.931237] scsi12 : usb-storage 2-1:1.0
> Jun 27 17:26:51 marge kernel: [ 524.930451] scsi 12:0:0:0: Direct-Access JMicron Generic 0114 PQ: 0 ANSI: 6
> Jun 27 17:26:51 marge kernel: [ 524.931228] sd 12:0:0:0: Attached scsi generic sg3 type 0
> Jun 27 17:26:51 marge kernel: [ 524.932964] sd 12:0:0:0: [sdd] Spinning up disk...
> Jun 27 17:26:53 marge kernel: [ 525.937848] ..ready
> Jun 27 17:26:53 marge kernel: [ 526.942395] sd 12:0:0:0: [sdd] 976773168 512-byte logical blocks: (500 GB/465 GiB)
> Jun 27 17:26:53 marge kernel: [ 526.943037] sd 12:0:0:0: [sdd] Write Protect is off
> Jun 27 17:26:53 marge kernel: [ 526.943048] sd 12:0:0:0: [sdd] Mode Sense: 47 00 10 08
> Jun 27 17:26:53 marge kernel: [ 526.943662] sd 12:0:0:0: [sdd] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
> Jun 27 17:26:53 marge kernel: [ 526.987919] sdd: sdd1
> Jun 27 17:26:53 marge kernel: [ 526.990055] sd 12:0:0:0: [sdd] Attached SCSI disk
> Jun 27 17:27:29 marge kernel: [ 563.227849] EXT4-fs (sdd1): warning: mounting fs with errors, running e2fsck is recommended
> Jun 27 17:27:29 marge kernel: [ 563.228043] EXT4-fs (sdd1): mounted filesystem with ordered data mode. Opts: (null)
>
> (This is on another machine, but it shows the same behavior without the patch.)
>
> Does somebody have a hint for a real fix?
I suspect the problem is in the USB bridge: The device reports FUA
support in its initial IDENTIFY, which gets translated to the correct
mode page bits, but when we send a READ/WRITE with the FUA bit set, the
bridge doesn't know how to translate it and gives us the error. There's
probably some way in USB to blacklist the bridge, but we need to
understand the consequences of the blacklist: The device is claiming a
writeback cache which means if it won't do either FUA or SYNC CACHE,
there's going to be a data integrity failure.
The fact that your initial hack works suggests that sync cache is
accepted by it, so we probably just need a USB blacklist fixing the FUA
problem.
James
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
