[Bug 219652] New: READ CAPACITY(16) not used on large USB-attached drive in recent kernels

[bugzilla-daemon@kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=219652

            Bug ID: 219652
           Summary: READ CAPACITY(16) not used on large USB-attached drive
                    in recent kernels
           Product: IO/Storage
           Version: 2.5
          Hardware: All
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P3
         Component: SCSI
          Assignee: linux-scsi@vger.kernel.org
          Reporter: bugs-a21@moonlit-rail.com
                CC: stern@rowland.harvard.edu
        Regression: Yes

Created attachment 307435
  --> https://bugzilla.kernel.org/attachment.cgi?id=307435&action=edit
diff without stamps of the usbmon output, good (-) to bad (+)

Upgrading from old mainline LTS kernel 6.6 to current LTS kernel 6.12, a large
3.00TB SATA drive connected via USB through an "Initio Corporation INIC-1610P
SATA bridge" (id 13fd:1e40) is falsely reported as a 2.00TiB capacity drive. 
According to the dmesg output, the newer kernel fails to identify the need to
call read_capacity_16(), resulting in a 32-bit size calculation.

I had reported this initially to the linux-usb mailing list.  Alan Stern
(CC'ed) wrote back, suggesting I redirect to linux-scsi, and include some
usbmon traces.  To upload the traces, I'm opening this on bugzilla.

A few notes:

* The USB-to-SATA bridge (via lsusb) is:
  ID 13fd:1e40 Initio Corporation INIC-1610P SATA bridge

* When booting into kernel 6.12.7, the external drive capacity
  is mis-reported.  Booting /back/ into 6.6.68, the capacity
  continues to be mis-reported.  One must now yank the USB
  cable (or power-cycle) to get the correct size detection again.

Here is a diff of the dmesg output when plugging in the drive.
From (-) is kernel 6.1.122
To   (+) is kernel 6.12.7:
 scsi host12: usb-storage 1-13:1.0
 scsi 12:0:0:0: Direct-Access     TOSHIBA  DT01ACA300       3.00 PQ: 0 ANSI: 4
 sd 12:0:0:0: Attached scsi generic sg1 type 0
-sd 12:0:0:0: [sdb] Very big device. Trying to use READ CAPACITY(16).
-sd 12:0:0:0: [sdb] 5860533167 512-byte logical blocks: (3.00 TB/2.73 TiB)
+sd 12:0:0:0: [sdb] 4294967295 512-byte logical blocks: (2.20 TB/2.00 TiB)
 sd 12:0:0:0: [sdb] Write Protect is off
 sd 12:0:0:0: [sdb] Mode Sense: 23 00 00 00
 sd 12:0:0:0: [sdb] No Caching mode page found
 sd 12:0:0:0: [sdb] Assuming drive cache: write through
-sd 12:0:0:0: [sdb] Very big device. Trying to use READ CAPACITY(16).
- sdb: sdb1
 sd 12:0:0:0: [sdb] Attached SCSI disk

I am attaching a diff of the usbmon output for the earlier kernel (good) and
current kernel (bad), in case the sequence of commands sent to/from the bridge
shows whatever it is that puts the bridge into a low-capacity state.  As I
think I can only attach one file to this posting, I'll make separate
attachments for the good and bad usbmon listings.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are the assignee for the bug.