1. Forum
  2. Usenet
  3. LINUX.DEBIAN.KERNEL

  • Bug#1010085: linux: Optimal transfer size 33553920 bytes not a multiple

    From Salvatore Bonaccorso@1:229/2 to Diederik de Haas on Sun Jun 15 17:20:01 2025
    XPost: linux.debian.bugs.dist
    From: carnil@debian.org

    Hi,

    On Mon, Apr 25, 2022 at 05:54:58PM +0200, Diederik de Haas wrote:
    On Sunday, 24 April 2022 00:10:01 CEST Diederik de Haas wrote:
    The drives have the exact same same characteristics:

    # fdisk -l /dev/sdb
    Disk /dev/sdb: 2.73 TiB, 3000592982016 bytes, 5860533168 sectors
    Disk model: WDC WD30EFRX-68E
    Units: sectors of 1 * 512 = 512 bytes
    Sector size (logical/physical): 512 bytes / 4096 bytes
    I/O size (minimum/optimal): 4096 bytes / 4096 bytes

    [ USB connected drive ]
    # sg_vpd -p bl /dev/sdb
    Block limits VPD page (SBC):
    ...
    Optimal transfer length granularity: 8 blocks
    Maximum transfer length: 65535 blocks
    Optimal transfer length: 65535 blocks
    Maximum prefetch transfer length: 65535 blocks
    <snip>

    [ SATA connected drive ]
    # sg_vpd -p bl /dev/sdb
    Block limits VPD page (SBC):
    ...
    Optimal transfer length granularity: 8 blocks
    Maximum transfer length: 0 blocks [not reported]
    Optimal transfer length: 0 blocks [not reported]
    Maximum prefetch transfer length: 0 blocks [ignored]
    <snip>

    Learned some more things:
    - partitioning is irrelevant (for this issue)
    - physical sector size of 4096 is very relevant; with 512 you don't get a
    warning, only an informational message saying 33553920 bytes is
    optimal transfer size (which is still 512 bytes short of 32MB)
    - USB connected drive reports Maximum/Optimum length of 65535 blocks, while
    SATA reports 0

    The last item explains the difference I saw and
    I missed it as I was focused on the partitioning.

    https://www.spinics.net/lists/linux-scsi/msg139591.html is where I got
    the various `sg_*` commands from.

    One of that message follow-ups was titled:
    [PATCH] scsi: sd: Optimal I/O size should be a multiple of reported granularity

    After a bit of searching that led me to the following: https://patchwork.kernel.org/project/linux-scsi/patch/20220302053559.32147-9-martin.petersen@oracle.com/
    or https://lkml.kernel.org/linux-scsi/20220302053559.32147-1-martin.petersen@oracle.com/ on LKML

    I feel a bit more comfortable ignoring the kernel warning, but I'm still
    not sure whether it is a harmless msg or an actual bug.
    If the maintainers consider this not to be a bug, feel free to close it.

    Looking at some older bugs, what should we do with this one (asking
    this question brings it on the list of bugs to discuss in the next
    kernel team meeting)?

    Regards,
    Salvatore

    --- SoupGate-Win32 v1.05
    * Origin: you cannot sedate... all the things you hate (1:229/2)