Commit e6f836ef7865 for kernel

commit e6f836ef7865fb454641febe080cf88a3c677186
Merge: 25fccf436e6e a4845f077415
Author: Linus Torvalds <torvalds@linux-foundation.org>
Date:   Sun Nov 8 11:59:25 2026 -0800

    Merge tag 'ata-7.4-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/libata/linux

    Pull ata fixes from Niklas Cassel:

     - Allow re-enabling NCQ, after EH disabled it, by writing the current
       queue depth to the queue_depth sysfs attribute (Igor)

     - Retry failed non-NCQ commands when their sense data asks for a
       retry, as is already done for NCQ commands. Previously, such
       non-NCQ commands were never retried (me)

     - Only check the power mode of ATA and ZAC devices before putting
       them in standby mode (on suspend, shutdown or device removal).
       libata never puts other devices, such as ATAPI devices, in standby
       mode, and some ATAPI devices reject the CHECK POWER MODE command
       (Giuseppe)

     - Fix several bugs in the MODE SELECT translation.

       Reject parameter lists with data after the mode page before the
       mode page is handled. Previously, a rejected command could leave
       the CDL enabled state tracked by libata out of sync with the
       device, and data after the control mode page was silently ignored.

       Reject mode pages with SUBPAGE CODE FFh, and mode pages with the
       SPF bit set but SUBPAGE CODE 00h, which were incorrectly accepted.

       The commit that passes the whole mode page to the MODE SELECT
       handlers might look like a cleanup, but it is a fix, and thus has
       a Fixes tag: the FIELD POINTER pointed 2 bytes (page_0 format) or
       4 bytes (sub_page format) before the invalid byte in the mode page
       (me, Shao)

     - Disable the enclosure management (SEMB) port of the Marvell
       88SM9705 port multiplier. The port never answers SRST, and the
       resulting EH retries delayed disk discovery by minutes (Marco)

     - Reject data-in commands, e.g. INQUIRY, submitted via SG_IO with a
       data-out buffer, whose pages may be pinned without write access.
       Writing a reply to such a buffer could oops the kernel (Zhao)

    * tag 'ata-7.4-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/libata/linux:
      ata: libata-scsi: reject data-in commands with a data-out buffer
      ata: libata-pmp: disable the SEMB port on Marvell 88SM9705
      ata: libata-scsi: set the field pointer when rejecting CDL
      ata: libata-scsi: pass the whole mode page to the MODE SELECT handlers
      ata: libata-scsi: fix MODE SELECT field pointer for block descriptor length
      ata: libata-scsi: fix MODE SELECT field pointer for unsupported subpages
      ata: libata-scsi: reject MODE SELECT mode pages with a wrong SPF bit
      ata: libata-scsi: do not accept subpage code FFh in MODE SELECT
      ata: libata-scsi: fix MODE SELECT handling of data after the mode page
      ata: libata-core: Check device support before querying power state
      ata: libata-eh: Retry non-NCQ commands when their sense data asks for it
      ata: libata-sata: allow re-enabling NCQ with the same queue_depth