drop 23 patches based on RC review feedback

Dropped patches:

- "ice: drop page splitting and recycling"
  (3a4f419f7509)
  Queues: 6.18
  Reason: intermediate step of the Page Pool conversion; without
          93f53db9f9dc ("ice: switch to Page Pool"), which is not
          queued, ice Rx loses page recycling (huge performance
          regression on all ice cards per the ice Page Pool author).
          Dropped on 10-14 (2d1c78fa20) and re-picked by the bulk
          regeneration.
  Report: https://lore.kernel.org/stable/fce25d6d-8fd7-4f83-8d8d-929871aff885@intel.com/

- "ice: remove legacy Rx and construct SKB"
  (9e314a3c525c)  [dependency]
  Queues: 6.18
  Reason: prerequisite of the dropped page-splitting patch (it does not
          apply without this one); same series.

- "ice: fix crash in ethtool offline loopback test"
  (a9c354e65659)  [dependency]
  Queues: 6.18
  Reason: Stable-dep-of e142dc4ef0f4; does not apply without the
          dropped series.

- "ice: fix rxq info registering in mbuf packets"
  (02852b47c706)  [dependency]
  Queues: 6.18
  Reason: Stable-dep-of e142dc4ef0f4; does not apply without the
          dropped series.

- "ice: change XDP RxQ frag_size from DMA write length to xdp.frame_sz"
  (e142dc4ef0f4)  [dependency]
  Queues: 6.18
  Reason: target of the dropped series; does not apply to the pre-series
          ice Rx code.

- "ice: fix broken Rx on VFs"
  (436fa8e7d1a1)  [dependency]
  Queues: 6.18
  Reason: Fixes: 3a4f419f7509 (does not apply without it).

- "s390/fpu: Fix false-positive kmsan report in fpu_vstl()"
  (14e4e4175b64)
  Queues: 6.12
  Reason: breaks every s390 GCC build of 6.12 (fpu-insn.h:381, 'size'
          undeclared in fpu_vst): the pre-6.14 variant layout puts the
          first hunk in the GCC fpu_vst(). Dropped on 2025-12-17
          (fdf3c3bffb) and re-picked by the bulk regeneration.
  Report: https://lore.kernel.org/stable/20251217135249.422394-1-naresh.kamboju@linaro.org/

- "bpf, s390: Clear fetch destination on faulting arena atomic"
  (cc3e12330599)
  Queues: 6.12
  Reason: calls bpf_atomic_load_reg(), which does not exist in 6.12, so
          the s390 JIT fails to build; the same backport was reverted
          from 6.12.y (7801e1156f68).
  Report: https://lore.kernel.org/stable/830fb27b-d5d3-4f7b-b074-d1b8762fbfde@oracle.com/

- "um: work around sched_yield not yielding in time-travel mode"
  (887c5c12e80c)
  Queues: 6.12, 6.6, 6.1
  Reason: references tt_extra_sched_jiffies, which only exists with
          0b8b2668f998 (v6.13+), so UML with time-travel support fails
          to link; the 6.12 copy was reverted from 6.12.y
          (6aac2c54762c) and 6.6/6.1 have the identical defect. No
          direct report; found in pre-flight review.

- "bnxt_en: Put the TX producer information in the TX BD opaque field"
  (34eec1f29a59)
  Queues: 6.6
  Reason: TX-refactor subset re-added without its follow-on fixes
          (18fe0a383cca, 12f2d033fae95, 0373d5c387f24), none of which
          applies to 6.6; leaves out-of-bounds TX ring indexing on
          XDP_TX completion and on the error path. The chain was
          dropped on 2025-07-03 (9290f87d40) and re-picked by the bulk
          regeneration. No direct report; found in pre-flight review.

- "bnxt_en: Add completion ring pointer in TX and RX ring structures"
  (7f0a168b0441)  [dependency]
  Queues: 6.6
  Reason: part of the bnxt TX-refactor subset dropped above; its
          follow-on fix 0373d5c387f24 does not apply to 6.6.

- "bnxt_en: Modify TX ring indexing logic."
  (6d1add95536b)  [dependency]
  Queues: 6.6
  Reason: part of the bnxt TX-refactor subset dropped above; its
          follow-on fixes 18fe0a383cca and 12f2d033fae95 do not apply
          to 6.6.

- "bnxt_en: Mask the bd_cnt field in the TX BD properly"
  (107b25db6112)  [dependency]
  Queues: 6.6
  Reason: target of the bnxt chain dropped above; it cannot ship
          without 7f0a168b0441 and 6d1add95536b.

- "ASoC: fsl_audmix: rework runtime PM handling in probe"
  (3359ba93d01a)
  Queues: 6.6, 6.1
  Reason: with CONFIG_PM=n, fsl_audmix_runtime_suspend/resume are called
          unconditionally but are still defined under #ifdef CONFIG_PM
          on these branches, so modpost fails with undefined symbols;
          needs 73abd9698960c, which is not queued. Dropped on
          2026-09-13 (892d9ffaab) and re-picked.
  Report: https://lore.kernel.org/stable/1a29a80d-1615-4110-8aa2-eaf931b579b6@w6rz.net/

- "usb: phy: fsl-usb: Fix use-after-free in delayed work during device removal"
  (41ca62e3e21e)
  Queues: 6.6, 6.1
  Reason: calls disable_delayed_work_sync(), which only exists since
          v6.10, so CONFIG_FSL_USB2_OTG builds break; the same backport
          was reverted from 6.6.y and 6.1.y (ff3f428df00b,
          edb0a2c6a663).
  Report: https://lore.kernel.org/stable/20260827154302.706689-1-ralf@mandelbit.com/

- "ptp: Replace timestamp event queue with linked list"
  (d26ab5a35ad9)
  Queues: 6.6, 6.1
  Reason: first step of the multiple-timestamp-event-readers feature
          series with no stable-visible benefit; its new Stable-dep-of
          e9f35294e18d is spurious (the max_vclocks fix applies
          without it). Dropped from 6.6 on 2025-02-05 (4618931e67) and
          re-picked. No direct report; found in pre-flight review.

- "net: ixp4xx_eth: convert to ndo_hwtstamp_get() and ndo_hwtstamp_set()"
  (c14e1ecefd9e)
  Queues: 6.1
  Reason: arm:ixp4xx_defconfig fails to build on 6.1:
          struct kernel_hwtstamp_config and the ndo_hwtstamp_get/set
          NDOs (v6.6) do not exist; the pair was reverted from 6.1.y
          (b2969caffe45, 9d2f0610c1c0).
  Report: https://lore.kernel.org/stable/8bb67921-d8e7-4535-bbec-249e9702ce62@roeck-us.net/

- "net: ethernet: xscale: Check for PTP support properly"
  (594163ea88a0)  [dependency]
  Queues: 6.1
  Reason: Stable-dep-of c14e1ecefd9e; only applies on top of the
          dropped ndo_hwtstamp conversion (reverted together in
          6.1.175).

- "scsi: core: Add struct for args to execution functions"
  (d0949565811f)
  Queues: 6.1
  Reason: first step of the scsi_execute_cmd() conversion; its
          temporary scsi_execute() macro never sets .sense_len, so
          callers with a sense buffer hit WARN_ON_ONCE and fail with
          -EINVAL (broke SPI Domain Validation). Reverted from 6.1.y
          (345b6b831980) at Martin Petersen's request and re-pulled as
          a dependency of the drivetemp fix.
  Report: https://lore.kernel.org/stable/yq1frvvpymp.fsf@ca-mkp.ca.oracle.com/

- "scsi: hwmon: drivetemp: Convert to scsi_execute_cmd()"
  (08e95a2ba4bb)  [dependency]
  Queues: 6.1
  Reason: Stable-dep-of 82163d63ae7a; needs the dropped scsi args
          struct patch.

- "hwmon: (drivetemp) Fix driver producing garbage data when SCSI errors occur"
  (82163d63ae7a)  [dependency]
  Queues: 6.1
  Reason: needs the dropped drivetemp scsi_execute_cmd() conversion.

- "s390/cio: Fix device lifecycle handling in css_alloc_subchannel()"
  (f65c75b0b9b5)
  Queues: 6.1
  Reason: double free of sch->lock (still kmalloc'ed on 6.1) and
          put_device() before device_initialize() on the error paths;
          the same backport was reverted from 6.1.y (5752c2c77957). No
          direct report; found in pre-flight review.

- "namei: don't reveal overmounted entries in refwalk"
  (8189ab688f5a)
  Queues: 6.6, 6.1, 5.15, 5.10
  Reason: withdrawn upstream by 214897acc8c2 ("This bug class is bigger
          than can be fixed with this simple patch"); the revert was
          posted as an AUTOSEL patch.
  Report: https://lore.kernel.org/stable/20261019135441.1420968-97-sashal@kernel.org/

Signed-off-by: Sasha Levin <sashal@kernel.org>