drop 6 patches and fix up a 5.15 backport based on RC review feedback

drop 6 patches and fix up a 5.15 backport based on RC review feedback

Dropped patches:

- "cpufreq: Make cpufreq_driver::exit() return void"
  Queues: 6.6
  Reason: KernelCI build break — 6.6 has drivers (nforce2, loongson2,
  ia64-acpi) whose int-returning exit() callbacks were not converted;
  the nforce2 prerequisite commits were never queued and ia64 has no
  upstream fix, so a clean 1:1 requeue is impossible
  Report: https://lore.kernel.org/stable/178467474625.1334.6331115540835663024@kernelci.org/

- "cpufreq: pcc: Remove empty exit() callback"  [dependency]
  Queues: 6.6
  Reason: prerequisite of the dropped patch above (both queued only to
  support bcb8889c4981)

- "cpufreq: qcom-cpufreq-hw: Fix possible double free"  [dependency]
  Queues: 6.6
  Reason: depends on the dropped API conversion (Stable-dep-of chain);
  to be requeued alone as a context-adjusted backport

- "reset: imx7: Correct polarity of MIPI CSI resets on i.MX8MQ"
  Queues: 6.6
  Reason: missing prerequisite 6d79bb8fd2aa ("media: imx8mq-mipi-csi2:
  Explicitly release reset") — without it the CSI block is left held
  in reset; completes GKH's drop from all older queues (commit
  b416574a40 missed 6.6); kept in 7.1 where the prerequisite is in
  the base
  Report: https://lore.kernel.org/stable/a64558dcb06d62931e73b638dca077ed4562779c.camel@pengutronix.de/

- "gpib: cb7210: Fix region leak when request_irq fails"
  Queues: 6.18
  Reason: upstream commit reverted (05d5d79440c2 reverts 2eae90a457ba,
  285ecb7d9e1b reverts 5ad284960558 — "Turns out not to be correct");
  causes a double release_region() via the detach path on attach
  failure; also recorded in scripts/dropped_commits to prevent a
  third re-pick; 7.1 keeps its fix+revert pair which nets out
  Report: https://lore.kernel.org/stable/26425ef4-454f-4087-98ec-d045247242d8@kernel.org/

- "net: ethernet: oa_tc6: mdiobus->parent initialized with NULL"
  Queues: 7.1, 6.18, 6.12
  Reason: author hold — oa_tc6 follow-up fix series still under review
  on netdev; to be requeued together with the rest of the oa_tc6 set
  once the fix series lands upstream
  Report: https://lore.kernel.org/stable/DM5PR02MB3369B73209A7D10B584B4C5483C22@DM5PR02MB3369.namprd02.prod.outlook.com/

- "dt-bindings: net: updated interrupt type to be active low, level triggered"
  Queues: 7.1, 6.18, 6.12
  Reason: companion driver change b542d13fab0f was already dropped at
  the author's request, leaving the binding example inconsistent with
  the shipping driver (doc says level-low, driver uses edge-falling);
  requeue with the full series later
  Report: https://lore.kernel.org/stable/DM5PR02MB3369B73209A7D10B584B4C5483C22@DM5PR02MB3369.namprd02.prod.outlook.com/

Backport fixup:

- "KVM: Move kvm_io_bus_get_dev() locking responsibilities to callers"
  Queue: 5.15
  Replaced the 5.15 backport content with the queue-5.10 backport of
  the same upstream commit (3a0724998162): the guard(srcu) form broke
  clang builds under -Werror,-Wdeclaration-after-statement (still
  enforced on 5.15.y); the 5.10 variant uses explicit
  srcu_read_lock/srcu_read_unlock and the affected functions are
  byte-identical between 5.10.y and 5.15.y
  Report: https://lore.kernel.org/stable/178467474395.1334.6033734617150761914@kernelci.org/

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