[PATCH 6.12.y 0/8] serial: 8250_dw: backport BUSY deassert series

["Ionut Nechita (Wind River)" <ionut.nechita@windriver.com>]

From: Ionut Nechita <ionut.nechita@windriver.com>

Hi Greg, Ilpo,

This series backports the 8250_dw BUSY deassert fix to 6.12.y, per
Ilpo's guidance from the request thread:

  https://lore.kernel.org/linux-serial/deb9499d-3245-7e38-9034-e533d4b5f512@linux.intel.com/

Background: we ship 6.12 LTS as part of a certified production
platform (StarlingX/Yocto) and are hitting the BUSY assertion issue
on Intel DesignWare 8250 UARTs - LCR writes get silently ignored
under Rx load, causing baud-rate / framing mismatches after
set_termios. A major LTS bump is a multi-month re-qualification we
can't justify for a single subsystem fix.

The original mainline series is 7 patches by Ilpo (plus its
dependencies). For 6.12.y, the resulting series here is 8 patches:

  Prerequisites (per Ilpo):
   1/8 serial: 8250: use serial_port_in/out() helpers       [dbd26a886e94]
   2/8 serial: 8250_dw: Comment possible corner cases ...   [bd8cad85561b]

  BUSY deassert series:
   3/8 serial: 8250: Protect LCR write in shutdown          [59a33d83bbe6]
   4/8 serial: 8250_dw: Avoid unnecessary LCR writes        [8002d6d6d0d8]
   5/8 serial: 8250: Add serial8250_handle_irq_locked()     [8324a54f604d]
   6/8 serial: 8250_dw: Rework dw8250_handle_irq() ...      [883c5a2bc934]
   7/8 serial: 8250_dw: Rework IIR_NO_INT handling ...      [73a4ed8f9efa]
   8/8 serial: 8250_dw: Ensure BUSY is deasserted           [a7b9ce39fbe4]

Notes:

 - Patch 6/7 of the original mainline series,
   commit e0a368ae7953 ("serial: 8250: Add late synchronize_irq() to shutdown to handle DW UART BUSY"),
   is *already* in 6.12.y as
   commit 0bae1c670aa8 ("serial: 8250: Add late synchronize_irq() to shutdown to handle DW UART BUSY")
   (Ilpo, 2026-02-03), so it is not re-sent here. Functionally this
   means patches 3-8 above land on top of the existing
   late-synchronize_irq() fix; the conflict in patch 3 (LCR write
   placement around the late synchronize_irq) was resolved
   accordingly.

 - Ilpo's other suggested prerequisites are *not* included as prereqs:
     * commit b339809edda1 ("serial: 8250: use guard()s")
     * commit fc9ceb501e38 ("serial: 8250: sanitize uart_port::serial_{in,out}() types")
     * commit c213375e3283 ("serial: 8250_dw: Call dw8250_quirks() conditionally")

   Reasoning:

     * commit b339809edda1 ("serial: 8250: use guard()s") is a
       large refactor and only the shutdown hunk was needed;
       instead, patches 3, 5, 6, 7 here are adapted to use the
       existing explicit uart_port_lock_irqsave/unlock_irqrestore
       form rather than the cleanup-based guard() introduced by
       that commit;
     * commit fc9ceb501e38 ("serial: 8250: sanitize
       uart_port::serial_{in,out}() types") only causes a trivial
       conflict in patch 8 in code that the BUSY change removes
       anyway (per Ilpo's note);
     * commit c213375e3283 ("serial: 8250_dw: Call dw8250_quirks()
       conditionally") only causes a conflict in patch 7 around
       the dw8250_setup_dma_filter() helper and the conditional
       p->handle_irq assignment, neither of which exist in 6.12.y
       and neither of which is needed for the BUSY fix.

 - Namespace export syntax: in 6.12.y both EXPORT_SYMBOL_NS_GPL()
   and MODULE_IMPORT_NS() apply __stringify(ns) to the namespace
   argument, so it must be a bare identifier. Mainline (where the
   upstream patches were written) accepts a string literal. Patches
   5 and 8 here use the bare-identifier form (SERIAL_8250) instead
   of the upstream string form ("SERIAL_8250"); without this fix
   the .vmlinux.export.c link step fails with "expected ':' or ')'
   before 'SERIAL_8250'". This is noted in the [Ionut: ...] block
   of the affected patches.

   Each of patches 3, 5, 6, 7 and 8 carries an explicit
   "[Ionut: adapt to 6.12.y - ...]" note describing exactly what
   was changed relative to the upstream commit.

Build:
 - Each patch builds individually on 6.12.87 to a complete vmlinux
   (bisect-safe), with CONFIG_SERIAL_8250=y, CONFIG_SERIAL_8250_DW=m
   on x86_64 defconfig.

Testing plan:
 - We will test on Intel platforms with DW APB UART
   (snps,dw-apb-uart) running 6.12.57-rt / 6.12.87-rt (PREEMPT_RT)
   to confirm the original symptom (LCR writes silently ignored
   under Rx load -> baud / framing mismatch after set_termios) is
   gone. Will report Tested-by once cycles complete.

Based on: linux-6.12.y at v6.12.87 (8bf2f55ef536).

Andy Shevchenko (1):
  serial: 8250_dw: Comment possible corner cases in serial_out()
    implementation

Ilpo Järvinen (6):
  serial: 8250: Protect LCR write in shutdown
  serial: 8250_dw: Avoid unnecessary LCR writes
  serial: 8250: Add serial8250_handle_irq_locked()
  serial: 8250_dw: Rework dw8250_handle_irq() locking and IIR handling
  serial: 8250_dw: Rework IIR_NO_INT handling to stop interrupt storm
  serial: 8250_dw: Ensure BUSY is deasserted

Jiri Slaby (SUSE) (1):
  serial: 8250: use serial_port_in/out() helpers

 drivers/tty/serial/8250/8250.h      |  25 +++
 drivers/tty/serial/8250/8250_dw.c   | 298 +++++++++++++++++++++++-----
 drivers/tty/serial/8250/8250_fsl.c  |   8 +-
 drivers/tty/serial/8250/8250_omap.c |   2 +-
 drivers/tty/serial/8250/8250_port.c |  66 +++---
 include/linux/serial_8250.h         |   1 +
 6 files changed, 319 insertions(+), 81 deletions(-)

-- 
2.54.0