[PATCH] Serial updates

From: Russell King (rmk+lkml_at_arm.linux.org.uk)
Date: 10/31/04

  • Next message: Alexander Gran: "[2.6.10-rc1-mm2] keyboard / synaptics not working" 
    Date:	Sun, 31 Oct 2004 17:51:14 +0000
To: Linux Kernel List <linux-kernel@vger.kernel.org>

    Ok, here's a major serial update. Items covered in this update:

    - unuse register_serial/unregister_serial.
      Instead, please use serial8250_register_port() and
      serial8250_unregister_port() to talk to the 8250 driver.

      The old interfaces have several restrictions:
      1. do not allow the struct device associated with a port to be known
         to the tty layer.
      2. have various restrictions on the size of IO addresses (see the
         HIGH_BITS_OFFSET stuff - which incidentally 8250_pnp got wrong.)

    - provide a mechanism for 8250 platform ports to be dynamically
      registered. We do this via platform devices - either one or
      multiple platform devices. You can have none, one, or as many
      as you desire. 8250.c couldn't care.

    - any ports listed in include/asm-*/serial.h will still be
      intialised in preference to platform device based ports for the
      time being. It is intended that everyone will move to using
      the platform device method.

    - this means that CONFIG_SERIAL_8250_NR_UARTS slightly changes
      definition. It is the number of _extra_ ports above those in
      include/asm-*/serial.h that 8250.h will support. If you have
      removed all ports from include/asm-*/serial.h, then it is
      obviously the total number of ports that 8250.c will support,
      and you need to make sure that it's large enough for your
      platform.

    - ppc64 broke in this merge. That's expected because I backed out
      benh's changes to the serial drivers. A "get you working again"
      patch is with benh as of last night pending his attention.

    The patch is about 50K, so won't fit through lkml, and is available
    here instead:

      http://www.arm.linux.org.uk/~rmk/misc/linus-serial.diff

    People who should test this patch as a minimum:

     - ia64 people (ACPI port discovery)
     - parisc people (GSC port discovery)
     - pnp using people

    Once this lot is in, I'll be following up with a set of patches which
    removes the serial device tables in include/asm-arm/arch-*/serial.h.

    Diffstat and changeset log follow:

     drivers/serial/8250.c | 371 +++++++++++++++++++++++--------------------
     drivers/serial/8250.h | 74 ++++++++
     drivers/serial/8250_acorn.c | 60 +++---
     drivers/serial/8250_acpi.c | 74 +++-----
     drivers/serial/8250_pci.c | 6
     drivers/serial/8250_pnp.c | 34 +--
     drivers/serial/au1x00_uart.c | 17 -
     drivers/serial/serial_core.c | 83 +--------
     drivers/serial/serial_cs.c | 28 +--
     include/linux/8250.h | 75 --------
     include/linux/serial_8250.h | 28 +++
     include/linux/serial_core.h | 10 -
     12 files changed, 420 insertions(+), 440 deletions(-)

    through these ChangeSets:

    <rmk@flint.arm.linux.org.uk> (04/10/31 1.2354)
            [SERIAL] Fix deadlock on removal of 8250 module.
            
            We must unregister all serial ports before driver_unregister()
            can complete. This means that we must unregister all ports in
            serial8250_remove, including our legacy ISA ports. We flag this
            special cleanup operation by setting serial8250_isa_devs to NULL
            and not handling our own platform device any differently from
            any others.

    <rmk@flint.arm.linux.org.uk> (04/10/31 1.2353)
            [SERIAL] Don't detect console availability using port->ops.
            
            Use !iobase && !membase rather than !ops for console port
            availability.

    <rmk@flint.arm.linux.org.uk> (04/10/31 1.2352)
            [SERIAL] 8250_acpi: Convert to use serial8250_{un,}register_port.

    <rmk@flint.arm.linux.org.uk> (04/10/31 1.2351)
            [SERIAL] Don't use UPF_AUTOPROBE, fix two build problems.
            
            The curse of the missing __devexit_p() returns, and asm-*/ obviously
            marks the end of the comment.

    <rmk@flint.arm.linux.org.uk> (04/10/31 1.2350)
            [SERIAL] 8250_pnp: Convert to use serial8250_{un,}register_port.

    <rmk@flint.arm.linux.org.uk> (04/10/31 1.2349)
            [SERIAL] serial_cs: Convert to use serial8250_{un,}register_port.

    <rmk@flint.arm.linux.org.uk> (04/10/31 1.2348)
            [SERIAL] 8250: Warn when ports with zero base_baud are registered.

    <rmk@flint.arm.linux.org.uk> (04/10/31 1.2347)
            [SERIAL] 8250_acorn: Convert to use serial8250_{un,}register_port.

    <rmk@flint.arm.linux.org.uk> (04/10/30 1.2345)
            [SERIAL] Undo "get_legacy_serial_ports" patch for PPC.
            
            This patch conflicts with work to properly integrate the device model
            into the serial layer, and provide architectures with a *clean* way
            to tell 8250.c about their serial ports at run time.

    <rmk@flint.arm.linux.org.uk> (04/10/30 1.2026.4.8)
            [SERIAL] 8250: prevent ports with zero clocks being registered.

    <rmk@flint.arm.linux.org.uk> (04/10/30 1.2026.4.7)
            [SERIAL] 8250: add probe and remove device driver methods.
            
            This change allows platform devices named "serial8250" to provide
            lists of serial ports to the 8250 driver at runtime, in addition to
            the hard coded table in include/asm-*/serial.h.
            
            The next step is to deprecate the tables in serial.h.

    <rmk@flint.arm.linux.org.uk> (04/10/30 1.2026.4.6)
            [SERIAL] 8250: Add platform device for ISA 8250-compatible devices.
            
            Add a platform device for ISA 8250-compatible serial devices listed
            in the table in include/asm-*/serial.h. Arrange for unregistered
            serial devices to be owned by this device.
            
            This enables power management for ISA 8250 devices, and starts to
            opens the way for architectures to dynamically provide their own
            lists of 8250 devices via platform device(s).

    <rmk@flint.arm.linux.org.uk> (04/10/30 1.2026.4.5)
            [SERIAL] Re-order 8250 serial driver initialisation/finalisation.
            
            Only register the 8250 serial driver with the device model after
            registering and setting up our internal uart ports. Do the reverse
            on module finalisation.

    <rmk@flint.arm.linux.org.uk> (04/10/30 1.2026.4.4)
            [SERIAL] 8250: move basic initialisation of 8250 ports.
            
            This moves the basic initialisation of 8250 ports from
            serial8250_register_ports() into serial8250_isa_init_ports()

    <rmk@flint.arm.linux.org.uk> (04/10/30 1.2026.4.3)
            [SERIAL] 8250: Fix resource handling.
            
            serial8250_request_std_resource() is now responsible for claiming
            the standard resources, _and_ calling ioremap if necessary.
            serial8250_release_std_resource() performs the complementary function
            in its entirety.
            serial8250_*_rsa_resource() perform the similar operations for RSA
            ports, with the exception that RSA ports can only be mapped into IO
            space.

    <rmk@flint.arm.linux.org.uk> (04/10/30 1.2026.4.2)
            [SERIAL] Clean up serial_core.c write functions.
            
            Since the tty layer now takes care of user space writes,
            __uart_user_write() and associated temporary buffer and temporary
            buffer semaphore have all become unnecessary. There's also little
            point in having __uart_kern_write() separate from uart_write(), so
            combine the two together.
            
            Adrian Bunk kindly provided the patch to remove __uart_user_write().
            The rest of the work is rmk's.
            
            Signed-off-by: Adrian Bunk
            Signed-off-by: Russell King <rmk@arm.linux.org.uk>

    <arjan@nl.rmk.(none)> (04/10/24 1.2026.4.1)
            [SERIAL] Remove dead code.
            
            serial8250_get_irq_map is no longer used anywhere in the kernel (it
            used to be used by the isapnp code but isn't anymore) so it's dead
            code, below is a patch to remove this. 

    -- 
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
                 2.6 Serial core
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/
  • Next message: Alexander Gran: "[2.6.10-rc1-mm2] keyboard / synaptics not working"

    Relevant Pages