Changes to support booting NetBSD/alpha

Along with the previous patch set for qemu-palcode, these changes
allow NetBSD/alpha to run in qemu-system-alpha.

- Allow a the minimum PCI device number to be passed to typhoon_init().
  Use this to specify a minimum PCI device number of 1 in the Clipper
  system emulation, since that's the lowest PCI device number with
  valid interrupt mappings on that system.
- Instead of passing just the CPU count in trap_arg2, also pass other
  configuration information.  Define the first config bit (bit 6) to
  reflect the "-nographic" option; PALcode will use this to initialize
  the Console Terminal Block to specify serial console.
- Attach a minimal i82378 SIO PCI node; some operating systems won't
  scan for ISA devices unless a PCI-ISA bridge is detected.  Real hardware
  would have had a Cypress or ALI bridge, but many Alpha models shipped
  with a i82378, and most operating sytems are not picky about the specific
  model.
- Modify the mc146818rtc emulation to ensure the timer is started as soon
  as the device is realized, and update the PF bit in REG_C even if it is
  not going to result in raising an interrupt, as the real hardware does.

Signed-off-by: Jason Thorpe <thorpej@me.com>
---
 hw/alpha/alpha_sys.h |   2 +-
 hw/alpha/dp264.c     |  16 ++++++-
 hw/alpha/typhoon.c   | 104 +++++++++++++++++++++++++++++++++++++++++--
 hw/rtc/mc146818rtc.c |   7 +++
 4 files changed, 123 insertions(+), 6 deletions(-)

Patchew URL: https://patchew.org/QEMU/20201007032417.26044-1-thorpej@me.com/



Hi,

This series seems to have some coding style problems. See output below for
more information:

Type: series
Message-id: 20201007032417.26044-1-thorpej@me.com
Subject: [PATCH] Changes to support booting NetBSD/alpha

=== TEST SCRIPT BEGIN ===
#!/bin/bash
git rev-parse base > /dev/null || exit 0
git config --local diff.renamelimit 0
git config --local diff.renames True
git config --local diff.algorithm histogram
./scripts/checkpatch.pl --mailback base..
=== TEST SCRIPT END ===

Updating 3c8cf5a9c21ff8782164d1def7f44bd888713384
From https://github.com/patchew-project/qemu
 * [new tag]         patchew/20201007032417.26044-1-thorpej@me.com -> patchew/20201007032417.26044-1-thorpej@me.com
Switched to a new branch 'test'
a3c92d1 Changes to support booting NetBSD/alpha

=== OUTPUT BEGIN ===
WARNING: Block comments use a leading /* on a separate line
#51: FILE: hw/alpha/dp264.c:75:
+    /* arg0 -> memory size

WARNING: Block comments use * on subsequent lines
#52: FILE: hw/alpha/dp264.c:76:
+    /* arg0 -> memory size
+       arg1 -> kernel entry point

ERROR: code indent should never use tabs
#56: FILE: hw/alpha/dp264.c:80:
+^I^I    bit  6   -> nographics option (for HWRPB CTB)$

WARNING: Block comments use a trailing */ on a separate line
#58: FILE: hw/alpha/dp264.c:82:
+       See init_hwrpb() in the PALcode.  */

ERROR: suspect code indent for conditional statements (4, 6)
#63: FILE: hw/alpha/dp264.c:87:
+    if (!machine->enable_graphics)
+      cpus[0]->env.trap_arg2 |= (1 << 6);

ERROR: braces {} are necessary for all arms of this statement
#63: FILE: hw/alpha/dp264.c:87:
+    if (!machine->enable_graphics)
[...]

WARNING: Block comments use a leading /* on a separate line
#67: FILE: hw/alpha/dp264.c:90:
+    /* Init the chipset.  Because we're using CLIPPER IRQ mappings,

WARNING: Block comments use * on subsequent lines
#68: FILE: hw/alpha/dp264.c:91:
+    /* Init the chipset.  Because we're using CLIPPER IRQ mappings,
+       the minimum PCI device IdSel is 1.  */

WARNING: Block comments use a trailing */ on a separate line
#68: FILE: hw/alpha/dp264.c:91:
+       the minimum PCI device IdSel is 1.  */

WARNING: Block comments use a leading /* on a separate line
#103: FILE: hw/alpha/typhoon.c:925:
+    /* Init the PCI-ISA bridge.  Technically, this would have been

WARNING: Block comments use * on subsequent lines
#104: FILE: hw/alpha/typhoon.c:926:
+    /* Init the PCI-ISA bridge.  Technically, this would have been
+       a cy82c693ub, but a i82378 SIO was also used on many Alpha

WARNING: Block comments use a trailing */ on a separate line
#112: FILE: hw/alpha/typhoon.c:934:
+       it to be there will see it.  */

WARNING: Block comments use a leading /* on a separate line
#125: FILE: hw/alpha/typhoon.c:970:
+/* The following was copied from hw/isa/i82378.c and modified to provide

WARNING: Block comments use * on subsequent lines
#126: FILE: hw/alpha/typhoon.c:971:
+/* The following was copied from hw/isa/i82378.c and modified to provide
+   only the minimal PCI device node.  */

WARNING: Block comments use a trailing */ on a separate line
#126: FILE: hw/alpha/typhoon.c:971:
+   only the minimal PCI device node.  */

ERROR: trailing whitespace
#128: FILE: hw/alpha/typhoon.c:973:
+/*                            $

ERROR: trailing whitespace
#129: FILE: hw/alpha/typhoon.c:974:
+ * QEMU Intel i82378 emulation (PCI to ISA bridge) $

ERROR: trailing whitespace
#130: FILE: hw/alpha/typhoon.c:975:
+ *                            $

ERROR: trailing whitespace
#132: FILE: hw/alpha/typhoon.c:977:
+ *  $

ERROR: trailing whitespace
#137: FILE: hw/alpha/typhoon.c:982:
+ *  $

ERROR: trailing whitespace
#142: FILE: hw/alpha/typhoon.c:987:
+ *     $

ERROR: trailing whitespace
#145: FILE: hw/alpha/typhoon.c:990:
+ */    $

ERROR: trailing whitespace
#162: FILE: hw/alpha/typhoon.c:1007:
+        VMSTATE_PCI_DEVICE(parent_obj, I82378State), $

ERROR: trailing whitespace
#164: FILE: hw/alpha/typhoon.c:1009:
+    },                        $

ERROR: trailing whitespace
#165: FILE: hw/alpha/typhoon.c:1010:
+};                            $

ERROR: trailing whitespace
#170: FILE: hw/alpha/typhoon.c:1015:
+ $

ERROR: trailing whitespace
#176: FILE: hw/alpha/typhoon.c:1021:
+ $

ERROR: trailing whitespace
#182: FILE: hw/alpha/typhoon.c:1027:
+}      $

ERROR: trailing whitespace
#185: FILE: hw/alpha/typhoon.c:1030:
+{   $

ERROR: trailing whitespace
#189: FILE: hw/alpha/typhoon.c:1034:
+    k->realize = i82378_realize; $

ERROR: trailing whitespace
#196: FILE: hw/alpha/typhoon.c:1041:
+}                             $

ERROR: trailing whitespace
#206: FILE: hw/alpha/typhoon.c:1051:
+        { },     $

ERROR: trailing whitespace
#208: FILE: hw/alpha/typhoon.c:1053:
+};  $

ERROR: if this code is redundant consider removing it
#226: FILE: hw/rtc/mc146818rtc.c:158:
+#if 0

total: 22 errors, 12 warnings, 192 lines checked

Commit a3c92d1aa5e6 (Changes to support booting NetBSD/alpha) has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.
=== OUTPUT END ===

Test command exited with code: 1


The full log is available at
http://patchew.org/logs/20201007032417.26044-1-thorpej@me.com/testing.checkpatch/?type=message.
---
Email generated automatically by Patchew [https://patchew.org/].
Please send your feedback to patchew-devel@redhat.com

Hi Jason,

On 10/7/20 5:24 AM, Jason Thorpe wrote:
> Along with the previous patch set for qemu-palcode, these changes
> allow NetBSD/alpha to run in qemu-system-alpha.
> 
> - Allow a the minimum PCI device number to be passed to typhoon_init().
>   Use this to specify a minimum PCI device number of 1 in the Clipper
>   system emulation, since that's the lowest PCI device number with
>   valid interrupt mappings on that system.
> - Instead of passing just the CPU count in trap_arg2, also pass other
>   configuration information.  Define the first config bit (bit 6) to
>   reflect the "-nographic" option; PALcode will use this to initialize
>   the Console Terminal Block to specify serial console.
> - Attach a minimal i82378 SIO PCI node; some operating systems won't
>   scan for ISA devices unless a PCI-ISA bridge is detected.  Real hardware
>   would have had a Cypress or ALI bridge, but many Alpha models shipped
>   with a i82378, and most operating sytems are not picky about the specific
>   model.
> - Modify the mc146818rtc emulation to ensure the timer is started as soon
>   as the device is realized, and update the PF bit in REG_C even if it is
>   not going to result in raising an interrupt, as the real hardware does.
> 
> Signed-off-by: Jason Thorpe <thorpej@me.com>
> ---
>  hw/alpha/alpha_sys.h |   2 +-
>  hw/alpha/dp264.c     |  16 ++++++-
>  hw/alpha/typhoon.c   | 104 +++++++++++++++++++++++++++++++++++++++++--
>  hw/rtc/mc146818rtc.c |   7 +++
>  4 files changed, 123 insertions(+), 6 deletions(-)
> 
> diff --git a/hw/alpha/alpha_sys.h b/hw/alpha/alpha_sys.h
> index e2c02e2bbe..4835b3d5ee 100644
> --- a/hw/alpha/alpha_sys.h
> +++ b/hw/alpha/alpha_sys.h
> @@ -11,7 +11,7 @@
>  
>  
>  PCIBus *typhoon_init(MemoryRegion *, ISABus **, qemu_irq *, AlphaCPU *[4],
> -                     pci_map_irq_fn);
> +                     pci_map_irq_fn, uint8_t devfn_min);
>  
>  /* alpha_pci.c.  */
>  extern const MemoryRegionOps alpha_pci_ignore_ops;
> diff --git a/hw/alpha/dp264.c b/hw/alpha/dp264.c
> index 4d24518d1d..de59ae78cb 100644
> --- a/hw/alpha/dp264.c
> +++ b/hw/alpha/dp264.c
> @@ -72,13 +72,25 @@ static void clipper_init(MachineState *machine)
>          cpus[i] = ALPHA_CPU(cpu_create(machine->cpu_type));
>      }
>  
> +    /* arg0 -> memory size
> +       arg1 -> kernel entry point
> +       arg2 -> config word
> +
> +       Config word: bits 0-5 -> ncpus
> +		    bit  6   -> nographics option (for HWRPB CTB)
> +
> +       See init_hwrpb() in the PALcode.  */
> +
>      cpus[0]->env.trap_arg0 = ram_size;
>      cpus[0]->env.trap_arg1 = 0;
>      cpus[0]->env.trap_arg2 = smp_cpus;
> +    if (!machine->enable_graphics)
> +      cpus[0]->env.trap_arg2 |= (1 << 6);
>  
> -    /* Init the chipset.  */
> +    /* Init the chipset.  Because we're using CLIPPER IRQ mappings,
> +       the minimum PCI device IdSel is 1.  */
>      pci_bus = typhoon_init(machine->ram, &isa_bus, &rtc_irq, cpus,
> -                           clipper_pci_map_irq);
> +                           clipper_pci_map_irq, PCI_DEVFN(1, 0));
>  
>      /* Since we have an SRM-compatible PALcode, use the SRM epoch.  */
>      mc146818_rtc_init(isa_bus, 1900, rtc_irq);
> diff --git a/hw/alpha/typhoon.c b/hw/alpha/typhoon.c
> index a42b319812..cfe7fd5098 100644
> --- a/hw/alpha/typhoon.c
> +++ b/hw/alpha/typhoon.c
> @@ -817,7 +817,8 @@ static void typhoon_alarm_timer(void *opaque)
>  }
>  
>  PCIBus *typhoon_init(MemoryRegion *ram, ISABus **isa_bus, qemu_irq *p_rtc_irq,
> -                     AlphaCPU *cpus[4], pci_map_irq_fn sys_map_irq)
> +                     AlphaCPU *cpus[4], pci_map_irq_fn sys_map_irq,
> +                     uint8_t devfn_min)
>  {
>      MemoryRegion *addr_space = get_system_memory();
>      DeviceState *dev;
> @@ -887,7 +888,7 @@ PCIBus *typhoon_init(MemoryRegion *ram, ISABus **isa_bus, qemu_irq *p_rtc_irq,
>      b = pci_register_root_bus(dev, "pci",
>                                typhoon_set_irq, sys_map_irq, s,
>                                &s->pchip.reg_mem, &s->pchip.reg_io,
> -                              0, 64, TYPE_PCI_BUS);
> +                              devfn_min, 64, TYPE_PCI_BUS);
>      phb->bus = b;
>      sysbus_realize_and_unref(SYS_BUS_DEVICE(dev), &error_fatal);
>  
> @@ -921,10 +922,21 @@ PCIBus *typhoon_init(MemoryRegion *ram, ISABus **isa_bus, qemu_irq *p_rtc_irq,
>      /* Pchip1 PCI configuration, 0x802.FE00.0000, 16MB.  */
>  
>      /* Init the ISA bus.  */
> -    /* ??? Technically there should be a cy82c693ub pci-isa bridge.  */
> +    /* Init the PCI-ISA bridge.  Technically, this would have been
> +       a cy82c693ub, but a i82378 SIO was also used on many Alpha
> +       systems and is close enough.
> +
> +       ??? We are using a private, stripped down implementation of i82378
> +       so that we can handle the way the ISA interrupts are wired up on
> +       Tsunami-type systems.  We're leaving that (and the rest of the board
> +       peripheral setup) untoucned; we merely need to instantiate the PCI
> +       device node for the bridge, so that operating systems that expect
> +       it to be there will see it.  */

Well, this is not the correct way to do that, so this patch
is unlikely to be accepted. We don't want Frankenstein models.

What is it you miss from the i82378? Why not implement the cy82c693ub?

The code you want to modify is in hw/alpha/dp264.c, see clipper_init().

>      {
>          qemu_irq *isa_irqs;
>  
> +        pci_create_simple(b, PCI_DEVFN(7, 0), "i82378-typhoon");
> +
>          *isa_bus = isa_bus_new(NULL, get_system_memory(), &s->pchip.reg_io,
>                                 &error_abort);
>          isa_irqs = i8259_init(*isa_bus,
> @@ -955,10 +967,96 @@ static const TypeInfo typhoon_iommu_memory_region_info = {
>      .class_init = typhoon_iommu_memory_region_class_init,
>  };
>  
> +/* The following was copied from hw/isa/i82378.c and modified to provide
> +   only the minimal PCI device node.  */
> +
> +/*                            
> + * QEMU Intel i82378 emulation (PCI to ISA bridge) 
> + *                            
> + * Copyright (c) 2010-2011 Herv\xc3\xa9 Poussineau
> + *  
> + * This library is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU Lesser General Public
> + * License as published by the Free Software Foundation; either
> + * version 2 of the License, or (at your option) any later version.
> + *  
> + * This library is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> + * Lesser General Public License for more details.
> + *     
> + * You should have received a copy of the GNU Lesser General Public
> + * License along with this library; if not, see <http://www.gnu.org/licenses/>.
> + */    
> +
> +#include "migration/vmstate.h"
> +
> +#define TYPE_I82378 "i82378-typhoon"
> +#define I82378(obj) \
> +    OBJECT_CHECK(I82378State, (obj), TYPE_I82378)
> +
> +typedef struct I82378State {
> +    PCIDevice parent_obj;
> +} I82378State;
> +
> +static const VMStateDescription vmstate_i82378 = {
> +    .name = "pci-i82378-typhoon",
> +    .version_id = 0,
> +    .minimum_version_id = 0,
> +    .fields = (VMStateField[]) {
> +        VMSTATE_PCI_DEVICE(parent_obj, I82378State), 
> +        VMSTATE_END_OF_LIST()
> +    },                        
> +};                            
> +
> +static void i82378_realize(PCIDevice *pci, Error **errp)
> +{
> +    uint8_t *pci_conf;
> + 
> +    pci_conf = pci->config;
> +    pci_set_word(pci_conf + PCI_COMMAND,
> +                 PCI_COMMAND_IO | PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER);
> +    pci_set_word(pci_conf + PCI_STATUS,
> +                 PCI_STATUS_DEVSEL_MEDIUM);
> + 
> +    pci_config_set_interrupt_pin(pci_conf, 1); /* interrupt pin 0 */
> +}
> +
> +static void i82378_init(Object *obj)
> +{
> +}      
> +
> +static void i82378_class_init(ObjectClass *klass, void *data)
> +{   
> +    PCIDeviceClass *k = PCI_DEVICE_CLASS(klass);
> +    DeviceClass *dc = DEVICE_CLASS(klass);
> +
> +    k->realize = i82378_realize; 
> +    k->vendor_id = PCI_VENDOR_ID_INTEL;
> +    k->device_id = PCI_DEVICE_ID_INTEL_82378;
> +    k->revision = 0x03;
> +    k->class_id = PCI_CLASS_BRIDGE_ISA;
> +    dc->vmsd = &vmstate_i82378;
> +    set_bit(DEVICE_CATEGORY_BRIDGE, dc->categories);
> +}                             
> +
> +static const TypeInfo i82378_typhoon_type_info = {
> +    .name = TYPE_I82378,
> +    .parent = TYPE_PCI_DEVICE,
> +    .instance_size = sizeof(I82378State),
> +    .instance_init = i82378_init,
> +    .class_init = i82378_class_init,
> +    .interfaces = (InterfaceInfo[]) {
> +        { INTERFACE_CONVENTIONAL_PCI_DEVICE },
> +        { },     
> +    },
> +};  
> +
>  static void typhoon_register_types(void)
>  {
>      type_register_static(&typhoon_pcihost_info);
>      type_register_static(&typhoon_iommu_memory_region_info);
> +    type_register_static(&i82378_typhoon_type_info);
>  }
>  
>  type_init(typhoon_register_types)
> diff --git a/hw/rtc/mc146818rtc.c b/hw/rtc/mc146818rtc.c
> index 7a38540cb9..596876cb43 100644
> --- a/hw/rtc/mc146818rtc.c
> +++ b/hw/rtc/mc146818rtc.c
> @@ -155,9 +155,15 @@ static uint32_t rtc_periodic_clock_ticks(RTCState *s)
>  {
>      int period_code;
>  
> +#if 0
> +    /*
> +     * Real hardware sets the PF bit rergardless if it actually
> +     * raises an interrupt.
> +     */
>      if (!(s->cmos_data[RTC_REG_B] & REG_B_PIE)) {
>          return 0;
>       }
> +#endif
>  
>      period_code = s->cmos_data[RTC_REG_A] & 0x0f;
>  
> @@ -944,6 +950,7 @@ static void rtc_realizefn(DeviceState *dev, Error **errp)
>      }
>  
>      s->periodic_timer = timer_new_ns(rtc_clock, rtc_periodic_timer, s);
> +    periodic_timer_update(s, qemu_clock_get_ns(rtc_clock), 0, true);
>      s->update_timer = timer_new_ns(rtc_clock, rtc_update_timer, s);
>      check_update_timer(s);
>  
>

> On Oct 6, 2020, at 9:42 PM, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
> 
> Hi Jason,
> 
> Well, this is not the correct way to do that, so this patch
> is unlikely to be accepted. We don't want Frankenstein models.
> 
> What is it you miss from the i82378? Why not implement the cy82c693ub?
> 
> The code you want to modify is in hw/alpha/dp264.c, see clipper_init().

The Clipper emulation *already* implements a Frankenstein model, a combination of hardware that DEC never shipped.  While some Tsunami-based systems did use the Cypress bridge, none of them, to my knowledge, had the CMD646 IDE controller.  The only system I'm aware of that shipped with a CMD646 IDE controller was the early version of the Miata (EV56 + Pyxis-based Digital Personal Workstation), and that system had the Intel i82378 PCI-ISA bridge.

The are a lot of differences in the Cypress PCI-ISA bridge, but most importantly, it has a different interrupt controller (the ELCR registers are programmed differently), and it has a different built-in IDE controller.  As far as I can tell, Qemu does not currently have emulation for the Cypress interrupt controller nor the Cypress IDE controller.

All Alpha systems that had a PCI-ISA bridge[*] shipped with, as far as I'm aware, one of 3 possible chips:

- Intel i82378 SIO
- Cypress 82C693
- Acer Labs M1533

...and the emulation provided by the existing Clipper model implements the Intel flavor (including the interrupt controller).  I just want software that expects one of those 3 chips to be there to work.  The existing emulation seems to assume that system software will just go out and probe the peripherals in the ISA space, but some operating systems (NetBSD in particular) support Alpha systems that have PCI without any ISA/LPC bus at all, and thus only go looking for ISA peripherals if a PCI-ISA bridge is present.  That is what my change here is intended to do, without changing any of the rest of the hardware configuration.


[*] ...ignoring the weirdo early models that had PCI + EISA + oddly-wired-up junk I/O chips.  Also, I don't have any direct experience with post-Titan models, so I'm not sure what they have...

-- thorpej

On 10/7/20 5:13 PM, Jason Thorpe wrote:
> 
>> On Oct 6, 2020, at 9:42 PM, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
>>
>> Hi Jason,
>>
>> Well, this is not the correct way to do that, so this patch
>> is unlikely to be accepted. We don't want Frankenstein models.
>>
>> What is it you miss from the i82378? Why not implement the cy82c693ub?
>>
>> The code you want to modify is in hw/alpha/dp264.c, see clipper_init().
> 
> The Clipper emulation *already* implements a Frankenstein model, a combination of hardware that DEC never shipped.  While some Tsunami-based systems did use the Cypress bridge, none of them, to my knowledge, had the CMD646 IDE controller.  The only system I'm aware of that shipped with a CMD646 IDE controller was the early version of the Miata (EV56 + Pyxis-based Digital Personal Workstation), and that system had the Intel i82378 PCI-ISA bridge.
> 
> The are a lot of differences in the Cypress PCI-ISA bridge, but most importantly, it has a different interrupt controller (the ELCR registers are programmed differently), and it has a different built-in IDE controller.  As far as I can tell, Qemu does not currently have emulation for the Cypress interrupt controller nor the Cypress IDE controller.
> 
> All Alpha systems that had a PCI-ISA bridge[*] shipped with, as far as I'm aware, one of 3 possible chips:
> 
> - Intel i82378 SIO
> - Cypress 82C693
> - Acer Labs M1533
> 
> ...and the emulation provided by the existing Clipper model implements the Intel flavor (including the interrupt controller).  I just want software that expects one of those 3 chips to be there to work.  The existing emulation seems to assume that system software will just go out and probe the peripherals in the ISA space, but some operating systems (NetBSD in particular) support Alpha systems that have PCI without any ISA/LPC bus at all, and thus only go looking for ISA peripherals if a PCI-ISA bridge is present.  That is what my change here is intended to do, without changing any of the rest of the hardware configuration.

Can you provide the information to reproduce your problem
with a NetBSD image please?

> 
> 
> [*] ...ignoring the weirdo early models that had PCI + EISA + oddly-wired-up junk I/O chips.  Also, I don't have any direct experience with post-Titan models, so I'm not sure what they have...
> 
> -- thorpej
> 
>

On 10/6/20 8:24 PM, Jason Thorpe wrote:
> Along with the previous patch set for qemu-palcode, these changes
> allow NetBSD/alpha to run in qemu-system-alpha.
> 
> - Allow a the minimum PCI device number to be passed to typhoon_init().
>   Use this to specify a minimum PCI device number of 1 in the Clipper
>   system emulation, since that's the lowest PCI device number with
>   valid interrupt mappings on that system.
> - Instead of passing just the CPU count in trap_arg2, also pass other
>   configuration information.  Define the first config bit (bit 6) to
>   reflect the "-nographic" option; PALcode will use this to initialize
>   the Console Terminal Block to specify serial console.
> - Attach a minimal i82378 SIO PCI node; some operating systems won't
>   scan for ISA devices unless a PCI-ISA bridge is detected.  Real hardware
>   would have had a Cypress or ALI bridge, but many Alpha models shipped
>   with a i82378, and most operating sytems are not picky about the specific
>   model.
> - Modify the mc146818rtc emulation to ensure the timer is started as soon
>   as the device is realized, and update the PF bit in REG_C even if it is
>   not going to result in raising an interrupt, as the real hardware does.
> 

This will have to be split up -- again, essentially by bullet point -- but it
looks reasonable enough.

You need to mind CODING_STYLE.  Run your patches through
./scripts/checkpatch.pl for a rough guide.


r~