Re: [PATCH v2 09/21] arm: virt: create sbsa_gwdt watchdog

[Igor Mammedov <imammedo@redhat.com>]

On Mon, 1 Jun 2026 18:37:05 +0200
Eric Auger <eauger@redhat.com> wrote:

> Hi Igor,
> 
> On 5/28/26 4:37 PM, Igor Mammedov wrote:
> > On Thu, 28 May 2026 11:59:06 +0200
> > Eric Auger <eauger@redhat.com> wrote:
> >   
> >> Hi,
> >>
> >> On 5/28/26 11:04 AM, Eric Auger wrote:  
> >>> Hi Igor,
> >>>
> >>> On 5/27/26 2:49 PM, Igor Mammedov wrote:    
> >>>> On Tue, 26 May 2026 15:58:08 +0200
> >>>> Eric Auger <eauger@redhat.com> wrote:
> >>>>    
> >>>>> Hi Igor,
> >>>>>
> >>>>> On 3/3/26 10:25 AM, Igor Mammedov wrote:    
> >>>>>> Allow to use SBSA generic watchdog with virt machine type.
> >>>>>> (includes conditional generation of corresponding FDT and
> >>>>>> ACPI GTDT descriptors)
> >>>>>>
> >>>>>> Use '-device sbsa_gwdt' to command line to enable it.      
> >>>>> I thank that in general Peter is not a big fan of dynamic instantiation
> >>>>> of sysbus devices. Don't you have means to instantiate it statically in
> >>>>> arm virt memory map?    
> > 
> > On Peter's part, there weren't any objection to the idea using -device.
> > (neither in v1 nor in this revision).
> > 
> >   
> >>>> /me neither
> >>>>
> >>>> in previous version there was a machine option to enable the feature
> >>>> to that was rejected in favor of device.
> >>>>
> >>>> in this case for platform  -device of sysbus based device, I don't think
> >>>> we can do better that dynamic sysbus.     
> >>>
> >>> sorry I may have missed some previous discussion details. As far as I
> >>> can read, I understand you could instantiate the wdt in a free MMIO
> >>> slot, unconfitionnaly just as it is done in hw/arm/sbsa-ref.c. Maybe add
> >>> a compat to avoid having it by default on older machine types. Then
> >>> about the type of the integration gtdt or wat, can't you have a device
> >>> compat?    
> >> After further reading of the previous version thread, I wonder if you
> >> shouldn't always instantiate the watchdog in the machine instead.
> >>
> >> Add a device property allowing to switch between GWDT or WDAT version.
> >>
> >> In arm virt,
> >>
> >> 1) if you discover the GWDT version is selected (default) you generate
> >> both dt and GTDT ACPI table
> >> 2) if you discover the WDAT version is selected, you only generate WDAT
> >> ACPI description and you output an error in case of DT boot.
> >>
> >> You may add a machine no_wdt flag to handle compats for older machine
> >> types if needed.  
> > 
> > rereading v1 thread it seems that there weren't much objection to adding
> > a dedicated virt machine property to enable 'wdat' [1] /like what is done to q35/.
> > 
> > problem was in conflating too generic whatchdog=none/auto/native/acpi into
> > one knob and discussion quickly went down rabbit hole of semantics.
> > 
> > For Q35, we don't really have a choice as it's already builtin device,
> > but for arm/virt we can do better than a bunch of compats and proxy
> > properties. (as Markus noted in v1 review, we've done that/still do
> > but it doesn't scale well)
> > 
> > I can go old route and revert to simplified v1 that would:
> >   * make gwdt builtin device on arm/virt board
> >     (i.e. default to SBSA native watchdog with GTDT ACPI table
> >      and relevant FDT entries)  
> would make sense to me.

forgot about why it's not default ATM,
the reason being that native/default GWDT causes issues on Windows
(only WDAT variant works).
the last 2 patches trying to do some damage control,
but what we can do on qemu side is limited.

and builtin gwdt we would end up with a mess of gwdt=on|of and wdat=o|n,

So I'd rather have watchdog as an optional device included to VM
when asked by user.
With side benefit reducing attack surface when device is not
needed compared to built-in by default variant.
 

> >   * add a machine.gwdt-wdat=on|off property
> >      1. that will make gwdt to use 1KHz frequency
> >      2. enable WDAT ACPI table
> >      3. disable GTDT ACPI table and relevant FDT entries
> >         as they are mutually exclusive with #1 point 
> >   * add a compat knob to disable gwdt on old machine types.  
> Either that or device prop setting as we have for ACPI PCI HP (technique
> you advocated at some point).

that was for symmetry with existing acpi pci hp feature on x86 if I recall
correctly. There is no much options for builtin devices, we either
 * use -global to tweak them  (historical variant)
 * more recent trend is proxy property in -machine
both not great variants, I've tried in v1 to propose generic
-machine watchdog=foo to at least limit a zoo of options, but that was rejected.
 

> > 
> > Re-spinning that wouldn't take much effort, all the code already exist,
> > just needs mostly reshuffling. 
> > 
> > 1) I think Daniel was ok having something like gwdt-wdat=on|off
> >    from libvirt pov.
> > 
> > That said,
> > Having watchdog as an optional -device is much cleaner compared to
> > adding knobs to machine (both internal compats and external to enable
> > specific watchdog).  
> I am OK with that too. However in general I think Peter prefers having
> them statically instantiated if possible. All the more so I think that
> having a watchdog, always instantiated, in its GTDT/FDT form (not WDAT
> ACPI only) in machvirt would make sense in general and does not make
> much trouble in terms of MMIO layout.

If there is no objections, lets go with -device variant

If dynamic sysbus resources is of concern, I probably can pin/make them
static, using device plug handler (virt_machine_device_plug_cb).
Let me try and see what could be done wrt static resources.

At this point series needs a rebase, due to conflicts.
So I'll respin it with staic variant if it works out or as it's now.

But while I'm at it, pls review the remaining not reviewed yet patches,
as they aren't likely to change. Especially generic GWDT fixes 18-21/21. 

> Good (well bad actualy) example
> of compats/props is set of
> > ITS ones, one eventually can untangle what depends on what
> > and how that works together, but I'd rather avoid doing that. 
> > 
> > Perhaps I went overboard with move to -device.
> > Anyways, all I can do is suggest in my opinion a better way,
> > but it's upto arm folks to pick a preferred approach.
> >   
> >> Would that make sense?
> >>
> >> Otherwise, can you point me the msgid where the -device was prefered?  
> > Sorry, rereading v1 thread, -device was just an idea to try
> > (but by no means a preferred/consensus one).  
> This is what I understood too.
> 
> Thanks
> 
> Eric
> >   
> >>
> >> Thanks
> >>
> >> Eric  
> >>>
> >>> Thanks
> >>>
> >>> Eric
> >>>     
> >>>>
> >>>>      
> >>>>>
> >>>>> Eric    
> >>>>>>
> >>>>>> Tested with Fedora 43:
> >>>>>>   FDT: -M virt,acpi=off -device sbsa_gwdt
> >>>>>>   ACPI: -M virt -device sbsa_gwdt
> >>>>>>
> >>>>>> Signed-off-by: Igor Mammedov <imammedo@redhat.com>
> >>>>>> ---
> >>>>>>  hw/arm/Kconfig           |  1 +
> >>>>>>  hw/arm/virt-acpi-build.c | 33 +++++++++++++++++++++++++++++++--
> >>>>>>  hw/arm/virt.c            |  2 ++
> >>>>>>  hw/core/sysbus-fdt.c     | 32 ++++++++++++++++++++++++++++++++
> >>>>>>  hw/watchdog/sbsa_gwdt.c  |  1 +
> >>>>>>  5 files changed, 67 insertions(+), 2 deletions(-)
> >>>>>>
> >>>>>> diff --git a/hw/arm/Kconfig b/hw/arm/Kconfig
> >>>>>> index c66c452737..1222efadd1 100644
> >>>>>> --- a/hw/arm/Kconfig
> >>>>>> +++ b/hw/arm/Kconfig
> >>>>>> @@ -36,6 +36,7 @@ config ARM_VIRT
> >>>>>>      select VIRTIO_MEM_SUPPORTED
> >>>>>>      select ACPI_CXL
> >>>>>>      select ACPI_HMAT
> >>>>>> +    select WDT_SBSA
> >>>>>>  
> >>>>>>  config CUBIEBOARD
> >>>>>>      bool
> >>>>>> diff --git a/hw/arm/virt-acpi-build.c b/hw/arm/virt-acpi-build.c
> >>>>>> index 719d2f994e..9c14b1d4d5 100644
> >>>>>> --- a/hw/arm/virt-acpi-build.c
> >>>>>> +++ b/hw/arm/virt-acpi-build.c
> >>>>>> @@ -64,6 +64,7 @@
> >>>>>>  #include "hw/virtio/virtio-acpi.h"
> >>>>>>  #include "target/arm/cpu.h"
> >>>>>>  #include "target/arm/multiprocessing.h"
> >>>>>> +#include "hw/watchdog/sbsa_gwdt.h"
> >>>>>>  
> >>>>>>  #define ARM_SPI_BASE 32
> >>>>>>  
> >>>>>> @@ -863,6 +864,8 @@ build_gtdt(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms)
> >>>>>>      const uint32_t irqflags = 0;  /* Interrupt is Level triggered  */
> >>>>>>      AcpiTable table = { .sig = "GTDT", .rev = 3, .oem_id = vms->oem_id,
> >>>>>>                          .oem_table_id = vms->oem_table_id };
> >>>>>> +    uint32_t gtdt_start = table_data->len;
> >>>>>> +    Object *wdt = object_resolve_type_unambiguous(TYPE_WDT_SBSA, NULL);
> >>>>>>  
> >>>>>>      acpi_table_begin(&table, table_data);
> >>>>>>  
> >>>>>> @@ -893,10 +896,15 @@ build_gtdt(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms)
> >>>>>>      build_append_int_noprefix(table_data, irqflags, 4);
> >>>>>>      /* CntReadBase Physical address */
> >>>>>>      build_append_int_noprefix(table_data, 0xFFFFFFFFFFFFFFFF, 8);
> >>>>>> +
> >>>>>>      /* Platform Timer Count */
> >>>>>> -    build_append_int_noprefix(table_data, 0, 4);
> >>>>>> +    build_append_int_noprefix(table_data,  wdt ? 1 : 0, 4);
> >>>>>>      /* Platform Timer Offset */
> >>>>>> -    build_append_int_noprefix(table_data, 0, 4);
> >>>>>> +    build_append_int_noprefix(table_data,
> >>>>>> +        wdt ? (table_data->len - gtdt_start) +
> >>>>>> +              4 + 4 + 4 /* len of this & following 2 fields to skip */
> >>>>>> +            : 0, 4);
> >>>>>> +
> >>>>>>      if (vms->ns_el2_virt_timer_irq) {
> >>>>>>          /* Virtual EL2 Timer GSIV */
> >>>>>>          build_append_int_noprefix(table_data, ARCH_TIMER_NS_EL2_VIRT_IRQ, 4);
> >>>>>> @@ -906,6 +914,27 @@ build_gtdt(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms)
> >>>>>>          build_append_int_noprefix(table_data, 0, 4);
> >>>>>>          build_append_int_noprefix(table_data, 0, 4);
> >>>>>>      }
> >>>>>> +
> >>>>>> +    /* ACPI 6.5 spec: 5.2.25.2 ARM Generic Watchdog Structure (Table 5-124) */
> >>>>>> +    if (wdt) {
> >>>>>> +        PlatformBusDevice *pbus = PLATFORM_BUS_DEVICE(vms->platform_bus_dev);
> >>>>>> +        SysBusDevice *sd = SYS_BUS_DEVICE(wdt);
> >>>>>> +        hwaddr mmio_base = vms->memmap[VIRT_PLATFORM_BUS].base;
> >>>>>> +        hwaddr rbase = mmio_base + platform_bus_get_mmio_addr(pbus, sd, 0);
> >>>>>> +        hwaddr cbase = mmio_base + platform_bus_get_mmio_addr(pbus, sd, 1);
> >>>>>> +        int irq = ARM_SPI_BASE + vms->irqmap[VIRT_PLATFORM_BUS] +
> >>>>>> +                  platform_bus_get_irqn(pbus, sd, 0);
> >>>>>> +
> >>>>>> +        build_append_int_noprefix(table_data, 1 /* Type: Watchdog GT */, 1);
> >>>>>> +        build_append_int_noprefix(table_data, 28 /* Length */, 2);
> >>>>>> +        build_append_int_noprefix(table_data, 0, 1);  /* Reserved */
> >>>>>> +        /* RefreshFrame Physical Address */
> >>>>>> +        build_append_int_noprefix(table_data, rbase, 8);
> >>>>>> +        /* WatchdogControlFrame Physical Address */
> >>>>>> +        build_append_int_noprefix(table_data, cbase, 8);
> >>>>>> +        build_append_int_noprefix(table_data, irq, 4); /* Watchdog Timer GSIV */
> >>>>>> +        build_append_int_noprefix(table_data, 0, 4); /* Watchdog Timer Flags */
> >>>>>> +    }
> >>>>>>      acpi_table_end(linker, &table);
> >>>>>>  }
> >>>>>>  
> >>>>>> diff --git a/hw/arm/virt.c b/hw/arm/virt.c
> >>>>>> index cab2e21e8a..a87991b591 100644
> >>>>>> --- a/hw/arm/virt.c
> >>>>>> +++ b/hw/arm/virt.c
> >>>>>> @@ -95,6 +95,7 @@
> >>>>>>  #include "hw/cxl/cxl.h"
> >>>>>>  #include "hw/cxl/cxl_host.h"
> >>>>>>  #include "qemu/guest-random.h"
> >>>>>> +#include "hw/watchdog/sbsa_gwdt.h"
> >>>>>>  
> >>>>>>  static GlobalProperty arm_virt_compat_defaults[] = {
> >>>>>>      { TYPE_VIRTIO_IOMMU_PCI, "aw-bits", "48" },
> >>>>>> @@ -3474,6 +3475,7 @@ static void virt_machine_class_init(ObjectClass *oc, const void *data)
> >>>>>>      machine_class_allow_dynamic_sysbus_dev(mc, TYPE_RAMFB_DEVICE);
> >>>>>>      machine_class_allow_dynamic_sysbus_dev(mc, TYPE_UEFI_VARS_SYSBUS);
> >>>>>>      machine_class_allow_dynamic_sysbus_dev(mc, TYPE_ARM_SMMUV3);
> >>>>>> +    machine_class_allow_dynamic_sysbus_dev(mc, TYPE_WDT_SBSA);
> >>>>>>  #ifdef CONFIG_TPM
> >>>>>>      machine_class_allow_dynamic_sysbus_dev(mc, TYPE_TPM_TIS_SYSBUS);
> >>>>>>  #endif
> >>>>>> diff --git a/hw/core/sysbus-fdt.c b/hw/core/sysbus-fdt.c
> >>>>>> index 89d0c46445..76aa1bc579 100644
> >>>>>> --- a/hw/core/sysbus-fdt.c
> >>>>>> +++ b/hw/core/sysbus-fdt.c
> >>>>>> @@ -36,6 +36,7 @@
> >>>>>>  #include "hw/display/ramfb.h"
> >>>>>>  #include "hw/uefi/var-service-api.h"
> >>>>>>  #include "hw/arm/fdt.h"
> >>>>>> +#include "hw/watchdog/sbsa_gwdt.h"
> >>>>>>  
> >>>>>>  /*
> >>>>>>   * internal struct that contains the information to create dynamic
> >>>>>> @@ -118,6 +119,36 @@ static int add_uefi_vars_node(SysBusDevice *sbdev, void *opaque)
> >>>>>>      return 0;
> >>>>>>  }
> >>>>>>  
> >>>>>> +static int add_sbsa_gwdt_node(SysBusDevice *sbdev, void *opaque)
> >>>>>> +{
> >>>>>> +    PlatformBusFDTData *data = opaque;
> >>>>>> +    PlatformBusDevice *pbus = data->pbus;
> >>>>>> +    const char *parent_node = data->pbus_node_name;
> >>>>>> +    void *fdt = data->fdt;
> >>>>>> +    uint64_t cbase, rbase;
> >>>>>> +    char *nodename;
> >>>>>> +    int irq;
> >>>>>> +
> >>>>>> +    cbase = platform_bus_get_mmio_addr(pbus, sbdev, 1);  /* control frame */
> >>>>>> +    rbase = platform_bus_get_mmio_addr(pbus, sbdev, 0);  /* refresh frame */
> >>>>>> +    irq = platform_bus_get_irqn(pbus, sbdev, 0);
> >>>>>> +
> >>>>>> +    nodename = g_strdup_printf("%s/watchdog@%" PRIx64, parent_node, cbase);
> >>>>>> +    qemu_fdt_add_subnode(fdt, nodename);
> >>>>>> +
> >>>>>> +    qemu_fdt_setprop_string(fdt, nodename, "compatible", "arm,sbsa-gwdt");
> >>>>>> +    qemu_fdt_setprop_cells(fdt, nodename, "reg",
> >>>>>> +                           cbase, SBSA_GWDT_CMMIO_SIZE,
> >>>>>> +                           rbase, SBSA_GWDT_RMMIO_SIZE);
> >>>>>> +    qemu_fdt_setprop_cells(fdt, nodename, "interrupts",
> >>>>>> +                           GIC_FDT_IRQ_TYPE_SPI, data->irq_start + irq,
> >>>>>> +                           GIC_FDT_IRQ_FLAGS_LEVEL_HI);
> >>>>>> +    qemu_fdt_setprop_cell(fdt, nodename, "timeout-sec", 30);
> >>>>>> +
> >>>>>> +    g_free(nodename);
> >>>>>> +    return 0;
> >>>>>> +}
> >>>>>> +
> >>>>>>  static int no_fdt_node(SysBusDevice *sbdev, void *opaque)
> >>>>>>  {
> >>>>>>      return 0;
> >>>>>> @@ -140,6 +171,7 @@ static const BindingEntry bindings[] = {
> >>>>>>      TYPE_BINDING(TYPE_ARM_SMMUV3, no_fdt_node),
> >>>>>>      TYPE_BINDING(TYPE_RAMFB_DEVICE, no_fdt_node),
> >>>>>>      TYPE_BINDING(TYPE_UEFI_VARS_SYSBUS, add_uefi_vars_node),
> >>>>>> +    TYPE_BINDING(TYPE_WDT_SBSA, add_sbsa_gwdt_node),
> >>>>>>      TYPE_BINDING("", NULL), /* last element */
> >>>>>>  };
> >>>>>>  
> >>>>>> diff --git a/hw/watchdog/sbsa_gwdt.c b/hw/watchdog/sbsa_gwdt.c
> >>>>>> index acb970e8b3..c4dd8005b7 100644
> >>>>>> --- a/hw/watchdog/sbsa_gwdt.c
> >>>>>> +++ b/hw/watchdog/sbsa_gwdt.c
> >>>>>> @@ -285,6 +285,7 @@ static void wdt_sbsa_gwdt_class_init(ObjectClass *klass, const void *data)
> >>>>>>      dc->realize = wdt_sbsa_gwdt_realize;
> >>>>>>      device_class_set_legacy_reset(dc, wdt_sbsa_gwdt_reset);
> >>>>>>      dc->hotpluggable = false;
> >>>>>> +    dc->user_creatable = true;
> >>>>>>      set_bit(DEVICE_CATEGORY_WATCHDOG, dc->categories);
> >>>>>>      dc->vmsd = &vmstate_sbsa_gwdt;
> >>>>>>      dc->desc = "SBSA-compliant generic watchdog device";      
> >>>>>    
> >>>>    
> >>>     
> >>  
> >   
>