Hi Igor, Jonathan,
On 6/20/25 6:13 PM, Jonathan Cameron wrote:
> On Fri, 20 Jun 2025 14:38:22 +0200
> Igor Mammedov <[email protected]> wrote:
>
>> On Fri, 20 Jun 2025 10:35:38 +0100
>> Jonathan Cameron <[email protected]> wrote:
>>
>>> On Mon, 16 Jun 2025 11:46:46 +0200
>>> Eric Auger <[email protected]> wrote:
>>>
>>>> Modify the DSDT ACPI table to enable ACPI PCI hotplug.
>>>>
>>>> Signed-off-by: Eric Auger <[email protected]>
>>>>
>>>> ---
>>>> v2 -> v3:
>>>> - use ACPI_PCIHP_SIZE instead of 0x1000 (Igor)
>>>> - use cihp_state->use_acpi_hotplug_bridge
>>> pcihp_state
>>>
>>> Takes a bit of searching to find the various bits of the
>>> same support on x86 but this seems to match up.
>>> Exactly when things are built does vary but not I think
>>> in a way that matters. e.g. I think on x86 the
>>> EDSM stuff is built whether or not we have pcihp enabled
>>> whereas here you've made it conditional on using acpi
>>> hp. Perhaps a tiny bit more description on that would be
>>> useful if you do a v4?
>> edsm should be built regardless of pcihp
>> (well intention was there, whether I messed it up or not I don't know)
>>
>> idea is that non hotplug ports can have a static acpi-index,
>> so it doesn't depend on pcihp.
> That makes sense - so here should that edsm feature be enabled whether
> or not we have pcihp_state->use_acpi_hotplug_bridge == true
>
> i.e. is it really a separate thing from the rest of this series?
Further studying this comment,
EDSM is invoked by code generated in aml_pci_static_endpoint_dsm() whcih
itself is invoked by build_append_pci_bus_devices()
So to me it means that if we generate edsm unconditionally we also need
to call the following block unconditionnally
+ aml_append(pci0_scope, build_pci_bridge_edsm());
+ build_append_pci_bus_devices(pci0_scope, vms->bus);
+ if (object_property_find(OBJECT(vms->bus), ACPI_PCIHP_PROP_BSEL)) {
+ build_append_pcihp_slots(pci0_scope, vms->bus);
+ }
which seems to be done that way in hw/i386/acpi-build.c/build_dsdt()
Igor, if I recall correctly you said that addition changes related to
"S%.02X" could change the guest ABI. And in that case this wouldn't be
guarded by any new option/compat. So that's annoying.
By the way I tested static acpi-index on ARM with resulting code and it
does not not seem to work - maybe I try with a wrong topology though
(pcie root port + virtio-net acpi-index)-. I have not checked on x86 yet.
So I wonder if it makes sense to do that refinement now. Maybe we can
check try to improve that afterwards?
What do you think?
Eric
>
> Thanks,
>
> J
>>> Reviewed-by: Jonathan Cameron <[email protected]>
>>>
>>>
>>>> ---
>>>> include/hw/acpi/pcihp.h | 2 ++
>>>> include/hw/arm/virt.h | 1 +
>>>> hw/arm/virt-acpi-build.c | 22 ++++++++++++++++++++++
>>>> hw/arm/virt.c | 2 ++
>>>> hw/arm/Kconfig | 2 ++
>>>> 5 files changed, 29 insertions(+)
>>>>
>>>> diff --git a/include/hw/acpi/pcihp.h b/include/hw/acpi/pcihp.h
>>>> index 5506a58862..9ff548650b 100644
>>>> --- a/include/hw/acpi/pcihp.h
>>>> +++ b/include/hw/acpi/pcihp.h
>>>> @@ -38,6 +38,8 @@
>>>> #define ACPI_PCIHP_SEJ_BASE 0x8
>>>> #define ACPI_PCIHP_BNMR_BASE 0x10
>>>>
>>>> +#define ACPI_PCIHP_SIZE 0x0018
>>>> +
>>>> typedef struct AcpiPciHpPciStatus {
>>>> uint32_t up;
>>>> uint32_t down;
>>>> diff --git a/include/hw/arm/virt.h b/include/hw/arm/virt.h
>>>> index 9a1b0f53d2..0ed2e6b732 100644
>>>> --- a/include/hw/arm/virt.h
>>>> +++ b/include/hw/arm/virt.h
>>>> @@ -79,6 +79,7 @@ enum {
>>>> VIRT_ACPI_GED,
>>>> VIRT_NVDIMM_ACPI,
>>>> VIRT_PVTIME,
>>>> + VIRT_ACPI_PCIHP,
>>>> VIRT_LOWMEMMAP_LAST,
>>>> };
>>>>
>>>> diff --git a/hw/arm/virt-acpi-build.c b/hw/arm/virt-acpi-build.c
>>>> index d7547c8d3b..a2e58288f8 100644
>>>> --- a/hw/arm/virt-acpi-build.c
>>>> +++ b/hw/arm/virt-acpi-build.c
>>>> @@ -34,6 +34,7 @@
>>>> #include "hw/core/cpu.h"
>>>> #include "hw/acpi/acpi-defs.h"
>>>> #include "hw/acpi/acpi.h"
>>>> +#include "hw/acpi/pcihp.h"
>>>> #include "hw/nvram/fw_cfg_acpi.h"
>>>> #include "hw/acpi/bios-linker-loader.h"
>>>> #include "hw/acpi/aml-build.h"
>>>> @@ -809,6 +810,8 @@ static void
>>>> build_dsdt(GArray *table_data, BIOSLinker *linker, VirtMachineState *vms)
>>>> {
>>>> VirtMachineClass *vmc = VIRT_MACHINE_GET_CLASS(vms);
>>>> + AcpiGedState *acpi_ged_state = ACPI_GED(vms->acpi_dev);
>>>> + AcpiPciHpState *pcihp_state = &acpi_ged_state->pcihp_state;
>>>> Aml *scope, *dsdt;
>>>> MachineState *ms = MACHINE(vms);
>>>> const MemMapEntry *memmap = vms->memmap;
>>>> @@ -868,6 +871,25 @@ build_dsdt(GArray *table_data, BIOSLinker *linker,
>>>> VirtMachineState *vms)
>>>>
>>>> aml_append(dsdt, scope);
>>>>
>>>> + if (pcihp_state->use_acpi_hotplug_bridge) {
>>>> + Aml *pci0_scope = aml_scope("\\_SB.PCI0");
>>>> +
>>>> + aml_append(pci0_scope, aml_pci_edsm());
>>>> + build_acpi_pci_hotplug(dsdt, AML_SYSTEM_MEMORY,
>>>> + memmap[VIRT_ACPI_PCIHP].base);
>>>> + build_append_pcihp_resources(pci0_scope,
>>>> + memmap[VIRT_ACPI_PCIHP].base,
>>>> + memmap[VIRT_ACPI_PCIHP].size);
>>>> +
>>>> + /* Scan all PCI buses. Generate tables to support hotplug. */
>>>> + build_append_pci_bus_devices(pci0_scope, vms->bus);
>>>> + if (object_property_find(OBJECT(vms->bus), ACPI_PCIHP_PROP_BSEL))
>>>> {
>>>> + build_append_pcihp_slots(pci0_scope, vms->bus);
>>>> + }
>>>> + build_append_notification_callback(pci0_scope, vms->bus);
>>>> + aml_append(dsdt, pci0_scope);
>>>> + }
>>>> +
>>>> /* copy AML table into ACPI tables blob */
>>>> g_array_append_vals(table_data, dsdt->buf->data, dsdt->buf->len);
>>>>
>>>
>>
