[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[RFC 0/3] virt: wire up NS EL2 virtual timer IRQ
From: |
Peter Maydell |
Subject: |
[RFC 0/3] virt: wire up NS EL2 virtual timer IRQ |
Date: |
Tue, 19 Sep 2023 11:12:37 +0100 |
This patchset is an RFC that wires up the NS EL2 virtual timer IRQ on
the virt board, similarly to what
20230913140610.214893-1-marcin.juszkiewicz@linaro.org/">https://patchew.org/QEMU/20230913140610.214893-1-marcin.juszkiewicz@linaro.org/
does for the sbsa-ref board.
Patches 1 and 3 are the usual dance to keep the ACPI unit tests happy
with the change to the ACPI table contents; patch 2 is the meat.
This is an RFC for two reasons:
(1) I'm not very familiar with ACPI, and patch 2 needs to update the
ACPI GTDT table to report the interrupt number. In particular, this
means we need the rev 3 of this table (present in ACPI spec 6.3 and
later), not the rev 2 we currently report. I'm not sure if it's
permitted to rev just this table, or if we would need to upgrade all
our ACPI tables to the newer spec first.
(2) The change causes EDK2 (UEFI) to assert on startup:
ASSERT [ArmTimerDxe]
/home/kraxel/projects/qemu/roms/edk2/ArmVirtPkg/Library/ArmVirtTimerFdtClientLib/ArmVirtTimerFdtClientLib.c(72):
PropSize == 36 || PropSize == 48
This is because the EDK2 code that consumes the QEMU generated device
tree blob is incorrectly insisting that the architectural-timer
interrupts property has only 3 or 4 entries, so it falls over if
given a dtb with the 5th entry for the EL2 virtual timer irq. In
particular this breaks the avocado test:
machine_aarch64_virt.py:Aarch64VirtMachine.test_alpine_virt_tcg_gic_max
I'm not entirely sure what to do about this -- we can get EDK2 fixed
and update our own test case, but there's a lot of binaries out there
in the wild that won't run if we just update the virt board the way
this patchset does. We could perhaps make the virt board change be
dependent on machine type version, as a way to let users fall back to
the old behaviour.
I'm putting this patchset out on the list to get opinions and
review on those two points.
thanks
-- PMM
Peter Maydell (3):
tests/qtest/bios-tables-test: Allow changes to virt GTDT
hw/arm/virt: Wire up non-secure EL2 virtual timer IRQ
tests/qtest/bios-tables-test: Update virt/GTDT golden reference
include/hw/arm/virt.h | 2 ++
hw/arm/virt-acpi-build.c | 16 ++++++++++++----
hw/arm/virt.c | 29 ++++++++++++++++++++++++++++-
tests/data/acpi/virt/GTDT | Bin 96 -> 104 bytes
4 files changed, 42 insertions(+), 5 deletions(-)
--
2.34.1
- [RFC 0/3] virt: wire up NS EL2 virtual timer IRQ,
Peter Maydell <=
- [RFC 1/3] tests/qtest/bios-tables-test: Allow changes to virt GTDT, Peter Maydell, 2023/09/19
- [RFC 3/3] tests/qtest/bios-tables-test: Update virt/GTDT golden reference, Peter Maydell, 2023/09/19
- [RFC 2/3] hw/arm/virt: Wire up non-secure EL2 virtual timer IRQ, Peter Maydell, 2023/09/19
- Re: [RFC 0/3] virt: wire up NS EL2 virtual timer IRQ, Ard Biesheuvel, 2023/09/19
- Re: [RFC 0/3] virt: wire up NS EL2 virtual timer IRQ, Leif Lindholm, 2023/09/19