[PATCH v4 00/19] KVM GICv3 emulation
eric.auger at linaro.org
Mon Nov 24 01:33:16 PST 2014
On 11/14/2014 11:07 AM, Andre Przywara wrote:
> This is version 4 of the GICv3 guest emulation series.
> As the previous series, this is based on v3.18-rc2. I didn't bother to
> update this to keep it diff-able with v3 of the patch series.
> I addressed most of the comments from Christoffer's last review round
> (mange tak!). There were numerous changes, though mostly reworking and
> style changes. To ease spotting them I updated the kvm-gicv3/v3 branch
> in the repo mentioned below to carry all the delta patches. Those
> patches are just for reference to see what has changed between
> v3 and v4. For review and all other purposes please use the v4 branch.
> For an changelog summary see below, also each patch carries a
> changelog now.
> Patch 02, 04, 06, 07, 09, 10, 11 have had no code changes compared to
> their previous counterparts.
> The (v3) patch 06 is gone now, the new patch 14 has changed dramatically
> and a new patch (16/19) has been added.
> I quickly tested this version with a GICv3 capable fast model in all
> endianness modes (LE guest on LE host, BE on LE, LE on BE, BE on BE).
> Both a GICv2 and a GICv3 guest were booted in all four combinations.
> In contrast to the v3 commit message, cross-endianness was working
> fine already with the previous patch series.
> A git repo hosting all these patches lives in the kvm-gicv3/v4 branch
> GICv3 is the ARM generic interrupt controller designed to overcome
> some limits of the prevalent GICv2. Most notably it lifts the 8-CPU
> limit. Though with Linux-3.17 Marc introduced support for hosts to
> use a GICv3, the CPU limitation still applies to KVM guests, since
> the current code emulates a GICv2 only.
> Also, GICv2 backward compatibility being optional in GICv3, a number
> of systems won't be able to run GICv2 guests.
> This patch series provides code to emulate a GICv3 distributor and
> redistributor for any KVM guest. It requires a GICv3 in the host to
> work. With those patches one can run guests efficiently on any GICv3
> host. It has the following features:
> - Affinity routing (support for up to 255 VCPUs, more possible)
> - System registers (as opposed to MMIO access)
> - No ITS
> - No priority support (as the GICv2 emulation)
> - No save / restore support so far (will be added soon)
> - Only Group1 interrupts support
> The first patches actually refactor the current VGIC code to make
> room for a different VGIC model to be dropped in with Patch 16.
> The remaining patches connect the new model to the kernel backend and
> the userland facing code.
> The series goes on top of v3.18-rc2.
> The necessary patches for kvmtool to enable the guest's GICv3 have
> been posted here before , an updated version will follow soon.
> There was some testing on the fast model with some I/O and interrupt
> affinity shuffling in a Linux guest with a varying number of VCPUs as
> well as some testing on a Juno board (GICv2 only, to spot regressions).
> Please review and test.
> I would be grateful for people to test for GICv2 regressions also
> (so on a GICv2 host with current kvmtool/qemu), as there is quite
> some refactoring on that front.
I tested your kvm-gicv3/v4 branch on Calxeda Midway with GICv2 host in
QEMU/VFIO passthrough use case. It behaves as expected.
> Much of the code was inspired by MarcZ, also kudos to him for doing
> the rather painful rebase on top of v3.17-rc1.
>  https://lists.cs.columbia.edu/pipermail/kvmarm/2014-June/010086.html
> Changes v3 ... v4:
> * bug-fix in handling GICv3 redistributor CFG register
> * move set/get_lr from gic_vm_ops back to vgic_ops (get rid of v3 06/19)
> * getting rid of init_emul() at all
> * rework guest GIC model initialization
> * use non-atomic bit-set and bit-clear functions
> * split up handle_mmio_misc* into multiple functions
> * refine handling of some reserved registers
> * use symbolic names for ICC_SGI1R_EL1 register fields (new patch 16/19)
> * move private parameter from MMIO accessors to struct kvm_mmio_exit
> * added documentation of new GICv3 guest device
> * added lots of comments
> * some renaming of identifiers
> * minor changes in style and code flow of various functions
> Changes v2 ... v3:
> * rebase to v3.18-rc2
> * adapt to new kvm_register_device() function
> * split up vm_ops patch and the GICv2 split-off patch to ease review
> * various smaller changes due to Christoffer's review
> * fix compilation for arm
> * remove support for trapping SGI sysreg accesses on arm hosts
> Changes v1 ... v2:
> * rebase to v3.17-rc1, caused quite some changes to the init code
> * new 9/15 patch to make 10/15 smaller
> * fix wrongly ordered cp15 register trap entry (MarcZ)
> * fix SGI broadcast (thanks to wanghaibin for spotting)
> * fix broken bailout path in kvm_vgic_create (wanghaibin)
> * check return value of init_emulation_ops() (wanghaibin)
> * fix return value check in vgic_[sg]et_attr()
> * add header inclusion guards
> * remove double definition of VCPU_NOT_ALLOCATED
> * some code move-around
> * whitespace fixes
> Andre Przywara (19):
> arm/arm64: KVM: rework MPIDR assignment and add accessors
> arm/arm64: KVM: pass down user space provided GIC type into vGIC code
> arm/arm64: KVM: refactor vgic_handle_mmio() function
> arm/arm64: KVM: wrap 64 bit MMIO accesses with two 32 bit ones
> arm/arm64: KVM: introduce per-VM ops
> arm/arm64: KVM: move kvm_register_device_ops() into vGIC probing
> arm/arm64: KVM: dont rely on a valid GICH base address
> arm/arm64: KVM: make the maximum number of vCPUs a per-VM value
> arm/arm64: KVM: make the value of ICC_SRE_EL1 a per-VM variable
> arm/arm64: KVM: refactor MMIO accessors
> arm/arm64: KVM: refactor/wrap vgic_set/get_attr()
> arm/arm64: KVM: add vgic.h header file
> arm/arm64: KVM: split GICv2 specific emulation code from vgic.c
> arm/arm64: KVM: add opaque private pointer to MMIO data
> arm/arm64: KVM: add virtual GICv3 distributor emulation
> arm64: GICv3: introduce symbolic names for GICv3 ICC_SGI1R_EL1 fields
> arm64: KVM: add SGI generation register emulation
> arm/arm64: KVM: enable kernel side of GICv3 emulation
> arm/arm64: KVM: allow userland to request a virtual GICv3
> Documentation/virtual/kvm/devices/arm-vgic.txt | 21 +-
> arch/arm/include/asm/kvm_emulate.h | 5 +-
> arch/arm/include/asm/kvm_host.h | 3 +
> arch/arm/include/asm/kvm_mmio.h | 1 +
> arch/arm/kvm/Makefile | 1 +
> arch/arm/kvm/arm.c | 23 +-
> arch/arm/kvm/psci.c | 17 +-
> arch/arm64/include/asm/kvm_emulate.h | 5 +-
> arch/arm64/include/asm/kvm_host.h | 5 +
> arch/arm64/include/asm/kvm_mmio.h | 1 +
> arch/arm64/include/uapi/asm/kvm.h | 7 +
> arch/arm64/kernel/asm-offsets.c | 1 +
> arch/arm64/kvm/Makefile | 2 +
> arch/arm64/kvm/sys_regs.c | 37 +-
> arch/arm64/kvm/vgic-v3-switch.S | 14 +-
> drivers/irqchip/irq-gic-v3.c | 14 +-
> include/kvm/arm_vgic.h | 34 +-
> include/linux/irqchip/arm-gic-v3.h | 44 +
> include/linux/kvm_host.h | 2 +
> include/uapi/linux/kvm.h | 2 +
> virt/kvm/arm/vgic-v2-emul.c | 805 ++++++++++++++++++
> virt/kvm/arm/vgic-v2.c | 3 +
> virt/kvm/arm/vgic-v3-emul.c | 1020 +++++++++++++++++++++++
> virt/kvm/arm/vgic-v3.c | 89 +-
> virt/kvm/arm/vgic.c | 1065 ++++++------------------
> virt/kvm/arm/vgic.h | 122 +++
> 26 files changed, 2469 insertions(+), 874 deletions(-)
> create mode 100644 virt/kvm/arm/vgic-v2-emul.c
> create mode 100644 virt/kvm/arm/vgic-v3-emul.c
> create mode 100644 virt/kvm/arm/vgic.h
More information about the linux-arm-kernel